It could be, but we won't know until we run them.Lycoming wrote:I think the sample size is too small, even going by number of races between the two of them since 2011, it probably isn't that unlikely to have such an unequal distribution of failures. That said, I havn't actually run the numbers, but there are ways of doing so, but I think you need to come up with some sort baseline KERS failure rate first.
Any reported failure in KERS that affects either race or qualifying performance would be a definition, to simplify we can take that to mean a KERS failure of any kind during Q1/2/3 and the race. That seems like a fairly decent and easy to interpret definition.
If we consider a KERS failure as our only dependant variable then it could be pretty straight forward hypothesis, there may well be more than one dependant variable involved, but initially you would want to see if there was simply some kind of failure rate difference:
-If yes you can explore further.
-If no then no exploration is possible until there is a statistical difference.
Our n number would simply be the number of races since the reintroduction of KERS.
In which case our hypothesis is that:
''Webber has a statistically greater failure rate of KERS than Vettel.''
...And our null hypothesis is therefore that:
''There is no statistical difference between the failure rate of KERS between Webber and Vettel.''
The frequency of failure rate is simply measured per race, with each race counting as a separate data point.
From there one would apply a statistical test, a Chi-squared test is where my stats book is directing me, thought I had initially thought a simple T test though?
