Causal Inference via Natural Experiments and Instrumental Variables

According to Laub and Sampson (2003), desistance from crime is made possible by “knifing off”—“offenders desist in response to structurally induced turning points that serve as the catalyst for sustaining long-term behavioral change” (p. 149). What turning points create are new situations that allow individuals to knife off the past, in part, by changing those routine activity patterns that led to trouble with the law prior to incarceration (Sampson and Laub, 2005). This idea is straightforward, but the corresponding intervention is extraordinarily complex, as the crippling expenditures on imprisonment and parole and the alarming recidivism rates in the United States clearly reveal. ¹