Does Porn Keep Men from Marrying or Not?

We live in the Age of Social Science. No political argument can be made without an accompanying study or survey.

As inexorable as the sun moving across the sky, each study is then criticized by opponents for its methodological flaws. When you don’t like the results, you can always go after the method.

All sides come to the political debate with their own scholars, and all sides have their own debunkers of social science studies.

One such debate is playing itself out on the question of pornography. Porn is much studied these days. As it has become an epidemic, policymakers, scholars, and activists are interested in the affect it has on the user and on society at large.

Professor Michael Malcolm of West Chester University in Pennsylvania and Dr. George Naufal of London-based Timberlake Consultants issued a draft paper in November looking at the affect Internet porn use may have on marital formation.

In their highly technical paper, the authors postulated, “Substitutes for martial gratification may impact the decision to marry.” They call Internet porn “an increasing low-cost substitute” and that Internet use itself “is negatively associated with marriage formation” but that “pornography consumption specifically has an even stronger effect.”

Using data collected for several years from the General Social Survey, the authors indeed found “…for young men there is a large degree of substitutionability between Internet and pornography usage and marriage—heavy Internet usage generally, and use of pornography specifically, are associated with lower participation in marriage.”

Their study got immediate coverage in the Washington Post, and The Independent in the UK, and likely many other places.

Jordan Weissmann of Slate says, “not so fast.” He calls the study “pure bunk” and “deeply silly.” And his criticism is based on methodology and the old bugaboo “correlation vs. causation.” Just because one thing seems to be connected to another thing, does not mean one caused the other. Weissmann telegraphs his general skepticism with concerns about porn use. He says the paper is getting a lot of attention “because, you know. Pornography. Tearing our society apart. Or something.”

Weissmann’s main concern is the two authors’ use of something called “instrumental variables.” Instrumental variables are the use of one seemingly unrelated thing to help explain another.

Weissman uses smoking and public health to explain. Suppose you’re measuring the effect of cigarette smoking on public health, “and your data tells you that people who smoke tend to be sicker than the rest of the population.” But there could be other reasons than smoking, so why not measure public health in correlation with increased cigarette taxes; higher cigarette taxes mean better health. That is an “instrumental variable.”

But, Weissmann says the “instruments” chosen by the authors are “strange.”

The authors used the father’s level of education to measure overall Internet usage since “children of educated parents have access to more technology.”

For porn use, the authors used urbanization because of “availability, speed and cost of Internet access.” Porn is bandwidth-heavy, and the authors looked at data from 2000-2004, a time when dial-up Internet was still common.

Weissmann says, “The problem is that there are lots and lots of other ways, aside from Internet or porn usage, that have having an educated dad or living in a city could change your propensity for getting married.” For instance, someone in the city might have more dating options and therefore puts off getting married.

It is unclear whether Weissmann contacted the authors to discuss his concerns. He quotes the authors but only from their paper.

The two authors shared their reactions to Weissmann’s piece with Breitbart News.

George Naufal told Breitbart News, “If one reads the paper carefully, all of the data issues are discussed in details. The wording in the paper is careful and takes into consideration the contribution of an instrumental variable approach while not ignoring the potential weakness of such methodology.”

Professor Malcolm went into much greater detail with us: “Regarding the criticism of our instruments, these instruments are common in the literature” and that such criticism as Weissman’s can be used on almost any “instrument.”

“Even the author’s own cigarette tax example could suffer from the same problem,” said Malcolm. “The idea is that cigarette tax is different in different areas, and this is an exogenous reason for variations in smoking rates. But there could be all kinds of unobserved differences between these locations that are endogenous to smoking behavior, and maybe different people move to areas with more favorable attributes that are correlated to the cigarette tax rates. Think Kentucky versus New York City.”

Malcolm said, “The deep causal structure is always difficult to disentangle, and I believe we have qualified the results properly (many are quoted in the Slate article). You will find these caveats in basically any paper that uses instrumental variables analysis. I would venture to say that the only time a researcher can be 100% sure that the causality is right, is to run a randomized experiment. But, again, no better data are available. Are there valid criticisms of this approach? Sure, but it’s not like we’ve gone off the rails. We would love to have better instruments, and we said so in the paper. But, in the absence of better data, this is not an uncommon approach in economics. Properly qualified, we think it is informative.”

So, once more, it is a battle among experts over social science and methodology.