ANALYSIS OF CONSUMER PREFERENCE IN PRODUCT
ATTRIBUTES: A CASE OF COMMON BEANS IN KIAMBU
COUNTY, KENYA
HANNAH WAIRIMU GITONGA
Al 03/14475/2009
A Thesis Submitted in Partial Fulfillment of the Requirements for the
Award of Degree of Master of Science (Agribusiness Management and
Trade) in the School of Agriculture and Enterprise Development of
Kenyatta University
March 2015
DECLARATION
I Hannah Wairimu Gitonga, declare that this thesis is my original work and has not been
presented for the award of a degree in any other university or any other award.
Signature: . . DatJD!rd!.?P.1,5.. .
Hannah Wairi u Gitonga (AI 03114475/2009)
Department of Agribusiness Management and Trade
SUPERVISORS
We confirm that the work reported in this thesis was carried out by the candidate under
our supervision and has been submitted with our approval as university supervisors.
Signatur~: . Date .. L ..{ ~t~~.I.!f.
Dr. Eric Bett (PhD),
Department of Agribusiness Management and Trade,
Signature: L......................... Date J.~..\.~..\.~13?\( .
Dr. Patrick Mbataru (PhD),
Department of Agribusiness Management and Trade,
11
DEDICATION
To my husband Chege and daughter Christine thanks for your patience and support.
111
ACKNOWLEDGEMENTS
I thank God Almighty for enabling me reach this far in my studies. His Grace saw me
through my ill health, hospitalization, recovery and gave me strength to complete my
studies. I sincerely thank my supervisors Dr. Eric Bett and Dr. Patrick Mbataru who
relentlessly guided me through the whole research work. Their technical guidance and
moral support played a key role in what I have been able to accomplish in regard to this
study. I also thank members of staff in the School of Agriculture and most especially in
the Department of Agribusiness Management and Trade for the support they accorded me
during my studies. I am highly indebted to the International Center for Tropical
Agriculture (CIAT) for funding the whole research work, through Kenya National Bean
Program. I am deeply grateful to Dr. Eliud Birachi of CIAT for the technical expertise he
extended to me in addition to the funding I got from CIAT. I also thank Mr. David
Karanja, coordinator Kenya National Bean Program for information on KAT bean
varieties as well as taking care of logistics during the entire research work. I would like to
thank Mrs. Grace Mbugua and KARl staff in Beans section for giving me valuable
information on Grain Legume Program (GLP) bean varieties. To KARl Thika team
which assisted in data collection led by Scolastica Wambua, Ministry of Agriculture staff
in Thika East and West Districts, I am sincerely grateful for doing the work diligently.
All my relatives and friends for enduring my busy study schedule. Lastly I am thankful to
my mum and late dad, who were visionary and set high standards for all their children -
your efforts were worthwhile and I am forever indebted to you.
v2.7 Methods of measuring willingness to pay 16
2.8 Analytical methods to determine willingness to pay 18
2.9 Critical review of consumer preference studies 22
CHAPTER THREE 25
3.0 MATERIALS AND METHODS 25
3.1 Introduction 25
3.2 Location of study 25
3.3 Sampling technique 26
3.4 Research instrument and data collection 29
3.5 Measurement of variables and data analysis 31
3.5.1 Hedonic model specification 33
CHAPTER FOUR 38
4.0 RESULTS 38
4.1
4.2
4.3
4.3.1 :
4.3.2
4.4
4.4.1
4.4.2
4.5.
4.5.1
4.5.2
4.5.3
4.5.4
4.5.5
Introduction 38
Socioeconomic characteristics, consumption and residence of respondents38
Evaluation of consumer preference for cornmon bean varieties .42
Consumer preference in cornmon bean varieties based on dwelling place .. .44
Respondents age and preference in cornmon bean varieties .45
Evaluation of consumer preference in attributes of cornmon beans 45
Consumer preference in bean attributes after pairwise comparison .45
Attribute ranking according to variety 46
Evaluation of consumer willingness to pay for ranked attributes 53
KAT X 56 Gituru 54
KAT B9 56
GLP 2 Rosecoco (Nyayo) 56
GLP 24 Canadian Wonder 56
GLP 585 Red Haricot. 56
VI
CHAPTER FIVE 57
5.0 DISCUSSION 57
5.1 Introduction 57
5.2 Respondents' socio economic characteristics 57
5.3 Consumer preference in common bean varieties 59
5.4 Consumer preference in attributes of common bean varieties 62
5.5 Effect of preferred attributes on willingness to pay price 69
CHAPTER SIX 76
6.0 CONCLUSION AND RECOMMENDATIONS 76
6.1 Introduction 76
6.2 Conclusion 76
6.3 Recommendations 78
REFERENCES 81
Vll
LIST OF TABLES
Table 3.1: Description of study area 26
Table 3.2: Probability proportional to size sampling for bean traders 28
Table 3.3: Description of variables that were evaluated 35
Table 4.1: Socioeconomic characteristics of respondents (n=212) 39
Table 4.2: weekly b.ean consumption based on region 42
Table 4.3: Consumer preference scores for common bean varieties (1-7) 43
Table 4.4: Different age groups' Preference in bean varieties 45
Table 4.5: Results of pairwise comparison of common bean attributes 46
Table 4.6: GLP 2 keeping quality ranking by different age groups 48
Table 4.7: Ranking of KAT X 56 Gituru's taste by different age groups 50
Table 4.8: Ranking of GLP 92 Mwitemania color by different age groups 51
Table 4.9: KAT X56 flatulence ranking by different age groups 52
Table 4.1O:Hedonic model results for bean attributes 55
V111
LIST OF FIGURES AND PLATE
Fig. 1.1: Decision making process 8
Fig 4.1: Respondents weekly bean consumption and monthly incomes 40
Fig 4.2: Respondents' weekly bean consumption and occupation 41
Fig 4.3: Bean varieties sold in the market 43
Fig 4.4: KAT B9 cooking quality ranking by different age groups 47
Fig 4.5: Market prices and willing to pay price for a kilo of beans 49
Plate 1: Seven bean varieties used in the study 30
IX
APPENDICES
Appendix 1: Map of larger Thika district showing study area 89
Appendix 2: List of released bean varieties since 1980 to 2010 90
Appendix 3: Data collection and analysis plan 92
Appendix 4: Summary of consumer preference literature 93
Appendix 5: Survey questionnaire 98
Appendix 6: Pairwise comparison of bean attributes 106
Appendix 7: Attribute ranking according to variety 107
Appendix 8: Evaluation of common bean attributes for consumer preference 109
Appendix 9: Consumers source of nutrition information of common beans 109
Appendix 10: Breusche-Pagan heteroscedasticity test results 1JO
Appendix 11: Regression results for seven bean varieties 110
l. Regression results for KAT X 56 Gituru ll0
2. Regression results for KAT B9 Gacuma 111
3. Regression results for GLP 2 Rosecoco .111
4. Regression results for GLP 92 Mwitemania .112
5. Regression results for GLP 24 Canadian Wonder 112
6. Regression results for GLP 585 Wairimu .113
7. Regression results for KAT B 1 Katheka 113
Appendix 12: Variety ranking based on consumers dwelling place 114
Appendix 13: Bean consumption frequency by different age groups 114
Appendix 14: Consumers education level and consumption frequency 114
Appendix 15: Correlation coefficients of attributes in bean varieties 115
xDEFINATION OF TERMS
Grain color: refers to the color beans impart into the food.
Grain size: refers to the expansion and visibility of the bean upon cooking.
Price: the market price of each variety during the time of study. Bean varieties have
different prices no matter the season.
Cooking time: refers to the duration a variety takes to cook. This study evaluated
whether the time taken to cook a variety was acceptable to the consumer.
Cooking quality: refers to the structure of the cooked bean, whether it mashes up or
remains whole.
Keeping quality: refers to the ability of the variety to stay fresh without spoiling. The
benchmark was two days, being the normal time well preserved boiled beans can
be stored under natural conditions without spoilage.
Flatulence: refers to the discomfort of excessive gas experienced after consuming beans.
The discomfort experienced is different for each variety.
Willingness to pay: refers to the amount of money consumers state they would pay for a
particular variety based on the attributes of that variety.
APA -
CGIAR
CIAT
DAO
FAO
GDP
GLP
GoK
KAT
KNBS
MOA
NARS
WTP
M-asl
Xl
ACRONYMS AND ABBREVIATIONS
American Pulse Association
Consultative Group on International Agricultural Research
Centro International de Agricultura Tropical
District Agricultural Officer
Food and Agricultural Organization
Gross Domestic Product
Grain Legume Programme
Government of Kenya
Katumani
Kenya National Bureau of Statistics
Ministry of Agriculture
National Agricultural Research Systems
Willingness To Pay
metres above sea level
XlI
ABSTRACT
Common bean (Phaseolus vulgaris L.) is an important source of livelihood and food for
approximately three million households in Kenya. Consumers appreciate common bean
more due to its nutritional value and health benefits. Between 2005 and 2009, a total of
403,604 MT of bean with a value of US$ 199,743,000 was produced in Kenya. The
Kenyan bean market has a deficit of 14,256 metric tons and is dominated by old
improved bean varieties, an indication of consumer preference for those beans. This is
despite new varieties being released into the market following intensive research and
breeding work done by research institutions. Consumer preference assessment gives
important information on acceptability of a commodity by consumers. The primary
objective of this study, therefore, was to analyze consumer preference for common bean
varieties by attribute sensory test and willingness to pay for preferred attributes. This
study focused on bean consumers and traders in two districts, Thika East and Thika West
of Kiambu County. The region was chosen as a test bed for this study due to high
utilization of common beans in most of the diets among the residents. Additionally the
two districts were selected because of their high population, diverse socioeconomic
characteristics of residents, and their rural and urban living setups. Semi structured
questionnaires were used to elicit information from 212 consumers and 67 traders who
were randomly selected. Bean variety preference was assessed using a preference scale of
1-7 score. A pairwise analysis of eight bean attributes was done to assess preference of
bean attributes. This was followed by assessment of attributes in seven bean varieties
using likert scale of 1-5 rank. A hedonic price model was used to analyze effect of
preferred attributes on price consumers were willing to pay. Data analysis was done using
descriptive and inferential statistics in Excel and SPSS software programs. Results
showed that beans were an important part of respondents diet with majority of
respondents (86%) consuming beans more than once a week. Rural respondents
consumed beans more frequently compared to urban respondents; difference in
consumption was statistically significant (p-value =0.025). Beans were popular with
women (83%) and were consumed by all age groups but there was more consumption in
the 31-40 years age group (26.8%). GLP 585 was ranked l ", GLP 2 was ranked 2nd and
KAT X56 was ranked 3rd in preference by 64.7%, 43% 39.8% respondents respectively.
GLP varieties were popular among urban respondents while rural respondents consumed
both GLP and KAT varieties. Consumers had preference for cooking quality, keeping
quality, color, taste, low flatulence and grain size attributes associated with GLP 585,
KAT X56, GLP 2 and KAT B1 varieties. Consumers were willing to pay a premium for
taste, price, cooking time and discount for grain size of GLP 585. They also discounted
grain size in KAT X56 and KAT B9. Other discounts were in color of KAT B9, taste of
GLP 2 and flatulence of KAT X56 varieties. Based on the findings of the study, it is
recommended that government supports breeding and improvement programs to ensure
seeds with preferred attributes are available and affordable to producers. This would
enhance acceptability and utilization of beans by consumers. It is further recommended
evaluation of KAT B l' s, low consumer preference, yet it has preferred attributes.
1CHAPTER ONE
1.0 INTRODUCTION
1.1 Overview
The chapter gives the basis of the study. It gives the background of the study,
statement of the problem under investigation, objectives, hypotheses, significance, scope
of the study and conceptual framework.
1.2 Background to the problem
Modem food industry faces the challenge of developing food products in
accordance with consumer needs (Bech et al., 1997). This is as a result of global and
regional integration which has exposed consumers to diverse commodities subsequently
changing their preferences. Research in common bean by National Agricultural Research
Systems (NARS) has been oriented towards increasing yield and producing surplus for
sale and improving nutrient content- biofortication, as a strategy for family auto-
sufficiency, alleviate malnutrition, hunger and poverty (Katungi et al., 2009), occasioned
by among others; increasing population growth and increasing cost of agricultural
products especially animal related. Emphasize on improved production technology
research, has left out consumer preference an important component in acceptability and
marketing of products.
The common bean is an important crop for small-scale farmers grown by more
than three million households in Kenya (Katungi, et al., 2010). It has short growth cycle
which permits production when rainfall is erratic. It also provides income to the
household and food to the consumer before harvesting of other long season crops such as
maize. Common bean is cultivated twice a year in March to April and September to
2October at altitudes between 600-2000 meters above sea level. Bean varieties have
different attributes which determine their attractiveness to consumers. These attributes
are heterogeneous, making each variety distinct. Wortmann et al. (1998) classified
common bean varieties into nine major classes according to color and size as follows:
1. Pure large reds. 2. Medium. 3. Small reds. 4. Red mottled. 5. Purple.
6. Yellow /tans. 7. Cream. 8. Navy/white. 9. Black.
In Kenya, the annual bean production in the period between 2005 and 2009 was
403,604 metric tons worth about US$ 199,743,000 (FAO, 2011) with an annual per
capita consumption of 14 kg to 66 kg (Spilsbury et aI., 2004; Rubyogo et aI., 2007). This
is an indication that bean trading can contribute towards injecting 80-90 billion Kenya
Shillings into the Gross Domestic Product, thus boosting the realization of Vision 2030
(GoK, 2007). However, the amount of bean produced in Kenya is not sufficient to meet
domestic needs. According to Kibiego et al. (2003) and Mauyo et al. (2010), the
unrecorded annual bean imports from Uganda are estimated to be 9,300 MT while
recorded imports are 1,700 MT. Katungi et al. (2009) places imports at 14,256 MT. The
deficit is expected to increase given the increasing population and urbanization (Kibiego
et aI., 2003). The deficit is an indication of local market failure to stimulate production.
Consumer's choice of bean type is influenced by among others, the food dishes to
be made such as; mixture of beans and maize popularly refered to as githeri in Kenya,
Ngata in Malawi, Kande in Tanzania. Bean sauce is another dish which is served with
accompaniments such as rice and chapati, a common pancake in East Africa. Animal
protein is expensive and has been attributed to negative health implications such as
urinary calcium excretion, associated with osteoporosis. On the other hand high
3potassium levels in legumes decreases urinary calcium (Massey, 2003). Nutrition content
of beans is about 60% carbohydrates, two-thirds of which is in the form of starch, 22% to
25% protein and very low fat content. According to Schwartz & Corrales (1989) beans
contribute one-sixth of total per capita protein intake in the East African highlands.
Compared to cereals, they are a valuable source of protein that supplements well, the low
quality proteins in cereal and are highly valued as staple food by consumers (Tapia, 1985;
EI-Tabey, 1992; Ruiz de Londono et al., 2000; Murray, 2010). Nutrition benefits and low
cost make common beans the alternative food choice to animal protein.
Common bean plays an important role in the soil fertility stabilization through
biological nitrogen fixation (Katungi et al., 2009). Rhizobium bacteria in bean nodules
supply the plant with fixed nitrogen, in form of ammonia, and get carbohydrates in
return. This factor is fundamental in mitigation of greenhouse gas emission. Excessive
use of fertilizers results in emission of climate change causing gas known as nitrous oxide
(N20) (Smith, 2008). Common bean therefore enhances sustainability of agricultural
production systems.
The important and diverse roles played by common bean, in the farming systems
and in consumer diets, makes it an ideal crop for achieving Millennium Development
Goal (MDG) one, five, six and seven; poverty and hunger eradication, improved maternal
health, low major disease incidences and sustainable environment.
Previous research on beans concentrated on agronomic aspects resulting in high
yielding bean varieties with little attention given to marketing aspects. As a result little is
known about consumer preference for the bean varieties. Market demand, which reflects
what consumers want, influences supply of commodities in the market. As indicated in
4Mishili et al. (2009) consumers are the beginning of the value chain from which the flow
of information about food preference moves back to retailers, manufacturers, farmers and
scientific laboratories. Anti-nutritional aspects of beans such as flatulence, long cooking
time may reduce their consumption (APA, 2010). Information on bean varieties/attribute
preference is therefore fundamental in enhancing utilization, development of bean market
and subsequently in stimulating production. It is also important considering the resources
and efforts which are directed towards development of alterriative varieties and
characteristics of agricultural commodities (Espinosa & Goodwin, 1991).
1.3 Problem statement
There is a wide range of bean varieties with different physical and sensory
properties. As a cheap and beneficial source of protein compared to other animal products
such as meat, there is need to know the effect of these different bean properties on choice
of beans varieties by consumers in order to enhance utilization. There has been extensive
research conducted on dry beans in relation to agronomic aspects which has resulted in
high yielding bean varieties that withstand biotic and abiotic stresses. Despite this, the old
improved low yielding varieties dominate market share and the country is bean deficit.
There is therefore a knowledge gap on what consumers prefer in the old beans that is
probably not in the recently released bean varieties whose agronomic properties have
been improved. Lack of consumer preference analysis, could be a factor that limits
utilization, subsequently low production of newly released varieties. The problem
therefore is insufficient information on the factors that determine choice of beans by
consumers in the market. The overall objective of this study is to fill this knowledge gap.
51.4 Overall objective
The overall objective of this study was to evaluate the consumers' preference in
common bean varieties and their willingness to pay for them.
1.4.1 Specific objectives
1. Evaluate the consumer preferred bean varieties in the market for attribute ranking.
2. Evaluate attributes that influence consumer preference for common bean varieties.
3. Evaluate consumer willingness to pay for preferred attributes in common bean
varieties.
1.5 Hypotheses
The study hypothesized that;
• There is no significant difference in consumer preference for different common
bean varieties.
• Consumer preference in attributes of different bean varieties is not significantly
different.
• Consumers' preference in bean attributes does not significantly influence their
willingness to pay for beans.
1.6 Significance of the study
Despite the many bean varieties that have been developed, the old improved bean
varieties continue to be popular with consumers. The study was important in establishing
the sensory attributes that made consumers prefer some bean varieties. Information from.
this study will provide insights to policy makers, government institutions and other
development agencies for research prioritization in common beans. Once the relevant
6institutions incorporate the recommendations in bean improvement and breeding
programs, it is expected that bean value chain will become more vibrant following supply
of beans with consumer preferred attributes. This in return will increase utilization and
trading of beans, subsequently contributing towards income generation along the bean
value chain, a healthier population through increased bean consumption and contribute to
the country's realization of food self sufficiency through an increase in participation of
bean producers in the bean value chain. Lastly the findings of this study will contribute to
the existing knowledge of consumer behaviour and especially in attribute preference in
relation to bean choice knowledge gap. The findings will also be a base for further
research in the same or other fields.
1.7 Scope of the study
The study focused on two value chain actors, (trader and consumer) from Thika
East and Thika West districts. The study concentrated on seven bean varieties, namely,
KAT X56 Gituru, KAT B9 (Red Haricot), KAT Bl Kayellow, GLP 2 Rosecoco, GLP 24
Canandian Wonder, GLP 585 Red Haricot and GLP 92 Mwitemania. All the selected
varieties were consumed in the study area as was established during an exploratory study
conducted in the study area in March 2012. It was therefore easy for consumers to
evaluate the beans since they were familiar with them.
1.8 Conceptual framework
Food choice is comprised of decision making process and factors influencing
these decisions. The study was based on consumer theory, where decision process
involves evaluation of alternatives and subsequently making a choice, which is equated to
purchase of commodity that will provide maximum utility, while factors influencing
7decision are the consumer needs and resources available. As Figure 1 shows, it was
hypothesized that consumers would evaluate characteristics of different bean varieties
and choose varieties with attributes that provided them with highest utility; the rank they
assigned an attribute depicted the level of utility it provided. Evaluation was based on
previous experience of the bean or information of available alternative varieties.
It was hypothesized that consumers made choices among the many varieties that
were available in the study area. They were therefore expected to assign levels to
varieties they preferred from one (1) in a descending order.
Attributes in the different varieties were expected to contribute to consumer
preference of a particular variety. The consumers were therefore expected to rank the
different attributes in beans. The way to measure consumer preference for attributes,
therefore, was by ranking attributes on a likert scale, 5-1: Exc.ellent to Very bad for each
variety a respondent consumed.
It was expected that consumers would pay a certain price for a variety depending on
the level of satisfaction provided by the attributes in the variety. By stating the amount
they were willing to pay for varieties, it was expected that consumers showed the value
they attached to attributes in those varieties. Hedonic price function was presented as:
Pj = aj + L~jZj + £j .................••........•....•.•..•...........••••.•••.•••••••..•.. (1)
Where: Pi = Bean price. Ui~j = Estimated coefficients.
Z, = A vector of bean attributes. ci = Random error.
Purchase would depend on individual differences, such as available resources and
motivatio~ consumer got after evaluating the attributes. Preferred attributes which was
the output of the study, can be used in development of appropriate bean technologies.
8This will lead to enhanced consumption, whose effect will be more trading in beans and
ultimately production intensification.
Consumer H Bean Varieties Intervention measures
Situation: Need 1 / Inform breeders on preferredrecognition attributes (Output) for
Evaluation of Bean technology development.
Attributes; '",/ ~maximize utility+ Increased bean consumption,trading and production
Willingness to Pay
for preferred bean !attributes Expected outcomes, Improved
nutrition, Increased incomes
Improved livelihoods
Fig. 1.1: Decision making process
Source: Modification of Engel et al. (1995).
9CHAPTER TWO
2.0 LITERATURE REVIEW
2.1 Introduction
The chapter consists of literature review that is relevant to this study. It covers
bean trading, improvement programs and bean attributes. It reviews methods of
measuring consumer preference and gives critical review of empirical studies on the
same. Overall, literature on bean preference concentrates more on varieties without
evaluating attributes of those varieties.
2.2 Socio economic characteristics
Socio economic characteristics such as age, income and gender playa crucial role
in acceptance of products in the market. They influence the amount of money spent on a
product and consumption patterns. A consumer may use enough amount of product
because he understands its nutritional value based on his level of education. On the other
hand a consumer may use less of a product because it is unaffordable (Mundua, 2010). In
Groote & Kimenju (2008) consumer preference for traits under study were influenced by
consumer socioeconomic and cultural background. Studies on socio economic
characteristics can inform traders and organizers of farmers' markets in coming up with
strategic locations for product outlets (Govindasamy, Italia, & Adelaja, 2002).
2.3 Bean marketing
Farmers grow beans not only for their own domestic consumption, but as a source
of income. Many farmers value beans as a fast-growing crop, which can be converted
easily and regularly to cash, especially during times of need. The availability of market
for beans domestically and regionally, according to Odendo et al. (2002) makes it a
10
potential income and food security crop. It is therefore important to understand its
consumers. According to Munene (1993) beans are accepted by different communities in
Kenya going by the beans in markets across all counties in Kenya. This study showed a
composition of 30 bean varieties in 21 markets surveyed with variation in prices.
According to Kimani et al. (2005) and Korir et al. (2005) farmers in Northern Tanzania
and Eastern and Southwestern Uganda produced red-mottled beans for sale in Nairobi
and other urban centers in Kenya. This indicates a ready market for dry beans.
The average annual bean import in Kenya is 14,256 metric tons (Katungi et al.
2009) with 9,300 tons informally imported (Kibiego et al., 2003; Mauyo et al., 2010).
Approximately, 70 % of beans in Thika market are imported (Karanja D. personal
communication March 23 2012). The above studies show existence of preference for
particular bean varieties whose deficit is compensated through importation.
2.4 Bean Improvement Research in Kenya
A Grain Legume Project was established at KARl Thika in early seventies, to
cater for bean research and development. It released six bean varieties in 1980s namely
GLP-2 (Rose coco), GLP-24 (Canadian wonder), GLP-1004 (Mwezi moja), GLP-x.92
(Mwetemania), GLP-x.1127(a) (New Mwezi moja), GLP-585 (Red haricot), (Munene,
1993). Appendix 2 shows the GLPs. KARl Katumani released two varieties in 1987, Kat
Bean 1 and 2 (KEPHIS, 2011). By 2008, twenty one more improved varieties had been
released into the market by Universities and other research institutions. Research has
mainly been geared towards mitigating for biotic and abiotic constraints in order to
increase yield. This is evident in last column of variety release list in appendix two.
However, consumers who purchase beans from the market will only buy beans that meet
11
their preference regardless of agronomic constraints, (Santalla et aI., 1999; Buergelt et
aI., 2009; GoK a, 2010).
The Katumani (KAT) bean varieties are improved bean varieties developed by
Kenya Agricultural Research Institute (KARl), Katumani. The institute in partnership
with the International Centre for Tropical Agriculture (CIAT) has been promoting the
varieties in different parts of the country under Tropical Legume II project. The project
was initiated to support dissemination and promotion of the improved bean varieties in
Central, Rift Valley and Western Kenya (Karanja, et al.,2012). A similar promotion
program was introduced in Tanzania by CIAT in partnership with East and Central
African Bean Research Network (ECABREN). The program has resulted in an increase
of beans in Tanzania which are exported to Kenya (Katungi et al 2010).
2.5 Consumer preferences
Consumer preference is a tool that is used in marketing research to gauge
consumer satisfaction (utility maximization) and acceptance of a given commodity,
(willingness to pay for a particular commodity). It helps reveal an option that has the
greatest anticipated value among a number of options. Modeling and measuring
consumer preferences is therefore useful in designing of new or upgrading products and
services.
Consumers choose from the market goods that will satisfy their needs given the
amount of money available. According to Economides (2010) to get the best choice,
consumers undertake several steps; they analyze choices available to them given their
limited funds, for example, the different bean varieties that are sold in the market. Next
they analyze their preferences given choices that are available. Highest preference is
12
given to the good whose attribute combination offers maximum utility. When consumer
chooses one level of attribute against a similar one in a different variety in order to
maximize utility, a tradeoff occurs; marginal rate of substitution. An additional unit of
attribute X will increase level of satisfaction of a consumer by the marginal utility of the
attribute X. This study applies Stated Preference Technique by Pearmain et al. (1991) as
explained in Abley (2000), to estimate utility. The technique uses individual respondents'
statement about their preferences in a set of options to estimate utility. The final stage is
to get an optimal choice by combining analysis of the preferences with available choices.
For a long time technology development has been focused on quantity of
commodities, leaving out consumer preferences; an important component in the
marketing chain. Without good acceptability/preference characteristics, a new crop
variety will find no market, and thus be unprofitable to the producer (Luse, 1980). One
way of measuring consumer preferences is by employing Willingness to Pay (WTP)
technique. Willingness-to-Pay is defined as the maximum price that can be charged
without reducing the individual's welfare and utilization of the product. Empirical studies
have documented that some market segments are willing to pay a premium for food
products with differentiated attributes. In a study by Padilla et al., (2007) consumers were
willing to pay 585 pesos more for homemade marmalade with a certified quality label.
Mclennon, (2002) documented that consumers were willing to pay for non-meat biotech
food, compared to biotech meat products.
2.6 Bean attributes in relation to consumption
Consumers start their decision making with attribute comparison and then turn to
brand evaluation in which trade-offs and comparisons of alternatives are made. The
13
existing research literature on customer evaluation of alternatives prior to choice reveals
the crucial role of identifying the attributes affecting the customer's decision in order to
understand customer choice among alternatives. Food intake is determined by among
others, food availability and cost, preparation time, palatability, bulk, anti-nutritional
factors and digestibility (Kaul, 1987). These factors have not always received a due and
balanced consideration in research.
Beans are heterogeneous in varieties and attributes, which appeal to consumers in
different ways (Mbugua et aI., 1997; Katungi et aI., 2010). Following sustained
popularity of old bean varieties, evaluation of consumer preferences has become
necessary before development of new varieties, for the farmers to produce marketable
varieties (Munene, 1993; Katungi et aI., 2011; Gichangi et aI., 2011). Bean attribute
evaluation is important to establish positions these characteristics are given by consumers
during the purchase process, for effective bean grain improvement, development and
business establishment.
Beans are found in different sizes namely: small, medium and large sizes. One
way of differentiating beans in the market is by grain size. According to Gichangi et aI.,
(2011) 69% farmers and 82% traders differentiate bean products by grain size in the
Central Rift Districts. According to Maryange et al. (2010) beans are classified small if
there are 900 seeds per kg, medium; 600-899 seeds per kg and large if there are less than
600 seeds per kg. Katungi et aI., (2010) classified beans in Kenya as follows: Small less
than 25g, Medium 25-40g, Large more than 40g per 100 seeds. Individual bean grain
sizes range from 0.7 to 1.9 centimeters (Mbugua & Munene, 1997).
14
The decision to take the attribute of taste into account when defining acceptance
was based on numerous studies that indicated taste as the single largest determinant of
food choice, directing consumers to eating, regardless of constraints of production
(Spilsbury et ai., 2004). In Deodhar and Intodia (2002) study of traits in clarified butter
that influenced daily price, it was found that consumers were willing to pay a premium
for branded clarified butter over the non-branded. Consumers attached economic
significance to flavor. In a similar study for rice characteristic, done in Ghana by Anang
et al. (2011) aroma had economic significance. Beans have diverse taste ranging from
beany to 'sweet taste (APA, 2010). It is important to establish whether taste influences
consumers preference in common beans.
Beans need to be cooked for long to ensure Lectin; a protein found in lentils is
well cooked. If not well cooked, Lectin can cause allergic reactions in some consumers.
Cooking time has implications for the rural and urban poor, gender equity and
conservation of biodiversity. For decades, 90% of consumers in Brazil have been
consuming a combination of bean and rice meal daily. However consumption of beans
has decreased significantly due to lifestyle changes which leave consumers with little
time to prepare and cook raw beans (Canada, 2009). Fast cooking food commodities save
on time and fuel cost. Different bean varieties have different cooking time (Maryange et
al., 2010) which range from three hours for unsoaked beans in Kenya to 103,96,56 and
105 minutes for CAL 96, MCM 5001, white Haricot and Uganda K2 varieties in Uganda
respectively (Kim ani et ai., 2005). Reduction in cooking time cuts down fuel
consumption, in the process reducing environmental degradation and fuel cost (Diamant
et al., 1989; Bressani & Chon, 1996).
15
Bean grain color is of great importance to some consumers such that producers
grow beans with particular color that is preferred in a particular area. In Maryange et al.
(2010) bean colors range from white all through the color spectrum to black; either plain
or speckled.
Beans contain oligosaccharide; 3-5 sugars bound together in a way that human
body cannot digest or absorb them. Bacteria in the intestines break and digest
oligosaccharides, producing gas in the process (flatulence). Flatulence can lead to
reduced consumption, a case in point is Brazil where there has been a steady decline in
pulse consumption due to health aspect related to flatulence effect of beans (Canada,
2009). Some beans have low flatulence effect such as the Manteca bean, grown in China
for centuries. This variety produces tannins in the seed coat that bind to calcium in the
intestines in ways that change the pH and chemistry of digested food enough to prevent
gas formation (American Bean Organization, 2008 ). Soya bean (medium sized purple
bean) grown in Northern Tanzania causes low flatulence, (Korir et aI., 2005).
Cooking quality refers to the cooked structure of the bean. The attribute takes into
account cooking time, density, hydration capacity, swelling capacity and whole grain
(Coelho et al., 2009). According to Coelho et al. (2009); Mwangwela in (Maryange et al.,
2010 ) there is notable varition of bean cooking quality in different genotypes, a clear
indication that the attribute is manifested diffently in the various varieties. Splitting and
mushing up of cooked beans is one of the undesirable cooking quality characteristics
(Maryange et aI., 2010 ).
16
2.7 Methods of measuring willingness to pay
To understand consumer behavior and relative importance of various attributes in
food purchase, various techniques have been applied (Kiesel & Villas-Boas, 2007).
Willingness to pay (WTP) techniques are devised to elicit people's monetary valuations
of costs and benefits for goods and services. They can be classified into two, revealed
preferences which can be derived from market data or experiments and stated preferences
derived from direct surveys or indirect surveys. Market data involves collection of
individual purchase data of a customer panel member or sales records from retail outlets.
The advantage of using this method is that real purchases are used instead of stated
purchase intentions. There is limitation in that it is not possible to estimate WTP for new
products or hypothetical products that are not yet in the market. In experiments, purchase
behavior is simulated by giving the subjects an amount of money and asking them to
spend on a specific selection of goods. It is not always possible to obtain the data
required in revealed preferences in order to estimate price-response function. For
example differentiated products have to be manufactured before they can be tested
experimentally. Practically, the expenditure and time needed to carry out experiments
make them less favored in product evaluations (Vo1ckner, 2006 ).
Stated preferences are methods of measuring WTP based on consumer statements.
They can be derived from direct surveys (contingent valuation) and indirect surveys
(choice modeling). Direct surveys can further be classified into - expert judgments and
customer survey. Indirect surveys comprise of conjoint and discrete choice analysis. In
indirect surveys customers are presented with product profiles with systematically varied
prices and are asked to indicate whether they would purchase the good at the given price
17
or not (Marbeau, 1987). In a study by Mennecke et aI., (2007) respodents were asked to
choose meat they liked from pictures of different beef cuts and combination of meat
origin, animal breeds, nutrition. One of the limitations of using indirect surveys is that the
customer must be willing to purchase the product as presented and at the given price -
status quo product, which is not realistic market behavior (Breidert, Hahsler, & Reutterer,
2006). Using a status quo product may not yield the correct WTP predictions due to
consumer heterogeneity; different participants might consider different products their
best alternatives.
Profiling is complex and difficult to present in some products. A profiling trial
done for this study yielded profiles which left out some of the attributes that had been
selected by consumers. Direct surveys require respondents to state how much they are
willing to pay for a specific product or bundle of attributes. The objective of direct survey
- contingent valuation methods is to provide the researcher with monetary valuations of
the target goods, whereas choice modeling methods target either monetary valuations or
preference order outcomes (Brown, 2003). Open-ended CV is a direct method asking the
respondents to state their maximum willingness to payor minimum willingness to accept
for a change in their utility compared to the status quo situation (Hanley, Maurato, &
Wright, 2001). In dichotomous-choice contingent valuation the respondents are instead
asked to choose whether they would accept or reject a fixed price for a certain product
(Koistinen, 2010). Some of the advantages of using direct surveys are that they are cost
effective and time efficient. They are flexible enough to include product combinations
and allow for individual level estimation. However the estimation might not give real
purchase behavior like in the market data and experiments techniques.
18
The approach has been applied in safety and environment related policy
evaluations (Breidert et aI., 2006). It has also been used to evaluate agricultural
commodities such as organic food products (Boccaletti & Nardella, 2000; Gil, Graca, &
Sanchez, 2000). It was also used in rice evaluation and indigenous vegetables (Moser,
Raffaelli, & Thilmany-Mcfadden, 2011). This study applied Stated Preference Technique
by Pearmain et al. (1991) as explained in Abley (2000) to estimate utility. The technique
uses individual respondents' statement about their preferences in a set of options to
estimate utility. Direct survey with open ended questions was used for data collection, to
elicit more information by giving respondents a chance to make independent choices,
unlike the dichotomous choice questions which limit respondent choice to status quo or
profiled products (Breidert et al., 2006).
2.8 Analytical methods to determine willingness to pay.
Two main approaches to measure consumer preferences are hedonic and discrete
choice models. Hedonic emanated from Lancaster, an American researcher, who came up
with Lancaster preference theory after expounding on the consumer theory of classical
economics on utility maximization (Lancaster 1966). From the theory he argued that
consumer's choice of a good was for satisfaction derived not from the good as a whole
but from the attributes of the good. Within the context of Lancaster preference theory, an
American economist Rosen (1974) introduced the first equilibrium model of market
supply and demand based on product characteristics. The concept underlying hedonic
model is that the price of a heterogeneous good is a function of the attributes of that good
(Mundua, 2010).
19
As explained in Picard (2010) discrete models such as logit and probit among
others identify importance of characteristics in commodity purchase decision but do not
explain the commodity price. Multinomial logit model is a discrete model that has been
used in willingness to pay studies. Its estimation procedure is the maximum likelihood
(MLE). It helps identify the important product characteristics in a purchase decision. It
however does not explain product prices but instead examines the variables that
determine whether a consumer makes a purchase or not. Random Utility Model which is
also a discrete model takes the sale prices as representative of market price available to
all consumers and not necessarily representative of characteristics of a product
(Palmquist, 2003). Repeat Sales Price Indexes analyze data of commodities that have
been sold at least twice, they show percentage growth in sale prices over time. They
however do not provide information on value of individual commodity characteristics or
on price levels. The hedonic regression on the other hand reveals the expected value of a
product given the characteristics and the expected contribution of each of the
characteristics to that value. The specification for hedonic model is the linear regression
model and the estimation procedure is the ordinary least squares (OLS). The concept has
been applied in many studies ranging from housing and automobile markets to
agricultural products.
Von Oppen (1978) was the first to define plant breeding goals by applying
hedonic estimation. He developed a preference index to evaluate the acceptance of new
food grains. In the study he indicated that evident and cryptic characteristics of a product
are related. This means that some cryptic characteristics can be inferred from evident
characteristics. This however may be difficult to establish in some products making it
20
necessary to evaluate individual characteristic instead of inferring. The notion that red or
yellow apples are sweet while green apples are sour as expressed by (Portugal, 2004),
may not apply in the case of beans.
Abansi et al. (1990) used hedonic pricing model to evaluate consumer preference
for rice quality. The results showed that consumers in Philippines were willing to pay
more for quality characteristics in rice. The study categorized consumers by location;
urban and rural. The findings of the study showed that both groups were price responsive
to changes in quality characteristics. However urban consumers attached higher value to
quality characteristics than rural group. It is therefore important to evaluate preference in
both urban and rural setting to establish whether there is any variation. In a study of wine
market, Schamel, Gabbert, & Witzke (1998) introduced a new dimension of a hedonic
analysis; regionl reputation. U.S. consumers preferred Chardonnay (white wine) to
Cabernet Sauvignon (red wine). This adds to the observation made above that region
may influence consumer preference of a product.
Dalton (2003) used hedonic price model to evalute consumption attributes
perceived important by rice consumers in West Africa. The study was to derive economic
value of upland rice and subsequently advice breeders on consumption traits to be
incorporated in the rice seed, which were not considered in the breeding programmes.
Results showed grain elongation and swelling were important in relation to the amount of
rice prepared and the amount that can effectively feed a household. The swelling
characteristic was perceived to increase in volume thus generating more food with less
grain. The value for the characteristic was 4.5 while for tenderness was 4.3 on the Likert
scale. An important observation was made in this study; yield served as the defining
21
factor for promoting a new variety for official release. However this trait was not
significant in determining willingness to pay. This means production traits do not
necessarily influence consumer choice or preference and thus do not necessarily have to
be included in attribute evaluation for consumer preference. As stated by Dalton (2003)
agricultural agencies should include a broader set of characteristics besides production
during product evaluations, in order to increase producer and consumer surplus in
agrarian economies.
Langyintuo et al. (2004) used hedonic pricing model to evaluate effect of cowpea
characteristics on prices in Cameroon and Ghana. Results showed that seasonality, grain
size, color and insect damage level explained a substantial price variation in both Ghana
and Cameroon. In a study in India and Nepal on Ricebean characteristics that influence
price, relevant characteristics choosen after literature review were moisture content,
Protein, fat, crude fibre, carbohydrates, ash, seed weight, foreign matter, water uptake
capacity, swelling capacity, and color diversity. Mishili et al. (2009) conducted a study in
Tanzania where they applied hedonic price model to analyse consumer preference for
bean grain quality characteristics. The i~vestigated variables included size of bean grains,
grain damage by bruchids, percentage of discolored grain and percentage of mix. Results
showed that consumers placed significant importance on cooking time. All the above
mentioned studies show that hedonic price model is appropriate in evaluating consumer
preference of agricultural products.
In Kenya, Chelangat (2005) conducted a research to explain pricing of three bean
varieties sold in Nakuru Municipal Market using a hedonic price model. The study
established that attributes in Red haricot and Mwitemania explained 22% and 13%
22
change in market price respectively. Flatulence, color and expansion were significant at
95% level of confidence. The study concentrated on consumers based in an urban set up
who depend on what is offered in the market and not what they were able to produce or
access from local producers as is the case with consumers in rural set up, where
production and consumption go hand in hand. Preference in rural set up may vary given
that products are easily acquired (Edmeades 2005).
2.9 Critical review of consumer preference studies.
Most of the research efforts have focused on demand of common beans in the
market and the results are therefore derived from the traders' perspective and not from
the consumers' point of view. Some of these studies were done by Munene (1993),
Mbugua et al. (2005), Katungi et al. (2010) and Gichangi et al (2011). They all
documented GLPs as the most popular varieties. In Munene (1993) the study was
inclined more to trading than consumers' perspective. Results showed there was price
variation between varieties but the reason for variation was not explained. A consumer
preference study based on evaluation of attributes would probably have given the reason
for price variation. The study by Mbugua et al. (2005) was a farmer participatory, where
grain quality characteristics both for consumption and production were evaluated. The
results were not clear for example GLP-2 variety was among the best ranked varieties
with an average score of 4.7. It was not explained whether the score was due to
production or consumption attributes of the variety. The study by Katungi et al. (2010)
indicated the preferred attributes were short cooking. time, color, seed size. However,
information from this study is derived from producers and traders perspective, leaving
out consumers who purchase beans from the market. In the bean marketing study done by
23
Gichangi et at (2011) in Central Rift Districts of Kenya, it was established that GLP
dominated the market. The results were derived from data collected from wholesale and
retail traders who indicated they preferred a differentiated crop either by color or size.
One of the recommendations was that consumer preferences should be evaluated before
embarking on introduction and promotion of market oriented beans.
In the study by Korir et at. (2005) on bean varietal preference in East African
markets, results showed that Maharage soya was the most preferred variety in Tanzania
while Nyayo was ranked number one in Kenya. This study did not compare attributes for
preference in each variety but gave overall varietal rank in different regions. It would
have been of great value to the breeders if the attributes in the preferred varieties were
known.
In a consumer preference study conducted by Laswai, Shayo, & Kundi (2008) on
sorghum and millet, local tradition varieties were more preferred than improved varieties.
The improved varieties had most of the desired attributes in relation to production such as
high yields. The study showed that there was no advocacy for production and utilization
of local varieties but they were dominating at the time of the study. One of the stated
reasons for their popularity was that they were more palatable than improved varieties.
The study did not elaborate what characterized palatability, information that could have
been important for future grain improvement.
Groote & Kimenju (2008) conducted a consumer preference study for color and
nutritional quality in maize in Nairobi, using dichotomus contingent valuation. Results
showed that there was a strong preference for white maize among urban consumers who
would ask for a 37% discount to buy yellow maize. Socioeconomic factors such as type
24
of marketing outlet, income and cultural background played a role in preference of the
two maize types. The study however did not address intrinsic attributes of the two types
of maize that influenced preference which would have greatly contributed in efforts to
promote fortified maize.
25
CHAPTER THREE
3.0 MATERIALS AND METHODS
3.1 Introduction
The section presents information on the tools used for the study and justification
thereof. It contains a brief of the study area and the technique used to arrive at the
sample. It also gives details on how data was collected and methods used for analysis.
3.2 Location of study
The study was done in Kiambu County. Thika West district was selected because
it is an urban setting while Thika East district and Kakuzi represented rural setting. Two
Thika Districts were selected because of diverse socio-economic orientation. The main
Thika town is an industrial town and population is therefore composed of consumers
from different backgrounds who are expected to have diverse preferences. Majority of
Thika residents purchase beans for consumption making it ideal for a consumer
preference study. It is also centrally located in terms of infrastructure among major bean
growing counties of Meru, Embu, Kirinyaga and the bean deficit areas in the tea zones of
Muranga and Kiambu counties. The main economic activity in the rural area is fanning.
The main market, Jamhuri, is a key outlet for both local and imported beans which means
there are many bean varieties. The main market actors are wholesalers and retailers who
supply beans to the study area and other county markets.
In the year 2010 an average of 8,300 hectares out of the total 44,615 hectares
arable land in Thika was allocated to bean production. This produced, approximately
70,650 bags of 90kg with an estimated value of Kshs. 364.2 million (GoK 201Qa).
Producers targeted two main markets in Thika town for sale of surplus beans. These are
26
Jamhuri market which has 153 bean traders and Madaraka market with 63 bean traders.
Annual per capita consumption was high at 60 kilograms. Table 3.1 presents a
summarized description of the study area.
Table 3.1: Description of study area
Parameter Thika East Thika West Source
Area (krrr') 493 382 (OoK,2010a)
Population 77,073 218,544 (OoK,2009)
Households 18,618 91,000 (OoK,2010a)
Altitude in m.asl Above 1500 1555-2400 (OoK,201Oa)
Flainfall inmm 500-900 500-900 (OoK,2010a)
Temperature range (Oc) 18.7-22.4 18.7-22.4 (OoK,2010a)
3.3 Sampling technique
Since the proportion of bean consumers in the study area was unknown,
estimation was done based on the bean consumption in the country and the per capita
bean consumption in the study area. Estimation was done as follows; Bean consumption
in Kenya in 2009 was 406,970 metric tons. Assuming annual per capita bean
consumption was 60 kg (Broughton et al., 2003; OoK, 2010b) and given that country
population was 40 million, total number of bean consumers was approximately 6,782,833
people. This was 16.95% (0.169) of the whole population.
The required sample was calculated using formula developed by Cochran (1963) and
explained in Israel (1992).
ZZPqn = -z- (2)e
Where: n = required sample size. z = confidence level at 95% (standard value of 1.96)
27
p = estimated proportion of bean consumers in the population.
q= 1 - p. e = desired level of precision 5% (standard value of 0.05).
Calculation of consumer sample size was therefore:
n = 3.8416 x 0.169(1-0.169)/0.0025 = 216.
Consumers were selected from all the six divisions in the two districts where a total of
216 people were interviewed. Research area was grouped into ten areas comprising of
both households and workplaces. Three areas were in the municipality; Makongeni 25,
Majengo 29, Thika district and Municipal offices 15. Two areas in Juja: Muchatha (20)
and Gacororo (20). Two areas in Gatuanyaga: Gatuanyaga and Munyu (22) Ngoliba:
Ngoliba and Mukawa (20). Kakuzi: Ithanga (22), Kakuzi (23) Mitumbiri: Thangira (20).
The study included both urban and rural areas, to eliminate possible bias of results that
could be attributed to easy access of beans in the rural areas where production takes
place. Thika town and surrounding estates such as Makongeni, Majengo, Juja and
government offices within Thika municipality were classified as urban while villages in
Gatuanyaga, Ngoliba, Kakuzi, and Mitumbiri were classified as rural. Four
questionnaires were incomplete and could not be used for analysis. The analyzed data
was therefore from 212 questionnaires.
Few markets were included in the study to gauge the preferred bean varieties and
magnitude of transactions. Identification of the markets with more than 14 bean traders
was done with the help of Divisional Agricultural Extension Officers and municipal
council staff and cereal traders' group officials in the relevant areas. A total of 67 out of
394 bean traders were interviewed for this study. 394 traders were approximately 4.6% of
total cereal traders in the study area. Traders from six markets were randomly selected as
28
follows; Jamhuri 25, Madaraka 11, Muchatha 10, Ithanga 7 Ngoliba 7 and Thangira 6.
Selection of traders was proportional to the total number of bean traders in chosen
markets as table 3.2 shows. Traders sample size was calculated using the Chochran
(1963) formula.
Table 3.2: Probability proportional to size sampling for bean traders
Market No. of cumulative Samples Sample
traders total (sampling interval of 6) size per
(x) market
Ithanga 40 40 6,12,18,24,30,36,42, 7
Jamhuri 153 193 48,54,60,66,72,78,84,90,96, 25
102, 108, 114, 120, 126, 132,
138,144,150,156,162,168,
174, 180, 186, 192,
Madaraka 64 257 198,204,210,216,222,228, 11
234,240,246,252,258,
Muchatha 58 315 264,270,276,282,288,294, 10
300,306,312,318,
Ngoliba 42 357 324,330,336,342,348,354, 7
360,
Thangira 37 394 366,372,378,384,390,396 6
Total (n) 394 302
Sample (s) 67 Sample 394 =6
67
interval
Source: GoK (201 Oa) and author.
As depicted in table 3.2, to get required samples per market, 394 which was the
total number of bean traders, was divided by 67, the total sample size, to give sampling
29
interval of six. This was successively added and market sample was given when the
cumulative total for each market was reached.
3.4 Research instrument and data collection
An exploratory study was conducted in Thika town where people were asked
which bean varities they liked. A similar study was conducted in Gatuanyaga and
Ngoliba areas where farmer groups were asked which bean varieties they liked. It was
established the popular bean varieties were Kat X 56 Gituru, KAT B9 (Red Haricot),
KAT Bl Kayellow, Gathika GLP 2 Rosecoco, GLP 24-Canandian Wonder, and GLP 585
Red Haricot. GLP 92 Mwitemania was not very popular. It was also established that
consumers considered such attributes as color, grain size, taste, cooking time, cooking
quality, keeping quality, flatulence, and price when purchasing beans. The seven bean
varieties were named with the help of KARl, Thika and Katumani researchers. It was
further established that the varieties fell into two clasifications based on the year of
release into the market. The classifications were Old Improved Bean varieties (1982-
1984) and New Improved Bean varieties (1989 to date). Appendix 2 shows details of
different varieties and the year the varieties were released. The GLPs used in this study
fell in the Old Improved Bean varieties while KATs fell in the New Improved Bean
varieties. Each variety was packaged in transparent polythene bags which were presented
to consumers by the enumerators for attribute evaluation in the main study. The seven
bean varieties presented in plate 1 were used in the study.
30
No PictureVariety Local
Name
Morphological
Characteristics
1 GLP2
Calima
Rosecoco,N
yayo
Medium red mottled
2 GLP 585 Wairimu
Red Haricot
Small red
3 GLP92 Mwitemania Medium cream
Pinto Sugar, mottled Pinto
Carioca
4 GLP24 Gituru Slim dark red kidney
Canadian shaped
Wonder
5 KAT Gituru
X56Canadia
n Wonder
Rounded large dark
purple kidney shaped
6 KAT B9 Gacuma
Red Haricot
Medium Red
7 KAT Bl Katheka,Ka
yellow
Medium
yellow/green round
shaped
Plate 1: Seven bean varieties used in the study
Source: KEPHIS (2011) and Author
31
A semi structured questionnaire was personally administered to consumers in
order to collect primary data. Administration of the questionnaire was done with the help
of research officers from KARl, Thika and Agricultural Extension Officers in the
different divisions of study area. Secondary data was collected from District Agricultural
Offices in Thika East and West Districts, KARl offices and from existing literature.
Information on socio-economic characteristics of respondents, their bean variety and
attribute preferences was collected for analysis. The required data was both descriptive
and diagnostic, in nature, thus fitting a survey design; the study helped establish popular
attributes of beans and frequency with which they were mentioned as preferred attributes
by consumers. It further helped evaluate influence of attributes on consumer preference
for bean varieties. The study further gauged the amount of money respondents were
willing to pay for varieties with preferred attributes.
3.5 Measurement of variables and data analysis
According to Mutai (2000) measurement is a procedure that assigns numerals to
events, characteristics or responses. Measurement of data facilitates it's analysis in order
to obtain statistical results capable of interpretation. Excel and the Statistical Package for
Social Sciences (SPSS) Version 16.0 were used to generate descriptive statistics
(frequencies, means, standard deviations, percentages, t-test values and skewness).
Appendix 3 presents information on source of data, method used for data analysis
and the expected results for this study. Socio-economic factors of the respondents playa
vital role in consumer choice of products in the market. It was therefore fundamental to
32
include them in the study to help establish whether there are any differences among the
consumer groups.
Bean consumers were asked to point out the varieties they consumed, they were
then asked to rate those varieties in order of preference, using a 1-7 (qualitative) scale;
one being the most preferred and seven the least preferred variety. This was to ensure the
rates given were as a result of consumer's experience with the beans. A chi square test
was used to test the hypothesis that there was no significant difference in consumer
preference of different common bean varieties. Respondents then evaluated eight
attributes in the seven bean varieties by assigning a rank to each attribute at 1-5 likert
scale; one being "very bad" and five "excellent" (qualitative). Description of variables is
presented in Table 3.3. After evaluating the attributes, consumers were asked open ended
questions on how much it cost them to purchase each of the varieties they consumed
(quantitative). They were further asked how much they were willing to pay per kilo of
each variety based on the evaluation they had done (quantitative). A chi square test was
used to test the hypothesis that consumer preference in attributes of different bean
varieties was not significantly different.
Descriptive statistics were computed on the sample data. The statistics were,
~ ~
sample size (n) and the proportions of participants in each response category (p I, P 2...
~
P k) where k represents the number of response categories. The test statistic for the
hypotheses was given by the formula;
33
~ "" (0 -E)2X=L
E (3)
Where O=observed frequency and E=expected frequency in each of the response
categories. The test compared the observed frequencies in a response category with the
frequencies to be expected if the null hypothesis were true. If the null hypothesis was
true, the observed and expected frequencies would be close in value and the x? statistic
would be close to zero. If the null hypothesis was false, the y} would be large.
Data on willingness to pay and attribute evaluation was used to fit a hedonic price
model. The model was used to estimate the relationship between willingness to pay and
the value consumers assigned to attributes in each variety. The model was tested for
goodness of fit using R-squared, analysis of residuals. Overall statistical significance was
checked with an F-test followed by t-test of individual parameters. The regression outputs
are summarized in Appendix 11 (1-7). The decision rule is that, "if the t-value of the regression
coefficient associated with an independent variable is greater than ,theabsolute critical t-value then
the independentvariable is significantat the given level of confidence".
3.5.1 Hedonic model specification
A multiple regression was done using a hedonic price model, introduced by Rosen
(1974), to check the significance of bean attributes in predicting the price consumers
were willing to pay for bean varieties. In the model the mean price of an ith variety would
be what consumers were willing to pay. It would be a function of the attributes in the
variety. The general form of hedonic pricing theory as specified by Rosen (1974) is:
Pi = Ui + L~iZi + £i (4)
Where: Pi = willingness to pay value. Ui= constatnt. ~i= implicit value of Zi
34
Z, = A vector of bean attributes. Ei= Random error.
The partial derivative of Pi with respect to Zr, a Pi / a Z, is referred to as the marginal
implicit price. It represents the amount consumers are willing to pay for a change in unit
of attributes.
This is taken as the value consumers place on a particular variety. This value comes about
by weighting the different attributes of the variety in relation to the utility they provide.
In the bean preference analysis, price consumers were willing to pay was regressed on
eight bean attributes namely; Color, Grain size, Price, Taste, Cooking time, Flatulence,
Cooking quality and Keeping quality. Likert scale 1 - 5 was used to rank preference for
attributes where five (excellent) was the highest rank and one (very bad) least rank
allocated to an attribute. Attributes and ranks used were as described in Table 3.3. The
linear model for each variety used in the study would take the form:
Pi = ~o + L~iZi + £i (5)
Where Pi= price(WTP) for bean variety i
~o = Constant L~i = implicit value of characteristic Z in variety i
Zi = quantity of the characteristic in variety i £i = Stochastic error term
Specification of the model into estimable form for this study was as follows for all
varieties:
Pi =fJo+fJlcoli+fJ2grzi+fJ3prci+fJ4tasti+fJ5ckdn+fJ6ckqlti+fJ7kpqlti+fJ~tui+ei (6)
Where;Pi =Price (Kshs) consumers were willing to pay for a kilo of common bean
variety i. fJo= Constant; this was the Pi intercept (value of Pi when Zi = 0). It gave
the value of the variety devoid of the attributes considered in the study. fJl-fJ8 = the
35
estimated coefficients of bean attributes. eol, grz, pre, tast, ektm, ekqlt, kpqIt, fltu = bean
attributes as defined in Table 3.3
e = Stochastic error term; the difference between observed value and predicted value of
dependent variable. Regression analysis calculates coefficients in a way that minimizes
sum of squared errors between actual values and predicted values of beans.
:Ee/=:E (S]-S2)2 (7)
S]= the stated willingness to pay price. S2 = estimated or predicted value of beans.
Coefficients were given by : ~2 = a~j / agrzj (8)pi
This means the percentage change in Pi brought about by a change in grz, holding all
other regressors constant. If ~2 = 4, then a 0.1 unit increase in grz, leads to a 0.4 percent
increase in Pi.
Table 3.3: Description of variables that were evaluated
Variable Variable Measurement ValueCode definition
1 wtp Willingness to Kenya shillings Quantitativepay
2 col Grain color Scale (5,4,3,2,1) Qualitative (Excellent-Very bad)
3 grz Grain Size Scale (5,4,3,2,1) Qualitative Excellent - Very bad
4 prc Price Scale (5,4,3,2,1) Qualitative (Excellent-Very bad)
5 tast Taste Scale (5,4,3,2,1) Qualitative (Excellent-Very bad)
6 cktm Cooking time Scale (5,4,3,2,1) Qualitative (Excellent -Verybad)
7 Ckqlt Cooking Quality Scale (5,4,3,2,1) Qualitative (Excellent-Very bad)
8 kpqlt Keeping quality Scale (5,4,3,2,1) Qualitative (Excellent-Very bad)after cooking
9 fltu Flatulence Scale (5,4,3,2,1) Qualitative (Excellent-Very bad)
36
NB: (5) Excellent (4) Good (3) Fair (2) Bad (1) Very bad
This study adopted linear functional form in regression analysis. To ensure
regression assumptions were satisfied, the data was tested for normality,
heteroscedasticity and collinearity.
One of the assumptions of linear regression analysis is that the residuals are
normally distributed. Estimation of coefficients requires that the errors are identically and
independently distributed this ensures the p-values for the t-tests are valid. As a result, all
variables were checked for distribution normality using histograms of the fitted model.
The histograms showed the results were confined within the normal distribution curve
and took the bell shape.
The regression data was tested for multicollinearity between the independent
variables, by running a tolerance and Variance Inflation Factor (VIF) assessment.
Tolerance = 1 - R2j, (9)
VIF= tOle!ance (10)
Where:
R2j is the coefficient of determination of a regression of explanatory j on all the other
explanatory variables. In other words tolerance is an indication of the percent of variance
in the predictor that cannot be accounted for by the other predictors, hence very small
values indicate that a predictor is redundant, and values that are less than .10 may merit
further investigation. A tolerance value less than 0.10 and VIF value of 10 and above
indicates a multicollinearity problem. According to Nzau (2003) cited in Kieti (2005), a
correlation coefficient more than or equal to 0.70 is an indication of a strong explanatory
interrelationship, which can lead to multicollinearity. Multicollinearity occurs when two
37
or more independent variables have the same effect on the dependent variable (perfect
correlation), leading to bias in ordinary least squares estimation (Wooldridge, 2002). In
the presence of multicollinearity, the estimate of one independent variable's impact on the
dependent variable Y while controlling for the other independent variables tends to be
less precise than if predictors were uncorrelated with one another. Multicollinearity
inflates the standard errors of coefficients making coefficients of some predictor variable
not significantly different from zero. This may lead to wrongfully accepting a null
hypothesis. In this study variance inflation factor (VIF) values for independent variables
in varieties were all less than 10 indicating absence of multicollinearity.
Another assumption of least squares regression is that the variance of residuals is
homogenous. This means that the distribution of residuals takes no pattern against the
fitted values. Heteroscedasticity which occurs if variance is not homogenous causes the
estimates of the variance to be biased resulting in biased standard errors. Using data with
heteroscedasticity can lead to biased inferences resulting from wrong hypothesis testing.
Data was therefore tested for heteroscedasticity using Breusch Pagan test and no
significant problem was noted. As appendix 10 indicates, the chi-square values were low
while the p-values were more than the 0.05 threshold indicating absence of
heteroscedasticity.
38
CHAPTER FOUR
4.0 RESULTS
4.1 Introduction
This chapter presents results of the study after data analysis. The results are
presented in four sections. The first section shows descriptive statistics results of socio-
economic characteristics of common bean consumers and study locations. The second
section presents descriptive statistics results of evaluation of consumer preference in the
different bean varieties as well as information on the most traded varieties in the markets.
Third section presents results of consumer preference in bean attributes respectively. The
fourth section presents hedonic regression results of the attributes influencing consumer
willingness to pay for the seven varieties. Data presented in this chapter was analyzed
using Microsoft Office Excel and SPSS v.16 statistical software.
4.2 Socioeconomic characteristics, consumption and residence of respondents.
Cross tabulation analysis was done to establish whether there was any relationship
between various variables under observation. Female respondents were 83.5%, while
male respondents were 16.5% of the total number of respondents, (Table 4.1). Difference
in bean consumption between these two groups was statistically significant (p-value =
0.000). There was more consumption of beans (more than thrice in a week) in the 31-51
age bracket (26%). This was followed by consumers with 51-60years, 20-31 years and
lastly consumer above 60 years of age at 20%, 15% and 11% respectively. As appendix
13 shows age groups 41 - 50, 51 - 60 had 20.1 % and 12.9% of bean consumers
respecti vely.
39
An average of 90% of the respondents were literate. The difference between
groups with college, secondary and primary education as indicated in Table 4.1 was
minimal at 21.8%, 36.9% and 31.6% respectively. Informal skills category had low
percentage of respondents at 7.8 percent. It consisted of respondents who had acquired on
job skills such as carpentry and masonry. There was statistically significant difference
(p-value=O.OOO) in bean consumption by respondents within the different education
levels. Respondents with secondary and primary education were the highest consumers of
beans at 36% each respectively. As appendix 14 shows, secondary education holders
were also least consumers with 44.8% consuming beans once a week.
Table 4.1: Socioeconomic characteristics of respondents (n=212)
Socioeconomic Frequency
Gender
Age
Education level
Main Occupation
Monthly Income
characteristics Percentage
male
female
20-30
31-40
41-50
51-60
above 60
primary
secondary
college
informal skills
none
Regular Employment
Trader
Housewife
Manual labor
Student
Unemployed
self-employment
0-5,000
5,000-10,000
10,001-15,000
15,001- 20,000
20,001- 30,000
above 30,001
16.5
83.5
22.5
26.8
20.1
13.4
17.2
31.6
36.9
21.8
7.8
1.9
18.6
5.7
26.2
3.3
1.0
3.8
41.4
42.7
16.6
12.6
10.6
5.5
12.1
35
177
47
56
42
28
36
65
76
45
16
4
39
12
55
7
2
8
87
85
33
25
21
11
24
As indicated in Table 4.1, most respondents earned Kshs. 0-5000 monthly.
40
Other respondents were distributed in the other five groups with between Kshs.
5001 to more than Kshs.30,000 monthly income. Results presented in Figure 4.5 show
respondents in the different income groups and their weekly consumption patterns.
Respondents in the Kshs. 5,000 and below, monthly income group were the major
consumers of beans at 42.9% of all respondents. Out of the total consumers in this group
approximately 47% consumed bean meals twice a week, 36.5% consumed bean meals
thrice a week, 18.8% consumed bean meal more than three times in a week, while only
8.2% consumed beans once a week. This group was followed byKshs. 5,100 - 10,000
income group which had 16.7% of all respondents. The above Kshs. 30,001 and Kshs.
10.000 - 15,000 groups had 12.1% and 12.6% respondents respectively. Percentages of
respondents consuming beans different times in a week were lowest in the Kshs. 20,001 -
30,000 income group at 5.6% followed by 15,001- 20,000 at an average of 10 percent.
-Thrice
Comparison of respondents monthly income and their consumption frequency
60
50
Percentage of 4030bean consumers 20
10o
-Once
-Twice
Respondents Monthly Income
Fig 4.1: Respondents weekly bean consumption and monthly' incomes
There was a statistically significant relationship between the respondents'
occupation and their monthly income (p-value = 0.000) at 5% significance level. As
indicated in Figure 4.2 majority of self-employedrespondents, most of whohad monthly
41
income of Kshs. 5,000 and below, were the highest consumers of beans.Most
housewives,who comprised26.2% ofrespondents (Table 4.1) and had a monthly income
of Kshs. 5,000 and below, consumed beans regularly(Figure 4.2). Consumers in regular
employment,who formed 18.6% of the respondents,comprisedof employees in formal and
informal sector. Majority in this group earned more than Kshs. 30,000, with their bean
consumption frequency falling way below that of housewives and self-employed
consumers. Their average beanconsumption was; 20.7o/00nce, 16.4%twice, 20.3%thrice,
and 18.2% more than thrice in a week.Students were the least bean consumers at3.4%
once per week andl.8% more than thrice in a week.
Comparison of consumers occupation and their bean consumption frequency
50
Percentage of bean ~~
consumers 20
10o
~~ 0(," .~ c-&R «...<~ ,;9~~ ~AO ~o N-:r;.
~v _"'-~':# "">"
~(,~
-Once
-Twice
-Thrice
- More than thrice
Respondents main occupation
Fig 4.2: Respondents' weekly bean consumption and occupation
Consumers from the urban area formed 55% of the total consumers in the study
while 45% were consumers from the rural areas. The difference in consumption
frequency between the two groups was statistically significant (p-value = 0.025) at 5%
42
significance level. As presented in Table 4.2 the urban respondents who consumed beans
once a week were 80% compared to 20% of rural consumers.
Table 4.2: weekly bean consumption based on region
Dwelling place Weekly Bean Consumption
once Twice Thrice More than Thrice
Urban 80 54.1 50.8 47.3
Rural 20 45.9 49.2 52.7
P-value 0.025
Approximately 53% of rural consumers consumed bean meals more than three times in a
week compared to approximately 47% urban consumers.
4.3 Evaluation of consumer preference for common bean varieties
Results indicated that 95% of interviewed consumers attached importance to
knowledge of bean varieties, while approximately 96 % of interviewed traders indicated
that consumers asked for particular bean varieties when purchasing. This was an
indication of choices being made when purchasing or cooking beans. Results indicated
that approximately 78% of total number of respondents consumed GLP 585 Wairimu,
59% consumed KAT X 56 Gituru, 50% GLP 2 Rosecoco (Nyayo) while 33% consumed
GLP 92 Mwitemania. Varieties released by KARl Katumani KAT Bland KAT B9 were
each consumed by 16% of the respondents. GLP 24 Canadian Wonder was the least
consumed (8%). Traders in the study area mainly sold the GLP varieties, originally
released from Kari Thika (Appendix 2). However as Figure 4.3 indicates KAT X 56
Gituru released from Kari Katumani was gaining popularity in the market with nearly
half of the 67 interviewed traders selling it.
43
Comparison of traders selling and not selling various bean
varieties
KAT B1 (Gathika) 4t9.?
GLP 94 (Mwitemania) 64.2'"~.•- GLP 585 Red Haricot (Wairimu) 97~.•'" 91eo: GLP 24 Canadian Wonder (Gituru)~ -don't sell=eo: GLP 2 Rosecoco (Nyayo) 71.6 -sell~~ KAT B9 (Gacurna) 77.6
KAT 56 (Gituru) 64.2
a 20 40 60 80 100
Percentage of bean traders
Fig 4.3: Bean varietiessold in the market
Out of the varieties presented to consumers in the study, three (3) GLPs varieties
were in the top four positions with only one KAT variety. GLP 585 Red Haricot was
ranked as the most preferred bean variety by both consumers and traders. Table 4.3shows
chi square test results for variety ranking on a scale of 1-7 from the most preferred to the
least preferred. The results were statistically significant save for KAT B1.
Table 4.3: Consumer preference scores for common bean varieties (1-7)
Rank N P=value
Variety
KAT X56Gituru
KAT B9 Gacuma
GLP 2 Rosecoco
GLP 24 Canadian Wonder
GLP 585 Red Haricot
GLP 92 Mwitemania
KAT B1 Kathika
Preference
(%) chi
39.8
6.1
43
56.3
64.7
38.8
18
76.915
13.727
91.924
18.500
222.205
29.806
6.818
0.000
0.017
0.000
0.002
0.000
0.000
0.235
3
6
2
5
1
4
7
47
6
34
5
101
26
6
44
The test results imply that consumer preferences were diverse in the different bean
varieties assessed. This gave ground to reject the null hypothesis that there was no
significant difference in consumer preference for different common bean varieties.
4.3.1: Consumer preference in common bean varieties based on dwelling place.
There was a relationship between consumers' residency and ranking of varieties
(Appendix 12). Rural consumers ranked KAT X56 variety highly while urban ranked it
fair this was significant (p-value =0.000) at 5% significance level. GLP 2 Rosecoco
ranking was statistically significant (p-value =0.000) at 5% significance where urban
respondents ranked it between 1 and 3 with each rank getting 75% and above of the
respondents. The GLP 24 Canadian Wonder variety was not popular with respondents
especially in the rural areas. However ranking by both urban and rural consumers was
statistically significant (p-value=0.093) at 10% significance level. GLP 585 variety was
given a preference score of 1 by 75.2% of urban consumers and 24.8% rural consumers.
It was ranked 2nd by 37.8% urban respondents and 62.2% rural respondents. The variety
was ranked 3rd by 20% urban consumers and 80% rural respondents. The difference in
rankings by the two groups was highly significant (p-value= 0.000) at 1% significance
level. GLP 92 Mwitemania was given a preference score of 2 and 4 by the rural
consumers while urban consumers scored it 5th and 6th, a clear indication of low
preference for it. These scores were statistically significant (p- value = 0.021) at 5%
significance level. A very high percentage (63.6%) of rural consumers scored KAT B1
Kathika 1, 2 and 4 with the scores being statistically significant (p-value = 0.045) at 5%
significance level.
45
4.3.2 Respondents age and preference in common bean varieties.
Results from the study (Table 4.4) indicate that the relationship between the
choice of bean varieties and consumers' age was statistically significant at 5%
significance level. Consumers aged between 20-30 years preferred GLP 585 Red Haricot.
The group between 31-40 years old preferred GLP 2 Rosecoco. Age group between 41-
50 years liked GLP 24 Canadian Wonder while groups in the age bracket of 51 years and
above preferred KAT B9 Gacuma and X56 Gituru.
Table 4.4: Different age groups' Preference in bean varieties
Age KATX56 KATB9 GLP2 GLP24 GLP585 GLP92 KAT
Group (%) Bl
20-30 18.7 3 24.4 23.5 26.5 12.9 9.1
31-40 18.7 18.2 33.7 23.5 27.2 21.4 21.2
41-50 17.1 24.2 23.3 41.2 21 22.9 24.2
51-60 17.9 24.2 7 0 11.7 18.6 21.2
Over60 27.6 30.3 11.6 11.8 13.6 24.3 24.2
Total' 100 100 100 100 100 100 100
x2 value 33.907 15.349 10.739 6.974 12.642 11.021 6.826
P value 0.000 0.004 0.030 0.137 0.013 0.026 0.145
4.4 Evaluation of consumer preference in attributes of common beans.
Results presented in this section are divided into two. Section 4.3.1 comprises of
pairwise comparison results of bean attributes, where eight attributes were compared
against each other to establish the preferred attribute in a pair. Section 4.3.2 comprises
variety attribute evaluation results to establish the varieties that had the preferred
attributes.
4.4.1 Consumer preference in bean attributes after pairwise comparison.
The summarized results of pair wise' comparison of bean attributes presented in
Table 4.5 indicate that cooking quality was the most preferred attribute at 21% of the
46
total rankings (comparison results presented in Appendix 6). Keeping quality was second
in preference with 20% of the total scores. Price was ranked third with an approximate
score of 12% of the total scores. Taste had 10% score while grain mix ranked better than
cooking time, grain size, color and flatulence at 9.97% scores. Cooking time which was
at 9.02% came in 6th in preference compared to the other attributes. Duration of bean
cooking was 2 to 3 hours for all varieties, however only GLP 2 Rosecoco and KAT B1
Kathika cooking hours were statistically significant (p-value 0.040 and 0.012)
respectively. Color had 8.35% while grain size was preferred to flatulence at 7.04% of
the total scores. Flatulence was the least preferred attribute at 1.85% of the total scores.
Table 4.5: Results of pairwise comparison of common bean attributes.
Variety Preference frequency Percentage
Cooking Quality
Keeping Quality
Price
Taste
Grain Mix
Cooking Time
Color
Grain Size
Flatulence
1174
1138
672
559
554
501
464
391
7
21
20
12
10
10
9
8
7
2
4.4.2 Attribute ranking according to variety
Each variety was evaluated to establish the varieties that had the attributes
consumers preferred as had been established during the pairwise comparison. As
Appendix 7 shows, respondents ranked GLP 585 Red Haricot good and excellent in most
of the attributes making it the most preferred variety in relation to its attributes. Age of
consumers was the only socioeconomic characteristic that was significant in evaluation of
consumer preference for attributes in the different varieties.
47
4.4.2.1 Cooking quality
GLP 585 Red Haricot was preferred for cooking quality attribute followed by
KAT B1 at mean rank 5 (p-value = 0.000) and 4 (p-value=0.003) respectively. KAT B9
Gacuma was 3rd with a mean rank of 4.19 (p-value 0.010) which was statistically
significant. As indicated in Figure 4.4 half of consumers in 41-50 age group ranked the
variety (3) fair for the attribute with the rest in the age group ranking it highly (good and
excellent).The results were statistically significant (p-value=O.013) at 5% level of
significance.
Evaluation of cooking quality attribute of KAT B9 by consumers in different
age groups
f 60~ Above 60 u 1 D. I<II~ 51-60 2.......•30'-' . 5.•..= ''''' ,.,,.., 0<II 41-50 -excellent"C 50= rv0 31-40 33.3 -goodQ. g'"2: 20-30 -fair••.. 25
0
<IICfI 0 10 20 30 40 50 60~
% of respondents
Fig 4.4: KAT B9cooking quality ranking by different age groups.
4.4.2.2 Keeping quality
GLP 585 had the highest mean rank of approximately 5 (p=0.000) for the keeping
quality attribute. It was followed by KAT B1 Kathika with 4 mean rank that was
statistically significant (p-value=0.009)at 5% significance level. KAT X56was third with
a mean rank of 4 (good). Evaluation of GLP 2 Rosecoco which was 4th in keeping quality
attribute by consumers in different age groups was statistically significant as Table 4.6
48
Table 4.6 indicates. Respondents in the age groups 20 - 40 ranked the variety either 4
"Good" or 5 "Excellent" while the rest ranked it either 2 "bad" or 3 "fair".
Table 4.6: GLP 2 keeping quality ranking by different age groups
Age Bad Fair good Excellent
(%)
20-30 0 0 34.7 16
31-40 0 25 34.7 44
41-50 16.7 75 18.4 24
51-60 16.7 0 2 12
Above 60 66.7 0 10.2 4
P value 0.001
4.4.2.3 Price
Price of GLP 585 Red Haricot was ranked 4 "good" (p=0.000), while all the other
varieties had a mean rank of three. Mean prices respondents were willing to pay per kilo
deviated by between Kshs. 16 and Kshs. 30 per kilo towards the lower side of mean
market prices for the different varieties. As indicated in Figure 4.5, KAT B1 Kayellow
was the highest priced variety at a mean market price of Kshs. 96 per kilo and willingness
to pay price of Kshs. 66 per kilo. The variance of market and willingness to pay price for
KAT B1 was Kshs. 30 per kilo making it the highest discounted variety by consumers.
GLP 92 Mwitemania had a mean market price of Kshs. 66 per kilo making it the lowest
priced variety. Consumers' stated price of Kshs. 50 per kilo for the variety was Kshs. 16
lower than the mean market price making the variety the least discounted variety. The
price consumers were willing to pay deviated from the market prices by between Kshs.
21 and Kshs. 23 per kilo for all the other bean varieties.
49
Market and willingness to pay prices for a kilo of beans
Canadian wonder
• Price the consumer is willing to pay
per kg
• Price the consumer pays per kg
W8lflJIlU
kayellow 1IWil __ 96.15
Mwitemania
Rosecoco •••••••• ~.81.39
Gacuma ••••• I.iIi& 80.96
60 80 100 120o 20 40
price per kilo (ksh)
Fig 4.5: Market prices and willingness to pay price for a kilo of beans
4.4.2.4 Taste
GLP 2 Rosecoco, 585 Red Haricot and KAT B1 Kathika had a mean rank of 5 for
taste that was statistically significant (p-value=O.OOO)for the 1st two and (p-value=O.Oll)
at 1% and 5% significance level respectively. KAT X56wasfourthwith a mean rank of 4
(p-value=O.OOO)significant at 5% level. More than 50% of respondents in 51 years and
above age groups ranked the variety excellent. Taste preference for this variety by
different age groups was opposite of GLP 585 where more than 50% consumers in the
20-50 and only 9% in the 61 and above years of age ranked it excellent. There was
distinct pattern in preference for taste of KAT X56by the different age groups.
Respondents in 20-40 years age bracket had a fair preference for the variety's taste while
respondents in the 41 and above age bracket had higher preference for the same variety.
As Table 4.7 shows, the ranking was statistically significant (P-Value=0.008) at 1%
Age Fair Good Excellent
50
level. GLP 92 Mwitemania and GLP 24 Canadian Wonder had high percentage of one (1)
to three (3) ranks (very bad - fair) making them least preferred for taste attribute.
Table 4.7: Ranking of KAT X 56 Gituru's taste by different age groups
20-30
31-40
41-50
51-60
Above 60
P value 0.008
41.7
33.4
16.7
8.3
o
20.7
24.1
20.7
10.3
24.1
11.5
11.5
11.5
28.8
36.5
4.4.2.5 Cooking time
According to respondents, cooking time took between one and three hours. GLP
585 Red Haricot had a mean rank of 4 (p=0.000) in cooking time. It was followed by
KAT B1 Kathika whose ranks for cooking time were statistically significant (p-
value=0.091) at 10% significance level. Approximately 27% of consumers said it took
them one hour to cook the variety, 24% said it took 2 hours while 16.9% said it took them
three hours to cook the variety. Results of attribute evaluation in GLP 2 Rosecoco
indicate that 21% of the consumers ranked it 4 "Good" for the attribute which was
statistically significant (p-value= 0.040) at 5% level. According to 51% consumers, the
variety took 2 hours to cook, 3 hours according to 24% consumers and 1 hour as reported
by 14% of the consumers. GLP 92 Mwitemania had a mean rank of3 (fair) making it the
least popular variety in relation to cooking time.
4.4.2.6 Color
Out of the total excellent rank scores for color attribute, GLP 585 Red Haricot got
41% making it the highest ranked variety (p=0.000) for color. It was ranked excellent by
51
26 % of consumers in the 20-30 age group, 30% by 31-40 age group, 20% by 41-50 age
group, 10% by 51-60 age group and 13.8% by the group with 60 years and above. The
results indicate that young consumers preferred GLP 585's color. GLP 2 Rosecoco was
second in excellent score (p=0.000) for its color attribute. KAT X56 Gituru was third
with a mean rank of 4 "Good" (p=0.000). GLP 92 Mwitemania had a mean rank of 3
making its color the least preferred. The variety had the highest percentages in "Very
Bad", "Bad" and "Fair" rankings for color compared to the other varieties. As indicated
in Table 4.8, the difference in color evaluation by consumers in the different age groups
was statistically significant at 5% significance level.
Table 4.8: Ranking of GLP 92 Mwitemania color by different age groups.
Age Very bad Bad Fair Good Excellence
20-30 0 0 8.7 28.6 7.1
31-40 66.7 0 8.7 19 42.9
41-50 0 33.3 21.7 23.8 21.4
51-60 33.3 44.4 26.1 14.3 0
Above 60 0 22.2 34.8 14.3 28.6
P value 0.043
4.4.2.7 Grain size
GLP 2 Rosecoco had the highest preference rank for grain size attribute which
was statistically significant (p-value=O.OOO) at 5% level. Approximately 24% of
respondents who ranked it excellent were mainly in the 20-50 age groups. KAT X56 was
second in ranking for grain size by 23% respondents who were mainly in the 51 years and
above age group. The least preferred variety for grain size was GLP 92 Mwitemania at a
rank of 4 which was statistically significant (p-value 0.000) at 5% level.
52
4.4.2.8 Flatulence
KAT B1 was highly preferred in relation to flatulence attribute with a mean rank
of five (p-value=O.OOO).KAT B9 and GLP 585 Red Haricot were second and third with
rank 4 (p-value=0.064) and 4 (p-value=O.OOO)preference ranking respectively. There was
a statistically significant difference in ranking of KAT X56 for flatulence by the
respondents in the different age groups. Respondents who were between 20-50 years of
age ranked the variety low for flatulence (Table 4.9), while respondents in the 51 years
and above age bracket ranked it high for flatulence.
Table 4.9: KAT X56 flatulence ranking by different age groups
Age bad Fair Good Excellent
20-30
31-40
41-50
51-60
Above 60
P value 0.046
36.4
9.1
36.4
o
18.2
22
36.6
12.2
14.6
14.6
15.4
10.3
12.8
28.2
33.3
13.8
10.3
20.7
17.2
37.9
Results of attribute evaluation in bean varieties as presented in appendix 7
indicate that GLP 92 Mwitemania was least preferred for 5 attributes; cooking quality,
keeping quality, cooking time, color and grain size. The variety had the least ranks for the
mentioned attributes compared to the other varieties. GLP 24 was second from last with
5 attributes which included cooking quality, keeping quality, price, cooking time and
grain size. Given the statistical significance in attribute ranks of all the varieties, it was
clear that consumer preference in attributes of different bean varieties was different. It
was therefore safe to reject the null hypothesis that consumer preference in attributes of
different bean varieties is not significantly different.
53
4.5 Evaluation of consumer willingness to pay for ranked attributes.
All the eight attributes in seven bean varieties were compared to the willingness
to pay price using a Hedonic price model. The important statistical parameters
considered in the regression analysis were:
Coefficient of determination (R2).
This measured the percentage of variation in the willingness to pay price explained by
all the attributes included in the model. It measured the overall strength of association in
the model.
F-statistics.
Statistical measure to establish whether the overall model results were statistically
significant. It tested the null hypothesis whether coefficients were equal to zero.
B coefficients.
These explained how much the willingness to pay price changed for a unit change
of the predictor variable.
Beta coefficients
These explained percentage changes in the model. They measured the percentage
in the willingness to pay price associated with a percentage change of anyone of the
attributes holding all the other attributes constant.
t-statistics.
These measured whether a given coefficient was significantly different from zero.
P>ltl showed the 2-tailed p-values used in testing the null hypothesis that the coefficient
(parameter) was 0, using 0.05 alpha level.
54
Hedonic price function model results presented in Table 4.10 and appendix 11
indicate that consumer preference for bean attributes was found to effect willingness to
pay in some varieties. The coefficients of determination (R2) for the different varieties
were as follows; KAT B9 Gacuma 42%, KAT B1 Kathika 34%, GLP 24 Canadian
Wonder 30%, KAT X 56 Gituru 25%, GLP 585 Wairimu 17%, GLP 92 Mwitemania
15% and lastly GLP 2 Rosecoco 13 percent.
4.5.1 KAT X 56 Gituru
Regression results for the variety presented in Table 4.10 indicate that 25% of
variation in the willingness to pay was explained by combined influence of all the
variables in the regression. The model was statistically significant (p-value=O.OOO)at 1%
level.
Table 4.10: Hedonic model results for bean attributes
Varieties Color Grain Size Price Taste Cooking Time Cooking Quality Flatulence Keeping Quality F test R2
KAT X coer -0.186 0.371 0.113 0.113 -0.008 -0.008 -0.277 -0.100 4.336 0.245
56 t. value -1.321 2.584 l.l10 0.748 -0.069. -0.067 -2.935 -1.014
Gituru P>t 0.189 0.011 0.270 0.456 0.945 0.947 0.004 0.313
KATB9 coer -0.691 0.745 -0.075 0.568 0.210 -0.527 0.223 -0.033 1.981 0.419
Gacuma t. value -2.226 1.814 -0.285 1.675 0.641 -1.519 0.915 -0.126
P>t 0.037 0.083 0.778 0.108 0.528 0.143 0.370 0.901
GLP2 coer 0.772 -2.490 1.478 -1.371 -1.289 0.174 2.209 -0.068 1.282 0.126
Rosecoco t. value -0.309 -0.920 0.697 -2.823 -0.471 0.063 1.201 -0.035
P>t 0.758 0.361 0.488 0.006 0.639 0.950 0.234 0.972
GLP24 coer -0.325 0.041 -0.181 -0.442 -0.118 1.281 -0.786 -2.935 3.907 0.302
Canadian t. value -0.621 0.058 -0.387 -1.415 -0.104 0.801 -0.922 -1.613
Wonder P>t 0.562 0.956 0.715 0.216 0.922 0.460 0.399 0.168
GLP 585 coer 0.048 -0.216 0.173 0.325 0.193 0.053 0.001 0.151 3.790 0.173
Wairimu t. value 0.375 -2.926 2.228 2.909 1.864 0.538 0.013 1.378
P>t 0.708 0.004 0.027 0.004 0.064 0.591 0.990 0.170
GLP92 coer 0.062 -0.482 0.038 -0.060 0.228 0.341 -0.309 -0.118 1.164 0.145
t. value 0.287 -1.651 0.162 -0.249 1.198 1.288 -1.481 -0.770
P>t 0.775 0.104 0.872 0.804 0.236 0.203 0.144 0.445
KATBI coer 0.100 -0.487 0.549 -0.815 0.034 0.211 -0.032 -0.522 1.827 0.335
Kathika t. value 0.207 -0.904 1.320 -1.175 0.057 0.549 -0.075 -0.950
P>t 0.839 0.377 0.202 0.255 0.955 0.589 0.941 0.354
56
KAT X 56 Gituru grain size variable had positive linear relationship to willingness to pay
while flatulence had a negative relationship to willingness to pay. Grain size and
flatulence were statistically significant at 5% and 1% significance levels respectively.
4.5.2 KATB9
The overall variance explained by all the 8 attributes was 42% (p=0.098) which
was significant at 10% significance level. Coefficients for color -0.691 (p=0.037) and
grain size -0.745 (0.083) attributes were significant at 5% and 10% significance levels
respecti vel y.
4.5.3 GLP 2 Rosecoco (Nyayo)
Predictor variables explained 13% of overall variance in the willingness to pay
pnce of the variety. Taste variable had a negative coefficient -1.371 which was
significant at 1% level.
4.5.4 GLP 24 Canadian Wonder
The R2 value for Canadian Wonder 0.302 was significant at 10% significance
level. None of the attribute coefficients were significant.
4.5.5 GLP 585 Red Haricot.
The variety had an R2 value of 0.173 which meant that the regression model
explained 17% of the overall variation in prices consumers were willing to pay. Price,
cooking time and taste positively influenced variation in the willingness to pay while
grain size had negative influence. As Table 4.10 shows, grain size and taste were
statistically significant at 1% significance level while price and cooking time were
statistically significant at 5% and 10% significance levels respectively. Results of both
GLP 92 and KAT B1 regression models were not statistically significant.
57
CHAPTER FIVE
5.0 DISCUSSION
5.1 Introduction
This chapter comprises interpretation of results and their implications. It is
divided into four sections. The first section discusses the socio economic results. The
second section is interpretation of results on consumer preference of bean varieties. The
third section is discussion of consumer preference of attributes in common beans. The
final section is a discussion of hedonic models results for the seven bean varieties.
Previous research findings are explored in order to possibly explain reasons for particular
outcomes in this study. Existing gaps are pointed out in the chapter for action to be taken.
5.2 Respondents' socio economic characteristics
Based on information of socio economic characteristics of respondents, there was
diversity in their education and financial status. Rural respondents (53%) consumed
beans more frequently than urban respondents (47%) this may be attributed to low cost of
making bean meals in rural settings compared to urban setting. This is supported by
results that showed that 86.3% rural respondents used firewood while only 13.7% urban
respondents used firewood to cook beans. The latter mainly used charcoal, (93.5%) to
cook beans, which is an expensive source of fuel at an average of Kshs. 150 in a week.
The other factor could have been availability of beans since they are grown in the rural
farms. Results are in line with research findings of Leterme & Carmenza (2002), where
rural population eats more pulses than the urban population, due to geographical
constraints that limit exchanges and favor consumption of locally produced foods. On the
58
other hand urbanization means people are out of home all day, thus have less time to
cook. They buy alternatives such as meat and processed food. This finding indicates that
beans may be exempted from a meal due to accessibility, preparation cost and time.
Majority of bean consumers in the study were female (82.9%) while only 16.6%
were male. This can be attributed to household roles of women in making major
decisions regarding food and are therefore conversant with attributes of the different food
products. Results are in line with other studies on role of women in food decisions. In
Mundua (2010), 55% of women were charged with household chores in relation to food
decisions. According to Soniia (1999) women were involved in bean handling right
from production to consumption. This is further strengthened by inability of men to
differentiate bean varieties and describe their attributes during this study. Women should
be involved in product evaluation and improvement, especially food related products.
Low income earners consumed beans more in a week with the majority (47%)
consuming beans twice a week. Consumption within the Kshs. 30,000 monthly income
earners was low at 12.1% of bean consumers. This could possibly be explained by the
affordability of other sources of protein such as animal protein which is more expensive
compared to beans for consumers in the middle and top economic stratum consumers in
the middle and top economic stratum (Leterme & Carmenza, 2002). There is also the
notion that beans are poor man's meat. These results are not isolated, in a study by
Broughton et al., (2003) the authors reported that bean consumption in the lower strata of
Latin American was 20% higher than in middle and upper strata. The notion that beans
only play an important role in the diet of the underprivileged can be reversed by actively
sensitizing consumers on importance of beans. Beans are a cheap source of protein
59
compared to meat (Perla et al., 2003) and offer a cheap source of food during off seasons
where previous harvest can be stored for long without getting spoilt unlike animal
products fruits and vegetables (Sathe et al., 2003). The health benefits of consuming
beans include control of diabetes, obesity and prevention of coronary heart diseases as
reported by Leterme & Carmenza (2002). Enhancement of knowledge benefits of beans
should be done through information dissemination which should be targeted mainly to
women to include it in their meal plans. Value addition of beans would produce products
that can be readily utilized to reduce preparation time.
The most frequent bean consumers were in the 31-51 years age bracket. This can
possibly be attributed to the group having disposable income enabling it make purchase
decisions as was outlined in Mundua (2010). This implies that preferred attributes should
be enhanced with the 31-51 age bracket consumers in mind since they are a sizeable
market outlet by being most frequent beans consumers more than thrice in a week at 26
percent.
5.3 Consumer preference in common bean varieties
The results obtained from traders on consumer behavior during bean purchase
showed that respondents were well aware of common bean varieties and they made
choices among the different varieties on offer in the market. This showed there were
differences in the varieties which made consumers choose one variety and not the other in
the market.
Grain Legume Project (GLP) bean varieties continued to dominate consumer
preference. There were three GLP varieties in the top four preferred varieties going by
the results of both consumers and grain traders. GLP 585 Red Haricot rank 1 (65%), GLP
60
2 Rosecoco rank 2 (43%) and GLP 92 Mwitemania rank 4 (39%) were the preferred GLP
varieties while GLP 24 Canadian Wonder popularity (rank 5) had declined. Data from
traders further strengthened this observation where the leading variety in relation to
traders selling the variety was GLP 585, followed by GLP 2 Rosecoco, KAT X56 and
GLP 92 being sold by the same percentage of traders. The basic principle in any business
is to offer a product that will appeal to consumers. The bean varieties sold by most
traders was an indication of their preference by consumers. Results were in line with a
study conducted by Gichangi et aI., (2011) where authors reported that GLP 585 was
preferred with a mean percentage of 91.5% in Central Rift Valley. According to Katungi
et aI., (2011), GLP 2, GLP 92, GLP 24 and KAT B1 had household share of 71.5 %, 87
%, 12.2 %, and 34.6 % respectively. The two studies were similar to this study in that
evaluation was for both KAT and GLP varieties, and was done by consumers. Other
studies indicated that GLP 2 also called Calima accounted for 20% of production due to
its high market preference (Wortmann et aI., 1998; Kibiego et aI., 2003)
Among the newly released varieties, KAT X 56 which is a Canadian Wonder
(rank 3) was preferred more. The variety was popular with both urban and rural
consumers and was clearly penetrating markets where older varieties were dominant.
Similar findings were recorded in Katungi et al. (2011); Gichangi et aI., (2011), in
Eastern, Rift Valley and Western Kenya where KAT X56 was more preferred among the
newly released KARl bean varieties. Farmers in Central Rift Valley consumed and sold
more of KAT X56 than other KATs while in Western Kenya, 35.8% of respondents in
the study, preferred KAT X56 to the other KAT varieties. Results of this study and
61
previous studies show consistency in consumer preference for KAT X56 in different parts
of Kenya.
There was a shift from GLP 585 to KAT X56 as the mainly preferred bean variety
in rural areas where KATs had been introduced through CIAT seed loan program. The
latter was ranked number one by 93% of rural respondents against 7% urban consumers.
Overall GLP 585 was ranked number one by 24.8% rural consumers against 75.2% urban
consumers. Despite presence of KATs in the rural areas, GLP 92 Mwitemania and GLP
2 Rosecoco still had a fair share of preference among the rural consumers. GLP 92 had a
preference 1 rank by 76.1% of the rural respondents against 23.9% of the urban
consumers. Preference for GLP 92 could be attributed to low market prices and high
production traits rather than culinary characteristics as was the case in Zambian
Copperbelt, and Eastern Kenya (Wortmann et al., 1998; Katungi et al., 2010).
The number of rural respondents who ranked GLP 2, KAT B9 and KAT B1 was
nearly the same in this study. Results for GLP 2 and KAT B1 were statistically
significant at 5% level while results for KAT B9 Gacuma were significant at 10% level.
GLP 24 Canadian wonder (p-value=0.093) was not popular among the rural consumers
going by the scores it was allocated. On the other hand GLP 2 and GLP 92 were popular
among the rural respondents despite presence of other new varieties in the market
(Wortmann et al., 1998; Kibiego et al., 2003; Katungi 2011). This was an indication that
GLPs were still relevant even with newer varieties entering the market.
It was noted that evalution of age-variety relationship was not common in
previous common bean consumer preference studies. Results of this study showed a trend
in consumer preference for the two types of varieties, GLP and KAT. The younger
62
respondents (20-40 age bracket) preferred GLPs while the older respondents (41 to >60
age bracket) preferred the KAT varieties. These results were statistically significant for
the varieties under study at 5% significant level apart from GLP 24 and KAT B1.
Low consumption of KATs may be attributed to lack of information about them
by most of the respondents. As indicated in appendix 9, half of consumers got
information about beans from fellow consumers and nutritionist in health and agriculture
sectors. Consumers who get information from fellow consumers are likely to use same
varieties as their informants, leaving out other varieties unfamiliar to their informants,
information channel is fundemental in preference of a product, (ASARECA, 2003;
Katungi et aI., 2010; Karanja, et al.,2012). However this is an area that requires further
investigation.
5.4 Consumer preference in attributes of common bean varieties.
Cooking quality was the most preferred attribute in comparison to all the other
attributes. This implied that consumers chose grains that were likely to remain whole
after long duration of cooking. Information from consumers indicated that different
varieties exhibit different textural characteristics after cooking; some are hard to cook
meaning they take long to cook while others crack or mash up easily. This characteristic
influences the final outlook of a bean meal and is therefore important to consumer when
making bean choice decisions. In the study area, beans are mainly used to make githeri,
which is more appealing and tasty if the grains remain whole after cooking. Based on the
type of dish to be prepared, cooking quality was therefore an important consideration
when choosing beans. These results are in line with results of Anang et al. (2011) and
Danbaba et al. (2011). In the first study, the authors observed that cooking quality was
63
one of the most preferred quality characteristics of rice in Tamale. According to the
second study people of Ogun Nigeria chose Ofada rice due to its cooking quality
attribute. In this study, GLP 585 and KAT B1 bean varieties were highly preferred for
this attribute followed by KAT B9 in third position. It was noted that GLP 585 was also
ranked as the most preferred variety an indication that cooking quality attribute could
have played a crucial role in preference of the variety.
The next important attribute was keeping quality. Putting into consideration the
high cost, man hours and cooking time required to cook beans, consumers chose varieties
that would last at least two days after cooking without getting spoilt and retain flavor
(Wortmann et aI., 1998). A popular bean meal githeri a mixture of beans and maize is
expected to last for an average of two days because making githeri involves cooking for
an average of two hours as 43% of respondents indicated. Duration of bean cooking
impacts on fuel cost which stands at Kshs. 199 in a week on average. GLP 585 and KAT
B1 were highly preferred by consumers for keeping quality. The continued popularity of
GLP 2 Rosecoco among consumers as also indicated in Kibiego et aI., (2003) and
Katungi et aI., (2011) may be attributed to high preference of the variety's keeping
quality attribute by consumers in different age groups. GLP 92 was ranked 'bad' for
keeping quality. These results agree with research findings by Katungi et al., (2011) in
Eastern Kenya where keeping quality attribute in beans was preferred across households.
Results on keeping and cooking quality attributes affirm Wortmann et aI., (1998) finding
that culinary qualities, just as yields, are considered important by women while doing
seed selection. However culinary qualities seem to have been given less prominence in
previous bean preference studies.
64
GLP 585 emerged as the preferred variety in relation to price. KAT Bl was the
highest priced variety in the market while GLP 92 was the lowest priced in the market.
Price of KAT B 1 has been influenced by the attributes of this variety as explained by
most respondents; low flatulence, sweet, good cooking and keeping quality. Despite its
good attributes the difference between mean price of what consumers were willing to pay
and the mean market price was the highest at Kshs. 30 compared to price differences of
the other varieties. The market price could have contributed to the variety getting the
least score in the variety preference ranking. On the other hand the low price of GLP 92
could have contributed to the variety being ranked as one of the preferred varieties. The
finding implies that price plays a fundamental role in acceptance of bean varieties in the
market. Results further showed that market prices of all varieties were higher than what
consumers were willing to pay. This led to mixed grains being an important alternative
for consumers to the pure stand of a variety. These findings are consistent with other
studies which showed that price was an important constraint that limits consumers '
access to preferred varieties because they could not afford (Wortmann et al., 1998;
Broughton et al., 2003; Katungi et al., 2010).
According to consumers grain mixes consist of different varieties which have
different grain sizes, taste and cooking times they also contain damaged grains and dirt.
This means they have to be selected to remove impurities before cooking, which takes
more preparation time than normal. The different varieties have different attributes which
bring inconsistency in structure of the final dish; some beans may be uncooked while
others overcooked. Prices of mixed grains in the market were low due to the negative
attributes associated with the lot. High ranking of mixed grains could therefore have been
65
occasioned by the high cost of pure stand lots. However uniform or pure stand grains
were a preference of all consumers going by results of variety preference rankings. These
findings are consistent with Wortmann et al. (1998) who observed that there was an
increase in preference by urban consumers for uniform samples of beans due to extension
sensitization especially in Kenya, resulting in increased production of uniform samples
and decline in marketed mixed grains and impurities.
Three varieties, GLP 2 Rosecoco, GLP 585 Red Haricot and KAT B1 Kathika
were preferred by respondents for taste attribute. The first two varieties were also ranked
as the most preferred varieties. This implies that taste contributes to preference of a
variety as well as demand in the market. Results of taste preference are in line with
studies by Katungi et al. (2010), where both GLP 2 and KAT B1 were preferred for their
taste. Results also indicated the older respondents liked taste of KAT X56, which
together with KAT B9 were next in preference. This 'could be attributed to exposure of
the two varieties to the group since majority of the older respondents were in the rural
areas where the KAT varieties had been introduced. This implies that there could be a
shift in taste preference from the old (GLP) varieties to other varieties if introduced to
consumers. Many studies some of which include Scott & Maiden (1998); Wortmann
(1998) and Mkanda (2007), have indicated the importance of taste in the food choice
process and that it influences demand, In Debaniyu et al. (2011), taste of cowpeas
statistically influenced demand in Niger state of Nigeria. Mazur (2011) emphasized that
taste, texture, appearance and cooking time were. the key attributes that should be
enhanced in development of bean value chain. Results in this study further showed that
two varieties, GLP 92 and GLP 24 were the least ranked for taste preference.
66
The last four attributes; cooking time, color, grain size and flatulence had low
overall percentages. This implied consumers considered cooking quality, keeping quality,
price taste and grain mix more important when making their choice decisions.
Cooking time variable was seen to have very little influence when consumers
choose beans in the market since nearly all varieties took between 2 -3 hours to cook,
apart from GLP 2 and KAT B1 which cook for 1-2 hours. GLP 585 was the preferred
variety for the attribute. Lack of difference in cooking time meant whichever variety was
chosen would have no advantage over the others. Though important as indicated in
Mishili et al. (2009); Maryange et al. (2010), the characteristic in this study had less
influence in choice decision making and therefore fell in low ranked attributes group.
This is supported by the fact that KAT B1 which had low cooking time was not ranked
among the most preferred bean varieties.
The low preference for color attribute implies that only few consumers were
influenced by color when making choice decision. This could possibly be due to the fact
that some beans do not retain the raw-grain color after cooking despite having nice grain
color. An example is KAT B1 whose yellow/greenish color is not retained in the final
meal which turns a pale brown. The findings are consistent with Scott & Maiden (1998)
who made an observation that grain color is less important compared to less cooking time
and good bean taste in Africa. If the attribute were to be considered, consumers would
prefer dark colored varieties as evaluation results indicated. GLP 585 Red haricot, GLP 2
Rosecoco and KAT X 56 were each ranked excellent by more than 50% of the
consumers. GLP 585 is red colored, KAT X56 is dark purple colored, while GLP 2 is red
mottled. Preference for dark colored varieties is due to the fact that the dark color imparts
67
to the final dish, giving it pleasant rich look for githeri and stew that can be used with
other accompaniments such as rice, chapati and noodles. The results were consistent
with Maryange et al. (2010), where it was observed that Kablanket bean variety which is
highly preferred in Tanzania, has similar characteristics. Other studies include
ASARECA, (2003), where red mottles and reds accounted for 50% of the market share in
Rwanda because they gave color to cassava and Irish potatoes. In Korir et al. ( 2005),
residents in Githurai Market, Kenya preferred GLP 585 because of its strong red color
that blended well with maize while in Katungi et aI., (2009) where brownish/purple or
reddish color beans were preferred for imparting red color to food. Further results in this
study indicated that color of GLP 92 which is cream mottled was less liked by
consumers. This could be due to the fact that the variety exhibits dull color which imparts
to the final dish.
Results show that large sized varieties were more preferred. GLP 2 Rosecoco
which is medium sized was highly preferred followed by KAT X56. This could be
attributed to the fact that large grains expand more, meaning that less amounts would be
required to make a meal compared to the small grained varieties. The results are in line
with study by Wortmann et al. (1998) where the authors observed that large and medium
sized seeds were preferred in Africa. However in this study grain size was not considered
by consumers as an important criterion of making choice decision. This can be proved by
the fact that majority of respondents preferred GLP 585 variety which is a small grained
variety while second in preference was GLP 2 which has large sized grains. The choice of
GLP 585 which is small grained as the preferred variety could have been due to the fact
that one lot of the variety would have more grains than the large size varieties. This
68
implies that food prepared with small grained varieties will have more grains and will be
more visible making the food look rich unlike food prepared with large grains.
Observation made in a more recent study by Katungi et al., (2009) indicated that small
and medium sized bean varieties were preferred by consumers in making githeri.
All beans caused flatulence, none could therefore be said to be free of flatulence.
Based on this, it was not a very good determinant of choice in the market. This was
probably the reason why the attribute was least preferred at 2% overall preference rank
compared to the other attributes. Evaluation results for low flatulence in varieties
indicated that KAT B1 followed by KAT B9 and GLP 585 were preferred, respectively.
It was however important to note that the first two varieties were the least preferred in the
variety rank. This further strengthens the observation that flatulence was not considered
by consumers as an important attribute when making choice decision. The finding may
appear to negate observation made by Katungi et al., (2011) where low flatulence was
highly considered in bean variety demand in drought prone areas of Eastern Kenya.
However it was observed that there was a rider in the findings, that flatulence was
considered by the top stratum households, this means the situation may not necessarily
apply to other household as was the case in this study.
These findings concur and contrast with earlier studies. In ASARECA (2003),
grain size, color and cooking time were the main preference criteria used by consumers.
In Korir et al. (2005) cooking time, grain color and flatulence were important attributes
in driving preference. Although the findings of the two studies agree with this study when
it comes to varieties that contain the said attributes, they differ in importance of the
69
attributes as drivers of preference. In the earlier studies, the attributes were important
preference drivers while they were less influential in the current study.
Results of attribute evaluation have shown that GLP 585 which was ranked
number one as the preferred variety had most of the preferred attributes. It was ranked
5"excellent" for the four important attributes considered in variety choice - cooking
quality, keeping quality, price and taste. It was also ranked highly in cooking time, color
and flatulence attributes. This means that ranking of this variety was based on the
positive attributes it has. The implication is that varieties with the preferred attributes are
likely to be chosen by consumers in the market.
Another observation was that KAT B1 had most of the preferred attributes;
cooking quality, keeping quality, taste, cooking time and flatulence. However it was not
among the preferred varieties. This could have been as a result of it's high market price at
an average of Ksh. 97 per kilo and little information about the variety since it was
consumed by approximately 15% of the respondents. This is an area that requires further
investigation.
GLP 92 was least preferred in most of the attributes namely; taste, cooking time,
color, size. However it was scored 4th in the variety preference rank which was on the
higher side. The reason for this rank could be the low market price of the variety. This
implies that consumers are likely to forego varieties with preferred attributes for low
priced beans. This finding also requires furthe investigation.
5.5 Effect of preferred attributes on willingness to pay price.
There was a change when the same variables were analyzed against what
consumers would be willing to pay in the Hedonic price analysis. Consumers considered
70
cooking quality an important attribute in the willingness to pay prices. All the other
varieties apart from KAT X56and KAT B9 had positive coefficients for the attribute
though none was statistically significant. This means the attribute attracted premiums in
the willingness to pay prices for GLP 2, GLP 24 GLP 585, GLP 92 and KAT B1
varieties. The results show GLP varieties attracted premiums while only one KAT B1 had
positive coefficients among the tested KAT varieties. Implication of the results is that
varieties with high cooking quality attribute, which was considered by consumers as one
of the most important bean attributes, would attract higher willingness to pay prices. The
results conform to results of Edmeades (2005), study of different banana varieties' effect
on farm gate prices. Endemic banana varieties which were considered superior in terms
of cooking quality captured higher farm gate prices.
Out of all tested varieties, keeping quality attribute influenced willingness to pay
price of only GLP 585 variety. This means one percentage change in the attribute would
change the stated price of the variety by 12%. Keeping quality just like cooking time was
seen to have a bearing in the cost of cooking beans. The more a variety kept without
getting spoilt, the less the frequency of cooking beans in a week, ultimately reducing the
amount of fuel used in a week. The attribute which was ranked as the second most
important attribute considered in bean choice process was ranked 5 "excellent" in GLP
585. The attribute therefore plays a crucial role in setting willingness to pay price.
The market price was the only extrinsic variable included in the regression
models. The variable explained willingness to pay price in most varieties but was
statistically significant in GLP 585. The offered prices of all varieties apart from KAT B9
and GLP 24 explained a certain percentage of the willingness to pay price. This means
71
that consumers found the prices of these varieties favorable. The fact that only GLP 585
was statistically significant makes the results conform to the attribute evaluation where
price attribute was highly ranked in GLP 585.
Taste attribute positively influenced willingness to pay of two KAT bean varieties
and one GLP variety. This is an indication that consumers were willing to pay a premium
for the new KAT bean varieties for their taste. Taste remains an important attribute when
making choices in the market. The taste of GLP 585 continues to be an important
component in variety choice process. GLP 2 and KAT B1 that had been preferred for
taste attribute (Katungi et al., 2010), posted negative coefficients in this study. This
means that a variety may have preferred attributes but this does not necessarily translate
to the amount of money the consumer is willing to spend on the variety. In hedonic price
analysis by Anang et al. (2011) taste was one of the attributes that defined quality of rice
most preferred by consumers. In ASARECA (2003), taste was one of the attributes
consumers looked for when purchasing bean varieties in D.R Congo.
Cooking time in GLP 585 explained 17% variation in willingness to pay for every
1% change in the attribute. Cooking time was an important attribute consideration in
choice process given that it had an implication in both cost of bean cooking and
availability of consumers for cooking. Cost of bean cooking stood at an average of
Kshs.199 in a week which is 17% of what consumers with a monthly income of Kshs
5000 earn in a week. This cost is high considering that it is for boiling beans only without
factoring in the accompaniments. GLP 585 was mainly preferred by the urban consumers
who are away from their homes most of the time. The discount consumers gave the
variety implied cooking time of the variety was favorable to their living situation in the
72
urban setting. Similar findings are reported in Mazur et al. (2008) where Ugandan
varieties K132 and NABE 4 were preferred by farmers for their short cooking time
amongst other qualities. In Katungi et al (2010), cooking time was listed as one of the
selected characteristics among others for GLP 24. As reported in ASARECA; (2003);
Korir et al (2005), cooking time played a key role in consumers variety choice. In Anang
et al. (2011), long duration in cooking attracted low premiums in rice. In study of
consumer preference for beans in Malawi, 53.5% ofthe respondents preferred beans with
short cooking time (Chirwa, 2007).
Results indicated that consumers would pay a premium for the dark colors of the
old bean varieties while they preferred the green color of the new KAT variety. Estimated
color coefficients had positive signs in GLP 585, GLP 92 and KAT Bl varieties. A 1%
increase in color unit of GLP 2 would increase the willingness to pay price by an average
of 3.6% while a 1% increase color unit of GLP 92 would improve the price by 5%. The
two varieties have dark colors which impart to the final dish explaining the positive
influence on the willingness to pay. The last variety KAT B1 is green colored and is
mainly used in dishes whose accompaniments include rice. This means the visual color of
beans influences the price consumers are willing to pay especially when the dish it is
going to be used in is put into perspective. This is in line with other studies (Korir et al.,
2005; Katungi et al., 2009; Katungi et al., 2010) dark colors such as reds were preferred
by consumers.
Grain size attribute was important in determining the pnce consumers were
willing to pay for varieties. Increase in size of grain improved the price consumers were
willing to pay while small grained varieties reduced the stated prices. A one percent
73
increase in grain size of KAT X56and KAT B9 improved the stated prices by 26% and
41% respectively. Grain size increase of GLP 585 reduced the willingness to pay price by
25% for the variety. Both KAT X56 and KAT B9 are medium sized grains while GLP
585 is small grained. Large grains are associated with expansion upon cooking which
means consumers require less amount of a variety to make a dish. Similar findings were
reported in Mishili et aI., (2009), where KAT B9 and GLP 92 had positive signs for grain
size in a consumer preference study for common beans in Tanzanian markets.
Flatulence attribute had a negative effect on the willing to pay prices in majority
of the varieties tested. This implies that the presence of the attribute in those beans
reduced the amount of money a consumer offered for the variety. In the varieties that had
positive coefficients, the explanation was that they had low flatulence which attracted a
premium in the willingness to pay price. The negative coefficients of flatulence attribute
in KAT X56, GLP 24, GLP 92 and KAT B1 meant that they reduced the percentage of
willingness to pay prices by an average of 28%, 52%, 27% and 2% per every 1% increase
of the attribute respectively. The varieties that had positive coefficients were GLP 2, GLP
585 and KAT B9. The low flatulence attribute in the two GLPs could have contributed in
the varieties being ranked among the most preferred varieties in this study. An important
example of bean variety with low flatulence is Soya bean found in Northern Tanzania
which is blended with other varieties to improve quality and attract premiums in the
market, (Korir et aI2005).
The willingness to pay price stated by consumers was explained by bean attributes
III some varieties. Cooking quality, keeping quality, price and taste attributes were
expected to influence the willingness to pay price after being ranked the preferred
74
attributes. Observations made were that the coefficients of the attributes were either not
statistically significant or they had opposite influence of what was expected. An example
is cooking quality attribute which was ranked as the most important bean attribute, with
GLP 585 and KAT B1 ranking highly for the attribute but were not statistically
significant. What this implies is that consumer preference for attributes and to some
extent varieties did not necessarily translate to willingness to pay for them. Out of the
four preferred attributes only two, price and taste, were statistically significant.
Coefficients for price of GLP 585, taste of GLP 585 and GLP 2 were statistically
significant at 5% and 1% levels respectively, making them the only attributes that
explained willingness to pay.
Results of Hedonic price model could possibly mean that there were other factors
other than what was included in this study that explained the willingness to pay prices.
These could possibly be general appearance of grains, level of grain damage by bruchid
and disease, level of foreign material among the bean grains or environmental effects.
Another possible reason could be that the prices consumers indicated they were willing to
pay were done arbitrary and did not match with the attributes ranking. These results are
not isolated. In Marreiros & Ness, (2009), where they reviewed theory on processes of
consumers' decision making, they observed that sensory preference is an indicator of
food acceptance which could or could not be a predictor of consumer behavior.
According to Richardson, MacFie, & Shepherd (1994); Raats et aI., (1995) in Mairreiro
et al (2009) and Asp (1999) taste was clearly a crucial parameter in determining food
acceptability. However, these authors argued that when buying behavior was examined, it
came out clearly that taste was not the only crucial determinant and in some cases was
75
way down the priority list. In other words preference of particular attributes does not
necessarily mean consumer will base purchase decision on the attribute preference rather
other factors may come into play such as expectation, perceived risks, perceived ethnic
origin, hunger, expectations of reward, and the level of uncertainty about a product's
identity (Bell & Marshall, 2003). According to Groote & Kimenju (2008), the open
ended format used in getting the willingness to pay price from consumers can be
problematic when respondents do not have enough information and stimuli to consider
thoroughly the values they would attach to a good with the preferred attributes, and might
thus not return realistic estimates.
76
CHAPTER SIX
6.0 CONCLUSION AND RECOMMENDATIONS
6.1 Introduction
The chapter presents the conclusions and recommendations. Conclusions are
based on findings of this study for the three objectives. Recommendations which are
targeted for the policy makers and agricultural sector practitioners were made to help
bridge the identified gaps.
6.2 Conclusion
The main purpose of this study was to find out which attributes influence
consumer preference in common beans. Results from this study have important
implications that can be used by institutions locally and internationally to enhance bean
varieties as well as develop new ones.
The results indicated that respondents consumed beans at least once in a week an
indication that beans are important in respondents' diets. All respondents consumed
beans as results showed with 55% of the total sample of consumers based in urban areas
and 45% in rural areas. This means the findings give a valid reflection of bean consumer
behavior in the two settings. Rural respondents consumed beans more frequently than
urban consumers; with an average of 53% rural respondents consuming beans more than
thrice in a week against 47% urban consumers. This is an indication of ready market for
beans which should encourage improvement of bean quality by incorporating desired
attributes during breeding.
77
Results from traders' and consumers' evaluation showed that consumers made
bean choices in the market an indication that consumer preference for the different
varieties was not the same. They further affirmed the observations made in the existing
literature that the older improved bean varieties, GLPs, were popular with consumers.
GLP 585 emerged as the most preferred variety followed by GLP 2 Rosecoco and KAT
X56 ranked 1, 2, and 3 respectively. The other varieties had low preference ranks
starting with GLP 92, GLP 24, KAT B9, and lastly KAT Bl with ranks of 4,5,6 and 7
respectively. KA T varieties were popular among rural respondents than urban
respondents, while GLP varieties were popular among both rural and urban consumers;
this was an indication that they were still relevant even with entrance of newer varieties
in the market. This leads to the conclusion that the GLPs have culinary characteristics
that appeal to consumers.
The pairwise comparison that was done on the seven bean attributes gave a clear
indication of the attributes consumers considered in bean choice. Cooking quality and
keeping quality, followed by price and taste were the first four preferred attributes which
means that the attributes are important in consumer bean choice. Cooking quality is the
ability of the cooked grain to remain whole without mashing up. Keeping quality
indicates cooked grains' ability to stay fresh without getting spoilt within three days. The
conclusion is that for a variety to be accepted by the consumers, it should possess either
of the two attributes as well as price and taste.
Price which was the only extrinsic attribute evaluated in this study was third in
preference list after cooking quality and keeping quality. This concludes that market
prices play a crucial role in consumer choice of beans in the market. This fact was
78
observed in preference for GLP 92 and non-preference for KAT Bl varieties which had
low and high market prices respectively. This implied that affordable market prices
contributed to consumer choice of a variety in the market.
It was evident from the results that preferred varieties had at least one of the four
important attributes; cooking quality, keeping quality, price and taste. This affirms the
conclusion made in this study that for a variety to be chosen by consumers, it should
possess at least one of the important attributes. The newly released varieties have better
production traits, however as this study showed, consumers considered culinary traits and
not production traits when making bean choices. The culinary traits which most of GLP
varieties (released earlier) possessed included; cooking quality, keeping quality and taste.
This means these attributes were the drivers of consumer variety preference.
The expectation that preferred attributes would influence willingness to pay price
was only true for price and taste attributes in GLP 585 and GLP 2 varieties.
6.3 Recommendations
Consumer preference for GLP 585, GLP 2 and KAT X56 varieties and the fact that
consumers made choices when purchasing beans implies that production and stocking of
these varieties will find ready market. In the existing literature producers prefer beans
that have favorable production traits such as drought tolerance and high yields without
taking into consideration consumption traits. This means that what is supplied to the
market is not necessarily what consumers want. Relevant stakeholders in the bean supply
chain; breeders, producers and traders should ensure there is sufficient supply of
preferred varieties to the many bean consumers. This can probably be realized by
79
ensuring that production of certified seed of the preferred bean varieties by Government
supported and private breeding institutions is supported to ensure the seeds produced are
available and affordable. This would enhance production of beans, by farmers, as an
agribusiness venture in the short term, rather than production for consumption.
The preferred bean attributes were cooking quality, keeping quality and taste. These
results give basis for both public and private seed breeders to come up with development
and improvement programs of these key attributes in beans as a long term intervention.
This information should also help traders to stock bean varieties with preferred attributes
to facilitate accessibility of desired varieties as a short term intervention. This would
possibly enhance utilization creating an agribusiness opportunity for both producer and
trader.
Results of attributes influence on consumers' willingness to pay were contrary to
what had been expected. This calls for further exploration to establish which attributes in
beans positively influences what consumers are willing to pay for the different bean
varieties. Such information would help breeders and traders plan on areas that can be
improved with assurance that consumers will be willing to pay for the improvement.
Despite the high consumer preference in attributes of KAT B1, it was not ranked
among the preferred varieties. Price of the variety was high compared to the other
varieties a clear indication that the attribute contributed to its low preference ranking.
Price was ranked by consumers as one of the important attributes in the pairwise
comparison. This information is fundamental to traders who can strategize to facilitate
accessibility of the variety to consumers at affordable prices. Local supply of certified
80
seed should also be enhanced to ensure there is an increase in production of the variety;
this can possibly lower the prices.
Evaluation of information channel is suggested. Results from this study indicated the
main information channel was fellow consumers. This narrowed knowledge base about
the different bean varieties to only what fellow consumers knew. A case in point was
minimal information among consumers of the KAT varieties, which had desirable
attributes. Information dissemination through ministry of Agriculture home economics
department and nutritionists in the ministry of Health would contribute in creating
awareness of important attributes leading to acceptance of varieties.
81
REFERENCES
Abansi, C., Lantica, F., Duff, B., & Catedral, I. (1990). Hedonic Model Estimation:
Application to Consumer Demand for Rice Grain Quality. Trans. Acad. Sci.
Techno!. 12 , 235-256.
Abley, J. (2000). Stated Preference Techniques and Consumer Decision Making: New
Challenges.to Old Assumptions. Cranfield, U.K.: Crankfield University ..
American Bean Organization. (2008 , 09 01). Beans/Bean Recipes/Bean Facts. Retrieved
september 1st, 2011, from American Bean Organization:
http://www.americanbean.org
Anang, B. T., Adjetey, S. N., & Abiriwe, S. A. (2011). Consumer Preferences for Rice
Quality Characteristics and the Effects on Price in the Tamale Metropolis,
Northern Region, Ghana. International Journal of Agriculture Vol. 1 (2) ,67-74.
APA. (2010, September 1st). Pulse Health Initiative . Retrieved July 31st, 2011, from
American Pulse Association: http://www.pea-lentil.com
ASARECA. (2003). BEANS: priority setting document by ECABREN Entebbe:
ASARECA Secretariat.
Asp, E. H. (1999). Factors Affecting Food Decisions Made by Individual Consumers.
Food Policy 24 (2-3) , 287 - 94.
Bech, A., M.Hansen, & Wienberg, l. (1997). Application of House of Quality in
Translation of Consumer Needs into Sensory Attributes Measurable by
Descriptive Sensory Analysis. Food Quality and Preferences Vol. 8 , 329-348.
Bell, R., & Marshall, D. W. (2003 ). The Construct of Food Involvement in Behavioral
Research: Scale Development and Validation. Appetite, 235 - 44.
Bressani, R., & Chon, C. (1996). Effects of Altitude Above Sea Level on the Cooking
Time and Nutritional Value of Common Beans. Plant Foodsfor Human Nutrition
49,53-61 .
Boccaletti, S., & Nardella, M. (2000). Consumer Willingness to Pay for Pesticide-free
Fresh Fruit and Vegetables in Italy. International Food and Agribusiness
Management Review, 297-310.
Breidert, C., Hahsler, M., & Reutterer, T. (2006). Review of Methods for Measuring
Willingness to Pay. Innovative Marketing.
82
Broughton, W., Hernandez, G., Blair, M., Beebe, S., Gepts, P., & Vanderleyden, J.
(2003). Beans (Phaseolus spp.) - Model Food Legumes. Plant and Soil 252: FAD
(pp. 55-128).
Brown. (2003). Introduction to Stated Preference Methods. Kluwer Academic Publishers.
Buergelt, D., Oppen, M. V., & Yadavendra, J. (2009). Quality Parameters in Relation to
Consumer's Preference in Ricebean. International Conference on Grain Legumes:
Quality Improvement, Value Addition and Trade. Gujarat, India: IIPR, Kanpur
India.
Canada, A. F. (2009). The Brazilian Consumer Behaviour, Attitudes and Perceptions
Towards Food Products. Ottawa: Government of Canada.
Chelangat, J. (Unpublished). Hedonic Price Analysis: A Case Study of Bean Pricing at
Nakuru Municipal Market and Its Periphery. Nakuru: Unpublished: MBA Thesis
Egerton University 2009.
Chirwa, R. (2007, June). Consumer Preference for Beans in Malawi and its Influence on
Price. Highlights CIAT in Africa ,pp. 1-2.
Chochrah, W. (1963). sampling Techniques, 2nd Edition. New York.: John Wiley and
Sons Inc.
Coelho, C., Vargas, V., Souza, C. A., Peraira, T., & Santos, J. C. (2009). Cooking Quality
of Common Beans as Influenced by Different Nitrogen Levels and Time of
Application. International Journal of Agriculture and Biology, 261-265.
Dalton, T. J. (2003). A Hedonic Model of Rice Traits: Economic Values From Farmers in
West Africa. 25th International Conference of Agricultural Economists (IAAE)
(pp. 1552-1560). Durban, S.A: Document Transformation Technologies.
Danbaba, N., Anounye, J., Gana, A., Abo, M., & Ukwungu, M. (2011). Grain Quality
Characteristics of Ofada Rice: Cooking and Eating Quality. International Food
Research Journal 18 , 629-634.
Debaniyu, F. 1., Yakubu, J. C., Kutigi, U. R., & Ingye, Y. 1. (2011). Cowpea Marketing
and Consumer Preference in Magama Local Government Preference in Magam
Local Government Area of Niger State, Nigeria. Journal of Agriculture & Social
Sciences, 1813-2235.
Deodhar, S., & Intodia, V. (2002, March-April 21). Does Ghee sold by any brand smell
as sweet? Quality attributes and Hedonic Price Analysis of Ghee. Indian Food
Industries, p. 2.
83
Diamant, R., Watts, B., Elias, L., & Rios, B. (1989). Consumer Utilization and
Acceptability of Raw and Cooked Black Beans (Phaseolus vulgaris) in
Guatemala .. Ecology of Food and Nutrition 22,183-195.
Economides, N. (2010). Notesfor Microeconomics. New York: New York University.
Edmeades, S. (July 24-27, 2005). A Hedonic Perspective to Estimating a Marginal Value
Functionfor Vqriety Attributes of a Subsistence Crop. Washington DC: IFPRl.
Engel, J., Blackwell, & Miniard, P. (1995). Consumer Behavior. Forth Worth: The
Dryden Press.
EI-Tabey, S. A. (1992). Hard-to-Cook Phenomenon In Legumes. Food Reviews
International 8 , 191-221 .
Espinosa, J. A., &. Goodwin, B. K. (1991). Hedonic Price Estimation for Kansas Wheat
Characteristics. Western Journal of Agricultural Economics. Vol 16. no.10 , 72-
85.
FAa.' (2011, May 17). faostatfao. Retrieved August 1, 2011, from FAOSTAT:
faostat.fao.org
Gichangi, A., Karanja, D., Ngari, M., Mutuoki, T., & Lelgut, D. (2011). Common Bean
Marketing in the Larger Nakuru District. Nairobi: KARl.
Gil, J. M., Graca, A., & Sanchez, M. (2000). Market segmentation and Willingness to
Pay for Organic Products in Spain. International Food and Agribusiness
Management Review, 207-226.
GoK. (201Oa).Annual Agricultural Report for Thika District:: Government Printer.
GoK. (2010b). Economic Review of Agriculture 2010. Nairobi: Government Printer.
GoK. (2009). Kenya Population and Housing Census. Nairobi: Government Printer.
GoK. (2007). Kenya Vision 2030. Nairobi: Government Printer.
Govindasamy, R., Italia, J., & Adelaja, A. (2002). Farmers' Markets: Consumer Trends,
Preferences and Characteristics. Journal of Extension.
Groote, H. D., & Kimenju, S. C. (2008). Comparing Consumer Prefernce for Color and•Nutritional Quality in Maize: Application of Semi-Double-Bound logistic Model
on Urban Consumers in Kenya. Food Policy, 362-370.
Hanley, N., Maurato, S., & Wright, R. (2001). Chice Modelling Approches: A Superior
Alternative for Environmental Valuation? Journal of Economic Survey Vol.15 ,
435-462.
84
Israel, G. D. (1992). Determining Sample Size. Florida: University of Florida.
Karanja, D., Kavoi, J., Sperling, L., Rubyigo, J. C., & Ogecha, J. (2012). Promotion of
Improved Bean Varieties in Nyaza, Western and Central Rift Valley Provinces of
Kenya. Kari Biennial Scientific Conference Proceedings. Nairobi: KARl.
Katungi, A.Farrow, Mutuoki, T., S.Gebeyehu, D.Karanja, F.Alemayehu, et al. (2010).
Improving Common Bean Productivity: An Analysis Of Socio-Economic Factors
In Ethiopia And Eastern Kenya. Cali, Colombia: CIAT.
Katungi, E., Farrow, A., Chianu, J., & Beebe, L. S. (2009). Common Bean in Eastern and
Southern Africa: a Situation and Outlook Analysis. Nairobi: ICRlSAT.
Katungi, E., Sperling, L., Karanja, D., Farrow, A., & Beebe, S. (2011). Relative
importance of common bean attributes and variety demand in the drought areas of
Kenya. Journal of Development and Agricultural Economics Vol. 3(8) ,411-422.
Kaul, A. K. (1987). Coarse Grains, Pulses, Roots and Tuber.Processing and Nutrition.
Bogor: FAO. .
KEPHIS. (2011). National Crop Variety List - Kenya. Nairobi: Government Printer.
Kibiego, M., Odhiambo, M., & Kimani, P. (2003). Analysis of Bean Marketing System in
Urban areas of Kenya. African Crop Science Society Conference Proceeding (pp.
587-590). Kampala, Uganda: African Crop Science Society.
Kiesel, K., & Villas-Boas, S. B. (2007). Got Milk? Consumer Valuation of Milk Labels
after Implementation of the USDA Organic Seal. Agricultural & Food Industrial
Organization.
Kieti, R. M. (2005). Application of Multiple Regression Analysis (MRA) in the Valuation
of Used Motor Vehicles. Nairobi Kenya: Unpublished MSC Thesis, School of
Built Environment University of Nairobi.
Koistinen, L. (2010). Consumers Relative Preferences for Meat Attributes and the Impact
of Carbon Footprint Infromation on Consumer Choice. Berlin: Masters Thesis
Unpublished.
Kimani, P. M., Buruchara, R., Ampofo, K., Pyndji, M., Chirwa, R. M., & Kirkby, R.
~005). Breeding Beans For Smallholder Farmer In Eastern, Central and Southern
Africa: Constraints, Achievements and Potential. Occasional Publication Series ,
pp. 1-28.
Korir, M., Odhiambo, M., Iruria, D., Kipsat, M., & Serem, A. (2005). Bean varietal
preference in East African markets and its implications to Breeding. African
85
Crop Science Conference Proceedings, Vol. 7. (pp. 803-805). Kampala, Uganda:
African Crop Science Society.
Lancaster, K.J., (1966). A New Approach to Consumer Theory. Journal of Political
Economy 74(2):132-157.
Langyintuo, A., Ntoukam, G., Murdock, L., Lowenberg-DeBoer, J., & Miller, D. (2004).
Consumer Prefernce for Cowpea in Cameroon and Ghana. Agricultural
Economics 30 (Elsevier) ,203-213.
Laswai, H., Shayo, N., & Kundi, S. P. (2008). Collaborative Project to Investigate
Consumer Preferences for Selected Sorghum and Millet Products in SADCC
Region of Africa. Morogoro, Tanzania: Sokoine University of Agriculture.
Leterme,P., & Carmenza, L. (2002). Recommendations by health organizations for
pulse consumption. British Journal of Nutrition , 239-242
Luse, R. (1980). The Need for Food Utilization and Processing Studies to Supplement
Nutrientional Evaluation. Food and Nutrition Bulletin Vol 2, No.1, p. 61.
Marbeau, Y. (1987). What Value Pricing Research Today? Market Research Society,
153-182.
Marreiros, C., & Ness, M. (2009). A Conceptual Framework of Consumer Food Choice
Behaviour. Largo dos Colegiais, Portugal: CEFAGE-UE.
Maryange, M., Nchimbi-Msolla, Sawargaonkar, S., Hudge, B., & Thanki, H. (2010).
Screening of 30 Advanced Common Bean Lines for Short Cooking Time Using
Different Methods. Electronic Journal of Plant Breeding 1(4) , 505-511.
Massey, L. K. (2003). Dietary Animal and Plant Protein and Human Bone Health: A
Whole Foods Approach. Journal of Nutrition 133 , 862s-865s.
Mauyo, L. W., Chianu, 1., Nassiuma, B. K., & Musebe, R. O. (2010). Cross-Border Bean
Marketing Perfomance in Western Kenya and Eastern Uganda. Journal Service
Science and Management 3 , 501-511.
Mazur, R., Nakimbugwe, D., Ugen, M., Musoke, H. K., & Vasanthakaal, H. (2008).
- . Enhancing Nutritional Value and Marketability of Beans through Research and
Strengthening Key Value Chain Stakeholders in Uganda and Rwanda. Iowa: Iowa
State University.
Mazur, R. (2011). Integrating Agricultural Production, Nutrition and Marketing -
Research and Development of the Bean Value Chain in Uganda. West Africa
CAADP Nutrition Program Development Workshop, (pp. 1-21). Dakar, Senegal.
86
Mbugua, G., & Munene, S. (1997). Determintion of Dry Bean Varieties Grown by
Farmers in Kenya's Major bean producing Areas. Nairobi: Government Printer.
Mbugua, G., Wachiuri, S., Karoga, J., Kimamira, J., Ndegwa, A., & Waiganjo, M.
(2005). Farmer-Participatory Evaluation Of Dry Bean Varieties With Multiple
Constraints Resistance In Central Kenya. Nairobi: KARl, Thika.
Mclennon, E. (2002). Analysis of Consumer Perception Toward Biotechnology and Their
Preferences for Biotech Food Labels. Louisiana: Louisiana State University .
Mennecke, B. E., Townsend, D. J., Hayes, & Lonergan, S. M. (2007). A study of the
factors that influence consumer attitudes towards beef production using conjoint
market analysis tool. Journal of Animal Science 85, 2639-2659.
Mishili, F. J., Temu, A. A., Fulton, J., & Lowenberg-Deboer, J. (2009). Consumers
Preferences As Drivers of the Common Bean Trade in Tanzania: A Marketing
Perspective. West Lafayette, USA: Purdue University .
. Mkanda, A. V. (2007, September). Relating Consumer Preferences to Sensory and
Physicochemical Properties of Dry Beans. Pretoria, South Africa: University of
Pretoria.
Moser, R., Raffaelli, R., & Thilmany-Mcfadden, D. (2011). Consumer Preferences for
Fruit and Vegetables with Credence-Based Attribute: A Review. International
Food and Agribusiness Management Review.
Munene, S. (1993). Retail Market Survey of Dry Beari Cultivars Sold in Kenya.
Proceedings of 3rd Multidisciplinary Workshop on Bean Research in Eastern
Africa (pp. 283-289). Kampala: CIAT.
Mundua, J. (2010). Estimation Of Consumer Preferences For Cowpea Varieties In Kumi
And Soroti Districts, Uganda. Makerere: Makerere University.
Murray, M. T. (2010). Affordable Natural Factors. Retrieved July 21st, 2011, from
http://www.affordable-natural-factors.com/dr murray beans and health.html
Mutai, K. B. (2000). How to Write Quality Research Proposal: A Complete and
Simplified Recipe. Thelly Publications, New York.
Nzau, B. M. (2003). Modelling the Influence of Urban Sub-Centre On Spartial and
Temporal Urban Land Value Patterns. Netherlands: Unpublished Msc Thesis,
International Institute for Geo-Information Science and Earth Observation (lTC).
87
Odendo, M., David, S., & Otsyula, R. (2002). Impact Of Root-Rot Resistant Bean
Varieties In Western Kenya: Application OfImpact Diagramming. Impact of New
Bean Technologies (pp. 141-146). Nairobi: KARl.
Padilla, C., Villalobos, P., Spiller, A., & Henry, G. (2007). Consumer Preference and
Willingness to Pay for an Officially Certified Quality Label: Implications for
Food Producers. Agricultura Tecnica (Chile) 67(3) , 300-308.
Palmquist, R. B. (2003). Property Value Models. Raleigh, USA: North Carolina State
University.
Pearmain, D., Swanson, 1., Kroes, E., & Bradley, M. (1991). Stated Preference
Techniques: A guide to Practice, Second Edition. Hague: Steer Davies Gleave and
Hague Consulting Group.
Perla, 0., Luis, A., Sonia, G., Maria, P., Juscelino, T., & Octavio, P. (2003). Effects of
Processing and Storage Time in Vitro Digestability and Resistant Starch Content
of Two Bean (Phaseolus vulgaris L.) Varieties. Journal of Science of Food and
Agriculture 83 , 1283-1288.
Picard, N., (2010) Econometric Models Working Paper 2.4. France: THEMA, University
de Cergy-Pontoise.
Portugal, L. A. (2004). Relevant Quality Attributes of Edible Dry Beans. Osnarbruck: Der
Andere Verlag.
Raats, M., Daillant-Spinnler, B., Deliza, R., & MacFie, H. (1995). Are Sensory Properties
Relevant to Consumer Food Choice. Food Choice and COnsumers.
Richardson, N., MacFie, H., & Shepherd, R. (1994). Consumer Attitudes to Meat Eating.
Meat Science, 36,57-66.
Rosen, S., (1974). Hedonic prices and implicit markets: produce differentiation in pure
competition Journal of Political Economy. 34-55.
Rubyogo, J. C., Sperling, L., & Kasambala, L. N. (2007). Developing seed systems with
and for the marginalized: case of common beans (Phaseolus vulgaris L.) in East,
Central and Southern Africa. Farmers first 20 Years Revisted. Future
Agricultures.
Ruiz London-o, N., Garcra, J., & Pachico, D. (2000). Impacto de la adopcio'n defrejol
(Phaseolus vulgaris) en Santa Cruz, Bolivia (Impact of the adoption of beans
(Phaseolus vulgaris) in Santa Cruz, Bolivia). Cali, Colombia: CIAT.
Santalla, M., Fueyo, M. A., Rodino, A. P., Montero, I., & Ron, A. d. (1999 ). Breeding
for Culinary and Nutritional Quality of Common Beans in Intercropping Systems
88
with Maize. Biotechnologie, Agronomie, Societe et Environnment 3 (4), , 225-
229.
Sathe, S., & Deshpande, S. (2003). Beans. Encyclopedia of Food Science and Nutrition
2nd Edition. Vol I , pp. 403-412.
Schamel, G., Gabbert, S., &Witzke, H. (1998). Wine Quality and Price: A Hedonic
Approach. Daniel Pick.
Schwartz, H., & Corrales, P. (1989). Bean Production Problems in the Tropics 2nd
Edition. Cali, Colombia: CIAT(Centro Internacional de Agricultura Tropical).
Scott, J., & Maiden, M. (1998). Socio-economic Survey of Three Bean-Growing Areas in
Malawi. occasional Paper Series No. 24. Kampala, Uganda: CIAT.
Wortmann, S. c., & Roger A. Kirkby, C. A. (1998). Atlas of Common Bean (phaseolus
vulgaris L.) Production in Africa. Feriva, South Africa: CIAT Publication No.
297.
Smith, P. (2008, January 25). Chemical Fertilizers Use Linked to Climate Change.
Retrieved November 3rd, 2011, from chemical-fertilizer-use-linked-to-climate-
change/: http://www.pinoypress.net
Soniia, D. (1999). Beans in the Farming System and Domestic Economy of Uganda: A
Tale of Two Parishes. Network on Bean Research in Africa, Occasional
Publication Series No. 28 .
Spilsbury, J., Jagwe, J., & Wanda, K. (2004). Evaluating the Marketing Opportunities for
Beans and Its Products in the Principle Beans Growing Countries of ASARECA.
IITA.
Tapia, H. (1985). Criterion of Selection and Intake of Common Beans in Nicaragua.
Current and Futuer Requirements in Food in Central America and Panama (pp.
42 - 47). Quatemala City: Republic of Quatemala.
Volckner, F. (2006 ). An empirical Comparison of Methods for Measuring Consumers'
Willingness to Pay. Marketing Letters, 137-149.
Von Oppen, M. (1978). A Preference Index of Food Grains to Facilitate Selection for
Consumer Acceptance.: ICRISAT, Economic Program. Hyderabad, India
Wooldridge, J. (2002). Introductory Econometrics. South-Western College.
89
APPENDICES
Appendix 1: Map of larger Thika district showing study area
.-*.•
LEGEHD
8IASHARAo GATUAHYAGA
DJUJABKAKUZIKOMO
t.lAKONGEHI
DUUHYU
/f(;TN lold
i TownD lDcItonel.dmilil•• tvit be.Mary
I Kibmt."
!!!!!!iiii~iiiiiii
Location of study locations
Source: KARl Nairobi.
90
Appendix 2: List of released bean varieties since 1980 to 2010
Variety Year Owners Optimal prodn Duration Yield Special attributes
name/code of alt range to (t /ha) .
release (Masl) maturity
(months)
Mwitemania 1982 KARlIKSC 900-1600 2-3 1.2-1.S Drought tolerant
(GLP 92)
pinto sugars
Rosecoco 1982 KARIlKSC IS00-2000 2-3 High yield Wide adaptation
(GLP 2) Attractive seed color
Good taste
Mwezi Moja 1982 KARlIKSC 1200-1600 2-3 1.2 - 1.5 Good performance
(GLPlO04) in dry areas Early
maturity Tolerant to
drought and bean fly
Canadian 1982 KARIlKSC 1200-1800 3-3.5 1.3 - 1.8 Moderately resistant
Wonder to angular leaf spot
(GLP-24)
GLP-92 1982 KARlIKSC 100-IS00 3 - 3.S 1.2 - 1.7 Wide adaptation
Pinto bean Resistant to halo
blight
GLP-S8S 1982 KARl ISOO-2000 2.5 - 3 1 - 1.S Suitable for high
Red haricot rainfall areas
Resistant to bean
common mosaic
virus
GLP-X 1127 1982 KARlIKSC 1000-IS00 2.S -3 1 - 1.5 Wide adaptation
New Mwezi Resistant to bean
moja common mosaic
virus Tolerant to
rust.
KatlBean 2 1987 KARl 1200-1800 2-3 1-1.2 Tolerant to shading
Kat X 16 1994 KARl 900-1600 2-3 1.5-1.8 High yielding
Kat XS6 1995 KARl 900-1800 2.S-3 1.5-1.8 High yielding
Kat X 69 1995 KARl 1200-1800 2-3 1.S-1.8 High yielding
KK22 1996 KARl IS00-1800 2.S -3 1.8-2 Tolerant to root rot
(RWR 719)
KatlBean 1 1987 KARl 1000-1800 2.S 1.2-1.5 Early maturity
(Katheka)
KK8 1997 KARl IS00-1800 2.S -3 1.8-2 Tolerant to root rot
(SCAM-
8011S)
KK IS 1997 KARl IS00-1800 2.5 - 3 1.8-2 Tolerant to root rot
(MLB
49/879)
Kat-Bean 9 1998 KARl 900-1600 2.S-3 1-l.8 Tolerant to heat
Wairimu 2008 Kenya SOO-1700 2.S - 2.8 l.S - l.7S Early, Heat tolerant,
Dwarf Seed Co Good for maize
intercropping,
excellent cooking
qualities
91
New Rose 2008 University 1100-2000 2.5 -3 1.3 - 2.3 Upright growth
Coco of Nairobi habit, Early,
Moderate resistance
to rust, common
bacterial blight,
angular leaf spot:
anthracnose, bean
common mosaic
virus & necrotic
virus, large grains
MieziMbili 2008 University 1000-2000 2.5 -3 1.2 - 2.26 Large grains, Early,
of Nairobi Resistant to floury
leaf spot, halo
blight, angular leaf
spot, anthracnose,
bean common
mosaic virus &
common bacterial
blight
Kenya early 2008 University 1100-1900 2.5 -3 1.07 - Large grains, Early,
of Nairobi 2.15 Moderately resistant
to, halo blight,
angular leaf spot,
anthracnose, bean
common mosaic
virus & common
bacterial blight
Kenya Red 2008 University 1000-2100 2.5 -3 1.09 - 2.8 Large grains,
Kidney of Nairobi Moderately resistant
to halo blight,
angular leaf spot,
anthracnose, bean
common mosaic
virus & common
bacterial blight
Super Rose 2008 University 1000-2100 2.5 -3 1.14 - 2.8 Medium maturity,
Coco of Nairobi Moderately resistant
to halo blight,
angular leaf spot,
anthracnose, bean
common mosaic
virus & common
bacterial blight
Kenya 2008 University 1030-2000 3 - 3.5 1.13 - Large grains,
Wonder of Nairobi 2.09 moderately resistant
to halo blight,
angular leaf spot,
anthracnose, bean
common mosaic
virus & common
bacterial blight
Kenya Sugar 2008 University 1000-1900 2.5 -3 1.08 - Early, Large grains,
Bean of Nairobi 1.81 Moderately resistant
to halo blight, bean
92
common mosaic
virus & common
bacterial blight 59
Kabete 2008 University 1300-2000 3 - 3.5 1.05 - Large grains,
Super of Nairobi 2.47 Resistant to floury
leaf spot, halo
blight, angular leaf
spot, anthracnose,
bean common
mosaic virus &
common bacterial
blight
Chelalang 2008 Egerton 1800-2200 2.5 -3.5 1.2 - 2.2
University
Tasha 2008 Egerton 1500-2000 2.5 - 3.5 1.1 - 2.1
University
28. Cianku 2008 Egerton 1500-2150 2.5 -3.5 1.0 - 1.9
University
Source: KEPHIS (2011)
Appendix 3: Data collection and analysis plan
Objective Type of data Data source Data Data Expected
collection analysis Output
method (priori)
Identification Bean varieties, Primary data Use Descriptive Consumer
of bean ordinal data from traders questionnaire (frequencies preference in
varieties (independent for preferred and bean varieties
preferred by variables) beans percentages) is different.
consumers. identification
Rank Bean Primary data Use of Descriptive Consumer
attributes that attributes, from questionnaire preference in
influence Ordinal data consumers for sensory attributes of
consumer (Independent evaluation of bean varieties
preference in variables) attributes is different.
beans.
Estimate the Kenya shilling Primary data Consumers' Hedonic Attributes are
pnces per kilo of both traders stated price model. linearly
consumers are beans and willingness to related to the
willing to pay quantitative consumers. prices for price
for the ranked ranked bean consumers
attributes varieties. are willing to
pay for a
variety.
Appendix 4: Summary of consumer preference literature
Title &Year Author(S) Objectives Analytical Methods Results - preferred attributes
Bean varietal M.K. 1. Gauge preference of bean Descriptive - SPSS Soya with low flatulence, fast cooking &
preference in Korir, et varieties traded and was used to generate sweet. Swahili, Muslim & institutions
East African al., consumed. % of respondents preferred Canadian wonder which is cheap.
markets and its 2. Determine the most ranking of the In Kenya preference was diverse, Nyayo,
implications to important quality varieties at each rank Red Haricot because of color
breeding 2005 characteristics that category.
consumers desire
Consumer Carlos 1. Analyze consumer Econometric model 1. Homemade jars achieved highest utility
Preference and Padilla, et preferences for an officially for the price in terms of appearance while lowest prices
Willingness to ai. certified label. attribute: R, = fio + yielded highest utility.
Pay for an 2. Quantify willingness to filP + fi2SCI + jhSC2 2. The most important attribute was
Officially pay for quality indicator. + j34AE + e. certified label followed by price and jar
Certified Quality appearance coming in 3rd. thus consumer
Label: behaviour is guided by quality label
Implications for attribute.
Traditional Food
Producers. 2007
Consumer Jennifer IConsumer rankings of Consumers ranked 1. No growth hormones ranked highest
Preference for Grannis, several specific product production attributes while local raising, within 250miles, ranked
Specific Neal H. attributes directly or in a Likert-scale of 1- lowest. Environmental and animal friendly
Attributes in Hooker & indirectly related to natural 5 latter being most attributes ranked high among consumers
Natural Beef Dawn production methods important value. willing to pay a higher premium and also in
Products. 2000 Thilmany 2. Assessment of how the steak segment.
rankings vary by consumers' 2. Most consumers were willing to pay a
willingness to pay for natural premium for natural meat that promotes
beef products. attention to all the mentioned production
Cowpea
Marketing and
Consumer
Preference in
Magama Local
Government
Area of Niger
State, Nigeria.
2011
Debaniyu
et al.,
l.Determine how marketing
channels have affected
cowpea mkting.
2. Evaluate extent to which
supply of cowpea addresses
demand in the study area.
3. Economic analysis of
cowpea mkting in the study
area.
Regression model
was used to examine
relationship between
cowpea demanded
and variables
practices.
The consumption pattern revealed that
income level of consumers, market price of
cowpea and its close substitute and the taste
of the product are major factors that
determine cowpea consumption.
4. Consumers and producers had preference
for larger grain sizes than small sized
cowpea grams.
Beef Consumer
Preferences in
Chile:
Importance of
Quality Attribute
Differentiators
on the Purchase
Decision
Pablo
Villalobos,
Carlos
Padilla,
Cristian
Ponce, and
Alvaro
Rojas
i) Evaluate importance of
quality attributes previously
described on the purchasing
behavior of beef consumers
in Chile.
ii) Determine implications
of consumer decisions on the
Chilean beef industry.
A conjoint analysis
technique was used to
analyze consumer
preference.
Conjoint Analysis: lowest price reached
highest utility. In country of origin Chile
attained highest utility because of its zoo
sanitary status. Quality assurance attribute
scored high utility level; it also ranked high
in terms of importance.
Comparing
consumer
preferences for
color and
nutritional
quality in maize:
Application of a
semi-double-
bound logistic
model on urban
consumers in
Kenya. 2008
Hugo De
Groote &
Simon
Chege
Kimenju
1. Describe and analyze
maize consumption patterns,
in particular which maize
products consumers
purchased and how they
prepared them.
2. Determine the willingness
to accept
Yellow maize, and compare
it to the willingness to pay
for maize with enriched
provitamin A carotenoid
content.
A contingent
valuation method was
used to evaluate
consumers'valuation
of yellow maize.
Consumers preferred white maize but were
willing to purchase yellow maize at a
discount of 37%.
Income doesn't influence preference for
color but high education group prefer white
maize.
Interest in fortified maize was registered
across the strata with an increase of interest
with increase in income. results indicate that
consumer preferences for the traits under
study are influenced by consumers'
socioeconomic and cultural background, in
particular
income, education, gender, and ethnic
group.
Collaborative H.S. 1. To find and characterize Appraisal technique The red local sorghums and the improved
Project To Laswai,N. relative preferences for was used for the rural varieties were less palatable but could be
Investigate B. Shayo various cereals among area while in the used for brewing.
Consumer andSTP different groups of urban areas there was Despite the improved varieties having
Preferences For Kundi consumers. direct involvement of desirable characteristics i.e. large seeds,
Selected respondents. white color, they still had negative attributes
Sorghum And Laboratory analysis e.g. Less palatability, low keeping,
Millet Products was done on sorghum susceptible to pest attack, high dehulling
In The Sadcc to establish nutrient losses.
Region Of Africa content.
Consumer Benjamin 1. Identification and ranking Target Factors that Kendall's ranking for preference was; Taste,
Preferences for Tetteh ofthe factors influencing identify consumer cooking quality, cooking time, aroma, price,
Rice Quality Anang et consumer preference for preference were impurities, source. Concordance analysis
Characteristics aI., quality characteristics of rice identified & ranked gave a 62% agreement of preferred
and the Effects in the Tamale metropolis. using Kendall's characteristics.
on Price in the 2.Quantification of the coefficient of Consumer preference for quality
Tamale effects of quality concordance: a characteristics effected price; more
Metropolis, characteristics of rice on statistical procedure premium was paid for rice with better
Northern Region, pnce Hedonic model was aroma. Sticky cooked rice and long duration
Ghana. 2011 used to determine the of cooking attracted low premium.
effect of quality
characteristics on
-------
pnce.
Consumer O.E. Evaluate the 200 banana Taste, size and/or number of fingers of the
Preference of Ayinde, Consumer preference of consumers were banana fruits were considered the most
Banana (Musa M.a. banana in Kwara State, interviewed using important attributes by the consumers, while
spp.) in Kwara Adewumi Nigeria. structured Appearance, color and shelf-life were
State, Nigeria and W.O. questionnaire. Data considered less important quality
Folorunsho analysis methods parameters.
were descriptive
analysis, ranking
method & Least
Significance
Difference.
Quality parameters.
Consumer Langyintu 1. Evaluate the relative Hedonic pricing The estimated regression results indicate
preferences for et al., importance of various model was used. that seasonality, grain size, color and insect
cowpea in cowpea characteristics damage level explain
Cameroon and determining cowpea prices in 93 and 97% of price variability in Ghana
Ghana. 2002 Ghana and and Cameroon, respectively.
Cameroon.
2. Compare these
characteristics
across markets in the two
countries.
Consumer Callie l.Evaluate and determine A conjoint model was most preferred attribute was color whole
Preference for Bryan consumer preferences for used to analyze watermelon price, seed content & Lycopene
Watermelons: A Evans certain consumer preferences sticker attribute.
Conjoint watermelon attributes, for various
Analysis. 2008 2. Determine if watermelon watermelon traits:
consumers who were flesh color, form,
surveyed prefer watermelons seed content,
for specific attributes or lycopene sticker
combinations of attributes, . .possession, pnce.
Consumer Fulgence J. Estimate the Hedonic price Seemingly Unrelated Regression (SUR)
Preferences As Mishili, premiums and discounts estimation technique results: In Buruguni Mkt in Dar consumers
Drivers Of The Anna A. negotiated by consumers for to measure strength discount damaged beans. Results for
Common Bean Temu, Joan various visual characteristics of consumer Morogoro mkt show consumers discount for
Trade In
Tanzania: A
Marketing
Perspective.
2009.
Consumer
preferences For
Quality
Characteristics
Along The
Cowpea Value
Chain In Nigeria,
Ghana And Mali
Consumer
Preferences For
Table
Cassava
Characteristics In
Pernambuco,
Brazil. 2009
Fulton and
Lowenberg
-DeBoer
Fulgence
Joseph
Mishili., et
al
Carolina
Gonzalez
& Nancy
Johnson
of beans in traditional open
markets in Tanzania.
1. Measure the preferences
of urban cowpea consumers
in selected West African
cities.
Determine the cowpea grain
quality characteristics that
command a price premium or
provoke a discount in
Ghanaian, Malian and
Nigerian markets.
1. Analyze the demand for
different cassava attributes.
2. Estimate the value
consumers give to implicit
attributes of cassava.
preferences in the
bean mkt.
Data was collected
from 6 mkts in
Ghana, 2 Mkts in
Mali, 3mkts in
Nigeria. Where
cowpeas were
purchased monthly
Hedonic Pricing
Method provided
statistical estimates of
premiums and
discounts.
Hedonic Price
method was used to
estimate value
consumers give to
attributes of cassava.
mixed beans. In the two regions farmers
discount for Red Canadian Wonder and
yellow compared to Soya Kablanketi. In
different mkts consumers were willing to
pay for mixed, natural shininess. Results for
bean size influence on prices were not
significant in all mkts while in one mkt
bruchid hole attracted a discount of2.3% of
average price per kilo.
In Ghana Consumers paid a premium for
ever increase of grain size, of 1-4.3%, Mali
they paid 1-1.3% and 1.2-1.4% in Nigeria.
Black eyed cowpea resulted in price
discount of7.1-13.2% in Ghana and Mali
mkts.
The results indicate that consumers in the
three countries are willing to pay a premium
for large grain size. They discount grains
with storage damage. Impact of price on the
cowpea quality characteristics e.g. skin
color & texture, eye color vary locally.
Ease of peeling (29%), cooking time (28%),
texture (16%) then color (11%). Price had
the lowest rank.
98
Appendix 5: Survey questionnaire
Section A
Site Description
Enumerator Date of interview-------------------------- ----------------
Village Town _
Division District, '---- _
Place of Data collection: 1. Home 2. Workplace. 3. Eatery
Respondents Socio-Economic Profile
Name of Gender Age Level of Business Occupation Income
Respondent Education ownership (Consumers) ina
(Traders) month
l=male 1=18-27 1. Primary 1. Self I=Regular
2=Female 2=28-37 2. Secondary 2. My Employment
3=38-47 3. College spouse. 2=Trader
4=48 and 4. Informal 3. Both of us 3=Housewife
above skills 4. My 4=Manual labor
5. None parents. 5=Student
6=Unemployed
7=Others
(Specify)
Codes
Section B: Bean Marketing and Trading (This section should be answered by
traders only).
1. For how long have you been in this business of selling beans? (in years or months)
2. Which is the source of your beans?
1. Local farmers 2. Farmers from other districts, 3. Local and
farmers from other districts 4. Other traders. 6. Other
traders and farmers. _
99
7.hnport 5.Fannersandlmport~. _
3. What makes you choose the above mentioned sources?
1. Quality of beans. 2. Availability . 3. Fair Price.
4. What attributes do you consider when procuring? (in order of priority 1-6)
A. Glossy B. Price C. Type D. Keeping E. F. Selling Others (please
appearance quality Origin quality explain)
5. Please list for me eight bean types that you sell including mixed and damaged.
1. 2. 3. 4. _
5. 6. 7. 8. _
6. Out of the 8 above, please rank 5 of the mostly purchased bean types by consumers.
Which types do you Type 1 Type 2 Type 3 Type 4 Type 5
purchase most (Please
list five in order of
priority)
Description (name)
Source: 1=Iocal
2=Import(name of
source)
3=Both
How much of each do
you purchase in a
month
what is the average
cost per bag
How much of each
type do you sell in a
month (bags)
what is the average
selling price per kilo
Post purchase Value addition
7. What are the challenges you experience after procuring beans? (Open ended)
1. Infestation of beans by bruchids 2. Rats. _
100
3. Bean rotting. 4. Others (Specify) _
8. What are the reasons for the above answers?
1. Bad bean quality. ___________ ,2. Unhygienic selling places.
3. Insufficient storage facilities. 4. Others. _
9. Do you sort mixed or damaged beans? 1. Yes 2. No. _
10. If yes above which of the sort and the unsorted do consumers most prefer?
1. Sort. 2. Unsorted. _
11. Is there any difference in prices of the two unsorted and sort? 1. Yes. __ 2. No.
12. Do you add any pesticide to your beans? 1. Yes 2. No _
13. If yes above, which of the dusted and the undusted do consumers most prefer?
1. Dusted 2. undusted. _
14. Is there any difference in prices between the two? 1. Yes. 2. No_
Target Market
15. Who do you sell beans to?
1. Direct consumers. 2. Other traders. _
3. Others (Livestock feed millers, Institutions etc.) 4. All three.
16. How would you rank the above customers in order of importance?
Customer Direct consumers Other traders Others
Rank
17. Where do the above mentioned customers live?
1. Within Thika districts. 2. Other districts. 3. Both. _
18. Are there differences in bean availability across the seasons? 1. Yes 2. No
19. When are beans mostly available?
101
1. May-June .2.January-February 3. Both _
20. How do prices compare during these periods and other periods?
1. High_2. Low. 3. No Change. _
21. In the two seasons mentioned above, which bean varieties are mostly available?
1. Imported varieties 2. Local varieties. 3. Both. _
22. Which varieties are available in the other seasons?
1. Imported varieties. 2. Local varieties. 3. Both. _
Section C: Consumer Behaviour (Direct consumers)
23. How do consumers behave when they come to your shop?
1. They demand for particular varieties. 2. They choose from what is
available.
Substitutes for beans
24. Please list for me the other pulses that consumers purchase and their prices
pulse Pigeon Peas Green Grams Cow Peas Dolicos Lablab
(Ndegu)
Price per kilo (Kshs)
25. What is their performance in terms of trading compared to common beans?
1. Isell more beans. 2. Isell more of other pulses than common beans.
3. There is no difference. ---
Section D: Consumer Preferences (This section is answered by consumers).
Consumer attitude towards common beans
1. Do you use beans when preparing meals at home? 1. Yes. 2. No_
102
2. How often do you cook beans in a week?
twice
1.0nce_. _2. Twice_.__ 3. More than
3. Which fuel do you use to boil beans? 1. Firewood 2. Charcoal_
3. Kerosene. 4. Gas. 5. Electricity _
4. How do you rate your source of fuel? 1. Easily Affordable _
2. Manageable. 3. Too expensive. _
5. Would you consider offering bean meals on an important occasion? 1. Yes_._2. No._
6. Ifno above what is the possible reason?
1. They are not good enough to give to visitors. 2. They would consider me
poor._
7. What would you say drives you to cook beans?
1. Nourishment. 2. Health benefits. 3. Both. _
8. Would you offer beans to a sick person? 1. Yes 2. No _
9. Are you aware of the health benefits of consuming beans? 1. Yes. 2. No
10. If yes above, where did you learn about health benefits of consuming beans?
1. Nutritionists_. _2. Fellow consumers. 3. Others (specify). _
Section E: Consumer behavior for utility maximization
11. Is identification of bean varieties important to you?
12. How many bean varieties do you know? Please name them
3. 4.__ 5. None. __
Sensory Evaluation
13. How important do you consider the following attributes?
(Pairwise Comparison: Please put initial of the important attribute in the clear cells).
1. Yes., ,2.No
1.__ 2.__
103
Attributes Color Grain Taste Flatulence Cookin Cooking Keepin Pric Grain
C size S B F g time I quality g eP mixM
Q quality
K
Color C
Grain size
S
Taste T
Flatulence
F
Cooking
time I
Cooking
quality Q
Keeping
quality K
Price P
Grain
mixes M
With me are some bean varieties, please rank their attributes using the scale below.
Attributes Ranks and descriptions
Grain color 5= Excellent 4= Good 3=Fair 2=Bad I=Very bad
Taste 5= Excellent 4= Good 3=Fair 2=Bad l=Very bad
Size 5= Excellent 4= Good 3=Fair 2=Bad l=Very bad
Low 5= Excellent 4= Good 3=Fair 2=Bad I=Very bad
flatulence
Price 5= Excellent 4= Good 3=Fair 2=Bad l=Very bad
Cooking time 5=Very 4=Acceptable 3=Fairly 2=Not 1=Unaccepta
Acceptability acceptable Acceptable acceptable ble at all
Cooking 5= Excellent 4= Good 3=Fair 2=Bad l=Very bad
quality
Grain mixes 1= not 2= do not like
Important
Food keeping 5= Excellent 4= Good 3=Fair 2=Bad l=Very bad
quality
104
14. Please use the above scale to rate the attributes in each sample.
Attributes Sample 1 Sample 2 Sample 3 Sample 4 Sample 5
Color
size
Price
Taste
Impurities & Grain damage
Cooking time
Mixed grains
Meal making suitability
Flatulence
Food keeping quality
15. How much would you consume and pay for each of the samples rated above in Kenya
shillings?
Variety Quantity Time Unit price Total
Amount in Kg l=day
2=week
3=month
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
16. Which of the following words best describe your preferences for the samples?
Code Description Sample 1 Sample 2 Sample 3 Sample 4 Sample 5
1 Very good
2 Good
3 Fair
4 Poor
5 Very poor
17. Which are the reasons for rating poor and very poor above (i) Poor .
(ii) Very poor .
105
18. If the samples were priced within your budget, how interested would you be in
purchasing them?
Code Description Sample 1 Sample 2 Sample 3 Sample 4 Sample 5
I Very Interested
2 Interested
3 Not interested
4 Undecided
19. Please name for me these samples
Samples Sample 1 Sample 2 Sample 3 Sample 4 Sample 5
Name
Substitutes of common beans
20. Please list for me how much you consume and pay for each of the following grains
Variety Quantity (kg) Frequency (per Month) Unit price Total
Pigeon peas
Green grams
Cow peas
Dolicos Lablab
THANK YOU FOR YOUR PARTICIPATION
Appendix 6: Pairwise comparison of bean attributes.
color Grain size taste flatulence Cook time Cook Keeping Price Grain mix
quality quality (GM)
Color (C) C(72.7) T(83.3) C(79.l) CT(77.9) Cq(84.6) KQ(84.1) P(99.5) C(50.7)
Grain size (GS) T(85.6) GS(7I.7) CT(88.9) CQ(96.2) KQ(97.l) P(65.6) GM(67.3)
Taste (T) T(81.6) CT(64.3) CQ(76) KQ(74) P(51.2) GM(52.5)
Flatulence (F) CT(74.5) CQ(79.1) KQ(79.2) P(80) GM(80.8)
Cooktime (CT) CQ(53.6) KQ(82.5) P(51.2) GM(51)
Cooking quality CQ(55.4) CQ(51.7) CQ(52.2)
(CQ)
Keeping quality KQ(64.4) KQ(52.7)
(KQ)
Price (P) P(55.8)
NB: Figures in parenthesis represent percentages.
107
Appendix 7: Attribute ranking according to variety
Attribute/ rank bad Fair good excellent
Color
KATX56 Gituru 0.8 8.9 33.3 56.9
GLP585 wairimu - 3 26.6 70.4
KATB9 Gacuma 9.4 9.4 28.1 53.1
GLP2 Rosecoco Nyayo - 5.8 34.9 58.1
GLP24 Canadian w. - 22.2 44.4 27.8
Gituru 17.1 32.9 30 20
GLP94 Mwitemania 7.1 17.9 42.9 32.1
KATl Kathika, Kayellow - - 20 80
Gaciku
Grain size
KATX56 Gituru - 11.4 50.4 38.2
GLP585 wairimu -
KATB9 Gacuma - 19.4 54.8 25.8
GLP2 RosecocoNyayo - 5.8 38.8 55.8
GLP24 Canadian w. - 17.6 58.8 17.6
Gituru
GLP94 Mwitemania 7.2 10.7 50 32.1
KATl Kathika, Kayellow - 20 50 30
Gaciku
Price
KATX56 Gituru 4.9 44.7 34.1 16.3
GLP585 wairimu 2.4 24.9 43.8 29.0
KATB9 Gacuma 6.5 19.4 51.6 22.6
GLP2 RosecocoNyayo 4.7 45.3 39.5 10.5
GLP24 Canadian w. 5.9 52.9 29.4 5.9
Gituru 8.5 40 40 11.4
GLP94 Mwitemania 17.9 32.1 35.7 14.3
KAT1 Kathika, Kayellow 10 40 50 -
Gaciku
Taste
KATX56 Gituru - 9.8 48 42.3
GLP585 wairimu 0.6 1.2 26.6 71.6
KATB9 Gacuma - 16.1 35.5 48.4
GLP2 RosecocoNyayo 1.2 5.8 44.2 47.7
GLP24 Canadian w. - 50 31.3 18.8
Gituru 7.2 32.9 35.7 24.3
GLP94 Mwitemania - 10.7 32.1 57.1
KATl Kathika, Kayellow - - 50 50
Gaciku
108
Cooking quality
KATX56 Gituru 2.4 13.7 56.5 27.4
GLP585 wairimu 0.6 3.6 33.7 62.1
KATB9 Gacuma - 12.5 56.3 31.3
GLP2 RosecocoNyayo 2.3 14.9 62.1 20.7
GLP24 Canadian w. - 37.5 50 12.5
Gituru 10 28.6 48.6 12.9
GLP94 Mwitemania 7.1 7.1 42.9 42.9
KAT1 Kathika, Kayellow 10 - 80 10
Gaciku
Flatulence
KATX56 Gituru 10.6 33.3 31.7 23.6
GLP585 wairimu 5.3 24.9 42 27.8
KATB9 Gacuma 12.9 12.9 38.7 35.5
GLP2 RosecocoNyayo 10.5 51.2 23.3 15.1
GLP24 Canadian w. 17.6 58.8 5.9 11.8
Gituru 12.8 41.4 35.7 10
GLP94 Mwitemania 3.6 25 10.7 60.7
KAT1 Kathika, Kayellow - 20 70 10
Gaciku
Keeping quality
KATX56 Gituru 4.9 11.4 44.7 39
GLP585 wairimu - 4.7 44.4 50.9
KATB9 Gacuma 3.2 25.8 35.5 35.5
GLP2 RosecocoNyayo 6.9 4.6 59.8 28.7
GLP24 Canadian w. - 37.5 50 12.5
Gituru 25.7 24.3 34.3 15.7
GLP94 Mwitemania - 17.9 21.4 60.7
KAT1 Kathika, Kayellow - 10 90 -
Gaciku
Cooktime rank / bean unacceptable Fairly acceptable Very acceptable
variety acceptable
KATX56 Gituru 0.8 24.2 53.2 21.8
GLP585 wairimu 1.2 3.6 50.3 45
KATB9 Gacuma 9.4 18.8 40.6 31.3
GLP2 RosecocoNyayo 2.2 14.8 63.6 18.2
GLP24 Canadian w. - 50 43.8 6.3
Gituru 25.7 28.6 34.3 11.4
GLP94 Mwitemania - 14.3 46.4 39.3
KATl Kathika, Kayellow - 20 20 60
Gaciku
109
Appendix 8: Evaluation of common bean attributes for consumer preference
Attributes Bean varieties
KATX56 KAT GLP GLP GLP GLP KAT
B9 2 24 585 92 Bl
Cooking Rank 4.08 4.19 4.02 3.75 4.57 3.61 4.21
quality Chf 132 9.30 71.58 3.50 169.72 49 14.29
Sig 0.000 0.01 0.000 0.174 0.000 0.000 0.003
Keeping Rank 4.17 4.03 4.1 3.75 4.46 3.29 4.43
quality Chi2 102 8.60 66.47 3.50 63.28 11.36 9.6
Sig 0.000 0.035 0.000 0,174 0.000 0.023 0.009
Price Rank 3.62 3.9 3.56 3.37 3.99 3.53 3.43
Chj2 47 13.5 43.02 11 59.57 48.43 10.21
Sig 0.000 0.004 0.000 0.012 0.000 0.000 0.037
Taste Rank 4.33 4.32 4.73 3.69 4.69 3.74 4.46
Chi2 31.37 4.90 97.26 2.38 225.58 34 9.07
Sig 0.000 0.086 0.000 0.305 0.000 0.000 0.011
Cooking Rank 3.95 3.94 3.99 3.56 4.39 3.26 4.25
time Chi2 69 7.30 119.70 5.38 139.66 19.43 4.78
Sig 0.000 0.064 0.000 0.068 0.000 0.001 0.091
Color Rank 4.46 4.25 4.5 4.06 4.67 3.49 3.93
Chi2 95 16.5 73.35 1.53 118.77 19.71 8.29
Sig 0.000 0.001 0.000 0.465 0.000 0.001 0.04
Grain size Rank 4.27 4.06 4.50 4.00 4.25 3.84 4.04
Chi2 29 7.00 33.23 6.16 80.54 90.14 23.43
Sig 0.000 0.036 0.000 0.047 0.000 0.000 0.000
Flatulence Rank 3.68 3.97 3.41 3.12 3.91 3.41 4.29
Chi2 50 7.30 63.88 12.50 99.25 41.94 21.71
Sig 0.000 0.062 0.000 0.006 0.000 0.000 0.000
Appendix 9: Consumers source of nutrition information of common beans
Source bean nutrition
information
Number of respondents Percentage of "yes" answer
Fellow consumers
Nutritionists
Radio
Television
Newspapers
110
111
37
26
22
96.4
95.5
86.5
61.5
45.5
110
Appendix 10: Breusche-Pagan heteroscedasticity test results
Varieties Variable Color Grain Price Taste Cooking Cooking Flatulenc Keeping
Size Time Quality Quality
KATX56 Chi2 1.570 1.540 1.530 1.530 1.030 1.030 1.650 1.530
Gituru P-value 0.211 0.215 0.216 0.217 0.310 0.310 0.199 0.215
KATB9 Chi2 0.080 0.890 0.680 0.840 2.830 0.680 4.050 0.010
Gacuma P-value 0.779 0.345 0.411 0.359 0.093 0.409 0.044 0.935
GLP2 Che 0.080 0.890 0.680 0.840 0.830 0.680 4.050 0.010
Rosecoco P-value 0.779 0.345 0.411 0.359 0.093 0.409 0.044 0.935
GLP24 Chi2 2.510 0.540 0.540 0.540 0.550 0.540 0.540 0.540
Canadian
Wonder P-value 0.107 0.461 0.460 0.459 0.464 0.462 0.461 0.462
GLP 585 Chi2 0.480 0.450 0.510 0.490 0.500 0.490 0.490 0.510
Wairimu P-value 0.489 0.479 0.474 0.484 0.479 0.483 0.485 0.475
GLP92 Chi2 0.830 0.180 0.710 1.960 3.960 2.550 5.960 0.110
Mwitemani
a P-value 0.362 0.671 0.400 0.161 0.046 0.110 0.015 0.742
KATBI Chi2 0.020 0.030 0.650 0.010 0.420 0.090 2.840 0.130
Kathika P-value 0.877 0.870 0.419 0.936 0.518 0.768 0.092 0.722
1. Regression results for KAT X 56 Gituru
Unstandardized Coefficients Standardized
Coefficients
Variable B Std. Error Beta Sig.
(Constant) 28.058 12.873 2.180 .031
color -.186 .141 -.140 -1.321 .189
grain size .371 .144 .257 2.584 .011
price .113 .101 .097 1.110 .270
taste .113 .151 .079 .748 .456
cooking time -.008 .121 -.006 -.069 .945
cooking qlty -.008 .122 -.007 -.067 .947
flatulence -.277 .094 -.280 -2.935 .004
keeping qlty -.100 .098 -.091 -1.014 .313
R2= 0.245
F=4.336
Appendix 11: Regression results for seven bean varieties
111
2. Regression results for KAT B9 Gacuma
Unstandardized Standardized
Coefficients Coefficients
Variables B Std. Error Beta Sig.
(Constant) 50.986 17.356 3.412 .002
color -.691 .310 -.510 -2.226 .037
size .745 .410 .411 1.814 .083
price -.075 .262 -.050 -.285 .778
taste .568 .339 .345 1.675 .108
cooking time .210 .328 .161 .641 .528
cooking quality -.527 .347 -.280 -1.519 .143
flatulence .223 .244 .184 .915 .370
keeping quality -.033 .260 -.023 -.126 .901
R2=0.419
F=1.98!
3. Regression results for GLP 2 Rosecoco
Standardized
Unstandardized Coefficients Coefficients
Variable B Std. Error Beta Sig.
(Constant) 70.149 18.047 3.887 .000
color -.772 2.493 -.039 -.309 .758
size -2.490 2.708 -.115 -.920 .361
price 1.478 2.121 .083 .697 .488
taste -1.371 .486 -.336 -2.823 .006
cooking time -1.289 2.737 -.068 -.471 .639
cooking quality .174 2.757 .009 .063 .950
flatulence 2.209 1.839 .151 1.201 .234
food keeping quality -.068 1.933 -.004 -.035 .972
R2 = 0.126
F=1.282
112
4. Regression results for GLP 92 Mwitemania
Unstandardized Standardized
Coefficients Coefficients
Variables B Std. Error Beta t Sig.
(Constant) 51.302 10.093 5.083 .000
color .062 .215 .051 .287 .775
size -.482 .292 -.304 -1.651 .104
price .038 .232 .024 .162 .872
taste -.060 .239 -.045 -.249 .804
cooking time .228 .190 .192 1.198 .236
cooking quality .341 .265 .241 1.288 .203
flatulence -.309 .209 -.207 -1.481 .144
keeping quality -.118 .154 -.109 -.770 .445
R2 = 0.145
F=1.164
5. Regression results for GLP 24 Canadian Wonder
Unstandardized Standardized
Coefficients Coefficients
Model B Std. Error Beta t Sig.
(Constant) -40.106 12.803 4.151 .009
color -.325 .523 -.192 -.621 .562
size .041 .706 .021 .058 .956
price -.181 .468 -.102 -.387 .715
taste -.442 .313 -.275 -1.415 .216
cooking time -.118 1.145 -.057 -.104 .922
cooking qlty 1.281 1.600 .640 .801 .460
flatulence -.786 .852 -.529 -.922 .399
keeping qlty -2.935 1.819 -1.548 -1.613 .168
R2 = 0.302
F = 3.907
113
6. Regression results for GLP 585 Wairimu
Unstandardized Standardized
Coefficients Coefficients
Std.
Variables B Error Beta t Sig.
(Constant) 75.173 12.347 6.088 .000
color .048 .129 .036 .375 .708
size -.216 .074 -.249 -2.926 .004
price .173 .077 .193 2.228 .027
taste 0.325 .112 0.254 2.909 .004
cooking time .193 .103 .166 1.864 .064
cooking qlty .053 .099 .045 .538 .591
flatulence .001 .067 .001 .013 .990
keeping qlty .151 .110 .124 1.378 .170
R2=0.173
F = 3.790
7. Regression results for KAT Bl Katheka
Standardized
Unstandardized Coefficients Coefficients
Variables B Std. Error Beta t Sig.
(Constant) 7.566 2.858 2.648 .016
color .100 .486 .066 .207 .839
size -.487 .539 -.283 -.904 .377
price .549 .416 .342 1.320 .202
taste -.815 .694 -.341 -1.175 .255
cooking time .034 .599 .014 .057 .955
cooking quality .211 .385 .112 .549 .589
flatulence -.032 .425 -.019 -.075 .941
keeping quality -.522 .550 -.249 -.950 .354
R2 = 0.335
F= 1.827
114
Appendix 12: Variety ranking based on consumers dwelling place
Variety/Rank 1 2 3 4 5 6 7
KAT 56 Urban 7 44.7 60.9 50 66.7
Gituru Rural 93 55.3 39.1 50 33.3
GLP 585 Urban 75.2 37.89 20 0 0
Wairimu Rural 24.8 62.2 80 100 100
KATB9 Urban 25 10 20 66.7 50
Gacuma Rural 75 90 80 33.3 50 100
GLP2 Urban 75 85.3 84.6 0 0 0
Rosecoco Rural 25 14.7 15.4 100 100 100 100
GLP24 Urban 50 100 100 0 50 50
Canadian Rural 50 0 0 100 50 50
GLP92 Urban 0 6.7 26.9 16.7 62.5 62.5
Mwitemania Rural 100 93.3 73.1 83.3 37.5 37.5
KATBI Urban 33.3 16.7 71.4 0 . 0 0
Katheka Rural 66.7 83.3 28.6 100 100 100
Appendix 13: Bean consumption frequency by different age groups
Age groups once Twice thrice more than thrice
20-31 30 20 28 15
31-41 30 30 23 26
41-51 20 21 12 26
51-61 10 10 12 12
above 61 10 20 23 11
Total 100 100 100 100
Appendix 14: Consumers education level and consumption frequency
Education Bean consumption in a week
Level
Once Twice Thrice More than thrice
Primary 24.1 33.3 25.4 36.4
Secondary 44.8 30 39 36.4
College 24.1 23.3 17.5 23.6
Informal 6.9 8.3 11.1 3.6
skills
None 5 1.6
115
KATX56
Appendix 15: Correlation coefficients of attributes in bean varieties.
flatulencecolor grain
size
price taste cooking
time
cooking
quality
keeping
quality
color
grain size
price
taste
cooking
time
cooking
quality
flatulence
keeping
quality
correlation
Sig
Correlation
Sig
Correlation
Sig
Correlation
Sig
Correlation
Sig
Correlation
Sig
Correlation
Sig
Correlation
Sig
.411
000
.142
.117
.502
.000
.280
.002
.404
.000
.. 165
.068
.171
.058
.411
000
.194'
.032
.354"
.000
.222'
.014
.124
.173
.285"
.001
.139
.126
.142
.117
.194'
.032
.207'
.022
.129
.155
.025
.786
.159
.078
.072
.431
.502
.000
.354"
.000
.207'
.022
.127
.162
.406"
.000
.264"
.003
.257"
.004
.280
.002
.222'
.014
.129
.155
.127
.162
.263"
.003
.310"
.000
.269"
.003
.404
.000
.124
.173
.025
.786
.406"
.000
.263"
.003
.239"
.008
.265"
.003
.165
.068
.285"
.001
.159
.078
.264"
.003
.310"
.000
.239"
.008
.263"
.003
.171
.058
.139
.126
.072
.431
.257"
.004
.269"
.003
.265"
.003
.263"
.003
KATB9 color grain
size
price taste cooking cooking flatulence keeping
time quality quality
Color
grain size
price
taste
cooking
time
cooking
quality
flatulence
keeping
quality
correlation
Sig
Correlation .505"
Sig .004
Correlation -.178
Sig .338
Correlation .430'
Sig .016
Correlation .504"
Sig .004
Correlation .060
Sig .750
Correlation .481"
Sig .006
Correlation .112
Sig .548
.505"
.004
.070
.707
.285
.120
.633"
.000
.196
.291
.389'
.031
.220
.233
-.178
.338
.070
.707
.106
.572
-.012
.947
-.087
.641
.036
.849
.050
.788
.430'
.016
.285
.120
.106
.572
.424'
.018
.345
.057
.409'
.022
.187
.313
.504"
.004
.633"
.000
-.012
.947
.424'
.018
.178
.329
.449'
.011
.448'
.011
.060
.750
.196
.291
-.087
.641
.345
.057
.178
.329
.260
.157
.163
.380
.481"
.006
.389'
.031
.036
.849
.409'
.022
.449'
.011
.260
.157
.151
.417
.112
.548
.220
.233
.050
.788
.187
.313
.448'
.011
.163
.380
.151
.417
116
KATBI color grain price taste cooking cooking flatulence keeping
size time quality quality
color correlation .747** .457* .439* -.024 .133 .090 .598**
Sig .000 .014 .019 .902 .498 .649 .001
grain size Correlation .747** .506** .530** .096 .035 .186 .515**
Sig .000 .006 .004 .626 .861 .343 .005
price Correlation .457* .506** .074 .153 -.105 -.126 .084
Sig .014 .006 .709 .436 .594 .523 .670
taste Correlation .439* .530** .074 .439* .379* .509** .435*
Sig .019 .004 .709 .020 .047 .006 .021
cooking Correlation -.024 .096 .153 .439* .030 .542** .134time
Sig .902 .626 .436 .020 .879 .003 .497
cooking Correlation .133 .035 -.105 .379* .030 .142 -.031quality
Sig .498 .861 .594 .047 .879 .470 .877
flatulence Correlation .090 .186 -.126 .509** .542** .142 .027
Sig .649 .343 .523 .006 .003 .470 .890
keeping Correlation .598** .515** .084 .435* .134 -.031 .027quality
Sig .001 .005 .670 .021 .497 .877 .890
GLP585 color grain price taste cooking cooking flatulence keeping
size time quality quality
color correlation .406** -.033 .357** .318** .276** .203** .409**
Sig .000 .672 .000 .000 .000 .008 .000
grain size Correlation .406** .038 .288** .199** .136 .015 .197*
Sig .000 .622 .000 .009 .077 .851 .010
price Correlation -.033 .038 .236** .391** -.056 -.143 -.070
Sig .672 .622 .002 .000 .473 .064 .364
taste Correlation .357** .288** .236** .271** .169* .171* .281**
Sig .000 .000 .002 .000 .028 .026 .000
cooking Correlation .318** .199** .391** .271** .067 -.012 .124time
Sig .000 .009 .000 .000 .389 .874 .110
cooking Correlation .276** .136 -.056 .169* .067 .187* .413**quality
Sig .000 .077 .473 .028 .389 .015 .000
flatulence Correlation .203** .015 -.143 .171* -.012 .187* .272**
Sig .008 .851 .064 .026 .874 .015 .000
keeping Correlation .409** .197* -.070 .281** .124 .413** .272**quality
Sig .000 .010 .364 .000 .110 .000 .000
117
GLP92 color grain price taste cooking cooking flatulence keeping
size time quality quality
color correlation .578- .142 .312- .298' .473- -.032 .075
Sig .000 .240 .009 .012 .000 .794 .537
grain size Correlation .578- 1 .255' .360- .098 .541- .179 .035
Sig .000 .033 .002 .419 .000 .139 .775
price Correlation .142 .255' 1 .466- .165 .383- .111 .013
Sig .240 .033 .000 .171 .001 .361 .917
taste Correlation .312- .360- .466- .406- .593- .178 .017
Sig .009 .002 .000 .000 .000 .141 .887
cooking Correlation .298' .098 .165 .406- 1 .205 .298' .316-time
Sig .012 .419 .171 .000 .088 .012 .008
cooking Correlation .473- .541- .383- .593- .205 1 .186 -.078quality
Sig .000 .000 .001 .000 .088 .123 .520
flatulence Correlation -.032 .179 .111 .178 .298' .186 1 .141
Sig .794 .139 .361 .141 .012 .123 .244
keeping Correlation .075 .035 .013 .017 .316- -.078 .141 1quality
Sig .537 .775 .917 .887 .008 .520 .244
GLP24 color grain price taste cooking cooking flatulence keeping
size time quality quality
color correlation .000 -.508' -.115 .145 .134 -.413 .267
Sig 1.000 .045 .671 .592 .622 .112 .317
grain size Correlation .000 .000 .133 -.503' .309 .119 .154
Sig 1.000 1.000 .624 .047 .245 .660 .568
price Correlation -.508' .000 .336 -.203 -.475 .236 -.475
Sig .045 1.000 .203 .452 .063 .379 .063
taste Correlation -.115 .133 .336 -.025 -.154 .154 -.277
Sig .671 .624 .203 .927 .570 .568 .299
cooking Correlation .145 -.503' -.203 -.025 .194 .464 .194time
Sig .592 .047 .452 .927 .472 .070 .472
cooking Correlation .134 .309 -.475 -.154 .194 .055 .857"quality
Sig .622 .245 .063 .570 .472 .839 .000
flatulence Correlation -.413 .119 .236 .154 .464 .055 -.165
Sig .112 .660 .379 .568 .070 .839 .540
keeping Correlation .267 .154 -.475 -.277 .194 .857" -.165quality
118
Sig .317 .568 .063 .299 .472 .000 .540
GLP2 color grain price taste cooking cooking flatulence keeping
size time quality quality
color correlation .335" .095 -.154 .345" .248' .039 .069
Sig .002 .384 .156 .001 .021 .719 .530
grain size. Correlation .335" .130 -.211 .125 -.014 -.236' .125
Sig .002 .234 .051 .250 .896 .029 .252
price Correlation .095 .130 -.101 .288" .057 .051 .070
Sig .384 .234 .355 .007 .603 .641 .523
taste Correlation -.154 -.211 -.101 -.117 -.136 .206 -.006
Sig .156 .051 .355 .282 .211 .057 .953
cooking Correlation .345" .125 .288" -.117 .520" .174 .004time
Sig .001 .250 .007 .282 .000 .108 .968
cooking Correlation .248' -.014 .057 -.136 .520" .243' .131quality
Sig .021 .896 .603 .211 .000 .024 .227
flatulence Correlation .039 -.236' .051 .206 .174 .243' -.088
Sig .719 .029 .641 .057 .108 .024 .422
keeping Correlation .069 .125 .070 -.006 .004 .131 -.088quality
Sig .530 .252 .523 .953 .968 .227 .422