<p><a href=“xiggi:”>quote</a>
Siserune, I’m not sure what was so vague in my comment. </p>
<p>For starters, it was a direct reply to. “Most PA-free rankings correlate with US News, so it would hardly be earth-shaking to drop PA or change its weight.” and my words were, “Siserune, the correlation of the PA and the ranking is circular, as the impact on the final rankings is so important.”</p>
<p>Is there really a need to explain this further? How hard is it to see that the PA (with its 25% weight) represents a major component and, unfortunately I may add, obliterates much of the more subtle differences that originate from the remaining 75%. The result is that a higher PA yields a higher total ranking. making a correlation circular, since it is exactly that higher PA that caused the higher ranking.
[/quote]
</p>
<p>The only way it would cause circularity is if the PA were so influential as to basically be the ranking, with the other components as low-impact window dressing. </p>
<p>That can actually happen, if the other 75% factors place all the schools in a very narrow range while PA varies widely, thus constantly serving as the tiebreaker. In other words, if the 25:75 split is fallacious and the actual effect (for example, the standardized beta-weights rather than the raw linear coefficients) of PA is to count as 90 percent of the rankings. </p>
<p>Part of my point about correlation was that this is unlikely to be the case. The other factors considered by US News are for the most part well correlated with each other and with the PA, so they will tend to rank schools in a similar way. Other rating systems (at least, those with goals similar to US News) also produce comparable lists, although using slightly different factors or different weights on the shared factors. One of the CC frequent posters, I think GAdad, mentioned that even “percent of on-campus students” will reproduce the US News ranklist; just as endowment per student or other objective measures will tend to tell the same story.</p>