Pole Position

[beelsebob]

The data there is too much summarised unfortunaly... . Wow I completely underesitmated this; I would need to have the data from every race and if the pole sitter had technical problems that race.

Well, firstly it would be interesting to see whether there were more race winners than pole-sitters. If there's no significant difference might not be worth looking at.

Huh? Every race has a pole sitter and unless the race was cancelled, every race had a winner. There will be no significant difference between those 2 numbers; that does not mean it is not worth looking into it.

Yes, but lots of races might have the same winner, or the same pole sitter, meaning that for n races, you will have between 1 and n race winners, and between 1 and n pole sitters. Those numbers may well be significantly different.

[timbo]

What beelsebob said.
Consider this, if you have poor reliability overall, than qualifying results may represent overall driver+car performance better. So you can have one driver getting lion's share of poles, but if he can make it to the finish in only half of the time others will have a chance for the win.
So, if reliability plays significant role in driver's pole/win ratio, we should have fewer pole-sitters than winners.

[turbof1]

What beelsebob said.
Consider this, if you have poor reliability overall, than qualifying results may represent overall driver+car performance better. So you can have one driver getting lion's share of poles, but if he can make it to the finish in only half of the time others will have a chance for the win.
So, if reliability plays significant role in driver's pole/win ratio, we should have fewer pole-sitters than winners.

Exactly. Apperently we were just saying the same. That's also why I need for every race every pole sitter and every race winner. Perhaps also other factors like sickness, crashes,... . Just to make sure not having too many omitted variable biases.

[timbo]

Perhaps also other factors like sickness, crashes,... . Just to make sure not having too many omitted variable biases.

I don't think that's important at least at the initial analysis as such factors will probably average out.