Online Dating

Introduction Questions

Video Summary

Finding partners for potential romantic, intimate relationships can be difficult. Recent data (Statista, 2020) shows that 61% of 18–29 year olds, and 44% of 30–59 year olds, have used online dating sites. But is there any evidence that the algorithms used by online dating sites are useful in finding mates?

Dating sites are mostly based on the principle that similarity drives attraction. Eli Finkel states that it is reasonable to think that if we are similar to someone, we are more likely to like them as well. However, he argues that similarity isn’t a strong predictor of relationship happiness. Research shows we tend to match mates on certain demographic characteristics, but similarity and dissimilarity in other areas equally relate to happiness.

Finkel states that we believe we know what we want in a romantic partner, but research shows we aren’t really accurate in these beliefs. Some matching algorithms claim to measure how you answer the question in addition to the answer itself, which can identify hidden personality traits. But just because algorithms use the scientific method, that doesn’t mean they are actually scientific themselves. Harry Reis states that the research can’t take into consideration all of the things that could happen in a relationship, and comparing potential relationships to successful ones isn’t always reliable.

No dating algorithms have been tested scientifically, but even though dating sites might not result in a successful relationship, the researchers state that these can be a great way to meet new people.

(Statista (2020). Share of Internet users in the United States who have used online dating sites or apps as of April 2017, by age group. https://www.statista.com/statistics/706499/us-adults-online-dating-site-app-by-age/)

Question 1

Online dating sites collect data so they can pair you with someone based on:

A. How closely your level of physical attractiveness matches the other person’s.

B. How well your strengths match up with the other person’s weaknesses.

C. How closely you match the other person’s values and personality.

D. Commonality of past experiences.

Correct!

Sorry, your answer is incorrect. Online dating sites collect data so they can pair you with someone based on how closely you match the other person’s values and personality.

People do often find partners whose level of attractiveness matches their own, but this is a relatively weak factor, and is not mentioned in the video in any case.

This notion of complementarity is rather popular, but research has indicated that many such differences are surface-level, and that there tends to be some underlying similarity that explains the relationship (e.g.; a dominant man and submissive woman sharing traditional sex-role values).

Common past experiences are important, but it’s the evaluation of those experiences that is more critical; a shared past experience of primitive camping won’t help if one person loved it and the other hated it.

Question 2

If Reginald and Esther were (accurately) considered well-matched by an online dating site’s algorithm, and they began a relationship, what might we predict about their relationship as it moves forward?

A. Their happiness will increase as the anxiety associated with fearing dissimilar values and beliefs is lessened.

B. Their happiness will start to decrease sooner than if they had to discover these similarities between them on their own.

C. Their happiness will remain stable over time, because people know all the important things about each other from the outset.

D. They aren’t any more likely to be happy in their relationship than if they were poorly matched.

Correct!

Sorry, your answer is incorrect. We might predict that they aren’t any more likely to be happy in their relationship than if they were poorly matched. As logical as the other options may sound, none of them are mentioned as ways of understanding the association between similarity and happiness. You may wish to design a study to test these hypotheses, though….

Question 3

You’re frustrated that you haven’t been able to make a love match through your online dating site. Which of the following might be a problem with the dating site’s recommendations?

A. You may not have the greatest insight into what you want in a dating partner, so the algorithm may be fine – but garbage in/garbage out.

B. The algorithms weigh certain responses differently than you would, so something really important to you is not considered as such by the algorithm.

C. The mathematical processes underlying the algorithm are not capable of making the nuanced judgments that make finding good matches possible.

D. Heuristics are much more accurate than algorithms are when it comes to decision making; the site simply uses the wrong process.

Correct!

Sorry, your answer is incorrect. You may not have the greatest insight into what you want in a dating partner, so the algorithm may be fine — but garbage in/garbage out.

Advanced algorithms may factor in differential weighting of responses, but such weightings are not chosen by the algorithm, they are chosen by the person.

There may be no algorithm truly capable of helping people find happy relationships, but the problem isn’t in the underlying math, it’s in the algorithm itself. In the same way, if you can’t balance your checkbook, you wouldn’t say that addition and subtraction aren’t capable of doing the job; the problem lies in the process.

Heuristics are decision-making rules that are inherently less accurate (although faster) than algorithms.

Question 4

There is a great deal of research available that looks at characteristics of couples who have positive relationship outcomes and ones that don’t. Shouldn’t this research support the claims of online dating sites so that they can help you find lasting relationships?

A. No; because online dating sites try to predict whether things will work out, while the research describes relationships that already exist.

B. Maybe, to the extent that the characteristics studied in the research literature are measurable via questionnaire.

C. Probably not, because the population using online dating sites is not directly comparable to the populations in the research literature.

D. Yes; if the research is understood properly and worked into the algorithms, it will increase its validity.

Correct!

Sorry, your answer is incorrect. The answer is no; because online dating sites try to predict whether things will work out, while the research describes relationships that already exist.

Not being measurable via self-report could be a problem, it’s true, but even if the characteristics being studied are measurable in this way, it’s pretty clear (according to Dr. Reis) that the work doesn’t apply.

There is nothing in the video to suggest that the population of people using online dating differs in theoretically important and statistically significant ways from the populations studied in the research literature.

Dr. Reis is pretty clear that the existing research doesn’t apply particularly well to online dating sites.

Question 5

How confident should you feel in the accuracy of the “match percentage” you receive from an online dating site?

A. Confident; if everyone puts in the most accurate information they have, the algorithm will work.

B. Not confident; the algorithms involved are extremely complex and so have a lot of potential for error.

C. Confident; the companies involved have done some extensive research to refine their algorithms.

D. Not confident; there haven’t been independent scientific reviews of the algorithms used by online dating companies.

Correct!

Sorry, your answer is incorrect. You should not feel confident as there haven’t been independent scientific reviews of the algorithms used by online dating companies.

The algorithm will produce a result regardless of the accuracy of the information; the system doesn’t know you’re lying. And does that algorithm really get at what it says it does? How do we know?

Greater complexity does offer more opportunity for error, but such errors are not foregone conclusions just because one has a complex algorithm.

The company may have done all the research it claims, but they also have a material interest in claiming their algorithm is the best, thereby driving more traffic to their site. The conflict of interest is apparent.