Prepared for Professor Linda Zhao’s course in Modern Data Mining (Stat 471, University of Pennsylvania, Spring 2016) and submitted as its final assignment. The author wishes to express his gratitude for the instructor’s (and assistants’) guidance and patience throughout the semester.

1 Introduction

University ranking systems are widespread and widely accepted as useful indicators of higher-education value. International lists gravitate toward the United Kingdom’s Oxbridge and American technology institutes ; in the U.S., the most famous ranking body, U.S. News and World Report, garners such demand that it charges a $30 annual subscription service for the full version of its considerations. Global university rankings reflect a number of implicit assumptions about education-related values and arguably influence admissions policies for many schools.

Most of these ranking systems rely heavily on admissions selectivity, admissions yield (how many of those admitted enroll), test scores of enrolled students, and research output of faculty. These heuristics obviously privilege those elite institutions with high volume of applicants and high standards for faculty research-that is, the Ivy League and similar.

This project proposes to consider instead two classification regimes concerned with service-learning and community engagement, a “broader mission” of higher-education institutions. In particular, the study will an exploration and analysis of the Community Engagement classification system (Carnegie Foundation for the Advancement of Teaching, now maintained by the New England Resource Center for Higher Education) and the Service rankings by the Washington Monthly.

1.1 Motivation

The University of Pennsylvania was the first Ivy League school on the Carnegie Foundation’s published list. (Cornell was classified in 2010.) After some cursory examination of the most recent list, most of the recognized institutions are (a) public and state-affiliated, either land-grant flagship or satellite campuses; (b) small, private, and often wealthy liberal-arts schools; or larger schools in urban settings. Penn is also the highest-ranked Ivy for Service in the Washington Monthly list (at 61st, just above Dartmouth). Overall, neither regime aligns with U.S. News and World Report’s, indicating that the “best” universities may not be doing the most good.

Why does Penn qualify as a (moderate) exception to the rule? What factors influence a university’s community engagement programming, classification, and rank? Copious data is available from the Department of Education’s College Scorecard, its Campus Safety and Security Data Analysis Cutting Tool, and the Carnegie Classification of Institutions of Higher Education.

A personal impetus for this project stems from my enrollment and interest in a course last semester on Urban University-Community Relations, focusing especially on Penn’s complicated past in West Philadelphia and the future of so-called Academically-Based Community Service (ABCS) courses.

1.2 Hypothesis

Those universities score better if located in urban environments (perhaps higher crime rates), with larger proportions of low-income, first-generation, and minority students, and with more students majoring in non-business fields, especially social policy or education. The primary goal is to understand principal components underlying community-engagement recognition; the secondary motivation is to create an effective classifier modeling inclusion in the Carnegie Foundation’s list (0 or 1) and to find under- and over-performers compared to the Washington Monthly rankings. Is Penn’s commitment to community engagement above the curve, in line with reasonable expectations?

2 Data Sources & Preparation

2.1 Carnegie Community Engagement classification

The Carnegie Foundation has classified 361 American higher-education institutions of various features as committed to community engagement, operationally and curricularly.

“Community engagement describes collaboration between institutions of higher education and their larger communities (local, regional/state, national, global) for the mutually beneficial exchange of knowledge and resources in a context of partnership and reciprocity.

“The purpose of community engagement is the partnership of college and university knowledge and resources with those of the public and private sectors to enrich scholarship, research, and creative activity; enhance curriculum, teaching and learning; prepare educated, engaged citizens; strengthen democratic values and civic responsibility; address critical societal issues; and contribute to the public good.”

More information is available on the <a href=“”>NERCHE website. There are understandable problems with the data – the classification is voluntarily pursued by colleges and universities, not as an award but as a step toward institutional progress. (Still, we might presume that a college which does not pursue Carnegie recognition would not qualify for high levels of engagement.)

2.2 Washington Monthly

The mission of the ranking system is explained on the website: “Unlike U.S. News and World Report and similar guides, this one asks not what colleges can do for you, but what colleges are doing for the country. Are they educating low-income students, or just catering to the affluent? Are they improving the quality of their teaching, or ducking accountability for it? Are they trying to become more productive-and if so, why is average tuition rising faster than health care costs? Every year we lavish billions of tax dollars and other public benefits on institutions of higher learning. This guide asks: Are we getting the most for our money?” Specifically, the Washington Monthly considers the following to develop “Service” rankings for 283 U.S. institutions: Peace Corps and ROTC involvements (rank); federal work-study funds spent on service; community service participation and hours served (rank); and service-related and financial aid resources (rank). While the service-exclusive rankings are not published explicitly, the website’s sorting tool allows for implicit ordering before downloading as an Excel sheet. A full list of variables and descriptions is available in the appendix.

2.3 Department of Education

The Campus Safety and Security Data Analysis Cutting Tool: Data from all postsecondary institutions, for all on-campus and noncampus violent crimes, summed across 2013 and 2014 but excluding non-forcible sexual crimes (incest, statutory rape). From the website, run by the Dept. of Education: “The data are drawn from the OPE Campus Safety and Security Statistics website database to which crime statistics and fire statistics (as of the 2010 data collection) are submitted annually, via a web-based data collection, by all postsecondary institutions that receive Title IV funding (i.e., those that participate in federal student aid programs).”

**Carnegie Classifications & Dept. of Education Data:** Fully described in its Appendix, this data (provided by the Dept. of Education and curated by Carnegie Classifications of Institutes of Higher Education) includes a wide variety of publically accessible information, most of which classifies schools by location, size, tuition, research level, et al.

A good deal of data cleaning was required. This represented the brunt of time which the author spent on the project, in fact. All data was labeled and sorted by the Department of Education’s “unitid” unique identifiers. Original and “clean” spreadsheets have been included in the final submission.

2.4 Nationwide Distribution

Which colleges are the focus of this study? Those which had (publicly available) rankings in US News and World Report, were featured in the Washington Monthly’s alternative rankings, and for which crime data was available from the Department of Education. All told, 182 schools qualify, a list that includes all eight Ivy League institutions and most other recognizable names.

A plurality of institutions represent California, New York, Massachusetts, Texas, Pennsylvania, Ohio, Illinois, and North Carolina. These states include some of the largest and best-educated populations in the Union. Generally, too, the Northeast and Mid-Atlantic regions (along with California) have some of the highest-ranked universities in the nation.

U.S. News & World Report, Top 50