Abstract
<p>Researchers often seek to measure affective experiences in daily life. Consequently, measures of positive and negative affect frequently take up large parts of ambulatory questionnaires. Optimizing questionnaire content and length is crucial in intensive ambulatory studies. In the current study, we developed and tested approaches to shortening ambulatory affect questionnaires. Specifically, we sought to build a preliminary item bank of 14 positive and 20 negative affect items calibrated with a graded response model using data from 14 individual ambulatory assessment studies (N = 3,818). We then compared different IRT-based strategies for shortening affect questionnaires to 2-4 items in simulation studies. Results indicate that computerized adaptive testing, but also simpler, (semi-) static item selection strategies achieve higher precision with the same number of items compared to random selection. We give concrete recommendations and highlight encountered issues and avenues for further refinement of the item bank.</p>