How to date pretty people if you are ugly (using maths)

I know the feeling of a woman not giving value to your abilities with maths. You are intelligent, funny and honest, but that cute blonde girl that you like seems not to care about that. She keeps dating gym boys which, at the end, break her heart. This human behavior is called assortative mating, and it says that people tend to date with people who have similar physical traits, such as skin color or weight.

Lucky you, maths are going to help you with that (small) problem. I have to be honest: I love the title of this post. I know, it sounds like total clickbait, but it’s not!

In this short but insightful text, I will review the paper called “Longer Acquaintance Predicts Reduced Assortative Mating on Attractiveness”, by Hunt et al. In their paper, the authors perform a study about the correlation between the difference in the beauty of the members of a couple and try to correlate it with other objective and measurable data.

Measuring couples beauty

The first step to correlate beauty with any other factor is, of course, to measure beauty. However, it may be difficult, as it’s an abstract feature which depends on the evaluator.

In the study of the paper, a total of 167 couples were assessed by different raters which gave independent punctuation to each member of the couple, ranging from -3 to +3. In order not to introduce a bias, the raters didn’t know anything about that person’s couple.

For a certain rater, we could plot the rating of one member of the couple against the rating of the other as:

Notice that the left-most plot corresponds to a scenario where beautiful people only date beautiful people. On the other hand, the right-most plot corresponds to the opposite scenario, where beautiful people only date ugly people (like me).

However, these are the two limit scenarios. We are interested in measuring a correlation index between the partners, where a correlation equal to 1 means a perfect correlation between beauties (similar to the left-most scenario) and a correlation equal to -1 measure the opposite (similar to the right-most scenario). If beauty is a redundant factor (such as knowing maths), then the correlation should be 0.

We can make use of Pearson’s correlation coefficient for this purpose, which is defined as:

Notice that the x would correspond to the marks of the raters for partner 1, while y would correspond to partner 2. We obtain the mean rating for both partner 1 and partner 2 (x_bar and y_bar, respectively) to normalize possible gender rating decompensations.

We obtain a different Pearson correlation coefficient (r_i) for each couple i.

Beating assortative mating with… longer acquaintance

Now that we have our beauty correlation index (r_i) for each couple, we can correlate it with other objective and measurable factors.

We could, for example, try to correlate it with the difference between the wages of the partners. Another funny factor could be to measure the sexual satisfaction of the partner with a greater rating. However, I’m not so sure that those factors can be included in a serious paper… without offending anyone.

Either way, in the paper the only factor that seems to have a correlation with beauty decompensation is to know the partner before dating him/her. That’s it, you have to be in the friend zone before you can finally date that cute girl/boy you like.

From now on, every time that your popular friend laughs at you because you are in the friend zone, tell him that you are just following math rules.

Credits to Enric Monte, my professor of Machine Learning at the Polytechnic University of Catalonia, for his special class on this topic.