> But early friendship bonds played an even bigger part than maternal relationships in the ways people navigated adult friendships and romantic partnerships, accounting for 4 percent of the variance in adults’ romantic partner- and best friend-specific attachment anxiety, and 10 to 11 percent in their partner- and best friend-specific avoidance.
Are those numbers r-squared figures? Seems like there's a lot more variance to be explained?
1. Maybe the measurements are just very noisy. In which case they may also have other biases.
2. Maybe there are systematic causes which the study didn't capture. If so, controlling for them might change the results.
Sigh. When I see a study headline like this I feel confident about two things. First, the study will have a weak design with no serious attention paid to causality, genetic confounding etc... second, the response to it will be full of people going "yes, that fits my N=1 anecdote" or "no that doesn't fit my N=1 anecdote", in other words, critiquing the weak methodology with an even weaker methodology (handwaving appeals to personal experience).
One reason social science is hard is there isn't much market for the truth. People just want a nice story to tell themselves.
Are those numbers r-squared figures? Seems like there's a lot more variance to be explained?