| Abstract: | Recently developed cognitive theories explain why some questionnaire items can be answered in a reliable and valid manner and others cannot. Those theories distinguish two components: (a) the interpretation or understanding of a question and (b) the role of memory. The present study investigates the ability of these two components to forecast the test-retest association coefficients of 207 pilot questionnaire items used in an international study of reading literacy in which 2 populations were involved: Grade 5 and Grade 2 of primary and secondary education, respectively. The analysis of the data showed that for both populations, both components forecast the relative sizes of the test-retest correlation coefficients. The results strongly suggest that if one wishes to use questionnaire items in research about relationships, then the items should be as unambiguous as possible. Moreover, the information needed to formulate an answer must be easily accessible in the respondent's memory. |
