Similarity in Ongoing Relationships

Definition and Measurement

Similarity in an ongoing relationship is how alike two relationship partners are at one point in time. Romantic partners and friends are generally the most similar on demographic characteristics such as age, education level, and socioeconomic status, as well as intelligence and cognitive abilities. To a lesser extent, partners are similar across behavioral and psychological characteristics, including personality, attitudes, emotional tendencies, interests, psychopathology, and sexual experience. This tendency of partners being alike is also known as homogamy. There is extensive evidence for homogamy and less evidence for heterogamy, or the tendency for long-term partners to be dissimilar to each other.

Researchers measure similarity in a variety of ways. The first is a discrepancy score. The absolute difference of two partners' scores is used to measure how similar two partners are on a given characteristic with lower scores indicating similarity. Often, absolute differences across a number of different questions measuring the same characteristic are averaged together. Discrepancy scores have some important limitations. First, greater scores indicate greater dissimilarity; however, there is no set point for when an individual stops being similar and starts being dissimilar. Second, discrepancy scores can be confounded with each partner's individual score. If an individual scores a 5 out of 10 on an agreeableness scale, he or she can have a maximum discrepancy of 5 from a partner, whereas if an individual scores a 10, he or she can have a maximum discrepancy of 9.

A second way of measuring similarity is a variable-centered correlation, where partners' scores are correlated on the same characteristic (height, extraversion, intelligence). This is a correlation between one partner's score with the other partner's score on the same characteristic across a sample of dyads. Higher correlations indicate greater similarity. This approach also has limitations. First, this approach can only estimate similarity one characteristic at a time and cannot show how similar couples are across multiple characteristics. Second, a variable-centered correlation shows levels of similarity across an entire sample and cannot estimate the similarity for one specific couple.

A third way of measuring similarity is by computing a couple-centered correlation. Here, a profile correlation of two partners' responses is computed. This is a correlation between one partner's responses on a number of items with the other partner's scores on those same items within the same dyad. This approach addresses some of the limitations of the other methods of computing similarity. It can be an index of when couples are similar (positive correlations) or dissimilar (negative correlations). It can also be used to measure similarity across a single characteristic, such as extraversion, or multiple characteristics, such as an overall measure of personality. However, computing this correlation requires that there be multiple items assessing the characteristic, so it cannot be used when the characteristic is measured by a single item, such as [Page 1497]height and age. Thus, although this method of computing similarity has the fewest limitations, it cannot be used for all characteristics.

...