Why Didn't You? SOC 3333 · Crime Lab · Midwestern State University

Every theory you have met so far asks why some people commit crime. Control theory refuses the question. It says the roots of crime are uncomplicated, that most offenses are easy and immediately rewarding, and that wanting to is not the rare ingredient. What is rare is the restraint. So the question worth asking is not why they did it. It is why the rest of us did not, and why any given person did not on any given day. This page starts with that inversion, then spends the rest of its time on the single strongest piece of evidence anyone has produced for the version of control theory that puts the restraint inside your own head.

Four moments, one question

Below are four ordinary situations. In each one, something could have gone differently and did not. For each, pick what you think actually held. This is an exercise in noticing, not a measurement of anything about you, and nothing here is recorded or scored.

What you just did

Look back at your four answers. Whatever mix you picked, notice what almost certainly did not appear in your reasoning: a claim that you wanted to do it less than someone else would have. The temptation was available in all four. Something else decided the outcome. That is the entire premise of this chapter, and Travis Hirschi built a theory on it in 1969.

The vocabulary of restraint

Formal
Official rules and the people who enforce them. Police, courts, deans, HR departments.
Informal
Disapproval from people who matter to you. The raised eyebrow, the disappointed parent, the group chat.
Direct
Somebody is actually watching right now. Supervision, a camera, a manager on the floor.
Indirect
Nobody is watching, but you would hate to let them down. The internalized audience you carry around.

Cross those two axes and you get the map. Control theory's bet is that the informal and indirect boxes do most of the work, and that the formal ones do far less than the criminal justice system likes to believe. Chapter 6 already showed you the empirical version of that claim: certainty of punishment barely registers because punishment so rarely arrives. Here it is again from the other direction.

One honest caution about this exercise. Asking people to explain their own restraint after the fact is not a research method. People are unreliable narrators of their own motives, and they tend to name reasons that make them look thoughtful. Treat what you just did as a way of getting the vocabulary into your hands, not as evidence for anything. The evidence starts two tabs from here.

Hirschi's four strands

Hirschi argued in 1969 that what holds people to conventional behavior is a bond with four separable strands. The strands are not causes that push you toward conformity. Their absence is what releases behavior that was available all along.

Attachment
Caring what specific people think and feel about you. The emotional strand.
Commitment
What you would lose. The degree, the job, the record, the standing. The rational strand.
Involvement
Time absorbed by conventional activity. Busy hands. The simplest strand, and the weakest.
Belief
Whether the rules feel legitimate to you in the first place. The strand about what you think the law is for.

Eight statements are below. Sort each one into the strand it belongs to. Some of them are genuinely arguable and the explanations say so.

Hirschi, T. (1969). Causes of Delinquency. University of California Press.

The general theory, and its biggest problem

In 1990 Hirschi did something unusual for a theorist. Working with Michael Gottfredson, he abandoned his own bond theory for a simpler one. Their general theory says there is a single trait, low self-control, that is set by roughly age ten through ineffective parenting and then stays put. It predicts not only crime but everything analogous to crime, since impulsive, short-sighted, risk-seeking people also smoke more, gamble more, and have more accidents.

The theory drew an obvious objection immediately, and it is the objection you should have too. If we identify low self-control by looking at who breaks rules, then saying low self-control causes rule-breaking says nothing at all. That is the tautology critique, and it is the reason the study on this page matters.

The design that answers it. Measure the trait in early childhood, using observers and parents and teachers, long before most of the outcomes exist. Then wait thirty years. If the childhood measure predicts the adult outcomes, the concept is doing real work rather than relabeling the thing it was supposed to explain. That is the Dunedin study.

1,037
children born in Dunedin, New Zealand, in 1972 and 1973
96%
still in the study at the age 32 follow-up
ages 3 to 11
when self-control was measured, using nine separate measures from observers, parents, teachers, and the children
Researchers split the cohort into fifths by childhood self-control. Among the fifth with the LOWEST childhood self-control, what percentage had a criminal conviction by age 32? For reference, the figure for the highest fifth was 13 percent.
25%

The gradient, across five different outcomes at age 32

lowest fifth in childhood self-control highest fifth

Moffitt, T. E., et al. (2011). A gradient of childhood self-control predicts health, wealth, and public safety. Proceedings of the National Academy of Sciences 108(7), 2693 to 2698. Income figure is annual household income under NZ $20,000. Retrieved August 2026.

The word doing the work in that title is gradient. This is not a finding about a troubled minority. Each step down the self-control distribution came with worse outcomes, all the way along, including among children who were entirely ordinary. That shape is much harder to explain away than a comparison of extremes would be.

Two further checks worth knowing

The confound check. The obvious objection is that childhood self-control is just a proxy for being poor or being smart. The authors reported that self-control predicted these outcomes independent of social class origins and IQ.

The sibling check. In a separate British cohort, researchers compared 509 pairs of same-sex fraternal twins raised in the same household. Within pairs, the sibling with lower self-control was more likely by age 12 to have started smoking, to be doing worse at school, and to be behaving antisocially, and those differences held after accounting for differences in IQ between the siblings. Same home, same parents, same neighborhood, different outcomes.

What the sibling check does and does not do. Comparing siblings removes everything the two children share, which is a great deal. It does not remove whatever made them differ in the first place, and it cannot tell you whether that was genetic, prenatal, or a difference in how the same parents treated two different children. It rules out a family of alternative explanations. It does not establish the mechanism Gottfredson and Hirschi proposed, which was specifically ineffective parenting.

Does the general theory survive testing

The Dunedin findings are real and they are strong. They are also not the whole scorecard. A theory this ambitious makes several distinct claims, and they have not fared equally. Predict each verdict before you reveal it.

Pratt, T. C., and Cullen, F. T. (2000). The empirical status of Gottfredson and Hirschi's general theory of crime: A meta-analysis. Criminology 38(3), 931 to 964. Twenty-one studies, 126 effect size estimates, 49,727 cases. Richmond-Rakerd, L. S., et al. (2021). Childhood self-control forecasts the pace of midlife aging and preparedness for old age. PNAS 118(3). Retrieved August 2026.

Which strand do programs actually reach

Here is a use for theory that goes beyond the exam. Hirschi's four strands each imply a different intervention. Attachment says give a young person an adult who cares about them. Involvement says fill their time. Self-control theory says go further upstream and work on parenting. All three of those exist as real program types, all three have been evaluated many times over, and the National Institute of Justice grades them at CrimeSolutions using three labels: Effective, Promising, and Ineffective.

Predict each rating before you look. The interesting part is not whether you get them right. It is whether the pattern matches what Hirschi said about which strand was strongest.

All ratings from CrimeSolutions, National Institute of Justice, crimesolutions.ojp.gov, practice profiles for After-School Programs and Mentoring for Youth Development and the program profile for GenerationPMTO. SMD is a standardized mean difference. Retrieved August 2026. Ratings are periodically re-reviewed and can change, so check the site before citing any of these in a paper.

Forum: does Dunedin rescue the general theory

Post your initial response and then reply to two classmates. Length and deadlines are in that week's module in the LMS.

  1. Say what you guessed for the conviction rate in the lowest self-control fifth, and what your reaction was to the real figure. Guesses are not graded and being far off is common.
  2. Explain, in your own words, how the Dunedin design is supposed to answer the tautology critique. Then say whether you think it fully answers it, and if not, what is left over.
  3. Pick ONE of these two findings and explain what it costs the claim that self-control is the general cause of crime: that the self-control effect was much weaker in longitudinal studies than in cross-sectional ones, or that social learning variables kept their predictive power even with self-control in the model. Do not pick both, because your replies depend on having picked one.
  4. Gottfredson and Hirschi say self-control is essentially fixed by about age ten. The age 45 follow-up found that adult self-control predicted midlife outcomes even after accounting for childhood self-control. Which do you find more convincing, and what would change your mind?
Replies: respond to two classmates who chose the other finding in step 3. Your job is not to agree or disagree. It is to say what their finding costs the theory that yours does not, and vice versa.

On writing this one well. The weak version of this post says the theory is right because the numbers are big, or says the theory is wrong because people can change. Both are available and neither is interesting. The strong version separates the claim that self-control predicts crime, which is very well supported, from the claim that self-control is the single general cause of crime, which is not. Almost everything worth saying in this forum lives in that gap.