It appears that marijuana legalization on the ballot may actually improve youth turnout. Over at FiveThirtyEight Harry Enten disputes this conventional wisdom, but I believe his argument suffers from two important flaws.

First, he looks at the Current Population Survey (CPS) to determine the percent of the overall electorate in 2008 and 2012 that was 18-29:

The Census Bureau found youth turnout rose by 0.2 points in Colorado, dropped by 0.9 points in Oregon, and dropped by 2.7 points in Washington from 2008 to 2012, an average 1.2-point drop across all three states. This drop is pretty much the same as the 1.5-point drop in young voters nationally, as measured by the CPS.

With the data set available from CPS, this seems like the wrong information to look at if you want to know if marijuana might encourage young people to vote — it brings too many variables into play that skew the results. For example, marijuana could cause an increase in the total number of young voters but if some other unrelated factors also brought out more older voters, the percent of the electorate made up by young people would remain unchanged.

A more direct way to look at the issue would be to compare only the change in number of young adults voting in marijuana legalization states to the behavior of young people in non-legalization states. A better data point to look at in presidential-year elections would be the percentage of 18-24 year old citizens who voted in 2008 and 2012, comparing the legalization states to the national average.

2008 % of 18-24 Citizens that Voted 2012 % of 18-24 Citizens that Voted Change
Colorado 50.9 56.1 5.2
Oregon 52.7 51.1 -1.6
Washington 46.7 42.1 -4.6
National 48.5 41.2 -7.3

Thanks to the Obama campaign, youth political engagement was very high in 2008, but it really dropped off nationally in 2012. In this environment, Colorado was one of only a handful of states where according to the CPS the percent of 18-24 year olds voting actually went up in 2012, while Oregon and Washington saw much smaller decreases than the national average. (A note of causation there is a relatively large margin of error for these subset in the CPS data.)

My second major issue with Enten’s analysis is that he inappropriately includes in his data set initiatives that would make only modest reforms to drug policy:

We can also look at prior years’ recreational marijuana ballot measures, including those that sought to legalize, decriminalize or lessen the penalty for recreational marijuana. For the 14 such ballot measures since 1998, the voting pool was made up of 0.2 percentage points fewer 18- to 29-year-olds, according to the CPS, compared to the prior similar election (i.e. the prior midterm for midterm years and the prior presidential election for presidential election years).

Looking only at the midterms, the 18-to-29 demographic rose 0.1 percentage points on average.

(emphasis mine)

For example, he includes in his table the 2000 California Proposition 36, “Probation and Treatment for Drug-Related Offenses.” While it was a positive improvement, Prop. 36 was not exactly a sexy initiative and not clearly about marijuana legalization. In general, it is hard to get people very engaged over actions that are perceived to be half-measures. I know from working in this space that full marijuana legalization creates real popular excitement in a way all other changes don’t. If you remove these non-legalization measures (like I have from his chart below) and look at only the initiatives that would actually legalize marijuana for adults, the results flip.

Even using Enten’s imperfect data choice, we see that on average when full legalization is on the ballot, 18-29 year olds make up 0.5 percentage points more of the voting pool. Looking at only midterm elections the increase is 1.3 points. While that is not a huge numbers, I can almost guarantee that in close elections the Democratic party would be willing to spend millions on an outreach program if they thought it could increase the youth share of the electorate by 0.5 points.

In addition, it could be argued you should only really look at four measures, California’s Prop 19 in 2010 and the three initiatives in 2012, because it was only starting in 2010 that several independent polls indicated legalization had a good chance of winning. That can make a big difference, since it is tough to get people excited about what is perceived as a lost cause.

Like with the three 2012 initiatives, youth turnout in California in 2010 performed better than the national average. The percentage of young people who decided to vote in 2010 compared to 2006 was down nationally but up in California. While that makes for a very small sample size, it is four for four.

Looking at the data from this perspective, full legalization on the ballot may actually have a modest, positive effect on youth turnout.

Adding in some subjective analysis based on observing the campaigns, I would I suspect it is more the case that legalization initiatives have the potential to drive youth turnout, but merely getting it on the ballot is only part of it. There seems to be a need for a decently well-financed legalization campaign seriously focused on young outreach to get the maximum effect, which is what the Colorado campaign did.

Given the large number of variables involved in an election and the small sample it might not be possible to definitively answer this question yet, but fortunately we will get some more data soon. We should likely seen around half a dozen full legalization initiatives on ballot in the next two election cycles.

Jon Walker is the author of After Legalization: Understanding the future of marijuana policy