The age crime curve is both a widely agreed upon and rigorously debated facet of criminology. 

A breadth of studies leveraging criminal data across continents and centuries, analyzing offense rates across different age groups, find a sharp spike of criminality in the teen years, peaking in the late teens and typically dropping off in the early 20s. That the curve exists, its general shape, and that it repeatedly demonstrates itself in data is widely agreed on; however, unpacking what the age crime curve means is a subject of significant debate. As statisticians and criminologists have gained access to increasingly better data, these new data have edified prior understanding while also creating new lines of inquiry. 

The Age and Crime Connection

Some criminologists (Hirsch 1983) have argued that the age crime curve is a persistent part of modern humanity, pointing to the multitude of curves demonstrated across various historic periods and societies, all bearing the same general shape. This consistency frustrates many hypotheses as to why this relationship exists, particularly when theories focus on specific societal pressures faced by teenagers of a given time and place.

Figure 1 shows Farrington’s (1986) age crime curve of British males from three separate years, using data from the U.K. Home Office: 

Line graph showing crimes per 100 population of British males, by age, from 1938, 1961, and 1983
Figure 1

Figure 2 shows Hirschi and Gott’s (1983) analysis of the age crime curve of criminal offenders in England and Wales in 1842-1844, separated for male and female offenders.

Line graph of age crime curve of criminal offenders in England and Wales in 1842-1844, separated for male and female offenders
Figure 2

Data showing the curve is often assembled in different ways, so while the general idea presented by the curve remains the same, what constitutes individual curves varies. In some cases, data has been assembled from citations, arrests, or convictions. Critics point out that convictions can occur as either the culmination of a long run of offending or at a significant delay from the date/age of the actual offense, thereby skewing data points later than the actual offending. More recently, researchers have used data assembled from self-reporting surveys, instead of relying on police and government-published statistics. Proponents argue these data collection methods create a more accurate and timely depiction of criminality (McVie 2004). 

Age, Frequency, and Criminal Offense in the Age Crime Curve

A frequent question provoked by the curve is whether the shape is driven by a high proportion of teenagers choosing to offend relative to other age groups, or whether there is a similar proportion of teenagers committing crimes relative to other age groups, just at a much higher frequency.

Analysis by Farrington (1986) and expanded upon by Nagin and Land (1993) suggests the former, that the peak in the age crime curve is driven mostly by an increase in the number of offenders. Nagin and Land, however, did find evidence of individual offenders committing crimes at a higher frequency as well.

Researchers have also attempted to understand the differences between generalist offenders, who will commit any or many crimes, and how it relates to the age crime curve. Their research has shown that specialization appears to develop over time, where offenders who continue committing crimes appear to specialize in a subset of criminal offenses.

Within the age crime curve, peak ages for specific offenses vary widely as well. Across many studies, researchers noted that crimes against property have an earlier median age relative to crimes against persons. Cline (1980) used FBI figures from 1977 to calculate median ages of arrest and found the lowest median ages were for vandalism, motor vehicle theft, arson, burglary, larceny-theft, and liquor law violations. He described these as “the offenses of adolescence.” But Hirsch et al argued that the clear age divisions found in official data were not so clear in the analysis of self-reporting surveys, meaning that the median offending ages observed by official data might instead be driven by the scale and severity of consequences for these offenses as perpetrators age.

Figure 3 shows contemporary American data, using NIBRS offender data and US census data for property crime and violent crime.

Line graph showing contemporary American crime rates by age range
Figure 3

The above figure matches the sharp increase in crime as described by prior research but also suggests a somewhat more gradual reduction in offense rates through the first half of adult life. 

Criminality and Age Peak Performance

The age crime curve is striking, particularly in the ways it tends to tell a similar story about crime across different societies in different eras. Generation after generation of young people experience a blossoming of criminality, alarming parents, teachers, school administrators, policymakers, and the population at large. After this brief spike in crime, most of the cohort will cease to offend, leaving a relatively small share who will grow into chronic offenders of varying capacity, some of whom will become highly specialized career criminals. These data points, and the fevered debate to best understand and respond to them, shed further light on the susceptibility of young people and the meaningful impact that programs explicitly targeting teenage youth can affect both individual outcomes and society at large. 

Published July 12, 2022

Sources: 

Cline, Hugh F. “Criminal Behavior over the Life Span.” Constancy and Change in Human Development, Cambridge, Mass.: Harvard University Press. 1980. 

Farrington, David P. “Age and Crime.” Crime and Justice. Vol 7. 1986. 

FBI Crime Data Explorer: https://crime-data-explorer.app.cloud.gov/pages/downloads 

Hirschi, Travis; Gottfredson, Michael. “Age and the Explanation of Crime.” The American Journal of Sociology. Vol. 89, no. 3. 1983. 

McVie, Susan. “Patterns of Deviance Underlying the Age-Crime Curve: The Long Term Evidence” 2004. 

Nagin, Daniel S; Land, Kenneth C. “Age, Criminal Careers, and Population Heterogeneity: Specification and Estimation of a Nonparametric, Mixed Poisson Model” Criminology. Vol. 31, no. 3. 1993. 

U.S. Census Data: https://data.census.gov/cedsci/table?q=United%20States&t=Age%20and%20Sex&tid=ACSST5Y2020.S0101