Large datasets provide intriguing — and concerning — insights into who we are drawn to and how important this is for our personal happiness.
This article is a translation. resume from the book by Seth Stephens-Davidowitz “Don’t trust your intuition: using data to get what you really want in life.”
Who to marry?
This could be the most important decision in a person’s life. Billionaire investor Warren Buffett certainly thinks so. He refers to whom you marry as “the most important decision you make.”
And yet, people rarely turned to science for help with this crucial issue. To be honest, science has little to offer. Researchers studying relationship science have tried to find answers. However, it turned out that gathering large samples of couples was difficult and expensive. Studies in this field typically relied on tiny samples, and different studies often showed contradictory results. In 2007, the distinguished scholar Harry Reis from the University of Rochester compared relationship science to a teenager: “stretched out, sometimes unruly, and perhaps more mysterious than we would like.”
But several years ago, the young, energetic, highly curious, and brilliant researcher Samantha Joel decided to change that. Like many in her field, Joel was interested in what guarantees successful relationships. However, she had a noticeably different approach. Joel didn’t just gather a new tiny sample of couples. Instead, she combined data from other existing studies. Joel reasoned that if she could merge data from existing small studies, she would have a large dataset and enough information to reliably determine what predicts relationship success and what does not.
Joel’s plan worked. She hired a large number of researchers to collect data on relationships—her team included 85 other scientists—and was able to create a dataset of 11,196 heterosexual couples.
The size of the dataset was impressive, as was the information contained within it. For each couple, Joel and her team of researchers had metrics on how happy each partner was in their relationship. They also had data on everything that could possibly be measured about the two people in those relationships.
The researchers had data on:
- demographic data (for example, age, education, income, and race)
- appearance (for example, how attractive did others rate each partner?)
- sexual preferences (for example, how often did each partner want sex? How adventurous did they want that sex to be?)
- hobbies and interests
- mental and physical health
- values (for example, their views on politics, relationships, and parenting)
- and much, much more
In addition, Joel and her team not only had more data than anyone else in the field, but they also had the best statistical methods. Joel and some other researchers mastered machine learning, a subset of artificial intelligence that allows modern scientists to uncover subtle patterns in large datasets. One could call Joel’s project “Marriage with Artificial Intelligence,” as it was one of the first studies to use these advanced methods to attempt to predict happiness in relationships.
Having formed her team and gathered and analyzed the data, Joel was ready to present the results — the results of perhaps the most exciting project in the history of relationship science.
Joel planned a talk in October 2019 at the University of Waterloo in Canada with a straightforward title: “Can We Help People Choose Better Partners?”
So, can Samantha Joel, teaming up with 85 of the world’s most renowned scientists, combining data from 43 studies, analyzing hundreds of variables collected from over 10,000, and using modern machine learning models, help people choose the best romantic partners?
No..
Samantha Joel told me in a Zoom interview that the number one lesson—and the most surprising one—from this data is how “unpredictable relationships seem.” Joel and her co-authors found that demographic data, preferences, and values of two people surprisingly have little impact on whether those two individuals will be happy in a romantic relationship.
And here it is, folks. Ask AI to find out if two people can build a happy life together, and it will be just as helpless as the rest of us.
Of course, it sounds like disappointment. Does data science really have nothing to offer us in choosing a romantic partner, perhaps the most important decision we will face in life?
Not quite. To be honest, Joel and her co-authors’ machine learning project offers important lessons, even if computers’ ability to predict romantic success is worse than many of us might have assumed.
To begin with, although Joel and her team found that the strength of all the variables they gathered to predict a couple’s happiness was surprisingly low, they identified a few partner-related variables that at least slightly increase the chances of being happy. More importantly, the astonishing complexity of predicting romantic success has contradictory implications for how we should choose romantic partners.
Think about it. Many people certainly believe that many of the variables studied by Joel and her team are important when choosing a romantic partner. They fiercely compete for partners with certain qualities, believing that these qualities will make them happy. If, on average, as Joel and her co-authors found,Many traits that are most sought after in the sexual market do not correlate with romantic happiness, which suggests that many people are looking for partners in the wrong way.
This brings us to another age-old question that has also recently come under scrutiny with the use of new revolutionary data: how do people choose a romantic partner?
In recent years, other research teams have studied online dating sites, sifting through large new datasets about the traits and characteristics of tens of thousands of single people to determine what predicts their romantic appeal. The results of the study on romantic desirability, unlike the research on romantic happiness, were conclusive. While data processing and analysis experts found it surprisingly difficult to identify the qualities in romantic partners that lead to happiness, data specialists discovered that defining the traits that are as attractive as catnip to the sexual market was remarkably easy.
A recent study essentially showed that it is possible not only to predict with high accuracy whether someone will swipe left or right on a specific person on a dating site, but also to astonishingly predict the amount of time a person will spend swiping. (People generally take longer to find someone who is close to their threshold of acceptability for dating.)
Another way to say all of this: XgoodIt’s difficult to predict romantic partners using data.DesiredRomantic partners can be easily predicted using data. This suggests that many of us are searching for a life partner in the wrong way.
So, what traits make people desirable to others?
Well, the first truth about what people seek in romantic partners, like many important truths about life, was articulated by a rock star long before scientists figured it out. As Adam Duritz of Counting Crows said in his 1993 masterpiece, “ Mr. Jones “We are all looking for ‘something beautiful.’ The usual attractiveness of a partner is the number one predictor of how many messages someone receives, for both men and women. We are also looking for:”
- someone tall (if it’s a man)
- someone of the desired race (although most will never admit it)
- someone wealthy
- someone from law enforcement (for example, a lawyer or a firefighter), if male
- someone with a sexy name (for example, Jacob or Emma)
- and someone just like us (people encounter someone with the same initials 11.3% more often)
It’s striking that alarming data from online dating sites indicates that lonely people are predictably attracted to certain qualities. But should they themselves be drawn to these qualities? If you resemble the typical single person, you are likely to click on individuals with the most desirable traits, according to researchers. But are you approaching dating the right way? Or are you doing everything wrong?
Let me remind you that I previously discussed the research by Samantha Joel and her co-authors. They found that it is surprisingly difficult to predict whether a person is happy with a romantic partner based on a long list of characteristics. There is no set of traits that guarantees romantic happiness or excludes it. And no algorithm in the world can predict with incredible accuracy whether two people will be happy together.
Nevertheless, certain traits had some predictive power, with factors that at least slightly increased the chances of a person being happy in their romantic relationships. Now I will talk about what…indeedpredicts romantic happiness, and how little this has to do with the qualities that people look for in a romantic partner.
Let’s say there is a person, John, and he is Sally’s partner. You want to predict whether John is happy in the relationship. You can ask John and/or Sally any three questions about themselves and use that information to predict John’s relationship happiness.
What questions would you like to ask? What would you like to know about the two members of this couple?
According to my reading of the study by Joel and her co-authors, as well as some other research in the field of relationship science, the three best questions to determine whether John is satisfied with Sally will have nothing to do with Sally; in fact, everything will be related to John. The best questions to predict John’s happiness with Sally might look something like this:
- “John, were you happy with your life before you met Sally?”
- “John, didn’t you have depression before you met Sally?”
- “John, did you have a positive effect before you met Sally?”
Researchers found that people who answered “yes” to such questions are significantly more likely to report being happy in their romantic relationships. In other words, a person who is happy outside of their relationship is much more likely to be happy within it as well.
Moreover, and this was quite striking, the way a person answered questions about themselves predicted their happiness in relationships about four times more accurately than all the traits of their romantic partner combined.
Of course, the realization that happiness outside of relationships can have a huge impact on happiness within those relationships is hardly a revolutionary idea. Think about this saying that was published in Daily Inspirational Quotes: “Nobody can make you happy until you’re happy with yourself first.”
This is the kind of quote that often makes cynical data fans like me roll their eyes. However, after reading the work of Joel and his co-authors, I am convinced that this quote is largely true.
This relates to an important moment in a data-driven life. We, data enthusiasts, can get very excited when we learn about a discovery that contradicts common beliefs or clichéd advice. It taps into our natural desire to know something that the rest of the world doesn’t. But we, data fans, also need to accept when the data supports conventional wisdom or well-worn advice. We must be ready to go where the data leads us, even if the conclusions are similar to those presented in Daily Inspirational Quotes.
So, as discovered by a group of 86 researchers and the author of Daily Inspirational Quotes, personal happiness outside of relationships is undoubtedly the most important predictor of happiness in romantic relationships. But what else predicts romantic happiness, aside from one’s prior mental state? What partner qualities are associated with romantic happiness? Let’s start with the partner qualities that are the least predictive of romantic happiness.
Machine learning models have found that among more than 11,000 long-term couples, the partner traits listed below are the least predictive of happiness with that partner. Let’s call these traits the irrelevant eight, as the likelihood that partners will ultimately be happy in their relationships is roughly the same when they pair up with people who have any combination of these traits:
- Race/ethnic affiliation
- Religious affiliation
- Growth.
- Class
- Physical attractiveness
- Previous marital status
- Sexual preferences
- Similarity with oneself
What should we do from this list of the Irrelevant Eight? I was immediately struck by the coincidence between the list of irrelevant features and another data-driven list that was discussed in this chapter.
Let me remind you that I previously discussed the qualities that make people the most desirable as romantic partners, according to Big Data from online dating sites. It turns out that this list of qualities that are most valued in the dating market almost perfectly aligns with the list of partner qualities that do not correlate with happiness in long-term relationships, according to a large dataset analyzed by Joel and her co-authors.
Let’s consider, for example, ordinary attractiveness. Beauty, as you may recall, is the most valuable trait in the dating market. Hitch, Hortasu, and Ariely, in their study of tens of thousands of singles on an online dating site, found that those who receive messages and get responses to their messages can largely be explained by their conventional attractiveness. However, Joel and her co-authors, in their study of more than 11,000 couples in long-term relationships, discovered that the ordinary attractiveness of a partner…it doesn’t matterfor predicting happiness in personal life. Similarly, tall men, men in sexually appealing professions, and representatives of certain races are highly valued in the dating market. But ask thousands of married couples in long-term relationships, and you will find no evidence that people who have managed to pair up with partners possessing these desirable qualities are happier in their relationships.
If I had to summarize the most important discovery in the field of relationship science made possible by this big data research in one sentence, it would be something like this (let’s call it the First Law of Love):In the dating market, people fiercely compete for partners with qualities that do not enhance their chances of romantic happiness.
Moreover, if I had to define qualities that are highly desirable, even if they don’t lead to long-term romantic happiness, I would describe many of them as shining qualities. These qualities immediately catch our attention. For example, almost all of us are quickly drawn to what is conventionally beautiful. However, as the data shows, these eye-catching, shining qualities do not impact our long-term romantic happiness. The data indicates that single people are predictably deceived by the allure of shine.
Из.the book “Don’t Trust Your”“Intuition” by Seth Stephens-Davidowitz. Copyright © 2022 Seth Stephens-Davidowitz. Original article at English Текст для перевода: ..