Effects of measurement and response errors on estimation based on centre sampling and two-phase centre sampling for elusive populations

Abstract

Accurate survey data is important for planning and decision making. The presence of measurement and response errors in survey data can result in biased estimates as well as estimates with large standard errors, affecting the power of the statistical test to be carried out. It is important to investigate the effects of response errors when computing survey data so as to obtain reliable information for use by statisticians and policy makers. This paper investigates the effects of measurement and response errors in elusive surveys where the population has capture problems. The technique of centre sampling (CS) which entails collecting data on especially migrant populations is studied. In the technique, observations are made on those subjects who visit specified centres within the survey time period. This paper also investigates the technique of two-phase Centre Sampling (CS). A simulation study is carried out to establish the effect of response and measurement errors in centre sampling (CS) and two-phase Centre Sampling (CS). Using profiles at different levels of multiplicities, it was found that the errors affect the statistical estimates, even in small proportions to true values resulting in large estimation errors in both the mean and the variance. It is therefore recommended that both measurement and response errors be put into consideration when designing and carrying out a survey for more accurate results.