Web matchmakers work hard on the science of love

The science of falling in love
Lori Gottlieb

I've been sitting in Neil Clark Warren's office for less than 15 minutes when he tells me he has a guy for me. It isn't surprising that the avuncular 71-year-old founder of eHarmony.com, one of the US's most popular online dating services, has matchmaking on his mind. The odd thing is that he is eager to hook me up without having seen my eHarmony personality profile.

I'd come to the eHarmony headquarters in Pasadena, California, to learn more about the sites scientifically proven and patented compatibility matching system. Apparently, the science wasn't working for me. The day before, after I'd taken the company's exhaustive (and exhausting) 436-question personality survey, the computer informed me that of the approximately 9 million eHarmony members, more than 40 per cent of whom are men, I had zero matches. Not just in my city, state, region, or country, but in the entire world.

So Warren suggests setting me up with one of his company's advisory board members, whom he describes as brilliant, Jewish and 38-years old. According to Warren, this board member, like me, may have trouble finding a match on eHarmony.

"Let me tell you why you're such a difficult match," Warren says, facing me on one of his bright floral sofas. He starts running down the backbone of eHarmony's predictive model of broad-based compatibility, the so-called 29 dimensions (things like curiosity, humour, passion, intellect), and explains why I and my prospective match are such outliers.

"I could take the nine million people on our site and show you dimension by dimension how we'd lose people for you," he begins. "Just on IQ alone - people with an IQ lower than 120, say. OK, we've eliminated people who are not intellectually adequate. We could do the same for people who aren't creative enough, or don't have your brilliant sense of humour. See, when you get on the tails of these dimensions, it's really hard to match you. You're too bright. You're too thoughtful. The biggest thing you've got to do when you're gifted like you are is to be patient."

After the over-the-top flattery wears off - and I'll admit, it takes an embarrassingly long time - I tell Warren that most people I know don't join online dating sites to be patient. Impatience with real-world dating, in fact, is precisely what drives many singles to the fast-paced digital meat market. From the moment Match.com, the first such site, appeared in 1995, single people suddenly had 24-hour access to thousands of other singles who met their criteria in terms of race, religion, height, weight, even eye colour and drinking habits.

Nearly overnight, it seemed, dozens of similar sites emerged, and online dating became almost de rigueur for busy singles looking for love. According to a recent Pew survey, 31per cent of all American adults (63million people) know someone who has used a dating website, while 26 per cent (53 million people) know someone who has gone out with a person he or she met through a dating website. But was checking off boxes in columns of desired traits the best way to find a soul mate?

Enter eHarmony and the new generation of dating sites, among them PerfectMatch.com and Chemistry.com. All have staked their success on the idea that long-term romantic compatibility can be predicted according to scientific principles. To that end they've hired high-powered academics, devised special algorithms for relationship-matching, developed sophisticated personality questionnaires, and put into place mechanisms for the long-term tracking of data. Collectively, their efforts mark the early days of a social experiment of unprecedented proportions, involving millions of couples and possibly extending over the course of generations. The question at the heart of this grand trial is simple: In the subjective realm of love, can cold, hard science help?

Although eHarmony was the first dating site to offer science-based matching, Warren seems like an unlikely pioneer in the field. Even though he earned a PhD in clinical psychology from the University of Chicago, in 1967, he never had much of a passion for academic research, or an interest in couples.

Fittingly, it was Warren's family, not academia, that piqued his interest in romantic compatibility. "When my daughters came along, that was a big pivot in my life in thinking about how do two people get together," he tells me. I started reading in the literature and realising what a big chance they had of not having a satisfying marriage. I started trying to look into it."

Soon he began a private practice of couples therapy with a twist. "People have always thought, wrongly, that psychotherapy is a place to go deal with problems," he says. "So when a couple would come in, I'd say, 'Tell me how you fell in love. Tell me the funniest thing that's happened in your marriage. If you want to make a relationship work, don't talk about what you find missing in it! Talk about what you really like about it."'

The basis of eHarmony's matching system sounds simple but profound. "In successful relationships," Warren says, "similarities are like money in the bank. Differences are like debts you owe. It's all right to have a few differences, as long as you have plenty of equity in your account.

"We want to put [our products] out there in a way that you'd say, 'This is common sense. This seems right, this seems like it would work.' Our idea of broad-based compatibility, I put it out there in front of you. Does that seem right?"

Whether or not it seems right on an intuitive level is almost beside the point. After all, eHarmony's selling point, its very brand identity, is its scientific compatibility system. That's where Galen Buckwalter comes in.

A vice-president of research and development for the company, Buckwalter is in charge of recruiting what he hopes will be 20 to 25 top relationship researchers away from academia - just as he was lured away by Warren nine years ago.

"Neil knew I lived and breathed research, and he had this idea to try to develop some empirically based model to match people," Buckwalter says. Relationships weren't Buckwalter's area, but he welcomed the challenge.

With the help of a graduate student, Buckwalter reviewed the psychological literature to identify the areas that might be relevant in predicting success in long-term relationships. "Once we identified all those areas, then we put together a questionnaire," he says. "It was probably close to a thousand questions. Because if you don't ask it, you're never gonna know. So we had tons of questions on ability, even more on interest. Just every type of personality aspect that was ever measured, we were measuring it all."

Because it wasn't practical to execute a 30-year longitudinal study, he and Warren decided to measure existing relationships, surveying people who were already married.

The idea was to look for patterns that produce satisfaction in marriages, then try to reproduce them in the matching of singles.

Buckwalter's studies soon yielded data that confirmed one of Warren's longtime observations: namely, that the members of a happy couple are far more similar to each other than are the members of an unhappy couple. Compatibility, in other words, rests on shared traits.

For Warren, a big question remained: What should be done with these findings? Originally, he had partnered with his son-in-law, Greg Forgatch, a former real-estate developer, to launch the business. Their first thought was to produce educational videotapes on relationship compatibility. After all, Warren had recently written his book, Finding the Love of Your Life .

"We tried so hard to make videotapes and audiotapes," Warren says. "I went into the studio and made lists. We came up with a hundred things singles need. But singles don't want education; they want flesh! They want a person."

To connect singles and create a data pool for more research, the internet seemed the best option. Based on a study of 5000 married couples, Warren put together the compatibility model that became the basis for eHarmony. Warren insisted on getting the matching system right before launching the site - and that didn't happen until August of 2000, during the dot-com bust. By 2001 he was contemplating declaring bankruptcy.

"And then," Warren recalls, "we found an error in our matching formula, so a whole segment of our people were not getting matched. It was an error with all the Christian people on the site."

This is a sensitive topic for Warren, who bristles at the widely held opinion that eHarmony is a Christian dating site. The company's chief operating officer is Jewish, and Buckwalter is agnostic. And while Warren describes himself as a passionate Christian, he worried about narrowing the site with too many questions about spiritual beliefs. Which is where the error came in.

Fortunately, a wave of positive publicity, featuring married couples who'd met through eHarmony and the naturally charismatic Warren, turned things around.

Today, eHarmony's business isn't just about using science to match singles online. Calling itself a relationship-enhancement service, the company has recently created a venture-capital-funded think tank for relationship and marital research, headed up by Gian Gonzaga, a scientist from the well known marriage and family lab at the University of California at Los Angeles.

Gonzaga thought twice about leaving the prestige of academia, but after interviewing with Warren, he realised that conducting his research under the auspices of eHarmony would offer certain advantages. He'd be unfettered by teaching and grant-writing, and there would be no sitting on committees or worrying about tenure. More important, since his research would now be funded by business, he'd have the luxury of doing studies with large groups of ready subjects over many years - but without the constraints of having to produce a specific product.

So far, the data are promising: a recent Harris Interactive poll found that between September 2004 and September 2005, eHarmony facilitated the marriages of more than 33,000 members - an average of 46 marriages a day. And a 2004 in-house study of nearly 300 married couples showed that people who met through eHarmony report more marital satisfaction than those who met by other means.

"We have massive amounts of data," Warren says. "Twelve thousand new people a day taking a 436-item questionnaire. Ultimately, our dream is to have the biggest group of relationship psychologists in the country."

But how useful is this sort of data for single people like me? Despite Warren's disclaimer about what a tough eHarmony match I am, I did finally get some profiles in my inbox. They included a bald man with a handlebar moustache, who was 14inches taller than me; a five-foot-four-inch attorney with no photos; and a film editor whose photo shows him wearing a kilt - and not in an ironic way. Was this the best science could do?

When I ask Buckwalter about this, he laughs. "The thing you have to remember about our system is we're matching on these algorithms for long-term compatibility," he says. "Long-term satisfaction is not the same as short-term attraction. A lot of people, when they see their initial matches, it's like, 'This is crap!?"'

In ads and on his website, Warren talks about matching people from the inside out. Was eHarmony suggesting that I overlook something as basic as romantic chemistry? "When we started out," Buckwalter says, "we were almost that naive. But now, eHarmony is conducting research on the nature of physical attraction.

"We're trying to find out if we can predict physical chemistry with the same degree of statistical certainty that we've used to predict long-term satisfaction through our compatibility matching. In general, people seem to be attracted to people who share their physical attributes," Buckwalter explains, noting that he has found some exceptions, like height preference. "We're still convinced that our compatibility-matching process is essential for long-term satisfaction, so were not going to mess with that," he insists. "But if we can fit a short-term attraction model on top of that, and it's also empirically driven, that's the Holy Grail."

Over at Chemistry.com, a new site launched by Match.com, short-term attraction is already built into the system. This competitor of eHarmony's was developed with help from Match.com's chief scientific adviser, Helen Fisher, an anthropologist at Rutgers University, whose research focuses on the brain physiology of romantic love and sexuality. Chemistry.com is assembling a multidisciplinary group of psychologists, relationship counselors, sociologists, neuroscientists and sexologists to serve as consultants. The company sought out Fisher precisely because its market research revealed that although a large segment of singles wanted a scientific approach, they didn't want it to come at the expense of romantic chemistry.

So what creates that chemistry? A constellation of factors, Fisher says. Sex drive, for instance, is associated with the hormone testosterone in both men and women. Romantic love is associated with elevated activity of the neurotransmitter dopamine and probably also another one, norepinepherine. And attachment is associated with the hormones oxytocin and vasopressin. "It turns out," she says, "that seminal fluid has all of these chemicals in it. So I tell my students, Don't have sex if you don't want to fall in love."

Romantic love is a basic mating drive, more powerful than the sex drive, Fisher maintains. "If you ask someone to go to bed with you, and they reject you," she says, "you don't kill yourself. But if you're rejected in love, you might kill yourself."

For Chemistry.com's matching system, Fisher translated her work with neurotransmitters and hormones into discrete personality types. "I've always been extremely impressed with Myers-Briggs," she says, referring to the personality assessment tool that classifies people according to four pairs of traits: introversion versus extroversion, sensing versus intuition, thinking versus feeling, and judging versus perceiving. Genes for the activity of dopamine are associated with motivation, curiosity, anxiety and optimism. Genes for the metabolism of serotonin, another neurotransmitter, tend to modulate one's degree of calm, stability, popularity and religiosity. Testosterone is associated with being rational, analytical, exacting, independent, logical, rank-oriented, competitive, irreverent and narcissistic. And the hormone estrogen is associated with being imaginative, creative, insightful, humane, sympathetic, agreeable, flexible and verbal.

"So I had these four sheets of paper," Fisher continues. "And I decided to give each a name. Serotonin became the Builder. Dopamine, the Explorer. Testosterone, the Director. And estrogen, I wish I'd called it the Ambassador or Diplomat, but I called it the Negotiator."

The 146-item compatibility questionnaire on Chemistry.com correlates users' responses with evidence of their levels of these various chemicals. One question, for instance, offers drawings of a hand, then asks:

Which one of the following images most closely resembles your left hand?
Index finger slightly longer than ring finger
Index finger about the same length as ring finger
Index finger slightly shorter than ring finger
Index finger significantly shorter than ring finger The relevance of this question might baffle the average online dater accustomed to responding to platitudes such as "How would you describe your perfect first date?". But Fisher explains that elevated fetal testosterone determines the ratio of the second and fourth finger in a particular way as it simultaneously builds the male and female brain. So you can actually look at someone's hand and get a fair idea of the extent to which they are likely to be a Director type (ring finger longer than the index finger) or a Negotiator type (index finger longer or the same size).

Another question goes like this:

How often do you vividly imagine extreme life situations, such as being stranded on a desert island or winning the lottery?
Almost never
Sometimes
Most of the time
All the time "Someone who answers 'All the time' is a definite Negotiator," Fisher says. High estrogen activity is associated with extreme imagination.

While other sites gather data based on often unreliable self-reports ("How romantic do you consider yourself to be?"), many of the Chemistry.com questions are designed to translate visual interpretation into personality assessment, thus eliminating some of the unreliability. In one, the user is presented with a book's jacket art. We see a woman in a sexy spaghetti-strapped dress gazing at a man several metres away in the background, where he leans on a stone railing. The sky is blue, and they're overlooking an open vista. "What is the best title for this book?" the questionnaire asks, and the choices are as follows:
A Spy in Rimini
Anatomy of Friendship: A Smart Guide for Smart People
A Scoundrel's Story

Even sense of humour can be broken down by type, with questions such as "Do you sometimes make faces at yourself in the mirror?" (people with a sense of humor do). According to Fisher, a Director likes people to laugh at his or her jokes; a Negotiator likes to be around someone funny so he or she can laugh at that person's jokes; an Explorer is spontaneous and laughs at just about anything; and a Builder, she suspects, generally isn't as funny as the others.

But how to match people up according to Fisher's four personality types, and under what circumstances, isn't so straightforward. One question, for instance, presents four smiling faces and asks: "Take a look at the faces below. Are their smiles sincere?"

Fisher says people with high levels of estrogen - usually women - have better social skills, and are better at reading other people. So users who choose the correct real smiles will be the Negotiators. This, she says, is an area where complementarity might be important. The problem with sites such as eHarmony, she believes, is that they place too much emphasis on similarity, whereas, in her view, falling in love depends on two elements: similarity and complementarity. "We also want someone who masks our flaws," she explains. For example, people with poor social skills sometimes gravitate toward people with good social skills.

Determining which works best - similarity or complementarity - may change with the circumstances. A young woman who's an Explorer, Fisher says, may be attracted to a Builder, someone who's more of a homebody, loyal, dependable and protective. "But the pair will be more compatible if their secondary personalities match - maybe they're both Negotiators underneath.

"Nobody is directly locked into any one of these temperament types," Fisher says. "That's why we provide each person with both a major and a minor personality profile. Do Explorers go well together? Do likes attract likes? Sometimes they do and sometimes they don't."

If this sounds a bit, well, unscientific, Fisher is the first to admit it.

This is why she decided to include an item on the Chemistry.com questionnaire that asks about the traits of a person's partner in his or her most successful former relationship: Was that person an Explorer, a Builder, a Director, a Negotiator?

At the same time, Fisher wants couples to be fascinated by each other early on. In other words, why waste time emailing back and forth to get to know a potential match over the course of several weeks, as eHarmony encourages its users to do, if there won't be any chemistry when they finally meet? Chemistry.com's guided 1-2-3-Meet system provides a step-by-step structure to get couples face to face as soon as possible for that all-important vibe check. Then there's a post-meeting chemistry check, where each person offers feedback about the date.

The goal is to incorporate this information into the algorithm to provide better matches, but it can also serve as an accuracy check of the data. Say, for instance, that Jack describes himself as a fashionable dresser, but Jill reports that he showed up for their date in flip-flops, cut-offs and a do-rag. If the feedback from a number of Jack's first meetings indicates the same problem, Chemistry.com will send him an email saying, "Jack, wear a pair of trousers".

When I ask Fisher how the sites scientific algorithm might change based on this user feedback, she says that perhaps the computer could pick up cues about a person's physical type based on the people he or she finds attractive or unattractive, then send that person closer matches.

It's also possible that user feedback could change the matching formula completely. "We always look at data," Fisher says. "If we find that Explorer/Builder to Director/Negotiator is working for more people, if we find the biochemistry is stronger, we'll adjust that in the formula."

Still, even a thoroughly researched biochemical model won't prevent glitches in the matching system. In Fisher's view, for example, no scientifically based site would pair her with the men she's dated, because, as she puts it, "they're all better-looking than me".

"This test doesn't pretend to be about chemistry," says Pepper Schwartz, who developed the Duet Total Compatibility System in conjunction with the two-year-old site PerfectMatch.com. "The chemistry test at Match - that's not about chemistry either. If I could concoct a test for chemistry, I'd make a zillion dollars."

A sociologist at the University of Washington in Seattle, Schwartz is the accessible, empathic, media-savvy love doctor who guides users through the treacherous dating trenches and onto the path of true compatibility.

According to the site, which calls her by the cutesy moniker Dr Pepper, Schwartz is the leading relationship expert in the nation, a woman who holds the distinction of being the only relationship expert on the web who's a published authority, as well as a professor at a major US university.

Unlike Warren, however, she neither founded the company (she was brought in by PerfectMatch's Duane Dahl), nor follows Warren's credo of simplicity. In fact, the nifty-sounding Duet Compatibility Profiler takes some complex deconstruction.

Like Fisher, Schwartz believes that both similarity and complementarity are integral to romantic compatibility. But while Fisher has more of an "it depends" attitude on the question of which of the two makes sense for a particular couple under particular circumstances, Schwartz has a more elaborately defined system, which she outlines in her latest book, Finding Your Perfect Match .

Schwartz's Duet model consists of a mere 48 questions and focuses on eight specific personality characteristics: romantic impulsivity, personal energy, outlook, predictability, flexibility, decision-making style, emotionality and self-nurturing style. On the first four, she believes, a well-suited couple should be similar; on the last four, however, a couple can thrive on either similarity or difference, provided that both people know themselves well enough to determine which works best.

Her questionnaire, she believes, will help users to think in a conscious way about who they are. As an example of the kind of introspection she hopes for, Schwartz cites the area of money. "It's a very important thing," she says, "and there's very little research on it, because nobody wants to talk about money."

When it comes to money, PerfectMatch asks users to get specific and honest about how important it is to them. They are asked questions such as "Do you feel you need to make extravagant purchases every once in a while?" Other tests generally stop at innocuous questions about whether people consider themselves fiscally responsible, but Schwartz ventures into un-PC territory with true-or-false statements such as "All other things being equal, I tend to respect people who make a lot of money more than people who have modest incomes."

I take the Duet test and am classified on the similarity scale as X, A, C, and V, that is, risk averse, high energy, cautious and seeks variety. The site then interprets the findings, which, to my surprise, rather accurately capture my personality . Yet the complementarity section of my test results - those traits on which my best match might be similar or different - reflect my temperament on only two of the four parameters. I am characterised as S, C, T and E that is, structured, compromiser, temperate and extrovert, but I'm neither a C nor a T.

Schwartz isn't ruffled by these inaccuracies. "PerfectMatch is the only scientific site out there that's completely transparent and user-operated," she says. "If you disagree with me, you can retake the test anytime and get a different profile that more accurately reflects the subtleties we may have missed. Or you can keep the same profile, but in addition to the matches we provide for you, you can do a search on your own."

This distinguishes PerfectMatch from eHarmony and Chemistry.com, she says. "In the Chemistry test," says Schwartz, "there is a question about where you'd like to live. And I chose the country. And I would - but the people I tend to prefer are in the city. So they sent me people from Bass Breath, Arizona. And there was no way I could change it! At PerfectMatch, we don't overdetermine people's answers that way."

What Schwartz is referring to is the bugaboo of all these compatibility-matching systems: nuance. Even if a site lets you choose physical characteristics like height there's no way it's going to guess your physical template. It could be lankiness in one case, it could be somebody's eyes in another. "We can't get that out of a questionnaire," she says. "Nobody can. So we say, go look at the pictures on our site, see who you find attractive, then look at their personality types and see if they're compatible [with you].

"The advantage to scientific matching," she says, "isn't to come up with some foolproof formula for romantic connection. Instead, the science serves as a reality check, as a way of not letting that initial rush of attraction cloud your judgement when it comes to compatibility."

Schwartz, who had been married for 23 years before she re-entered the dating pool, empathises with PerfectMatch users. "I know what dating is like," she says. "I'm doing it, too. You start to burn out, and you need to find a certain amount of positive reinforcement. So if we can cut down the really inappropriate personalities for you, we can help out."

Of course, before the days of Myers-Briggs and PerfectMatch and academic departments devoted to deconstructing romantic relationships, there were matchmakers. And today, despite the science, they're still thriving. One of the US west coast's largest matchmaking agencies is called Debra Winkler Personal Search, and its slogan is the opposite of scientific: "The art of the perfect match". Indeed, in the FAQ section of the company's website, the reply to "How do you go about matching members?" reads as follows:

"Our matchmakers use a combination of tools, including experience and intuition, when matching members. We start with basic demographic information such as age, religion, location, physical requirements and other preferences. Personality profiles are also used but not relied upon exclusively. In the end, however, it comes down to your personal matchmaker."

Winkler founded the company 18 years ago and sold it in 2003, leaving its day-to-day operations to Annie Ahlin, who worked with Winkler for 14 years and until November was the company's president.

Intuition is a big part of determining long-term compatibility, Ahlin says. Many of the agency's clients are people who have tried scientific matching online but had no luck. Ahlin believes she knows why. "When you're reading a profile online, or looking at a photo, it's one-dimensional. It's that person's PR for themselves. There's no substitute for sitting down with a person one-on-one to get the full picture."

While the Winkler clients fill out personality profiles similar to the ones found online, the difference, Ahlin believes, is the hour-and-a-half interview. Some of these matchmakers have a psychological background, but others are recruited for different reasons. "We go for people who have a heart, are good listeners, are empathetic, and who just have a feel for matching people for the long term," Ahlin says.

Ahlin estimates the agency's success rate at 70 per cent - meaning that 70 per cent of clients either end up in a relationship engineered by their matchmakers or get engaged to someone they've met through the agency. But unlike the studies being done at eHarmony, there's no follow-up to determine how long these relationships or marriages last, or how satisfying they are down the line.

Ahlin and her matchmakers use feedback forms like those on Chemistry.com to learn how a match went after two clients have met in person. But whereas the Chemistry.com people classify this step as part of their scientific research, Ahlin says simply, "This way, you know what it is that works so you can get closer the next time, it helps us with intuition."

Often when Ahlin talks about intuition, she describes the same principles that the scientists I spoke with use in their empirically based matching systems. For instance, in matching couples, she follows what is essentially the similarity- complementarity model.

Like Fisher and Schwartz, Ahlin believes that similarity and complementarity are situational models. "Each person is unique and contradictory," she says, "and you can't just group people into big categories, the way the personality profiles do ... We're always adjusting. It's not a scientific process, its an intuitive one."

It may well take a generation before we learn whether the psychological, anthropological or sociological model works best. Or maybe an entirely different theory will emerge. But at the very least, these dating sites and the relationships they spawn will help us to determine whether science has a place, and if so, how much of a place, in affairs of the heart.

Meanwhile, until these sites start sending me better dating prospects, I figured I'd take Warren up on his offer to introduce me to the 38-year-old single board member he thought would be such a good match for me. But when I asked a company spokesman about him, I was told that he had recently begun seeing someone. Did they meet through eHarmony? My potential soul mate declined to answer.

(Lori Gottlieb is author of Stick Figure: ADiary of My Former Self and Inside the Cult of Kibu: And Other Tales of the Millennial Gold Rush)