Data on Arrest Records Aren't Always by the Book
Old Reports Suggesting One-Third to One-Half of All Men Are Apprehended Had Flaws, but New Studies Confirm a High Rate
Researchers who announced the stunning arrest rates in 1967 were stumped by data deficiencies, such as their inability to tell whether the same person was being counted more than once -- an often overlooked point the researchers made in their own report. Today, data problems in crime measurement persist. Reporting by local law-enforcement agencies is incomplete, and criminologists say local data aren't calculated in a uniform way across the U.S.
To mitigate gaps and inconsistencies in the numbers, the Federal Bureau of Investigation extrapolates from the numbers it does have to get a nationwide total.
That 52% of American men will be arrested originated in a report from the President's Commission on Law Enforcement and Administration of Justice. Ronald Christensen, then a graduate student in physics who held degrees in electrical engineering and law, was asked to crunch the numbers on arrests and convictions. He set out to calculate how many men and women at each age had been arrested, counting juvenile arrests. Mr. Christensen assumed that current arrest rates would hold.
The most difficult assumption concerned first-time arrests, Mr. Christensen says.
Mr. Christensen drew upon files from San Mateo, Calif., Washington, D.C., New York State and the Federal Bureau of Investigation, each one suggesting that between one-quarter and one-third of all arrests were what he called "virgin arrests." But each source had its problems. For instance, San Mateo is hardly representative of the country, and its files lacked juvenile arrests.
So Mr. Christensen says he went with a "conservative" estimate, that about one-eighth of all arrests are first-time bookings based on extrapolations from Philadelphia juvenile records. That figure helped him to estimate an individual's chances of being arrested over an entire lifetime, because each first-time arrest adds to the overall count of people who have been arrested.
Still, the resulting 52% estimate for men (four times the probability for women) surprised Mr. Christensen's co-authors, including Alfred Blumstein, then director of the commission and now a criminologist at Carnegie Mellon University.
Mr. Christensen, now president of the statistical consultancy Entropy Ltd. in Lincoln, Mass., recalls that Mr. Blumstein at the time said, "That can't be right. Go back and check your numbers." But he did, and they checked out, Mr. Christensen recalls. Mr. Blumstein recalls that he was surprised by the finding.
But these numbers come with some big caveats. First, averaging out the arrest rate for men and women yields a much lower proportion -- one person out of every three -- who will be arrested in his or her lifetime. Separately, much depends upon a big assumption that, if wrong, would produce an outsize error in the 52% arrest statistic. If the virgin-arrest rate were inflated by 10%, the paper itself states, that would produce a 10% error in the resulting arrest estimate.
Also, the number is easily misunderstood. The proportion of American men who had been arrested was below 52%, because some hadn't yet been arrested but would be in their lifetimes. In 1965, the proportion of American men who already had been arrested was closer to 40%, according to Mr. Christensen. Another reason researchers arrived at the high 52% figure was that the number includes people arrested as juveniles but never as adults. Juvenile arrests are often considered a different category because many states seal those records from employers.
Taking those factors into account, the proportion of men who had been arrested as adults might have been as low as 30% in 1965, Prof. Blumstein says.
Even that arrest rate might not be the same today. There have been many changes in U.S. crime and law enforcement in the intervening four decades. And there haven't been many efforts to update the study's findings. A 1978 report commissioned by the Labor Department found that one-fourth to one-third of adults in the labor force had an arrest record. That was consistent with the 1965 report, but the author of the 1978 study cautioned that the source data were problematic. For instance, the analysis relied partly on data that might not have been rigorously calculated, the report said.
Similar results to the 52% arrest rate emerged from a 1987 study by sociologist Robert Tillman, who found that one-third of California men born in 1956 were arrested between the ages of 18 and 29. Prof. Tillman, now a professor at St. John's University in Queens, N.Y., says he was inspired to study the question because he read the 1965 estimate and thought, "Oh, you can use this data to get a better estimate." He acknowledges that his study had limitations, because it included those born outside California who committed crimes there, while excluding Californian natives who had arrest records elsewhere. "I concluded it wasn't a huge problem for the estimates," says Prof. Tillman.
Several criminologists, including those who conducted the earlier studies, say they don't know of any new national estimates. Prof. Blumstein says the proportion of American men arrested may have risen to roughly 60% over a man's lifetime as drug arrests rose, though he calls this speculation. Other researchers point to sharply declining violent-crime rates in the last 15 years, and lower arrests per capita, to suggest that the lifetime arrest rate might be lower today than it was decades ago.
The 1967 report was "likely cited thousands of times in just about every intro textbook," Christopher Maxwell, associate professor of criminal justice at Michigan State University, said in an email. "Unfortunately, we haven't made much progress on closing the data holes."
If anything, data have gotten less reliable during the economic downturn, according to Robert Friedmann, professor of criminal justice at Georgia State University. "Budget cuts are creating horrendous gaps in the system," Prof. Friedmann says.
The FBI serves as a clearinghouse of sorts for local crime statistics. Law-enforcement agencies and employers -- with prospective employees' permission -- can access individuals' rap sheets (Records of Arrest and Prosecution) from the bureau. There are about 70 million nonduplicate rap sheets, according to FBI estimates. This might include some people who have died, but also is likely to exclude many juvenile arrests. Also, some states only include convictions in their reports to the FBI. Still, dividing 70 million by 263 million -- the number of Americans age 10 and older last year, according to the Census Bureau -- yields a rap sheet for 27% of American men and women. Add in juvenile arrests, and account for men's higher arrest rate, and men could have a lifetime chance of arrest of around 50%, consistent with the 1967 report.
That report's finding on arrest rates, while attention-grabbing, was ancillary to its main purpose of making law-and-order recommendations to President Johnson. One of those recommendations, partly adopted, was to limit the FBI's role as a central database of crime statistics, a reaction to overreaching by then-FBI director J. Edgar Hoover. Prof. Blumstein recognizes the irony that the report helped make it tougher to estimate arrest rates in the future. "There is always the trade-off between privacy and efficiency," he says.
Write to Carl Bialik at email@example.comPrinted in The Wall Street Journal, page A15