<p>UCLAri,</p>
<p>sorry, i was pretty busy lately and im going to my first interview tomorrow, so i probably wont have as much time.</p>
<p>but anyways, i agree with you. so i suggest to compare stats from rejected students.</p>
<p>“If that’s the population, then it’s the population” (btw how do you quote here?)</p>
<p>i have to disagree on that statement. because there is always a thereotical mean, and the population we can see is the estimate of the thereotical mean. thereotical mean is constant whereas the estimated mean can change depends on the size of sample. so the accepted population is not fixed by theory (it hasnt reached the thereotical maximum). for example you use that mean to estimate if another person has some number in his GPA and MCAT, how likely is he accepted to that school. so what i was saying is that since the sample size is fairly small, the standard deviation is very large, and null hypothesis is unlikely to be rejected.</p>
<p>i dont find comparing mean of the entire berkeley accepted students to a particular school make too much sense, because standard deviation in berkeley accepted students are higher than means of a particular school, despite of consistency over the years in both x- and y- axes (i only can assume you mean GPA and MCAT scores). please remember that difference in means do not represent dynamic of distribution over the years. if the shape of distribution, not mean is consistent over the years, that would be more convincing.</p>
<p>its such a shame that they do not publish std deviation, because comparison of means is meaningless without proper stat treatment.</p>