Odongo, L. O.2014-05-302014-05-302005Journal of Statistics Vol. 1 (1): pp. 1-81117-1421http://ir-library.ku.ac.ke/handle/123456789/9716doi.org/10.4314/eajosta.v1i1.39150In this article an application of a kernel based nonparametric approach in constructing a large sample nonparametric confidence interval for a shift parameter is considered. The method is illustrated using the Cauchy distribution as a location model. The kernel-based method is found to have a shorter interval for the shift parameter between two Cauchy distributions than the one based on the Mann-Whitney test statistic.enBest Asymptotic NormalCauchy distributionKernel estimatesMann-Whitney test statisticNonparametric confidence intervalShift parameter.Article