A Resampling Technique for Complex Survey Datal

SUMMARY In this article a new boot~trap technique of variance estimation for complex survey data known as "Rescaling Bootstrap Without Replacement" has been developed for some important sampling designs. The comparison of proposed bootstrap variance estimation technique has been done with different methods of bootstrap and jackknifed variance estimation through simulation. It has been observed that for estimation of variance of a non-linear statistic, the proposed rescaling bootstrap without replacement technique works well and its perfonnance is almost comparable with the jackknifed variance estimator for most of the non-linear statistics considered in this study.