How epidemiology and car sales data could have narrowed the Long Island murderer search to 5 suspects in about 5 minutes...but didn't.

Until yesterday, I’d never heard of the Gilgo Beach serial killings on Long Island. I lived in New York during the time of the murders, but I guess being a medical student meant I wasn’t keeping up with local news. (Fortunately, I do remember where I was when Sully landed that plane on the Hudson, though, so I wasn’t completely shut off from the real world!)

Long story short: a group of linked cold cases has apparently been solved. Within a minute of reading about the case, though, I wondered why it had taken so long for the investigators to find their man. After all, the suspect had two features that should have narrowed the search from 1 in 1,500,000 to 1 in 5 (or maybe 10). With a little epidemiology and market research, detectives may have been able to get there.

Here’s how….

First, the suspect has a rather unusual epidemiological feature: he is at least 6’4 in height. Investigators were told this by a witness. What’s more, that witness told investigators that the potential suspect drove a green (or dark) Chevrolet Avalanche. To me, just that information was akin to checkmate in three moves.

Using epidemiology and car sales to narrow your search for a killer.

Let’s think about how powerful just the information we’ve discussed so far would and should have been to investigators.

Source: US Census.

As you can see, just about 1% of adult males in the United States are 6’4 or taller. Still, that’s a lot of people. So, how common were Chevrolet Avalanche’s back in 2010? Well, we’ve got that too.

In 2010, the Chevrolet Avalanche was the 131st most sold car in America. Among the 11.5 million vehicles sold that year, just 20,515 were that make, or 0.177% (1 in 565). Assuming 0.18% of cars on the road were Chevy Avalanches, we can just do basic multiplication to figure out how many leads this information could have identified.

If you are a detective, the key is narrowing things down to a manageable number of suspects to really investigate. 500 would be way too many. 100 would still be a heavy lift. But what if you could get it down to just 10? It would be worth a deeper dive into 10 suspects.

So…

Let’s run the numbers!

It’s time to play detective. Our biggest weapon? Math.

Assume the killer lived in Suffolk County, NY, where all the victims were found. Population: 1.5m

Assume the killer is a man (50% reduction of pool): 750,000 suspects remaining.

Assume the killer was aged 16-65 years (33% reduction of pool). 500,000 suspects remaining.

Assume the killer was 6’4 or taller (99% reduction of pool): 5,000 suspects remaining.

Assume 0.2% (rounding up) of registered vehicles are Chevy Avalanches (99.8% reduction of pool): 10 suspects remaining.

Assume 50% of cars are anything close to a dark color (because they are): 5 suspects remaining. (Note: if you assume green, as the witness said, you’re probably done).

So even before a shred of DNA or other evidence, we’ve got a chance here.

Now you just head on over to the state’s department of motor vehicles with your handy subpoena and have them run a search. How many licensed drivers are listed as 6’4 or taller who also have a Chevy Avalanche registered to them in Suffolk County, NY? (Ever noticed that your license has your height on it? The DMV knows how tall you are). “Oh, and what are their names?” you ask. “How many of them work in Manhattan?” (Investigators knew the suspect made phone calls from there).

Pretty soon, you’ve got your prime suspect.

From here, you can start really poking around. Find out if any DNA on a pizza box discarded by your prime suspect matches DNA from a hair found on a victim’s body (this is indeed what detectives did last year, when the case got new life). It’s tough because of the decay to those bodies (they were not found immediately. But there’s enough mitochondrial DNA to sequence and get something meaningful. The results of that DNA test can rule out 99.96% of the population (all but 1 in 2,500). But it doesn’t rule out your suspect! That’s good news. The case seems to be getting stronger and stronger.

Now you start digging through phone records and internet searches. It’s all what you’d expect from someone this creepy and evil. It’s not for the faint of heart. (Also, it doesn’t help the guy’s case that he was constantly Googling for updates on how the investigation into the women he has been accused of killing was progressing.)

What if the approach I advocated for leads nowhere? Well, then you re-examine one of the assumptions. Or you follow other leads. The point is, that if the above approach had been taken, this case could have been solved a long, long time ago. (It bears mentioning that once the detectives started looking into the Chevy in 2022, the rest of the investigation appears to have played out extremely effectively—almost virtuosically in some aspects. They’ve got cell phone pings from his phone and the victim’s phones in the same place. They know the suspect’s wife was always traveling when the murders occurred. It goes on and on.)

The importance of being average.

Things are not always “this easy.” But this suspect made two “mistakes.” First, he decided to be 6’4. (I jest, but this fact is a major part of this). Second, he bought a Chevy Avalanche instead of the Ford F-Series, the latter of which was the most commonly bought vehicle in America in 2010 (and still is now). If the suspect had been 5’10 and driven a Ford F-150, he might be free today. Around half of all US adult males are between 5’8 and 5’11, and around 4.6% of all cars on the road are Ford F-150s. That would leave around 11,500 suspects in Suffolk County, NY—and still 5,750 if you limited the color to the darker hues. The detectives would have had a much harder time starting with >5,000 suspects than they did with 5, 10, or even 50. Being tall and buying an Avalanche are two reasons this suspect is likely to spend the rest of his life in prison.

So why did it take over a decade for an arrest to be made? The crux of it seems to be when the investigators gained knowledge of that pickup truck. In a bizarre part of the story (see page 7 of the case against the defendant), the suspect met with one of his victims once before murdering her on a subsequent meeting. During the first meeting, though, he was seen by people other than the victim. One witness of the first meeting provided investigators with information about the suspect’s height and that his vehicle was a Chevy Avalanche. (How the witness happened to notice the car make and model impresses me, by the way. We entertain guests in our home from time-to-time and I couldn’t tell you what car most of our friends drive, let alone people who only visited once after sunset.)

If it turns out that investigators knew either of these two key facts about the suspect back in 2010, the delay in solving the case is 100% on them. But if that information was somehow never solicited from the witness, or was somehow obscured (I read somewhere that the vehicle may have been registered to the suspect’s brother in the past, which, if true, could have thrown the cops off the scent; It’s unclear), it would be less embarrassing.

The lesson? If you want to get away with murder, be average. Or maybe, you know, just be average by not murdering anyone.