When searching for a romantic partner, most people have an idea of what their ideal partner should be like. Common preferences might include traits such as intelligence, a good sense of humor, or physical attractiveness. But how much do these ideal partner preferences really influence who we find attractive or end up dating? A large-scale international study, conducted in collaboration with the Psychological Science Accelerator, sought to answer this question by testing whether the traits people say they want in a partner actually predict how they feel about their romantic interests in real life.
The study, which is the largest of its kind, found that while people’s preferences for certain traits do have some predictive power, the strength of this influence varies significantly depending on how “preference-matching” is measured. The research suggests that while we may know generally what we find attractive, the specific traits we think we want in a partner don’t always strongly predict who we are attracted to or satisfied with in a relationship.
The findings have been accepted for publication in the Journal of Personality and Social Psychology.
For decades, researchers have tried to understand what drives human mating preferences. Ideal partner preferences, or the traits people desire in a partner, have been a cornerstone of theories explaining human romantic relationships. However, whether these preferences actually predict how much people like their romantic partners has been a matter of debate. Previous studies have produced mixed results, with some supporting the idea that people are happier with partners who match their ideals, and others suggesting that this matching has little to no effect on relationship satisfaction.
This inconsistency prompted researchers to take a closer look at the matching hypothesisโthe idea that people will evaluate a romantic partner more positively if that partner’s traits align closely with their ideals. By conducting the study on a global scale and using a variety of rigorous analytical methods, the researchers aimed to provide more definitive answers about the role of ideal partner preferences in romantic relationships.
“My main line of work unpacks how the process of romantic relationship formation and maintenance happens in real life. People commonly tell you theyโre looking for certain traits in a partner as a part of this process โ like attractiveness, intelligence, or kindness โ and Iโve long wanted to understand how peopleโs ideas about the traits that they like (i.e., their ideals) coincide (or not) with the traits that they actually like,” said study author Paul Eastwick, a professor of psychology at the University of California, Davis.
“I embarked on the current article about 5 years ago. At this time, this whole literature on partner preference matching was fracturing, because people started adopting all of these different measurement approaches without really have a strong sense of what those approaches meant, conceptually or statistically. So I wanted to put a large team together and submit a registered report where we committed to run ALL the statistical approaches โ the good ones and the bad ones.”
The researchers collaborated with the Psychological Science Accelerator, a globally distributed network of researchers that facilitates large-scale, cross-cultural studies in psychology. This collaboration allowed the researchers to gather data from a wide array of participants (more 10,000 participants drawn from 43 different countries), ensuring that the studyโs findings would be broadly applicable and culturally inclusive.
The participants were a mix of individuals currently in romantic relationships and those who were single, providing a broad perspective on how ideal preferences might influence both ongoing partnerships and initial romantic interests. The participants were on average 28.5 years old, with a wide age range that included younger adults as well as older individuals.
Participants first completed a block of questions where they rated the desirability of 35 different traits in an ideal romantic partner. These traits were chosen based on previous research and included characteristics such as attractiveness, intelligence, a sense of humor, and kindness. Participants rated each trait on a scale from 1 (not at all desirable) to 11 (highly desirable).
Then, participants were asked to evaluate four individuals they knew personally, with a focus on people of their romantically preferred gender. Participants in relationships were instructed to include their current romantic partner as one of these individuals, while single participants were asked to identify someone they would most desire to have a romantic relationship with. In addition to their romantic partner or desired partner, participants rated three additional peers, which allowed the researchers to analyze patterns of attraction and partner preference across multiple individuals.
Participants rated each of these individuals on the same 35 traits they had identified as important in an ideal partner. They also provided an overall evaluation of their romantic interest in or satisfaction with each individual. This evaluation included questions about their level of romantic interest, how much the person aligned with their ideal partner, and their desire to maintain a close relationship with the person.
One of the most interesting findings was the distinction between “normative” and “distinctive” preference-matching. Normative matching refers to the alignment between a partner’s traits and traits that are generally considered desirable by most people, such as kindness or intelligence. Distinctive matching, on the other hand, involves traits that are particularly important to an individual, even if they are not widely sought after.
The researchers found that normative preference-matching had a stronger influence on romantic evaluations than distinctive matching. In other words, people were more likely to evaluate their partners positively if those partners had traits that were widely seen as desirable, rather than just matching their personal ideals.
The researchers also examined whether the predictive power of preference-matching varied depending on whether participants were evaluating a current partner or someone they were not yet romantically involved with. The matching effects were slightly stronger in established relationships, but the differences were small.
“I was surprised to see the effect sizes were pretty much the same whether or not people were in relationships. I had assumed that single people would show smaller effect sizes. But it turns out that you get similar effects, regardless of whether you are asking people in relationships to report on their partners vs. single people reporting on someone they might like to date.”
When researchers looked at how specific traits influenced romantic evaluations, they found that certain traits like extraversion and religiosity had stronger predictive power than commonly studied traits like physical attractiveness or intelligence. This suggests that the importance of matching on specific traits may depend on the cultural context and the specific attributes valued in different regions.
Eastwick explained three primary findings from the study:
“1. If you care about normative matching โ do people like partners when they perceive those partners to have consensually desirable (e.g., ‘attractive’) rather than consensually undesirable (e.g., ‘easily upset’) traits โ the effects are fairly large. I interpret these effects to mean ‘People have a shared understanding that some attributes are more evaluatively positive than others.’ Important, but somewhat obvious.”
“2. If you care about individual differences in desire โ sometimes called distinctive-matching โ you need to make certain statistical corrections that separate out the normative component. When you do this, across all 35 traits, you get significant โ but pretty modest โ effects, in the .10-.20 range. I interpret these effects to mean ‘If I think you match my particular unique preferences across 35 traits, I feel a little more positively about you.'”
“3. If you care about single attributes in isolation (i.e., do people who have a higher ideal for ‘attractiveness’ actually desire ‘attractive’ partners more?), these effects are tiny on average, and theyโd be very hard to detect in a typical study. Some are larger (religious is notable in this respect), and some are no different from zero (i.e., warmth/trustworthiness). I interpret these effects to me: ‘If I say I desire a particular trait like attractiveness more than you do, that doesnโt mean very much.'”
The study also found some expected gender differences. Men generally placed a higher value on traits related to physical attractivenessโsuch as being attractive, having a nice body, and being sexyโwhile women prioritized traits associated with earning potential, including ambition, financial security, and having a good job. These findings align with traditional gender stereotypes and previous research that suggests men are more focused on physical appearance and women on resource-related traits when considering potential partners.
However, the researchers also analyzed “revealed preferences,” which reflect the actual impact of these traits on participants’ romantic evaluations of specific individuals. This analysis revealed a more nuanced picture. While men and women differed in their stated preferences, these differences were less pronounced.
In practice, both men and women tended to rate partners more positively if they were physically attractive, though women underestimated the importance of physical attractiveness in their stated preferences more than men did. Similarly, while womenโs stated preferences indicated a strong preference for partners with high earning potential, this trait was less influential in their actual romantic evaluations compared to what their stated preferences suggested.
“Because matching effects for single attributes in isolation are tiny (#3 above), this fact explains the lack of gender differences in revealed preferences,” Eastwick explained. “If the people who say they like attractiveness donโt especially desire attractive partners (relative to people who say they donโt like attractiveness), then you canโt expect to find that men (i.e., a group of people who say they like attractiveness) especially desire attractive partners (relative to women, a group of people who say they donโt like attractiveness). Those tiny single-attribute matching effects are a mathematical constraint that render gender differences more or less impossible to detect.”
Despite its large scale and rigorous methodology, the study had some limitations. “Even though our sample is very large and diverse in terms of world region, most of them had at least some amount of post high school education,” Eastwick noted. “So we canโt claim that this sample is representative of the world in that sense.”
Additionally, the study relied exclusively on self-reported data collected at a single point in time, which limits the ability to draw conclusions about cause-and-effect relationships. To fully understand the causal impact of ideal partner preferences, future research will need to incorporate experimental approaches.
“Iโm also very interested in what ideals do causally โ do ideal partner preferences actually cause any of the outcomes that we attribute to them? So weโre developing manipulations in my lab to address this major challenge, too,” Eastwick said.
The study, “A Worldwide Test of the Predictive Validity of Ideal Partner Preference-Matching,” was authored by Paul W. Eastwick, Jehan Sparks, Eli J. Finkel, Eva M. Meza, Matรบลก Adamkoviฤ, Peter Adu, Ting Ai, Aderonke A. Akintola, Laith Al-Shawaf, Denisa Apriliawati, Patrรญcia Arriaga, Benjamin Aubert-Teillaud, Gabriel Banรญk, Krystian Barzykowski, Carlota Batres, Katherine J. Baucom, Elizabeth Z. Beaulieu, Maciej Behnke, Natalie Butcher, Deborah Y. Charles, Jane M. Chen, Jeong Eun Cheon, Phakkanun Chittham, Patrycja Chwiลkowska, Chin Wen Cong, Lee T. Copping, Nadia S. Corral-Frias, Vera ฤubela Adoriฤ, Mikaela Dizon, Hongfei Du, Michael I. Ehinmowo, Daniela A. Escribano, Natalia M. Espinosa, Francisca Expรณsito, Gilad Feldman, Raquel Freitag, Martha Frias Armenta, Albina Gallyamova, Omri Gillath, Biljana Gjoneska, Theofilos Gkinopoulos, Franca Grafe, Dmitry Grigoryev, Agata Groyecka-Bernard, Gul Gunaydin, Ruby Ilustrisimo, Emily Impett, Pavol Kaฤmรกr, Young-Hoon Kim, Mirosลaw Kocur, Marta Kowal, Maatangi Krishna, Paul Danielle Labor, Jackson G. Lu, Marc Y. Lucas, Wojciech Maลecki, Klara Malinakova, Sofia Meiรner, Zdenฤk Meier, Michal Misiak, Amy Muise, Lukas Novak, Jiaqing O, Asil A. รzdoฤru, Haeyoung Gideon Park, Mariola Paruzel, Zoran Pavloviฤ, Marcell Pรผski, Gianni Ribeiro, S. Craig Roberts, Jan P. Rรถer, Ivan Ropovik, Robert M. Ross, Ezgi Sakman, Cristina E. Salvador, Emre Selcuk, Shayna Skakoon-Sparling, Agnieszka Sorokowska, Piotr Sorokowski, Ognen Spasovski, Sarah C. E. Stanton, Suzanne L. K. Stewart, Viren Swami, Barnabas Szaszi, Kaito Takashima, Peter Tavel, Julian Tejada, Eric Tu, Jarno Tuominen, David Vaidis, Zahir Vally, Leigh Ann Vaughn, Laura Villanueva-Moya, Dian Wisnuwardhani, Yuki Yamada, Fumiya Yonemitsu, Radka ลฝรญdkovรก, Kristรฝna ลฝivnรก, and Nicholas A. Coles.
