Can you predict compatibility with an algorithm? Can you quantify love?
Baylor has a reputation as a “Ring by Spring” school. But what about those without a BU beau? On November 5 at 5:08pm, 1,622 Baylor students received an email with the name of one other Baylor student, informally entering an agreement to marry that person if they remain Ring-less by Spring.
These emails were sent out by the Marriage Pact, a matchmaking organization that uses a unique questionnaire and algorithm to identify your most compatible match (the history of the Marriage Pact is discussed in a previous article).
Ryan Marsh, a sophomore at Baylor, met up with his match last week.
He decided to fill out the Marriage Pact questionnaire “just for fun” but “didn’t expect anything to come of it.” He met up with his match at George’s and said he had a fun time. Ryan said that the results of the algorithm this year were “almost perfect, which is scary. We have the same cat.”
“You have the same cat?” I asked.
Ryan continued, “I emailed her and she didn’t email me back for a whole day because she was adopting a cat. I’m a very big cat person, so I asked for pictures of the cat and she sent them to me and we have the same cat.”
Including cats, Ryan and his match had a lot of similar interests, but Ryan didn’t think the questionnaire was completely perfect. “They didn’t really ask a lot about your personality, just your beliefs and what you were studying in school. It was more observable traits rather than internal traits,” such as interests or personality.
Ryan suggested adding questions like “Why did you choose your major?” “What’s your favorite childhood memory and why?” or “What’s your favorite TV show and what makes it your favorite?” which he says would be more difficult to put through an algorithm.
Overall, Ryan said, “I can see us being good friends, but I can’t see us being anything more than friends,” but he is open to taking the questionnaire again next year for fun.
Ryan’s suggestions raise some interesting questions about the Marriage Pact’s methods of data classification. Can you really predict compatibility with an algorithm? How do you find one set of questions that works for everyone? For these questions, I went to an expert: Baylor professor and psychometrician Dr. Grant Morgan.
Morgan describes classification as “a very natural process that we as humans engage in because we are surrounded by so much data all the time that there’s no way we could attend to it all. Classification is really a data simplification process.” (Though this, also, he notes, is a simplification of the subject.)
Morgan teaches graduate level Quantitative Methods and researches psychometrics and classification. He also goes to HEB for his groceries, where he puts classification into action to locate an unusual ingredient.
“I went to the HEB on Hewitt Drive in Woodway, and I needed tahini. I’d never bought tahini before and so I thought as I walked into the store, ‘Where is tahini?’ If grocery stores weren’t structured the way they are, you would just start at the place closest to the door and just walk up and down aisles until you saw tahini.”
First, Morgan asked, “What is it? What things is it similar to? It’s not something we eat a lot in the American diet, so I’ll go to the aisle labeled ‘International.’”
He went there, but didn’t see it. Then he pondered further, “What is tahini? What is it exactly? It’s roasted sesame seeds. It had been ground up with probably a little oil, let’s see, roasting things and grinding them up—sounds like peanut butter.” (On the nut butter aisle, he did in fact discover the sought after tahini).
These qualities (in the case of tahini, roasted and grinded nut butters) are examples of “salient features of some observation that help us to classify” tahini, or another subject of interest. For the Marriage Pact, the subjects of interest are the people profiled through the questionnaire.
“I think that [The Baylor Marriage Pact] has got a kind of metaphorical grocery store. They’ve got the items all set up or the shelves all set up. They’re administering this instrument to figure out which shelves to put people on.”
Morgan says, “I think that [The Baylor Marriage Pact] has got a kind of metaphorical grocery store. They’ve got the items all set up or the shelves all set up. They’re administering this instrument to figure out which shelves to put people on.” He added, “I really don’t like that metaphor, but I think it actually kind of fits.”
“And everyone’s trying to find their tahini,” I added.
“That’s the title of this article—Finding Your Tahini,” Grant said.
In the case of the Marriage Pact, “There’s this set of items that are intended to be manifestations of all these different aspects of a person’s personality, values, interests, those kinds of things” which is a sort of “profiling.” Morgan added, “That verb (profiling) has gotten a negative connotation over time, but mathematically that’s what is happening—you’re creating these profiles based on what we know from the empirical literature.”
Classification is a process used in psychometrics, another research focus of Morgan. And what is psychometrics? “Psychometrics as a discipline is focused on how we can make decisions about latent variables, [which are] variables that we believe exist but can’t directly measure. And the way we do that is we develop these instruments that are hopefully reflective of the latent variable.”
The Marriage Pact is one of those instruments, and in this case, the latent variable is theoretical compatibility. How then does the Marriage Pact identify the salient features by which to profile, classify, and arrange those who take the survey? Here’s what the Marriage Pact’s Data Principles and Practices have to say:
“Based on your values, the Marriage Pact algorithmically ‘interviews’ all other participants at your school on your behalf to find the best person with whom you could make a marriage pact…Your long-term compatibility with a person depends not on your surface-level traits, but on your core values—what you care about deep down.”
Your long-term compatibility with a person depends not on your surface-level traits, but on your core values—what you care about deep down.
Marriage Pact
To acquire this “values data,” the Marriage Pact asks 50 questions, each of which can be answered on a scale from one to seven. Questions from this year’s survey include:
- I find politically incorrect humor funny (1- strongly disagree; 7 – strongly agree)
- It’s okay that my partner drinks
- It is more important to protect someone’s feelings than to tell the truth
- I’m comfortable with my child being gay
- Exercising is an important part of my identity
- My partner can be “just friends” with an ex
- Are you smarter than most people at the university?
- It’s important my child(ren) be raised outside of the U.S. for some period of time
- I always vote (1 – never; 7 – even in local elections)
- The world needs (1 – more realism; 7 – more imagination)
How are these responses scored? The Marriage Pact is built on a variation of the Deferred Acceptance Algorithm (a foundation shared by the dating app Hinge), which “seeks to orchestrate a stable match between two equal-sized groups of participants, typically men and women. The objective is to secure a stable matching, where no two participants would desert their current partners for one another.” (However, this does not exclude the possibility of a stronger unrequited desire among matches.)
In addition to empirical weighting (which psychometricians call “loadings”), Morgan says, “There can also be a weighting imposed on scoring so that if ‘You believe in God’ is one item and ‘You like macaroni and cheese’ is another item, they probably shouldn’t be weighted equally if the purpose of the answers is to find a life partner, hopefully.” We don’t know if the Marriage Pact has imposed importance onto various questions. Is one’s stance on politically incorrect humor as important as one’s ideal parenting approach?
Even with a seemingly objective questionnaire and algorithm, Morgan says, “The results of this are an expression of some value system.” The Marriage Pact acknowledges the importance of values, claiming to determine matches “based on your values,” but really it determines matches based on your values within their values. The framework of their value system is formed by the questions they find important. “So I hope people are aware that they’re accepting a value system by accepting the results. I’m not saying that’s bad, I’m just saying be aware of it.”
Even with a seemingly objective questionnaire and algorithm, Morgan says, “The results of this are an expression of some value system.”
Whether determining which questions hold the most importance, or what results to accept, we are faced with what Morgan calls a balancing act: “Do you let the data make every decision or do you say ‘This is what we believe’?”
This is easier said than done. According to Morgan, “Even doctoral students are hesitant to impose their values on decisions. They want to be the translator for the data.”
Most of us aren’t PhD students, but many of us still face the temptation to offload the responsibility for our most important decisions.
Morgan encourages his students to resist that temptation. “Just this week in class, I reminded students who are hesitant to make decisions: ‘You’re pursuing a terminal degree, you’re going to have to make decisions. And is it scary? Sometimes it is. Is it hard? Sometimes it is. But you wanted this. That’s what comes along with being an expert.’”
The Marriage Pact is an easy opportunity to give up our agency, but I believe it is better to have courage and make our own decisions. Even the wrong ones can be instructive.
After an illuminating breakup, I formulated a questionnaire of my own to reflect my revised understanding of my values and preferences. I identified the qualities I was looking for and transformed each data point into a question, resulting in my The Questionnaire (which I do use on dates). I believe that one of the many merits of dating is how it teaches you about your own preferences, character, and values. In searching for someone to complement ourselves, we are also searching for ourselves.
As someone who took the Marriage Pact questionnaire “for the bit,” I honestly don’t really like the way this instrument turns people and matchmaking into economics (though I am a big fan of my own DIY psychometrics). Even though it has some success and the end decision does belong to the users, I disagree with this system on principle. One of the selling points of the Marriage Pact (and online dating platforms) is the opportunity to allow data to make decisions for you. Whether people are truly handing over the keys of their dating life to economics or not, I take issue with the attitude of submission. I do not want to give up the search just to find myself in the machine. Instead, let us own our agency, make decisions, and become the expert interpreters of our lives’ date-a.
