Using the Rasch Measurement Model in Psychometric Analysis of the Family Effectiveness Measure (FEM)

Thursday, 25 July 2013: 1:35 PM

Linda L. McCreary, PhD, RN1
Karen M. Conrad, PhD, RN2
Kendon J. Conrad, PhD3
Christy K. Scott, PhD4
Rodney R. Funk, BS5
Michael L. Dennis, PhD3
(1)Department of Health Systems Science, University of Illinois at Chicago College of Nursing, Chicago, IL
(2)Program Metrics, LLC, Oak Park, IL
(3)Chestnut Health Systems, Normal, IL
(4)Chestnut Health Systems, Chicago, IL
(5)Chestnut Health Systems, Bloomington, IL

Learning Objective 1: Describe the psychometric properties of the EFF and the IFF, subscales of the Family Effectiveness Measure that assess effective and ineffective family functioning.

Learning Objective 2: Discuss the psychometric information provided by Rasch analysis and explain its advantages in assessing psychological and behavioral measures used in nursing research.

Purpose: Existing measures, developed primarily with mainstream American families, may not validly assess family functioning in other populations. Guided by the Rasch measurement model and classical test theory, this study was conducted to assess psychometric properties of the Family Effectiveness Measure (FEM), developed with low-income African-American single-parent families, and to make refinements based on results.

Methods: A convenience sample of 607 predominantly unmarried, low-income African American adult women participated. A cross-sectional design using Rasch analysis assessed unidimensionality, response category functioning, item fit, person reliability, differential item functioning (DIF) by race and parental status, and item hierarchy. Criterion-related validity was tested using correlations with five variables related to family functioning.

Results: The original 42-item FEM measures family effectiveness with both positive and negative (reverse-scored) items. Rasch analysis suggested that the original FEM actually measures two separate constructs: effective (EFF) and ineffective family functioning (IFF). The 20-item EFF was shown to be a psychometrically sound measure, made more efficient by deleting 22 items (6 misfitting effective family items and 16 ineffective family items). The IFF, consisting of the 16 ineffective family items, was less psychometrically strong. A weak, negative correlation between the EFF and IFF supported existence of two constructs. Collapsing the 5-point Likert-scale response categories into a 4-point scale optimized response category functioning. Both EFF and IFF evidenced good reliability and strong criterion-related validity. Neither EFF nor IFF showed DIF by race; one IFF item showed DIF by parental status.

Conclusion: In contrast to the prevailing conceptualization of family functioning as a single construct assessed by positive and negative items, Rasch analysis suggested the existence of two separate constructs: effectiveness and ineffectiveness. While the EFF is a strong, efficient measure of family functioning, the IFF requires additional development and testing. Rasch analysis provided important information that strengthens measurement of family functioning.