Predictors of Consent in a Randomized Field Study in Child Welfare

Background:Numerous factors play a role in individuals’ willingness to consent to participate in a randomized field trial (RFT). These factors are important to understand, as they affect which participants are included in the study and the generalizability of the findings (Weiss et al., 2008). This study examined factors that could influence consent among families recruited to participate in a randomized study that tested in-home Parent Management Training-Oregon model (PMTO) for children in foster care with serious emotional disturbance (SED). The study aim was to build knowledge about the key predictors of consent to an RTF in a real-world child welfare setting.   

Methods: The project setting was a statewide demonstration project funded under the federal Permanency Innovations Initiative which sought to reduce long-term foster care. The sample comprised 1,115 (647 treatment and 468 comparison) families who were recruited to participate in the study between September 2012 and September 2014. The study was a randomized consent design (Zelen, 1990). Bivariate and multivariate analyses were conducted to identify predictors of consent. The bivariate analyses consisted of crosstabulations with chi-square tests for dichotomous variables and t-tests for continuous variables. Odds ratio were calculated to measure effect size for crosstabs and Cohen’s d statistic was calculated to evaluate effect size for the t-tests. Logistic regression was used to test the predictive validity of the available variables from a multivariate perspective. Steps were taken in the multivariate analysis to maximize the potential for discovery of possible confounding or interaction effects in the pool of predictor variables.

Results: Bivariate analyses revealed the strongest effects were in the moderate range for dual reunification cases (i.e., when more than one family was seeking reunification with a child) (OR=2.22, p<.001), and cases where the child was removed because of parental incarceration (OR=2.67, p=.02). In the multivariate analyses, the strongest predictor of consent was whether the case was a dual reunification case, with dual reunification cases’ odds being twice as great they would consent (OR=2.22, 95% CI=1.48-3.33, p<.001). Single male headed families were approximately twice as likely to refuse to consent compared to single female headed families (OR=.50, 95% CI=.31-.80, p=.004). A significant but weak effect was observed for the treatment group, indicating an increased likelihood in their consenting to the study (OR=1.50, 95% CI=1.14-2.00, p=.004).

Conclusions and Implications: The findings suggest that having two families identified for possible reunification played an important role in predicting the likelihood of agreeing to participate in a research study. The reason for a child’s removal from the home also proved to be a significant factor, with parental incarceration increasing likelihood of consenting while abandonment reduced the likelihood. Furthermore, primary caregiver’s gender also influenced how likely families were to participate in an RFT. Future studies could examine provider/clinician variables that may act as significant predictors of families’ willingness to consent to a study. Our findings demonstrate the importance for researchers to consider family structural and caregiver gender differences when recruiting participants.