Pornography Use and Marital Separation:

Evidence from Two-Wave Panel Data



With the increased availability and consumption of pornography in the US, researchers have been studying the effect of it’s use (see article for bibliography). Perry points out that most studies look at the effect of porn use on relationship quality not on relationship stability. Focusing on heterosexual married couples and using data from the Portrait of American Life Study (PALS), Perry “examined whether viewing pornography in 2006, either at all or in greater frequencies, predicted a higher likelihood of experiencing a marital separation by 2012.”


Two questions were pulled from the PALS survey to measure the variables. For the first variable, respondents in the 2012 survey identified if they “had a separation due to marital difficulties” in the last six years. The second variable was assessed in 2006 with an eight-point Likert scale response (never to 1x/day or more) to the question “In the past 12 months, how often have you viewed pornographic materials?”.

Perry controlled for several 2006 variables including marital satisfaction, sexual satisfaction, previous divorce/annulment, age, children (in or out of home), region of the country, race, education, and household income. As the PALS study looks specifically at the impact of religion on daily life, variables were also available for religious commitment, conservatism, and tradition and results were controlled for these variables.


About 35% of married participants in 2006 reported some level of porn use.

Any use of porn in 2006 was correlated with experiencing a marital breakup by 2012 (r = .15, df = 443, p < .001). When testing for the amount of porn use in 2006 with separation, a curvilinear relationship was found (see chart).

As pornography viewing frequency increased from “never” in 2006, the likelihood of marital separation by 2012 also increased, but only to a point. Those who indicated they watched pornography about “2 or 3 times a month” in 2006 had the highest likelihood of experiencing a marital separation by 2012 (nearly 1 in 3). At higher frequencies of pornography use, however, this likelihood appeared to decline to where those who reported viewing pornography at least once a day in 2006 were slightly less likely to experience a marital separation by 2012 compared to those who “never” used pornography in 2006.


Perry then split the porn use into three groups – never using, those using up to 2-3x per month, and those using weekly or more. Results showed a statistically significant difference in those who used up to 2-3x per month but those who used weekly or more were not significantly different from those who never used or those who used up to 2-3x per month. Perry pointed out that the small number of the sample who used weekly or more (5%) made finding statistical significance problematic.

“Findings affirmed that married pornography users were more than twice as likely to experience a marital separation in the 6 years following their reported pornography viewing compared to those who did not view pornography.”

“Tests for interactions with gender also showed that the associations between pornography use and marital separation held regardless of whether the participant was male or female.”


So, does the main point of the research hold up (blue text above)? Technically, yes. But Perry points out several issues.

  1. That reported porn use in 2006 shows a significant prediction to separation in 2012 does not mean the porn use caused, or even contributed to the separation.
  2. PALS data did not ask for type of porn use, or if used in isolation or as a couple.
  3. Many factors could not be controlled for that are known variables (i.e., personality characteristics of spouses) nor was the data dyadic. Perry points out that those with higher use might be more open, even viewing it with their spouse where those with more infrequent use might be hiding it.

Another limitation of the study is the lack of data on porn use at time 2. PALS data did not ask the pornography question in 2012 so porn use over time is not known. It is possible that a percentage of subjects reporting weekly or more use in 2006 might have stopped use by 2012 and those with moderate use had continued to escalate use. This is not known.

I have several thoughts on why higher use did not predict separation, but none of them can be tested with the data available. It is possible that higher use was equated with greater transparency – i.e., both spouses were aware of the use and it wasn’t a distractor in the marriage. They might even use it together. It is possible that moderate use came with high levels of shame, self condemnation and partner blame (i.e., if you would have sex with me more often I wouldn’t look at the porn) or self blame (you look at porn because I’m not enough). We know these attributions are highly destructive. So, like much of research, we have more questions than answers.

Finally, even though technically accurate that those who used porn were more than twice as likely to experience a marital separation in the 6 years following reported use, the effect was not large. While statistically significant (p<.001) an r value of .15 is a very small value. If memory serves, this means the porn use accounts for less than 5% of the variance. This suggests there are much bigger factors at play in the separation than the porn use.

So What? …

Overall you would be technically accurate if you reported a proven link between report of porn use in marriage and separation. This is what the data showed. However, keep in mind that when the researchers broke down the data, these results were not valid at higher levels of use. Finally, take care not to suggest a strong relationship nor in any way imply a causal relationship (i.e., “use of porn shown to lead to separation” would be inaccurate).

If you hear, or read, someone reporting porn use has been proven to lead to marital separation, maybe you can gently help them be accurate.


Perry, S. L. (2017). Pornography Use and Marital Separation: Evidence from Two-Wave Panel Data. Arch Sex Behav. doi:10.1007/s10508-017-1080-8