Assessing the Ordinality of Response Bias with Item Response Models: A Case Study Using the PHQ-9

Author

Venessa N. Singhroy, The Graduate Center, City University of New YorkFollow

Date of Degree

5-2018

Document Type

Dissertation

Degree Name

Ph.D.

Program

Educational Psychology

Advisor

Jay Verkuilen

Committee Members

Irvin Schonfeld

Renzo Bianchi

Howard Everson

Claire Wladis

Subject Categories

Applied Statistics | Occupational Health and Industrial Hygiene | Quantitative Psychology | Social Statistics | Statistical Methodology | Statistical Models

Keywords

ordinality, monotonicity, clinical assessment, item response theory, PHQ-9, depression

Abstract

Improper scale usage in psychological and clinical assessment is an important problem. If respondents do not use the scales in a consistent manner, the reliability of a composite is likely to be attenuated. This is particularly problematic when particular items are singled out for special treatment or when subscales are of interest, not just a total score. This study used both non-parametric and parametric item response theory (IRT) methods to gain further insight into the validity of the PHQ-9, a dual purpose instrument that assesses the severity of depressive symptoms using nine Likert-scale items and allows the investigator to establish provisional diagnoses of depressive disorders. The data was collected by Bianchi et al. (2015) across three separate cross-cultural samples of teachers. The analysis indicated that scale monotonicity was preserved, violations to ordinality occurred among a subset of items resulting in inconsistent scale usage within the different samples, and that language differences in the test administration primarily accounted for the differences in scale usage.

Recommended Citation

Singhroy, Venessa N., "Assessing the Ordinality of Response Bias with Item Response Models: A Case Study Using the PHQ-9" (2018). CUNY Academic Works.
https://academicworks.cuny.edu/gc_etds/2583