Amie Gordon, Assistant Professor of Psychology

Amie Gordon has always been interested in relationships. As a preteen, she stole her older sister’s romance novels. But even at that age, such surreptitious reading mostly led to frustration: Why did the characters make such bad decisions? Why didn’t they just talk to each other? Why, in essence, were they so bad at relationships?

Now an Assistant Professor of Psychology, Gordon continues to ask those same sorts of questions, only now about real relationships: How and why do romantic relationships form? How are they maintained? And why do they often fail? Recently, for example, she has explored how various stressors—such as work stress, lack of sleep, or parenting demands—affect how people perceive and respond to their partners.

But studying romantic relationships, especially at crucial turning points such as first dates or the transition to parenthood, is logistically difficult. How can investigators find enough couples experiencing these moments to establish a decent sample size? And what about finding couples who are willing to have their relationships tracked over time? Ideally, what researchers need is to have both a very large pool of couples and a way to monitor their developing relationships closely. A flyer on the department bulletin board is not going to cut it.

Well, it turns out that technology may have already provided a tool to do just that: dating apps. Apps such as Tinder, Match, or Bumble already gather enormous amounts of data (including both quantitative data such as who matches with whom and qualitative data such as the contents of private messages) about vast numbers of users. 

But existing dating apps were designed for business, not academic research. While academics have collaborated with commercial app companies in the past, it is not an ideal partnership. The goal of a commercial app company is to keep users engaged (and thus either paying for the service or viewing ads). That means creating as hassle-free an experience as possible. But psychologists need data—including information such as users’ subjective thoughts about their partners and relationships. Such data is only gatherable via surveys, questionnaires, or interviews. And nobody wants to fill out a survey, especially in an app they just paid for.

So Gordon’s lab is designing their own dating app. By explicitly positioning the app as a research tool, Gordon’s team is letting users know up-front there will be surveys—and that their data will at least be helping to advance the science of compatibility rather than, say, helping some company display more ads for Valentine’s flowers.

The idea to develop an app was suggested to Gordon after a talk she gave in 2021. She thought it sounded like a great idea for someone but not necessarily a good fit for her work. Existing research drawing on dating apps has (naturally) focused mostly on how relationships are initiated, and Gordon is more interested in how existing relationships are maintained. But graduate students in her lab do study relevant phenomena like similarity, compatibility, and chemistry—phenomena which are still poorly understood by science, in part because of how difficult it is to study them.

“To examine those things, you really have to catch people before their relationships begin, see what they're like at that time, and then see what they're like in the relationship,” Gordon explains. “You have to see how they change over time. I realized that a dating app is one of the few practical ways to gather that kind of early data—even from before people first meet.”

Soon Gordon began to see the lack of a dedicated, research-focused dating app as a major missed opportunity for studying not only how (and why) relationships form but also how they are maintained over time. After all, just because previous app-based studies have focused mostly on relationship formation does not mean that future studies must do the same. 

“I’m very excited about it all because I think it’s going to result in an amazing data set,” Gordon says. “After we get people to join the app, we will be able to see the whole relationship formation process. And once we have those couples recruited, we will provide questionnaires and other ways to keep track of their thoughts about their relationships and their partners. Then, because we can monitor the development of their relationships, we can recruit them for other studies focused on ongoing relationships and see how they develop over time.”

Partial development wireframe from the in-progress dating app

But developing an application like this one is a complex process. Before software coding and testing (which will be handled by U-M’s in-house software developers) can even begin, the team first needs to plan out exactly what the app needs to do and how it will function for users. Even for a noncommercial app, that often means striking a balance between user preferences and data-collection needs.  After all, there will be no data if no one will use the app. For example, Gordon and her team are interested in what factors make people compatible—as well as whether factors that actually determine compatibility are the same things people believe are determinate. Are people’s supposed “dealbreakers” really dealbreakers, or do they just think they are? To explore those questions, researchers would ideally need complete control over which potential matches users are shown to users. But many users are unwilling to use an app that will not allow them to filter potential matches by, say, age or political affiliation.

The team’s current compromise is to both give users the ability to opt in or out of filters and to build in a certain amount of time where all users are forced to go without them. “When people open the app, they are given the choice: Do you want to curate your matches using filters, or do you want to let us do it for you? We will then have a blurb explaining why social science says you should probably let us do it (prior research suggests that people don’t actually know what they want). But we are also going to have something called Serendipity Sundays, where on Sundays everyone is forced to go without filters and let us show them any potential matches we want.”

As Gordon and her team continue to plan out various features, the U-M software engineers are beginning to build the user-facing side of the app. Once the application is developed, the next challenge will begin: Getting enough people (but not too many, at least at first) to use it. For a brand-new app like this one, that brings yet another seeming catch-22: On the one hand, you need a small enough pool of users to make beta testing and bug fixing manageable. On the other hand, the app needs a sufficiently large sample size to operate as intended. After all, a dating app cannot function for long without enough potential dates to keep users swiping.

Gordon plans to pilot the app to a group of students (perhaps the Greek system) in Fall 2023. Their goal is to then roll it out to all U-M students (across all three campuses) in Winter 2024—and, if things go well enough, possibly even expand its reach outside of U-M in years to come. 

Either way, the application will provide researchers with a novel way to follow large numbers of new and established relationships closely—which, for Gordon, makes it worth overcoming the many challenges.

“This is a unique opportunity to see relationships from before they even begin all the way through to the end,” she says. “We can potentially continue to follow up with people in the future and learn a lot of things that have been very difficult to learn in the past. For example, we still don’t know if the kind of ‘chemistry’ that happens early in relationships is the same as (or even closely related to) the things that matter to people three, five, or even 20 years in. Previous research has looked at some of those things in isolation, but never together. This is potentially a feasible way to capture the complete lifespan of a relationship, and I think it’s going to be very cool.”