“I took the Marriage Pact questionnaire and it matched me with my boyfriend!”
People slide into our DMs all the time and tell us this. It isn’t uncommon for our algorithm to match existing couples to each other, since it is not uncommon for couples to be already similar to each other.
At first glance, this might suggest our algorithm has reached the perfect level of accuracy for predicting compatibility —If we are already matching existing couples to each other, why should not we expect that all the matches we give out will make perfect relationships?
That would be an easy interpretation, and the relationship science team does not settle so easily. The more we thought about this, and the more we researched about similarity in relationships, we faced a very intuitive question:
Are existing couples compatible because they were similar to each other in the first place, or did they grow similar to each other over the course of their relationship?
These are not necessarily mutually exclusive concepts, yet we were curious to explore whether one of them mattered more when explaining the similarity between couples. The relationship science literature defines the former as homophily, or assortative mating, the tendency of similars pairing up at a higher rate than dissimilars, and describes the latter as attitude convergence, people growing similar to each other over time due to the time spent together.
A longitudinal study conducted by eHarmony Labs in collaboration with UCLA matched participants with a pool of potential compatible matches depending on how they answered a relationship questionnaire and collected data on how these relationships progressed over years.
Results indicated that, even though the pool of potential compatible matches one received displayed significantly more similarity in personality and values than what random matches would have displayed, couples who eventually married had even greater levels of initial similarity than the matches who were not pursued. Further, similarity in personality didn’t change over time for the married couples, which is not surprising considering change in personality is gradual across adulthood. As far as attitude convergence is concerned, the only supporting evidence was convergence in interests and emotional experience for the married couples.
These results provide significant evidence favoring the homophily hypothesis over the attitude convergence one. It seems likely that the couples who match each other when they fill out the Pact are actually similar in the first place!
Does this mean we should expect that all the matches we give out will make perfect relationships? Maybe. More likely than not.
Meanwhile, we will keep researching and informing you.