To build a robust, replicable, and generalizable family science we must ensure that our research includes samples that are large enough that we can test effects reliably and are diverse enough to speak broadly to families’ experiences. This can be challenging for family science researchers who focus on family processes because many of the features of high-quality family process research make the experience quite onerous for participants; often multiple family members must participate, and data is typically collected through intensive methods, such as video observation or daily diaries. These methodologies allow us to capture rich and detailed data about family processes, but can make it difficult to achieve a large and diverse sample. Fortunately, there are a number of promising methods already in use in family science, or currently being deployed in other related fields, that offer good prospects for family science researchers seeking to improve the samples used in their research by increasing sample size and/or diversity. This article highlights innovative methods that will be useful in overcoming some of the sampling challenges facing family science researchers, focusing on creative ways to use existing datasets, including secondary data analysis and integrative data analysis, and methods that can be deployed when collecting new data, including accessing alternative data sources such as digital trace data, collecting observational data remotely, methods for reaching underrepresented groups, and big-team science.

Objective: Government initiatives undertaken to improve the earning potential of disadvantaged unmar- ried parents assume that job training and additional schooling will strengthen these families, yet alternative models predict that these same interventions could overwhelm couples’ limited resources, undermining family stability. Method: We use 3 waves of dyadic data and propensity score analysis to test these competing perspectives by examining the effects of job-related and school-related interventions on 3-year marriage rates. The sample consists of unmarried new parents averaging $20,475 in household income, 52% of whom are African American and 20% of whom are Hispanic/Latino. Results: Marriage rates decreased, from 17% to 10%, for couples in which men participated in school-related interventions. Mediation analyses indicate that school-related interventions reduce the amount of time men spend with their child and the amount of money they contribute to their household, reducing marriage rates in turn. Marriage rates were unaffected by women’s participation in school-related interventions, and by men’s and women’s participation in job-related interventions. Conclusion: Implementing economic interven- tions that increase income while minimizing demands on the limited resources of economically distressed couples may prove necessary for strengthening society’s most vulnerable families.