5. The problems of samples and questionnaires
1. Introduction
The subject of samples and questionnaires is rather general because the subject matter is vast and the number of potential situations is infinite. This lack of mathematical sophistication together with the generality sometimes inculcates the feeling that the subject is a bit 'waffly' and not very important.
This is a big mistake sets do not just happen - they are the result of decisions to collect particular items of information (and not to collect others), in a particular way, over a particular time period. These decisions are absolutely vital to the success or failure of any statistical analysis since, as is perhaps obvious, the validity of such analysis will depend on the kind, quality and quantity of data available.
So far we have tended to assume that the data to be analyzed are available to us as if by magic. This assumption seems to be common in statistics texts: indeed many of them, at least in the social sciences, do not even consider problems of data collection, concentrating instead on data analysis. In fact, however, deciding what to collect, collecting, compiling and verifying the data set are probably the most important aspects of any statistical study.
They are almost always the most time-consuming elements of a study because the availability and power of relatively inexpensive micro-computers mean that much of the data analysis can be undertaken extremely rapidly. If the basic data set is full of errors or is simply inappropriate to the question being studied, the analysis based upon it will be worthless and may lead to e ...