Sex For Me But Not For Thee?

Robin Hanson links to an interesting article about which activities young adults count as “having sex.” The article purports to reveal a double-standard: people are less likely to call a given act sex when they did it, and more likely to call it sex when their long-term partner did it. Here’s the abstract:

The purpose of this study was to determine if undergraduates (N=839) apply the same standard to themselves when labeling a behavior ‘‘having sex’’ as they apply to their significant others if those persons engage in the same behaviors outside the relationship. Using a between-participants design, one form asked participants if each of 11 behaviors constituted having sex if they engaged in the activity; the other form asked participants if each of the same behaviors constituted having sex if their significant other engaged in the activity outside their relationship. Participants answering for themselves were less likely to indicate a behavior was having sex for all behaviors except penile–anal and penile–vaginal intercourse. Men were also more likely than women to indicate most behaviors were having sex. The authors discuss what they define as a definitional discontinuity in undergraduate emerging adults’ definitions of having sex. Fundamental attribution error (FAE) and emerging adulthood literature are used to explain the findings. Health and relationship implications are identified. [emphasis added]

These are fascinating results, but I don’t think they really demonstrate what the authors say they do.

Let’s take an example: does oral sex count as sex? It turns out that men are less likely to say “yes” when they’re the ones who did it, and more likely to say “yes” when it’s their girlfriends who did it. And the same is true for women; they’re more likely to call it sex when their boyfriends did it than when they did it themselves.

The authors interpret this as evidence of a self-favoring bias known as the Fundamental Attribution Error. People wish to see themselves in a favorable light and so they cut themselves some slack, but they are less forgiving when it comes to other people.

But let’s look at how the survey questions were actually phrased. On Form A, respondents were asked to classify an activity involving themselves, such as:

A person had oral contact with your genitals.

On Form B, respondents were asked to classify an activity involving their S.O. (defined as an actual or hypothetical boyfriend/girlfriend or spouse), such as:

Another person had oral contact with your S.O.’s genitals.

Note that the phrasing of the latter activity instantly implies cheating. It mentions an S.O., which by definition means there exists a committed relationship. (The past tense phrasing leaves open the possibility of the event having occurred before the relationship, but the “S.O.” phrasing nevertheless brings the possibility of cheating to mind.) The phrasing of the former activity, on the other hand, does not imply cheating; no S.O. is ever mentioned.

So here’s an alternative hypothesis: the differential results may represent a framing effect rather than a self-serving bias. People tend to define “having sex” in a broader way when faithfulness is an issue, but in a narrower way when the issue is perceived as simply definitional. Thus, a person might consider oral sex to be not-quite-sex as a matter of definition, but count it as sex for purposes of policing loyalty. And that could be true even if the potential cheater is oneself.

This is not to say that there’s no FAE here, just that something else could easily be going on. To distinguish FAE from the framing effect, you’d want to phrase the activities in a more symmetrical fashion, something like this:

A person, not your current S.O., has oral contact with your genitals.

versus

Another person, not you, has oral contact with your S.O.’s genitals.

A difference in answers here would isolate the FAE, since in both cases cheating is clearly involved. On the other hand, if you wanted to isolate the framing effect, you could phrase the activities like this:

A person has oral contact with your genitals.

versus

A person who is not your current S.O. has oral contact with your genitals

This pairing focuses entirely on one person (the respondent) while differing on whether the possibility of cheating is invoked.

The larger point is that the existing question-pairs differ on two dimensions (the person involved and whether an S.O. is mentioned). But to isolate a particular kind of difference, they need to differ on just one dimension.