The Validation of the Spanish Adaptation of the Protective Factors Survey

Purpose. The child maltreatment prevention field lacks valid and reliable Spanish instruments. FRIENDS National Resource Center and the Center for Public Partnerships and Research, University of Kansas developed and tested a Spanish version of the Protective Factors Survey (S-PFS). Many maltreatment prevention programs use the PFS, which measures families’ Family Functioning/Resiliency, Concrete Supports, Social Supports, Nurturing and Attachment (subscales), and Knowledge of Parenting/Child Development (individual items).

The purposes of this study are to determine whether the S-PFS is metrically invariant across time; subscales of the S-PFS are correlated; and the S-PFS is a valid and reliable measure.

Methods.

Study Design. This study is a three-form planned missing pretest-posttest design, which reduces participant burden. Participants are randomly assigned to complete a proportion of questions from each validation measure. Missing data resulting from planned missing designs can accurately be recovered using modern missing data techniques (full information maximum likelihood [FIML]).

Data Collection. Between June 2012 and April 2013, maltreatment prevention agencies were recruited through national list serves, conferences, and webinars. The seven participating agencies agreed to provide 12 hours of service and to serve a minimum of 10 Spanish-speaking clients. 141 pretest and 148 posttest surveys were collected between September 2012 and July 2013.

Measures. Agencies received survey packets with four instruments to complete at pre- and posttest (S-PFS, Family Resource Scale [FRS], revised Difficult Life Circumstances [DLC], Child Abuse Potential Inventory [CAP]).

Sample Characteristics.  Participants received an average of 30.6 service hours (SD=5.4, range 13-36). Primarily, participants were female (86.5%), Latino (92.1%), partnered (70.8%), 35 years of age or younger (62.0%), and earning $30,000 or less annually (69.2%). Most participants received a high school education or less (75.5%).

Statistical Approach. We conducted confirmatory factor analyses (CFA) with the R package lavaan 0.4-14. Missing data was handled through FIML.

Results. Using S-PFS pretest data, we fit a CFA model to the four subscales, finding good fit, χ²(84) = 148.097, p < 0.001, CFI = 0.929, TLI = 0.911, RMSEA = 0.074 (90% CI: 0.054-0.093), SRMR = 0.072. Regarding content validity, all standardized factor loadings besides one were greater than 0.65. Each subscale correlated between 0.49 and 0.90. Concerning stability over time, we fit S-PFS data to a strong invariance model, χ²(384) = 657.57, p < 0.001, CFI = 0.869, TLI = 0.851, RMSEA = 0.066 (90% CI: 0.057-0.075), SRMR = 0.072, which indicates the S-PFS’s measurement structure did not differ between pre and posttest. Predictive validity results from an autoregressive panel model indicated that S-PFS scores account for 39% to 68% of the variance in posttest scores. For concurrent validity, correlations between the S-PFS subscales and FRS, CAP, and DLC subscales at pre and posttest were as expected (-0.60 to 0.47). Cronbach’s alpha values indicated very good internal consistency across pre and posttest, with all values greater than 0.65.

Implications. The S-PFS represents a viable measurement tool for maltreatment prevention programs to use with Spanish-speaking families. The results from this study provide support for the validity and reliability of the S-PFS.