A recent study in the peer-reviewed journal PLOS ONE examined the effects of medical marijuana legalization (MML) on crime rates. The work has garnered some notable media attention. The authors of the study noted that their "findings run counter to arguments suggesting the legalization of marijuana for medical purposes poses a danger to public health in terms of exposure to violent crime." Further investigation of this study is required.

The study used data "between 1990 and 2006 [that] were obtained from the Federal Bureau of Investigation's Uniform Crime Reporting (UCR) Program." As the authors note, "between 1990 and 2006, the following 11 states legalized marijuana for medical use, with the year the law was passed in parentheses: Alaska (1998), California (1996), Colorado (2000), Hawaii (2000), Maine (1999), Montana (2004), Nevada (2000), Oregon (1998), Rhode Island (2006), Vermont (2004), and Washington (1998)." One obvious question comes to mind: how do you examine MML effects on crime rates using a dataset that ends in 2006 when there are states such as Rhode Island that only legalized medical marijuana in 2006? And a two-year post-MML dataset for states such as Montana and Vermont, which legalized the medical use of this narcotic in 2004, seems very short for trend analysis.

Overall, primary concerns with the study appear to be the short, and inequivalent (i.e., not all MMLs occurred in the same year among the states) pre- and post-MML timeframes chosen for these states, as well as the limited -- and somewhat unusual -- choice of sociodemographic control variables (e.g., percent of the civilian labor force unemployed; the total employment rate; percent of the population living below the poverty line; real per-capita income; the proportion of residents aged 15-24; the proportion of residents aged 25-34, the proportion of residents aged 35-44 years; the per-capita rate of beer consumption; the proportion of residents with at least a bachelor's degree; the percent of the state's population that lived in a metropolitan area; number of prison inmates per 100,000 residents; and the number of police officers per 100,000 residents). Certainly having "beer shipments (31-gallon barrels) per 100k" as a variable seems odd. Some of these variable may also display extremely high intercorrelation of predictors, or multicolinearity, which is a major faux pas in traditional multiple linear regression approaches.

When I look at the FBI's UCR data for violent crime rates, which are available between 1960 and 2012, I see some potentially different results. Controlling for all possible underlying variables is clearly desirable, but effectively impossible. The choice of control variables can often predetermine the findings, a problem commonly observed in econometrics. For a first-cut analysis, looking at trends in crime rates before and after MML in each state can be informative. If this approach agrees with the study results, we can assume the study's results are likely on solid footing. If not, perhaps the study did not reach the correct conclusions and warrants further examination.

The following table shows the date of MML in each state, along with the change in violent crime rate before and since MML. To compare with the post-MML change in violent crime rates, the corresponding change in the US national violent crime rate is provided, along with the pre-MML change in crime rate (using the same time length as for the respective post-MML period) for both the state and the nation as a whole. In other words, if a state had MML in 2000, the post-MML period would go from 2000-2012 and the pre-MML period from 1988-2000 (i.e., 12 years in both directions).

In the period before MML, eight (Alaska, California, Colorado, Maine, Nevada, Oregon, Rhode Island, and Washington) of the eleven states had violent crime rates that were declining faster than, or equivalent to, the national rate (highlighted in green). The remaining three states (Hawaii, Montana, and Vermont) had pre-MML violent crime rate changes that were not declining as much as the national rate (highlighted in red).

Compare this to the equivalent length post-MML period. Now only two of the states (California and Oregon) have violent crime rates declining more rapidly than the national rate. The other nine states all have rates that are either increasing post-MML (Maine, Nevada, Rhode Island, and Vermont) or not decreasing as rapidly (Alaska, Colorado, Hawaii, Montana, and Washington). The violent crime rate trend reversal in Maine, Nevada, Rhode Island, and Vermont following MML is notable. All these states had declining violent crime rates before MML, and now have increasing violent crime rates since MML, whereas the corresponding national rate has declined rapidly in the post-MML periods.

The national violent crime rate peaked in 1991, similar to that of Alaska (1995), California (1992), Colorado (1992), Hawaii (1995), Nevada (1994), Rhode Island (1991), and Washington (1992). The violent crimes rates peaked at different times in Maine (1977), Montana (2007), Oregon (1985), and Vermont (1979). The graphs below show the violent crime rates (per 100,000 population) since 1960 for each state, as well as the national rate. The red dashed lines indicate the date the MML law was passed. Unfortunately, the mainstream media rarely shows the public actual data like this, thus allowing unsupported generalizations to routinely be made that the public cannot check for themselves.

Keep in mind that these eleven MML states hold more than 63 million people, or 20 percent of the American population. Thus, their crime rate trends significantly influence the national rate. But the plots show apparent reversals or slowing of declining violent crime rates in each MML state at the same time as, or soon after, legalization.

These results aren't consistent with a study claiming MML had no effect on (or even a reduction of) violent crime. Consequently, the topic deserves far more attention before drawing the type of broad conclusions portrayed in the media. More rigorous follow-up studies are necessary. One could reasonably conclude that MML may have increased violent crime in many of these states.