Even a sample size of 5 allows us to reject some claims. Consider claim - "this die is fair". If it comes up a six 5 times out of 5, we're justified in being suspicious.
Well sure, your example here is objective based on pure numbers. Statistically there is much less than a fraction of a percent of that occurring by chance alone, so being suspicious is absolutely justified.
But a dice is the antithesis of human behavior. The example I was talking about was almost completely subjective in nature and with 5 samples from an (assumingly) untrained observer (given he/she was a reporter, not a clinical researcher) leads to a whole lot of observation/experimenter bias.
So while my original if n=5, reject may not have been 100% accurate, in the case of human behavioral analyses, I stand by it.
