[Application of a simple random sampling method on surveys at the community level].

ABSTRACT

OBJECTIVE: To study the feasibility of a simple random sampling on surveys at the community level and to evaluate the quality of samples under survey. METHODS: A simple random sample of households was taken, based on the electronic listings of community households from Gongshu and Xiacheng districts of Hangzhou city. One of the adults aged 18 to 64 years in the sampled households was identified with KISH method to finish a questionnaire survey. More than 500 people from the sample size was required in each district. RESULTS: Of 950 sampled households in Xiacheng district, 511 (53.8%) finished the survey while 506 (36.7%) out of the 1380 sampled households in Gongshu district did. The proportions of non-response due to the following reasons as none with eligible age in the households, relocation of the original household, mass relocation of the community, and errors in the household listings etc. were 38.3% and 43.5% respectively, in the two districts. Proportions attributed to non-response and refusal to response of sampled household or individual were 8.0% and 19.9% respectively. No statistical significant differences in age and gender were found between the surveyed samples and the population in the sampled households, or in gender between the populations in the sampled households and in Hangzhou city. However, the population in the sampled households showed a more aging population structure than the population in Hangzhou city. CONCLUSION: In a geographically limited area, using a simple random sampling method to do the survey is feasible, based on the electronic listings of household. Enough time spent during the household visit guarantees the interviewers to get a representative sample of the sampling frame. There is an urgent need for the timeliness, completeness and accuracy of electronic household listings to be improved.