<p>Phuriku:</p>
<p>Groovygeek’s argument rests on 2 main points:</p>
<p>1) that posters to CC are statistically representative of the entering class
2) that one can draw usable data from a sample pool ratio of 100:4500</p>
<p>Addressing point #1, I would say that the simple fact, as one poster pointed out, that there are many more Asian posters on the UChicago boards than there are in the student body at large skews the data fatally. Again, I don’t have the exact numbers of how many Asians/Whites/Black/Latino there are on CC, but this simply reinforces the point that there is no way that we can reliably argue that posters to CC are statistically and reliably representative of the student body at large. </p>
<p>Addressing point #2, yes, you can use a sample group of 2 if you’d like, but it would still give you an extraordinarily high standard deviation. Saying that one can reasonably draw a statistically reliable conclusion from a sample ratio of 100:4500 based on a highly questionable assumption (that the CC posters are representative of the Chicago student body), simply doesn’t hold any water.</p>
<p>Moreover, there is the further assumption that the acceptance criteria have somehow changed over time based on comparisons over 2 consecutive years. Again, working with a data set of 2 (2 entering classes, back to back) would leave you with an enormous standard deviation. You would need to compare entering class data over a much longer period of time to draw reliable conclusions. This would be an incredibly difficult exercise to prove and until one actually does some serious and clever data mining, there is no way that we can accept such a charge on anecdotal evidence.</p>