Date of Award

1-1-2019

Document Type

Dissertation

Degree Name

Ph.D.

Department

Quantitative Research Methods

First Advisor

Kathy Green, Ph.D.

Keywords

Adoption, Item response theory, Psychometrics, Rasch

Abstract

A unidimensional Rasch approach was used to explore whether the data collected through the National Survey of Adoptive Parents of 2007 (NSAP) for the well-being items represented a single latent construct and to establish a base model for comparison. A consecutive approach was then used as an exploratory tool to draw out potential multiple dimensions. Finally, multidimensional item response theory (MIRT) was used to confirm the results of the consecutive approach findings while comparing with the unidimensional baseline. Items within the survey were evaluated for scale function as well as invariance.

The comparison of three approaches (unidimensional, combined consecutive, and 2-dimensional MIRT) found that the combination of Consecutive Dimensions A and B yielded the best fitting model for these data sets. The nested 2-dimensional MIRT model showed better fit than the unidimensional model, but concerns with item position and inconsistent error terms supported the combined consecutive model.

The use of IRT and MIRT analysis techniques helped strengthen the survey by identifying items within the survey that relate to identified constructs. The comparison of three approaches provides practitioners with an example of how to use a consecutive approach in Rasch for exploratory purposes when dimensionality has not already been established.

The NSAP survey was developed to gather data from a large cross-section of adoptive parents in the United States. The well-being subsection gathered data on the parent-child relationship with the intent to assist adoption practitioners, policy-makers, and researchers. Since only twelve of the thirty-nine items were utilized within the models, the data collection opportunity was not fully captured. This lost opportunity of data collection supported the idea of survey development partnerships between topic content experts and psychometricians, when building measures, to maximize the effectiveness of the tool as well as the data gathered.

Provenance

Received from ProQuest

Rights holder

Michael A Furno

File size

170 p.

File format

application/pdf

Language

en

Discipline

Statistics

Share

COinS