In a readership survey conducted in London, one interview in every three had been tape recorded. Through follow-up intensive interviews, conducted the next day, estimates were developed of the accuracy of certain of the information collected in the first interview. This situation provided an opportunity to compare: a. the estimated accuracy of information collected in the readership survey when a tape recorder was used with b. the estimated accuracy of such information when a tape recorder was not used. The recorded and the non-recorded samples were closely matched, on an empirical basis, in order to increase the meaningfulness of the comparison of (estimated) accuracy with and without the recorder being used. The comparison made here is of 189 recorded estimates (from 105 people) and of 430 non-recorded estimates (from 226 people). The small numbers involved make this a small-scale study, so that its results must be regarded as indicative and not final.
The market researcher depends, for much of the information he needs, upon the respondent's memory of his/her own behaviour, particularly buying behaviour and behaviour relation to the mass media. Amongst other things, the market researcher may need to know; when specific kinds of buying occurred; the frequency of such buying, what brand was bought, the quantity bought. He may want to know about the respondent's recent exposure to press, television, radio, cinema, posters and possibly certain things about the nature of that exposure. These and other of the things he may want to know depend in part upon the respondent drawing upon memory - and doing so, moreover, under the conditions of the survey interview.
The testing of groups of people, under rigorously controlled condition, has considerable potential as a tool of market research, In saying this, I am not referring to the use of small groups (of 6-8 people) for discussion purposes. What I am writing about, however, is a very different use of groups. It has proved-a relatively simple matter to bring ordinary members of the public together in groups of about 40 at a time and to test them on an almost unlimited range of topics - tests of knowledge, assessments of attitude and studies of reactions to carefully prepared stimuli. Taken forty at a time, group numbers can be aggregated to give as many cases as are required. Provided certain serious sampling difficulties can be overcome - and I shall be speaking about this at some length - group work of this kind can be used to provide a valuable supplement to the information gathered through the ordinary survey interview. I am thinking particularly of Information derived through the kinds of test I have just mentioned: tests of knowledge and of ability? assessments of certain attitudes, comparative studies of-reaction under controlled conditions. As a technique, it is relatively inexpensive and has the special advantage of offering quick results. A further advantage of the method is that a full record can be kept of precisely what was said by the test administrator and of precisely what processes the respondents, were put through.
This paper describes a technique for the matching of population samples. The matching is achieved through prediction and the method turns-upon the combining of empirically developed predictors to give the best available predictive (or matching) composite. The development of this composite is through the principle used in-simple biological classification, which is to be sharply distinguished from the principle inherent in multiple correlation. The mathematical bases of the method are very simple and appear to be superior to the usual maximising formulas. Empirical matching is to be contrasted with matching by custom or hunch.
This paper describes a technique for the matching of population samples. The matching is achieved through prediction and the method turns-upon the combining of empirically developed predictors to give the best available predictive (or matching) composite. The development of this composite is through the principle used in-simple biological classification, which is to be sharply distinguished from the principle inherent in multiple correlation. The mathematical bases of the method are very simple and appear to be superior to the usual maximising formulas. Empirical matching is to be contrasted with matching by custom or hunch.