University Futures, Library Futures: a multi-dimensional model of US higher education institutions

Over the last several months, OCLC Research has been working to develop a framework for exploring emerging directions in US higher education, to better understand the institutional needs that academic libraries will need to support and advance in years to come.

We undertook this work as part of an ongoing collaboration with Ithaka S+R on our University Futures, Library Futures project, generously supported by The Andrew W. Mellon Foundation.

Our framework is informed by an extensive literature review examining current scholarship on the US higher education landscape, which identified a range of factors important in shaping institutional direction that are imperfectly or inadequately captured in current typologies. For example, it is widely recognized that the changing demographics of student enrollment (e.g. increasing ethnic diversity, growing share of adult and part-time learners) is altering institutional approaches to advancing student success in colleges and universities with different enrollment profiles, yet the commonly used basic Carnegie categories reveal little about the character of the prevailing educational model. A Baccalaureate College–Arts & Sciences institution that is actively growing an online baccalaureate completion program targeting adult learners will look very different from a Baccalaureate Colleges–Arts & Sciences institution with a robust full-time, first-time undergraduate enrollment pipeline. Similarly the institutional profile of a Doctoral Universities: Highest Research Activity institution pursuing an aggressive strategy to diversify its undergraduate enrollment while improving retention and completion, will be different from a similarly classed research university with a highly selective undergraduate enrollment.

In this context, we adopt the working assumption that colleges and universities do and will increasingly attempt to differentiate themselves to succeed in a highly competitive market. We further hypothesize that academic library services may benefit from – and some may indeed already are benefiting from – undergoing a disruption to allow them to adapt to rapid changes in academic institutions’ business models and value propositions.

The task we undertook to fulfill was to push beyond the commonly used Carnegie Classification of US higher education institutions in order to shed new light on US higher education institutions and their libraries’ fit. In our collaborative work with Ithaka S+R it was agreed that OCLC would be developing a working model of US higher education institutions, while Ithaka S+R would be charged with developing a library services framework. These two pieces will eventually come together, but more on this in future posts.

It may be observed — quite rightly — that the Carnegie Classification has evolved to provide a much more nuanced profile of institutional character and activity than is reflected in the basic classification. Yet, combining the multiple variables of the complete classification (undergraduate and graduate instructional profiles, size and setting, enrollment profile) for statistical analyses can be cumbersome and most analyses use the basic categories in isolation. The approach we have taken in our project is to combine elements of the more robust Carnegie classification by selectively incorporating the same underpinning statistical indicators into a simpler framework. We have explored ways in which a taxonomy of US higher education institution could both incorporate Carnegie concepts and categories, and move beyond it.

The University Futures, Library Futures project population includes 4-year plus, public and private nonprofit US colleges and universities. We first created a representative random sample of 100 institutions, and evaluated the sample and population to compare variance on key parameters, to ensure the sample was representative.

Next, to base our model on nationally available statistical indicators, we turned to IPEDS 2015 institutional survey data and proposed indicators for each of the categories. We applied these categories to roughly 25 institutions in our sample for review by group.

In addition, following recent work by Martin Ruef and Manish Nag, we attempted to apply a text analysis of self-professed identity as expressed in colleges’ and universities’ mission statements on their websites. With assistance of our OCLC colleague, Shenghui Wang, we applied word-clustering and phrase-clustering to these mission statements. These efforts resulted in hundreds of phrase clusters, which we clustered manually into a dozen super clusters for further analysis that resulted in two very large networks of related phrase clusters, through which it was not evident how further work could be supported by. We did decide, however, to retain mission statements as a possible source of indicators.

Next, we engaged in rigorous investigation of the sample of 100 institutions; which we approached from both qualitative and quantitative angles:

• We extracted an initial set of IPEDS institutional survey data for the sample of 100 institutions.
• We compiled a deck of screenshots of the top-level Web presence (landing page) of the sample institutions to examine the visual signaling of institutional purpose and mission
• We compiled university mission statements for the 100-institution sample (for consistency, we used the URLs reported to IPEDS). We then anonymized these mission statements and attempted to extract worlds of meaning based on their self-professed identity, ideas, language, vocabulary, and style.

This investigation led us to propose a preliminary working model, based on a preliminary analysis of mission statements and web-sites. Our initial proposed model included the following categories: values-forming, values-norming, career-convenience, which we later revised to elite, place-based, convenience.

A more developed version of our model emerged as we introduced three distinct dimensions to our working model: a primary dimension that included categories of global elite, research, career-convenience, liberal arts; alongside a secondary dimension that included the binary categories of residential/selective, convenience/access, and a tertiary dimension that introduced the likewise binary categories of religious, secular.

We subsequently decided not to pursue our investigation of the tertiary religious/secular dimension because we detected some discrepancies between reported IPEDS data and self-professed institutional identities reflected in mission statements. We regard this as a potential area for further research.,

We then cumulatively included additional IPEDS statistics to refine our typology of the first and second dimensions and support an analysis. We consequently corrected our sample to exclude three institutions that were lacking sufficient IPEDS data, which were required for typing.

At this stage, our model yielded reasonably believable typing, but in which the dominant type in each dimension canceled out any other scoring an institution might have had. This caused us to push further to produce more nuanced typing, which would allow for dual or triple identity and reflect a more realistic view of colleges’ and universities’ complex educational activity.

This realization that we could construct a considerably more detailed and accurate model of our project population drove our current and most developed version, which includes the following two dimensions:

The primary educational activity dimension includes three major categories:

• Research that is focused on the doctoral level (IPEDS indicators: doctor’s degree – research/scholarship (DRVC2015) as percent of all doctorates, research expenses as a percent of total core expenses GASB (DRVF2015) and FASB (DRVF2015), Carnegie Classification 2015: Basic (HD2015) R1-R3).
• Liberal education that is focused on arts and sciences at mostly the baccalaureate level (IPEDS indicators: Carnegie Classification 2015: Undergraduate Instructional Program (HD2015) Arts and Sciences, Bachelor’s degree (DRVC2015) as percent of all degrees).
• Career preparation that is focused on professional training and preparation for the labor market (IPEDS indicators: Carnegie Classification 2015: Undergraduate Instructional Program (HD2015) Professions, Certificates of 2 but less than 4-years (DRVC2015), Certificates of 1 but less than 2-years (DRVC2015), Certificates of less than 1-year (DRVC2015) as percent of all degrees and certificates).

We conceive of this dimension as capturing WHAT institutions are offering

The mode of provision dimension captures how institutions are providing their educational offers to students. We find it useful to distinguish between the extent to which an institution provides its educational offering in two modes:

• A “Traditional-Residential” mode of provision (IPEDS indicators: Full-time enrollment (DRVEF2015) as percent of Total enrollment (DRVEF2015), Grand total (EF2015B All Students total  Age under 25 total as percent of Grand total (EF2015A  All students total).
• A “New-Traditional-Flexible” mode of provision (IPEDS indicators: Percent of undergraduate students awarded Pell grants (SFA1415), Number of students receiving an Associate’s degree (DRVC2015) as percent of all degrees, Part-time enrollment (DRVEF2015) as percent of total enrollment, Percent of students enrolled exclusively in distance education courses (DRVEF2015), Black or African American total (EF2015A All students total) as percent of all students, Hispanic total (EF2015A All students total) as percent of all students, Grand total (EF2015B  All Students total  Age 25 and over total) as percent of all students).

We conceive of this dimension as capturing HOW  and FOR WHOM institutions are providing their educational offers.

Our working model currently accounts for the share of institutional educational activity that is dedicated to each of the three major areas of research, liberal education, and career-preparation, and for the degree to which institutional activity is provided in a more residential or a more convenient/flexible mode. We depict the first dimension in triangular form, with each of the three areas represent one corner of a radar graph:

We depict the second dimension as a continuum, with residentiality reflecting one end and convenience/flexibility reflecting the opposite:

 

We are currently working on finding an approach to adequately visualizing both dimensions at the same time. If you have suggestions to offer, or comments on our approach, we would be glad to hear them! Please get in touch by email (steinr@oclc.org or malpasc@oclc.org) or by leaving a comment below.