Users opted for sameness more often than chance would predict in all the characteristics examined in this section. This concurs with the overwhelming evidence gathered by relationship researchers (see surveys in Brehm et al. 2002, Fisher 1992) that actual similarity and perceived similarity in demographics, attitudes, values, and attractiveness correlate with attraction (and, later, relationship satisfaction). However, users demonstrate this homophily to differing degrees for different characteristics.
Dyads were much more likely than chance to choose the same value for characteristics relating to the life course. Values for marital status and wanting children were the same in dyads 64 percent and 54 percent more often, respectively, than would occur with random pairings. The number of children users already have was the same in dyads 39 percent more often than chance. These were the three most strongly bounding characteristics.
Physical build was the same among dyads 28 percent more often than chance would predict. This finding rests on similarity-seeking among a few possible values for build, such as "average" and "athletic," that encompass both genders; many of the other possibilities, like "petite" and "body-builder," are strongly gendered and thus very unlikely to be the same in a heterosexual dyad. Physical appearance — a self-reported rating of attractiveness — was the same among dyads 23 percent more often than chance.
Among lifestyle choices, including smoking habits, drinking habits, and pet preferences, only smoking was the same in dyads more than 20 percent more often than chance would predict. Most dyads (68.7 percent) were the same in drinking habits, but this is because 75.6 percent of male users and 77.9 percent of female users identified themselves as "Social/occasional" drinkers. Thus, the expected probability of sameness was also high for this characteristic, rendering the high actual similarity unremarkable.
Pets — both general preferences regarding them and specific pets already owned — proved the least bounding of any characteristics. Users picked dyadic partners who shared their preferences only about 10 percent more often than chance would predict. Homogeneity on these characteristics did not matter to users nearly as much as other characteristics.
Religion was the same in dyads 17 percent more often than chance. More than half of the active users of the Site identified themselves as Christian, and about a third chose "Prefer not to answer," a very high percentage compared to other characteristics. Given the distribution of religions (Table A.X) among users who did answer, we might reasonably presume that a large number of "Prefer not to answer" respondents are in fact Christians, even if we allow that non-Christians might be more likely to choose "Prefer not to answer."
If this is the case, the bounding strength of religion might appear lower than it is because of users' reluctance to specify their religion. It might also be true that, as some analyses suggest (e.g., Williams & Lawler 2003), having similar religiosity is more important than sharing a specific religion.
The overwhelming majority of dyads (81.2 percent) shared the same race, but, as with drinking habits, this high rate of similarity is only moderately better than chance (14 percent). Because 83.7 percent of users were Caucasian, the rate of similarity expected by chance was also high, 71.1 percent.
In general, characteristics were slightly more bounding among the subset of reciprocated contacts, but the difference was small and roughly equal across characteristics. Although the difference is small, it suggests that users were slightly more likely to respond to an initiation from a more similar other.
Was this article helpful?