Affiliations: [a] William S. Middleton Memorial Veterans Hospital, Madison, WI, USA | [b] University of Wisconsin-Madison, Madison, WI, USA | [c] University of North Texas, Denton, TX, USA | [d] Yale University School of Medicine and VA Northeast Program Evaluation Center, New Haven, CT, USA
Correspondence:
[*]
Address for correspondence: Shana K. Bakken-Gillen, Wm. S. Middleton Memorial Veterans Hospital, 2500 Overlook Terrace, Madison, WI 53705, USA. Tel.: +1 608 770 3572; [email protected]
Abstract:
OBJECTIVE: The purpose of this study was to explore factors associated with employment outcomes of veterans with substance use disorders.
METHODS: Chi-squared automatic interaction detector (CHAID) was used to explore how interactions between participant entry characteristics and treatment participation variables relate to employment outcome. Participants included 46,641 veterans with substance use disorders from the Veterans Health Administration’s Compensated Work Therapy (CWT) Programs during the period of 1993 to 2005 The data used in this study were administrative archival data routinely collected by the Northeast Program Evaluation Center (NEPEC) from all CWT programs nationwide.
RESULTS: Factors most highly associated with competitive employment outcome included greater treatment intensity, defined as higher weekly mean earnings and longer treatment duration; higher vocational functioning prior to admission defined as a shorter length of time since employed for at least a month; and participation in a transitional work experience position on the VA grounds.
CONCLUSION: The CHAID analysis was able to construct an optimal model using participant entry characteristics and treatment participation variables to explain the variance in the dependent variable, employment outcome. Through the process of segmentation of the sample into mutually exclusive homogeneous subgroups with different probabilities of positive outcomes, CHAID provided detailed information about interactions between the participant entry characteristics and treatment participation variables on the one hand and the outcome variables on the other.
Keywords: Employment, vocational rehabilitation, substance use, veteran, data mining
