Life risk in early motor racing

[pertti_jarla]

Once again, from the preface to the Hill-Von Trips book "The Limit", I read the claim that in those days a racing driver had a 33% survival rate. A few years ago I did some counting on the chances of a 50's and 60's championship point scoring F1 driver getting killed racing/practising/testing. Roughly 1/3 seemed to be the chances of getting killed, not surviving. The risk was slightly higher in the 1950s, but still about 1/3.

I have lost my original calculations, but today I listed all drivers taking part in the 1950 F1 season, not including Indy 500. Of all the participants 10 out of 46 got killed racing/practising/testing a racing car that year or in the future. The risk for championship point scoring drivers was higher, 6 out of 16 got killed, that is 1/2,7

1960 season: of all F1 racers (excluding Indy) 12 out of 57 = 1/4,75, of championship point scoring drivers 6/21 = 1/3,5

So, I wonder where that "33% survival rate" comes from? Of course I have only listed the F1 participants.

[555-04Q2]

I think they were just throwing a random number around. If 1 in 3 drivers were dying, the sport would have died along with it, not to mention after 3 seasons you could possibly have had none of the original drivers from the first season left!

Sounds like and old wives story to me ;)

[pertti_jarla]

Yes. The exact quote is: "Between 1957 and 1961 twenty Grand Prix drivers died. Many more suffered terrible injuries. By some estimates, drivers had a 33 percent rate of surviving"
If you give a survival rate like that, you should also give some kind of a time frame, for example "a 33 percent rate of surviving five years". After all, ultimately we have a 0% rate of surviving. If that 33% refers to surviving the years 1957-61, it seems like nonsense. This would have been easy for the author to check.

[pertti_jarla]

I did some counting on the years 1957-1961.
-During those years 116 drivers started in F1 races, not including Indy 500, including non-finishers. Of these 24 lost their lives in racing related accidents during this period or later, that is 1 in 4,9 or about 80% chance of surviving.
-40 of these drivers scored F1 championship points during their careers, 15 of these got killed racing/practising/testing/ during this period or later, that is 1 in 2,7. Which is quite a high risk.
-Of all these 116 starters 14 lost their lives during the period 1957-61. The claim "twenty Grand Prix drivers died" could be including Indy drivers.

[Mintexmemory]

History is littered with 'facts' established by poor understanding of statistics by authors or poor translation into English. It's amazing how many people can't get conversion of fractional probability into % probability correct.
Did you count all F1 races between 57-61 or just World Championship races? What may also be confounding data is the presence of a GP driver (pre-57) who is killed during the period under study but hadn't driven in F1 during that period. Does Hawthorne's death figure in your numbers - death in a vehicle accident but not racing?

[Mark]

A 33% survival rate means that that chances of you being killed was 66%, which doesn't seem right. There were units involved in D-Day which had a better survival rate than that.

[anfield5]

Quite, getting killed dead twice in every three races seems a bit harsh.

I think (not sure) that the quote wasn't meant to be the odds of a particular driver meeting his/her maker, but any driver i.e in any given race there is a 66.67% chance of a driver being killed. If there are 20 starters in said race, the odds of a specific driver dying would be 1 chance in 30 or a 3.33% chance of death (worked out by multiplying the 2/3 chance of a death by 1/20 as a given driver is 1 of 20 in the race). This figure seems more realistic and probably quite accurate, obviously the more drivers in the race the longer the odds of a specific driver dying.

A specific example could be the indy 500 where there were 33 starters. 2/3 x 1/33 = 2/99 or a 2.02% chance that any given driver would perish in the race, rather than indicating that 22 of the 33 were not expected to make it home.

[Mark]

I think it was in a season, rather than in a race, but still.

[anfield5]

I think it was in a season, rather than in a race, but still.

You are more than likely correct.

Then in a 15 race season with 20 starters in each race you would multiply 2/3 by 15/300 (a specific driver starts 15 times out of 300 starters) so the maths is the same i.e 15/300 is the same as 1/20. :)

[pertti_jarla]

As I said, talking about chances of survival is a bit meaningless unless you specify a time frame. I counted all drivers who got killed directly because of an accident in any kind of racing car, racing/practising/testing, no matter when it happened (including, for example Bonnier who lost his life in 1972). I left out traffic accidents, like Farina and Mike Hawthorn, but there were only a couple of cases like that. I also counted separately those who got killed in 1957-61.
I only included drivers who had taken part in F1 championship races, because "The Limit" talked about "Grand Prix drivers", but accepted all kinds of races as causes of fatality.

[D28]

Thanks, it is essential in such analysis to clearly define what is being counted in the sample, as you have done. Some lists of F1 deaths simply ignore any fatalities outside of F1, thus missing Clark, McLaren, Pedro Rodriguez and others.
Looking at the quote,
"Between 1957 and 1961 twenty Grand Prix drivers died. Many more suffered terrible injuries. By some estimates, drivers had a 33 percent rate of surviving"

I found a source that largely confirms the first part, it lists 21 GP drivers killed in the period, for all motorsport, excluding Indy.
The problem is the second assertion beginning with "by some estimates". A common mistake is to extrapolate a simple fact into a statistical rate without any background calculations. We don't know who or how such estimates were made, but the statement is false which you have clearly pointed out.
I don't what the precise calculated death rate would be, but much lower than 66%.

[Mark]

It would be interesting to compare the deaths in all motorsport. Which would be difficult to compile. But when you add in all motorsport and not just F1 you'll see a picture of a much more dangerous sport.

[D-Type]

Figures for all motorsport do exist . The most comprehensive database I know of is that compiled by Motorsport Memorial (motorsportmemorial.org) but it is not possible to filter easily or count fatalities. It may also be possible to compile figures from coroner's courts but I don't know if anybody has tried. Someone may have done an unpublished PhD study but I doubt it. The main difficulty with these global statistics is relating them to the number of participants or the number of starts. It might be possible to make a stab at total participants by totalling the number of competition licences but even that would be incomplete. The number of starts can only be an estimate to try and allow for some, particularly amateur, drivers starting only a few races and some, like US sprint and midget racers, who raced several days a week.

I think the original statistic was meant to be a 33% chance of being killed rather than of surviving. I think it also related to a time base of a whole career and includes all forms of racing. If you take a typical 1950s or 1960s grand prix grid and count up the number of drivers who would be killed at the wheel of a competition car and you'll get a figure of somewhere around 1 in 3 or 1 in 4 depending on which you choose and your exact criteria..

Here's a couple of figures based on points scoring drivers as data is readily available and points scorers is closely related to regular competitors. In his Grand Prix! books, Mike Lang includes a list of points scoring drivers. The 1950-1965 book lists 105 drivers, at least 33 were killed driving a racing car (31%); the 1966-1973 book lists 70 drivers, at least 18 were killed at the wheel (26%); and the 1974-1980 book lists 55 drivers, at least 8 killed (15%). I say 'at least' as Lang lists only those killed in the period covered by the book and I have had to add those from later and may have missed one or two.

[D28]

"I think the original statistic was meant to be a 33% chance of being killed rather than of surviving".

Right, the comments by others above suggest that the author has the survival and death rates roughly reversed.
Autor Michael Cannell is described as a sports writer for well known New York publications and editor of the New York Times House and Home section.
He admits to not being a "car guy" or a racing fan. The question raised is would the mistake be intentional, or is it a misprint.
Given his writing experience with reputable journals, I would doubt that is a mistake. Excerpts of the book are available on line for free; from what I read, I would expect it to have limited interest to readers here.