Sunday, November 14, 2010

Separating "physique" and "skill" effects

Sometime back, I had posted on the impact of Sachin Tendulkar's reputation on his opponents. In that context, I had argued that the physical size and on- and off-field behaviours (in addition to their obvious skills) of players like Hayden and Richards may have contributed substantially to the overall reputational intimidation (or mental disintegration) exercised on their opponents.

Amol Agarwal had mailed me an interesting question about how to separate the relative contributions of individual skill and physical factors towards intimidating opponents. In other words if...

Net intimidation, I = intimidation due to "skill-effect", I(s) + intimidation due to "physique effect", I(p)

I, say, for a superstar batsmen, is a function of the difference in the performances (in terms of runs per over or strike rate or even bowling averages) of the opposing bowlers against the superstar and the other regular batsmen. But the big challenge is how to separate I(s) and I(p)? For simplicity, let us also assume that I(s) also includes the contribution of skill-itself (in addition to intimidation due to the over-sized skill).

I have thought long and hard and have not been able to come with a satisfactory enough solution. But any answer(s) would surely be interesting. Are there any instrumental variable that can be used to separate the two effects?

3 comments:

Sridhar said...

how about a regression on:

Player i's strike rate against a specific bowler = f(batting average of player i, weight*height of player i)

Anonymous said...

You are attempting to measure an unmeasurable and varying affect (on some people) due to unmeasurable and varying factors.

If u can solve this, u can solve the current economic recession.

gulzar said...

sridhar, it would still not be able to separate out the two effects. the regression would give only a measure of the combined effect.

anon, i tend to agree with you. however, there is atleast one measurable here - the physiques of the batsmen - apart from the total effect (difference in the averages etc).

i was thinking about the possibility of proxies - eg. performance against teams which do not know the batsman (but in this age, the superstar's fame travels ahead of him!), against superstar bowlers (who are less likely to be intimidated by his skill than by physique?) etc.