Converting negative symptom dimension scores across SANS and PANSS

Several factor-analytic studies of the Scale for the Assessment of Negative Symptoms (SANS) and Positive and Negative Syndrome Scale (PANSS) have identified two dimensions. These dimensions reflect expressive (EXP) and experiential or Motivation and Pleasure (MAP) negative symptoms. However, the extent to which these factor solutions yield similar dimension scores across the two scales remains unclear. This study compared equipercentile linking and linear regression data harmonization methods for cross-scale EXP and MAP conversions and evaluated their accuracy and reliability. While conversions between methods were similar, regression provided a slight edge over equipercentile linking in most comparisons. Most conversions had moderate or better reliability (ICC > 0.6, p < 0.01). EXP conversions generally had higher ICCs than MAP, with EXP >60% and MAP <60% posterior probability of exceeding ICCs>0.75. Some factor score conversions yielded higher reliability than others. However, when considering both EXP and MAP together, Ahmed's factor scores for SANS and Jang's or Khan's for PANSS performed the best. This study is the first to compare equipercentile linking and regression frameworks for scale harmonization as well as to examine the conversion reliability and accuracy for the EXP and MAP dimensions across SANS and PANSS. It is also the first to employ a Bayesian approach, which provides additional information through probability distributions for choosing the most favorable conversion. The conversion formulae from this study and their reported metrics lay the groundwork for analyses of the EXP and MAP even when assessed with different clinical scales.