Re: Our Relationship
Using statistics to quantify the most qualitative emotion: love
As a scientist who studies online-dating data, I’ve spent a lot of time quantifying how other people fall in love. I began to wonder whether it was possible to apply the same methods to my own relationship. I told myself it was for the sake of science: I was acting out of professional curiosity, and would understand others’ relationships better by putting myself under the mathematical microscope.
But it would be more honest to admit it was also because I missed my boyfriend. After spending three years at college with him, I had left to study for a year at Oxford; this was my equivalent of flipping through photo albums.
But what data to use? We rarely text or take pictures. But in the four years since we began dating, we’ve exchanged an average of four emails a day, which works out to more than 5,500 emails. If we had just typed out literature to each other, we would’ve recently completed Harry Potter and the Goblet of Fire, though we won’t finish Infinite Jest for another six years or In Search of Lost Time for another 19. When I told my boyfriend I wanted to statistically analyze our emails, we had the following conversation:
Him: I think you should ask my permission to do that.
Me: I wouldn’t ask your permission to read the emails. Why should reading them using a computer be any different?
Him: You’re going to find some weird pattern and break up with me.
Me: Either that will be warranted by the data, in which case it’s a good thing, or it won’t, in which case I’m a bad statistician. Are you saying I’m a bad statistician?
This is what scientists call “obtaining consent.”
It was dark and drizzling when I walked back alone to my Oxford dorm room, curled up around my laptop, and dove into the digital record of our relationship. I was unsurprised to find that our emails became more frequent after I left for England. But I felt a jolt when I discovered that I sent my boyfriend far more emails than he sent me.
I closed my program, made myself a cup of tea, called my boyfriend, and asked him why, according to the data, I missed him more than he missed me. He said that wasn’t true, he just preferred to talk on the phone rather than send me emails. I went back to the data to see if it substantiated his claim, and indeed it did: He used “call me” and “phone” more frequently in his emails.
Having avoided one potential breakup, I returned to the data and looked at how the average length of our emails changed over time. I found large spikes corresponding to the first three times we were apart for our university’s spring, summer, and winter breaks. These lengthy emails turned out to be, unsurprisingly, the sort of pour-your-soul-out messages that accompany first infatuation. The content of the emails changed over time in other ways as well. For example, we used the word “promise” more frequently early in our relationship, often to make the sort of charming but trivial pledges that build trust—“I promise not to kill you,” or “I promise never to make you go to a yacht club.” On the other hand, we began to use nicknames and endearments only later in our relationship—promises replaced by pet names.
Then I wondered if differences in our personalities would show up in our emails. I compared the words I used with the words he used; this revealed that, contrary to gender stereotypes, I am probably more aggressive. I am responsible, for example, for more than 95 percent of the profanity in our emails. He is much more likely to use the phrase “I am not sure,” and is also responsible for 60 percent of the incidences of “sorry.” I have a penchant for bleaker topics, and am more likely to mention “pain,” “cancer,” and “suicide.” I am also more likely to make sweeping generalizations about men, as evidenced by my more frequent use of “boys” and “male.”
We each bring up our interests: He, the quadrilinguist, mentions Greek, Latin, and Italian. I use words related to statistics. Our language is distinctive in other ways. He, the New Englander, is much more likely to use the word “dandy” (as in Yankee Doodle); I, who when comfortable with someone begin talking like a frat boy, am much more likely to use the word “bro.”
To avoid getting dumped, I will stop sharing details of our emails and will instead share two larger lessons I learned about love. The first is that statistics can be unexpectedly, painfully powerful. I had long known the joy of slicing out truth with a statistical scalpel, but here the heart I’d cut into was my own. Why does my boyfriend apologize more than I do? Why have our emails gotten shorter? What if I still want promises, not just pet names? Another statistician of love, the founder of the dating site OkCupid, once said that analyzing people’s relationships made him “very grim” because he had to “embrace the darkness.” I always found this a bit melodramatic, but perhaps he simply empathized better than I did with the lovers who went under his knife.
I was originally planning to make an app to allow anyone to analyze their own relationship, but it isn’t clear to me I’d be offering something worth having. There are far more unpleasant things you could find. What if your partner’s emails are less affectionate on days when they have meetings with that attractive co-worker? What if you no longer send them flirtatious emails, or only tell them they’re attractive after 10 p.m. on Fridays and Saturdays, when you tend to be drinking? What if you discussed Plato and Proust with your ex, and with your current partner you only talk about what’s for dinner? (I did not perform a comparative analysis across my relationships because I figured my exes dislike me enough already.)
You might argue that if these things are true, it is better to be aware of them. But I am not sure that love is best looked at so clear-eyed. Relationships are weird: pas de deux that in the moment are endearing seem bizarre when coldly quantified; full of truths that might be softer if only dimly perceived. If you never thought of your partner as unaffectionate until you learned they say “I love you” 20 percent less frequently than you do, what benefit have you gained?
The second lesson I learned is about the limits of statistics. My relationship is not fully captured by my emails: What I remember are the moments themselves, not their digital shadows. The entire email record of my relationship can itself be attached to an email. It is but a hundredth of a hundredth of a hard drive, a pinch of electron fairydust that cannot contain four years of tears and touches. And my emails are not fully captured by my algorithms, which would react the same way if I took every carefully crafted message and scrambled the words into random order. Writing this piece alone, what I want is my boyfriend; what I have are some line graphs. If I lose this much when I study a single relationship on which I’m an expert, God knows what I’m losing when I apply the same approach to tens of thousands of people I’ve never met.
So those are the twin and opposite warnings I’d pass on to those who would reduce love to a line graph: You don’t know what you’re missing, and you don’t know what you’ll find. Perhaps the moments that count most can’t—or shouldn’t—be counted.