[olliesaunders]

The data really doesn’t look that statistically significant to me: http://hbr.org/2011/06/defend-your-research-what-makes-a-tea...

[trevelyan]

N = 192 teams.

[tst]

That doesn't mean so much if you don't know the effect size and variance. In the end you have to look at the confidence intervals.

[catch23]

So is 192 the magic number? As long as we get 192 samples, we have enough data for our experiment!

[trevelyan]

When I did quant in grad school the magic number was around 30 in the sense you could presume anything less wasn't going to have much power. 192 passes my smell test for an expected distribution of classroom grades. By all means check the research, but I'd suspect "correlation is not causation" and "not a random sample" are more likely problems.