Reducing the burden of psychological questionnaire measures through selective item re-weighting

Questionnaire measures are central to many areas of study within the psychological sciences. However, they often place a heavy burden on participants; questionnaires are frequently lengthy and unengaging, and with participants often required to complete multiple measures within a single study, this results in lower data quality, increased cost and a poor participant experience. Here, we introduce a straightforward method for creating short versions of existing measures that are able to accurately determine participants' sum scores, subscale scores or factor scores. Our method, referred to as Factor Score Item Reduction with Lasso Estimator, uses Lasso-regularized regression to select items and weight them such that true scores can be predicted accurately from a reduced item set. We demonstrate the performance of this method on an example dataset, and provide code and guidance for implementing the approach.