This is another example of the horrid state of modern “journalism”. You may see reports that Boca Raton is one of the most dangerous cities for DUI deaths, covered in various “newspapers” such as the Miami Herald, the Florida Times Union and the Sarasota Herald Tribune.
The reports are based on a “study” by that famous research university – “Value Penguin” (a clickbait driven consumer finance website). Author Matt Timmons has a bachelor’s degree in Philosophy and East Asian Studies, so obviously he’s an expert in statistics. Linda Robertson of the Miami Herald describes Timmons as “an insurance analyst.” Timmons describes himself as a “technical writer.”
Timmons “study” is statistical nonsense. It starts off with this factoid:
There were 515 deaths due to DUIs in Florida in 2017
In that same year Florida had a death rate of 672 per 100,000 people. With 21 million people that makes DUI the cause of 0.3% of all deaths. With a total of 3112 Florida traffic deaths in 2017, DUI deaths account for less than 20% of all traffic deaths.
The other problem with the number 515 is it’s small for statistical purposes especially if you’re going to do a breakdown by cities. Florida has 412 incorporated municipalities. That works out to an average of barely more than 1 DUI death per city a year. That creates sample size problems, which anyone slightly familiar with statistical analysis would recognize.
The small sample size makes it easy to include attention grabbing numbers. This study says Sarasota is the “the most dangerous city in Florida for DUI deaths.” It indicates Sarasota had “12.2 DUI deaths per 100,000 residents.”
But Sarasota doesn’t have 100,000 residents. It has only 60,000. So that 12.2 number works out to about 7 deaths. If they happen to have a bad year with a few extra deaths it vaults them to the top of the list. And many small cities will have no deaths in a particular year or even a few years, making them appear safe. Value Penguin hints at this problem when it mentions that the number is up dramatically from the previous period.
What’s really going on is the cities at the tops of the lists had bad years. Large cities like Miami don’t make the list because they’re big enough that an extra couple of accidents don’t move the needle.
When it comes to our beloved Boca Raton, Value Penguin has this to say:
Boca Raton is the most dangerous city in the Miami area and the 8th most deadly city for DUIs statewide.
Seriously? This is ridiculous. We don’t know how Timmons came up with his numbers but the odds are he got something very wrong. Driving in Boca has its moments, but it’s delightful compared to the insanity of driving in Miami. So this is where I started to suspect Timmons might be making things up. Did Timmons use data for all of Boca Raton, but compare to the population of just the city of Boca Raton (85,000 people vs 215,000 people)? No, it’s not that sensible.
I checked for Florida traffic deaths and found this pdf report for 2017 from the FLHSMV. Right off the bat page 12 of that pdf indicates there were 986 traffic deaths with alcohol, drugs, or both alcohol and drugs confirmed, nearly double the number used by Timmons.
Starting at page 45 is a breakdown of drug and alcohol crashes by county. Palm Beach County, of which Boca is a small part, had a total of 12 alcohol, drug, or drug and alcohol confirmed fatalities. Miami-Dade County had 30 such fatalities. Miami-Dade has twice our population but nearly 3 times as many DUI deaths, so on a county level Palm Beach is safer.
You can see the extreme variability when looking at those numbers. Broward had a terrible year in 2016 with 31 DUI deaths but only 13 in 2017. That’s not an improvement, it’s just the randomness of life and death in our world. Similarly in 2015 Miami-Dade had 48 DUI deaths.
Timmons says he used the FARS database, but it’s not clear whether his analysis of that data is accurate or if the data itself is accurate. It seems odd at least that he used the 515 deaths when FLHSMV reported nearly double that number.
I ran my own query on the database for Boca Raton (city code 290, county code 99) and found this
The right side shows that only one of the 13 fatalities reported for 2017 had police-reported alcohol involvement. It’s the far right column with nearly all indicating 0 (no involvement), one indicating 8 (not reported) and one indicating 1 (alcohol involved). It appears one other accident, number 7, may have also had drug involvement (THC, benzodiapenes and oxycodone). So that’s 2 fatal alcohol/drug involved accidents for the year for Boca Raton. The longitudes all seem to be inside the city lines, though that’s hard to tell. For a city population of 85,000 that’s less than the statistic reported by Timmons. If you take into account greater Boca Raton, including the unincorporated parts of West Boca, the number is closer to 1/3 what Timmons reported.
On the fun side the Herald comically reports:
The good news is that you’re at much less risk in Miami and Fort Lauderdale. And Miami Beach, Miami Lakes, Palmetto Bay, Florida City, Opa-locka and Sunny Isles Beach are among the safest cities.
Anyone who lives anywhere near Miami sees a warning flag in the claim that Opa Locka and Florida City are “safe.” In fact they’re among the most dangerous places to live according to … wait for it … yes, you guessed it … The Miami Herald.
A newly released ranking of the “most dangerous” cities in the U.S. lists two Miami-Dade County cities — Opa-locka and Florida City — in the top five … Opa-locka (2,807 violent crimes per 100,000 people) and Florida City (2,551 per 100,000 people)
I should also mention that Timmons ignores the statistical impact of pedestrian deaths. Pedestrian-vehicle accidents are far more likely to produce fatalities because cars are so much heavier and pedestrians are unprotected. So a city with more pedestrian activity is going to have more traffic fatalities. That doesn’t make drivers any less safe.
Altogether this “news” story getting attention from supposedly reputable newspapers shows the state of modern journalism. Pathetic.