Stability of response styles in a personality measure: Evidence with clinical adolescents

Abstract Background In counselling and psychotherapy, individuals may exhibit response styles when completing personality measures, potentially distorting comparisons of inter‐ and intra‐differences. However, previous research has primarily focused on nonclinical samples and cross‐sectional data, leaving gaps in our understanding of the stability of response styles and their impact on clinical samples. This study aimed to fill gaps in knowledge by applying a mixed Rasch model to explore response style stability in a personality measure in clinical adolescents. Methods The study sample consisted of 290 clinical adolescents (112 males and 178 females) who were experiencing depressive symptoms and receiving continuous treatment for at least 2 years. Each participant completed a personality assessment at four time points with a six‐month interval between each assessment. To comprehensively examine the stability of response styles over an extended period, this study utilised mixed Rash modelling to analyse enduring response patterns using data collected at the initial (T1) and fourth (T4) assessments, which were spaced 2 years apart. Results Using mixed Rasch modelling, the results showed that the extreme response style (ERS) and middle response style (MRS) were the two most common response styles in the five domains of the Big Five Inventory‐21 at T1 and T4. The ERS and MRS remained stable throughout the mental health intervention. There was no gender‐related relationship found for response styles in the data at separate time points (T1 and T4) or in the longitudinal data (long‐format data). Importantly, the findings revealed that odds of a clinical adolescent changing their personality response style (ERS → MRS or MRS → ERS) increased by 1.15 times with each unit increase in depression score. Conclusion Researchers and practitioners should be aware that the stability of the ERS and MRS for a personality measure may result in response style‐based explanations for findings obtained from longitudinal data, such as trait stability or trait change over waves.