Question order and context effects on multi-items scales questions

Dissertação apresentada como requisito parcial para obtenção do grau de Mestre em Estatística e Gestão de Informação

This study examines question order effects, in particular context effect, on multiple-item scales questions and the impact that these effects can have in validity and reliability of measurement. For this purpose, it was performed a survey research with two versions of ECSI questionnaire, in which the position of the indicators of two constructs was changed. In the two groups of items studied, only one of them presented differences when they are grouped or separated in the questionnaire. Although even in this group of questions the effect is diminished, the results suggest that context effects are not necessarily negative for marketing research, since grouping related questions together can result in a higher correlation between them and, consequently, lead to a decrease of variation and an increase of reliability. However, the emergence of context effect seems to be closely related with question content and the level of similarity between the items. Future research should examine the generalization of these results across other service industries and with other question content.