All ETDs from UAB

Advisory Committee Chair

Robin Gaines Lanzi

Advisory Committee Members

Scott Snyder

Peter Hendricks

David Redden

Document Type

Dissertation

Date of Award

2018

Degree Name by School

Doctor of Philosophy (PhD) School of Public Health

Abstract

This study applies Rasch analysis to validate two scales in the Parenting for the First-time project: Beck Depression Inventory (BDI-II) and Parenting Stress Index (PSI). The main purpose of this study is to evaluate quality of the two scales by using Rasch analysis to examine the unidimensionality, reliability, fit statistics, group and time invariance, and optimal response categories. Meanwhile, this study also compares the results of Rasch model and Classical Test Theory (CTT) to assess the advantages of Rasch analysis. Result: Both Rach analysis and CTT show the evidence of unidimensionality of BDI-II in the three administrative periods: Prenatal, 6 month, and 12 month. For parenting stress scale, the results of both analysis shows that there are two dimensions: Childrearing (CRI) and Self Stress (SSI). Overall, the Rasch person reliabilities index are less than Cronbach alphas for the three scales (BDI-II, CRI, and SSI) in each administrative period. Most of items in BDI-II do not remain invariant across two age groups (<19 and >=19). Item 29, 31, 32, 33 (misfit items) of CRI cannot keep invariant across the age groups. No DIF items are found in SSI scale. Differential Test Function (DTF) analysis shows that, except for several items, BDI-II roughly functions similarly over the three administration periods. Both DTF analysis for CRI and SSI also shows that the two measures are time invariant (except for one CRI item showing outside the 95% boundary of DTF chart). Response category optimization shows that collapsing one response level may generate a better reliability statistic for BDI-II. Although collapsing one response level increases the person reliability for CRI and SSI, higher person reliability may not result in a “good” category probability curve. The small increase in reliability is less important than the scales performing in an acceptable manner. Conclusion: Rasch analysis is a complementary and alternative method of classical test theory (CTT) for evaluating the quality of a measure. In this study, both Rasch and CTT presented similar results in term of reliability and validity. However, Rasch analysis provides more detailed information on person ability, item difficulty, targeting, and misfitting items to improve instrument design.

Included in

Public Health Commons

Share

COinS