1. Unobtrusive Measures In Data Collection.
- Author
-
Sechrest, Lee
- Subjects
INFORMATION science ,RESEARCH ,HYPOTHESIS ,QUESTIONNAIRES ,MANAGEMENT ,INTERVIEWING ,DECISION making ,AUTOMATIC data collection systems ,ERRORS ,INTERVIEWING in marketing research - Abstract
It is important that we recognize that all our measures are imperfect and, hence that their use involves error. An error will be as serious in its implications if made on the basis of one measure as on the basis of another. Errors of prediction arc errors, and no greater comfort is to be taken in the making of one kind of error than another. The measures to be discussed in this paper, unobtrusive and nonreactive measures, are imperfect, but their degree of validity can ordinarily be known, and to the degree that they are valid, they are as good as any other measures. For example, if birth order correlates .30 with academic achievement, then it is as good a basis for making a decision as a test that correlates .30. However, there is a tendency to disparage certain kinds of measures on the ground that they are far from perfect, without applying the same stringent standards to more conventional tests It must be reiterated that the approach taken here does not represent a rejection of widely accepted and useful techniques of interviewing and testing. They have shown their value. But they have also had their weaknesses revealed, and it is the purpose of the foregoing presentation to recommend ways of supplementing interviews and questionnaires so that their shortcomings are counterbalanced. When sources of error are complementary rather than overlapping, great strength in measurement is achieved. [ABSTRACT FROM AUTHOR]
- Published
- 1971
- Full Text
- View/download PDF