What's Ahead for the Carnegie Classifications?
By Alexander C. McCormick

From the January 2000 AAHE Bulletin

Bringing the Carnegie Classification Into the 21st Century

The Carnegie Classification has served higher education well for three decades, but the time has come for a new approach that recognizes the many dimensions of commonality and difference among institutions.


It is time to rethink the Carnegie Classification system. To some, the very notion of changing the Carnegie Classification borders on heresy. But heretical notions are sometimes needed to bring about positive change.

The Carnegie Classification of Institutions of Higher Education was developed in 1970 by the Carnegie Commission on Higher Education (a project of the Carnegie Foundation for the Advancement of Teaching) to aid its policy research. The Commission needed a taxonomy that reflected the great variety of American higher education while providing "more meaningful and homogeneous categories than are found in other existing classifications." The "existing classifications" at that time typically differentiated institutions according to simple descriptors such as level (two- or four-year, with four-year sometimes divided into doctoral and nondoctoral) and control (public or private, with private sometimes divided into sectarian and nonsectarian). Such taxonomies failed to capture important differences, grouping very different institutions and separating institutions with much in common, such as public and private research-intensive universities.

The Classification was developed as a research tool. When the Classification was published in 1973, Carnegie Commission Chair Clark Kerr wrote that "the Commission decided to make the Classification available in published form because we felt that it would be helpful to many individuals and organizations that are engaged in research on higher education" (emphasis added).

In the three decades since its development, the Carnegie Classification has been adopted by a wide range of users: researchers, policy makers, funding agencies, institutional personnel, even news magazines. Institutions commonly invoke their Carnegie Classification as a concise and consensually understood means of self-description. Ironically, the Classification’s wide acceptance may be its greatest liability, as its present uses have far exceeded its original purpose.

Strengths and Weaknesses of the Present System

It is important to note the Classification’s strengths, which in large measure account for its durability.

  • The Classification is simple. The category definitions are straightforward, and require no complex statistical manipulations. It is easy to see how an institution’s classification follows from the data.
  • The Classification is objective. Institutions are classified according to public data on selected dimensions of institutional activity. In other words, they are classified according to behavior. To the extent that the Classification is thought to characterize institutional mission, it is mission as enacted by the institution.
  • The Classification is analytically useful. Countless studies have demonstrated the discriminant utility of its categories with respect to a wide range of institutional, student, and faculty characteristics as well as student outcomes.
  • The Classification provides a basis for shared understanding. It gives researchers, policy makers, and institutional personnel a common language for characterizing institutions, such that research findings and other analyses are readily interpreted and compared.
  • The Classification has the ring of truth. Many of those who know higher education find that the institutional groupings make intuitive sense — they hang together. There is disagreement on this assessment, however: Some argue that it is based on prominent institutions that are clustered in relatively small categories.

The Classification’s weaknesses fall into two broad categories: internal ones, which derive from the way the categories are defined and labeled; and external ones, which derive from the way the Classification is interpreted and used. Following is a brief summary of the Classification’s internal weaknesses. There are practical reasons for many of these, but a reasoned critique must acknowledge them.

  • Although the Carnegie Classification represented a great leap forward in describing the diversity of higher education in the United States, it is nevertheless a great simplification to distill that diversity into a handful of characteristics and a single way to group institutions.
  • The order in which categories are presented, combined with the use of numbered subcategories, permits the interpretation that the Classification values some categories over others — that it is indeed a ranking, with Research Universities I at the head of the list.
  • Although the Classification is often described as characterizing institutional mission, it does not attend to the traditional components of mission equally. Research is measured explicitly (if imperfectly); instruction is addressed only indirectly, through degree conferrals and field coverage; and service is absent.
  • Certain types of institutions are much more finely differentiated than others. Doctorate-granting institutions — 7 percent of the institutional universe in 1994 — are divided into four subcategories. Yet two-year colleges — 40 percent of the universe, encompassing a considerable diversity of institutions — are not divided into any subcategories.
  • With a few exceptions, the category definitions use absolute cutoffs — a particular number of degrees at a certain level, a particular number of fields represented, or a particular level of federal funding. While this is an element of the Classification’s desirable simplicity, it introduces a confounding effect of institutional size. As a result, average enrollment declines systematically from Research I to Master’s II (see graph, below). In The Rise of American Research Universities (Johns Hopkins University Press, 1997), Hugh Davis Graham and Nancy Diamond document the differences that result when one replaces an aggregate measure of research activity (as is used in the Classification) with measures that tap research activity on a per capita basis. For example, suppose two institutions with comparable sets of schools, departments, and programs receive roughly equal amounts of federal research dollars. If one has half as many faculty members as the other, the similarity with respect to aggregate funding may conceal considerable differences with respect to the activities of individual faculty members.
  • There are a number of technical shortcomings associated with the measure of research activity used in the Classification. While a thorough explication is beyond the scope of this article, a brief summary is in order. The Classification has traditionally used total federal obligations (as reported by the National Science Foundation) to capture research activity. While this source has the considerable advantages of reliability and objectivity (the data are reported by federal agencies, not institutions), it has several shortcomings as the sole measure of research activity. Not all research is federally funded (this measure favors fields heavily dependent on federal support, such as biomedical research); not all federal obligations are related to research; not all federally funded research is included; and these data cannot track the pass-through of funds between institutions. Each of these factors compromises our ability to assess an institution’s true level of research activity.

Next, there are external weaknesses — difficulties related to the Classification's impact and usage:

  • Its broad acceptance reinforces the notion that there is only one meaningful way to group institutions, and that those in different categories have little in common.
  • Its interpretation as a status hierarchy can have important consequences for institutions, especially their access to human and fiscal resources. As a result, many institutions; especially doctorate-granting institutions — have adopted "moving up the Carnegie Classification" as an explicit institutional goal. In effect, the Classification plays a role in shaping the system that it means to describe: The research tool has become an unintended policy lever. This has implications for institutions’ ability to innovate and, in combination with the uneven attention to the components of mission described above, raises questions about system-level change and adaptation. To the extent the Classification exerts normative pressure on the system, there are strong incentives for institutions to conform to particular models of institutional activity, and indeed to particular indicators of activity.

The last point is an important one. It is especially ironic in light of the recommendations made by the Carnegie Commission in one of the first reports to use the Classification, New Students and New Places (McGraw-Hill, 1971): to preserve or increase institutional diversity and resist homogenization. Indeed, the Commission discouraged expansion of "research-type universities granting the Ph.D." — the very model that many institutions emulate in response to the interpretation of the Classification as a status hierarchy. (This is not to dispute the great value and importance of the research university model, of course, but simply to highlight the unintended role the Classification has played in promoting it and thereby shaping institutional aspiration and action.)

Toward a More Flexible and Comprehensive System

Now let us look to the future and think about how the Classification might be adapted to the present realities of higher education, and how its weaknesses might be remedied. By 2005, we at the Carnegie Foundation envision replacing the present single scheme with a series of classifications that will recognize many dimensions of institutional commonality and difference. The first step toward such a system will involve the development of a series of indicators that describe institutions and what they do. This will simply be an extension and elaboration of the present approach, and will include improved indicators of research activity, but also indicators that characterize instruction and service activities. Each of these would likely be based on multiple measures, with appropriate controls for size as needed. We also envision indicators that capture important elements of institutional resources. After a "menu" of indicators is developed, they will be combined to produce a series of classifications. We also anticipate making the indicators available to users interested in creating their own, customized classifications. Another dimension of the new system will be differentiation of two-year colleges. (Several promising efforts are currently under way by analysts outside the Carnegie Foundation, and we are monitoring these efforts with interest and enthusiasm.)

A system of multiple classifications offers several advantages:

  • It acknowledges the complexity and diversity of American higher education, by recognizing that institutions that resemble one another when viewed through one lens may look quite different through another lens.
  • It better differentiates the classification enterprise from ranking: There will no longer be a single way to group institutions, exerting a singular homogenizing influence.
  • Users of the new classification system will have to think carefully about the dimensions of institutional variation that are most relevant to a given purpose, and then would select the classification (or classifications) that best distinguishes institutions along those dimensions.
  • Under such a system, there will be less risk associated with institutional innovation: Adaptation in one direction would not necessarily come at the expense of distinctiveness in other areas.

Such a system is not without risk or difficulty. The taken-for-granted nature of the present system will be undermined: Different analysts will employ different classifications, making it more difficult to compare findings from multiple studies. There might be disagreement about the appropriate classification for certain purposes, such as composing peer groups. And the complexity of the new system might discourage its acceptance: The community might reject it altogether, or simply embrace just one of its new classifications — effectively reproducing the present system. But we are persuaded that the potential benefits of a new approach are worth these risks. This is an ambitious undertaking, and its success will depend in large measure on the participation and assistance of the higher education community. Over the next few years, we will be counting on that community to help us develop a system that meaningfully characterizes the great diversity of U.S. higher education.

In the interim, the Carnegie Foundation has announced plans to issue a revised edition of the Carnegie Classification for release this summer. This revision is primarily intended to update the Classification, which was last revised in 1994 and is now seriously out of date. Our plans for the 2000 revision address some of the shortcomings set forth above while preserving the basic methodology and most of the categories of the present system (details are available on our website, www.carnegiefoundation.org). We are currently assessing feedback in response to those plans. This is an interim solution as we engage the challenging tasks we have set for the 2005 revision.


Alexander C. McCormick is a senior scholar at The Carnegie Foundation for the Advancement of Teaching. Comments can be sent to classification@carnegiefoundation.org.


2000 Edition of the Carnegie Classification:

Summary of Changes to Category Definitions
Only those categories for which substantive changes are planned are listed. Precise cut points are subject to change pending analysis of new data. See the Carnegie Website for a full list of the 2000 Preliminary Classifications.
1994 Edition 2000 Edition
Research Universities I
50 or more doctorates per year, and $40 million or more per year in federal support
Doctoral/Research Universities I
50 or more doctorates per year across at least 15 disciplines
Research Universities II
50 or more doctorates per year, and $15.5-40 million or more per year in federal support
Doctoral/Research Universities II
10 or more doctorates per year across at least 3 disciplines, or 20 or more doctorates per year overall
Doctoral Universities I
40 or more doctorates per year across at least 5 disciplines
[Included in above categories]
Doctoral Universities II
10 or more doctorates per year across at least 3 disciplines, or 20 or more doctorates per year total
[Included in above categories]
Baccalaureate (Liberal Arts) Colleges I
40 percent or more of bachelor’s degrees in liberal arts fields, and restrictive in admissions*
Baccalaureate (Liberal Arts) Colleges I
At least half of undergraduate awards are bachelor’s degrees, and at least half of bachelor’s degrees in liberal arts fields
Baccalaureate Colleges II
Less than 40 percent of bachelor’s degrees in liberal arts fields, or less restrictive in admissions*
Baccalaureate Colleges II
At least half of undergraduate awards are bachelor’s degrees, and less than half of bachelor's degrees in liberal arts fields
Associate of Arts Colleges
[no explicit distinction from Baccalaureate Colleges]
Associate’s Colleges
Less than half of undergraduate awards are bachelor’s degrees

*Admissions selectively indicated by achievement profile of entering students (high school class rank, and standardized test scores)

Average Enrollment by Carnegie Classification, 1994
chart




Copyright © 2008 - American Association for Higher Education and Accreditation