Trial Marriage Reconsidered: Evidence for a Positive Causal Effect of Cohabitation on Marital Stability

Felix Elwert, Harvard University

Previous research rejects the well-known "trial marriage" hypothesis, which posits that pre-marital cohabitation decreases the risk of marital dissolution. This paper proposes a revised theoretical perspective and new statistical methods to reevaluate the empirical support for the trial marriage hypothesis. The paper has three parts. The theoretical part of the paper demonstrates that the trial marriage hypothesis concerns the total causal effect of cohabitation. The methodological part of the paper shows that previous empirical estimation strategies are doubly biased toward rejection. The empirical part of the paper suggests an improved test of the hypothesis, using Robins's marginal structural models for complex longitudinal data. Preliminary estimates from the National Survey of Family Growth suggest that, contrary to findings established in the literature, the trial marriage hypothesis is supported by the data.