Stratified Cluster Sampling And Its Applica-Ation On The Improved RRT Model For College Students’ Cheating On Examination

Statistical methods for sample surveys of sensitive issues has been attentioned by many experts at home and abroad, From the survey of simple random sampling, stratified simple random sampling, stratified cluster sampling, multistage sampling to a larger sample size have achieved certain results.This paper improves the RRT method on the sensitive issue of statistical sampling survey and design the questiongnaire with the improved RRT model of stratifiedcluster sampling method.In2012September to2013June,using stratified cluster sampling,We select20clusters of undergraduates and12clusters of college students of Gansu Normal University for Nationalities of total1639students to interview the students cheating and carries on the parameter estimation, derived variance estimation of confidence intervals,variance characteristics and other digital features.The final conclusion is as follows:the rate of20clusters of undergraduates,12clusters of college students, and32clusters cheating on exams incidence is15.82%,24.76%,18.349%.The estimation variance is0.00026,0.00018,0.00015.The estimated value of the confidence interval is [0.126595,0.189805],[0.221305,0.273895],[0.159486,0.207494] of95%of Gansu Normal University for Nationalities.And the same survey (class) and the same survey indicator (cheating),come to the conclusion that the correlationship coefficients of twice survey datas is0.90397(12clusters of college students),0.94048(20clusters of undergraduates),0.94808(32clusters). They are greater than0.9and the P value is less than0.01.It indicates that the retest reliability is higher.so the paper shows that the methods of investigation and formula are reliable, effective and scientific. It will provide a reliable basis for Gansu Normal University for Nationalities preventing cheating in exams.