The last time unemployment rates were this high, in the Great Depression, John Dillinger was making withdrawals from banks along with Ma Barker and Bonnie and Clyde.

It might seem that in the current recession, as waves of "marginal" workers and others lose their jobs, more crime is inevitable. But an analysis of crime data and unemployment rates for the last 20 years shows no strong relationship.

Scholars disagree strongly on the subject. Some say recessions mean escalating crime; others argue that there is no evidence to link rising unemployment with rising crime. One problem is that crime statistics are notoriously unreliable, partly because only a fraction of crimes committed are reported to authorities.

The 1982 recession is a kind of laboratory, testing these theories in the real world of crime.

Looking at nationwide crime statistics in seven categories--auto theft, larceny theft, burglary, robbery, property crime, violent crime and overall crimes committed--and using various statistical methods, The Washington Post compared the incidence of crime in the seven categories with nationwide unemployment rates from 1960 through 1980.

Sometimes there seemed to be a connection. In the recession year of 1975, when unemployment reached 8.5 percent, the total crime index reached a peak, with historic highs in violent crime, robbery, burglary and auto theft.

Some of those peaks were reached again in 1979 and 1980 as the crime rate crept up again after dropping when the 1975 recession ended. In general, both the crime and unemployment rates have tended to rise since 1970.

On the other hand, 1961 was also a year of high unemployment, although by today's standards it seemed tame at 6.7 percent. Yet in that year there were drops in violent crime, murder, rape, robbery and aggravated assault. Furthermore, while the unemployment rate declined from 1961 to 1969, the crime rate rose substantially.

The analyses show only a slight relationship between crime and unemployment. When unemployment rises it is likely that most types of crime will rise only a little more than they would otherwise.

The question of the unemployment-crime relationship, if any, has consumed much ink and computer time in the academic world. M. Harvey Brenner, a sociologist at Johns Hopkins University in Baltimore, has published controversial findings that there is a link between unemployment and many crimes, particularly murder.

Brenner says he believes a 1 percent increase in the unemployment rate is associated with a 5.7 percent rise in murders and a 4 percent increase in the number of people in state prisons. He also reports that each 1 percent rise in unemployment is tied to a 4.1 percent increase in suicide, a 1.9 percent rise in the overall mortality rate, and increases of 4.3 percent among men and 2.3 percent among women of people newly admitted to mental hospitals.

While Brenner stands by these figures six years after he presented them to a congressional committee, he cautions that a recession does not automatically mean more crime. He believes it is safest to say that all things being equal, more unemployment will result in higher crime rates.

Brenner says the most important aspect of unemployment is joblessness among males aged 15 to 24. If the ratio of unemployed young males to all unemployed people rises, then he said increases in crime are particularly likely. Many types of crimes, even rape, are affected, he said.

A principal reason for this relationship, according to Brenner, is stress. When people are out of work, they feel tension. They may also feel victimized, and there may be economic reasons for committing thefts or burglaries, so some of their number turn to crime.

He emphasizes that there are many other causes of crime: increasing use of alcohol and narcotics, urban growth, development of an underclass and demographic and family changes.

Brenner said that in the Great Depression, the rates of homicide and imprisonment reached peaks that were not reached again until very recently.

Other scholars have roundly criticized Brenner's work and have reached different conclusions from the same data. The statistical analysis is extremely complex because the data are not always reliable and comparable, and because there are many factors to be taken into account.

Frederick C. Nold, president of a research company in Palo Alto, Calif., worked with some of Brenner's data and found different results. In particular, Nold said the most important factor is the proportion of young males in the unemployment figures. The overall rate of unemployment is much less important, Nold said.

"It doesn't demonstrate much of a relationship between crime and overall unemployment," he said.

Michael K. Block, a University of Arizona economist, said the evidence is so mixed that no conclusions can easily be drawn. "Undoubtedly there's some effect. Getting at it is very difficult," he said.

Harvard professor James Q. Wilson, an authority on crime, criticized Brenner's work, and said, "Overall unemployment seems to bear little or no relationship to crime."

Wilson noted that there are two ways to examine the relationship of unemployment and crime. Researchers can look at crime and unemployment statistics for a number of years and compare them. Or they can take a cross-sectional approach, getting areas around the country with high and low unemployment rates and checking whether high unemployment areas have more crime.

When other factors are controlled, Wilson said, neither approach demonstrates a significant relationship.

Reliable national data are not available for the Depression years, Wilson said, adding that a review he conducted of local studies on unemployment and crime found no overall relationship.

"There's no basis for a prediction that a deepening or continuing of the recession will lead to increases in the crime rate," Wilson said.

Staff writer Kenneth E. John contributed to this report.