When crime rates begin to increase; when residents, business owners, tourists, or visitors begin to feel unsafe; when local politicians want to appear tough on crime; and when high-profile violent crimes are splashed across the headlines, the most common policy response is to hire more cops. For example, in his 1992 presidential campaign, Bill Clinton promised to ”fight crime by putting 100,000 new police officers on the streets” (Clinton 1992, 72). Once in office, he fulfilled that campaign promise with the enactment of the 1994 Crime Act. Clinton’s rationale for wanting to hire more police was simple and intuitively appealing:
Our crime bill fulfilled a commitment I made to the American people to put 100,000 new police officers on the street in community policing. It’s an old-fashioned idea, really. It means put the police back on the street, in the neighborhood, working with neighbors to spot criminals, shutting down crack houses, stopping crime before it happens, getting to know children on the street and encouraging them to stay away from crime. (Clinton 1996)
In a 1995 Atlantic Monthly article, one prominent police reformer, Adam Walinsky, suggested that the United States needs at least 500,000 new police officers to protect us from rising crime rates.
This affinity for hiring more police officers to reduce crime appears to transcend time and place—it is a universal instinct for achieving safer communities. From tiny developing nations to heavily populated industrial democracies, the pressure to hire more police is omnipresent. Research has shown that growth in the world’s policing industry has outpaced population growth, which means the ratio of police to citizens is increasing worldwide (Maguire and Schulte-Murray 2001).
Although the idea of hiring more police to control crime is understandable, the evidence for its effectiveness is weak. Social scientists have studied the relationship between the number of police officers (known as ”police strength”) and crime rates for many years. If we look closely at this body of research, we find that it is exceedingly complex and often reaches unclear or contradictory conclusions.
Why is the research on the relationship between police strength and crime so complex? One reason is that it suffers from what statisticians call an ”identification” problem. In other words, the statistical equations used to model the relationship between police and crime often lack sufficient information for the model to be identified or solved. The reason for this statistical difficulty is that the relationship between police and crime is reciprocal or simultaneous— the amount of police may influence the crime rate, and the crime rate may influence the amount of police. If we find a positive relationship between police and crime (as one increases, so does the other), is it because places with high crime hire more police, or because there are more police officers available to detect and report crime? If we find a negative relationship (as one increases, the other decreases), is it because places with high crime cannot afford to hire a sufficient number of police, or because more police means less crime? For these reasons, merely finding a statistical relationship (or a correlation) between police and crime is not enough to conclude that one ”causes” or exerts an influence on the other.
Due to the model identification problem, complex statistical methods are often used to assess the relationship between police strength and crime. These studies are difficult to understand and evaluate for somebody without advanced statistical or econometric training, thus they are often inaccessible to those who can use them most such as police chiefs, politicians, and the media.
In 2000, a pair of university criminologists, John Eck and Edward Maguire, reviewed every published study that had ever investigated the relationship between police strength and violent crime. These twenty-seven studies contained eighty-nine separate estimates of the effects of police on violent crime. Of these estimates, 49.4% found no relationship, 30.3% found a positive relationship (more police, more crime), and 20.2% found a negative relationship (more police, less crime). The authors concluded that they ”could not find consistent evidence that increases in police strength produce decreases in violent crime” (Eck and Maguire 2000, 217).
One of the most well-known papers reviewed by Eck and Maguire was written by the popular economist Steven Levitt (1997), who found that hiring more police results in less crime. In a replication of Levitt’s (1997) analysis, McCrary (2002) criticized his methods and concluded ”In the absence of stronger research designs, or perhaps heroic data collection, a precise estimate of the causal effect of police on crime will remain at large.” Levitt (2002) then replied by using different methods and finding the same results as his previous study. Though social scientists continue to use better data and more sophisticated methods, the results are often not very illuminating for practitioners and policy makers trying to decide how to spend taxpayer’s money to reduce crime.
Two more recent studies took advantage of the increase in police presence during periods of terrorist threat—a sort of ”natural” experiment—to examine the effects of police on crime rates. Klick and Tabarrok (2005), for instance, found that the heightened police presence associated with terror alerts resulted in no effect on violent crime, but a substantial decrease in auto thefts and thefts from autos. That analysis relied on daily crime data from Washington, D.C. Similarly, DiTella and Schargrodsky (2003) found that auto thefts decreased significantly during the heightened police presence that occurred in the aftermath of a terrorist incident at a Jewish center in Buenos Aires. In another recent study, Corman and Mocan (2003) used data collected over time in New York City, and found that increased police presence had an effect on motor vehicle thefts, but not on other types of offenses. Thus, three careful, recent studies have now found that increased police presence can reduce auto theft.
Taken together, the studies that were completed after Eck and Maguire’s review of the research suggest that police strength may have effects on some offenses but not others. Altogether, combining the most recent research with the earlier findings summarized by Eck and Maguire, it seems that the scientific evidence on the effects of police strength on overall crime rates and violent crime rates is unclear.
Why are the research findings inconsistent and often contradictory? One reason is that different research methods often produce different findings. Another reason might simply be that increasing the number of police officers may have a dramatic effect in some places and times but not in others. Moreover, it may deter some offense types, such as those committed outdoors (like auto theft), but not others. In other words, this universal ”cure” for unsafe communities may only work in some places, at some times, and under some conditions.
There are good reasons to support this interpretation of the evidence. As Sherman (1997) notes:
Hiring more police to provide rapid 911 responses, unfocused random patrol, and reactive arrests does not prevent serious crime. Community policing without a clear focus on crime risk factors generally shows no effect on crime. But directed patrols, proactive arrests and problem-solving at high-crime ”hot spots” has shown substantial evidence of crime prevention. Police can prevent robbery, disorder, gun violence, drunk driving and domestic violence, but only by using certain methods under certain conditions.
The research evidence suggests that what police do may be far more important than how many of them there are. Hiring more police may have a large effect when the officers are assigned to perform the tasks of problem-oriented policing or hot-spot policing that have been found highly effective in reducing crime. When officers are deployed inefficiently or they are asked to perform tasks having a low probability of reducing crime, then the number of police may not matter. This conclusion is consistent with the findings of a committee of social scientists convened recently by the National Research Council to review research on policing. The committee found that previous research on the effects of police strength ”are confounded with . . . the effects of changes in the strategies of policing” (National Research Council 2004, 225).
Some communities in the United States are chronically understaffed and could benefit from hiring more police officers. Other communities have too many police—where scarce community resources might be used to greater benefit by hiring more teachers or social workers or investing in more effective crime prevention strategies. There is no magic number of police—no ratio of police to citizens that is right for every community. The challenge is looking carefully at police operations, making sure that police are using the right tactics and strategies in the right places at the right times—and then hiring a sufficient number of police to get those things done.
