Differences between complete and incomplete couples in physical health research: Implications for methods and generalizability.

RATIONALE: Couples-based interventions to facilitate health behavior change and improve disease outcomes are gaining attention from researchers. Dyadic research, however, poses unique methodological challenges that have raised questions about research samples and the generalizability of findings.

OBJECTIVE

The current study examined whether couples in which both partners participated (i.e., complete couples) in a couples' health research study systematically differed from those in which only one partner participated (i.e., incomplete couples).

METHODS

Between January 2014 and November 2015, an online survey was advertised on Facebook to people who were both engaged to be married and living in the Denver, Colorado metropolitan area. When the first member of a couple completed the survey (initially recruited participant), they provided their partner's email address, which prompted the research staff to invite the second partner to complete the same online survey. Constructs assessed included demographics, health behaviors, general health status, and relationship quality. Participants answered questions about themself and their partner. Approximately one-third of the partners of initially recruited participants also participated. Data from initial participants in complete couples (N = 265) were compared to data from initial participants in incomplete couples (N = 509).

RESULTS

Chi-square tests and independent samples t-tests indicated participants in incomplete couples had significantly lower relationship quality, worse health behaviors, and poorer health status compared to those in complete couples. Reports on partner health behaviors also differed in the same direction between the two groups. Participants in complete couples were more likely to be White, less likely to have children, and had more education than in incomplete couples.

CONCLUSIONS

Findings suggest that studies requiring both members of a couple may recruit samples that are less diverse with fewer health concerns than research that only requires individual participation if their partner refused participation. Implications and recommendations for future couples-based health research are discussed.