Sampling in the Kwun Tong industrial community research programme


Chan, Ying Keung

TitleSampling in the Kwun Tong industrial community research programme
PublisherSocial Research Center, the Chinese University of Hong Kong
Publication DateFebruary 1972
Keywords:Sampling (Statistics)
Social sciences -- Research

Kwun Tong

Abstract/ Concluding Remarks:In Kwun Tong, the population is now reaching 460,000. But the district is small in area - around 12 sq. km., and residences are rather concentrated; travelling within the district is not so time-consuming, and the cost is low. Thus we did not need t o apply the techniques of cluster sampling or multi-stage sampling for the purpose of reducing the travelling of fieldworkers; another reason is that these methods would make statistical analysis more complicated and yield larger bias. For Life Quality Survey and Health System Survey, large sized representative samples were required, as well as precision of sub-divisions of the population. Therefore stratified sampling method was most preferable and was actually employed since ancillary information was accessible.

We used housing type and location as criteria f or stratification because they were important variables for both studies. But for other surveys in other districts, it may not be the same. In fact, when a sample design is requested, the project investigator should provide the person responsible for sample designing information about the project, especially the purposes and variables considered as most important. The project investigator should notice also the amount of ancillary data which are currently available and can help the designer t o work out a best-fit sample frame.

In case supplement is required so as to match the desired sample size as has been done in Kwun Tong, new assignments to fieldworkers should be taken from the list prepared in advance, following the sequence and not exceeding the required number which can be determined by computation of reporting of the initial sample. This is very important when stratified sample is designed to be a self-weighted one.

Strict control methods cannot eliminate the possibility of cheating by fieldworkers; anyway, good control is necessary when processing our fieldwork. Supplement must be released carefully if they are required. Cases of refusal, not-at-home, or ineligible should be reported in detail as well as number of calls and interviewing time. These not only can re duce errors in the procedure of fieldwork but also can provide adequate information for estimating the effects of non-response and non-coverage in the sample.

Specific statistical problems have not been discussed in detail in the paper. Readers can refer to specialized literature on that field. Here we just intend to discuss problems in practising sample survey, particularly “sampling” in the Kwun Tong District, to explain how we explored alternatives to solve these problems, and hope to provide reference for future sample surveys in other communities which may or may not be similar to Kwun Tong.

NoteTypescript printing
Bibliography: leaf 21
Images Preview
Sampling in the Kwun Tong industrial community research programme (PDF)1.83 MB