You're making multiple unsupported assumptions and simplifications in these couple sentences that might not hold, so your conclusion becomes invalid. One example: the age of people in marriages compared to people in unmarried relations will be different and with it a variety of variables, such as the presence of kids. Another: single women are more interested in marriage than single men, so there could be even more differences between married/unmarried couples. Or let's take cheating - I could be completely satisfied throughout 99% of the duration of the relationship until the point when cheating occurred. Or we know that women are more likely to divorce men earning less than them or stay-at-home dads and the earning power of men and women in marriages changes as compared to unmarried men and women. Etc, etc. Social science gets very messy very quickly.