The Asshole Survey

First off, I apologize for the severe lateness in this post. I shall try to post more frequently (hopefully!).

Going back to the subject of assholes, I often wonder what makes a person an asshole. Using less statistically forceful language, I wonder what factors are associated with assholeness. A previous post of mine, The Search for the Elusive Asshole, talks about why we need to find ways to identify assholes and therefore pick out the non-asshole, good men in the world, if they do indeed exist.

So, let's come up with some ideas of what characteristics could be associated with assholeness (note that I say 'associated' because we're not talking about actual causal effects at this stage). We have to be careful not to list traits that are themselves characteristics of assholenes, like being mean, chauvinistic, etc. What we are looking for is a trait that isn't itself indicative of assholeness, but is something that assholes often tend to be.

Lets take, for example, drinking. Now, drinking itself doesn't make a person an asshole (unless they're a mean drunk, but that's another story), but perhaps assholes are more likely to drink more than non-assholes. Seems plausible, no?

Now it's time to look at the survey data:

Asshole rating is on a scale of 1-5 where 1 is not an asshole at all, 3 is neutral, and 5 is total asshole. The numbers in parentheses are standard deviations for all the geeky types who want to see it.

The mean asshole rating actually shows that men are distributed around the middle of the spectrum, which is neither total asshole or total saint. Actually, the average of 2.97 means that men are on average slightly less assholeish than 3, though statistical tests show that this difference is not significant.

So, now it's time to do all the real fancy statistical work. The table below lists the results of several regressions:

Ok, now I will try my best to explain all these numbers in layman's terms. The numbers reported are the coefficients of each variable (drinks, weight, smoker, etc.) and the number in parentheses are the standard deviations. The asterisk indicates the coefficient is significant at the 95% level. I'm not going to go into what statistical significance means, let's just say at 95%, we're pretty damn sure the answer is close to the number we have.

The first regression shows how the asshole rating changes for each additional drink the guy consumes in one week. This regression does not taking into account any other factors that might influence both drinking and assholeness that would affect our results (something that is called bias in statistical jargon). That's why all the other variables are listed as (-) because they weren't included. What this shows is that on average for every 1 drink per week a guy consumes, their asshole rating goes up by 0.173. What the "constant" term means is that for a guy who consumes 0 drinks, they're average asshole rating is 2.118. So a guy who drinks 15 drinks a week goes from a pretty nice guy to a total jackass. Pretty crazy huh?

Now, you must be thinking that there must be some factors that would skew these results. Regression 2 goes one step further and takes into account some variables that might have an effect on the coefficient of drinks. It seems that smoking has a rather large association with being an asshole, but this result is unfortunately not significant at the 95% level. Although most of the coefficients here are not significant at the 95% level, controlling for them is still important, because we can see that the change in the coefficient of drinks means that there other characteristics bias the results and by controlling for them we filter them out.

One interesting possibility is that the girls they date may be associated with assholeness. After all, there's the whole "dating bad boys" phenomenon, and some girls just have really bad taste (for the sake of confidentiality, this survey will not be used to name names). I'm not going to say that some girls make guys assholes or that assholes will only date some girls, that's a whole different can of worms I'd rather not open, for fear of my safety. Anyway, in order to account for the possible "girl-factor", if one actually exists, the third regression also includes girl fixed effects, thereby controlling for each girl's own unique characteristics. Interestingly enough, this makes the coefficient on drinks even greater, suggesting that the girl-factor does indeed exert some bias into the results.

So, what's the conclusion from all of this? I'd be hesitant to conclude with any certainty that more drinking causes a guy to become more asshole-like. However, there does seem to be some positive association between how much a guy drinks and how likely he is to be an asshole. Whether this is the effect of the drinks itself or some other omitted variable is yet unknown. Clearly more research is necessary for the good of science.