class: center, top, title-slide # Criminal Careers and Incapacitation ## SOC371 ### Chuck Lanfear ### Feb 1, 2020<br>Updated: Feb 1, 2021 --- # Overview 1. Incapacitation vs. Selective Incapacitation 2. Criminal Careers Research * Delinquency in a Birth Cohort * Participation vs. Frequency * Substantive Implications 3. Age, Crime, and Self-Control 4. Predicting High Rate Offenders * RAND Studies * Problems of Actuarial Prediction --- class: inverse # Incapacitation vs. Selective Incapacitation --- # Incapacitation Segregate criminals via imprisonment to prevent future crime ??? Cannot commit crimes, at least in public, if not in public -- * Incapacitation vs. deterrence + Behavioral vs. non-behavioral + Direct vs. indirect ??? Deterrence is behavioral: Use punishment to prevent everyone or the punished from committing future crime Incapacitation is not behavioral: You're just stuffing people in prison so others are not exposed to them; opportunity vs. motivation Incapacitation only directly effects whoever is incapacitated Deterrence has both direct (specific) and indirect (general) effects They're tied together though--probably hard to have incapacitation without deterrent effect -- * Modest effects on crime + Incarcerating one person prevents *their* crime + Most age out of crime + May be replacement of offender + Negative second-order effects ??? Incapacitation does guarantee those folks cannot commit crimes, but it appears to have weak effects on crime Long sentences ineffective as folks age-out anyway: Time spent in prison when not going to offend is time wasted To some degree there may be replacement: Remove one offender, someone else takes same role May have consequences: Criminogenic learning in prison, damage to neighborhoods they come from -- * **Expensive** + Greater than $30,000 per year on average in US + Loss of productivity ??? Warehousing in prison is extremely costly Combined with the modest effect, it is incredibly inefficient You get an added loss from that person no longer working and paying taxes Compare to work or education that can have direct cost but then increase tax revenue -- .centernote[*Incapacitation is a poor tool for general crime control*] --- # Selective Incapacitation Selectively imprisoning high rate offenders ??? While general incapacitation is not efficient, it might be a way to handle chronic offenders unresponsive to other strategies -- * How effective would this be? * Depends on who offends and how often ??? First question is how effective would it actually be to target chronic offenders -- * Problem is predicting future offending + Actuarial Models ??? Second question is how do we determine who will be chronic offender Need to be able to predict who will commit crimes in the future--this is an actuarial problem: risk analysis Crimes in past, no matter how numerous, don't necessarily guarantee many in the future --- class: inverse # Criminal Careers Research --- # Delinquency in a Birth Cohort Tracking criminal careers ??? Criminal careers describe the criminal involvement of individuals during the period between their first crime (known crime, or arrest) and last. Focus is only on criminal involvement--not other elements of life course. We'll get to life course studies later! For now, focusing on criminal careers and relevance to incapacitation. -- * Philadelphia 1945 cohort (Wolfgang, Figlio, & Sellin) * 9,945 boys * 35% with police record by 18 * Chronic offenders: 6% of cohort commit 50% of crimes ??? Wolfgang, Filgio, & Sellin conducted first cohort life course analysis of deviance All boys born in Philadelphia in 1945 (tracked via school and police records), tracking all recorded deviance from at least age 10 to 18. Rather than just knowing how many crimes committed or how many offenders there are, this allows tracking trajectories of criminality in full population This deals with issues like inadvertently comparing people raised under different conditions -- * Philadelphia 1958 cohort * 28,000 * 7.5% of cohort: > 5 arrests, 61% of crimes * Racine, WI (Shannon) * 25% account for 80% of arrests ??? Similar results in cohorts separated by 13 years -- Selective Incapacitation: Lock up this ~6%, cut crime in half ??? General idea is if most crime is concentrated in small group, why not lock up that group? --- # Blumstein & Cohen * Decompose crime rate into participation rate and frequency rate * Participation rate: Proportion of pop arrested * Frequency rate: Average arrests per arrested `\(\text{Crime Rate} = \text{Participation Rate} * \text{Frequency Rate}\)` `\(\frac{\text{# of Arrests}}{100,000} = \frac{\text{Arrested People}}{100,000} * \frac{\text{# of Arrests}}{Arrested People}\)` ??? Blumstein and Cohen take a demographic decomposition approach to this problem Overall crime rate is function of how many are involved in crime at a given time and how many crimes they each commit -- * Implications + If participation high, frequency low, selective incapacitation won't work well + If participation is low, frequency high, selective incapacitation has great potential ??? Separating participation and frequency is informative--it implies policy responses If most crime is concentrated in small number of chronic offenders, you can in principle isolate them If crime more uniformly spread out, each person in prison gains you little --- # Empirical Composition * Participation Rates + 50-60% of urban males arrested in lifetime + 25% for index crimes (14% white, 50% black) + Recidivism: 85-95% ??? Participation is surprisingly high, and most who participate have repeat crimes Large race difference in participation -- * Frequency Rates (Arrests in DC & Detroit) + Aggravated Assault: 2-3 per year + Robbery: 3-5 per year + Property Offenses: >5 per year + Auto Theft: 3 in DC; 9 in Detroit + No difference between white and black offenders ??? On average those arrested for a given crime are arrested multiple times in a year But frequency of crime looks basically identical across race Race predicts only participation, not frequency --- # Self-Report Frequency Greenwood & Abrahamse (1982) *Selective Incapacitation*. Rand Corporation. * Survey of 2100 male prison and jail inmates * Inmate reports before incarceration + Robbery * 50% of robbers <4 per year * 10% of robber >70 per year + Burglary * 50% of burglars <5 per year * 10% of burglars >195 per year ??? Prior studies looked at arrest. What about self-reports? Rand study to evaluate benefits of selective incapacitation; interviewed inmates about crime frequency prior to prison; ask rather than infer rates We'll visit their conclusions later Found high concentration of offenses in small population --- # Key Points * Crime frequency highly skewed + May be a subpopulation for incapacitation -- * Race differences are in participation not frequency + No justification for differential treatment ??? Blumstein & Cohen: Both "ethically wrong" and "empirically incorrect" to treat offenders differently by race; race differences in initiation but not frequency The criminal legal system gets involved only once people commit crimes If there's no racial difference after participation takes place, there's no justification for differential treatment Empirically could be justification for differential treatment in preventing participation--but that's not legal system, that's social policy (which they say is justified empirically and normatively); makes sense to use race to target social programs but not justice. -- * Incapacitation will not work in criminal markets ??? In criminal markets like drug selling there are more potential workers than available drugs. Take someone out and the role will just be filled with a new person --- class: inverse # Age, Crime, Self-Control --- # Age-Crime Curve .image-full[  ] ??? Recall the age-crime curve. Offending mostly in narrow range of adolescent years. Theory has to explain this. --- # Moffitt and Age-Crime Curve .image-full[  ] ??? Terrie Moffitt and colleagues suggest the age crime curve can be decomposed into two contributors Life-Course Persistent Offenders start early and continue longer Adolescence-Limited Offenders age into crime, approach persistent offenders, then age out rapidly --- # Life-Course Persistent Offenders * Serious underlying problems + Cognitive deficits + Attention Deficit Hyperactivity Disorder + Psychopathy ??? This is a psychological and social-psychological theory of crime It is also a taxonomic or typological theory: There's a qualitative difference between these types of offenders -- * Early maturation, large physical size ??? Group may mature earlier either mentally or physically, more likely to be larger in stature (it gives advantage in deviance) -- * Cumulative disadvantage ??? Cognitive and emotional issues often result of accumulated disadvantage: abuse, neglect, difficult environment. -- * Temper tantrums in late adulthood + Poor and unstable jobs + Poor and unstable marriages ??? Angry outbursts and problem behavior compromise all aspects of their life --- # Adolescence-Limited Offenders * "Normal" adolescents * Learn deviance from peers (partying, fighting, vandalism, theft) * Learn from life-course persistents * Age out due to life course transitions --- # Gottfredson & Hirschi * Gottfredson & Hirschi: "Ignore this research!" + Reflects unmeaningful statistical variation + Offenders differ by frequency only--same age patterns + Cannot identify career criminal before fact + Identify career criminal by 5+ offenses, but by then ceases to be career criminal due to age ??? Gott & Hirschi are curmudgeons of crim--all that matters is self-control, which is mainly determined in early childhood. There's more than self-control but their critiques of taxonomic approaches seems supported Different careers appear to be mainly variations around general crime curve -- * Analogy: For heart attacks, prevention is better + Don't focus on career criminals + Invest in children ??? So-called career criminals are only result of earlier failures Easiest, cheapest, most effective is to invest in children Nobel prize winner James Heckman share perspective on investment--cost savings in criminal justice more than pay for early childhood programs of many kinds. -- .centernote[*Can we predict the chronic offender?*] ??? Gottfredson & Hirschi say all that matters is social control and we can't identify career criminals until they're aging out --- class: inverse # Actuarial Prediction ??? Actuaries measure risk, often in industries like insurance and banking. Actuarial prediction is about assessing future risks based on present knowledge. --- ## Predicting High Rate Offenders Greenwood & Abrahamse (1982) * 1970s Context * Rehabilitation failures * No evidence for specific deterrence * Only certainty appears to deter (and it is *low*) * Prisons overcrowded ??? Use self-reports of crime and background data to predict high-risk offenders Study began in late 1970s; current vein of policy had been tailoring sentences to match offender needs Did not appear to be helping with crime--question is whether they were doing any real rehabilitation in prisons Regardless, public and politicians were abandoning and wanted alternate approaches Their study found prob of arrest and conviction is about 3% per crime. -- * Prediction Scale (Binary Measures; 4+ predicts "High-Rate") + Incarcerated >50% of 2 years before recent arrest + Prior conviction for crime in question + Juvenile conviction before age 16 + Commitment to juvenile institution + Heroin or barbiturate use as a juvenile + Heroin or barbiturate use within 2 years of recent arrest + Employed <50% of 2 years before recent arrest ??? These rules only applied to robbery and burglary, and scale is calculated separately for each Less crimes they felt were not appropriate for prison and more serious crimes, particularly homicide, are rarely done more than once by any one offender --- ## Sentencing Rule and Simulations * Sentencing rule: High rate offender gets 8 years in prison, all others 1 year ??? Greenwood & Abrahamse supposed giving offenders different sentences, then simulated what would happen provided actual prison pops -- * California (Texas) simulation results: + Robbery -15% (-10%), imprisoned robbers -5% (+30%) + Equivalent robbery reduction to non-specific 25% increase + Burglary -15% (-10%), imprisoned burglars +7% (+15%) + Poor classification accuracy * 55% false positives on high-rate (25% low, 30% medium) * 24% false negatives on low-rate (16% medium, 8% high) ??? CA has much higher number of chronic offenders; Texas has lower frequencies, so selective incap less effective False positives are low-rate offenders placed in higher category; 14% of true low-rate offenders ranked as high False negatives are high-rate offenders placed in lower category; 11% of true high-rate offenders ranked as low These look pretty bad, but the existing discretionary system--in which many similar factors are idosyncratically considered--does worse with about 80% more false positives and negatives and about 42% classification accuracy. -- * Reanalysis + After 24 months, prediction scale does 5-10% better than chance; not significant + May effectively identify low-risk offenders (Auerhahn 1999) + Legal correlation versus causation issues ??? Follow-up analyses have found it actually has pretty bad accuracy There are suggestions the scale may actually identify low-risk offenders well, just doesn't distinguish high-rate That is, most people with high number of scale items don't commit a lot of crime, but some do. Nearly no one with few or no scale items commit lot of crimes Maybe better to selective not incapacitate low risk Lastly, some have raised issue that use of purely correlation, not causal, factors linked to future crime may violate due process --- # Actuarial Prediction Problems * Changes sentencing from doing justice to crime control ??? No longer about punishing individuals or reforming them, but rather inferring future behavior and hoping locking up likely candidates reduces future crime Regardless of what either public or politicians often want, stated goals of legal system are justice not crime control -- * Used unemployment as predictor--victim blaming ??? They acknowledge controversial but don't really address it; basically say could drop or replace it with something else -- * Ignores agency, reform, and regression to mean ??? Once classified, folks are assumed to behave accordingly in future. Gives no opportunity for reform. Imagine considering not incapacitating but directing resources at potential high rate Very high rate offenders caught may be those in middle of spike in behavior--in general large increases in anything (not power distributed) tend to be followed by reductions, crime included. Might be incapacitating folks after they've already gotten past high rate. -- * Builds in racial bias ??? If structural racism impacts any of the predictors in an actuarial model, it builds that bias into system. This is a substantial problem with modern sentencing models using machine learning and other techniques. -- * Career lengths are usually short + Average less than 6 years for serious offenses ??? Even if you successfully predict, chances are that person ages out before much incapacitation --- class:inverse # Questions --- # For Next Time * Hirschi, Travis. [1969] 2001. *Causes of Delinquency.* New Brunswick, N.J: Transaction Publishers. Chapter 2 * Sampson, Robert J., and John H. Laub. 1990. "Crime and Deviance over the Life Course: The Salience of Adult Social Bonds." *American Sociological Review* 55:609-27. Things to pay attention to: * Control Theory + Negative position: Not about why we break law, but why we obey it + Basic assumption: Motivation for delinquency is universal * Contrast between theories of stability and change in crime + Note context of debate between sides + Consider implications for incapacitation and policy