TY - JOUR
T1 - Multigroup Propensity Score Approach to Evaluating an Effectiveness Trial of the New Beginnings Program
AU - Tein, Jenn-Yun
AU - Mazza, Gina L.
AU - Gunn, Heather J.
AU - Kim, Hanjoe
AU - Stuart, Elizabeth A.
AU - Sandler, Irwin
AU - Wolchik, Sharlene
N1 - Publisher Copyright:
© 2018, © The Author(s) 2018.
PY - 2018/6/1
Y1 - 2018/6/1
N2 - We used a multigroup propensity score approach to evaluate a randomized effectiveness trial of the New Beginnings Program (NBP), an intervention targeting divorced or separated families. Two features of effectiveness trials, high nonattendance rates and inclusion of an active control, make program effects harder to detect. To estimate program effects based on actual intervention participation, we created a synthetic inactive control comprised of nonattenders and assessed the impact of attending the NBP or active control relative to no intervention (inactive control). We estimated propensity scores using generalized boosted models and applied inverse probability of treatment weighting for the comparisons. Relative to the inactive control, NBP strengthened parenting quality as well as reduced child exposure to interparental conflict, parent psychological distress, and child internalizing problems. Some effects were moderated by parent gender, parent ethnicity, or child age. On the other hand, the effects of active versus inactive control were minimal for parenting and in the unexpected direction for child internalizing problems. Findings from the propensity score approach complement and enhance the interpretation of findings from the intention-to-treat approach.
AB - We used a multigroup propensity score approach to evaluate a randomized effectiveness trial of the New Beginnings Program (NBP), an intervention targeting divorced or separated families. Two features of effectiveness trials, high nonattendance rates and inclusion of an active control, make program effects harder to detect. To estimate program effects based on actual intervention participation, we created a synthetic inactive control comprised of nonattenders and assessed the impact of attending the NBP or active control relative to no intervention (inactive control). We estimated propensity scores using generalized boosted models and applied inverse probability of treatment weighting for the comparisons. Relative to the inactive control, NBP strengthened parenting quality as well as reduced child exposure to interparental conflict, parent psychological distress, and child internalizing problems. Some effects were moderated by parent gender, parent ethnicity, or child age. On the other hand, the effects of active versus inactive control were minimal for parenting and in the unexpected direction for child internalizing problems. Findings from the propensity score approach complement and enhance the interpretation of findings from the intention-to-treat approach.
KW - children of divorce
KW - effectiveness trial
KW - intention-to-treat principle
KW - preventive intervention
KW - propensity score analysis
UR - http://www.scopus.com/inward/record.url?scp=85046807505&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85046807505&partnerID=8YFLogxK
U2 - 10.1177/0163278718763499
DO - 10.1177/0163278718763499
M3 - Article
C2 - 29635949
AN - SCOPUS:85046807505
SN - 0163-2787
VL - 41
SP - 290
EP - 320
JO - Evaluation and the Health Professions
JF - Evaluation and the Health Professions
IS - 2
ER -