America's Driving Problem with Drinking

Drinking Driving and DUI's

‘Gasoline and alcohol don’t mix,’ says the American slogan. Of course they do. Our urban planners mix them all the time and in great doses. See the zoning codes for confirmation.
— Ray Olderburg, The Great Good Place

A Simple Comparison

My post about bars and parking lots has been getting some attention recently so I decided to expand on that idea with some additional data. The data didn’t show me what I was looking for but it did provide some interesting insights about America. 

For many years now it has been my contention that America has a problem not with drinking, but with driving. Sometimes I tell people that I became a planner in order to ensure everyone can safely imbibe and safely get home. I decided to try and explore this idea with some simple data. 

Here’s the comparison that got me thinking about this. If drunk driving is a result of excessive drinking and a lack of restraint than we should be able to find a clear relationship between the two. Drunk driving is a huge and well publicized issue in the US but what about other places? Americans, on average, drink a little less alcohol than people in the “old county” so places like England should be a tragic hellscape of automobile crime, right?

data sources: health and social care information centre, drinkdriving.org.

 As an added bonus I just invented a flag for the off chance that Texas and California secede from the Union and merge into a separate republic. You're welcome.

The UK presents a good comparison because it is very similar to the US in terms of culture, attitudes, and politics but has some notable differences. To even the playing field a bit I decided to compare the UK to California and Texas rather than the entire US. Together they approximate the 60 million people in the UK. The British drink just a little more on average than Cali-Texans but the difference in DUI arrests and DUI’s per capita is absolutely staggering.

With roughly the same number of people and just a little less drinking, Cali-Texans somehow end up with Eight Times the number of drunk driving arrests. There is something fundamentally different about how we drink and how we drive in the United States compared to Britain. The British don’t actually drive that much less than Cali-Texans either so the answer must by somewhere else. Maybe the issue is less about having NO option to drive than it is about having the option NOT to drive? I wonder how many bars in the UK look like Westville Pub versus Buffalo Wild Wings?


Here is the most sobering part of this comparison. About 10,000 Americans are killed each year in drunk driving incidents. The UK has averaged about 500 per year. Twenty times as many Americans die from drunk driving as British people. Even if we account for the difference in population, the fatality rate is five times as high. 

I was inspired to revisit this idea by a Citylab article by Sommer Mathis. She does a great job of pointing out how anti-drunk driving efforts focus almost single-mindedly on drinking while ignoring driving. I’m hardly the first one to notice this gap . Research is already pointing out the link between public transit and drunk driving . As I discovered though it’s hard to get good data.

Searching for A Link

When I Did The Math to try and demonstrate the relative impact of car-reliance on drunk driving here in the US I couldn't really find a pattern. I compared DUI arrests, vehicle miles traveled, and alcohol sales between States. There is an unsurprising connection to per capita drinking but I don't think VMT was the right metric. As we saw with the UK example people may drive to work every day but its how they get home from the bar that matters for DUI's.

Some appear to support my hypothesis. Wyoming, for instance, is ranked first in the nation for both per capita driving and DUI’s per capita but only 12th for drinking. Massachusetts on the other hand is 17th for per capita drinking but only 44th for DUI’s and 43rd for driving. There is a stark difference in the level of urbanization and transportation options between the two as well. This post  makes a similar comparison.  

I suspect that the state level, at which I could readily find all three data, is too big a scale to make this comparison. If you were to compare tighter geometries I bet you would tease out this relationship. I'd be shocked if the DUI rate in New York City was not lower than in say Houston . What I really want to find is significant variation in auto-dependency. I don’t think that states vary enough to tease out the impact on drunk driving.

And Now For Something Completely Different

This is a ternary plot which is a fun way to show how a feature is split between three elements. In this case the consumption of three types of alcohol. It also serves as a legend for the classes on this map.

The data I found wasn't really useful for what I intended it for but I decided to map it anyways. Things go decidedly off-topic here but I just wouldn't feel right if I didn't include some maps. Maybe I just wanted to end on a lighter note.  Interestingly, the alcohol consumption data is broken down between beer, wine, and spirits. I thought it might be interesting to take advantage of that to look at alcohol preference across the nation. 

By and large America skews strongly towards beer. Though there are some interesting exceptions. Idaho seems to like their wine more than you would guess. Wyoming and North Dakota, perhaps in keeping with pioneer tradition, prefer hard liquor. 

That's what I call a BAR graph. Get it?

When you line states up by beer preference a more interesting pattern emerges. The beer-dominant states on the left tend to be less affluent states while the wine and whisky states on the right are among the wealthiest. Does alcohol preference track with income? Trough the wizardry of common household excel statistics we can find out.

Sure enough, there is a clear relationship between income and beer preference. For every additional $1,500 worth of median income you can expect to find 1% lower preference for beer. If we run the same model to compare income to wine we find that as median income rises by $2,300 the preference share for beer shifts by 1.6% to make room for a 1% increase in wine preference and .6% increase in liquor. This makes sense considering the qualitative relationship. Wine and liquor tend to be more expensive and are marketed as more sophisticated beverages. I wouldn't be surprised to find that craft beer follows the same pattern as wine with regard to income.

In case you'd like to do your own excel (or better) wizardry here are some links to the DUI, Ethanol, and VMT data I used.