<p>
According to my data (from the AEI study), Juniata’s graduation rate is 76%. That’s above the Pennsylvania average (64.6%), but still not especially high. The University of Chicago has many required courses but its graduation rate is 90%.</p>
<p>If you want to compare Juniata to another small college, consider St. John’s College, where virtually all courses are required. Its graduation rate is not all that much lower than Juniata’s, at 71%.</p>
<p>On the other hand, Chicago’s rate is a little lower than some of its peer schools, and St. John’s rate is much lower than most of its peer schools (schools of similar size, cost, and SAT scores or selectivity). </p>
<p>Reed (76%, much lower than many of its peers) has at least one strong course requirement (their Humanities 110 sequence). More significantly, it has an unusual oral comprehensive exam requirement at the end of the 3rd year. My understanding is that questions can be asked on the content of any course the student has taken. After that hurdle, there is a major senior thesis requirement.</p>
<p>All three of these schools, by the way, have among the highest per capita rates for Ph.D. production. They are hard, but good.</p>
<p>Student body is still probably one of the highest correlations. Chicago and Reed have significantly stronger students than Juniata.</p>
<p>It would be interesting to compare to a small private religiously-affiliated school, which often maintains a core curriculum.</p>
<p>I entered the percent of classes with less than 20 students into my data. It was correlated .51 with graduation rates. I am working on an urbanization variable.</p>
<p>There is a variable in the IPEDS database called “degree of urbanization”.</p>
<p>11 City: Large<br>
12 City: Midsize<br>
13 City: Small<br>
21 Suburb: Large<br>
22 Suburb: Midsize<br>
23 Suburb: Small<br>
31 Town: Fringe<br>
32 Town: Distant<br>
33 Town: Remote<br>
41 Rural: Fringe<br>
42 Rural: Distant<br>
43 Rural: Remote<br>
-3 {Not available} </p>
<p>Variable Description
Locale codes identify the geographic status of a school on an urban continuum ranging from “large city” to “rural.” They are based on a school’s physical address. The urban-centric locale codes introduced in this file are assigned through a methodology developed by the U.S. Census Bureau’s Population Division in 2007. The urban-centric locale codes apply current geographic concepts to the original NCES locale codes used on IPEDS files through 2004. </p>
<p>I entered this into my data but the correlation with graduation rate was only -.05, very weak. But it seems to add significantly when combined with other factors.</p>
<p>I did a stepwise multiple regression and “degree of urbanization” was one of the variables selected. Overall, the regression accounted for 86.5% of the variability in graduation percent using the following factors:</p>
<p>SAT CR 25th
yield
public or private
peer assessment
size of freshman class
percent of students in science, engineering, comp sci, math squared
SAT CR 25th squared
admissions percent squared
percent os students receiving grant from institution
average amount of grant from institution
percent receiving aid
degree of urbanization
percent of classes with under 20 students</p>
<p>SAT CR 25th accounted for 69% of the variability by itself.
PA accounted for an additional 7%.
yield added 3%
public or private added 1.5%
SAT CR 25 squared added 2.5 %
urbanization added about 1.5%</p>
<p>The above 6 variables alone account for 85.6% of the variability</p>
<p>interesting re: yield contribution.</p>
<p>Yes, yield must account for something unique.</p>