Our curious result from our research is that having a higher level of average caps had a negative influence on a team’s prospects of advancing to the quarter-final and semi-final. Also, we could not find empirical support for the claim that a higher level of caps is key to eventual success in RWC – which is not to say we can discount the notion, it just means we did not find support for it.
The context is that it has frequently been suggested in the past that experience counts in RWCs. Is having a total of approximately 600 caps in the team an optimal number? For more background on this, see these ESPN, Telegraph Rugby Average age and caps, and the Roar articles. Also h/t to the New Zealand Herald for penning this recent article on Records that could fall at the Rugby World Cup.
So naturally, we were keen to examine average number of caps during our research process. What did we find? Well, we found a curious, seemingly counter-intuitive result, and again with clear signs of stage effects.
The particular result was that in the early stages of the tournament, teams with a higher level of average caps, were less likely to progress. Perhaps contrary to common wisdom, we could not establish an effect in the latter stages of the tournament – i.e. of itself, increasing a team’s average level of caps did not seem to affect its success prospects later on in the tournament.
So what could be going on? On reflection, we rationalise the findings as follows. It could be that weaker teams have smaller playing pools, so simply they don’t have the necessary talent available to replace their superstars in the latter stages of their careers. As a result, their team’s average caps level increases, but their likelihood of progression reduces.
What is much harder to rationalise is that the effect seemed to disappear as the tournament progressed. On the face of it, our results did not establish that the average level of caps had the positive effect on results that might be expected. So we can’t say that it had no effect, we just couldn’t establish that there was an effect. It may be that there are small sample problems at play (as we have cautioned previously). It might be that if teams in the latter stages tend to have equivalent levels of average caps, that there is no difference between them in that regard.
We conclude for the moment that this is something of a puzzle.