Perceptions of Ideal and Former Partners ’ Personality and Similarity Pieternel Dijkstra

The present study aimed to test predictions based on both the ‗similarity-attraction‘ hypothesis and the ‗attraction-similarity‘ hypothesis, by studying perceptions of ideal and former partners. Based on the ‗similarity-attraction‘ hypothesis, we expected individuals to desire ideal partners who are similar to the self in personality. In addition, based on the ‗attraction-similarity hypothesis‘, we expected individuals to perceive former partners as dissimilar to them in terms of personality. Findings showed that, whereas the ideal partner was seen as similar to and more positive than the self, the former partner was seen as dissimilar to and more negative than the self. In addition, our study showed that individuals did not rate similarity in personality as very important when seeking a mate. Our findings may help understand why so many relationships end in divorce due to mismatches in personality.

In the last few decades numerous studies have been carried out on the characteristics individuals value most in a mate.Several studies have, for instance, shown that individuals, particularly men, highly value a potential mate's physical attractiveness (e.g., Buss, Shackelford, Kirkpatrick, & Larsen, 2001).Much more scarce are studies that relate individuals' own characteristics to those they desire in a potential mate.With regard to these ‗relative' mate preferences two hypotheses have been presented.First, according to the ‗similarity-attraction hypothesis' the more similar two individuals are, the higher the attraction between them will be (e.g., Byrne, 1971;Byrne & Nelson, 1965).It is usually reassuring and comforting to meet others who are like ourselves.Similar others do not only validate our beliefs about the world and ourselves but also facilitate harmonious interactions, reducing the risk of conflicts and disagreements (e.g., Byrne, 1971;Morry, 2005;Rusbult, Kumashiro, Kubacka, & Finkel, 2009).The second hypothesis is usually referred to as the ‗complementarity hypothesis'.According to this hypothesis, individuals feel most attracted to potential partners who complement them, an assumption that reflects the saying that -opposites attract‖ (e.g., Antill, 1983).Complementary individuals are assumed to be so attractive because they enhance the likelihood that one's needs will be gratified (e.g., De Raad & Doddema-Winsemius, 1992).For example, young women who lack economic resources may feel attracted to older men who have acquired economic resources and therefore may be good providers (Eagly & Wood, 1999).

The importance of similarity in personality
Mismatches in personality are often one of the major causes of divorce (e.g., Amato & Previti, 2003).In a study conducted among divorcees (n = 8,145) in The Netherlands, where the present study was conducted, about 40% reported dissimilar personalities as the major cause of their break-up (De Graaf, 2006; publication by the Dutch Central Bureau for Statistics).For individuals who seek a mate for a long-term, harmonious relationship, finding a similar mate is therefore important, or at least, should be.In general, however, partners have been found to have only somewhat similar personalities (e.g., Barelds & Dijkstra, 2007;Botwin et al., 1997;Dijkstra & Barelds, 2008;McCrae et al., 2008;Rammstedt & Schupp, 2008).Several explanations have been forwarded for this discrepancy.For instance, Barelds and Dijkstra (2007) argue that individuals usually do not take the time to get to know each other properly before they get romantically involved.Indeed, these authors showed that individuals who took the time to get to know each other's personalities and became friends before they got romantically involved -in contrast to those who became romantically involved relatively quickly -ended up with a partner with a more similar personality.Other studies suggest that people are simply unaware of or confused about the characteristics that are important in a mate (e.g., Dijkstra & Barelds, 2008;Eastwick & Finkel, 2008).
The present study builds on and extends work by Dijkstra and Barelds (2008).Dijkstra and Barelds (2008) found that singles looking for a long-term mate, as can be expected on basis of the ‗similarity-attraction' hypothesis, looked for someone similar in terms of extraversion, agreeableness, emotional stability, conscientiousness and openness to experience.Nonetheless, when individuals were asked about their preferences for a similar or complementary mate, they found that the majority of individuals (86%) indicated that they desired a complementary partner rather than a similar one.The problem with this study, however, was that the authors asked people about their mate preferences in general, not, more specifically, about their mate preferences with regard to personality.A possible explanation for their (seemingly) contradictive findings is therefore the possibility that individuals took the general question about similarity/complementarity to be about other characteristics than personality, such as age, height, intelligence or religion.The first aim of the present study was therefore to replicate and extend Dijkstra and Barelds' (2008) study, by assessing both partner criteria in terms of personality and the importance individuals attach to a partner's similarity specifically with regard to personality.
We would like to argue that, in general, individuals who are seeking a mate tend to underestimate the importance of finding a partner with a similar personality.When asked about their preferences for a mate people may partially draw upon lay theories of romantic attraction rather than their true desires for a mate.In general, the notion that -opposites attract‖ is a relatively popular lay theory of romantic attraction: people often think that individuals who possess complementary characteristics are highly attractive as a partner (Barelds and Dijkstra, 2007).In contrast, looking for someone who is similar to oneself may be perceived as -boring‖.These popular lay theories may confuse people and lead them away from their true partner desires (Dijkstra & Barelds, 2008).As a result, individuals may underestimate the importance of finding a similar mate in terms of personality.It is important to examine this issue.If indeed individuals underestimate the importance of having someone with a similar personality as a potential long-term partner, it may help explain why people often start relationships with individuals with such different personalities than they have themselves, increasing their chances of relationship break-up.

Similarity and former partners
To date, studies on partner similarity have exclusively focused on perceived similarity in the current or ideal partner (e.g., Barelds & Dijkstra, 2007;Buunk & Bosman, 1986;Dijkstra & Barelds, 2008;Klohnen & Luo, 2003), neglecting perceptions of similarity of former partners.With regard to former partners, another hypothesis seems to be relevant.In addition to the ‗similarity-attraction' hypothesis, the ‗attraction-similarity' hypothesis has been postulated (Morry, 2003;2005).This latter hypothesis claims that attraction (i.e., love, liking) leads to heightened perceptions of similarity.According to Morry (2005), in ongoing relationships, projection of the self onto the other person is the result of the attraction between two individuals, with higher attraction leading to higher perceived similarity.Whereas the ‗similarity-attraction' hypothesis seems to be most relevant in those situations in which individuals do not know each other (very well), the ‗attraction-similarity' seems to apply mainly to relationships that already exist.It must be noted, however, that the effect of attraction on similarity does not preclude the effect of similarity on attraction.In fact, a mutual effect is the most likely scenario.Byrne, Lamberth, Palmer, and London (1969) found that recent information about attitude similarity influenced liking more than earlier information about attitude similarity, indicating that similarity influences attraction.At the same time, research on projection indicates that attraction also influences perceptions of similarity (Morry, 2003;Morry et al., 2005).
Because the relationship with the former partner already exists, sometimes even for a long time, the ‗attraction-similarity' hypothesis seems to be highly relevant for the understanding of perceptions of former partners.In most cases, individuals do not feel attracted any more to their former partner.A survey study in The Netherlands showed that, the first two years following the divorce, about 60% of ex-partners described their contact as hostile and recently had engaged in serious conflicts with their ex-spouse (Fischer, De Graaf & Kalmijn, 2005).Research also shows that people tend to devalue former partners.Gray and Silver (1990; see also Amato & Previtti, 2003), for instance, found that participants thought they were less responsible for the break-up than the exspouse, and portrayed themselves more as a victim (and less as a villain) compared to their ex-spouse.In terms of personality, Geher and colleagues (2005) found that, compared to the current partner, individuals rated their former partner as less open, less conscientious, less agreeable, more neurotic and as having a more problematic attachment style.Geher and colleagues (2005), however, asked participants to provide ratings of their current and former partner's, but not self-ratings.As a result, they could not relate ratings of the former partner to self-ratings and establish levels of similarity between participants and their former partners.We do not know of a single study that assessed similarity between individuals and their former partner.Examining this issue is important, however.It may help understand why people often devalue their former partners, and why they may hate the ones they formerly loved.The third goal of the present study was therefore to examine the level of similarity between the self and the former partner.Based on the ‗attraction-similarity' hypothesis, we expected former partners to be perceived as dissimilar to the self in personality.

The present study
In sum, the present study examined three hypotheses, that is, the prediction that individuals seek a similar mate in terms of personality (Hypothesis 1), the prediction that people attach relatively low importance to similarity in personality (Hypothesis 2), and the prediction that former partners are perceived as being dissimilar in personality to the self (Hypothesis 3).Studies on mate selection usually focus on the mate preferences of relatively young individuals (i.e., those in their twenties and early thirties).Although indeed many people in this age category are eligible, the last couple of decades have shown a clear increase in the number of ‗older' single people.For example, whereas in 1996 in The Netherlands 300.000 men between 34 and 54 years were single, this number had risen to 419.000 in the year 2006, most of which do seek a long-term partner (De Graaf, 2006).The present study therefore aimed at examining a group of single individuals that is heterogeneous with regard to age, including younger, middle-aged and older individuals.

Participants and procedure
Participants were 871 (612 women, 259 men) members of two dating sites, one for college-educated singles under fifty looking for a long-term mate (www.match4me.nl;n = 421) and one for singles over fifty looking for a long-term mate (www.50plusmatch.nl;n = 450).Mean age was 50.18 years (SD = 11.32,range 19-78).Participants were, by means of an email message, invited to participate in an online study on relationships.

Measures
Personality.Personality characteristics were assessed by an abridged version of Shafer's 30-item bipolar rating scale designed to measure the Five-Factor Model of personality.Ten itemstwo for each factor -were selected that Shafer (1999) found to have the highest factor loadings.The same ten item instrument was used in a study by Dijkstra and Barelds (2008) on partner preferences.The items that were selected were: Openness --uncreative-creative‖ and -unartistic-artistic‖, Conscientiousness --lazy-hardworking‖ and -unresponsible-responsible; Extraversion --shy-outgoing‖ and -quiet-talkative‖; Agreeableness --headstrong-gentle‖ and -vengeful-forgiving‖; and Neuroticism --at ease-nervous‖ and -unagitated-tense‖.The response scale was a five-point semantic differential type scale ranging from one (the left trait in the trait-pair describes me very well) to 5 (the right trait in the trait-pair describes me very well).For each of the five factors greater values indicated higher levels of extraversion, neuroticism, conscientiousness, agreeableness and openness respectively.
The participants answered the personality items three times in total, thus providing self-ratings, ratings of their ideal partner and ratings of their former partner.
With regard to the ratings of their ideal partner, participants were asked to rate the personality characteristics of their ideal long-term partner.In addition, participants were asked to what extent their most recent former long-term partner possessed these personality characteristics.Cronbach's alphas (across ratings, i.e., self, ideal partner and former partner) were .77for Neuroticism, .71for Extraversion, .83 for Openness, .75 for Agreeableness, and .60 for Conscientiousness.
To examine the internal validity of this instrument, a Principal Components Analysis with Varimax rotation was first conducted (across ratings), extracting five factors.Congruence coefficients (Tucker, 1951) were then computed between the obtained rotated factor structure and the expected (a priori) structure (simple structure).
These congruence coefficients (with a theoretical maximum of 1.00) were .95for Neuroticism, .97 for Extraversion, .98 for Openness, .86 for Agreeableness, and .96for Conscientiousness.These values indicate a strong correspondence between factors (e.g., Haven & Ten Berge, 1978;Lorenzo-Seva & Ten Berge, 2006), and therefore strongly support the internal validity of this instrument.
Importance of similarity.Following the self-ratings of personality, participants were asked how important they think it is that a partner resembles them in each of the following five domains: Personality, intelligence, attractiveness, attitudes and sense of humor.These domains and/or similarity in these domains have been found to be important to attraction and relationship satisfaction (personality, e.g., Barelds, 2005; intelligence, e.g., Watson, Klohnen, Casillas, Nus Simms, Haig & Berry, 2004; attractiveness, e.g., Yela & Sangrador, 2001; attitudes: e.g., Hatfield & Rapson, 1992; humor: e.g., Ziv, 1989).All items were answered on five-point scales.With regard to personality participants had to indicate how important they feel it is that a partner is similar to them (1), or is their opposite (5).With regard to attitudes and sense of humor, participants were asked to indicate how important it is to them that a partner is either similar (1) or different (5).With regard to a partner's intelligence/attractiveness, participants had to indicate how important it is to them that a partner is less intelligent/attractive (1), or more intelligent/attractive (5) than they are.Note that we did not ask participants the extent to which they prefer a partner who is similar in intelligence/attractiveness in contrast to one who is opposite.We felt that asking participants the extent to which they preferred a partner who is more or less intelligent/attractive than themselves, makes more sense in terms of framing and conveys more information about participants' preferences.Thus, in the case of intelligence and attractiveness both ends of the scale refer to dissimilarity, whereas the midpoint (3) refers to similarity.
Similarity of the former partner.In addition, participants were asked to what extent their most recent former long-term relationship partner had resembled them in each of five domains: Personality, intelligence, attractiveness, attitudes and sense of humor (see ‗Importance of similarity').All items were rated on five-point scales.With regard to personality, participants had to indicate the extent to which their former partner was similar to them (1) or their opposite (5).Next, participants were asked the extent to which their former partner resembled them with regard to attitudes and sense of humor (1 = similar, 5 = different).Finally, participants rated the extent to which their former partner was less intelligent/attractive (1) or more intelligent/attractive (5) than they are.Each of the measures described above was presented to participants on a new screen (the three personality ratings were also presented on separate screens).
Participants were not able to go back to the previous screens, making it harder for participants to compare their responses to their previous responses.

Ideal partner personality
To test Hypothesis 1, the prediction that individuals prefer mates who resemble them with regard to personality, correlational analyses were conducted relating participants' own personality scores to those of their ideal romantic partner (see Table 1).Correlations were computed for men and women separately, in order to examine potential sex differences.Previous studies have consistently found sex differences regarding personality (e.g., Feingold, 1994), and also differences regarding the importance of personality characteristics in the ideal mate (e.g., Botwin, Buss, & Shackelford, 1997).We therefore decided to analyze the data separately for men and women.Confirming Hypothesis 1, for all five personality characteristics positive significant correlations were found (p < .001),ranging from .27 for Openness to .54 for Conscientiousness, indicating that people clearly desired a partner who is similar to themselves with regard to these five personality factors.A significant difference between the correlations for men and women was found only for Neuroticism (test for independent correlations; Z = 2.84, p < .01),with the correlation for women being significantly higher (r = .47,p < .01)than the correlation for men (r = .29,p < .01).The other correlations for men and women were highly comparable in size, and no other significant differences were found.Next we examined whether, besides rank order similarity (i.e., the correlations reported in Table 1), there also was an indication of mean level similarity between participants' personality and their ideal romantic partner's personality.For this purpose, mean personality scores of self and the ideal romantic partner were computed.To examine potential sex differences, a repeated measures MANOVA using Target (self, ideal and ex-partner) as a within subjects factor and participant Sex as a between subjects factor was conducted on the five personality scales.The MANOVA revealed significant multivariate main effects of Sex [F(5,865) = 10.66,p < .001]and Target [F(10, 860) = 190.84,p < .001],as well as a significant interaction effect [Sex × Target; F(10,860) = 14,37, p < .001].Note: means with different subscript letters differ significantly between columns (p's < .01).Subscript numbers refer to between sex comparisons per personality scale: Means with different subscript numbers differ significantly between men and women (p's < .01)Table 2 shows that men's ideal partners are less neurotic, more extraverted, more open, more agreeable, but less conscientious than they are themselves.Women's ideal partners are less neurotic, more open, and more agreeable than they are themselves.In addition to seeking a similar mate, individuals in general seem to look for a mate who possesses more positive traits than themselves.In addition, sexdifferences were found for most self-rated personality scales, with women rating themselves more extraverted, open, agreeable and conscientious than men.Moreover, significant sex differences were found for an ideal partner's personality with regard to neuroticism, extraversion, agreeableness and conscientiousness, with women generally setting higher standards than men (except for Openness).

Perceived importance of similarity
To test the prediction that people attach relatively low importance to similarity in personality (Hypothesis 2), mean scores were next calculated for the item that assessed the importance of similarity in personality between participants and their ideal partner.The means are reported in Table 3, separately for men and women.Table 3, in addition, lists the mean scores for the other domains (intelligence, attractiveness, attitudes and sense of humor) for which the importance of similarity was assessed.Subscript numbers refer to between subject comparisons.Means with different subscript numbers differ significantly between men and women (p's < .01).Subscript letters refer to within-subject comparisons (ideal partner vs. ex-partner).Means with different subscript letters differ significantly within sex (p's < .01).
The mean scores for the personality similarity item seem to support Hypothesis 2: For both men and women the mean scores were around the theoretical midpoint of the scale, indicating that individuals perceived similarity with regard to personality to be of relatively little importance.Score frequencies support this conclusion: 14.9% of the participants scored 1 or 2 (indicating a preference for a similar ideal partner), 67.6% scored 3 (indicating a preference for neither a similar nor a complementary ideal partner), and 17.4% scored 4 or 5, indicating a preference for a complementary partner.
Differences between men and women were non-significant (Χ 2 = 4.64, > p).In addition, Table 3 shows that men valued high intelligence significantly less and high attractiveness significantly more in an ideal mate than women, whereas women valued similar attitudes and sense of humor in an ideal mate significantly more than men.

Former partner's personality
To test the prediction that former partners are perceived as being dissimilar in personality (Hypothesis 3), correlational analysis were conducted relating participants' own personality scores to ratings of the former partner's personality.In line with Hypothesis 3, non-significant correlations were found for Neuroticism, Openness, Agreeableness and Conscientiousness (|r|s < .05,p = ns; see Table 1).For Extraversion a marginally significant negative correlation emerged (r = -.07,p < .05;see Table 1).
These findings suggest that people perceived their former partner not be similar to themselves.Neither, however, did they perceive their former partner to be their opposite in terms of these FFM traits.Also non-significant correlations emerged between the ideal romantic partner's personality and the former partner's personality, with one exception: A small significant correlation was found for Openness (r = .13,p < .001).
As individuals had had a former partner who was more open, they desired a more open ideal partner.
The means in Table 2 show that, on average, men perceived their ex-partners to be more neurotic, less open, less agreeable, and less conscientious than themselves whereas women perceived their ex-partners to be more neurotic, less extraverted, less open, less agreeable and less conscientious than themselves.Note that, for both sexes, the difference between the self and the former partner was particularly large for Agreeableness.Thus, it seems that individuals hold relatively negative perceptions of their former partner's personality.With regard to their former partner's personality, only one significant sex difference was found: Ratings of the level of openness of the ex-partner were higher for men than for women.
Finally, Table 3 shows that former partners were judged clearly more dissimilar with regard to personality, attitudes and sense of humor than the ideal partner, and as significantly less attractive and intelligent than the ideal partner.For both men and women mean similarity scores for former partners are above the theoretical midpoint of the scale, and therefore lean towards complementarity.The frequency distribution supports this conclusion: 9.0% of participants scored 1 or 2 (indicating similarity), 30.1% scored 3 (indicating neither similarity nor complementarity), and a majority of 60.8% scored 4 or 5 (indicating complementarity).These results indicate that expartners in general are not only judged unfavorably, but are also considered more complementary than similar to the self.

Discussion
Our findings strongly support the -similarity-attraction‖ hypothesis: Individuals clearly desire a potential partner with a similar personality.Nonetheless, individuals did not seem to attach much importance to finding a similar mate.This was not only true when individuals were asked about their ideal mate's personality, but also with regard to other characteristics of their ideal partner, such as his or her attractiveness and intelligence.In line with Dijkstra and Barelds' (2008) study, our findings suggest that people are not aware of the importance of finding a similar mate in terms of personality.
Our findings fit well with the observation that, more in general, people seem to be confused about, or unaware of, the characteristics they value in a mate (Eastwick & Finkel, 2008; see also Kurzban & Weeden, 2007;Todd, Penke, Fasolo & Lenton, 2007).Eastwick and Finkel (2008) found that participants' ideal preferences, assessed before a speed-dating event, failed to predict what inspired their actual desire at the event.
According to these authors, people often lack the introspective awareness of what influences their judgments and behaviors in dating situations.Whatever the reason is of this confusion or unawareness, our study shows that people often also underestimate the importance of similarity in personality.
In addition to finding a mate with a similar personality, our study showed that individuals seek a mate who is slightly ‗better' than they are: They prefer a mate who is somewhat less neurotic, more agreeable, more conscientious, more open and more extraverted than they are themselves.This finding is in line with Rusbult and colleagues' (2009) assertion that individuals seek a partner who resembles their ideal self.According to these authors, individuals feel closer to their ideal self when their partner possesses key elements of their ideal selves, a phenomenon they refer to as the ‗Michelangelo-phenomenon'. In addition, we found several gender differences with regard to the ideal partner.In our opinion, the most interesting one is the finding that women seemed to set higher standards for an ideal mate's personality.More than men, women look for an agreeable, emotionally stable (i.e., low neurotic), and extraverted long term partner.Our finding that women set higher standards for a mate than men is consistent with results of previous studies that have shown women to be more selective in their choice of both short term and long term mates than men (e.g., Geary, 2010).
Women seem to set especially high standards regarding characteristics that, from an evolutionary psychological point of view, are important to them.First, high extraversion and low neuroticism make up a broader trait that has been labeled ‗social dominance' (e.g., Dijkstra & Buunk, 1998), a trait that is indicative of future social status (Sadalla, Kenrick & Vershure, 1987).In addition, women's relatively high standards for a partner's agreeableness may reflect the importance women attach to a man's nurturing ability.Whereas social dominance is important for success in the context of intrasexual competition over scarce resources, a nurturing and agreeable personality is important in the context of child caring and the intimate relationship with the partner.

Negative perceptions of former partners
Our study also found support for the ‗attraction-similarity hypothesis: former partners were perceived as being dissimilar to the self in terms of personality.In contrast to the ideal partner, individuals judged their former partner less favorably than themselves: they perceived their ex-partner to be more neurotic, less open, less agreeable, and less conscientious than themselves.Our findings fit well with previous studies that have revealed negative perceptions of former partners (e.g., Fischer, De Graaf & Kalmijn, 2005;Gray & Silver, 1990).Especially the large discrepancy between self-ratings and ratings of former partners in terms of agreeableness seems to support previous studies' claims that, in general, former partners are blamed for the relationship break up and seen as ‗evil' (Amato & Previtti, 2003).Following the ‗attractionsimilarity hypothesis', the relatively dissimilar view individuals hold of their former partner should be attributed to lowered feelings of sympathy individuals hold towards their former partner, indicating that former partners are not actually less similar, but only in the perception of the beholder.If this line of reasoning is correct, our study shows that potential feelings of resentment and dislike are not limited to lowered perceptions of agreeableness, but may be generalized to other aspects of a former partner's personality.Former partners are not only seen as relatively unagreeable, but also as relatively neurotic, less open, and less conscientious.Our finding that former partners are perceived as being dissimilar to the self in terms of personality may, however, also be explained differently.It is possible that individuals indeed had partners who were dissimilar to themselves.As noted before, mismatches in personality are a frequently mentioned cause for relationship break-up.If former partners indeed have dissimilar personalities, our findings underline how difficult it is for many people to select a mate with a similar personality, or, alternatively, how little value individuals put on finding a similar partner in terms of personality.

Strengths, Limitations and Future Studies
The present study contributes to the literature for several reasons.Research on partner personality similarity with regard to either the Big Five or the FFM factors is scarce, especially among individuals who are in the process of looking for and attracting a partner, as is research on how important people perceive personality similarity to be.
A noted before, studying this topic is highly relevant.It may shed light on the high relationship break-up rates, and may especially help explain the high number of breakups that can be attributed to mismatches in personality.In addition, the present study examined a large sample that was heterogeneous with regard to age, whereas most previous studies on personality similarity have used relatively young individuals.
Finally, the present study is the first to show that former partners are perceived as relatively dissimilar to the self.Our finding that individuals hold these perceptions of their former partner may help understand why so many ex-partners, even though the relationship has ended, often hold such problematic relationships.
Our study also suffers from limitations.One concerns the generalizability of our findings.Although by now internet dating has become a socially accepted way of finding a mate, those people who engage in it may differ from those who try to find a mate differently.As a result, it is possible that our findings are not (entirely) representative of singles that are looking for a mate.Future studies may therefore crossvalidate our findings in a group of singles who are dating in real life.In addition, the lack of additional biographical data (e.g., ethnicity), makes it impossible to determine to which extent this sample is representative of the population of single people in The Netherlands.In addition, we did not have information on the length of the relationship with the former partner, nor on the time that had passed since the break-up with the former partner.Both these variables might have been relevant in the context of the present study.For instance, the longer people are in a relationship with someone, the better they should be able to accurately describe their partner's personality characteristics.In addition, when a relationship break-up is relatively fresh, individuals might experience more negative emotions than when a break-up happened a long time ago.This in turn might affect the ratings of the former partner.
The present study's results, as well as the results found in previous studies (e.g., Eastwick & Finkel, 2008), may be used to educate people, especially singles, about what really matters in long-term relationships, for instance, similarity in personality, instead of complementarity.This seems especially important, because, as noted in the introduction section, the belief that ‗opposites attract' is a popular one.In our opinion, it is important for authors of articles such as the present one, to spread their findings among the general public, for instance, through the popular media.More specifically, in the context of the present paper, a press release went out to inform the media, and thus the larger public, about our research findings.We hope that the present study inspires others to study the intriguing question of why people sometimes fall in love with individuals that differ so much from their initial desires, and increase the chances of having a troublesome relationship.

Table 1 .
Correlations between personality scores for self, ideal partner and former partner

Table 2 .
Table 2 reports the mean scores separately for men and women.Mean personality scores for self, ideal partner and former partner (SD's between brackets) by participant sex

Table 3 .
Mean scores on the items reflecting the importance of similarity (ideal partner) and perceived similarity (former partner) in five domains by participant sex (SDs in parentheses)