Key points are not available for this paper at this time.
Background: Short, accurate, rapidly completed psychometric tests are in high demand in psychiatry.Time is always limited.Researchers often try not to overburden patients included in studies measuring many variables.However, shorter psychometric tests may demonstrate poorer reliability or lower sensitivity.Computerised adaptive testing (CAT) could be a solution to this important trade-off.Aim: This study aimed to evaluate whether the Paradox of Self-Stigma scale (PaSS-24), a pen and paper test evaluating three dimensions of self-stigma, could be transformed into a computerised adaptive test to reduce the number of items administered.Method: The PaSS-24 items were calibrated using the item response theory.A Monte-Carlo simulation was performed to evaluate the number of items needed to ensure the same test reliability as the initial scale.We simulated 50,000 participants with various levels of self-stigma. Results:Results showed that two of PaSS-24's three subscales could be substantially shortened while maintaining similar reliability.The correlations between simulated and estimated scores were close to unity, indicating that the CAT did not sacrifice accuracy for brevity.Conclusions: Simulated data suggested that shortened psychometric CAT could achieve similar reliability to the initial PaSS-24 despite the latter already being brief.CAT provides an opportunity to give in to the pressure of using short scales in psychiatry without compromising on reliability and accuracy.
Golay et al. (Tue,) studied this question.