Próba weryfikacji wiarygodności odpowiedzi na pytania wieloalternatywne
Sztabiński, Paweł B.
MetadataPokaż pełny rekord
Multiple-choice questions constitute a particular case of cafeteria questions; they are composed of: 1. the introductory statement specifying the fragment of reality to which should relate the alternatives; 2. the task specifying the number of choices to be made by a respondent; 3. the qualitative alternatives being the relevant answers. On the basis of methodological manuals the directives, for construction of multiple-choice questions are formulated as follows: 1. the scopes of particular alternatives should not overlap; 2. the list of alternatives should be exhaustive; 3. all alternatives should be formulated at the same generality level; 4. the alternatives should be presented in a certain order. However, all the problems connected with the construction and application of multiple-choice questions have not been, so far, solved satisfactorily. Doubts and reservations may arise in connection with some problems, as follows: 1. the list of alternatives may be a source of bias; 2. the task may be a source of bias; 3. the item “other - specify” may not fulfill its function; 4. the length of the list of alternatives may be excessive and the way of its presentation may cause difficulties. Results of an empirical research are presented whose purpose was to obtain a preliminary orientation concerning the following problems: 1. efficiency of the item “other” in the cafeteria; 2. reliability of answers; 3. efficiency of the request, addressed to respondents, asking them to familiarize themselves with the list of alternatives and to give thoughtful answers. Analysed were answers to two multiple-choice questions demonstrated to respondents with the use of a respondent card. The results obtained show that the inclusion, in cafeteria, of the item “other” is rather inefficient, the reliability (reproducibility) of the answers turned out to be very low. Inefficient was also the above-mentioned request addressed to respondents.