UNIPOL-GC

Certain traits in personality and attitude measurement, particularly clinical traits and extreme beliefs have extreme (rightly skewed) item distributions and are, possibly, best modeled by assuming that they behave as unipolar variables or quasi-traits (Lucke, 2013, 2015; Reise & Waller, 2009). A unipolar trait is one that: (a) takes only positive values, (b) is not equally meaningful at both ends of the continuum, and (c) has a rightly skewed distribution in the target population. For example, in a clinical scale that measures symptoms, the low end merely reflects absence of pathology while the upper end reflects the different levels of severity of those people who suffer from the pathology, and so, is far more relevant or meaningful. Furthermore, most of the individuals are expected to have low trait levels and be piled-up at the lower end. UNIPOL-GC (for unipolar graded and continuous) is able to fit two models intended to measure unipolar traits, in which the latent trait distribution is assumed to be lognormal. They are: (a) the Log-Logistic graded response model (LL-GRM; Reise et al. 2021) and (b) the Log-logistic continuous response model (LL-CRM; Ferrando et al. 2023), Essentially, both models can be viewed as transformed versions of the standard Item Response Theory (IRT) Samejima’s Graded-Response and Continuous-Response Models (GRM and CRM) in which the trait scale and distribution has been changed. More in detail, the standard models assume that the trait of interest has a normal distribution and is equally meaningful at both ends, whereas the LL-models assume this distribution to be lognormal, and the trait to be more meaningful at its upper end. This alternative modeling implies a very different functioning from that of the standard models. Thus, at the calibration stage, the item response functions are not ogives but power functions. And, at the scoring stage, the shape of the information curve is the opposite to that derived from the standard models. Unipol-GC has been developed in R version 4.1.2 and has been successfully tested in previous versions such as 3.6.2. It runs in any operating system that supports R (Windows, Linux, Mac OS). If you think the program can be useful in your research or practice, please, give it a try. You can freely download here the package and a detailed manual. A working example that shows how the program functions is also included so as to be used with the xsim10.dat database.

References

Ferrando, P. J., Morales-Vives, F., Hernandez-Dorado, A. (2023). Measuring unipolar traits with continuous-response items: Some methodological and substantive developments. Educational and Psychological Measurement.
https://doi.org/10.1177/00131644231181889

Lucke, J. F. (2013). Positive trait item response models. In R. E. Millsap, L. A. van der Ark, D. M. Bolt, and C. M. Woods (Eds.), New developments in quantitative psychology (pp. 199–213). Springer.

Lucke, J. F. (2015). Unipolar item response models. In S. P. Reise and D. A. Revicki (Eds.), Handbook of item response theory modeling: Applications to typical performance assessment (pp. 272–284). Routledge/Taylor & Francis Group.
https://doi.org/10.4324/9781315736013

Reise, S. P., Du, H., Wong, E. F., Hubbard, A. S., & Haviland, M. G. (2021). Matching IRT models to patient-reported outcomes constructs: The graded response and log-logistic models for scaling depression. Psychometrika, 86(3), 800-824.
https://doi.org/10.1007/s11336-021-09802-0

Reise, S. P., Rodriguez, A., Spritzer, K. L., & Hays, R. D. (2018). Alternative approaches to addressing non-normal distributions in the application of IRT models to personality measures. Journal of Personality Assessment, 100, 363–374.
https://doi.org/10.1080/00223891.2017.1381969

Reise, S. P., & Waller, N. G. (2009). Item response theory and clinical measurement. Annual review of clinical psychology, 5, 27-48.
https://doi.org/10.1146/annurev.clinpsy.032408.153553

Rosseel, Y. (2012). lavaan: An R package for structural equation modeling. Journal of statistical software, 48, 1-36.
https://doi.org/10.18637/jss.v048.i02