How Googling Unmasks Child Abuse

By Seth Stephens-Davidowitz (New York Times)
July 13, 2013

During the Great Recession, child abuse and neglect appeared to decline. Incidents reported to local authorities dropped. “The doom-and-gloom predictions haven’t come true,” Richard Gelles, a child-welfare expert at the University of Pennsylvania, told The Associated Press in 2011.

The real story about child maltreatment during the recession is a grim one. I spent months studying this topic, using a number of different data sources, including Google search queries. I found that the Great Recession caused a significant increase in child abuse and neglect. But far fewer of these cases were reported to authorities, with much of the drop due to slashed budgets for teachers, nurses, doctors and child protective service workers.

Here’s what caused the initial optimism: from 2006 to 2009, the number of cases reported to child protective services decreased by 1 percent nationally. Remarkably, the biggest drops were in the states hardest hit by the recession. Nevada, despite an unemployment rate that rose as high as 13.3 percent, showed a 17.5 percent drop in reports of maltreatment of children.

The first clue that the official statistics were misleading comes from looking at the most extreme forms of abuse and neglect, which are least susceptible to reporting pressures: child-fatality rates. During the downturn, there was a comparative increase in these rates in states that were hardest hit by the recession. From 2006 to 2009, Nevada’s fatality rate from abuse or neglect rose 50 percent.

But child fatalities are, thankfully, rare. I also used a novel technique for studying child maltreatment: an analysis of anonymous, aggregate Google searches. (I am currently an intern at Google, but I finished the doctoral research on which this essay is based before joining the company.) Google queries provide an immensely powerful database, particularly on sensitive topics that people don’t discuss freely with pollsters or authorities or even the friends and family members they know best. Online, often unobserved, we tend to be very honest.

I examined a heart-wrenching category of searches: those likely to have been made by recent victims of abuse who were old enough to use Google. These searches included “My dad hit me” or “Why did my father beat me?” I also examined a more common class of Google queries: those that include the words “child abuse” or “child neglect.” In some sense, this Google data is like a survey of how many people suspected child maltreatment at a given time. If you see something that worries you, you may well ask Google about “child abuse signs” or “child abuse effects.”

Sure, some people search for “child abuse” for other reasons, but these irrelevant searches can often be parsed out. The Google numbers are so large — many orders of magnitude larger than in any survey or poll — that the overall rates are telling. And they are likely to capture suspected incidents that we would never know about otherwise.

After declining for many years in the United States, the searches that seem to have come from abuse victims themselves rose as soon as the Great Recession began. On weeks that unemployment claims rose, these searches rose as well.

Searches that appear to have originated with people who suspect abuse also provide evidence that the increase is caused by the economic downturn. Controlling for pre-recession rates and national trends, states that had comparatively suffered the most had increased search rates for child abuse and neglect. Each percentage point increase in the unemployment rate was associated with a 3 percent increase in the search rate for “child abuse” or “child neglect.”

If the Great Recession increased suspicion of child maltreatment, why were fewer cases reported? Keep in mind first that many, probably most, suspected cases are never reported. Even primary care doctors, who are legally mandated to report suspected child abuse, admit in surveys that they do not report 27 percent of suspicious incidents.

It is certainly plausible that an economic downturn could lower the rate at which suspect cases are reported. Budgets were slashed in hard-hit states, particularly on social programs directed toward children. Overworked teachers, doctors and nurses may be that much less likely to go through with the reporting process. Shorter hours and more thinly stretched staffs at child protective service agencies may make it harder to report cases. There is abundant anecdotal evidence of people attempting to report maltreatment by phone facing long wait times and hanging up.

When you compare places that Google search data suggest have similar levels of abuse or neglect, you find that the less an area spends on social services for children, the lower its reported rates of child maltreatment. My research also shows that when a particular group’s budget is reduced, it reports fewer cases of maltreatment. Cut resources for teachers, for example, and teachers report fewer of their suspicions.

THERE are four take-aways from this research. First, the maltreatment of children is yet another cost of the Great Recession, one that will be felt long after the economy fully recovers. The evidence from medical researchers and psychologists is overwhelming: as adults, victims of child abuse or neglect will face higher probabilities of mental illness and criminal behavior and lower probabilities of employment and stable family lives.

Second, we should be skeptical of statistics based on official reports of crime in general, not just child abuse or neglect. According to government surveys, between 2006 and 2010, throughout the United States, 52 percent of violent crimes, 60 percent of property crimes and 65 percent of rapes and sexual assaults were never reported to the police. When reported crime drops, it is always possible that this is a result not of a decline in crime itself but of factors that make it more difficult to report crime.

Consider, for example, Detroit. Its police department recently reported 20 percent reductions in some major crimes. Might this be because of dwindling police department resources? The average time it takes to get a response to an emergency call to the Detroit police is now 58 minutes, and many precincts have stopped taking crime reports in evening hours.

Third, Google search data can fill holes in our understanding of crime generally. If your iPad was stolen, whom would you be more likely to tell: the police or Google? Indeed, I have found that Google queries for “stolen iPad” and “stolen iPod” yield meaningful information about property crime rates. Searches like “I was just raped” and “rape hot line” might help us measure a city’s true rape rate. (Here’s a disturbing side note: Rape proves the most difficult to measure crime using Google data, but not because of women’s historical reluctance to report rapes. The problem is that the majority of rape-related Google searches are typed in by people looking for pornography.)

Fourth, and most important, the contrast between the search data and the reported data tells a sad story about social services in this country. Just when more children are searching for help, we decimate the budgets of the very people who might actually do something to protect them.

Seth Stephens-Davidowitz is an economist who recently received a Ph.D. from Harvard.