Machine Learning Identifies the Most Important Predictors of Infidelity
Researchers have used a machine learning approach to identify factors that predict someone's likelihood to cheat in a relationship. Their findings were published late August to the Journal of Sex Research.
Authors Laura M. Vowels, Matthew J. Vowels, and Kristen P. Mark ran an algorithm through in-depth survey data from 891 participants gathered in 2014. The anonymous questionnaire asked about sexual satisfaction, relationship satisfaction, sexual behaviors, romantic and sexual desires, a range of demographic information, as well as whether or not the participant had had sex with someone other than their partner during their most recent or current relationship. About a third of survey takers admitted to this kind of infidelity – 43.4% of men and 25.7% of women.
So what factors predicted cheating? In men, the algorithm identified lower relationship satisfaction and a lack of romantic love as the two largest drivers. A higher desire to masturbate solo and decreased sexual satisfaction also contributed, though to a lesser extent. Interestingly, men who did not receive oral sex in the last month also were much more likely to cheat – it was the fourth most powerful predictor of infidelity.
In women, the two strongest predictors of infidelity by far were lower levels of romantic love and a higher desire to masturbate by oneself. Lower sexual satisfaction, longer relationship length, good physical health, and lower relationship satisfaction also contributed to the likelihood of infidelity.
The authors stress that these are just associations, and no single variable alone predicts somebody's likelihood to cheat – all of the factors add up to forecast infidelity.
As the authors concluded:
The present study provides the most robust and reliable evidence of factors associated with past in-person and online infidelity. The results showed that relationship variables were the most robust predictors of infidelity whereas demographics and individual differences variables were not consistently associated with infidelity. These results suggest that intervening in relationships when difficulties first arise may be the best way to prevent future infidelity. Furthermore, because sexual desire was one of the most robust predictors of infidelity, discussing sexual needs and desires and finding ways to meet those needs in relationships may also decrease the risk of infidelity.
Source: Laura M. Vowels, Matthew J. Vowels & Kristen P. Mark (2021) Is Infidelity Predictable? Using Explainable Machine Learning to Identify the Most Important Predictors of Infidelity, The Journal of Sex Research, DOI: 10.1080/00224499.2021.1967846