Take a group of 18- and 19-year-old women, college freshmen and sophomores. Then test them to find out who has the most social anxiety: who's most nervous about dealing with other people, particularly in public situations. What would be the most difficult thing you could ask these high-social-anxiety women to do? How about this:
I would like you to prepare and deliver a four-minute talk. This talk will be videotaped and viewed later by several professors and graduate students.... It is extremely important that you do the best job that you can with this talk.... Your talk should be about the most difficult time in your life and how you coped with it.
Now, give them five minutes to prepare, and allow their boyfriends to "help."
That's what a team led by J. Gayle Beck did; their goal was to see how socially anxious women and their romantic partners handled a difficult social situation. They asked women with low social anxiety and their partners to do the same task; 45 women in all participated. Of course, what the researchers were really interested was to see how the couples interacted while they prepared the speeches; in the end none of the women had to give a speech, and they were told their preparation session had been videotaped and would be analyzed for insights into how their relationship worked.
You might think that highly socially anxious women (which I'll abbreviate as HSA) would be more distressed about this than women with low social anxiety (which I'll abbreviate as LSA). You might also think that HSA women who weren't satisfied with their relationships would show more have more negative interactions with their partners than HSA women who were satisfied. And you'd probably speculate that if the boyfriends of HSA women made negative comments or behaved negatively during the preparations, that HSA women would show even more distress.
Beck's team predicted all three of these results, and were surprised to find that none of the predictions were supported by the study. They studied all the videos and rated the women along three dimensions:
- Positive: Specific analysis of the problem, statement of feelings, asking for help, positive response to helper
- Negative: Demanding help, criticizing, blaming, accusing, rejecting helper, whining, complaining
- On Task: Staying focused on the assignment.
The boyfriends were rated on a similar scale.
The researchers could find no significant differences in behavior between the HSA and LSA women or their partners when the results were averaged across all participants in a group. But when each group was divided into subgroups of high- and low-satisfaction with their relationships, a significant difference was observed. Among high-satisfaction women, HSA women showed significantly more negative behavior than LSA women.
The researchers speculate that women who are satisfied with their relationships may fell more secure expressing their emotions when they are nervous or anxious. Since the LSA women probably weren't as anxious about the speech, they had no reason to show any signs of discomfort, but HSA women did. HSA women who were unsatisfied with their relationships, on the other hand, were not comfortable sharing their anxiety with their partners.
And what about when the boyfriends behaved negatively? Again unexpectedly, HSA women behaved more negatively when their boyfriends behaved more positively to them. Among low-social anxiety women, there was no difference in behavior regardless of how their boyfriends behaved. Why did the highly-anxious women behave worse when their boyfriends were being nice?
Beck's team believes that these women are more comfortable behaving negatively because they know their boyfriends are supportive. The women with unsupportive boyfriends don't respond in kind because they feel they will only get more negative responses back from them.
GAYLEBECK, J., DAVILA, J., FARROW, S., & GRANT, D. (2006). When the heat is on: Romantic partner responses influence distress in socially anxious women Behaviour Research and Therapy, 44 (5), 737-748 DOI: 10.1016/j.brat.2005.05.004
I'm not sure the title of this post (which I saw in the Now on Scienceblogs header, btw), then, is really that accurate, since the study is more along the lines of "When are highly-anxious women most likely to show that they're anxious?"
Personally, I'd like to see the test repeated with some sort of physiological measurement, although I'm not sure how that would be set up from a logistics standpoint. Especially with the way we're socialized to be demure and nice, I'd agree with the team that a lot of the women who didn't show distress could very well be more anxious.
I'm not a statsy guy, but don't these results seem a little weak? I think there's a rule of thumb about not looking for interactions unless you also have a main effect. In this study, they found nothing in regression, (test 1), then it sounds like they did a median split (also generally a no-no), sounds like they still found nothing between HSA and LSA (test 2), they also found nothing between High satisfaction and low satisfaction (test 3), and then they looked at interactions (tests 4-7?), where they found one result (among high satisfaction, there's a difference between HSA and LSA in negativity). I was confused a bit by post in regard to second finding, is that all HSA women, or the HSA women in the high satisfaction relationships? Seems like the latter (assuming boyfriends who provide positive support are correlated w/high satisfaction relationships), and now I'm just starting to get confused w/all the tests and the fact that there were only a couple of findings. I'm not trying to go all bonferroni on this, but it feels a bit weak. Having said all that, I'm realizing now that I date an HSA woman and she always respond negatively to me when she's stressed, and it gets worse the more supportive I am, and I think she's satisfied w/the relationship...
Though I do think the content and idea of the study is interesting--The nature of the interactions between "reported" satisfaction, and anxiety do not seem to be thought through enough for the testing. I think this test shows how the causal relationships in subjects like this are epistemologically quite murky.
I have to agree with Tage; this does seem a bit like data dredging to me with a little hindsight bias thrown in to justify the findings.
Proofread your work, please.
I'm not sure how you can judge how anxious someone is based on a video. The results show whether or not the subject appears anxious based on the interpretation of the person viewing the video, not whether the subject actually is anxious.
As a personal example, I happen to be a hyper, fidgety-type person, which other people sometimes misinterpret as a sign of anxiety at times when I am not anxious at all. I just have a speedy metabolism.
More accurate results could have been obtained by simply asking the subjects to rate how anxious they felt on a 0-10 scale.
Women with nice guys for boyfriends treat them like crap? Say it ain't so... who were the nitwits who formulated these null hypotheses. They clearly are out of touch with human nature.
One thing I fail to understand in the research setup: high social anxiety seems to be a thing that only women suffer from?!