This repository contains source data related to the technical report, "University Twitter Engagement: Using Twitter Followers to Rank Universities," published August 19, 2017. The goal of this study is to determine how the number of Twitter followers relates to academic prestige. We also consider additional web-based attributes that correlate with rankings but do not require expensive data collection. Please read the tech report which provides important context and details before proceeding.
Table Of Contents
- Feedback / Questions?
The raw, input data for our core analyses was obtained from the following sources:
Athletic expenditures for men's and women's sports teams was obtainted from the U.S. Department of Education reports related to Equity in Athletics Data Analysis. This database consists of athletics data that are submitted annually as required by the Equity in Athletics Disclosure Act (EADA), via a Web-based data collection, by all co-educational postsecondary institutions that receive Title IV funding (i.e., those that participate in federal student aid programs) and that have an intercollegiate athletics program. The list contents as they appeared during our data collection is shown in the
Expert Ranking Lists
This source data was extracted from the 2015-2016 academic rankings of U.S. News, Times Higher Education, Academic Rankings of World Universities and Money Magazine for the top 264 universities in the United States. These institutions were ranked based on 12 indicators that measure their academic research performance and their global and regional reputations. Note: The ranking data was collected during August 2016. It is possible the ranking pages have been updated subsequent to when the data were originally collected. The list contents as they appeared during our data collection is shown in the
The 2015 NCSE Endowment Market Values file was downloaded in a PDF format from the National Association of College and University Business Officers (NACUBO) website. It lists the self-reported market value of endowments for those U.S. and Canadian institutions who responded to a NACUBO survey. We scraped the PDF to convert it into a machine readable format
endowment/university_endowments_2014-2015.csv, then matched the institution to universities on the expert ranking lists. In cases where an endowment was associated with a university system (e.g., University of Minnesota Foundation), we used DBpedia to determine the proportion that should be allocated to a particular university of interest (e.g., University of Minnesota-Twin Cities).
The Integrated Postsecondary Education Data System (IPEDS) is the core education collection program for the National Center for Education Statistics (NCES). The
IPEDS folder contains two zipped files downloaded from IPEDS. These files contain directory information for every institution in the 2015 IPEDS universe. This information includes name, address, city, state, zip code and various URL links to the institution's home page, admissions, financial aid offices and the net price calculator. It identifies institutions as currently active, institutions that participate in Title IV federal financial aid programs for which IPEDS is mandatory. It also includes variables derived from the 2015 Institutional Characteristics survey, such as control and level of institution, highest level and highest degree offered and Carnegie classifications. For our research, IPEDS was used to standardize institution names across ranking lists and served as the source for undergraduate enrollment numbers.
To select the universities of interest to our study, we begin with the 351 American colleges and universities classified as Division I by the the National Collegiate Athletic Association (NCAA). For evaluation, we consider a subset of 65 universities associated with the Power Five conferences: SEC, ACC, Big Ten, Pac-12, and Big-12. The
NCAA folder contains several files which provide information related to athletic expenditures and conference membership.
The entire dataset used to perform the ranking analysis is published as a JSON in the
Twitter_Accounts folder. The file identifies the official primary and affiliated secondary Twitter accounts associated with each of the 264 institution in our dataset along with selected attributes for each profile (i.e., screen name, user name, URL, protected, follower count, following count). The Twitter data is used to calculate University Twitter Engagement (UTE) score which is the friend and extended follower network of primary and affiliated secondary Twitter accounts referenced on a university's home page. Please note this is a large file that contains over 1 million Twitter profiles, and is best opened with a utility such as JSON Editor Online.
In rank-to-rank comparisons we observed a significant, positive rank correlation (τ=0.6018) between UTE and an aggregated reputation ranking (ARR) which indicates that UTE could be a viable proxy for ranking atypical institutions normally excluded from traditional lists. In addition, we significantly reduce the cost of data collection needed to rank each institution by using only web-based artifacts and a publicly accessible Twitter application programming interface (API).
ODU Ranking Lists
ODU_Ranking_Lists folder contains the complete university listings ranked by our specific criteria of
adjusted reputation rank (ARR),
expenditures, enrollment and endowment (EEE), and
university Twitter engagement (UTE). Please refer to the tech report for more information regarding the methodology used to construct each ranking list.
Feedback / Questions?
Contact Corren McCoy at firstname.lastname@example.org.
Looking for more from the Web Science and Digital Libraries Research Group at Old Dominion University? Click here to read blog postings and more about research projects and education updates.