Originally Posted by chasingthedon
Charles Davis looked into this in his book Best of the Best, limiting his investigation to those occurences between 1980 and its publication in 2000. There were 113 times when the nightwatchman was used and 89 similar situations when a nightwatchman was not used.
He found that
a) the averages of the nightwatchmen was virtually the same in either case (15.2 usual average vs 15.0 in nightwatchman situations);
b) the performance of the team overall was negatively impacted in the majority of cases. He based this on the fact that (for example at the fall of the second wicket) in 33 out of 38 cases where a nightwatchman was used at the fall of the second wicket, the final score of the team was lower than would have been statistically expected based on the score at the fall of the second wicket. In those 36 cases where a nightwatchman was not used at the fall of the second wicket, it was 50-50 vs the expected score.
For the fall of all first, second and third wickets, only 27% of the final team totals were higher than expected when a nightwatchman was employed.
That seems extremely flawed - all the data shows is that night watchmen are more likely to be used in low-scoring matches. As you'd expect, as when wickets are tumbling it's likely that one will be near the end of the day.