Search

This study discusses context effect, that is, non-sampling errors that arise from the structure of the questionnaire. Although it is difficult to quantify these measurement errors, they are systematic, i.e. show the same regularities when using questionnaire items in the same structure. To the authors' knowledge, this is the first study whose focus is particularly on the effect of question order, specifically part-whole contrast effect on Hungarian data. A survey experiment was conducted to measure this effect: the order of questions related to the importance of certain elements of the country's democratic functioning, was changed on a randomly divided sample of a national survey. Overall, the results provide sufficient statistical evidence to support the claim that question order has an effect on the pattern of answers to certain questions. [ABSTRACT FROM AUTHOR]