Lincoln Steffens, a close colleague of Jacob Riis and a brilliant journalist in his own right, wrote in his autobiography that it was a common practice for reporters around the turn of the century to create the impression of a dramatic increase in violent crime simply by reporting more stories about specific murders, muggings and rapes. He claims even to have once single-handedly created a crime wave in New York.

The better part of a century later, a number of usually thoughtful commentators are suggesting that the FBI is doing today essentially what Steffens had done earlier. The FBI has recently reported that violent crimes increased in the United States by 13 percent from 1979 to 1980 and that the number of violent crimes per 100,000 U.S. residents increased by almost 30 percent from 1977 to 1980. However, because the FBI statistics are derived from data about crimes reported to the police, they are subject to forces that are unrelated to the true crime rate, such as the willingness of victims to report crimes to the police and the potential for police departments to alter unofficialy their procedures for recording the data that they pass on to the FBI.

The critics of FBI statistics claim that a much better indicator of the true crime rate is the "victimization" survey data collected by the Census Bureau. The victimization series was instituted in 1972 to provide national estimates of the extent to which persons, households and businesses are victimized by robbery, rape, assault, larceny, burglary and auto theft. The series was designed largely to cast light on the "dark figure" of crime -- episodes not reported to the poice -- and thus provide a more reliable measure of change in the crime rate. The basic method of obtaining the victimization data consists of asking people whether they have victims of crime within the past six months. This survey is second only to the decennial census in terms of the number of persons interviewed. According to the victimization series, the violent crime rate showed an insignificant increase from 17.1 serious violent episodes (rapes, robberies and aggravated assaults) per 1,000 residents in 1977 to 17.3 in 1979, the most recent year for which statistics are available. A number of prominent authorities are using these statistics as evidence that the level of violent crime is actually stable.

Unfortunately, however, these data, like those of the FBI, have their limitations. Some people cannot recall precisely when a crime occurred -- they may "telescope" the event either foreward or backward in time. Others fail to mention the crime altogether. About 25 percent of crimes reported to the police have been found to be unmentioned in follow-up victimization interviews. Other crimes may be reported neither to the police nor in the victimization survey. Some people selected for the survey cannot be found; others refuse to respond. Moreover, the victimization survey ignores the most serious crime of all, homicide; obviously, homicide victims make poor survey respondents.

The homicide rate, in fact, serves as an alternative measure of violence that had distinct advantages over both the FBI violent crime series and the victimization series. The victim is nearly always found, so the crime is rerely unreported. The date of the event is nearly always accurate to within a day or two. Homicide is the extreme culmination of every type of violent crime -- robbery, rape and assault -- hence it serves as an index of all violent crime. It is, in fact, correlated with other crime. And it is derived from two totally separate sources: police statistics provided to the FBI and death certificate statistics provided to the FBI and death certificate statistics provided to the U.S. Public Health Service.

The agreement between the two homicide series produced by these disparate sources is remarkable by most conventional standards of reliability for social science statistics. The following graph displays the two series for the 30-year period 1950 through 1979:

Note that for the years 1977 through 1979 both series show an upward trend: an annual increase of 5 percent according to the FBI data and 7 percent according to the data from the Public Health Service. Preliminary estimates of homicides for 1980 by both agencies indicate a continuation of this rise in violence. After declining slightly in the mid-1970s, violent crime really does appear to have resumed an upward course, starting in 1977.

The debate over the current trend in violent crime is especially curious because it is so shortsighted. When we look back to less than 20 years, rather than only three or four, we find homicide rates that are not even half as large as the current ones. In 1962 the homicide rate stood at 4.5 per 100,000 residents, according to the FBI, and 4.8 according to to the Public Health Service. At this moment it appears to be well over 10 in both series.

More fundamentally, crime in the United States is too high by any reasonable standard. People are deeply concerned about violent crime, and their concern is supported by the numbers. Violent crime is an extremely serious problem, and it appears currently to be getting worse.