Do these Google searches really reveal our deepest sexual anxieties?

This weekend in the New York Times, a Harvard-trained economist gave us a look at our deepest sexual anxieties via an unusual source: Google searches. Seth Stephens-Davidowitz mined Google searches to find that married people are more concerned about not getting sex than not getting love; “sexless marriage” is desperately typed into Google three and a half times as often as “loveless marriage.” And if you’re worried about your favorite body part being too small or too smelly, you are not alone. Stephens-Davidowitz says American men Google “how to make my penis bigger” or “longer” 8,000 times a month, while women are most likely to Google why their vaginas smell like “fish, vinegar, onions, ammonia, garlic, cheese, body odor, urine, bread, bleach, feces, sweat, metal, feet, garbage and rotten meat” (in that order). The most popular term attached to “my girlfriends vagina” in Google searches is “smells” and “stinks,” though “smells like condoms” also pops up from suspicious searchers. There’s a lot more data in the article.

Stephens-Davidowitz’s article is premised on the idea that when people are most anxious about their bodies — and how people are going to react to them in a naked state — they turn to Google. It’s not a terrible premise. Our Google searches are our psyche laid bare; if I had to choose between making my email public or making my Google search history public, I’d probably hand over my inbox first. It’s less embarrassing to ask a faceless search engine about weird smells from below than to ask a friend. But the article is a reminder that while Google may be faceless, it does have a brain and a memory. All of those embarrassing searches get logged and catalogued, and then, after being stripped of your identity and joined with other searches, returned to you in a Google Trends analysis or in a New York Times editorial.

Mining the popularity of Google searches to find out what it reveals about who we are is a favorite party trick by Stephens-Davidowitz. He has previously written articles for the Times about what pregnant women’s Google searches reveal about food cravings around the world; how Google searches for the n-word reveal racism across the country, and how many dudes in America are actually gay as revealed by their Google searches. Even though the data is anonymous and aggregated, the analysis can become privacy invasive. Based on the contents of their Google searches, Stephens-Davidowitz claimed that residents of West Virginia, western Pennsylvania, eastern Ohio, upstate New York and southern Mississippi are the most racist people in America.

Stephens-Davidowitz is a bit vague about his methodology in the “Searching Sex” study. He makes very specific claims in his article about how often certain terms are searched which is not data Google makes public. “Every year, in the United States, there are more than seven million searches looking into breast implants,” he writes. The article includes charts with the number of average monthly searches for particular terms; “sexless marriage” is searched 21,090 times per month on average while “My boyfriend won’t have sex with me” gets Googled 805 times per month, he tells us. (Oh, the pain of being confronted with the “I’m feeling lucky” button every time you reveal to Google that you’re not getting lucky.)

When asked about his sources, Stephens-Davidowitz says he used Google Trends and Google Adwords. Google Trends though only reveals the “interest” in a particular term over time, by rating the relative popularity of that search in comparison to all searches on Google. Perhaps he’s done some “data magic” to combine his Trends data with his Adwords data and come out with hard numbers. He has written before about “hacking” Trends data to reveal more than it was supposed to. In the racial study he did, the use of the offensive term he was looking for was too seldomly searched in some states to register in Trends, so he came up with a workaround. From his paper: “Get search volume for ‘weather+nigger(s),’ searches that include either ‘weather’ or ‘nigger(s).’ Subtracting search volume from ‘weather’ from search volume from ‘weather+nigger(s)’ will give approximate search volume for ‘nigger(s).'”

Regardless, it’s hard to say if his sex searching methodology is sound. He uploaded an Excel sheet to his website that includes the data behind this article but it’s hard to parse where the numbers there actually come from. We may find out more in the future. His website notes that he has a forthcoming paper about methodology co-authored with Google chief economist Hal Varian titled “Google Trends: A Primer for Social Scientists.”

Even if his method is sound, his study seeks out our sexual anxieties based on his ideas about what they should be. He is telling us the prevalence of searches for particular terms he designated. He expected people would be concerned about smelly vag and dicks that don’t measure up, and the data backed him up. Were we to get to peek into Google’s vast database, the findings might be less predictable than what we get here. What Stephens-Davidowitz can tell us about our anxieties is limited by his imagination about what our anxieties are.