Recently I have noticed that our knowledge concerning crime and crime prevention is moving in multiple directions. On one side, confidence is building that we now know enough about what works, especially in prevention with individuals at risk for offending, to justify pushes to mandate increased use of evidence-based programs. We have come far enough to now be grappling with the challenges of the next stage, including: trying to scale-up effective programs, understanding what makes program implementation succeed or fail, and maintaining investment to develop and test new programs. This is a sea change from the age of "Nothing Works" of 30 years ago.

At the same time, our relative ignorance concerning crime trends and their drivers has lately become much more apparent. The broad and sustained crime decline that we have enjoyed since the mid- to late-1990s was not anticipated by crime experts. Even more surprising is that economic conditions of the Great Recession have not spurred increased crime. This has gone on long enough to produce recent expressions of puzzlement from many crime experts.

This begs the question: Why have we made so much more progress in developing evidence-based prevention programs for individuals than in understanding macro-level crime trends? Two basic explanations occur to me.

First, at the individual-level we use experiments (and quasi-experiments) to confidently isolate the effects of one intervention from the effects of other things. And because we can test one idea at a time, we don’t have to figure out which interventions are the most powerful before trying to make headway against less powerful—but still significant-- factors. We can pursue stronger and weaker forces in parallel. But in macro-level studies, we can rarely experiment with crime’s drivers, and hardly ever in a way that isolates their effects. Instead we must simply observe them as they happen, and we are less able to confidently isolate the effects of particular macro-level factors.

Second, our knowledge accumulates much faster for individual-level intervention studies than for macro-level studies for several reasons. To find out if something works at the individual level, each study needs to involve tens to hundreds of program participants. But a macro-level study needs data from thousands or millions of people. Also, most individual-level intervention studies take months to years of data while macro-level studies can require decades of data. And some important macro-level effects on crime can take decades to develop, and even longer to learn about. A case in point: two much-debated “crime decline” hypotheses concern whether changes in abortion law or the removal of lead from children's environments have reduced teen-age crime. Put together, all of this means that we have--and will probably always have--many more distinct, independent studies of individual-level interventions than of macro-level effects.

We are at an optimistic moment for criminal justice policy because we have made notable progress in developing the evidence about what works with individuals, and also because persistently low crime rates create an opportunity for more thoughtful and evidence-based policy. Even so, if my speculations here are even remotely correct, it’s chastening to think that it will likely be decades before we understand enough about macro-level crime trends to anticipate either large crime increases or decreases.