Below, Robert Sampson shares five key insights from his new book, Marked by Time: How Social Change Has Transformed Crime and the Life Trajectories of Young Americans.
Robert is a sociology professor at Harvard University and an affiliated research professor at the American Bar Foundation. He is also a member of the National Academy of Sciences.
What’s the big idea?
Societal changes over time shape people’s life outcomes in powerful ways, challenging much of what our common wisdom predicts about individual propensities to commit crime.
1. Just a few years can make a difference.
Darnell Jackson, one of our study members from Chicago, was born in the mid-1990s, while Andre Lewis was born in the 1980s. Although from the same city, the same poverty level, the same family background, and the same neighborhood background, they held different tickets in the birth lottery of history. They came of age in different social worlds.
By the time they were twenty years old, people born in the 1980s, like Andre, were arrested at a rate more than double that of kids born in the mid-1990s, like Darnell. This difference shaped their futures, because becoming arrested is a snare that creates multiple obstacles to success in life. Take education, for instance. In our study, being arrested as a teenager, even if not incarcerated, is linked to a 25 percent lower likelihood of graduating from a four-year college.
There are also cohort differences in exposure to violence and early mortality. As crime declined from its heights in the 1990s, later cohorts encountered a less violent world in early adolescence. Importantly, they were less likely to witness gun violence, carry concealed weapons, and to later use guns when they became adults. There were even differences in the odds of being murdered by age 40 between those born in the early 1980s and the mid-80s.
Just a few years can produce dramatic shifts in being arrested and exposure to guns and violence—even within the same generation.
2. The usual suspects don’t explain cohort differences.
You might think these differences in arrest rates, violence, and gun use are due to demographic composition or early-life circumstances that varied across cohorts. But that’s not the case. Even after we account for commonly studied childhood and family differences, cohort gaps in life trajectories persisted.
In looking at arrest in adolescence into adulthood, I controlled for a range of factors including demographic differences like race and ethnicity; individual differences, such as self-control or grit; family factors, such as parental education, employment, income, welfare receipt, depression, and family criminal history; and neighborhood factors during childhood, such as unemployment, poverty, crime levels, and racial segregation. Yet after controlling for these early-life differences, cohort trajectories still diverged markedly.
“Even after we account for commonly studied childhood and family differences, cohort gaps in life trajectories persisted.”
Cohort differences in arrest rates are strongest for the most disadvantaged, particularly poor Black youth. For this group, arrest rates were more than two times higher in older cohorts compared to younger cohorts. Put differently, the disadvantage gap declined over time. Even among the poorest of the poor, things got much better, which is not something we might expect given the popular narrative that the world is getting worse.
If it does not come down to individual differences, race, family, or neighborhood factors in childhood, what are the mechanisms driving the birth lottery of history?
3. Institutional and behavioral changes drove cohort divergence.
One explanation is rapidly changing policing practices in adolescence and young adulthood. Much has been written about how the drug war led to mass incarceration. However, kids coming of age in recent cohorts were exposed to declining drug arrests. For example, drug arrests in Chicago decreased from the mid 1990s to 2021 by about 90 percent.
Second, what is widely known as “broken-windows policing” (where public disorder is targeted) also shifted dramatically over time. Disorder arrests fell by nearly 100 percent. That’s surprising given the prominence of broken windows theory in public discourse as a way to reduce crime—but crime and disorder arrests both fell.
Arrests for other crimes also declined, including violent and property crimes. But here, policing practices only played a small role. The decline in violent and property-crime arrests in Chicago was driven primarily by decreases in offending behavior, not changes in police practices.
What drove these behavioral changes in crime? I argue that a constellation of factors is responsible:
- Urban revitalization occurred throughout America on a broad scale. Relatedly, increases in the number of first-generation Americans through immigration helped this revitalization, in turn helping bring down crime rates starting in the 1990s.
- Increases in community-based organizations, especially for young people, reduced dangerous and unregulated environments.
- Increasing technological surveillance in public, and tracking by parents, likely decreased opportunities for violence. Younger cohorts also had less unstructured time.
- Toxic exposure to lead plummeted, with the largest absolute improvements experienced within poor Black neighborhoods. Lead exposure is a known correlate of delinquency.
4. Social change degrades our ability to predict.
Risk assessment instruments, especially those relying on prior criminal records, are widely used in the criminal justice system. The problem is they’re based on ideas of inherent criminal propensity and of consistent predictive power across time.
Risk-assessment instruments trained on the older cohort overpredict the arrest probability of the younger cohort by nearly 90 percent. That is a huge overprediction, and it is due to changing crime and the changing relationship of risk factors with arrest.
History is baked into criminal records, leading to what we call cohort bias. It’s the systematic difference between the younger cohort’s actual arrest patterns and their predicted arrest patterns based on a risk assessment instrument trained on older cohort data. Cohort bias exists across all racial groups. This finding means cohort bias represents a potential source of inequality separate from racial or ethnic bias.
“History is baked into criminal records, leading to what we call cohort bias.”
The implications for justice are troubling. When cohort bias generates risk overpredictions, individuals may be denied bail, sentenced more severely, or denied parole due to inflated risk assessments. Through these mechanisms, cohort bias can trigger further criminalization, exacerbating existing racial inequalities.
Also troubling is that risk assessment instruments commonly go ten or even fifteen-plus years between updates, even though the world is constantly changing. Prediction is literally behind the times. As social change accelerates, outdated predictions will likely only get worse.
5. The character trap.
We typically view character as an immutable core of personality traits that guides our actions and defines who we are. It has a distinctly moral dimension that leads us to think—or at least talk—about individuals as good or bad.
Consider how criminal behavior is commonly explained. We invoke supposedly stable propensities in individuals’ dispositions, characters, and, less frequently, environmental background. And when certain individuals or groups repeatedly commit or are arrested for crimes, they are typically deemed to possess a criminal propensity or criminal nature. This idea manifests in terms like chronic or predatory offender, sometimes reaching extremes like superpredators.
Certainly, there are troubled and violent individuals, and I am not advocating that we dismiss the idea of individual responsibility or punishment. But blindness to social change in practice, especially by the criminal justice system, drives the character trap, a self-reinforcing cycle that labels and punishes individuals based on assumptions about their innate criminality. This can backfire, leading those criminally labeled to conform to those negative expectations. The character trap functions by hyper-focusing on allegedly stable individual qualities while neglecting changing social environments.
“We need to be thinking about moral societies more than moral individuals.”
For example, self-control is a component of character that is widely assumed to be stably associated with criminal propensity. And indeed, in my data, low self-control is related to arrest. But individuals from the 1980s cohort with high levels of self-control had the same arrest rates at age 20 as individuals from the 1990s cohort with low levels of self-control. This is a different way of viewing the problem. Self-control does not mean the same thing across time.
Cohort differences in arrest, violence, guns, mortality, and other life outcomes should cause us to rethink assumptions about stable character and propensity. They also require a distinct theoretical and policy perspective—one focused on cultivating social character instead of just individual character. We need to be thinking about moral societies more than moral individuals. This is essential for a just future, as we are all marked by time.
Enjoy our full library of Book Bites—read by the authors!—in the Next Big Idea App:
