Abstract: Screening for Post-Traumatic Stress Disorder after a Major Disaster: A Differential Item Functioning (DIF) Analysis of the Impact of Event Scale-Revised (IES-R) (Society for Social Work and Research 23rd Annual Conference - Ending Gender Based, Family and Community Violence)

506P Screening for Post-Traumatic Stress Disorder after a Major Disaster: A Differential Item Functioning (DIF) Analysis of the Impact of Event Scale-Revised (IES-R)

Schedule:
Saturday, January 19, 2019
Continental Parlors 1-3, Ballroom Level (Hilton San Francisco)
* noted as presenting author
Jennifer First, Doctoral Candidate, University of Missouri-Columbia, Columbia, MO
Wes Bonifay, PhD, Assistant Professor, University of Missouri-Columbia
J Brian Houston, Associate Professor, University of Missouri-Columbia, Columbia, MO
Background: Community disasters (e.g., tornadoes, hurricanes, chemical spills) are potentially traumatic events that can result in a high prevalence of post-traumatic stress disorder (PTSD) among individuals exposed. The Impact of Event Scale Revised (IES-R, Weiss & Marmar, 1996) is a 22-item measure often used to screen or evaluate individuals for the development of post-disaster PTSD. Given the increased rates of PTSD in communities experiencing a disaster, the ability to accurately assess and diagnosis PTSD is important for post-disaster research and practice settings. Measurement equivalence refers to the measurement condition in which the underlying trait (e.g., PTSD) is being measured equally across different groups (e.g., gender, race, age). Item response theory (IRT) methods of differential item functioning (DIF) analyzes if different groups within a sample have an unequal likelihood of endorsing a measure’s item, despite having equal levels of the latent trait. Few studies have evaluated the IES-R for DIF across various group memberships in post-disaster settings. Thus, the purpose of the following study was to analyze the IES-R for DIF across subgroups of gender (men vs. women) and age (18-49 vs. 50+ years old) in a sample exposed to a disaster. 

Methods: Data for the current study was drawn from a cross-sectional survey of adult residents of Joplin, Missouri who were exposed to the May 22, 2011 EF-5 tornado (N = 438). IRT analyses were conducted using R statistical programming software and the “mirt” and “lordif” packages. IES-R item responses were evaluated for item and model fit using the generalized Sχ2 item fit statistic and generalized partial credit model (GPCM). DIF analyses of the IES-R were performed by implementing a hybrid ordinal logistic regression/item response theory (OLR/IRT) approach based on the likelihood ratio χ2 test (Choi, Gibbons, and Crane, 2011). To quantify the magnitude of DIF found, McFadden’s pseudo R2 was computed. To control for Type I error, empirical thresholds were derived from Monte Carlo simulations.

Results: Results from the Sχ2 statistic and GPCM calibration found all 22-items in the IES-R displayed adequate item and model fit (CFI = 0.95; TLI = 0.94; RMSEA= 0.07). DIF detection across gender groups found one item exhibited non-uniform DIF. For age groups, DIF detection found one item exhibited uniform DIF and one item exhibited non-uniform DIF. However, the R2 effect sizes for the detected DIF items were small or negligible (< 0.13) and did not produce differential measurement error at the test functioning level for both age and gender subgroups.

Implications: Social workers often work with disaster survivors from various groups, and therefore the validity of clinical measures across groups is essential, since bias can potentially lead to inappropriate treatment guidelines and services. This study’s results provide evidence of the functional equivalence of the IES-R for identifying post-disaster PTSD symptoms across gender and age subgroups. These results indicate that if differences in rates of probable PTSD are observed between men and women or younger and older adults, these differences are unlikely to be caused by item or test functioning biases in the IES-R.