Suchergebnisse

The hold of sample size on the thinking of the consumers & manufacturers of survey res (SuRes) reflects partly the sympathetic magic of 'bigness', & partly the reality which has become badly distorted in the process of transmission that: other things being equal, a large sample is better than a small one. The distortion comes from ignoring the qualification 'other things being equal'. Despite the benefits which can result from use of large samples, the emphasis on sample size to the exclusion of other considerations represents a threat to SuRes & market res. The pressure of competition has led to a need to reduce SURes costs, desirable where it is achieved by increased efficiency but it is easier to play the 'numbers game' of 'sample size ueber alles' & achieve the appearance of maintaining quality without the reality by use of large but cheap & inefficient samples. Clients & practitioners must be educated to the idea that sampling error rather than sample size is the criterion of sample adequacy. The controversy about quota vs 'probability' sampling continues due to the psychol that the important thing was sample size, not the way in which the sample was selected. Even the concern with non-response rates is a fairly recent development & earlier thinking in the SuRes field was concerned only with how many cases you finally got & not with how many R's 'got away' in the course of securing the sample. Methods exist which give the appearance of reducing costs without loss of quality, & the knowledge for detecting the fallacy is not commonly available. Thus, it is easy to misuse the Polits-Simmons Method to give the client the same size sample at a lower cost without the client being aware that the same size sample does not imply the same quality of sample. As long as clients continue to judge samples solely in terms of size, commercial org's are faced with a choice between sacrificing quality or going out of business. Sampling error, rather than sample size, must be emphasized in the selling of survey res. Unless preventative action is taken, it is predicted that there will exist in the SuRes field an analogy to Gresham's Law & 'bad res will drive out good.' Modified AA.

This study was conducted to determine the speed and accuracy of form recognition as a function of the size of target forms for various amounts of detail resolution. The stimulus forms were generated by filling in, on a statistical basis, some of the cells of a 90,000-cell matrix. The subjects were shown a "briefing target" and instructed to locate that target on a display containing numerous other forms. The significant finding is that both search time and errors remain invariant until the visual angle subtense of the targets falls below 12 min; at values below 12 min performance deteriorates. This effect is independent of the range of resolutions investigated. The implications of these findings to equipment design are discussed.