[Originally to be titled “Marijuana: I Was Wrong”, but looking back I was suitably careful about everything, and my reward is not having to say that.]
Five years ago, I reviewed the potential costs and benefits of marijuana legalization and concluded that there wasn’t enough evidence for a firm conclusion. I found that using some made-up math, the effects looked slightly positive, but this was very sensitive to small changes in how made-up the math was.
The only really interesting conclusion was that most of the objective costs or benefits of legalization came from road traffic accidents. Either stoned driving would increase such accidents, killing thousands. Or people using marijuana instead of alcohol would decrease those accidents, saving thousands. I concluded:
We should probably stop [emphasizing direct] health effects of marijuana and imprisonment for marijuana-related offenses, and concentrate all of our research and political energy on how marijuana affects driving.
Using the best evidence available at the time, I predicted that marijuana legalization would probably decrease road traffic accidents. Now several states have legalized marijuana, data are in, and we have some preliminary evidence on how marijuana affects driving. And I was wrong.
A study by the Highway Loss Data Institute in June of last year finds that states that legalized marijuana saw insurance claims for auto accidents increase about 3% over the general national trend for the time. An updated study by the same group finds 6% according to insurance claims, and 5.2% according to police reports.
These are usually contrasted with a 2017 study that finds legalization states did not have significantly increased rates of car accident fatality. What is going on?
This study finds a (non-significant) increase of 2.7%. This very nicely matches the non-fatal collision study covering the same period, which finds a 3% increase in total collisions. But because there’s a lower sample size of fatal collisions compared to total collision, the fatality result fails to reach significance. Probably the reason these two results are lower than the 5.2% – 6% result is because the 5.2% – 6% result is newer, and marijuana sales have been increasing every year after legalization.
If this interpretation is true, we should expect that a mature legal marijuana industry causes about a 5% increase in car crashes and fatalities. Score one point for “obvious things” in its fight with “clever attempts to draw counterintuitive conclusions because of substitution effects”.
In the current set of nine states with legalization, the 5% increase would amount to an extra 300 deaths per year. If the country as a whole legalized, that would make about 1800 extra deaths per year. Using my totally made-up math model from the previous post, this is enough to shift the net effect of marijuana legalization from positive to slightly negative. This is especially true if the alternative to legalization is decriminalization, which has many of the benefits of legalization but fewer costs.
But again, given how weak the math here is and how dependent it is on a lot of assumptions, this probably shouldn’t taken too seriously. Tomorrow we could find out that I interpreted the fatality study wrong and marijuana really does cause uniquely non-fatal car accidents. Or that we should be ignoring all of this and paying attention to the effects on chronic pain. Or that marijuana causes cancer. Wait, no, that one was last week. Screw it.
People pointed out on the original thread that all this quantification of the objective harms and benefits of marijuana left out something important: a lot of people like it. Fair. This is hard to think about, but here are some things that help guide my intuitions:
1. Marijuana still is definitely not as bad as alcohol or smoking, which aren’t banned
2. Marijuana still is probably worse than SSRIs, which are banned without a prescription (though it’s hard to go to jail for having them; consider them “decriminalized”). Don’t tell me this a fake comparison; they’re both psychoactive drugs that purport to make you calmer and happier.
3. About two thirds of drunk driving deaths are the drunk driver themselves, and stoned driving is probably the same way. We might choose to focus only on the one-third of fatalities that happen to bystanders if we believe people should be allowed to make bad choices that only hurt themselves.
4. Everyone expects the marijuana market to keep expanding in the states where it already exists, so these numbers may increase.
5. Marijuana taxes, spent intelligently, could easily save more lives than these accidents cost.
6. Marijuana taxes won’t be spent intelligently
7. If marijuana really does increase cancer risk by a few percent, that could easily outweigh everything else and make it a giant public health disaster.
8. But bacon also increases cancer risk by a few percent, is already a giant public health disaster, and we don’t worry about it that much.
9. If the above calculations are true, preventing national legalization of marijuana would save half as many lives as successfully implementing Australia-style gun control in the US.
There wasn’t meant to be a conclusion to all of these: they help guide my intuition, but in so many different directions that I still don’t have a real position.
Min-maxing isn’t how we do laws; the rule of thumb seems to be that unless Death could replace its scythe with X, we keep X legal.
???
not really.
Lots of things that kill some people but aren’t as deadly as getting slashed with a scythe are banned or heavily restricted in many countries.
I would argue that there is a far larger number of things do kill people, but are barely regulated or not regulated at all.
When trying to decide whether to make something that kills peole illegal we never do the min-maxing analysis that Scott Alexander does here, we mostly look if the body count is so high that it would be a hassle to keep it legal.
In fact, scythes themselves are completely legal and can be purchased on Amazon:
https://www.amazon.com/Zenport-K310-Clearing-Sickle-Aluminum/dp/B01FJMKND2/ref=sr_1_19?ie=UTF8&qid=1541615670&sr=8-19&keywords=sythe
That’s really more of a sickle.
Scythes have long handles, in order to reach close to the ground.
I’m reasonably certain something here is going over my head, but to be sure: I was referring to this fella here and the tool of his trade.
Rings true to me – damn near had my head taken off by an unlicensed hair braider in Utah once.
I’m glad the piles of corpses from plastic straws on San Francisco streets are now just a fading memory!
But the ones in the ocean aren’t.
Is it coincidance that San Francisco’s ban on plastic straws happens to coincide with growing awareness of what plastic does to wildlife?
Those straws aren’t coming from San Francisco so it’s really not relevant.
What turned the gaze of San Francisco’s authorities on plastic straws is relevant and I’m betting it was the corpses.
Whether it makes sense or not, San Fran is at least eating the inconvenience of banning plastic straws themselves. They put their money where their, er, mouths are. Good for them, even if it turns out to be incorrect.
A) First! EDIT: Nope, second actually; still cool though.
B) Points for being as perplexed on this as the evidence demands. Nobody else seems to be, but hey, that’s life in 2018 America, I guess?
I see this “First! Second! Third! First!” stuff on other blogs and I have to say I find it really irritating. This isn’t the first time I’ve seen it on SSC recently, and I really hope it doesn’t catch on.
A particular person pretty much only ever does it once.
It’s stupid and sophomoric, which is why it’s fun, but only the once then it’s out of your system.
Isn’t reducing crime the most important benefit of legalising marijuana? The violence and tax-evasion associated with gangs and, of course, the imprisoning people for what should be a very minor transgression (smoking pot).
That would be covered by decriminalization, and would not require making it legal to farm and publicly sell. Scott’s approach is specifically about options that go beyond decriminalization.
De-crim usually only covers the end consumer, black/grey market sales personnel still faces jail time and there could still be crime and possible groups engaging in turf wars even with de-crim.
It takes a big chunk of people out of the criminalization equation, but not everyone. The dutch have had that problem for decades because what they have is “extreme de-crim”. You still cannot legally grow to supply the coffee shops, hence the business is still in the hands of shady people and the govt sees no taxes from them.
https://www.talkingdrugs.org/netherlands-paradox-cannabis-policy-front-door
That’s fair. I tend to think of the vast majority of police interactions around illegal drugs are for minor infractions, with relatively few (in terms of number of interactions and people directly affected) from dealers. The fact that the dealers and producers have a much larger impact per capita is more relevant than I was giving credit.
Evidence suggest that this does not happen:
https://sites.northwestern.edu/hxl642/
It appears that marijuana legalization has lead to an increase rate of crime committed by marijuana dealers as they have had to go into new criminal ventures.
At best we can hope time sorts it out, but so far the empirical evidence is that drug dealers commit more crimes post legalization than prior.
I would expect that as a short term effect that desperate drug dealers might do some other crime instead but that a long term effect is that fewer people become drug dealers in the first place if there is less money and fewer “jobs” in the field. Same for organized crime; they may become more violent short term but reduced resources should shrink them long term. So for both reason I would expect a short term increase in crime but long-term decrease, with the long term effect being more significant.
I would think the many effects related to the criminal justice system would be orders of magnitude more important than the medical/accident effects mentioned in the post. I’m also surprised that none of that was mentioned here, although I suppose Scott is coming from a framework in which he’s likely to consider medical issues first.
Sending people to jail for smoking marijuana is just plain wrong. It’s still wrong even if there’s a net benefit to society in terms if cancer, car crash injuries/fatalities etc.
Don’t send people to jail, but don’t make it super duper easy to get, and don’t make it so it’s actively promoted by people getting rich off of it. Decriminalising better than legalising?
I’m pretty sceptical about the tax arguments. Don’t think in terms of we can get this tax and spend it on this good. Think in terms of total tax take – will this really increase this over the long term, do we want it to be increased? If we have same total tax, is collecting some through dope tax (much) better than alternatives?
Think in terms of reducing criminal element – good. But also in terms of increasing corporate element – likely bad (?).
Think in terms of how it affects society as a whole over the long term? Realise we just don’t know. Go back to the wrongness of sending people to jail. Stop doing that while reducing the risks of decriminalistation/legalisation as much as possible. Make changes slowly.
I believe that most people imprisoned for marijuana are dealers, not users.
It’s hard to find stats on that because prosecutors love to throw every charge possible at people.
So you get people saying things like “oh those guys aren’t just in jail for possessing a small quantity of marijuana” and people then assume that it means they were kingpins running giant evil enterprises when in reality they were also charged with a thousand and 1 extra charges like resisting arrest for looking at the officer funny and assaulting the officers fist with their face.
Since dealing carrying a heavier sentence they can try to tack that on anyone if they feel like it. Which is an easy choice since they can claim that having plastic food bags in your kitchen is evidence that you were intending to sell even if you only had a couple grams of weed and that doesn’t stop them from charging you for possession as well.
haha!
That’s hard to say. In my home, the Australian state of Victoria, a “traffickable quantity” of cannabis is defined as 250g of plant material or 10 plants. My understanding is that those quantities could equally well apply to a long-term supply for personal use, or for friends-and-family type growers who are probably less of a risk to society than a large-scale commercial operation.
Decriminalization loses on of the biggest potential benefits: eliminating cartels and black markets.
Yeah but you know what would achieve better results? Criminalizing driving.
One thing that is puzzling to me is why not so few countries appear to have strict rules against drunk (and high) driving, and enforce them. From my tender youth in the late eighties and until now, drunk driving was simply Not Done. You either had a designated driver (who stayed sober), or you used public transportation, or you walked. Yet it seems that while many countries have laws against DUI, the limits seem lax (we had 0.5 ‰, now it’s 0.2 I think), and you only get caught after a serious accident (we’ve always had random police checkpoints). Why not attack and punish the combination of intoxicants and driving, if that’s what is causing the harm?
I think that different countries have wildly varying incidence of DUIs, and each adjust their laws until getting to what they consider acceptable levels of death by DUI.
It’s a cultural thing. It’s hard and unpopular to make visible the deaths by DUI and you have to start taxing alcohol, adding road controls, etc. Add that some countries have quirks in their laws and people will fight legally against road controls (like, well, America), and you will see it is an uphill battle.
The thing is there is a technical solution that would reduce this a lot, breathalizers in each car, but most people consider it too much.
So at the end of the day we accept a certain level of fatalities by this. The balance of more deaths by Pot Related Driving Deaths vs less crime is hard to make. If it was just people incarcerated it’d be one thing, but those incarcerated people have families that are effected by their incarceration, the incarceration itself will have effects in future employability and may lead to more crimes down the line.
But drug related incarceration in America is not color blind, while pot related driving deaths may be, so I could see the argument being a racial one.
Back to the question, I think it’s two reasons:
-We like driving and car privacy too much to make it acceptable to have a breathalizer in each car.
-There’s a little libertarian in most of us that says I shouldn’t be randomly stopped by the cops to prove my innocence.
Remember though,
Don’t drink and drive,
Because there are people who text and drive,
And they’ll hit you,
And you’ll get the blame…
Someone had to invent the concept of “designated driver.” Maybe Scandinavian police checkpoints caused Scandinavians to invent it, but it was able to spread to other countries without ramping up serious enforcement. American TV advertising (both designated drivers and images of death from drunk driving; also seatbelts) seems to have had a big effect on car fatality rates, reducing the American per mile levels below those of western Europe for a decade or two. And when Western Europe copied the idea, they pulled back ahead.
Maybe harsh enforcement also would have been effective, or might still be, but I don’t think that the Scandinavian experience shows much. (other than designated drivers and social pressure, but we’ve already copied those)
Probably because the threat from drunk driving is greatly exaggerated and we have other laws that actually deal with the consequences: battery and manslaughter.
This was my stance for a while as well but I think if we’re going with policies that are never going to get enacted but would actually solve the problem I would like intoxication to be viewed the same as public nudity.
Meaning it’s fine to order how ever many mind altering substances and use them at home or at specific venues that cater to that sort of thing but as soon as you leave you’re sent to jail including fining any nightclub or bar that doesn’t check on their patrons before it lets them back out the door. You’d need some other way of getting rid of them but such services already exist and having the rest of your evening cut short should actually encourage people to drink responsibly if nothing else does.
Driving high is almost as bad as driving drunk — they’re not mutually exclusive and getting high might even increase the risk of driving drunk, too.
Putting people in jail and labelling them as felons, for the rest of their lives, is too much punishment for smoking dope. Long term social benefit to having fewer criminals.
Smoking dope is likely to be directly detrimental to cognitive skills, and will have long term social costs; possibly quite high long term costs. This might well provide more clear examples, like winos do today, to encourage younger folk to be more responsible in their drinking / toking habits. Heavy use might well be more damaging and costly than either smoking or drinking.
Getting the tax cash is necessary for the politicians to support it, and relative to other tax collection, this “vice tax” is less socially bad than wealth or income or payroll taxes. Legalize & tax rather than merely decriminalize. Perhaps tax plus more regulations like registration of auto, identification as a user thus accepting of a blood test at every traffic accident / infraction.
Reducing the budget deficit, and thus borrowing by the state, is a very intelligent use of the revenue. Most increased state spending is less intelligent, tho far more politically popular.
I would venture to say that the same folks liable to drive high in an irresponsible manner are the same people liable to do so drunk. The data would be difficult to obtain save for police records and names of arrestees, but I would think that there would be strong crossover.
What effect does incarceration have on lifespan? That’s hard enough to answer, but then there’s also the effects on the children of those incarcerated.
Even if 100% of stoned driving fatalities were the stoned drivers themselves, we couldn’t dismiss these deaths on the ground that it only hurts themselves, since they have family and friends who will be negatively affected by their death, not only emotionally but often also economically. But I guess it’s something that can be said in most cases where people want to apply the rule that we should allow people to make bad choices that only hurt themselves. I’m not sure whether this observation should count for or against the point I just made.
Yes, and?
With the possible exception in case of children, one doesn’t have an enforceable emotional or economic duty to family or friends.
Well, this strikes me as very dubious, but that’s irrelevant to the point I was making. Scott was saying that, if people should be allowed to make choices that only hurt themselves, then we may have to ignore stoned drives who killed themselves and count only the other deaths from stoned driving in the moral calculation. I’m just saying that, from the fact that a stoned driver only killed himself, it doesn’t follow that he only hurt himself. Now, if you add the assumption that people have no enforceable economic/emotional duties to family and friends, then you may be able to get the original conclusion anyway. But this assumption is logically independent from the one Scott was considering.
Similar reason to why suicide was decriminalized.
If Bob has the right to hop on a plane to the far side of the globe, cut ties with everyone and leave behind all his family and friends and inflicting “harm” in the form of depriving them of his presence without any warning beyond avoiding financial fraud… then it’s hard to argue that Bobs behavior should be legally restricted in such a way as to forbid him from inflicting similar “harm” should his behavior get himself killed.
The flip side is insidious: there is no limit to how much you can claim right to restrict the lives of all around you to convenience yourself if you lay claim to a right on your behalf to their continued well-being, presence and existence.
Put another way, eating bacon kills people. It doesn’t follow that just because you’ll miss your wife if she dies that you have any right to make a law to have her thrown in jail should she eat bacon.
Yes, I actually agree with all that (well, not quite actually, but it’s irrelevant to the point I was making), but the point I was trying to make is precisely that the criterion “something is okay as long as it only hurts you” won’t do the job the people who use it want it to do, because many of the things they want to use it to say they are permissible would actually become impermissible on that criterion.
Would I be right in saying that Bob actually doesn’t have that right if he is paying child support? In general, don’t americans’ financial obligations (in particular also taxation) follow them around the world?
That’s probably true, with some qualifications, but it affects only the antecedent. If Bob has the right to disappear, it’s hard to explain why he doesn’t also have the right to suicide.
The only argument that holds any weight with me is the same one made about the death penalty — if it turns out later that it was a mistake, there are no take-backs.
that’s why I added a caveat of “beyond avoiding financial fraud”
if Bob has no intention to commit any kind of financial fraud then he’s free to disappear.
This one strikes me as a questionable assumption from the get-go. After all, a decent number of those one-fatality-drunk-driver cases are likely to be suicides, where the decreased inhibitions of the alcohol allow someone to finally push it over the line and just go for it. Marijuana for all of its intoxicating effects doesn’t to my knowledge usually end up causing people to act rashly or impulsively in the same way.
Sometimes it can, and some of those cases are almost certainly just that, but it seems more intuitive to think that the relative proportions are going to differ from those involving alcohol. So I’m not certain it’s something that could safely be assumed as a rule-of-thumb-until-proved-otherwise.
I disagree, as far as both alcohol and MJ have effects of decreasing reaction time and awareness. Alcohol may increase impulsivity and decrease inhibition, but so many traffic fatalities occur in split seconds, and any alteration of one’s awareness will increase mistakes, driving alone or not, traffic or not.
The assertion was that a significant fraction of drunk driver-only fatalities were intentional suicides, and that a smaller fraction of high-driving fatalities would be suicides.
The net effect would be that the fraction of high-driving deaths per accident would be lower, but the number of non-driver fatalities per accident would be the same for high-driving and drunk-driving.
It seems to me that in considering whether or not marijuana should be sold at corner delis we should be thinking mainly about questions such as: (1) Is weekly use of marijuana more or less likely than weekly use of beer to become daily use? (2) Is daily after-work use of marijuana more or less likely than daily after-work use of beer to become throughout-the-day use? (3) Is throughout-the-day use of marijuana more or less likely than after-work use of beer to be compatible with keeping your job, maintaining friendships, and enjoying hobbies? (4) is throughout-the-day use of marijuana more or less likely than throughout-the-day use of beer to be compatible with keeping your job, maintaining friendships, and enjoying hobbies?
If the answers to (1) and (2) are “Yes”, then (3) should be the focus of attention. But note that the answer to (4) isn’t as obvious as it might seem to those who picture someone drinking beer after beer throughout the day as compared to someone using marijuana upon awakening, then at lunchtime, then after work: the picture should rather be of someone’s drinking one beer with breakfast, one with lunch, and a couple more in the evening.
And the assumption throughout should be that if marijuana is sold at corner delis it will be used much more frequently by many more people than is presently the case.
(I’ve always felt that a marijuana-high is a lot weirder, a lot more alienating, than an alcohol-buzz is, but then again marijuana has always had a much more intense effect on me then it seems to have on most people. (Many years have passed since my last use of marijuana.))
Those are great arguments to use to ban beer, not so great to legalize pot.
Smoking pot all day every day is probably not a good life choice, but it is not very comparable to drinking all day. For one, marijuana is certainly habit-forming and probably a little bit addictive, but nothing like booze or nicotine. For another, people who smoke 24/7 are not really “high” the way you would think if you’ve only smoked casually; regular heavy use feels different, almost to the point of seeming like a different drug. In particular I’m much more sanguine about riding in the passenger seat of someone who’s been high all week than someone who’s recently gotten high for the first time in days.
However, I think you’re missing the main point about drug use in general, which is that it’s a mistake to get hung up on the question “Should people use drugs?” as we have very limited ability to affect that. What we can affect is whether they get arrested for doing drugs, and if your heuristic is that they should not get arrested for pot if pot is clearly and obviously less harmful than alcohol, then I think the jury is pretty much back on that one.
Well, it’s a valid and worthwhile question (as long as you break it down for each specific drug rather than pretending that drugs are a homogenous natural category) … it’s just a mistake to confuse “Should people use drugs?” with “Should we use the criminal justice system to punish people who use drugs?” … and a lot of people seem to have difficulty understanding that those are two wildly different questions.
Is it really completely implausible that the increase in non-fatal accidents due to marijuana is greater than the increase in fatal accidents due to marijuana? I seem to recall some research that drunk drivers tend to drive faster, while stoned drivers tend to drive more slowly, which would suggest that more of the latter might be expected to produce an increase in non-fatal accidents specifically.
This seems right to me. Also, I’m pretty sure that the ratio of fatal to non-fatal car accidents is going down due to better safety equipment in cars and better medical tech in hospitals. If fatalities are actually increasing in legalization states despite this, we’re probably underestimating legalization’s effect if we only blame it for the net increase.
And isn’t supposed to be good to relax when an accident is immanent? I could see different fatality results being reasonable if the physiological results are opposite.
Caveat: I’m not sure the physiological results are opposite. Is alcohol a stimulant or relaxant? Seems like it is subject and dose dependent.
I’ve only heard the claim that one should be relaxed in an accident in the context that drunk drivers are relaxed.
Being relaxed via chemicals often means that your reaction time is slowed, rather than being relaxed and also very focused, as in driving a racecar for example. Neither THC dominant MJ or Alcohol are beneficial for focus, CBD is different as so much as ive read. I dont use MJ and drink rarely.
I certainly don’t expect either to make you a better driver. Just trying to make sense of a statistical increase in accidents without an equivalent increase in fatalities.
But this post isn’t ignoring fatal accidents. It claims that legalization increased accidents by a statistically significant 3% and increased fatalities by a statistically insignificant 2.7%. It’s too soon to tell, but it’s not looking good for that hypothesis. (I don’t know why there isn’t a fatality number to go along with the later 5-6% number.)
———
I’m surprised that Scott didn’t mention the claim that stoned drivers drive slow in his previous post, because without it, it seems to me like a poor argument. If marijuana substitutes for alcohol, that doesn’t do you any good unless stoned drivers are safer. And without the causal mechanism, one should be skeptical of the end-to-end studies he cited.
>Marijuana still is probably worse than SSRIs, which are banned without a prescription. Don’t tell me this a fake comparison; they’re both psychoactive drugs that purport to make you calmer and happier.
Here’s a question that struck me while considering this: if some random generally-well-tolerated SSRIs weren’t highly marketed and incentivized drugs but rather relatively-recent research chemicals with grey legality, would they become popular on their own merits? (Drugs do not need to have obvious flashy effects to become popular RCs, though it helps. Exhibit A: kratom.)
I feel like it’s plausible, but they’d be conceptualized in a very different way to how they currently are. I don’t think your comparison is false at all, and it’s worth considering, though probably only by people who have an outside-the-Overton-window perspective on SSRIs* to avoid all the normies making the conversation way more boring.
* anything that can be described as the ‘fifty Stalins’ of this topic counts as ‘inside the Overton window’
—
>Tomorrow we could find out that I interpreted the fatality study wrong and marijuana really does cause uniquely non-fatal car accidents.
I agree with Protagoras that this, or a retreat position from this, is plausible.
Some possible confounders
1. Perhaps the reported rate of car accidents has increased rather than the actual rate. People have less reason to be paranoid about reporting certain accidents now. Possible check: Did multi car accidents increase or stay flat? I’m not going to pass on being rear ended because you have a joint in your glove box and don’t want the Fuz involved, if multi car accidents stayed the same or had a much smaller increase that would be a sign pointing to few or no extra crashes and just higher reporting.
2. Its a short term blip. Iirc in Portugal use after decriminalization went up in the short term and then went down in the long term. Will have to wait and see on that one.
You bring up a good point in your #1. I think this is a bigger factor than may be supposed.
Kudos to Scott for being open about this, but putting the ‘I was wrong posts’ in the US election news desert almost feels like a coverup, ala zuckerburg
So, in essence there are more fender-benders but no evidence of more fatalities.
I wonder if this supposed few percent increase of “cannabis related collisions” happened in similar numbers before, but have now just found a new label because there is a higher awareness that a driver involved MIGHT be stoned (hence more testing by the police, resulting in more “uncovered” cases with a cannabis connection) – in other words a statistical zaro-sum game. We all know drugged driving is a “control offense”, if you want to lower the statistical numbers of intoxicated driving you can just lower the number of drunk driving checkpoints or drug tests ordered, voila, less intoxicated driving in the statistics.
The important claim here isn’t that “cannabis related collisions” went up. It’s that “insurance claims for auto accidents” went up after legalization.
Actuary here. My boss e-mailed the HLDI study to me last year when the first one came out, and again this year with the update. I’m extremely skeptical of it.
-It uses Collision coverage claim frequencies. It should have also looked at liability Property Damage and Bodily Injury. Property damage is something like a mirror image of Collision (it covers the other guy’s auto if you’re at fault in the accident, whereas Collision will cover damage to your own car even if you are at fault). I would expect that they’d get similar findings if they did the same analysis on PD or BI, and if they didn’t I’d distrust their statistical methods (which are basically trend-fitting). These other coverages are conspicuously missing from the analysis.
-It looks like only about 7% of the adult population moved from being non-users to cannabis users after legalization (past-month users, anyway). See this document, second table on past month use .
Assuming this is right (if there is a better source for this, I haven’t found it, so please share), that means 7% of the adult population is driving total accident frequency up by 3% (the effect found in the last year’s study). In other words, *this* population’s accident frequency is up by 3%/7% = 43%. That seems like an implausibly large effect for “past month cannabis use on driving”, unless we assume that all new smokers are high all the time. Note that the individual state effects are all larger than 3% effect for the combined states, 14% for Colorado according to last year’s study. (Simpson’s paradox at work?) If the total number of new cannabis users is relatively small, this makes large effects look pretty implausible.
-I’m not really buying their “control” states. Are Idaho, Montana, Nevada, Utah, and Wyoming really like Colorado, Oregon and Washington? HLDI is basically saying that these states should behave the same (as in, they should have the same accident trends) and the only difference is marijuana legalization. I think these states are culturally different enough that their trends will probably diverge for other reasons. (The later report acknowledges that it’s hard to control for differences between states, and that their analysis did the best they could.)
-It would have been nice to see some kind of back-testing to see if their statistical methodology works. As in, if you did the same regressions from ten years ago, well prior to legalization, would Colorado *really* have the same trend as its control states? It seems like this would have been good “statistical hygiene.”
-It’s important to note that they didn’t find “accident frequencies rose 5% after legalization”. What they found was “Our statistical model said that accident frequency should have risen by x%, but it actually rose by x plus 5 % after legalization.” Note that in the later report, Colorado’s accident frequency is actually *down*, just not down as much as the statistical model says it should be.
I e-mailed my boss to tell him I had reservations about the study, and he heard me out and agreed that these were good reasons to be skeptical.
Good comment, I was hoping for a quick overview of the driving study. Did they control for the age characteristics of the states? I would guess Colorado, Oregon, and Washington are younger than Utah, Wyoming and Idaho, which might affect the expected trends.
They tried. From the link in Scott’s post:
“The population proportion of young drivers (20–24) and male drivers by state and year
were included in the regression models to account for variation in state demographics…”
But even they concede:
” Although the regression models included several variables intended to control for state differences, the three study states (and their controls) differ in ways that are difficult to quantify.”
Good on them for admitting this point.
Is this so implausible? Let’s say drivers get in auto accidents once per 10 years (I don’t know what is the real rate, I guess that depends on what is an ‘accident’). With 43% higher rate, they get in an accident once per 7 years. I think an individual driver could well not even notice such an increase in expected accident rate. My guess would be that any person’s daily variation in, say alertness or stress levels would cause much bigger differences in accident rates.
Another thing is that maybe legalization also increased the cannabis use of the already-users, so the effect does not need to be entirely because of the new users.
I guess “implausible” is subjective, but it’s certainly a bigger effect size than I have ever seen anyone claim. Especially if we’re just talking about someone who used *at some time* in the past month, not someone who’s actively high. A 43% increase in accident frequency is actually a big deal. As an actuary who does predictive modeling, I’d love to have access to a variable that identifies these poor drivers, if it were a real effect.
Sorry, what was your point about the individual driver not noticing? I agree that they wouldn’t notice. (Typical claims frequencies for Collision are in the 3-5% per car-year range, so even in a deterministic world it would take a long time to catch on.)
Intensity of use by current users might also go up, but we’re still dealing with a small fraction of the population and a large (implied) effect size per user.
Very good comment. I like the point about coverage mirroring and about control states a lot.
“It’s important to note that they didn’t find “accident frequencies rose 5% after legalization”. What they found was “Our statistical model said that accident frequency should have risen by x%, but it actually rose by x plus 5 % after legalization.” Note that in the later report, Colorado’s accident frequency is actually *down*, just not down as much as the statistical model says it should be.”
This pretty much says it all.
Is cannabis smoke less carcinogenic than tobacco smoke, or do people tend to smoke less cannabis than tobacco even when their availability is similar?
A quick googling seems to suggest that legal cannabis is still much more expensive than tobacco, is price the limiting factor in consumption? Is the price going to plummet as the market expands, or is there any intrinsic reason that makes cannabis more expensive to produce than tobacco?
> Is cannabis smoke less carcinogenic than tobacco smoke, or do people tend to smoke less cannabis than tobacco even when their availability is similar?
Smoking is bad not because of anything specific to tobacco, but because of the inhalation of smoldering organic compounds. It doesn’t matter what dried plant you smoke: tobacco, marijuana, wood chips, pine needles. Pound for pound it all pretty much has the same health impact. What makes tobacco uniquely dangerous isn’t the fact that smoking it is bad, it’s that it’s much better than nearly any other plant at getting large number of people to smoke it.
By comparison, cannabis users smoke far lower quantities than tobacco users. One cigarette contains approximately one gram of dried plant material. Therefore a pack a day user smokes 140 grams a month. By comparison, modern cannabis strains have very high THC combinations, and only small quantities are needed to get high.
A very heavy cannabis user maybe will smoke a quarter ounce of high quality cannabis a week. That’s 7 grams. Or in other words 95% less than the pack a day smoker. About the equivalent of one cigarette a day. And that’s a heavy user. Anecdotally, I’m a casual user, and maybe might go through an eight an ounce a month of dispensary weed. That’s the equivalent of smoking less than one cigarette a week.
I thought that this could make a case for selecting/engineering tobacco to be higher in nicotine in order to reduce consumption, but then I looked up the pharmacokinetics on Wikipedia and it says that nicotine is metabolized much faster than THC (although cotinine, a less potent active metabolite of nicotine has a half-life comparable to THC), so it won’t work.
Maybe engineering tobacco to contain a high amount of cotinine but little or no nicotine could do the trick.
Its not that simple. Nicotine alone has been shown to be non addictive in certain studies, so there are implications that it is other chemicals in Tobacco smoke that add/ create the addictive potential, esp. in terms of withdrawal and perhaps anti-depressant effect.
Isn’t this inconsistent with the observed success rate of nicotine replacement therapy for smoking cessation?
I would quibble with a lot of this (for example, pot smokers smoke less in quantity, but they also hold the smoke in a lot longer; also, a very heavy pot users is probably going through more like a quarter a day than a week) but it’s worth pointing out that the advent of vaporizers has made this more or less moot, in places where pot is legal. In fact, other than not getting arrested, this is probably the biggest upside of legalization, the fact that it allows pot users to switch to vaporizers and edibles, which have very near zero adverse health impact (other than the possible health impact of being high at all).
What about smoking marijuana vs eating it? Is the reason smoking is more common that it takes less to get you high? If so, then a decrease in price would result in some shift from smoking to eating, which ought to reduce the health effects.
It’s also easier to control and maintain how high you get. Getting too high is unpleasant, and not getting high enough is undesirable.
Eating has a long lag time, and generally large doses taken at a single time. There are also potential issues with unknown strength and other things can change the effects, such as emotional state and other foods eaten.
People get a much more controllable high by smoking. It hits faster and each puff is a relatively small dose. People can smoke small amounts at intervals until they get as high as they want and then stop.
If passing around a joint in a group, each smoker to feel the effects and choose to take more or not when the joint comes around again.
I suppose vaping would provide same dosing ability of smoking with no (or less?) cancer risk.
Yes, vaping combines the “no lung cancer” upside of eating pot brownies with the “immediate feedback on dosage” upside of smoking a joint. It is IMO inarguably the best way for someone new to marijuana to experiment safely. Every story I’ve heard of someone who had a very bad experience after legalization starts the same way: “I went to the pot shop and bought some edibles, then I took one, but I didn’t feel very high, so I took another one…”
AFAIK the only health effect from eating marijuana is that you get high.
Also fat if it’s in brownies, as it traditionally is.
I understand this is because THC is liposoluble, hence edibles are made by extracting THC into vegetable oil or butter, then using it to make rather fatty foods (I suppose you could directly eat the plant matter, but maybe it tastes bad?).
Given that THC also induces hunger, putting it into fatty and sugary foods like brownies seems like the perfect recipe for obesity.
It might seem like it, but findings consistently indicate that users of marijuana tend to have lower body mass indices than nonusers.
It is much less carcinogenic than tobacco, but it’s still somewhat of a mystery why that is. As Doug points out, marijuana contains a lot of the same smoldering organic compounds as tobacco, and so it would be expected to cause cancer for similar reasons, but no study to date has been able to find even definitive evidence of a correlation between cancer and marijuana, which is trivial to do with tobacco. And I don’t buy that it’s because people smoke less of it than tobacco: there are stoners who smoke nonstop all day; that should be enough to see some sort of correlation if it really is as dangerous as tobacco.
I was under the impression that typical marijuana smoke was more carcinogenic than typical cigarette smoke, since cigarettes usually have a filter and hand-rolled joints usually don’t.
As usual with illegal drugs, though, good and unbiased studies are almost impossible to find.
Either way, the comparison with eating bacon seems WAY off.
Yeah, it’s a great mystery because there are many reasons why you would expect it to be more carcinogenic than tobacco, and yet it appears to be mostly harmless. I’m not so sure the comparison with eating bacon is way off; we don’t have good data yet, but I wouldn’t be surprised if weed is less harmful on average than bacon.
In the U.S. at least, there’s no good reason why there aren’t good and unbiased studies given that it’s been effectively legal in California since 1996 (it was only medically legal until recently, but it was trivial to get a medical prescription for any vague reason like “I sometimes have trouble sleeping”, and once you had a medical prescription there were no rules on how much you’re allowed to consume). The ability to do research is severely restricted by the Federal government unfortunately.
Doesnt this fail to account for the amount of fillers and other chemicals in your average cigarette vs joint? I think a cigarette vs cigar comparison would be quite valid here. Obviously both are carcinogenic , but Cigars have much less added compounds, along with a wrapper made of tobbacco itself, and not paper
Do you have a source for it being “mostly harmless”?
@Christianschwalbach
Cigars are less cancerous for a different reason. People smoke cigars and cigarettes differently. Much more cigarette smoke is inhaled into the lungs.
Filters don’t effect the carcinogenic risk of cigarettes, except to perhaps increase it. IIRC, they increase the draw strength needed and therefore encourage deeper lung penetration.
And the browning of the filter? That isn’t anything removed but simply an effect of exposing the filter to heat.
The idea that filtered cigarettes are healthier is a PR gimmick promoted by the cigarette companies.
Interestingly. I’ve googled and found some works regarding cigarette filters vs. increasing rates of some cancer types. Those works show that there is indeed more marketing bullsh*t involved that promote “healthier” cigarettes.
I’d say that it’s not exactly correct (because of air dilution): ventilated filter cigarettes have lower resistance to draw which results in bigger puffs.
> but no study to date has been able to find even definitive evidence of a correlation between cancer and marijuana
It should be said that the absence of evidence is not the evidence of absence.
There are three possible explanations for why it’s difficult to find statistically signifiant correlation.
1) The heaviest marijuana users still smoke significantly less than even moderate cigarette smokers. One gram a day is pretty much the upper limit for which you can find a population of any decent size. That’s equivalent to 1/20 of a pack a day smoker. Therefore you’d need 20 times the statistical power to achieve the same statistical significance as you would with a study of pack a day smokers.
2) There are a lot of heavy cannabis users, but not that many consistently heavy cannabis users. People vary their consumption quite elastically given work, family and social circumstances. College age potheads get a job or have a kid than cut back or quit altogether. This happens much less frequently with tobacco because 1) it’s much more addictive, and 2) it’s much more functional.
For an accurate longitudinal study on mortality you need a user base that keeps its usage consistent over years. Ballpark, I’d estimate that this probably increases the statistical power by a factor of at least 4 compared to equivalent cigarette studies.
3) Cannabis users skew much younger than cigarette smokers. The reality is cancer and mortality cluster among the elderly. And there aren’t that many 70 year old stoners. Marijuana could have the same hazard ratio impact on all-cause mortality, but you’re going to have to observe a hell of a lot more young people to have a statistically significant number of deaths. 40 year olds have about 90% lower mortality than 70 year olds. So you’ll need approximately 10 times as much statistical power.
So marijuana smoking can have the same all-cause mortality impact as tobacco smoking, but still not show up on anything other than gigantic studies. Based on the three effects above, you probably need 800 times the statistical power. That means a sample size 28 times larger than pack-a-day smoker studies.
EDIT: Math
+1
Those are good points.
Probably not many, though? A very large fraction of tobacco smokers smoke 10-20 cigarettes a day, so the effect on e.g. lung cancer becomes immediately noticeable.
Also, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1277837/:
One reason is that marijuana contains a lot less isotopes. Tobacco contains significant traces of radioactive polonium, lead, and a few other elements, which builds up in the lungs and (like anything radioactive) emits tiny rays that occasionally hit a cell the wrong way and turn it into cancer. I don’t think it’s know how much of tobacco-related cancer is caused by these isotopes (as compared to tobacco-related cancer caused by the other crap in tobacco) but the fact that potheads get less lung cancer than one might expect could be relevant.
I heard somewhere that hand-rolled cigarettes are less carcinogenic because they don’t have all the additives that commercial cigarettes do. If that’s true, then it could partly explain why hand-rolled marijuana is also less carcinogenic.
The nicotine content of cigarettes vs. plain tobacco leaves may also play a role in that. The latter have more nicotine so you need less hand-rolled cigarettes to achieve the same effects.
From my casual observation (mostly talking to my acquaintances), vaping and/or edibles seem to be more popular than smoking. These options were pretty much unavailable when it was still illegal. This should have a large impact on whatever cancer risk there is.
No, or at least not by much. To a first approximation, breathing in any kind of smoke is really bad for you. Woodsmoke is about as bad for you as cigarette smoke on a concentration basis, for example (you’re just not deliberately breathing it directly into your lungs).
Yes. It doesn’t take a lot of cannabis to get you high, and a cannabis high lasts a lot longer than nicotine’s effects (a couple hours vs. a few minutes).
My big concern (which doesn’t seem to be addressed) is the rate at which marijuana use trades off against alcohol use. Alcohol is far more addictive than marijuana and sometimes causes violent and antisocial behavior, which AFAIK marijuana does not. Most of the harms of marijuana are also harms from alcohol. If marijuana legalization causes a decrease in alcohol use, then that might be enough to flip the sign over to “positive”.
I support high alcohol taxes, but am troubled by forbidding people who prefer to get blackout drunk from getting blackout drunk. If those people would find excessive marijuana use an acceptable substitute, then IMO high alcohol taxes and low to moderate marijuana taxes are a reasonable compromise that protects public health while respecting other people’s preferences.
In general, I don’t support criminalizing things unless there’s a good reason to believe that they should be illegal, and I’m uncertain how large a role pro-marijuana activism has played in reducing the number of people who are in prison for marijuana use. But certainly if the benefits are equivocal then one’s activism energy is probably best directed somewhere else.
Is there any substitution effect between alcohol and marijuana, or do they actually promote each other consumption? Anecdotally, I’ve noticed that people who like to get stoned also tend to like to get drunk, often at the same time.
People who like to use drugs will, in general, use more drugs. That doesn’t actually provide any information about whether there’s a substitution effect or a complementary effect. For example, if a person likes drugs and prefers to spend every weekend intoxicated and has access to both marijuana and alcohol, they may spend a third of their weekends drunk, a third of their weekends high, and a third of their weekends both drunk and high. Conversely, if they only have access to alcohol, they will be drunk on 100% of weekends. Giving this person access to marijuana has reduced their alcohol consumption, even though the outside observer would see that they still enjoy getting drunk quite a lot.
The complementary effect model matches my anecdotal observations. Living in San Francisco, I have access to a mind-boggling array of drugs. I hardly ever use caffeine pills (for example) because they make me jittery and don’t provide the calm euphoric focus associated with modafinil, but probably if I only had access to caffeine pills I would use them quite a lot.
I can see two situations where marijuana and alcohol are complements. First, if being both drunk and high is significantly better on the axes on which people make their drug-using decisions, such that people prefer to be drunk and high on occasions when they would otherwise prefer to be sober, then access to marijuana may increase alcohol consumption. I don’t think that’s true. Second, there may be a “gateway drug” effect, where people start with marijuana, which they consider to be safe, and that gets them into drug-using social groups which cause them to drink. I don’t expect a marijuana-to-alcohol gateway drug effect to be significant right now, although it might be significant in a world where marijuana is the default intoxicant.
My primary concern about a “marijuana displaces alcohol” model is that people who consume marijuana may be different from people who consume alcohol, or people may consume marijuana may consume it on different occasions than they consume alcohol, so that legalizing marijuana causes an increase in overall drug use and little effect on alcohol use. But I’d still support legalizing marijuana because it is relatively nonaddictive and we should in general allow people to decide for themselves what to put in their own bodies.
There could be an alcohol-to-marijuana gateway drug effect, since alcohol makes people more impulsive, therefore I suppose it makes them more likely to try other drugs, and if marijuana is highly available then the other drug is going to be marijuana. This can be bad both because both the acute and chronic effects of alcohol and marijuana presumably add up.
On the other hand, legal marijuana may act as a substitute for more the dangerous drugs that are still illegal.
Personally, I think legalising it is dumb, but as many here say people should be allowed put what the hell they like in their own bodies, so if legalisation is coming, so be it.
I think sensible users (like social drinkers) will do okay, and the kind of people who drive drunk will drive stoned (because I see no reason they would suddenly become responsible by swapping one intoxicant for another), and the kind of people who like to drink themselves into oblivion will probably mix marijuana with that and other legal or illegal drugs, pills and potions – for example, look at this genius here (guy was arrested for a fatal hit-and-run, tested positive for cannabis and coke, if you buy his defence he was already high on weed before the collision and then went and took coke on top of it):
And of course, young people being idiots because they’re young and young people are idiots:
I don’t know the source for this (I think I just heard it in conversation, so feel free to take it with a grain of salt), but my understanding is that marijuana causes somewhat less impairment than alcohol on its own, but combining marijuana with alcohol, even in low amounts, is far worse than either on its own.
On top of that, I believe consuming alcohol when high was more common than not (although maybe that was just among those who were arrested for DUI, which would be entirely different). Anecdotally, whenever I’ve smoked pot in a social gathering, there were almost always drinks involved, too.
Meta-analyses show a clear relationship between beer then grass leading to one being on one’s ass, while grass then beer correlates strongly with being in the clear.
It’s just science.
Anecdotally, I’ve found even very small doses of marijuana to be an excellent substitution for certain kinds of alcohol abuse.
Specifically, when I’ve been hard-pressed in my work/personal life I’ll be tired and apathetic at the end of the day, and being unable to motivate myself to do anything will just kill the evening by getting drunk. This is a bad cycle, as it tends to make the next day stressful too (working with a hangover is stressful).
About a quarter the dose of edible marijuana that people would use to typically get high entirely removes the reason and impulse to drink, and doesn’t substantially alter my ability to act normally or function. Haven’t been able to test the effect long-term though, as it’s only freshly legal here.
This is disappointing, I had hoped the substitution effect would be true. The ambiguity of the case for marijuana weakens my stance in favor of legalizing / decriminalizing all drugs.
I assume many accidents are from combining marijuana and alcohol, as I was under the (possibly mistaken) impression marijuana alone did not seriously impair driving. I’m not sure if that makes any difference, but I am curious.
How would you think about the possibility that stronger DUI laws could lower deaths? Or that technological advances and increasing popularity of ride-sharing could also reduce such accidents in the medium to long term?
As an experienced stoned driver, here’s my take on how seriously marijuana impairs driving:
It is extremely easy to maintain control of the vehicle while stoned, i.e., stay in your lane without swerving. This is obviously not true of alcohol. Reaction time is slightly lower (though not nearly as low as when drunk), but you can easily compensate for this by driving slower and maintaining more distance behind the car in front of you than you normally would, and at least in my experience I always have the presence of mind to make sure that I’m keeping an appropriate distance. The lower reaction time does put me at slightly greater risk from mistakes by others though, since I cannot react as quickly to bad drivers.
Given this, it is very surprising to me as well that the substitution effect doesn’t appear to be true.
How certain are you of your personal assessment of your reaction time while stoned?
If marijuana impairs you without making it self-obvious that you are impaired, it could be more dangerous than alcohol… highly impaired but aware of your impairment might be better than lightly impaired but unaware.
I once saw a video where they had volunteers get drunk, drive a closed course, and then give a self assessment of their performance after they had sobered up.
Most remembered doing pretty well. They might say things like “I might have knocked over a cone or two, but overall I did alright because I knew I was drunk so I drove super careful”. Some even said they drove better when drunk.
The participants were absolutely shocked when shown video of their own performance, in which they careened around the course with no regard to the demarcated lines and hit dozens of cones. They said things like “Is this a prank? There’s no way that’s me”
People who drive drunk are hilariously bad at self assessing their performance. I am sceptical that marijuana users will fare better, but I suppose it’s possible. Alcohol is associated more with false confidence, after all.
If you drive under the influence regularly, please have your ability to do so independently evaluated. You may be putting yourself and others at tremendous risk without even knowing it.
Please stop doing that. I get that there are situations and points in life where it seems inevitable, I’ve been there and done that. However, if you taboo ‘driving’ for a moment and reconsider that you’re commandeering over a ton of metal and plastic at speeds your body couldn’t achieve unaided it becomes clear that ‘driving’ is already, by itself an activity dangerous to yourself and others which we as a society only allow because it represents an agreeable compromise between our desire to get somewhere and our desire to stay alive.
You are unilaterally defecting from that agreement by lowering your ability to interact with traffic and then getting behind the wheel. More concretely: Your lower reaction time is an issue when kids step onto the road from behind a car. The risk you impose on ‘bad drivers’ is negligible at most speeds and assuming you’re both in fairly modern cars.
I found this:
This is some really bad total harms analysis.
1) alcohol increases probability of accidents by 17.8x compared to 1.65x for cannabis. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5678710/
2) Alcohol consumption decreases with legalisation. https://www.forbes.com/sites/thomaspellechia/2018/01/22/alcohol-sales-dropped-15-percent-in-states-with-medical-marijuana-laws/
Therefore, cannabis sales would have to increase many many times before the driving accidents argument were viable. And this is before any consideration of personal health effects (which obviously favour substitution from alcohol to cannabis). You’re picking up noise in your epidemiological level stats (potentially shifting demographics).
Related to the issue of noise, it seems like we’re witnessing an inherently transient phenomenon. It’s like measuring the temperature of an object which has just been taken out of the fridge and put on a heating pad and making judgments on the basis of its current temperature. Even if this effect were fundamental and not a result of shifting demographics, making the call on “legalization: good or bad?” is still premature.
A secondary issue is the fact that the total driving fatalities figure is so absurdly large that tiny changes in its percentage can appear to swamp all other concerns. This is an uncomfortable situation for any kind of consequentialist analysis, where the total harms are dominated by what is basically a contingent side effect. (I say contingent, because driving safety can vary, and has historically varied, much more significantly than +/-5% due to other relatively simple interventions – seat belts save 8 times more lives than stoned driving kills, for example – and slightly futuristic interventions like self-driving cars could literally nullify the harms caused by legalization.)
Perhaps we should be re-evaluating how we treat SSRIs and other non- narcotic drugs. If I’m on SSRIs and forget my bottle at home while travelling, why shouldn’t I be able to go to a pharmacy and self-prescribe more? Have a “yes, this could be a really bad idea” form to sign and go ahead.
Likewise for other insanity such as 400mg ibuprofen tablets, 1mg folic acid tablets, and D50.
What’s insane about 400mg ibuprofen tablets?
You can’t buy 400mg ibuprofen tablets without a prescription, yet the bottle of 200mg you can buy OTC has directions on them to take 2.
That’s because they’re taking advantage of the placebo effect where “take 2 200mg tablets” helps more than “take 1 400mg tablet”.
One component that’s not included in Scott’s analysis is the positive impact of higher quality weed and smokeless delivery methods.
Smoking any sort of dried plant material introduces health risk. It doesn’t matter what the plant is, because the impact pretty much comes from smoldering organic compounds. Marijuana is much less a public health problem than tobacco because marijuana users smoke far less total material. (One pack a day habit equals 140 grams of tobacco a month, versus maybe 7 grams for a heavy cannabis user.)
One of the hallmarks of legalization is much higher quality weed with significantly higher concentrations of THC. That’s a major public health victory because it means the user smokes less plant material to achieve the same high. For example if THC levels increase from 5% ditch weed to 25% Ghost Train Haze, then we’d expect to see a commensurate reduction in quantity of 80%.
Along the same lines, it’s pretty clear that legalization strongly promotes smoke-free routes of administration. Things like edibles and vapes require a whole economic infrastructure to deliver in a convenient way. In a black market, nobody’s investing the time and capital to produce those things en masse. But in a dispensary those products are front and center. That’s another public health boon, because smoking drastically more risk than vaping or edibles.
In other words, I’d modify add point 7a: Smoking marijuana may increase cancer risk, but legalization could actually prove to be a major ameliorating factor.
Yes, but the same change may be dramatically increasing the risk of fatality in automotive accidents. When smoked the effect of the drug come on almost immediately but when consumed orally they come on hours later– potentially after the user’s situation changed and put them behind the wheel of a car.
Scott, with this change in view aren’t you basically being the Man Of One Study (ok, two, but by the same group so it’s really like 1.5 max) that you rightly warned us to beware awhile back?
That’s before even getting into the deontological argument that bodily autonomy >> your studies. Both this and the childcare argument seem unusually weakly reasoned for you, though to be fair you are still doing better than 95% of political pundits, so there’s that.
i’m ignoring this analysis because it seems obvious to me that self-driving cars will become widespread soon.
Until then, what…
No, no, that’s why it is so important. Legalizing pot will cause an increase in traffic accidents, which will in turn make legislators approve self-driving cars, which then will eliminate almost all traffic accidents.
Next task: ending world hunger by legalizing MDMA.
Curiously, all of the states that have recreational marijuana legalization already have below-average levels of car accident fatalities per-capita, as of 2016 (source). I’m not sure how to interpret that in light of this, but as a hypothesis, it might be a substitution effect – states with legalization may well have lower per-capita levels of alcohol abuse, but once weed becomes more widely available, people who wouldn’t drive drunk instead decide to drive stoned.
Are you reading the table correctly? Perhaps you read the first row, Alabama as representative?
Colorado is almost perfectly average, both per-mile and per-capita. Washington is below average (ie, better) and Oregon is above average. [WA is a full standard deviation better than average in both metrics. Oregon is 2/3 of a sd worse than average per mile, but only 1/8 worse per capita, as much as CO is better per capita. CO is 3% of a sd better than average per capita]
Oregon drivers are just idiots; I don’t know why they can’t drive as well as in Washington. Above-average rain predictably causes an interstate-clogging crash. Rain.
Being a resident of a state that borders both WA and OR, I would venture to say that more WA potential drivers take public transit or commute shorter and more consistent routes to work or wherever. In Oregon you have some larger urban areas spread out in a wider area , not to mention more suburban sprawl in the Portland area vs a very North-South splay for Seattle, and of course Spokane, Olympia and Tacoma are smaller cities….
Also, east west traffic in Washington crosses long stretches of more arid land, whereas that effect in Oregon is perhaps less pronounced.
Yeah, Sea-Tac is neatly constrained by geography while Portland sprawls both from Vancouver, WA north of the Columbia to barely more than 20 miles north of Salem along the Willamette/interstates and through West-East suburbs.
It’s way overserved for its low density by a relatively convenient mass transit, though. Less sophisticated than Seattle’s or BART, but it’s there.
BART may be expensive, but it’s not sophisticated. I’ve never seen a worse public transit system of comparable size.
@Nornagest: Yeah, “expensive” would have been a much better choice of word. You might say it’s a… crappy use of tax dollars.
Ah, sorry. I got the wrong average.
It could just be a cultural effect. Liberalism (in the classical sense) tends to coexist with relatively peaceful, well-behaved populations. It’s easier to give your populace more freedom when they’re less prone to social pathologies. The most socially restrictive places in the US tend to be those settled by the Scotch-Irish, who have a history and culture of wild, antisocial behavior.
You see a similar distribution of “dry counties”. The localities that ban alcohol tend not to do it because everyone there eschews alcohol. They do it, because at least some sliver of their population loves alcohol a little too much.
Well, few of those regions have open MJ laws at this point, so it is yet to be seen eh?
@Scott, any opinion about the effect being transient as new users try now legal pot, and maybe don’t understand how it effects them? Might we see a smoothing and reduction in these numbers to background levels in the coming years? Can we look at Portugal or Amsterdam for data? I imagine there’s a large increase in accidents when left-side driving countries switch to right-side driving, but concluding that the right side is inherently deadlier would be unwarranted. Just a thought.
I don’t think I said it on the “Nominating Oneself For The Short End Of A Tradeoff” thread but I think there’s really something there with that intuition and I appreciate you making the post. Unfortunately, it’s hard to phrase in utilitarian terms. Very easy in libertarian ones, though.
It’s easy to phrase in utilitarian terms.
If I know that smoking marijuana increases my chance of dying in a traffic accident by x% and do it anyway, that’s evidence that the utility to me of smoking is greater than the disutility of the extra risk.
It depends on my being rational and well informed, as do many utilitarian arguments for libertarian conclusions, but given that it’s straightforward.
This assumption is quite dubious when addictive drugs are involved. But of course if we assume that the state should save people from themselves when addictive things are involved, then we should start with banning alcohol, tobacco, sugar and probably porn before we get to marijuana.
You can’t justly “ban sugar” when it’s the main macronutrient in every fruit. Banning certain processed foods reliably seen to lead to diabetes, maybe.
And shouldn’t porn be equal to marijuana and not ahead of it? Both make you sit on the couch and giggle foolishly. 😛
I meant refined sugar, high fructose corn syrup and things like that, not bananas.
Good on you for being confused on this issue.
My thought is that it’s not logical for marijuana to be illegal when tobacco is legal. Screw this utilitarian triage.
Oh, and as to substitution effects? You know as a psychiatrist that ndividual reactions to drugs are highly vaeiable, so it would harm individuals to set public policy based on such an assumption from some study. I drink wine in moderation for its purported health benefits and because it feels good. A marijuana edible (no cancer thanks) made me paranoid rather than a mellow hippie.
420 blaze it
I have a general principle that goes something like
It’s wrong to restrict “innocent” people’s freedom to solve a problem that can be reasonably solved by restricting only the freedom of those who are culpable.
If you’re worried about driving deaths from marijuana, then make a one-strike-and-you’re-out law that says that one DUI means you lose your license for life, and if you get caught again after that, you get a minimum of a year in prison.
It’s like we grow up and forget how unfair it was when the teacher punished us for “fighting” when all that happened was the school bully ran up and punched you.
As usual, human freedom is given a value of exactly zero in this analysis. The losses suffered by those arrested and/or convicted of marijuana use or marijuana sales are also valued at zero.
Not just that, but the medical/health/lifespan-related effects of those arrests, botched arrests, mistaken address home raids, prison violence, gang and cartel violence, etc. are left out along with the secondary effects for things like having two non-incarcerated parental figures without added difficulty in seeking employment due to prior or arrest or conviction, which might affect things like, say, a good environment for early-childhood development? I am really scratching my head in light of the last two posts.
Measuring all the factors involved accurately is really hard. Signing the net effect of ending pot prohibition? Not so much.
The losses suffered by those arrested for marijuana are valued at 70,000 quality-adjusted life-years, and the losses suffered by those convicted for marijuana are valued at 10,000 quality-adjusted life-years. Please read the post I linked to saying that it was where I discussed the value of things like arrests and convictions.
Black market economies are a blight, and their costs are WAY underestimated. They promote and reward fraud and violence, they erode proper respect for the law, and the corrupt law enforcement priorities and the agents themselves. It would be hard to put a number on this, but probably outweighs a few thousand increased traffic deaths (many of the users themselves). You would have to balance this against lives saved by foregoing a violent trade and freeing legal resource to actually reducing crime and solving murderers.
I can’t vouch for the accuracy of this, but there is a widespread narrative according to which decriminalization has made marijuana trafficking less profitable for the cartels, so they have shifted their efforts to heroin, and this has been a big driver of the opioid crisis (my understanding is that heroin is an order of magnitude cheaper than it was 20 years ago). So in accordance with the law of unintended consequences, it may be that the biggest negative consequence of marijuana legalization/decriminalization has been the soaring rate of opioid-related death and crime (this is not to say cheap and readily available heroin is the *only* driver of the crisis, but it does seem to be a big one).
Are the economies of scale really that good?
How much more heroin do you have to be importing to drop the unit costs that much?
I don’t know much about the economics of the illegal drug market, but I read an interesting book about organized crime during prohibition. It sounded from that as if the economy of scale was in corruption. If you were going to buy a judge or police officer, it made sense to use him to cover as large a volume of bootlegging etc. as possible.
And a judge or police officer would probably not let themselves be bribed by a small scale amateur drug dealer, but if Pablo Escobar shows up and offers them plata o plomo, they might be much more willing to accept the bribe.
Not sure about your general question, but I did a bit of sleuthing on prices. It looks like the ‘cheap heroin’ story I mentioned above is based on its cheapness *relative to prescription opioids*, not historic prices. In terms of historic prices, these guys document a giant fall in prices from 1993-2008:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5725335/
And these more recent data from the UN show something similar until 2008, followed by fluctuations up and down with no particular pattern since then:
https://www.unodc.org/wdr2018/prelaunch/7.5_Standardized_prices_of_cocaine_and_heroin_in_the_United_States_and_Western_Europe.xlsx So heroin is a lot cheaper than it was in the 90s, and a lot cheaper than prescription drugs, but not a lot cheaper than it was at the beginning of the current epidemic, I think.
I think something like this did happen, but it happened several decades ago, when we started fighting the drug war in earnest and poured resources in to border control and interdiction efforts. I don’t think the cartels have been a big supplier of pot in a generation, excepting maybe the border states.
I googled around a bit, and I’m still basing this all on thirdhand reports, but this more recent version of the story seems to be based on changing levels of seizures of various drugs at the border since 2009. This Washington post article sums it up: https://www.washingtonpost.com/world/the_americas/losing-marijuana-business-mexican-cartels-push-heroin-and-meth/2015/01/11/91fe44ce-8532-11e4-abcf-5a3d7b3b20b8_story.html?utm_term=.75a9176ed510
Is it still the case that we have no effective way to measure marijuana intoxication, other than field sobriety tests and the like? It seems like comparing one state to another is a very blunt instrument, and the only way to really know how much marijuana affects DUIs is to know how stoned the people who get in accidents are, like we do with booze.
I actually really dislike the way California is handling legalizing marijuana. Suddenly we have these huge billboards everywhere advertising the drug. I guess people figure it’s big business now and they need to put their logo everywhere. Public spaces suddenly are smoky everywhere. You really cannot walk around certain neighborhoods in San Francisco without smelling it everywhere.
Amsterdam seems to have legalized it without having either of these problems. I never saw advertising for marijuana while I was there other than the storefronts of shops. It was pretty well contained to the “coffee shops” and I never had to deal with secondhand smoke while I was there.
I honestly would have preferred just keeping weed illegal. I think all the benefits are mostly illusory.
People who drink a beer or two after work are not going to smoke weed instead. Marijuana stays in your system for days, and frankly, it makes you stupid. A lot of people report that they don’t even get high the first time they smoke it. Weed probably could displace antidepressants for some people (although there is apparently also some chance of having a psychotic break when you’re on it, so I’m not sure the FDA would approve it for that purpose either…) But your boss probably doesn’t want you to be giggling and stupid while you’re on your antidepressant, so… again, not very useful.
The kind of half-hearted legalization we’re doing is not going to break up the cartels. They’ll still have their cocaine business, and their heroin business, and lots of other stuff. I don’t think anyone is proposing to legalize all that stuff, so it’s not like the drug war is going to end. The federal government not even on board with the limited legalization that California is doing, so criminal organizations are still going to be part of the picture.
We already know that any kind of carbon smoke generates small particles which migrate from your lungs to your arteries, and cause heart disease. It doesn’t matter whether you burn wood, diesel fuel, coal, paper, cigars, whatever. Those PM2.5 particles are going to escape and cause problems. We almost succeeded in wiping out cigarettes in the first world. e-cigarettes were starting to wipe out the remaining demand for cancer sticks. Now we’re back to square one with weed. One step forward, two steps back.
I feel like there are a lot of better drugs we could have legalized. MDMA is a good example. It would be excellent for psychotheraphy and the kind of lowering of inhibitions that alcohol provides. But it doesn’t make you stupid or stay resident in your fat cells for weeks. Almost all of the ADHD drugs (like modafinil) could probably just be made legal over the counter with very few consequences. As Scott wrote, the process for getting a prescription for them is mostly just a bureaucratic Kabuki dance anyway.
+1
I’d like to see smoking (anything) made illegal outside of particular confines. People need to air out their houses on occasion, enjoy their yard, etc….
https://en.wikipedia.org/wiki/Law_Enforcement_Action_Partnership
There are quite a few ex law enforcement people who are pushing for full legalisation of all drugs.
That’s because they haven’t technically legalised it. Rather, they have a policy of non-enforcement against the retail end of the market that takes place within the coffee shops, while production and supply are still controlled by the criminal market.
I have also heard it suggested that most countries would be be better placed than the USA to legalise recreational drugs and reap the benefits of taking the market away from the criminals without suffering the harms of the wide scale commercialisation and advertising that you talk about, but because the USA has such robust free speech laws, preventing restrictions on advertising, it is pretty much all or nothing for you. I am not sure how accurate that is, though.
People can get high without smoking, by vaporizing oil with THC in it. It’s incredibly safe and cheap and effective, but it’s really difficult to make the oil yourself. As a result, vaporizers are common and cheap where pot is legal, and incredibly rare where pot is illegal. In other words, this is an argument for legalization.
“Public spaces suddenly are smoky everywhere. You really cannot walk around certain neighborhoods in San Francisco without smelling it everywhere.”
I don’t know how you could possibly attribute this to legalization. SF has always been incredibly weed friendly, especially if you’re including Haight, etc in “certain neighborhoods”. If you’ve been paying attention, SF has smelled of weed for half a century now.
That was literally the tourist attraction value of Haight. For the last 30 years or so the police let the homeless kids in the park sell overpriced crappy weed and baby boomers who remembered getting stoned in college got to relive the good times (or at least see people who were).
I don’t know if there is more or less smoke there now (but it’s Haight it’s semi-official designated weed area in SF) but eliminating the scuzzy dealers in the park (even if I found them useful) is a plus even if it might have helped put a few bucks in their pockets.
There’s been a sharp increase in pedestrian deaths in recent years. From the New York Times:
“Where Pedestrian Deaths Are Up, Is Marijuana to Blame?”
By Neal E. Boudette
Feb. 28, 2018
https://www.nytimes.com/2018/02/28/business/pedestrian-deaths-marijuana.html?smid=tw-nytimes&smtyp=cur
Smartphones, however, could be a more important cause.
Here’s an earlier NYT article on the unexpected increase in traffic deaths:
“Traffic Deaths Up More Than 10 Percent in First Half of 2016
“By DANIEL VICTOR OCT. 5, 2016
“Traffic deaths in the United States rose 10.4 percent in the first half of this year compared with fatalities from the same period in 2015, maintaining a steady and troublesome climb.
“The numbers were released on Wednesday by the National Highway Traffic Safety Administration, which noted that Americans drove about 50.5 billion more miles in the first six months of 2016 compared to the total miles in the first half of 2015, an increase of 3.3 percent.
“But that does not account for the rise in the number of deaths: to 17,775 in the first six months of 2016 from 16,100 in the same period in 2015.”
Something that’s striking is that the rise in traffic deaths was rather similar in scale and timing to the rise in homicides during the BLM era following Ferguson in August 2014. (The total number of homicides in the US grew 22% from 2014 to 2016).
I haven’t seen any data on how hard traffic cops were working at stopping speeders and the like, but it’s possible that the tendency of cops to retreat to the donut shop when being criticized by The Establishment (i.e., the Ferguson Effect) may have had an effect on traffic safety as well as homicides.
I’m a little wary of looking at differences by state in traffic deaths in reference to marijuana laws because there have been recently some sharp changes in homicide deaths related to political events. For example, the homicide rates in St. Louis, Baltimore, and Chicago shot up following political triumphs of BLM over the local police.
My impression is that police effort can’t be assumed to be constant over time. It seems fairly sensitive to the political and cultural environment, which can shift rapidly and significantly. I was struck by an old time journalist’s observation that during the high crime 1970s-1980s in NYC, NYPD officers might have been averaging only about 2 hours per 8 hour shift in actual policing. The pro-police Rudy Giuliani’s big accomplishment, in this view, was doubling that to perhaps four hours of work per day.
Here’s one possibility that I’ve never heard discussed: Perhaps police don’t try to enforce traffic laws as hard after marijuana is legalized because the expected value of traffic stops to the government from fines is lower after marijuana is legalized.
“But bacon also increases cancer risk by a few percent, is already a giant public health disaster, and we don’t worry about it that much”
“We might choose to focus only on the one-third of fatalities that happen to bystanders if we believe people should be allowed to make bad choices that only hurt themselves.”
https://iea.org.uk/publications/killjoys-a-critique-of-paternalism/
This is a very good primer on the different categories of cost and benefits when evaluating policies. There is a difference between costs you impose on others and costs you impose on yourself. You will only incur a cost if at the same time you are getting a benefit that is at least as much, i.e when you do your math counting costs that people choose to assume you have to assume there is a benefit that is at least as high.
On the general issue of external effects from risking your own life …
It’s clear that letting individuals make the choice doesn’t give the optimal result, since the choice effects others–there are external effects with both signs (consider the benefit to your heirs). Suppose the value of your life to you is 1000 and to other people 100. If the value to you of something that imposes a .01 probability of death is more than 10 (assume you are risk neutral) you take it. But if it is less than 11, you shouldn’t take it.
This would be an argument for restrictions if we had a mechanism for imposing them, such as an omniscient and benevolent deity, that could estimate the net cost and benefit more accurately than you do. But most of the time we don’t. You have inside information on both the value of your life to you and the value to you of smoking, bacon, or whatever. You have an incentive to make the choice that is almost optimal–counting value to you but not to others. Nobody else has either that information nor that incentive.
I remember an old book that calculated the value of life implicit in a bunch of different regulatory decisions. It varied wildly, I think over a couple of orders of magnitude. They cannot all be right, or even close to right.
Age also affects reaction times — bigly. Should we set the maximum driving age down to 60?
Even factoring in that these are relative rates, I still think we should have a look at the way automobiles have changed in the last few decades. Compared to the 1990s, cars:
1. Have quicker steering wheels. (Smaller motions of the steering wheel have bigger effects.)
2. Worse visibility. The rear windows have been moved back for aerodynamics.
3. Controls are far more busy. They even have $#(*%&$* touch screens.
4. Engines are more powerful. This is useful getting on the interstates, but also a temptation.
5. Front wheel drive is the norm. (OK, this goes back a bit further.) Front wheel drive is better in snow perhaps, but it produces instability on wet roads. Fish tails get magnified if you take your foot off the accelerator.
These factors magnify the effects of driver impairment, whether we are talking alcohol, dope, or age.
Old people tend to drive older cars. What happens when future old people drive today’s cars?
“A study by the Highway Loss Data Institute in June of last year finds that states that legalized marijuana saw insurance claims for auto accidents increase about 3% over the general national trend for the time. An updated study by the same group finds 6% according to insurance claims, and 5.2% according to police reports.
These are usually contrasted with a 2017 study that finds legalization states did not have significantly increased rates of car accident fatality. What is going on?”
You suppose some complex hypotheses about sampling and significance to answer this question. Why not try the obvious? Legalization increases accidents but does not increase fatal accidents. The kind of behavior that leads to fatal accidents is very different than the kind of behavior that leads to fender benders. Fatal accidents come from speeding, aggressive driving, and excessive multi-tasking. Fender benders, getting rear ended, and side scrapes come from hesitating before taking actions. If marijuana causes slow responses or an increase in hesitation but doesn’t cause an increase in speeding or aggression then we would expect to see exactly what the data shows.
I think you may be thinking of car crashes as a more homogenous class than they actually are.
I’m telling you, that’s a fake comparison. And it’s a fake comparison that only someone who has never actually tried marijuana firsthand would make. My partner and I are fairly prolific bakers, and we live close enough to Oakland to deliver a brownie (Or a pan even, for a proper sample size), if you’d like to experience just how fake a comparison it is.
Why are marijuana legalization advocates burdened with proving it’s a good thing? I’m sorry, I thought this was AMERICA, land of the free?
To say nothing of the human tragedy that is making a felon out of someone for smoking a benign substance that makes them sit on the couch giggle.
I think that regardless of the right or wrong of marijuana, this case, it seems to me, illustrates the practical limits of utilitarianism with regard to measurement. In most cases, where utilitarianism would seem like a good model, we just can’t trust the measurement to give us a reliable ‘utils’ calculation. So we revert to some back-justified virtue ethics.
Bayesian updating seems like a good solution but it too has practical limits on how often it can trigger action without people’s lives descending into chaos.
So, perhaps stating a starting position as priors is not that useful. Maybe better to say, many people smoke pot, it does less harm than other things that are legal, it should be legal. How much less harm or how much more harm than doing nothing is not within the scope of the calculations, so I will support the ‘right’ thing to do, unless I have a LOT of strong evidence that it is the ‘wrong’ thing to do. Small perturbations in utility are simply too noisy to use for any useful decisions.
A few reasons to think the effect goes the other way:
1) Replacing illegal weed with legal weed creates huge opportunities from stronger product and less carcinogenic routes of administration (edibles etc..) as well as other harm reduction strategies that may balance out the increased usage.
2) Replacement effects of other intoxicants with weed. This is especially important for opiates but alcohol also causes a huge amount of damage and if we have even a small substitution effect that could have outsized influences. No I don’t expect to see this right away but phase in as people choose to avoid opiates because of another alternative and it becomes more socially normal (e.g. it will be younger generations who replace alcohol)
3) Simulator based studies of the effect of MJ on accidents actually are kinda mixed as to the net harm or benefit of being stoned while driving. Apparently stoned drivers are less coordinated, slower to react etc.. but also tend to drive more slowly reducing damage when they do get in accidents and partially compensating. This may change with increased familiarity but I suspect skill will as well.
4) That traffic study was very unconvincing. The supposed control states differed by substantial amounts in the net change in their accident rates and the changes were from 1-8%. Given there were only 3 states that switched and a complicated statistical model this looks even weaker and more subject to publication bias than himmicanes (obviously no model that just said ‘ehh too little data’ was going to get published).
5) The elimination of the externalities associated with the illegal marijuana trade. Others have mentioned people serving time in prison but it not only creates direct violence at grow operations but helps create a criminal underworld that can’t report on other crimes (both as victims and reporting other criminals. Many people who are ok growing pot would report people planning a robbery if they weren’t in an illegal industry and vulnerable to retribution creating a pool of potential criminals.
6) Remember the states which have legalized attract the most problematic users from other states so initial negative effects will diminish over time.
But I do think this really makes the compelling case that gun control isn’t really about concern for saving lives. It’s a cultural issue stemming from a distaste at the idea that people enjoy owning and using guns.
This doesn’t mean it’s not a net positive. Probably is. Merely that it is appropriately way the fuck down the list of things people would care about especially given the difficulty in getting it enacted.
Unfortunately I haven’t had access to a computer in the last 7 days, but I feel like there’s still insufficient evidence to draw a conclusion. Greg S. makes most of the points I would have, but there’s a bit more I’d like to add:
The article that references the HLDI study also mentions a Federal study that shows no correlation between marijuana usage and accident frequency. Here’s that survey Another study considers that the effect increases accident rates 200% Both of these however, are less than the Seven times likelihood of getting into a fatal collision. As Greg mentioned before, if amount of smokers that increased in those states legalized are only in the low teens, then that is insufficient to cause such a greater increase.
The HLDI used neighboring states as a control. I assume they did this so they could make sure states of similar size/geography would be compared. However, the HLDI study only looks at frequency of claims in a big-picture view. It did not mention controlling for, or inspecting the rates of: accidents/claims/collisions between driver/property and driver/driver collisions, and the resident states of the driver and property. This becomes important, because for a link between marijuana usage and increased recklessness in driving to be tenable, we should expect to see increased driver-driver and driver-property collisions and claims where both parties are residents of the same state.
Because the HLDI takes a big picture view though, we’re missing a significant possible confounder: increased collisions and accidents as a result of drivers from jurisdictions where recreations marijuana isn’t legal traveling to states where it is legal, and increasing accidents by way of simply driving more to get high, and then driving back home. You can’t just hand-wave something like Pot tourism in a country that is highly automobile centric.
If the HLDI used a more granular level, like counties as opposed to states, and compared counties of both parties of a collision, then we could draw a more valid conclusion from that. We would then expect to see increased collisions inside counties that contain stores or dispensaries and between counties where one or both of those counties has a store or dispensary (Compared to levels of collisions prior to a dispensary being constructed) However, based on my hypothesis, I would expect to see no increase within counties that contain stores or dispensaries, but a significant increase in cross-county collisions where only one of the counties contain stores or dispensaries, and the other is a county in a non-legal jurisdiction, and potentially also increased intra-county accidents on border counties.