Wake Forest Taser study - 0 (or 2?) out of 5

~UPDATED~ The summary that I read at the Advocate didn't indicate that they'd found anything significant (à la Ho). I had to dig further to find that there actually were two fatalities included in the study, but they were barely worth mentioning. Given the Taser Proof Issue, they shouldn't be quite so quick to sideline fatality data.


Advocate staff writer Mr. Steven Ward wrote an article [LINK] contrasting the use of tasers in Baton Rouge and other parts of Louisiana, to the Wake Forest University School of Medicine 2009 study, "Safety And Injury Profile of Conducted Electrical Weapons Used By Law Enforcement Officers Against Criminal Suspects."

The Wake Forest study is a bit weird in that they gathered data from just six law enforcement agencies (quite lazy in my opinion) over the period June 2005 to June 2008.

They found very little to report on in the way of taser-induced injuries and deaths, which indicates 'something'. But what does it indicate? Taser safety? Or a flawed study?

Let's check...

UPDATED: The study included two taser-associated deaths, but these were discounted because the autopsy reports didn't link them to the taser. Duh! I guess that's good enough for those that haven't been following the news lately.

According to the List of The Dead [LINK] at the Truth ... Not Tasers blog, there were about 230 (#124 to #354) taser-associated deaths in North America during this period (including 11 or 12 people in Canada).

Let's focus on just the USA for this post. 230 subtract 12 in Canada = about 218 taser-associated deaths in the USA during the period June 2005 to June 2008 (inclusive). Let's round it down to 200 just in case we've included an extra month at one end or the other.

So we have (at most) two of these about 200 taser-associated deaths that made it into the Wake Forest study. Only two? [ding ding ding - sound of alarm bells going off...] What's going on with the Wake Forest study?

Is the study truly representative?


The six law enforcement agencies included in the report cover a total population of just a bit over eight million (8,059,001 according to the numbers provided in the left column summary here [IBID]).

US population during this period is about 305 million. So the study covers about 2.6% (8M/305M) of the US population. If the study was large enough to be representative, then we would expect that they'd also have about 2.6% of the taser-associated deaths included in their study. 2.6% of about 200 is about 5.2 (let's call it 'five').

A truly representative study would therefore be expected to include about FIVE taser-associated deaths in the data set.

But they have just two, and both were discounted into the disregarded column.

Ding ding ding.

Obviously the Wake Forest study may have been honestly randomized at the outset, but the intermediate results clearly indicate that the data is not representative of the much-larger, all-encompassing experiment that is being performed on the entire population of the USA.

Two isn't five. And given that AI reported that something like 37% of all taser-associated autopsy reports DID mention that the taser was a contributing factor, it seems strange that they would structure the study to avoid stumbling across this same ratio. [LINK]

Anybody have a parrot cage that needs some paper to line the bottom?


Many statistical analysis will include the mathematical disclaimer "...(19 times out of 20)".

The most generous interpretation of this Wake Forest Taser Study is that it just happens to be the 20th.


Advocate writer Mr. Ward then goes on to review some of the egregious taser incidents that have occurred in Louisiana, including the Winnfield incident where 21-year-old Baron 'Scooter' Pikes was alledgedly tasered-to-death. ("The coroner found the death of Mr. Pikes to be homicide by Taser." [LINK])

Winnfield, LA was not obviously included in the Wake Forest study.


Update details - Ah, they did have 'two'. Not the expected five, but two. "Two subjects died, but autopsy reports indicate that neither death was related to the Taser." [LINK] Given the Taser Proof Issue, then this may be an example of circular logic of the highest possible order.

Furthermore, I wonder how much denominator washing this study includes in the "nearly 1000 incidents" studied? How many were full-on taser hits to the chest? How did the details of the taser hits correlate to the 'un-related' deaths?


Commentary - It well past time to start discounting these sorts of studies with such limited input data. There exists a much larger experiment happening all across the USA and Canada and beyond. Hundreds and hundreds of taser-associated deaths. Autopsy reports reportedly mentioned tasers as a contributing factor in as many as 37% of the reports studied. Prof Savard plotted increased risk of death is linearly proportional to taser exposure. I'll bet that the X26 is more dangerous than previous taser models. I'll bet that taser hits to the chest are not-coincidentally far more lethal than hits to the back or extremities. Let's get the complete data and run the numbers.