Non-Gaussian Liability Distribution for Depression in the General Population.

Area of Science:

• Psychiatry
• Psychometrics
• Epidemiology

Background:

• The distribution of depression risk in the general population is often assumed to be normal.
• This assumption simplifies statistical modeling but may not reflect empirical reality.

Purpose of the Study:

• To empirically assess the true shape of the latent depression risk distribution.
• To evaluate the performance of the Davidian-Curve Item Response Theory (DC-IRT) method for estimating complex latent distributions.

Main Methods:

• Utilized the National Health and Nutrition Examination Survey (NHANES) data from 2005-2018 (n=36,244).
• Applied the Davidian-Curve Item Response Theory (DC-IRT) to estimate continuous latent depression density.
• Conducted simulations to test DC-IRT performance with large samples and realistic items, including analyses on subsamples.

Main Results:

• The estimated latent depression risk distribution was found to be left-skewed and bimodal.
• DC-IRT successfully recovered complex latent distributions, even when not apparent from sum scores.
• Estimation accuracy varied with sample size and model selection method.

Conclusions:

• The assumption of a normal distribution for depression risk in the general population is not empirically supported.
• Depression risk exhibits a complex, non-normal latent structure.
• DC-IRT is a viable method for uncovering nuanced latent distributions in large datasets.