Which statement did Sutherland himself label 'the principle of differential association'?
Choose an answer, then check it.
Criminal Justice & Criminology · Criminology
Social Learning Theory of Crime
On this page 9 sections
In 30 seconds
Edwin Sutherland argued that criminal behavior is learned the way anything else is: in communication inside close personal groups. His 1947 statement runs to nine propositions, and its core is an Excess of definitions Sutherland's sixth proposition: offending follows when the meanings a person has acquired that favor violating the law outweigh those that oppose it. It is a ratio between two kinds of meaning, not a count of associates. Full entry → favorable to breaking the law over definitions unfavorable to it. Ronald Akers later rebuilt the theory on operant conditioning. Association and attitudes are among criminology's steadiest correlates, but whether delinquent friends cause delinquency or delinquent youths simply find each other is still argued.
Why this matters
Almost every claim you will hear about peers and crime rests on this theory, usually without naming it: that a young person fell in with the wrong crowd, that attitudes toward the law can be changed, that separating someone from a group changes what they do. Courses in criminology, sociology, and social work expect you to state the mechanism rather than the slogan. The theory is also the best available teaching case for a problem that runs through all social science: when two things go together because people choose each other, no amount of correlation settles which one moved first. Learning to say exactly what the data can and cannot show here transfers everywhere.
The college version
The 1947 statement, and what it was answering
Sutherland's textbook ran through several editions; the one criminologists mean when they say Differential association The process by which a person is exposed to competing patterns of meaning about the law through interaction with others, and the resulting balance between the two. Full entry → is the fourth, published as Principles of Criminology by J. B. Lippincott in 1947. Its theory chapter separates two kinds of explanation. One is situational: criminal behavior arises in what he calls the person-situation complex, where the situation that matters is the one as the person involved defines it. The other is historical or genetic - an account of the life experience that produced a person who would define a situation that way. The nine propositions are the second kind. Sutherland was arguing on two fronts. Against inherited theories of criminality, he wanted a mechanism running entirely through ordinary social interaction, which is why his first proposition is a denial as much as an assertion. Against explanation by general drives - the money motive, frustration, striving for status - he set a demanding test: an account that explains lawful behavior just as well as criminal behavior has explained nothing. Respiration is necessary for burglary and for teaching school. That test is proposition nine, the quietest and most exacting item on the list.
The nine propositions
One: criminal behavior is learned - negatively, it is not inherited, and a person untrained in crime does not invent it any more than an untrained person invents machinery. Two: it is learned in interaction with other people in a process of communication, verbal and also gestural. Three: the principal part of that learning occurs within intimate personal groups. Sutherland attaches a negative clause modern readers often skip - impersonal agencies of communication, and he names picture shows and newspapers, play a relatively unimportant part. Four: what is learned has two parts, the techniques of committing the crime, sometimes very complicated and sometimes very simple, and the specific direction of motives, drives, rationalizations, and attitudes. Five: that direction is learned from definitions of the legal codes as favorable or unfavorable; in some settings a person is surrounded by people who treat the law as rules always to be observed, in others by people whose definitions favor violating it, and in American society, he writes, the definitions are almost always mixed. Six: a person becomes delinquent because of an excess of definitions favorable to violation of law over definitions unfavorable to violation of law - the proposition Sutherland himself calls the principle of differential association. Seven: differential associations may vary in frequency, duration, priority, and intensity. Eight: the process involves all the mechanisms involved in any other learning, and so is not restricted to Imitation Patterning one's own conduct on behavior observed in others and on the consequences that followed for them; in Akers' account it matters more for starting a behavior than for sustaining it. Full entry →. Nine: criminal behavior expresses general needs and values but is not explained by them, since noncriminal behavior expresses the same.
Definitions, and what excess does not mean
A definition, in this tradition, is an attitude or meaning a person attaches to an act - the verbalizations that make a particular offense justified or unjustified in a particular situation. Some cover almost anything, some attach to one offense, some only to one situation. The excess principle is a claim about the balance between favorable and unfavorable, not about how many rule-breaking people someone knows. Sutherland is explicit that the principle concerns counteracting forces: people become criminal both through contact with criminal patterns and through isolation from anti-criminal ones, and neutral associations - his example is learning to brush your teeth - matter mainly by occupying time. So a person with many law-breaking acquaintances and a dense supply of definitions against law-breaking is not predicted to offend. The theory is also, in principle, falsifiable: Matsueda states the individual-level hypothesis as three conditions - techniques, an excess of favorable definitions, and objective opportunity - so that crime in the absence of all three, or no crime when all three are present, would count against it. The catch is measurement. Nobody has ever counted definitions. Sutherland conceded that a precise description would state the modalities quantitatively and reach a mathematical ratio, that no such formula existed, and that developing one would be extremely difficult. Researchers still use survey scales standing in for a ratio no study has ever computed.
The modalities: frequency, duration, priority, intensity
Proposition seven says associations vary along four dimensions. Frequency and duration Sutherland treats as obvious and needing no explanation - how often, and for how long. Priority is about earliness: patterns formed in early childhood may persist through life. But he immediately qualifies it, saying the tendency has not been adequately demonstrated and that priority seems to be important principally through its selective influence. That phrase matters more than its length suggests. Sutherland is conceding that being shaped early mostly works by steering whom a person later ends up around - a selection process, sitting inside the theory from the beginning. Intensity, he says plainly, is not precisely defined; it concerns the prestige of the source of a criminal or anti-criminal pattern and the emotional reactions attached to the relationship. The upshot is that a definition heard once, early, from someone who matters can outweigh many hours with people who do not - which is why textbook illustrations use a long-standing close friendship as the strong case and a new acquaintance from class as the weak one.
Burgess and Akers, 1966: adding the machinery
Proposition eight says the learning involves all the mechanisms involved in any other learning, without saying what those mechanisms are. That gap is what Robert Burgess and Ronald Akers set out to close in A Differential Association-Reinforcement Theory of Criminal Behavior, published in Social Problems in 1966. Taking up C. R. Jeffery's suggestion that operant behavior theory be brought in, they restated Sutherland's nine propositions as seven, each mapped onto the Sutherland proposition it reformulated. Proposition one became the claim that criminal behavior is learned according to the principles of operant conditioning, absorbing Sutherland's first and eighth. Intimate personal groups became those groups that comprise the individual's major source of reinforcements. The last of the seven states that the strength of criminal behavior is a direct function of the amount, frequency, and probability of its reinforcement. Two things are worth noticing. The count changed - seven, not nine - a detail study guides routinely get wrong. And the motive was measurability: Sutherland had named a process he could not specify, and behavioral psychology offered a vocabulary with observable referents. Sutherland and B. F. Skinner were colleagues at Indiana University in the late 1940s, and Sutherland never used any of it.
Akers' four elements
The 1966 revision drew criticism, and Akers eventually dropped the serial list altogether, restating the theory around four core elements. Differential association is the interaction that supplies models, definitions, and patterns of reinforcement; Akers widened it well past Sutherland's intimate personal groups to include indirect ties, reference groups, and symbolic exposure through media - a deliberate reversal of Sutherland's dismissal of picture shows and newspapers. Definitions are the attitudes and meanings the person holds, general or specific, split into positive definitions, which approve of the act outright, and neutralizing definitions, which concede the rule and excuse this instance of breaking it. Differential reinforcement In Akers' theory, the balance of rewards and punishments a person has experienced and anticipates for an act, operating through positive and negative reinforcement and positive and negative punishment. Full entry → is the balance of experienced and anticipated reward and punishment attached to the behavior, running through positive reinforcement, negative reinforcement, positive punishment, and negative punishment. Imitation is modeling one's own behavior on observed behavior and its consequences for the model; Akers likens it to vicarious reinforcement, drawing on Bandura's demonstration that a pattern of behavior can be acquired by watching someone else perform it, and treats it as mattering more for starting a behavior than for maintaining it. The general psychology of observational learning is a separate lesson; what is criminological here is the claim that these four processes, operating on law-violating conduct, produce it.
Neutralization: definitions you can actually hear
Sutherland's rationalizations are hard to measure because they are internal. Gresham Sykes and David Matza gave them a shape in Techniques of neutralization Five justifications named by Sykes and Matza - denial of responsibility, denial of injury, denial of the victim, condemnation of the condemners, appeal to higher loyalties - that suspend a rule the person otherwise accepts, before the act rather than after. Full entry →: A Theory of Delinquency, published in the American Sociological Review in December 1957. Their starting point is a paradox: delinquents largely do accept conventional morality and show guilt and shame. So the question is not how a rival value system gets installed but how the ordinary one gets switched off for the occasion. They named five techniques - denial of responsibility, denial of injury, denial of the victim, condemnation of the condemners, and appeal to higher loyalties. Everyone remembers the list; the load-bearing claim is the one about timing. Sykes and Matza's concern was with justifications that, in their words, precede deviant behavior and make deviant behavior possible. Neutralization is preparation, not an alibi. Later work added techniques - the metaphor of the ledger, the defense of necessity, everybody does it - and the hardening thesis, on which repeated neutralized offending gradually weakens conventional belief until neutralizing is no longer needed. Structurally, neutralizations are definitions in Sutherland's sense made observable, which is why measures of pro-criminal attitudes usually turn out to be measuring them.
What the evidence supports, and what it does not
The reference point is a meta-analysis by Pratt and colleagues published in Justice Quarterly in December 2010. It swept the leading criminology and criminal justice journals from 1974 to 2003, collecting 133 studies that measured social learning variables, which produced 246 statistical models, 704 effect-size estimates, and 118,403 cases. Read the results carefully, because the two standard summaries differ in emphasis. Akers and Jennings note that all four elements produced significant independent mean effects, ranked largest to smallest as differential association, definitions, differential reinforcement, imitation, and that only 10 of 55 moderator analyses reached significance across sampling frame, race, sex, and age - which they read as stable general effects. The published abstract is blunter: relationships of crime and deviance to differential association and definitions are quite strong, while those for differential reinforcement and modeling or imitation are modest at best, and effect sizes for three of the four elements shifted significantly with model specification and research design. Both describe the same numbers. The pattern worth carrying away is that the components bearing the empirical weight are the ones Sutherland already had, and the operant machinery Burgess and Akers contributed is the least well supported part. Association and attitudes are among the most consistently reproduced correlates in criminology - which is a claim about correlation.
Selection or socialization?
The standing objection is that the correlation is spurious. Travis Hirschi argued it for decades, and under the self-control theory he later wrote with Michael Gottfredson it sharpens: children with low self-control make poor friends, end up in groups with each other, and offend - so both the offending and the friendships are downstream of one trait. Birds of a feather, not contagion. Notice that the theory left the door open. Sutherland wrote that it is not necessary, at this level of explanation, to explain why a person has the associations they have, and he had already said priority operates mainly through selective influence. Akers too accepted that people select into groups. Researchers have pushed back four ways: adjusting for covariates such as social bonds and self-control; adding a lagged measure of prior delinquency; fitting fixed-effects models that use only within-individual change and so absorb every stable personal characteristic; and fitting stochastic actor-based network models that estimate friendship formation and peer influence at once. A peer effect survives all four. That is real triangulation, and also the ceiling: randomly assigning adolescents to friendship groups is not feasible, so every estimate rests on observational data and unobserved bias can never be ruled out. The estimated effect also shrinks when it is estimated longitudinally, when selection is explicitly modeled, and when peers report their own behavior instead of the respondent guessing at it. McGloin and Thomas's conclusion is the fair one: posing this as selection or socialization misreads both Sutherland and Akers, since both processes operate, and the productive questions are how, when, and for whom.
Boundaries
This is one theory family among several, and it is deliberately narrow. It says nothing about why some neighborhoods supply more definitions favorable to crime than others; Sutherland handled that separately with what he called Differential social organization Sutherland's group-level companion concept, under which a group's crime rate reflects the relative strength of its organization in favor of crime versus its organization against crime. Full entry →, the idea that a group's crime rate reflects the relative strength of its organization for and against crime. Strain, social disorganization, labeling, and routine activity are parallel explanations with their own lessons, and the framework placing them sits in this unit's opening topic. Two cautions. The theory describes a process, not a category of people: Sutherland devoted a section of the same 1947 book to white-collar criminality, treating business offending on the same terms as street crime. And empirical support is not a warrant for any particular policy - what follows from evidence about attitudes and associations is contested, and this lesson takes no position on it. This is educational material about criminological theory and research. It is not legal advice, not clinical advice, and not a guide to any intervention.
Eli explains
The same idea, in plain words
Explain it like I’m 10
People do not learn crime from thin air. They learn it the same way they learn slang, cooking, or how to act at a family dinner - from people they spend real time with, who show them how something is done and, just as importantly, what to say about it afterward. That second part is the whole theory. If everyone around you treats shoplifting as unthinkable, you collect reasons not to. If some people around you treat it as no big deal, or as something a store deserves, you collect reasons why it is fine this once. Sutherland's claim is that whichever pile of reasons gets bigger is the one that decides. Ronald Akers later added the part Sutherland skipped: whether the behavior keeps happening depends on what actually follows it - who laughed, who was impressed, whether anything bad happened.
Picture it like this
Think of an accent. Nobody sits a child down and teaches one. It arrives from the people the child is around the most, the earliest, and cares about the most, and it turns out to be very hard to hear in yourself. Attitudes toward rules travel the same way, through the same channels.
Where the picture stops working
The analogy breaks in two places. Accents are near-universal - everyone ends up with one - while crime is not, so an accent cannot show you why two children in the same house diverge. And you do not choose your accent; a large part of this lesson is about the fact that people partly choose their friends, which is exactly the complication researchers have spent seventy years trying to measure around.
Worked example
A researcher studies a 16-year-old, call her R, who has begun shoplifting with three friends. Applying Akers' four elements gives four separate questions rather than one. Differential association: who supplies her exposure, how often, for how long, since when, and how much do they matter to her? Definitions: what does she say about it - not in general, since she may well condemn theft in the abstract, but about this act, in this store, in this moment? If she says the chain writes it off anyway, that is a denial of injury, one of Sykes and Matza's five, functioning as a definition favorable to the act. Differential reinforcement: what has followed? Group laughter and status are positive reinforcement; nothing happening is the absence of punishment, which reinforces just as effectively. Imitation: did she watch someone else do it first? Now the hard part. Every one of those associations is also a choice R made. A skeptic will say she picked friends who already matched her, and the correlation says nothing about direction. What would move the argument is within-person change - did her offending rise after the friendships formed, holding everything stable about her constant - which is why fixed-effects and network models, not cross-sectional surveys, are where this dispute is actually fought.
Key takeaway
Social learning theory says crime is learned in interaction - techniques plus a supply of justifications - and that the balance of those justifications decides. Its association and definition components are among criminology's best-replicated correlates; whether that reflects influence, selection, or both is still being measured.
Quick check
3 questions here, of 5 in this lesson’s practice set. Answers stay hidden until you check.
Sutherland's seventh proposition names four modalities along which differential associations vary. What are they?
Choose an answer, then check it.
What did Burgess and Akers (1966) change about Sutherland's theory?
Choose an answer, then check it.
Study tools & related lessonsYou’ll learn to · Common mistakes · Easily confused · Key vocabulary · Related
You’ll learn to
- State Sutherland's nine propositions in order and explain what each one is doing in the argument.
- Define the excess-of-definitions principle and distinguish it from counting a person's delinquent friends.
- Explain the four modalities of association and identify Sutherland's own reservations about priority and intensity.
- Distinguish Sutherland's differential association from Akers' social learning theory, and name the four elements Akers added or retained.
- Apply the techniques of neutralization to a described case and explain why Sykes and Matza placed them before the act rather than after it.
- Evaluate the evidence for social learning variables, including what the selection-versus-socialization dispute does and does not resolve.
Common mistakes
Reading the theory as 'people copy what they see', so that learning means imitation.
Sutherland ruled that out explicitly. Proposition eight says the learning involves all the mechanisms involved in any other learning, and negatively, that it is not restricted to imitation. Imitation only becomes a named component decades later, in Akers, and it is the weakest-performing of his four elements.
Reversing the excess principle - saying that exposure to more definitions unfavorable to the law leads to offending. This reversal appears in published course material, so it is worth checking yourself against the original.
The direction is favorable over unfavorable. A person becomes delinquent because of an excess of definitions favorable to violation of law over definitions unfavorable to violation of law. Definitions unfavorable to law-breaking are the counterweight, and Sutherland stresses that isolation from anti-criminal patterns matters as much as contact with criminal ones.
Testing the theory by counting how many delinquent friends someone has.
The claim is about a ratio of definitions weighted by frequency, duration, priority, and intensity - not a headcount. One early, intense relationship can outweigh several casual ones, and a person with many rule-breaking acquaintances but a dense supply of definitions against rule-breaking is not predicted to offend.
Treating techniques of neutralization as excuses offenders make afterward.
Sykes and Matza's argument is about sequence: they were concerned with justifications that precede deviant behavior and make deviant behavior possible. A post-hoc account is rationalization; neutralization is what lowers the guard beforehand. Losing that distinction turns a causal claim into a description of interview transcripts.
Concluding either that peer influence is proven causal or that it is entirely selection.
Neither is supportable. A peer effect survives covariate adjustment, lagged dependent variables, fixed-effects models, and network models that estimate friendship selection directly - but no observational design identifies the true causal effect, and the estimate shrinks under the more demanding specifications. Sutherland and Akers both allowed selection; the current view is that both processes operate.
Easily confused
Sutherland's differential association (1947) vs. Akers' social learning theory
Sutherland specifies the content and the social location of learning but leaves the learning mechanism unnamed; Akers, after Burgess and Akers (1966), supplies the mechanism from operant conditioning and adds differential reinforcement and imitation as separate components. Sutherland confines learning largely to intimate personal groups and discounts media; Akers extends it to indirect ties and symbolic exposure.
Definitions vs. Differential reinforcement
Definitions are meanings held before and during the act that make it justified; differential reinforcement is the balance of consequences that follow the act and shape whether it recurs. In Pratt and colleagues' meta-analysis, definitions relate strongly to offending while differential reinforcement is modest at best.
Neutralization vs. Rationalization
Same content, different timing, and only one of them is a causal claim. Neutralization precedes the act and makes it possible; rationalization is offered after the fact and explains nothing about why the act occurred.
Differential association (individual level) vs. Differential social organization (group level)
The first explains why a particular person offends, through the balance of definitions they have acquired. The second explains why one group's crime rate exceeds another's, through the relative strength of organization for and against crime. Sutherland required the two to be consistent, since a group rate is a summary of individual acts.
Key vocabulary
- Differential association
- The process by which a person is exposed to competing patterns of meaning about the law through interaction with others, and the resulting balance between the two.
- Definitions (favorable and unfavorable)
- Attitudes, motives, verbalizations, and rationalizations that make a specific act justified or unjustified for the person holding them; they may be general, specific to one offense, or specific to one situation.
- Excess of definitions
- Sutherland's sixth proposition: offending follows when the meanings a person has acquired that favor violating the law outweigh those that oppose it. It is a ratio between two kinds of meaning, not a count of associates.
- Modalities of association
- The four dimensions along which exposure varies - frequency, duration, priority (how early), and intensity (prestige of the source and emotional weight of the relationship).
- Differential reinforcement
- In Akers' theory, the balance of rewards and punishments a person has experienced and anticipates for an act, operating through positive and negative reinforcement and positive and negative punishment.
- Imitation
- Patterning one's own conduct on behavior observed in others and on the consequences that followed for them; in Akers' account it matters more for starting a behavior than for sustaining it.
- Techniques of neutralization
- Five justifications named by Sykes and Matza - denial of responsibility, denial of injury, denial of the victim, condemnation of the condemners, appeal to higher loyalties - that suspend a rule the person otherwise accepts, before the act rather than after.
- Selection versus socialization
- The dispute over causal order in the peer-offending correlation: whether being around rule-breakers changes behavior, or whether people already inclined to break rules choose each other.
- Differential social organization
- Sutherland's group-level companion concept, under which a group's crime rate reflects the relative strength of its organization in favor of crime versus its organization against crime.
Sources & references
- Principles of Criminology, fourth edition (theory of criminology chapter, pages 6-9) — Edwin H. Sutherland; J. B. Lippincott Company (scan in the Digital Library of India collection, Internet Archive)
- A Differential Association-Reinforcement Theory of Criminal Behavior, Social Problems 14(2):128-147 (Autumn 1966) — Robert L. Burgess and Ronald L. Akers; Society for the Study of Social Problems / Oxford University Press
- Social Learning Theory, chapter 12 in The Handbook of Criminological Theory (ed. Alex R. Piquero), pages 230-240 — Ronald L. Akers and Wesley G. Jennings; John Wiley & Sons
- Techniques of Neutralization: A Theory of Delinquency, American Sociological Review 22(6):664-670 (December 1957) — Gresham M. Sykes and David Matza; American Sociological Association
- Revisiting neutralization theory and its underlying assumptions to inspire future information security research (WISP 2022 Proceedings, 17th Pre-ICIS Workshop on Information Security and Privacy) — Wael Soliman and Hojat Mohammadnazar; Association for Information Systems (self-archived copy in JYX, University of Jyvaskyla)
- Empirical Status of Social Learning Theory: A Meta-Analysis (Justice Quarterly 27(6):765-802, December 2010; NCJ 232709) — Travis C. Pratt, Francis T. Cullen, Christine S. Sellers, L. Thomas Winfree Jr., Tamara D. Madensen, Leah E. Daigle, Noelle E. Fearn and Jacinta M. Gau; abstract hosted by the Office of Justice Programs NCJRS Virtual Library
- Peer Influence and Delinquency, Annual Review of Criminology 2:241-264 (2019) — Jean Marie McGloin and Kyle J. Thomas; Annual Reviews (author copy hosted by the University of Maryland Department of Criminology and Criminal Justice)
- Sutherland, Edwin H.: Differential Association Theory and Differential Social Organization, in Encyclopedia of Criminological Theory (eds. Francis T. Cullen and Pamela Wilcox), pages 899-907 — Ross L. Matsueda; SAGE Publications (publisher-hosted PDF on study.sagepub.com)
- 9.3 Differential Association Theory, in Introduction to Criminology — Zachary Rowan and Michaela McGuire, in Hassan and Lett (eds.), Kwantlen Polytechnic University / Pressbooks
- 9.4 Social Learning Theory, in Introduction to Criminology — Zachary Rowan and Michaela McGuire, in Hassan and Lett (eds.), Kwantlen Polytechnic University / Pressbooks
- 9.5 Research in Learning Theories, in Introduction to Criminology — Zachary Rowan and Michaela McGuire, in Hassan and Lett (eds.), Kwantlen Polytechnic University / Pressbooks
- Generative AI and criminology: A threat or a promise? Exploring the potential and pitfalls in the identification of Techniques of Neutralization (ToN), PLOS ONE 20(4):e0319793 — Federico Pacchioni and colleagues; Public Library of Science
- 5.10. Learning Theories, in SOU-CCJ230 Introduction to the American Criminal Justice System — Brian Fedorek, in Burke, Carter, Fedorek, Morey, Rutz-Burri & Sanchez (Open Oregon Educational Resources / Pressbooks)
- Transmission of Aggression Through Imitation of Aggressive Models (Journal of Abnormal and Social Psychology, 63, 575-582) — Albert Bandura, Dorothea Ross and Sheila A. Ross; reproduced by Classics in the History of Psychology, York University
EliExplains lessons are original prose written from the open, credible references above. See Copyright & Licensing.
Researched 2026-08-19
Educational content only. It is not medical, legal or professional advice. Found an error? Tell us.