Why people become attracted to each other

Once upon a time, John met Jenny. They fell in love. They married, had children and presumably will live happily ever after. This is a common love story, and leads us to wonder why people become attracted to each other. Is it because they are similar? Research (Byrne, 1971) has shown that people are attracted to those immediately similar to them and this could lead to marriage. However, arguments for complementarity contend that opposites do indeed attract, and this attraction would also lead to long-term relationship and marriage. Such views are further reinforced by research like that of Shiota and Levenson (2007), which suggest that complementary couples are more satisfied in the long run. Thus, this essay seeks to evaluate the concepts of ‘similarity’ and ‘complementarity’, analysing the relevant researches in the context of a romantic heterosexual relationship.

Whilst there are many theories given on the process of mate selection, of particular interest is the recent emergence of the popular notion that ‘opposites attract’. However, this idea is open to interpretation, because it is expected that if a couple is not alike, they would tend to have more conflict, which will reduce the quality of their relationship (Pieternel & Dick, 2008). Felmlee (2001) has shown that relationships developed from attraction based on complementarity often end prematurely. Nonetheless, there is also evidence supporting the complementary need theory (Winch, 1954), contending that for attraction and therefore a happy marriage, there must be potential gratification of needs for both John and Jenny. An example of such ‘need-gratification’ is when younger females tend to be more attracted to older males who are financially stable (Eagly & Wood, 1999, as cited in Pieternel & Dick, 2008). Nevertheless, there has not been enough support in recent findings for Winch’s hypothesis, and reasons for this will be discussed.

Best services for writing your paper according to Trustpilot

Premium Partner
From $18.00 per page
4,8 / 5
Writers Experience
Recommended Service
From $13.90 per page
4,6 / 5
Writers Experience
From $20.00 per page
4,5 / 5
Writers Experience
* All Partners were chosen among 50+ writing services by our Customer Satisfaction Team

Conversely, the contrasting thought to the concept of complementarity is the established theory of similarity; simply put, that ‘birds of a feather flock together’. According to Hill, Rubin and Peplau (1976), there is a tendency for people who are similar in “physical attractiveness, religion, education, age, and even height” to be attracted to each other. However, there has been suggested that such tendencies may not exist, due to invalid testing procedures. This thought is suggested by research showing only small degrees of similarity between spouses’ personality in marriage (Eysenck, 1990), and in some reported studies (Antill, 1983; Peterson et al., 1989) no degree of similarity was observed, because couples are paired on a random basis. The theory of similarity’s influence on one’s attraction therefore needs examination in its methodological aspects and will be treated accordingly in this essay.

Examination of Similarity and its theory in Romantic Relationships

Summary of Similarity and its theory

‘Similarity’, the more accepted theory, suggests that we will be more romantically attracted to people who are similar to us in aspects like physical attractiveness, intelligence, socio-economic background, and overall attitude towards life. According to Byrne (1971), this can be explained through the theories of classical conditioning, and in particular, the idea of positive reinforcement. In a relationship consisting of two like-minded individuals, sharing the same viewpoint allows them to feel that their opinions are validated, and thereby their own confidence increases, enhancing the relationship.

However, there is a distinction between the perceived and actual similarity. Many have argued that actual similarity is not as important as perceived similarity. A suggestion for this is that the individual experiences the positive reinforcement; regardless of them believing that the similarity is there even though it is not (Montoya, Horton & Kirchner, 2008). However, a possible flaw in this is that if only one partner experiences such feeling of similarity whilst the other does not, then the attraction may not exist. Nonetheless, there is also contradicting evidence of people whose actual similarity is low, but are still highly attracted to another, which support the idea of ‘complementarity’ (Winch et al., 1954). This is important, as it reminds us that similarity is only a positive correlate to the process of attraction, and is not the absolute factor that determines the formation of a romantic relationship.

Methodological Concerns in Measuring the Effects of Similarity in the Formation of Attraction

It has been shown in various recent researches – post-dating Byrne’s studies on attraction – that whilst similarity exists in the formation of attraction, mate selection still operates on a random basis (Antill, 1983; Peterson et al., 1989); or at most, according to Eysenck (1990), the presence of similarity is only lightly significant. The reason for this can be attributed to the methodologies use to measure the effects of attraction. Such is the claim by the meta-analysis conducted by Montoya, Horton and Kirchner (2008), where they concluded that whilst similarity does lead to attraction, this occur mainly in a laboratory setting and not in existing relationships.

This criticism is the product made by the use of the “bogus stranger”, which resulted in artificial responses and lack of ecological validity. The method originates from Byrne’s studies in which given a set of characteristics similar to their own, the participants are asked if they will be attracted to an absent imaginary person unknown to them (Byrne, Clore & Smeaton, 1986). This involves no interaction between the participant and the stranger, as well as requiring the assistance of a third person called a “confederate”, a trained interviewer. While the technique has been heavily criticised, its focus on attraction is clear-cut and minimises other external factors that could influence how much the stranger can be perceived as attractive to the participants. Other advantages to this method are that it is inexpensive and not very time consuming, and thus is viewed as more open for a wider pool of participants.

Another technique is one that involves little to some interaction between the participants and the confederate or fellow participant in the context of a dyad relation. This can last from a few minutes to several hours. However, there have been some contradicting evidences (Dryer & Horowitz, 1997) produced through the use of this method, which some has criticised because the nature of the exchange that involves mainly shallow and polite niceties. A final method that has been in favour recently is the use of couples in an existing relationship. Whilst this is a strong method in that it examines the attraction that then result in a real relationship, its weakness is the sample is often small because of its exhaustive nature, expensive and time consuming. (Montoya, Horton & Kirchner, 2008).

Summary of “Birds of a Feather Don’t Always Fly Farthest” (Shiota & Levenson, 2007)

Given these concerns, it is interesting to see that the study of Shiota and Levenson (2007) proposes the alternative for the concept of similarity in the context of marital satisfaction as opposed to attraction. It is a longitudinal study that examines the effects of high level of similarity in the Big Five Personality on marriage satisfaction, which predicts the possible outcome in terms of a linear trajectory that is then supported by the negative correlates or negative result. The authors discuss the significance of this result in terms of different stages of marriage life, which contain different roles and responsibilities that will affect their satisfaction of marriage. For example, the first stage is explained as the newlyweds who still try to please their partner and therefore, their similar personalities will then be important for their daily social interactions. However, later in their mid-marriage life, when their passion has waned off, this will result in conflicts on issues like child rearing rather than agreement because “spouses competing with each other in similar performance domains and clashing when attempting to complete the same task” (Shiota & Levenson, 2007, p. 672).

In addition, the authors also provide a quick overview of some of the weaknesses in past studies. An example of this is the lack of distinction in examining the similarity of husband and wife’s personalities as opposed to how an individual’s personality might affect the marital satisfaction. Another is when past studies have chosen to only examine similarity in existing relationship in a cross-sectional view. The author improved the study’s method by conducting a long term study but at the same time, also provide a linear trajectory that can be used to predict further development from the given twelve years study. Some limitations that have been self-identified by the authors include the cultural and generational effects that were resulted from the voluntary nature of the sample’s selection and also for the convenience of the research.

More importantly, in their discussion, Shiota and Levenson (2007) have also explicitly referred to how complementary couples in terms of the Big Five personality may perhaps achieve higher level of marriage satisfaction. Their research also investigated on older couples who have been married for at least thirty-five years and again, this show the decline in marriage satisfaction that occurs over time in couple who share a high level of similarity. However, as stated by the authors, this may not be true for other social domains of personality such as the dominant/deferent trait that was tested in Winch’s studies on complementarity. Nonetheless, the study provides the stepping-stone on the investigation of the concept of complementarity that will be discussed as followed.

Examination of Complementarity and its theory in Romantic Relationships

Defining ‘complementarity’

In this essay, ‘complementarity’ can be understood as people who are different in their personality liking one another because they would fill in the gaps present in one another’s life. Formal definition for this term is borrowed from Winch’s definition of ‘need-complementarity’ in which an individual will be attracted to “that person who gives the greatest promise of providing him or her with maximum need gratification” (Winch et al., 1954, p. 242). Other terms used are ‘complementary’, an adjective used to describe how compatible two partners are even when they are different; and ‘complementariness’, a noun that is used by Winch to refer to the degree in which two people can complement.

Winch’s Study and Theory of Complementary Needs in Mate-Selection

Winch, Ktsanes and Ktsanes (1954) made a proposition, the “theory of complementary needs in mate-selection” that supports the idea of ‘opposites attract’. This is made after Winch has conducted a study that involves twenty-five, 19 to 26 years old, white, middle-class, childless married couples who had been married for less than two years, and in which at least one member of each couple was an undergraduate student (Winch et al., p. 245). He uses triangulation to investigate his supported hypothesis, which is people marry based on their complementary needs. The different methods that Winch has used within triangulation include a “need-interview”, which is to identify the type of need pairs present; a case history interview of participants, and TAT, “an eight-card thematic apperception test” (Winch et al., p. 244). However, there are possible limitations within Winch’s method namely the choice of participants where they are clearly culturally biased towards the developed, Western views and the generational effects of participants involved.

Winch’s theory requires that one of two conditions; either Type II or I must be met for complementarity between partners to exist. Type I is the difference in “intensity”, where one partner will be highly “expressive” in a certain “need-pattern” but the other partner will be very low in expressing the same pattern. For instance, a person who needs to be dominant in their relationship will complement with someone who have a very small amount of this need. Conversely, Type II is the difference in “kind” of the interactions between two individuals where both partners will be highly expressive but their needs are opposite to each other. An example is a person whose need to be dominant complements with a person who has the need to be deferent. It is important that there are two types of complementary needs because in the case of “abasement” as one of the need pair, there are situations where both “abasement-autonomy” and “abasement-hostility” exist, which gives a different continuum that cannot be explained sufficiently by Type I. (Winch et al., 1954)

Evaluation of Winch’s Need Complementarity

One of the weaknesses in Winch’s study (Winch et al., 1954) is the lack of evidence in which there has not been enough proof to support his theory despite numerous attempts being made (Bowerman & Day, 1956; Schellenberg & Bee, 1960). However, Levinger (1964) attempts to justify these discrepancies by explaining that there are different needs for different level of relationship as identified by Winch when he postulates his theory, i.e. a married couple’s needs to complement differ from two friends’ needs to complement. Therefore, since most replicated studies uses the Edwards Personal Preference Schedule (Edwards, 1953, as cited in Levinger, 1964), which is aimed for a “general peer relations” to test for the validity of Winch’s theory of need-complementarity for married couples, the results are therefore considered to be inconclusive in regards to Winch’s theory.

Furthermore, Winch’s theory has also been criticised for insufficient explanation on the two types of complementarity. According to Levinger (1964), given the example of a hypothetical man who is average in both “intensity” and “kind” in his needs, it would not be possible for him to meet a complementary partner because both the conditions stated for Type I and Type II required a person to be on either end of the spectrum in the “intensity” of the personality traits. However, Levinger offers a solution to this problem by suggesting that if both partners express the same behaviour in equal amount and has the need for it in moderation then they would complement as well as be similar to each other. Nevertheless, Levinger’s seemingly logical explanation of his suggestion is questionable because he based his assumption yet on another hypothetical example, which is like using X to prove X. Levinger assumes that if there is a given situation where both partners express the same need in moderate quantity and it has also been shown that they complement with each other’s need, then they are showing complementarity. Nonetheless, Levinger acknowledges the loose “footing” of his hypothesis and suggest that it should be tested for empirical evidence.

Another weakness in Winch et al. (1954)’s theory that has also been stated by Levinger (1964) is that Winch does not give clear guidelines to determine how a need would complement or be similar to another. Rather, Winch only gives examples of need-complementary pairs such as dominance-deference and nurturance-succorance. This lack of clarification may question the validity of the theory because it would be difficult to replicate the study using different complementary-need pairs such as practicality-impracticality, which can be a possible need pair. Nonetheless, this problem of deciding the criteria for which characteristic complements another, as offered by Levinger can be inferred by Schutz’s theory of need compatibility (Schutz, 1958, as cited in Levinger, 1964). Even though it is limited to only Type I in Winch’s theory but it provides the basis for complementarity where, for two people, person A and B, to complement one another in a relationship, person A should “express” the same amount of specific “need” that person B wants to “receive” and vice versa. Nonetheless, Edwards (1953, as cited in Levinger, 1964) points out that Schutz’s method would be difficult to test in married couples because it was intended to be used for testing “general peer relations” like friendship. Alternatively, we can also use the various circumflex models for interpersonal behaviour (Kiesler, 1982; Wiggins, 1982, as cited in Orford, 1986) to determine the possible complementary need pairs. However, similar to the criticism of Schutz’s way, these models were composed for the purpose of non-romantic relationships such as friendship and family interactions rather than the romantically-natured relationship between married couples, which might demand a different kind of structuring in Kiesler’s Interpersonal Circle.

Empirical Research on the Complementarity Hypothesis

Perhaps the main weakness in conducting researches to support the complementarity hypothesis in married couples is that there is the lack of clear criteria to determine what is considered as complementarity or more specific to Winch’s (1954) study, the conditions required to form the complementary need pairs. This is important because complementarity is often simplified down to the loosely used phrase of “opposites attract” for ease when explaining to the general population. This is troublesome because there is no clear definition of what is considered as “opposites” or how can a certain thing “attracts” another thing. For example, it is assumed that the general population of men are attracted to the opposite gender, which is women, demonstrating the complementariness. However, how can we explain the similar attraction between men and men, also known as homosexuality? This is where the definition of complementarity can be misleading and ambiguous. If this is taken back to the studies conducted to find support for the complementarity hypothesis, we can see that this confusion on what complements and what does not in a romantic relationship reflected in the methods of various studies, in particular the one followed below.

Review of “Complementarity in marital relationships” (Saint, 1994)

A somewhat recent study conducted by Saint (1994) aims to seek support for the theory of complementarity as a factor in mate selection for marriage. Using a questionnaire that contains nine statements, which the participants have to choose on a scale of nine from strongly disagree to strongly agree, Saint surveys twenty-eight couples who have been married for an average of twenty years. According to Saint, the mean age in the study for men are forty-seven years old and for women, it is forty-two years old and they are located in Oxford, England. Saint has concluded that result does not have enough support for the complementarity hypothesis.

There are many weaknesses in the method that Saint (1994) uses to source participants. By using the method of “door-to-door solicitation”, there is an implication that the participants are concentrated in local area since it is unlikely that the researchers will travel long distance to recruit participants. This assumption is supported when Saint states the geographical demographics of participants are “suburbs of Oxford” and “nearby village”. This small and narrow geographical population sampling can suggest a high degree of cultural bias in which there is a tendency to focus on the western individualistic views and also, questions on whether the results and the conclusion drawn can be used for the general population. However, Saint clarifies this doubt by concluding that this study is a good indicator for complementarity’s little impact on the Western’s selection of marriage. Thus, this gives the study its strength but still, it should be maintained that modern society are slowly changing and integrating both the Western and Eastern views (Zhuang, 2004). Therefore, the study cannot be used exclusively to explain that complementarity is not very influential in one’s selection for marriage, regardless whether it is from an Eastern or Western’s viewpoint. An example is interracial marriage (Lewis, Yancey & Bletzer, 1998), which would be possible fifty years ago but now is a common occurrence in our society.

Saint (1994) comments on the weak support of his study for the complementarity hypothesis, which can be reflected back to the design of questionnaire. To some extent, it is a forced response despite the varying scale of agreeableness to a statement. Even if this assumption is wrong and that participants do not feel restricted in their response, the scale of nine levels is still a weakness in Saint’s method. This is because given such a range for different responses; there is only a small sample of twenty-eight couples. More importantly, there is also the scarcity in the numbers of statements that can be used ascertain the reliability of answers and given the big range of at least eighteen different responses between couples, the only logical conclusion would be that the results would be “statistically insignificant”.

When summarising the results, Saint gave three statements that result in a significant negative correlation, which indicates the presence complementarity. They are “when socialising I seek a high level of eye contact”, “using public transport is a stressful experience”, and “when socialising I will raise the level of my voice to make myself heard”. Saint indicated earlier in his abstract in that he aims to investigate complementarity within the social domain of “social dominance, social confidence and communication initiation”. Whilst there are presence of these elements embedded in each statement but the relevance to married couple is not substantial, i.e. anyone can do the questionnaire whether with their friends or family members, and the chance to obtain a negative correlation is still probable. This indicates that these statements are not valid for testing and therefore, questions Saint’s conclusion that complementarity have little influence on mate selection for marriage.

Yet, we must not forget the strengths in the design of Saint’s (1994) study. One such strength is how individuals (husband and wife) have to work independently to complete the questionnaire. This increases the confidentiality of the answers because there are less comparisons and attempts to modify one’s answer, which engenders more honest responses obtained from participants and reliable information. Furthermore, questionnaires are economical when compared to other types of measurements like interview, easy to control and is replicable to confirm the findings. Thus, if this study is to be repeated, improvements could be made on the sample size and demographics. Moreover, changing the nature of the statements and increasing the number of statements to increase the reliability of the response could also help in seeking support for certain trends in participants’ answers.

Hence, from the studies of Winch (1954) though not so much of Saint (1994), it can be said that complementarity do exist in married couples. However, further investigations need to be made to confirm the claim that complementarity increases marital satisfaction over time.


It has been implicitly suggested throughout the essay that attraction is linked with similarity and couples have higher marital satisfaction when they have complementary needs. However, there is no definite conclusion that this is the case. Criticisms of Winch’s theory demonstrate the uncertainty and lack of clarification in certain aspects of his concept of need-complementarity. An example of this is the undefined need pairs such as dominance/deference, which later affects later studies like that of Saint (1994). However, in his evaluation of Winch’s concept, Levinger (1964) also explain that the replicated studies that did not support Winch’s theory is not conclusive either because the method used is more suitable for testing non-romantic dyadic relationship.

Likewise, the concept of similarity leading onto attraction may seem obvious and indeed there are numerous studies to support this hypothesis. However, the majority of them use the method of a bogus stranger to determine the effects of attraction. In addition, the use of partners in existing relationships was not common because of limited time and resource. This lead to the conclusion in that perhaps similarity leading to attraction is only valid in laboratory environment and not real life situations.

In both cases, the main problem that prevents proper investigation of the issue lies in both the methodological and theoretical aspect of the studies. A possible suggestion for improvements could be to have longitudinal researches instead of cross-sectional that involve couples in existing relationships. This is the biggest obstacle for researchers because of limited budget and time. Another is to increase the sample size theoretically by doing a meta-analysis of all the studies conducted over the years. In addition, more accurate and fitting instrument could also be developed to measure the degree of similarity or complementarity such as a revised version of Edwards Preference Schedule that is targeting romantic relationship and not general peer relationships.



Byrne, D. (1971). The attraction paradigm. New York: Academic Press.

Journal Articles

Antill, J. K. (1983). Sex role complementarity versus similarity in married couples. Journal of Personality and Social Psychology , 45, 145-155.

Aube, J., & Koestner, R. (1995). Gender Characteristics and Relationship Adjustment: Another Look at Similarity-Complementarity Hypotheses. Journal of Personality , 63, 879-904.

Bowerman, C. E., & Day, B. R. (1956). A Test of the Theory of Complementary Needs As Applied to Couples During Courtship. American Sociological Review , 21, 602-605.

Byrne, D., Clore, L. G., & Smeaton, G. (1986). The Attraction Hypothesis: Do Similar Attitudes Affect Anything? Journal of Pereonalily and Social Psychology , 51, 1167-1170.

Dijkstra, P., & Barelds, D. P. (2008). Do People Know What They Want: A Similar or Complementary Partner? Evolutionary Psychology , 6, 595-602.

Dryer, C. D., & Horowitz, L. M. (1997). When Do Opposites Attract? Interpersonal Complementarity Versus Similarity. Journal of Personality and Social Psychology , 72, 592-603.

Felmlee, D. H. (2001). From appealing to appalling: Disenchantment with a romantic partner. Sociological Perspectives , 44, 263-280.

Hill, C. T., Rubin, A., & Peplau, L. A. (1976). Breakups before marriage: The end of 103 affairs. Journal of Social Issues , 32, 147-168.

Kiesler, D. J. (1983). The 1982 Interpersonal Circle: A Taxonomy for Complementarity in Human Transactions . Psychological Review , 90, 185-214.

Levinger, G. (1964). Note on Need Complementarity in Marriage. Psychological Bulletin , 61, 153-157.

Lewis, J. R., & Yancey, G. (1997). Racial and Nonracial Factors That Influence Spouse Choice in Black/White Marriages . Journal of Black Studies , 28, 60-78.

Montoya, M. R., Horton, R. S., & Kirchner, J. (2008). perceived similarity Is actual similarity necessary for attraction? A meta-analysis of actual and perceived similarity. Journal of Social and Personal Relationships , 25, 889-922.

Orford, J. (1986). The Rules of Interpersonal Complementarity: Does Hostility Beget Hostility and Dominance, Submission? Psychological Review , 93, 365-377.

Saint, D. J. (1994). Complementarity in marital relationships. The Journal of Social Psychology , 134, 701-703.

Schellenberg, J. A., & Bee, L. S. (1960). A Re-Examination of the Theory of Complementary Needs in Mate Selection. Marriage and Family Living , 22, 227-232 .

Shiota, M. N., & Levenson, R. W. (2007). Birds of a Feather Don’t Always Fly Farthest: Similarity in Big Five Personality Predicts More Negative Marital Satisfaction Trajectories in Long-Term Marriages. Psychology and Aging , 22, 666-675.

Winch, R. F., Ktsanes, T., & Ktsanes, V. (1954). The Theory of Complementary Needs in Mate-Selection: An Analytic and Descriptive Study. American Sociological Review , 19, 241-249.

Zhuang, X. (2004). We or I? Collectivism-Individualism in Chinese and American Values. Sociology. University of Victoria.

You Might Also Like

I'm Alejandro!

Would you like to get a custom essay? How about receiving a customized one?

Check it out