Convictions per 1,000 offenders

From 1981 to the latest year of conviction data (1994 in the United States, 1995 in England), the number of persons convicted of --

• murder per 1,000 alleged murderers has risen 43% in the United States (340 in 1981 rising to 487 in 1994) but fallen 12% in England (631 in 1981 falling to 555 in 1995) (figure 25)
• rape per 1,000 alleged rapists has risen 94% in the United States (97 in 1981 rising to 188 in 1994) but fallen 63% in England (272 in 1981 falling to 100 in 1995) (figure 26)
• robbery per 1,000 alleged robbers has risen 29% in the United States (17 in 1981 rising to 22 in 1994) but fallen 40% in England (10 in 1981 falling to 6 in 1995) (figure 27)
• assault per 1,000 alleged assaulters has nearly tripled in the United States (9 in 1981 rising to 25 in 1994) but fallen 66% in England (41 in 1981 falling to 14 in 1995) (figure 28)
• burglary per 1,000 alleged burglars has risen 40% in the United States (10 in 1981 rising to 14 in 1994) but fallen 78% in England (27 in 1981 falling to 6 in 1995) (figure 29)
• motor vehicle theft per 1,000 alleged vehicle thieves has more than doubled in the United States (7 in 1981 rising to 18 in 1994) but fallen 77% in England (53 in 1981 falling to 12 in 1995) (figure 30).

Are persons committing a crime unlikely to be caught and convicted in both countries?

In both countries, persons committing a crime are unlikely to be caught and convicted. The major exception is murder.

According to latest statistics (derived from figures 25-30 covering 1994 in the United States, 1995 in England), the likelihood of conviction was about --

• 50% for murder in both countries
• 20% for rape in the United States and 10% in England
• 2% for robbery in the United States and less than 1% in England
• 2% for assault in the United States and 1% in England
• 1% for burglary in the United States and less than 1% in England
• 2% for motor vehicle theft in the United States and 1% in England.

However, these estimates must be interpreted cautiously. For example, the rape conviction likelihood is inflated because it is based on the number of convictions divided not by the total number of rapes but by just the number recorded by police.

In other respects, all of these estimates (including rape) are conservative because, to be precise, the data used to calculate them estimate the likelihood of an offense leading to conviction, not the likelihood of an offender being convicted sooner or later. To estimate the likelihood of an offender being convicted, the number of different persons convicted is divided by the number of different persons committing the crime. But in available data, per- sons who are convicted more than once or who commit crime more than once during the year are counted over and over again.

Such overcounting is particularly prevalent in the data on the number of persons committing crime because the typical offender commits more than one crime over the course of a year (the typical violent offender commits from 2 to 4 violent crimes per year and the typical property offender commits from 5 to 10 per year according to Blumstein and others, 1986, page 4).

Some adjustment can be made for their overcounting. For example, if the typical robber in the United States commits 4 robberies per year, the likelihood that a person committing robbery will be convicted of robbery during the year is nearly 8%, not 2%.

At first glance that might seem like a small difference but a 1 in 12 chance of conviction (the equivalent of 8%) is considerably greater than a 1 in 50 chance (the equivalent of 2%). Furthermore it should be stressed that, despite the adjustment, the 8% is still conservative because it is the probability that a robber will be convicted just of robbery during the year. Obviously the likelihood that a robber will be convicted of robbery or some other offense during the year is greater than just the probability that he will be convicted of robbery. How much greater is not known but for two reasons it is probably substantial.

One reason has to do with the practice, common both in England and the United States, of downgrading offenses. Because of downgrading, when persons are convicted of a crime, the crime they are convicted of is often less serious than the one for which they were originally arrested and charged (in the United States, downgrading occurs in 40% of violent crime convictions and 25% of nonviolent convictions according to Reaves, 1998, tables 26 and 27).

The other reason is related to the fact that persons who commit a crime typically commit more than one a year. Importantly the crimes they commit typically differ from one another. For example, a man committing a robbery one day may commit a burglary the next. While he might not be caught for the robbery, he might be caught and convicted for the burglary. Consequently his chance of conviction is greater than just the chance he takes of being convicted of the robbery.

Over periods longer than a year, the probability of an offender being convicted can be quite high. For example, Farrington (1989, pages 339-423) found that 58% of English males who admitted to committing burglary were eventually caught and convicted of burglary at least once before reaching age 33.