Bryan Caplan  

Public Opinion on Marijuana Legalization: What the Data Say

Group-Serving Bias: Bloodla... Scott Sumner Explains Himself...
I'm all in favor of speculation on questions where the data are unclear or non-existent.  But if excellent data exists on a question of interest, we should consult it.  The General Social Survey has been asking Americans "Do you think the use of marijuana should be made legal or not?" for decades.  (Variable identifier GRASS).  Tyler puts great emphasis on the parent/non-parent divide, and Megan McArdle backs him up.  Is the GSS on their side?

They definitely get the sign right.  Here are the results from a simple regression of opposition to legalization on year, age, education, and parental status:
All else equal, people with kids are 8 percentage-points more opposed to legalization than people without kids.  The effect is comparable to 27 years of trend cultural change, 27 years of age, or 6 years of education.

But throwing in more variables sheers the effect of parental status - and gives us better yardsticks for comparison.  Here's what happens if you add controls for ideology, church attendance, and gender:

Notice: Adding a few controls roughly halves the effect of having kids.  The gap between parents and non-parents is smaller than the maternalistic gap between women and men.  Parenthood matters less than a single step on the GSS's 7-point ideology scale, and much less than two steps on its 8-point church attendance scale.

Bottom line: Parents are more opposed to legalization, but there's no reason to put great emphasis on it.  Religion, ideology, and education are far more important cleavages.

P.S. I must report with smug satisfaction that most of the apparent effect of education is a disguised effect of IQ.  Intelligence once again makes people think like economists.

Comments and Sharing

COMMENTS (12 to date)
Bman writes:

> The gap between parents and non-parents is smaller than the
> maternalistic gap between women and men.

Probably not. The difference is (.042 - .050) = -.008. Without X'X one cannot be sure this is not statistically significant, but, given that the SE of each coefficient is .006 and .005 respectively, it seems pretty unlikely.

Comparing the .080 from the first regression and the .042 from the second is not so straightforward (though it can be done) and doing so would likely substantially temper that it "halves the effect".

Nonetheless, very interesting post and great to consult the data and when it is exists and gives a strikingly clear result.

Hyena writes:

Perhaps your arrow is flipped. Name me an economist with a low IQ.

Tyler Cowen writes:

Here it's the absolute effect which matters, not the adjusted.

agnostic writes:

Here's one more variable thrown in that shows the greater impact of demographic than ideological changes, the cluelessness of many libertarians about how certain of their policies would destroy certain others, and why the failure of legalization in California shouldn't be so surprising (except to east coasters who have no idea what California is like).

Keep the variables already used, then throw in Hispanic status (HISPANIC, recoded as non [1] or Hispanic [2-50]). Here are the estimates for B and SE(B):

HISPANIC , 0.095 , 0.019
YEAR , -0.006 , 0.002
AGE , 0.002 , 0
EDUC , -0.014 , 0.002
KIDS , 0.024 , 0.014
POLVIEWS , 0.052 , 0.004
ATTEND , 0.037 , 0.002
SEX , 0.056 , 0.012

It's not shown, but now the T-statistic for the KIDS variable is no longer significant (0.088, surely wouldn't pass after correction for multiple hypotheses). The others still hold up as significant.

The most powerful factor by far is whether or not someone is Hispanic, even with these other things being equal. It's just under twice as powerful as political views and sex, and just under three times as powerful as church attendance. About four times as powerful as parental status.

Unadjusted, non-Hispanics are 36.1% in favor of legalization, Hispanics 27.1% in favor -- a gigantic split, greater than the one between men and women. It doesn't generalize to racial differences, as blacks and whites are equally in favor.

So the open borders policy that libertarians love will quickly dissolve support for legalization, and most other related policies.

This also shows why its defeat in California isn't too surprising, given how different the make-up of the population is from, say, 20 years ago when it was still in the popular imagination. Back then it was possible to view it as full of European social liberals playing Mahler or R.E.M from their Volvos. Now it's more like pickup trucks full of Mexicans blasting ranchero music.

Patrick writes:


While a number of people have been trying to warn Bryan about the negative externalities associated with open borders, I think you're the first to make the point this well, and with hard data. Good job.

I hope Bryan addresses this. I want to believe he understands something I don't because I don't like the idea of sacrificing one libertarian ideal to preserve others. But it's better to try to cope with an undesirable tradeoff than make sub-optimal choices out of delusional thinking.

Evan writes:
So the open borders policy that libertarians love will quickly dissolve support for legalization, and most other related policies.

Almost, but not quite. It depends on what you mean by open borders. An open citizenship program might have that effect. But an open guest worker program would have no such effect, since guest workers can't vote.

So Patrick, if agnostic is right, we don't really need to sacrifice the ideal of open borders. We just need to make sure that we let in an unlimited amount of guest workers, not an unlimited amount citizens.

That being said, I'm not sure agnostic's data is exactly a silver bullet. From what I understand, Hispanics are one of the least politically active groups in the country (unfortunately I heard this on the radio and haven't yet found a webpage with this data to link to). When you measure the damage caused by Hispanic support of liberty-limiting programs, you have to factor in how little they vote.

If this is true, one of the worst things for liberty conservative politicians could do is antagonize hispanics, since it might spur a great amount of them into political action.

Lastly, agnostic, you should avoid things like this passage:

Back then it was possible to view it as full of European social liberals playing Mahler or R.E.M from their Volvos. Now it's more like pickup trucks full of Mexicans blasting ranchero music.

This passage weakens your message considerably. It makes it sound like you find these Mexicans aesthetically unappealing, and that that's the real reason you oppose open borders, rather than any worries about politics.

Patrick writes:


Guest workers can have children. They are citizens. They can vote. The troubling thing is that second and third generation descendants of immigrants have a tendency to de-assimilate. A particularly striking case is Nidal Hasan (not Hispanic, but the aggrieved child of immigrants). Diversity is a tough nut to crack.

Also, Harry Reid just won reelection by mobilizing the Hispanic vote. Not libertarianism's finest hour. Their political potency is now a fact of life.

wcudon writes:

This is a controversial issue. There have been many studies about marijuana and its effect on performance standards. I believe that despite these studies if people want to smoke this substance they will. Unfortunately, many are caught with this illegal substance and have to face the consequences. However, this does not deter people from engaging in the act of "smoking weed". Personally, I feel that if people want to smoke or engage in this activity they should benefit the community while doing so. I believe that marijuana should be taxed like alcohol and tobacco and that if it is we will be able to use profits to help with our budget deficits, which will in turn push our economy higher on the recovery period. While I am saying we can benefit I am not sying I approve of people walking around high all the time but I do feel that if there is some kind of regulation like taxes that the population could benefit from risky behavior.

RPLong writes:

The problem with the marijuana issue is that consuming marijuana is an objectively stupid thing to do - but so is making it illegal.

So really it's a question of "which kind of stupid are you prepared to live with?"

In general, libertarians have a good handle on that question (as it applies to many situations, in fact), but the general population is still a long way away from being able to understand that not everything that is objectively stupid should be federally prohibited.

As more and more Americans become marijuana consumers, I don't expect this divide to narrow much, either. But this is just one means by which we're all paralyzing our abilities to reason...

Dr. T writes:
Intelligence once again makes people think like economists.
Given the way some economists think, that may not be a compliment. It would be better to say that intelligence tends to make people think like rational economists.
Big Ric writes:

It is obvious that in the recent years the controversy of the legalization of marijuana is not only a political problem but also a social and economic problem. We see a lot of graphs who is for the legalization and who is against it based on gender, age, ethnicity, parenthood, education, and political views. I see this as being in a category that I like to call "splitters". I think this scientific term can be applied here because we must not focus on who this could potentially hurt and who this could potentially help. The government needs to focus on how this will affect the entire population.

There have been studies that have shown marijuana to be harmful not only to oneself but also to the general population. However a lot of people say, "How is it any different from smoking or drinking?" They are both harmful to oneself and the general public. To that I would have to say, "Well aren't two harmful things enough." If we legalize marijuana, where will it go next? Why not just legalize cocaine, heroin, and other drugs? We can gain the same economic value from taxing these like alcohol and tobacco. I guess the point here is that we need to draw the line at marijuana so that is does not go any farther.

Susana writes:

this dearth connected with useful scientific trial details -- randomized, handled, double-blind assessments involving big patient populations -- will be the biggest hindrance facing marijuana's legitimacy. While quite a few smaller tests were performed of which support each of those sides belonging to the health discussion, political in addition to economic benefits will form the main impetus for any legalization of marijuana. for just a state want California, whose budget deficit concentrations to quantities of dollars, the possible tax earnings from legalized pot is a lot of dough that is now going away in smoke a cigarette.

Comments for this entry have been closed
Return to top