Almost without exception, Americans believe that violent crime is increasing. In the short run, they are right: Violent crime did increase between 1985 and 1990. But what really worries most people is not the short-run trend but their sense that violent crime has been climbing steadily for a long time and that the future will only bring further increases. Such worries are linked to anxiety about drugs, permissive childrearing, hedonism, declining academic standards, the growth of the ghetto underclass, and our collective inability to compete with the Japanese. Taken together, these fears have convinced many sensible people that American society is on the skids.
America certainly has more violence than other rich countries. Murder rates are far higher in the United States than in Europe, Japan, or even Canada. We also have more rapes, robberies, and assaults than other rich countries. But this is nothing new. Crime rates have always been much higher in America than in other affluent nations. Indeed, violence is part of our national mythology. We shed more blood settling our frontier than any other New World nation, and we made more movies glorifying the bloodshed. Our struggle over slavery was also far bloodier than any other nation's. We have lived with this grim heritage for a long time.
For those who fear that American society is coming unglued, however, the question is not how America compares to other countries but whether our traditional ways of containing violence have broken down. Here the answer is more ambiguous. America is more violent today than at many times in its past. But it is no more violent than it was during most of the 1970s. Thus, there is no obvious reason for thinking that chaos is just around the corner.
The best available indicator of long-term trends in violence is the murder rate. An American's chance of being murdered was relatively low in the 1950s and early 1960s. It doubled between 1964 and 1974, remained high from 1974 to 1980, declined significantly between 1980 and 1985, and edged back up in the late 1980s. In 1989 the murder rate was higher than it had been from 1983 to 1988, lower than it had been from 1972 to 1982, and higher than it had been from 1950 to 1972. Victimization surveys -- that is, surveys asking people whether they have been the victims of crimes -- suggest that non-lethal violence has followed the same trajectory. Furthermore, black-white differences in the incidence of violence have been diminishing, not increasing.
Nonetheless, most Americans are convinced that America has become much more dangerous. One reason is that American cities really are considerably more violent than they were between 1945 and 1965, when middle-aged Americans were growing up. But even younger Americans, who grew up in the late 1960s and 1970s, think America has become more violent. Here the explanation is subtler. When most of us think of the past we think of our childhood. Most middle-class Americans grow up in placid residential neighborhoods where violent crime has always been quite rare. Middle-class adults lead less sheltered lives. They usually work in cities rather than suburbs, and they expose themselves to risks to which they would never expose their children. When today's children grow up and remember their youth, they too will think the world has grown more violent, even if the crime rate remains unchanged.
Still another reason we think America has become more violent is that American journalists are concentrated in New York and Washington. The latter really did experience an extraordinary increase in lethal violence during the 1980s, and this increase got far more attention than what happened in San Diego or Atlanta or Omaha.
The mass media also have a very selective approach to crime statistics. When crime declines, as it did in the early 1980s, editors assume the decline is only temporary and give it very little air time. When crime increases, as it did in the late 1980s, both journalists and editors see the increase as a portent of things to come and give it a lot of play.
On July 18, 1990, for example, The New York Times carried a front page story with the headline "Number of Killings Soars in Big Cities Across U.S." The story focused on trends between January, 1988, and June, 1990. It referred briefly to the fact that the national murder rate was considerably lower in 1988 than in 1980 but concentrated on explaining the recent increase, not the earlier decline.
Two weeks later the Senate Judiciary Committee released a report predicting that the number of murders would reach an all-time high in 1990. This was front-page news across the country. But what the report neglected to mention, and most journalists also failed to note, was that the population will also reach an all-time high in 1990. Once we take this into account, the projected murder rate for 1990 is no higher than it was during most of the 1970s.
The same journalistic bias operates when government agencies issue contradictory reports. Every year both the Federal Bureau of Investigation (FBI) and the Bureau of Justice Statistics (BJS) issue reports showing how the crime rate changed during the previous year. The FBI almost always reports that violent crime increased, and these reports almost always make the evening news. The BJS almost always reports that violent crime was constant or declined. Its reports, while more reliable than the FBI's, get far less attention.
But this contrast raises another question. Why should two government agencies constantly reach different conclusions about trends in violent crime? Neither the FBI nor BJS fudges its data for political reasons. But whereas BJS relies on survey data that are collected in the same way year after year, the FBI relies on administrative data, which are generated in a slightly different way every year. These incremental changes make FBI statistics on non-lethal violence almost useless for analyzing changes over time.
Police Estimates of Crime
Every year most local police departments calculate the number of crimes committed within their jurisdiction. They forward these estimates both to the local news media, which usually give them considerable attention, and to the FBI, which uses them to estimate the number of crimes "known to the police" nationwide. These FBI estimates suggest that violent crime has increased by a factor of four since 1960 and is now at an all-time high.
The FBI's index of violent crime has four components: murder (including nonnegli-gent manslaughter); forcible rape; robbery (which must involve either force or a threat of force); and aggravated assault (which includes both assault with a weapon and assaults resulting in serious injury). Because robbery and aggravated assault are far more frequent than rape or murder, police and FBI estimates of violent crime depend largely on the number of robberies and aggravated assaults.
Figure 1 shows trends in the FBI's estimate of the number of robberies and aggravated assaults per 100,000 persons in the population. Both crimes appear to have increased dramatically between 1960 and 1980. Both declined slightly in the early 1980s. But aggravated assault began to rise again after 1983, pushing the overall index of violent crime to an all-time high in 1988. (Preliminary data for 1989 show another small rise.)
Taken at face value, these figures confirm the popular view that crime is on the rise. Before accepting this conclusion, however, we must ask what the numbers really measure. The FBI does not estimate the number of crimes committed in a given year. It estimates the number of crimes "known to the police." If the police do not see a robbery or assault taking place, and if no one reports it to the police, it will not end up in the FBI's records. Likewise, if someone complains to the police, but the police do not record the complaint, do not get around to investigating it, or conclude that it was legally unfounded, the incident will not be reported to the FBI.
The proportion of all robberies and assaults recorded by the police was traditionally quite low. In 1973, when the Census Bureau conducted its first National Crime Survey (NCS), citizens' reports plus a survey of business establishments suggested that there had been about 1.4 million robberies during the year, which was almost four times the number of robberies recorded by the police. NCS respondents also reported 1.7 million aggravated assaults, again about four times the number recorded by local police departments.
If the police had collected and recorded crime statistics the same way year after year, it might be reasonable to assume that their records included about the same fraction of all crimes today as in 1973. The true crime rate would then be much higher than Figure 1 implies, but Figure 1 would still provide an accurate picture of trends over time.
In reality, however, the police are constantly changing the way they record crimes. The Census Bureau began conducting the NCS in 1973 because both criminologists and the police believed that a lot of crime was going unreported. The first NCS confirmed these suspicions. Partly as a result, the Justice Department embarked on a major effort to help local police departments improve their record-keeping. Rules for recording crimes were formalized, records were computerized, police departments spent more on record-keeping, and officers on the beat began spending more time on paperwork. These changes have not had much effect on the proportion of citizens reporting crimes to the police, but they have increased the proportion of citizen complaints recorded by the police and reported to the FBI.
What the Victimization Surveys Show
Victimization surveys also underestimate the amount of violence in America. To begin with, survey respondents either forget or choose not to report a lot of the violence they experience. NCS respondents report very little domestic violence, for example. Even when violence involves strangers, some people repress the memory fairly quickly. (NCS asks people whether they have been robbed or assaulted at any time in the past six months. Respondents report substantially more incidents in the two or three months just before the survey than in earlier months.) Lapses of memory appear to be especially common among people for whom violence is an everyday occurrence and among people who have not finished high school. As a result, the NCS is especially likely to underestimate victimization rates for such people.
In addition, the NCS undersamples certain groups with very high rates of victimization. It does not even try to interview the homeless, for example, and it misses a large fraction of individuals who are seldom at home. (The more time you spend away from home, the more likely you are to be robbed or assaulted.)
Nonetheless, the NCS is conducted in the same way every year, so most of these biases are likely to be constant. So while the NCS almost certainly underestimates the level of violence in America, it should be a quite reliable guide to trends in violence.
Figure 2 shows NCS estimates of trends in robberies and aggravated assaults per 100,000 persons. There is no clear trend in either kind of violence from 1973 to 1981. After 1981 both robbery and aggravated assault began to decline, although aggravated assault increased again in 1988.
The NCS data therefore tells a very different story from the FBI data in Figure 1. According to the NCS, the aggravated assault rate fell 10 percent between 1973 and 1988, while the robbery rate fell 19 percent. According to the FBI, the aggravated assault rate rose 85 percent during this period, while the robbery rate rose 21 percent.
The same pattern holds for forcible rape. According to the NCS, rape has followed almost the same trajectory as aggravated assault, holding steady during the 1970s and declining in the 1980s. In 1973 citizens reported about three times as many rapes to the NCS as the police reported to the FBI. By 1988 citizens reported only 37 percent more rapes to the NCS than the police reported to the FBI. As a result, the FBI recorded a 53 percent increase in the incidence of rape between 1973 and 1988, while the NCS recorded a 30 percent decline.
Police estimates of trends in crime could differ from NCS estimates because victims are reporting a smaller proportion of violent crime to the NCS or because they are reporting a larger proportion to the police. Neither of these theories, however, is very plausible. The response rate in the NCS remains extraordinarily high compared to other surveys, and there is no evidence that respondents have become less cooperative. Likewise, the proportion of NCS respondents who say they reported crimes to the police has hardly changed since the Census began conducting the survey. (Victims said they reported 57 percent of all aggravated assaults and robberies to the police in 1987-88, compared to 53 percent in 1973-74.)
The most likely explanation for the increase in robberies and aggravated assaults reported to the FBI is, therefore, that the police are recording more of the violence that citizens report to them. In 1973, for example, citizens said they reported about 861,000 aggravated assaults to the police. The police recorded only 421,000. By 1988 citizens said they reported 940,000 aggravated assaults to the police, and the police recorded 910,000. The same pattern recurs for robbery and rape.
What about Murder Rates?
Murder is easy to define, hard to conceal, and sufficiently serious to ensure that the police rarely ignore it. As a result, almost all experts agree that murder statistics are more complete than statistics on other crimes. When a husband assaults his wife with a kitchen knife, for example, neither she nor anyone else is likely to report this fact either to the police or to a victimization survey. Even when the wife does report the assault, the police have traditionally been reluctant to record it unless she was willing to press charges. If a husband kills his wife, however, her death will almost always be discovered and the fact that she was murdered will usually be obvious, even if her husband's guilt is not. The same logic often applies to violence between strangers. If two teenage gangs get in a fight, they will usually try to conceal it from the police. If one teenager is killed, the others may try to conceal his death, but they seldom succeed.
Not only is murder better recorded than other crimes, but there is less reason to think that the police have changed the way in which they record murders. Not all murders are detected, of course. But murderers have always had the same incentive to conceal a victim's body, and there have been no major technological breakthroughs that make this easier today than in the past. Most experts therefore regard the murder rate as a society's most reliable index of long-term trends in violence. Accordingly, we must ask whether trends in murders "known to the police" have followed the same upward trajectory as trends in other crimes known to the police, or whether trends in murder have edged downward, as victims' reports of robbery and aggravated assault have since 1973.
Figure 1 suggests that police estimates of murder have followed a very different trajectory from their estimates of other violent crimes. The FBI recorded 17 aggravated assaults for every murder in 1960,21 in 1970, 29 in 1980, and 43 in 1987. The same pattern holds in somewhat less dramatic form for robbery. We can put this same point slightly differently by saying that while the FBI's estimate of the aggravated assault rate quadrupled between 1960 and 1988 and its estimate of the robbery rate more than tripled, its estimate of the murder rate rose "only" 63 percent. That is a lot, but far less than the FBI assault data suggest.
Could these differences between the trend in murder and the trend in other crimes of violence be real? When violence declines, we might expect lethal violence to decline even faster than non-lethal violence. But when violence is increasing, as it did between 1963 and 1974, why should non-lethal violence rise faster than lethal violence? One possible answer is that hospitals have become increasingly adept at saving the lives of people who arrive in the emergency room with gunshot or stab wounds. But those who engage in violence have also been using guns with more firepower, lowering their victims' chances of survival. The survival rate among assault victims may nonetheless have risen. But survival rates cannot have risen enough to explain the entire discrepancy between trends in aggravated assault and trends in murder. And improvements in medical technology can hardly explain the rising ratio of robberies and rapes to murders in police and FBI data.
Figure 2 shows that murder rates derived from police records and victimization rates for aggravated assault and robbery derived from the NCS all declined by roughly the same percentage from 1973 through 1988. This suggests that the ratio of murders to aggravated assaults and robberies was relatively stable after 1973. If it was also stable prior to 1973, FBI estimates of murder should constitute a good indicator of trends in the overall level of violence prior to 1973.
The murder rate roughly doubled between 1963 and 1974. I infer, therefore, that the overall level of violence also doubled during this period. Since 1973, both murder statistics and the victimization surveys show some decline in violence, though today's levels are still far above those that prevailed from 1950 to 1965. (Figure 2 does not include 1989. Preliminary FBI estimates suggest that the 1989 murder rate was about 8.6 per 100,000, compared to 9.4 in 1973 and 5.1 in 1960.)
Murder in Big Cities
The public's conviction that violence is rising derives primarily from news reports about what is happening in big cities. Here again, however, the story depends on which years and which cities you compare.
In 1980 America had six cities with more than a million residents: Chicago, Detroit, Houston, Los Angeles, New York, and Philadelphia. Between 1980 and 1989 the murder rate rose 32 percent in Detroit, 13 percent in New York, and 11 percent in Philadelphia. It fell 13 percent in Chicago, 27 percent in Los Angeles, and 32 percent in Houston. Averaging across all six cities, the murder rate per 100,000 residents fell from 29.2 to 28.1.
The story was much the same for the five cities with 750,000 to 1,000,000 inhabitants in 1980 (Baltimore, Dallas, Phoenix, San Antonio, and San Diego). The only big city with a really large increase in murder between 1980 and 1989 was Washington, DC, where the murder rate more than doubled.
We could, of course, tell a very different story if we picked a different baseline year for describing how America's big cities have changed. If we use 1960 or 1965 as our baseline year, almost every city shows a big increase in murder. If we use 1970 or 1975, there is not much change. If we focus on trends for the past year or two, as news reports usually do, we find significant increases in most cities. If the level of violence were increasing steadily over time, trends would look the same no matter what year we chose as a baseline. In reality, however, violence fluctuates a lot in relatively short periods. As a result, the recent upturn does not prove any clear long-term trend.
Race and Crime
Almost everyone is convinced that crime has increased more in poor black neighborhoods than elsewhere in America. Indeed, this assumption is so widespread that some black leaders mutter darkly about a white conspiracy that encourages blacks to murder one another. Some even use terms like "genocide."
Table 1 shows trends in murder rates by race from 1950 through 1987. Looking first at trends from 1950 through 1985 (before the crack epidemic hit), three broad conclusions stand out:
- Blacks are far more likely than whites to die violently. This has been true for as long as we have records.
- Blacks' chances of being murdered show no consistent increase over time. The black murder rate declined significantly during the 1950s, shot up during the 1960s, remained essentially constant during the 1970s, and fell in the early 1980s. As a result, a black man or woman's chances of being murdered were about the same in 1985 as in 1950.
- The white murder rate, while still much lower than than the black rate, has clearly risen over the past forty years. Whites were about twice as likely to be murdered in 1985 as in 1950. The gap between a black man or woman's chances of dying violently and a white man or woman's chances of dying violently has narrowed since 1950.
The crack epidemic became national news in 1986. According to both the police and the news media, crack unleashed an unprecedented wave of violence in poor black neighborhoods. Table 1 suggests, however, that crack's effect on the black community was quite limited in 1986 and 1987. The murder rate for black men between the ages of 15 and 24 rose dramatically between 1985 and 1987, presumably because selling crack is an adolescent profession. But older black men's chances of being murdered did not change appreciably, and the murder rate actually fell slightly among white men, even though most Hispanics are classified as white.
The National Center for Health Statistics, which collects the data in Table 1, has not yet released figures for 1988 or 1989. Judging by FBI data and police reports from big cities, the 1985-87 trends are likely to have continued, with murder rising among 15- to 24-year-old blacks but not changing much among other groups.
About 90 percent of those arrested for murder are of the same race as their victim. This figure has declined slightly since 1976, when the FBI first reported it, but the change has not been large. Trends in the race of murder victims should, therefore, provide a quite reliable guide to trends in the race of those who commit murders.
* * *
When we turn to robbery and aggravated assault, we can use victims' reports to determine the race of assailants. Table 2 shows the proportion of adults who said they had been victims of an aggravated assault by a black man or woman from 1973 through 1987. The number of people reporting an aggravated assault by a black fell by a quarter during this period; the number reporting they were robbed by a black fell 30 percent. Assaults and robberies by whites also declined, but less than those by blacks.
Murder statistics and victimization surveys, therefore, tell a consistent story. Black violence has not increased since the early 1970s; if anything, it has declined. There is still far more violence among blacks than among whites, but the gap is narrower than it used to be.
The Effects of the Baby Boom
Arrest data suggest that men between the ages of 15 and 24 are about three times as likely as older men to commit violent crimes. This means that when the proportion of men between the ages of 15 and 24 increases, as it did from 1964 to 1974, violent crime is likely to increase. When the proportion of men between 15 and 24 declines, as it did in the 1980s, crime is likely to decline.
But if age-specific crime rates had remained constant, neither the baby boom of the 1950s nor the ensuing baby bust would have had a very large effect on the level of violence. Men over the age of 15 commit most violent crimes. The proportion of such men who were under 24 rose from 20 percent in 1960 to 25 percent in 1975 and then fell back to 20 percent in 1988. If individuals of different ages had gone on committing crimes at a constant rate, such fluctuations in the percentage of 15 to 24 year olds would have raised the level of violent crime by about 8 percent between 1960 and 1975, and would then have lowered it about 7 percent between 1975 and 1988. Since violence roughly doubled between 1960 and 1975 and dropped by about 15 percent between 1975 and 1988, age-specific crime rates must have risen a lot between 1960 and 1975 and must have fallen a little after 1975.
Demographic change may also have influenced age-specific crime rates, however. When the number of young men rises, the balance of power shifts away from those who support authority toward those who resist it. During the 1960s and early 1970s the young sometimes seemed to feel they owned the streets, indeed, the whole society. This feeling may not only have made young revolutionaries more assertive but may also have made young hoodlums more aggressive. The graying of the baby boomers after 1980 may, in turn, have helped delegitimate violence.
Has "Getting Tough" Made a Difference?
When the crime rate began to climb in the mid-1960s, conservatives urged a "get tough" policy that emphasized swift, severe punishment. Demands for swifter trials and sentencing were defeated by the "due process" clause of the Constitution. If anything, justice moves even slower today than it did twenty years ago. Conservative demands for more severe punishment did eventually have an effect, however.
Reliable statistics on the risk of punishment for any given offense are hard to come by, but certain broad trends are fairly clear. Murder rates suggest that the number of violent crimes committed by adults roughly doubled between 1960 and 1975. Yet the fraction of the adult population in federal and state prisons fell slightly during these years. Thus if violent offenders accounted for a constant fraction of the prison population, they must have been serving only about half as much time per offense in 1975 as in 1960. It is not easy to determine how much of this change to blame on declining arrest rates, how much to blame on declining conviction rates among those arrested, and how much to blame on the fact that those convicted spent less time behind bars. Reducing the expected cost of violence must, however, have had some effect on the frequency of violence.
After 1975, the expected cost of violence began to rise. First, while the police continued to make arrests for about half the violent offenses they recorded, arrests rose considerably faster than victimization rates. Thus, if victimization rates are our best indicator of the underlying trend in violence, the percentage of violent offenders getting arrested must have risen. At the same time, those who went to prison were staying longer. The net effect of these changes was that violent offenders could expect to spend more time in prison. Judging by murder and victimization rates, the violent crime rate was about 10 percent lower in 1988 than in 1975. Yet the fraction of adults in state and federal prisons more than doubled during this period. In part, this was because we were locking up more people for drug-related offenses. But those who committed violent crimes could also expect to spend considerably more time in prison in 1988 than in 1975.
Locking up criminals can affect the crime rate in at least three ways:
- Imprisonment reduces the number of crimes prisoners commit while they are behind bars. A panel convened by the National Academy of Sciences in the mid-1980s estimated that if this were the only effect of incarceration, doubling the prison population would reduce crime by about 10 percent.
- Imprisonment is also likely to increase the number of crimes prisoners commit after being released, both because it exposes them to a lot more violence while they are in prison and because a prison record makes it harder to integrate oneself into society after release.
- Locking up convicts deters others from engaging in violence. The magnitude of this effect is controversial. Most conservatives think deterrence is very effective. Most liberals think it ineffective. Criminologists have not been able to resolve this controversy.
The overall effect of changes in imprisonment is uncertain. But changes in the expected costs of violence could have played a significant role in altering age-specific crime rates.
Crime as a Political Issue
The feeling that crime is out of control influences Americans' views on issues as diverse as defendants' rights, capital punishment, drugs, school desegregation, and housing policy. The feeling rests on our belief that violence has risen steadily for the past quarter century, despite the best efforts of both liberals and conservatives to control it. I have tried to show that crime has not, in fact, risen steadily. But this does not mean the public is wrong in thinking that neither liberals nor conservatives know how to reduce it.
The stock liberal response to crime has been that we must provide disadvantaged youths with better job opportunities. Such a policy would have many benefits, but reducing violence is not likely to be one of them. Between 1964 and 1969, for example, we cut male unemployment from 4.6 to 2.8 percent, and raised real wages 1 percent a year. Violent crime rose about 50 percent during these years. Between 1979 and 1983, in contrast, we raised male unemployment from 5.1 to 9.9 percent and lowered real wages by almost 2 percent a year. Yet violent crime fell by about a sixth. Thus while there may still be some relationship between economic opportunities and violence, it is certainly not very strong.
* * *
The stock conservative response to crime is punishment. Punishment can work wonders when it is both swift and certain. (Very few people rape, rob, or murder someone else if they know a policeman is watching.) But a country as big, chaotic, and rights-oriented as the United States will never manage to punish more than a small fraction of those who engage in violence. In practice, therefore, the conservative strategy boils down to locking up a small minority of violent offenders for longer periods. Our experience since 1975 suggests that this approach may reduce violence a little but that its effects are quite modest.
One thing liberals and conservatives agree on is that the criminal justice system has broken down in a number of big cities. The number of crimes has risen faster than the number of police available to investigate them. The number of arrests has risen faster than the number of prosecuting attorneys and judges needed to deal with them (partly because more defendants now have attorneys whose job is to make each case take more time). Convictions have also risen faster than the number of prison cells we would need to incarcerate those the public wants locked up. This last development might be a blessing in disguise if it had led us to spend more money experimenting with alternatives to incarceration, but we have only done a little of that, so the actual consequence is that a lot of people are released on probation who should not be.
The breakdown of the criminal justice system in big cities is not new. It was already apparent in the late 1960s and early 1970s. But its effects are cumulative. When the criminal justice system is unable to punish any but the worst offenders, ordinary citizens see no reason to risk retribution by cooperating with the police or prosecutors. The longer this situation persists, the more citizens have first-hand experience of the system's paralysis and the more widespread cynicism becomes. Furthermore, when punishment is slow and erratic, its deterrent effect diminishes, both because fear of imprisonment depends more on speed and certainty than on the length of sentences and because failure to punish a crime signals tacit moral acceptance. If we do not bother to punish those who commit a crime in some way that criminals regard as costly, they can hardly be expected to feel they have violated a strong social taboo.
While politicians have not done much to reduce crime, they sometimes win elections by claiming they will do something. Both Democrats and Republicans love to run "Willie Horton" campaigns, in which they fault their opponent for being "soft" on crime. Since these claims are all hot air, an election that focuses on crime is an election wasted. But the only way to prevent such waste, I fear, is a constitutional amendment forbidding any reference to violent crime in political campaigns.
In the absence of such an amendment, we should obviously try to raise the level of public debate about crime. No one who has followed the debates of the past quarter century is likely to feel much optimism about our chances, however. Few subjects inspire as much nonsense as violent crime.
Violence is a product of many conflicting influences, all of which are constantly changing. In most periods of history these changes roughly offset one another. If the proportion of the population between the ages of 15 and 24 falls, for example, the benefits are likely to be offset by some other change, such as the opening of new drug markets that everyone wants to control. Occasionally, however, a lot of the factors that influence violence all change in the same direction at roughly the same time. When this happens, the level of violence can change substantially.
Between 1965 and 1975, for example, the number of young men increased, young black Americans became increasingly disillusioned and angry about the way American society treated them, and the Vietnam War schooled a whole generation of young men in killing. At the same time, respect for authority eroded, those who had traditionally exercised authority became more cautious about using it, and the chances of being punished if you broke the law declined. With all these forces pushing in the same direction, it is hardly surprising that violence became more common.
Unfortunately, the explanations people proposed for the increase in crime during these years consisted largely of identifying one of the factors listed above and claiming that it was "the" explanation. Skeptics then responded by showing that this particular explanation could not possibly account for the entire observed change. Soberer folk advocated multi-factor explanations, but they got little attention. Multi-factor explanations are complicated and boring. Worse yet, they seldom serve the political needs of reformers. Those who want to lock people up, for example, do not want to acknowledge that locking people up will, at best, have only a modest effect on violence. Those who want the government to provide young people with jobs are equally reluctant to admit that this will have only a modest effect on street crime.
* * *
Instead of trying to understand the causes of crime, therefore, we should probably concentrate on assessing the proposed cures. We know, for example, that an extraordinarily high proportion of those on death row were physically abused as children. We also know that our current system for dealing with abused children is a disaster. Because adults worry more about the rights of parents than the rights of children, they are very slow to intervene even when they suspect abuse. Once we do intervene, our concern for parental rights forces us to shunt abused children from one temporary foster home to another rather than placing them immediately in a permanent adoptive home. This prospect makes social workers even more reluctant to intervene.
Our prisons are also a national scandal. We could make them less brutal by spending more money, and we could probably teach more inmates marketable skills and better ways of dealing with their psychological problems. Such changes would probably reduce recidivism a bit. But even if we had the will to change all this, the benefits would be slow in coming and would never turn Boston into Toronto or San Francisco into Vancouver. Even so, reinvigorating the criminal justice system, making prisons work better, and intervening more decisively to protect children from abusive parents are all worthwhile in their own right.