Is time of the essence? The impact of different recall periods on participant responses to the Depression, Anxiety and Stress Questionnaire (DASS-21)

There are a significant number of instruments available to measure mental health symptoms, which can limit the comparability and integration of data. Retrospective harmonization facilitates comparison of data from different sources; however, researchers must test measurement invariance testing before harmonization can be conducted. To date, harmonization research has typically focused on the semantic content of scale items and limited attention has been given to the recall period (e.g., one week, one month). Three groups of university students completed the DASS-21 scale based on their symptoms in the “past week” (n=134), “past month” (n=134), and “past six months" (n=127). The latent structure of the DASS-21 was first estimated using confirmatory factor analysis. Then, DIF was tested using a moderated mediation MIMIC model with the recall period group as a covariate. CFA results supported a three-factor model fit the data. MIMIC models indicated that none of the DASS-21 items displayed uniform or non-uniform DIF across groups. However, the “six months” group had significantly higher mean depression scores compared to the “one week” group. The lack of DIF between groups indicates that the recall period of the DASS-21 did not affect item response. The results suggest that different recall periods may not be a barrier to retrospective harmonization, although further research should identify the mechanisms underlying participant symptoms ratings across different recall periods.