Shorter tests through the adaptive use of planned missing data in sampling designs
When conducting a large-scale survey or panel stude, researchers must balance topical breadth and depth with factors influencing respondent behavoir, such as motivation, compliance, concentration, and so forth. Lengthy test are assumed to result in increased nonresponse and repondent burden, modified answering behavoir and eventually, increased measurement error. The first empirical investigation in this thesis explored changes in subject answering behavior that are related to test length and the relative position of items within a 485-items questionnaire. Apart from altered answering behavior, concentration, and motivation of the subjects towards the end of the test, it could be shown that the measured traits changed as a function of relative positioning in a test. Seriour doubts arise concerning whether the same charateristic in measured at the beginning and end of a long survey, even if the questions are indentical.