Social Graph Analysis
NodeXL is a template for Microsoft Excel which makes network analysis easy and rather intuitive. We used this tool for data collection to import the Twitter networks and to analyze the various social media interactions. There are limitations established in the Twitter API which regulate
the amount of data collected per hour by any one user. Therefore, due to rate limiting, NodeXL will inherently only import the 2,000 most recent friends and followers for any Twitter account. To improve the response time of the API, we further restricted our data collection to the 200 most recent tweets for both the university and each of its follower accounts.
For our first look at the alumni associations, we clustered the data based on an algorithm in NodeXL which looks at how the vertices are connected to one another. The clusters, as shown in Figure 1, are indicated by the color of the nodes. The clusters themselves revealed some interesting patterns. The high level of inter-association connectivity, as measured in
follows,
tweets and
mentions, was unexpected. We would have thought that each association operated within the confines of its own Twitter space or that of its parent organization. As we examine the groupings in this network, it is not unreasonable that we would observe connections between Old Dominion University (
@ODUAlumni), Norfolk State University (
@nsu_alumni_1935) and Hampton University (
@HamptonU_Alumni) as all three are located within close proximity of one another in the
Hampton Roads area. But, then we must take notice of Hollins University (
@HollinsAlum), a small, private women's college in Roanoke, VA, which has a connection with ten (10) other alumni associations; more connections than any other school. Hollins is one of the smallest universities in our group with enrollment of less than 800 students. Since Twitter is primarily about influence, in this instance, we can probably assume the
follows serve as a means to observe best practices and current engagement trends employed by larger institutions. While Hollins University is well connected, as are many of the other schools, at the opposite end of the spectrum we find Liberty University (
@LibertyUAlum), a large school with more than 77,000 students. Liberty University remains totally isolated with no follower connections to the other alumni associations. You might minimally expect some type of connection with either Regent University (
@RegentU) since both share a similar mission as private, Christian institutions or other universities within close physical proximity such as Randolph College (
@randolphcollege).
|
Figure 1 - Connectivity of Alumni Associations |
Twitter Followers, Enrollment, and Selectivity
We normally measure the popularity of a Twitter account based on the number of followers. Instead of simply quantifying the follower counts of each alumni association, we sought to understand if certain factors, actions or inherent qualities about the institution might influence the relative number of followers. First, we considered whether more active tweeters would attract more alumni followers. As shown in Figure 2, the College of William and Mary (
@wmalumni) has generated the most tweets over its lifetime, approximately 6,200 or 2.5 tweets per day. But, we also observe the University of Mary Washington (
@UMaryWash), which has approximately half the student enrollment, a similar Twitter life span, 50% percent less tweets at 2,800 or 1.3 per day, with only a slight difference in the number of followers, 4,400 versus 3,800 respectively. While the graph shows that schools such as Virginia Tech (
@vt_alumni) and the University of Virginia (
@UVA_Alumni) have more followers with fewer lifetime tweets, the caveat is that these public institutions have the benefit of considerably larger student populations which inherently increases the pool of potential alumni.
|
Figure 2 - Lifetime Tweets Versus Followers |
Next, we considered whether a higher graduation rate, or alumni production, would result in more followers. We obtained the most recent, 2014 overall graduation rates for each institution from the
National Center for Education Statistics, with reported overall six-year graduation rates ranging from 34% to 94%. A 2015 Pew Research Center study of the
Demographics of Social Media Users indicates that among all internet users, 32% in the 18 to 29 age range use Twitter. This is a key demographic as we would expect our alumni associations to be primarily focused on attracting recent undergraduates. We also factored in selectivity, a comparative scoring of the admissions process, using the categories defined in the
2016 U.S. News Best Colleges Directory. In this directory, colleges are designated as
most selective,
more selective,
selective,
less selective or
least selective based on a formula.
As we look at Figure 3, we observe a positive correlation between admissions selectivity and the institution's overall graduation rate. Schools which were least selective during the admissions phase also produced the lowest graduation rates (less than 40%) while schools which were most selective, experienced the highest graduation rates (around 90%). It isn't surprising that improved graduation rates positively affect the expected number of alumni Twitter followers. We'll leave it as an exercise for the reader to extrapolate how closely each institution's annual undergraduate enrollment, graduation rate and expected level of engagement on Twitter corresponds to the actual number of followers when all three factors are considered.
|
Figure 3 - Followers Versus Graduation Rate |
Potential Reach of Verified Followers
Users on Twitter want to be followed so we looked carefully at who, besides alumni and students, was following each of the alumni associations. Specifically, we noted the number of
Twitter verified followers; accounts which are usually associated with high-profile users in "music, acting, fashion, government, politics, religion, journalism, media, sports, business and other key interest areas." In addition to an abundance of local news reporters and sports anchors, regional politicians and career sites, other notable followers included: restaurant review site Zagat (
@Zagat), automaker Toyota USA (
@toyota), musician and rapper DJ King Assassin (
@DjKingAssassin), the Nelson Mandela Foundation (
@NelsonMandela), the President of the United States Barack Obama (
@BarackObama), Virginia Governor Terry McAuliffe (
@GovernorVA) and artist and singer Yoko Ono (
@yokoono). It's a safe assumption that some of the follower relationships with verified users were probably established prior to 2013. This is the year in which Twitter instituted new rules to
kill the "auto follow" which was a programmatic way of following another user back after they follow you. Either way, the open question would remain as to why these particular users would follow an alumni association when there are no readily apparent educational ties.
Twitter doesn't take follower count into consideration when verifying an account, but it's not unusual for a verified account to have a considerable following. Since the mission of an alumni association is essentially about networking and information dissemination, we also measured the potential reach or level of influence across the followers' extended network obtained from the verified accounts. No single university had more than 70 verified accounts among its followers. However, when we look at their contribution, in Figure 4, as a percentage of the combined reach achieved by all followers of each alumni association, these select users accounted for as little as 1.6% for Virginia Military Institute (
@vmialumni) to as much as 95.8% for Longwood University (
@acaptainforlife) of the institution's total potential reach (i.e., followers of my followers).
|
Figure 4 - Potential Reach Percentage of Verified Accounts |
Finally, we examined how each follower described himself in the description (i.e., bio) portion of their Twitter profile by extracting the top 200 most frequently occurring terms for each alumni association. A
for the alumni of each university is shown in Figure 5. If we further isolated the descriptions to the top ten most frequently occurring words, we observed a common pattern among all alumni followers. In addition to the official or some derivative of the institution name (e.g., JMU, NSU, Tech), we find the terms love, life, and some intimate description of the follower as a mom, husband, student, father or alumni. If the university has an athletic department, we also found mention of sports and, in the case of our two Christian universities, Liberty and Regent, the terms God, Jesus, and Christ were prevalent. In 22 of 23 institutions, the alumni primarily described themselves using these personal terms. Conversely, the alumni followers at only one institution, the University of Richmond (
), described themselves in a more business-like or academic manner with more frequent mention of the words PhD, career, and job.