cat brain.log | less

Getting it down on `paper`

Ranking Colleges by Reading Level

Google recently released a new feature that allows you to see the reading level for a given domain. I was curious what this meant for universities, and wondered which universities’ websites had the most advanced reading level documents.

Top 10 (excerpt):

School Rank1 Advanced Intermediate Basic Num Results
Cornell University 1 98 1 1 34300000
Johns Hopkins University 2 75 21 2 36600000
Stanford University 3 66 28 4 28800000
Massachusetts Institute of Technology 4 60 29 10 21200000
University of California-Berkeley 5 52 38 8 14900000
Harvard University 6 91 6 2 12400000
Pennsylvania State University* 7 97 2 1 11300000
University of Michigan* 8 65 29 5 12400000
Westfield State College 9 41 49 8 12100000
Oregon State University 10 58 34 7 8780000

The results can be found here: University Ranking, Source Code.

My assumption is that the higher the reading level, the more intelligent the professors are at their respective universities. With this premise, one can use the search results to rank the universities accordingly.

This assumption comes with several caveats. One such example includes the situation where publications are not hosted under the university’s primary domain. While I haven’t found this to be the case, I cannot prove otherwise. The results are clearly biased in favor of universities with more publications online. This is justified in that the prodigious universities will publish more papers than their less substantial counterparts.

The ranking algorithm: (3*advanced + 2*intermediate + 1*basic) * num_documents

The primary caveat is that Google returns the percentage of documents for one of 3 reading levels, as well as the number of search results. By multiplying the result count by the percent, we calculate the number of documents at the advanced, intermediate and basic reading levels. However, we can’t simply sort our results by raw numbers alone. For example, a school that has 100 advanced level, 800 intermediate level, and 100 basic level documents should not be ranked lower than a school that has 101 advanced level, 100 intermediate level, and 800 basic level documents. In fact, the basic level documents negatively weights on the second school. For this to be the case, one must apply scaling factors to each of the reading levels. Unfortunately, these scaling factors may be selected arbitrarily, and may also be subjective.

Final caveat: The comparison is done on reading level and documents available online. For a school that emphasizes publishing papers, the results will be weighted in its favor. For a school that emphasizes the arts, the advanced reading level metric becomes less important.

References:
List of American Universities
Google Search [site:log.largevoid.com]

 

Comments

No comments so far.

(comments are closed)