This year's rankings are based on reputational surveys completed by 266 philosophers throughout the English-speaking world (an increase of nearly 50% from 2002). Evaluators were asked to evaluate faculty lists that did not include the university's name along the following dimension:

Please give your opinion of the attractiveness of the faculty for a prospective student, taking in to account (and weighted as you deem appropriate) the quality of philosophical work and talent on the faculty, the range of areas the faculty covers, and the availability of the faculty over the next few years.

Because evaluators do not have reliable access to information about the quality of graduate teaching and mentoring for most departments, this was not a component of the evaluation. As always, students are advised to talk to current students before enrolling at any program.

Evaluators scored departments on the following scale: 5-Distinguished; 4-Strong; 3-Good; 2-Adequate; 1-Marginal; 0-Inadequate for a PhD program. .5 fractions were permitted. There is some difference of opinion as to how faculty size affects the ratings. Many people think (and the editor concurs) that larger faculties fare better in reputational surveys; some think this is an unfair bias (it is simply more likely that an evaluator will know someone on a large faculty), while others think this is a warranted advantage (larger faculties have more to offer). Others believe that per capita quality matters more than size, so that large faculties are at a disadvantage insofar as they may have more underproductive members. Readers may consult the faculty lists used in the evaluation to see how faculties differ in size and composition (i.e., the percentage who are part-time, or affiliated, and so on).

We present, below, two alternative overall rankings of departments, one based on the "scaled" mean score for each department, and one based on the raw mean score. Scores were standardized, to offset any slight distortions resulting from evaluators awarding different average scores (some gave lots of 4s and 5s, some gave almost none), and from not all evaluators scoring all departments. Standardizing the results puts all the evaluators on the same scale. First, the scores were centered by subtracting the mean score for each evaluator from each of that evaluator's scores (each observation); then, each observation was divided by the root-mean-square of that evaluator's scores. After scaling, the distribution of each evaluator's scores has a mean of zero and a variance of one. Note that the scaled scores are for each region, and are not comparable across geographic regions.

While scaling corrects for some distortions, it also introduces some, once again attributable to not all evaluators scoring all departments; when this selective scoring has a fairly systematic pattern (say, the evaluator mostly scores "top" departments, but scores a few lower ranked departments as well), the scaled score of the lower ranked departments will be artificially depressed relative to the raw mean.

In the first ranking, the rank of each department is based on the "overall" scaled mean, i.e., the standardized score of the faculty as rated by all evaluators. The overall scaled mean is followed by the "raw" (unscaled) mean based on all evaluators and then by the total number of evaluators who scored that department. In the next two columns the department's rank in the 2002-04 and 2001-02 Report is given (note, however, that scores were not standardized those years, though scaling, as one can see from the raw scores, has only a modest impact at the margins). The "local" scaled and raw mean reflects the scores given only by evaluators in that national region. There were 161 American evaluators who scored U.S. departments; 49 U.K. evaluators who scored U.K. departments (a 75% increase from 2002); 19 Canadian evaluators who scored Canadian departments (a 25% increase from 2002); and 15 Australasian evaluators who scored Australasian departments (a 90% increase from 2002). In the second set of rankings, the rank of each department is based on the "overall" raw mean, and then the rest of the data is as before. The rankings are unchanged for Australasia , whether raw or scaled mean is used, so we print only one set of rankings for that region. For purposes of comparing rank this year with prior years, it makes more sense to look at the raw mean ranking. For purposes of comparing departments across geographic regions, one must also look to raw mean, since scaling was done relative to departments in that geographic region.

Rankings by Scaled Mean

Note: Although housed at separate institutions and evaluated separately, St. Andrews and Stirling run a joint graduate program.

Rankings by Raw Mean