ABSTRACT
Most if not all measures collected in survey research contain some amount of measurement error. Psychometric theory (e.g., Nunnally & Bernstein, 1994; Traub, 1994) makes clear that especially when only a few questions are used to measure a specific underlying concept, that concept will be measured with considerable measurement error. For that reason, psychological tests tend to use scales that consist of a large number if items (30-60 items are not uncommon in psychological measurement), in order to attain a sufficient reliability for the combined scale score. In contrast, the measures used in surveys are typically short, because there is limited room in the questionnaire. Faced with the choice between including fewer topics in the questionnaire, while having more questions for each topic, versus having more topics and fewer questions, most survey researchers opt for the latter (Heath & Martin, 1997).
