Can Data-Driven Tools Predict How Long Relationships Will Last?
Advancements in online dating and social media platforms mean it may have never been easier to find a partner who is compatible with you, at least in theory. Internet dating platforms feed algorithms with information about those seeking a relationship in order to find the best match for them. Machine learning in the form of computer matching has become extremely sensitive and accurate.
A new study analyzes whether this predictability can be applied to a relationship. Is it possible to foresee from the start whether it will last? And, do we want to know the probability of when a relationship may run its course and end?
Psychologists of Friedrich Schiller University Jena, Germany, and the University of Alberta, Canada, have looked into this question and have come to a clear conclusion. “Predictions as to the longevity of a relationship are definitely possible,” said Dr. Christine Finn from the University of Jena.
As part of the long-term study called “pairfam,” she held regular interviews over seven years with nearly 2,000 couples, 16 percent of which broke up during this period.
Pairfam, short for “Panel Analysis of Intimate Relationships and Family Dynamics project,” is a longitudinal research endeavor among four German universities. Investigators have been studying the development of 12,000 individuals of various ages since 2008. The study, which is funded by the German Research Foundation, will run into 2022.
Finn said, “Right at the outset of a relationship, one can find typical features — that is to say certain prediction variables — that provide information on whether or not the relationship will be long-lasting.”
In psychology, there are currently two scientific models, which describe the course of a relationship in different ways, she said.
One posits that all couples are initially more or less equally happy. If the relationship ends in separation, this can be traced back to problems that only developed over the course of the couple’s time together.
The second model assumes that the two individuals in a couple start at different levels of happiness. They generally maintain these levels, but a more negative initial situation increases the likelihood of failure.
According to Finn, the new findings suggest that there is actually a combination of the two models.
“We too can confirm that there are differing levels to begin with. In addition, happiness declines in both groups. However, in those who later separate, this happens significantly faster, meaning that a person who starts off unhappy becomes increasingly unhappy.”
Therefore, the beginning of a relationship can reveal something about how it will progress. The researchers in Jena determined satisfaction by, for example, asking couples to what extent they considered that their needs were being met.
In general, people who have similar needs, for example a need for closeness, but who also want to be able to continue pursuing their own interests, usually stay together longest.
Researchers say the expert machine learning algorithms can provide couples advance information on the probability of their staying together. But is such information useful?
Christine Finn is skeptical. “It is not our intention to further reinforce the general trend for optimization and only to have a relationship that is result-oriented, with the prospect of it being long-lasting,” she said.
“Even if couples split up after a time, it can still be a valuable and important phase in their lives, which might have a positive influence on the next relationship. Furthermore, couples can also consciously influence and work on their mutual interests and on cultivating closeness as well as independence. No relationship is doomed to fail from the outset.”
Nauert PhD, R. (2020). Can Data-Driven Tools Predict How Long Relationships Will Last?. Psych Central. Retrieved on December 4, 2020, from https://psychcentral.com/news/2020/03/04/can-data-driven-tools-predict-how-long-relationships-will-last/154681.html