Synthetic Control Group Models

To estimate the effect of an exogenous intervention on a treated unit, such as the implementation of a new criminal justice policy in a given state, a control unit is necessary. Comparing the treated unit’s (let’s say the state of California) time series of the outcome of interest (i.e. violent crime) to the population in which the treated unit is nested (i.e. national time-series) post-intervention would not yield interpretable findings, because it is unknown whether the difference in crime rates was caused by the intervention or some other factor. To navigate this obstacle, California’s crime rates would be compared to a weighted combination of other states chosen to optimally match California’s pre-intervention violent crime trends.

Continue Reading Synthetic Control Group Models

Prop 36

An inefficient reliance on incarceration as a means to reduce crime has led to massive incarceration costs and widespread demand for reform. Criminal Justice policymakers fear they will not be re-elected if they support reforms that are perceived to endanger public safety. The current study examines an instance of the diminishing returns paradox, California’s Substance Abuse and Crime Prevention Act. Using synthetic control group methods, this study evaluates whether SACPA threatened public safety or cost more than it saved, as critics predicted. The results suggest UCR Part 1 property crimes increased and aggravated assault decreased following SACPAs enactment.

MML

In November, 1996, California voters approved Proposition 215, which legalized the use of marijuana for medical purposes. Campaigns for and against the Proposition focused largely on the medical need for and effectiveness of marijuana. Although the debate emphasized cancer, AIDS, and spinal cord injuries, the Proposition’s language included virtually all chronic conditions, thereby effectively decriminalizing most marijuana use (Vitiello, 1998). Cultural values played a smaller role in the campaign. Opponents argued that legalization would “send the wrong message” to youth, for example, and this fear was realized to some extent. Public opinion surveys conducted before and after the election found increased acceptance of marijuana by Californians but no change in actual use (Khatapoush and Hallfors, 2004). The potential collateral consequences of legalization played no role in the public debate over Proposition 215. Would easier access to marijuana lead to an increase in driving under the influence, for example, and if so, would the higher prevalence of driving under the influence lead to an increase in traffic fatalities?

Continue Reading MML

Public Safety Realignment

In May 2011, the Supreme Court affirmed a 9th Circuit Court ruling that ordered the state to reduce its prison population. Although California’s prison population fell dramatically post-AB109, critics argued that releasing prisoners would lead to an increase in property crime. To test this hypothesis, Lofstrom and Raphael (2013) compared post-AB109 crime in California with crime in a matched synthetic control group of states (Abadie, Diamond and Hainmueller, 2010). They found an effect in Part I Auto Theft but no other crime. In a replication of the Lofstrom-Raphael synthetic control group analysis, we find no effect on any crime. We attribute this difference to two factors. First, the pre-intervention matched synthetic control group used by Lofstrom and Raphael was less than ideal. Second, due to the passage of time, we were able to construct a longer, better match control group.

Computationally intensive permutation testing

Statistical tests used in Time-series intervention models assume a long time-series of white-noise observations. Since these assumptions are seldom warranted in Synthetic Control Group designs, conventional significance tests are not available. In light of this obstacle, the exact significance of the post-intervention difference between a treated time series and its synthetic control must be calculated from a permutation test model.

Continue Reading Computationally intensive permutation testing