User blog comment:The Milkman/Citadel Review: Entertaining and Inconsequential/@comment-4599927-20130310233855/@comment-4121017-20130311172907

You're not going to like this, but Lily's right in this one. I did the calculations. I did a 1-Proportion Z test on my TI-83 Plus calculator (with the null hypothesis being greater than, say 75%, with the alterative being less than that percentage. IIRC from some of the Twitter feeds, the statistic is 65%, with the sample size 1000.) And let's make the alpha value is .05. According to my calculations, the p-value is 0, less than the alpha. That means the null hypothesis will be rejected in this case.

Lily, you said that there is a little more than a thousand, right? Did Clevernoob get his results from just his forums or from BSN as well? If so, the sample size shouldn't be bigger than the population.