Innovative Higher Education

, Volume 42, Issue 2, pp 97–111 | Cite as

Institutional Uses of Twitter in U.S. Higher Education

  • Royce Kimmons
  • George Veletsianos
  • Scott Woodward


This study employed data mining and quantitative methods to collect and analyze the available histories of primary Twitter accounts of institutions of higher education in the U.S. (n = 2411). The study comprises a sample of 5.7 million tweets, representing 62 % of all tweets created by these accounts and the entire population of U.S. colleges and universities. With this large, generalizable dataset, researchers were able to determine that the preponderance of institutional tweets are 1) monologic, 2) disseminate information (vs. eliciting action), 3) link to a relatively limited and insular ecosystem of web resources, and 4) express neutral or positive sentiment. While prior research suggests that social media can serve as a vehicle for institutions to extend their reach and further demonstrate their value to society, this article provides empirical and generalizable evidence to suggest that such innovation, in the context of institutional social media use, is limited.


Social media Twitter Dialogic communication Data mining 

Social media have become a ubiquitous aspect of U.S. life over the last decade (Perrin 2015). The term social media refers to “a group of Internet-based applications that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of User Generated Content” (Kaplan and Haenlein 2010, p. 62). Examples include platforms such as Facebook, Twitter, YouTube, and Pinterest. Significantly, over 84 % of institutions of higher education have at least one Twitter account, making the platform one of the most popular social media tools used by institutions (Barnes and Lescault 2013). Faculty, staff, and administrators have attempted to leverage Twitter to reach and interact with students, alumni, prospective students, and the broader community (Barnes and Lescault 2013; Beverly 2013; Herrmann 2010; Kelly 2013; Linvill et al. 2012).

Twitter, with its widespread use, succinct communication method, and following functionality has become a staple of contemporary educational institutions. Due to its ubiquitous adoption and potential, understanding how institutions use Twitter is important. Thus far, research has suggested that institutional Twitter messages are largely monologic, marketing-oriented, and positive in nature (Beverly 2013; Veletsianos et al. 2016). Institutional Twitter posts appear to paint a positive picture of academic life and rarely reveal extended interaction and communication between the institution and others. Current studies however, suffer from a number of limitations that curtail the generalizability of these findings. Such limitations include limited sample sizes (number of institutions and number of tweets), failure to differentiate between types of institutions, and sample bias (such as only gathering data from highly ranked institutions). The study we report here addressed these limitations by exploring Twitter use by all institutions in the U.S.. with the exclusion of community colleges.

Review of the Literature

Research on the use of social media in education over the last decade has flourished. A recent review of the literature identified about 1000 empirical papers published in conference proceedings and academic journals between 2010 and 2015 (Pasquini et al. 2016). Given the focus of this study, the breadth of the literature on social media, and the fact that different tools have different affordances, we focused our examination of relevant literature on Twitter. In reviewing this literature, it appears that institutions’ uses of Twitter are varied. In particular, institutions use Twitter as a news platform (Herrmann 2010), recruitment tool (Barnes and Lescault 2013; Kelly 2013), and public relations device (Beverly 2013; Linvill et al. 2012). Each of these uses will be considered in turn.

Research has shown that the majority of institutional university communications on Twitter currently take the form of one-way messages aimed to disseminate information from the institution (Bélanger et al. 2014; Linvill et al. 2012; Palmer 2013; Yolcu 2013). This has been termed the “megaphone” model of Twitter use (Gallaugher and Ransbotham 2010). This finding appears to hold true for institutions in the United States (Linvill et al. 2012) as well as in other countries, including Australia (Palmer 2013), Turkey (Yolcu 2013), and Canada (Bélanger et al. 2014; Veletsianos et al. 2016). For example, one study analyzed the tweeting habits of the top 100 schools according to the U.S. News & World Report rankings and found that the most widespread intitutional use of Twitter is as a news, event, and announcement platform (Herrmann 2010). Findings such as this indicate that, although many institutions are using Twitter, the vast majority are not capitalizing on the platform’s dialogic affordances. In other words, they do not appear to use Twitter as the efficient two-way communication platform it is designed to be.

Other research has shown how institutions use Twitter as a recruitment tool. For example, using a sample of schools from all 50 U.S. states, Barnes and Lescault (2013) found that colleges and universities are using social media to both recruit and research prospective students. They found that 71 % of the institutions sampled said they believed Twitter was an effective tool in recruiting prospective students, while 13 % reported actively researching prospective students via social media as part of their admissions process. Related research shows that, although more admission offices are using Twitter, there is no empirical evidence demonstrating that this strategy is effective in actually recruiting more or better students (Kelly 2013).

Researchers have also reported on how institutions use Twitter as a public relations tool. Because of its widespread use and dialogic or conversational nature, many have seen Twitter as an ideal public relations tool which colleges and universities might use in engaging with their communities (Waters et al. 2011). Szymańska (2003) asserted that “the main goal of public relations in higher education institutions is the thoughtful creation of a positive image of them in the social and economic environment” (p. 471). Beverly (2013) argued that the local community is one of the most important publics for colleges and universities to engage, and institutions “must make aggressive efforts to build and maintain mutually-beneficial relationships with their host communities” (p. 42) because community support is directly proportionate to the community’s perception of the institution. “Twitter is one of the most popular social media tools that can be used to accomplish this goal” (p. 53). One study examined the images shared by Canadian public universities on Twitter and found that institutions present themselves in an exaggeratedly positive manner to the public (Veletsianos et al. 2016), suggesting that aggressive marketing efforts may go hand-in-hand with exaggeration and misrepresentation.

Although many institutions in the U.S. are also using Twitter to connect with various non-student publics, such as by reaching out to alumni, eliciting fundraising help (Davis et al. 2012), and connecting with their athletic team fans (Watson 2009), recent research has highlighted how ineffective colleges and universities have been at using Twitter to build relationships by employing dialogic communication principles. For example, in Beverly’s (2013) analysis of the tweets of 155 universities during a two-week period, he found that “many colleges and universities are not following the most-commonly accepted relationship-building strategies, such as dialogic and two-way communication” (p. iii). Linvill et al. (2012) found that only 29.5 % of college and university tweets in their study of 113 institutions were following dialogic principles. Both of these studies are limited, however, due to the relatively short time frame involved in each study.

In addition to institutional use of Twitter, other research has been conducted concerning how scholars (Kimmons and Veletsianos 2016; Veletsianos and Kimmons 2012, 2013, 2016; Veletsianos et al. 2013), university presidents (Barnes and Lescault 2013; Borysenko 2014), and college and university instructors (Jacquemin et al. 2014; Junco et al. 2013; Junco et al. 2011; Welch and Bonnan-White 2012) are using Twitter. Scholars are using social media to share specifics about their teaching, research, and professional practice (Greenhow et al. 2009; Veletsianos and Kimmons 2012). College and university presidents are also using Twitter to engage with their publics. Barnes and Lescault (2013) found that 55 % of a sample of college and university presidents from all 50 states in the U.S. have a Twitter account. Some presidents report that they feel Twitter makes them more approachable to their students and provides students with direct access to ask questions and to give feedback, while also providing them another way to monitor student and public perception of their institutions (Borysenko 2014).

Research regarding instructional uses of Twitter in higher education classrooms is not without controversy. For example, researchers found that retention, achievement, and overall course satisfaction increased when college and university teachers used such tools (Junco et al. 2011). However, a similar study found no significant difference in these areas between students in control groups and students who used Twitter for classroom purposes (Welch and Bonnan-White 2012). A more recent study argued that multiple components are essential for improved outcomes, namely “faculty participation on the platform, integration of Twitter into the course based on a theoretically driven pedagogical model and requiring students to use Twitter” (Junco et al. 2013, p. 273). Meanwhile, others found that, while Twitter can provide a useful hub to link news articles relevant to course topics, students often find it a cumbersome tool to be used for formal student-teacher interactions (Jacquemin et al. 2014). Thus, although merely using Twitter in the college classroom does not automatically improve student outcomes, its use can prove beneficial when integrated intentionally.

The Study

Research Question

The guiding research question of this study was as follows. How do institutions in the U.S. use Twitter, and are there differences in use based upon institution type? To answer this question, we articulated several secondary research questions as follows:
  • What types of tweets do institutions post?

  • How do institutions utilize links on Twitter?

  • What sentiment is expressed in the tweets?

  • To what extent do tweeted images demonstrate positive narratives about college life?

Each of these secondary research questions will be considered in turn, and we will also consider how the results may vary by institutional type.


For this study we used a combination of data mining and quantitative methods to answer the research questions. Given the scope of the questions being asked, it was appropriate to include all colleges and universities in the U.S. in this research study (n = 2411 university and college accounts; n = 5.7 million tweets), though community colleges were excluded. Next, we provide more detail about our sampling, data collection, coding, and analysis methods.


Using a publicly available list from The University of Texas at Austin (2015) and The Carnegie Classification of Institutions of Higher Education (2010), we developed a web script that opened all available institution homepages and scanned their contents to detect links to institutional Twitter accounts (cf. Kimmons 2015a, b). These links were then stored in a MySQL database. The links pointed to Twitter accounts for institutions, internal offices, personnel (such as the president or Provost), and other entities (such as the New York Times). The researchers manually read information on each account including location, description, and screen name to determine if the account was an official institutional account. Some institutions had more than one official account (e.g., an account for the university and another account for the athletics department). In these cases, only the primary institutional account was retained for further analysis. In total, 2411 accounts were collected in this manner, representing all institutional accounts referenced by college and university websites in the U.S.

Data Collection

Data mining for this study consisted of extracting all account data and millions of available institutional posts from Twitter. The researchers wrote a series of custom PHP/MySQL scripts in order to query the Twitter API (n.d.) methodically and extract account profile information along with tweet histories. The user information collected included screen names, descriptions, and locations as well as metadata on overall Twitter use, such as tweet counts, followers, following, and creation dates. The Twitter API only provides access to a limited number of tweets in this manner, meaning that only the most recent 3500 tweets were retrievable for each institutional account. Data cleaning required many additional steps to make the data suitable for analysis, including tasks like unshortening URLs of links. (This step was necessary because the majority of links in the dataset were processed via a URL shortener, such as or

Data Analysis

Data were analyzed by research assistants as well as by algorithms developed by the research team (hereafter referred to as “machine methods”). We used machine methods in coding cases where little variability and complexity was anticipated, such as in the case of determining dialogic vs. monologic nature of tweets. Whenever possible, we preferred machine methods to human coding methods because they could be applied to the entire data set. On the other hand, we could only perform human-driven methods on a comparatively small (though statistically representative) set of random tweets. Therefore, we reserved human coding for requirements that were either too complex for machine analysis or wherein the accuracy of machine coding merited additional human support. In these cases, we used human methods on statistically appropriate random samples of data, thereby allowing results to be generalizable to the entire population; and human results are provided alongside machine analyses to allow the reader to infer similarities and differences in the methods.

The most complex machine-coding mechanism used in this study involved classifying each tweet as either having positive, negative, or neutral sentiment. An open-source sentiment classifier called phpInsight was used for this purpose (Hennessey 2015). This approach had been used in a previous study (Veletsianos and Kimmons 2016), but in the current study we present results alongside human-coded results to determine to what extent the two methods generated similar findings. We conducted additional analysis on each research question and explain the results below.


Data collection enabled the identification of 2411 primary U.S. college and university Twitter accounts. User metadata revealed that these accounts had created 9.2 million tweets in their lifespan, which ranged from 1 to 8 years. In total, 5.7 million tweets were accessible and retrieved for analysis, representing 62 % of all tweets created by these accounts. The rest of the tweets were not accessible through the Twitter API. The average account posted 3819 tweets over the course of its lifespan (Median = 2487; S.D. = 4425; Min = 1; Max = 77,613). We provide results below, organized by the secondary research questions.

Research Question 1. What Types of Tweets Do Institutions Post?

Tweets were categorized as either dialogic or monologic. Dialogic tweets were defined as those that responded to another tweet or user. Some examples included the following:
  • “Thanks for sharing the info about our [open position]. Appreciate the shout out.”

  • “We’ll let facilities management know there is an issue.”

  • “We’re pleased that you enjoyed your visit to our university today! Thank you for the hashtags highlighting some of the best…”

Conversely, monologic tweets were those that broadcasted content in a unidirectional manner. Some examples of monologic tweets included the following:
  • “Thanks to our cheerleaders for putting this together for our football team this morning!”

  • “Visiting with future [university] students at [a local high school]…”

  • “[University magazine] mails out this week - check out the new issue online!”

To clarify, monologic tweets might include references to people (e.g., cheerleaders or football teams), but they would only be classified as dialogic if they responded to their Twitter accounts.

Few tweets (10.1 %) were dialogic in nature, indicating that the platform was generally used as a broadcast tool. Similarly, tweets were categorized as action tweets or informational tweets. Action tweets were defined as those that called upon the reader to respond in a particular way (e.g., read, retweet, follow, come, join us). These were identified by a script that searched for common action words within tweet content. Only a small minority of tweets included these signifiers (12.2 %), indicating that most tweets were intended to provide information rather than to elicit action (cf. Fig. 1). As we consider these two findings together, we discover that the vast majority of tweets (78.8 %) are monologic actions to disseminate information and that the least common type of tweet is a dialogic invitation to action (1.1 %; cf. Table 1).
Fig. 1

Visual comparison of monologic vs. dialogic and information vs. action factors

Table 1

Tweet type as categorized by monologic vs. dialogic and information vs. action factors






78.8 %

9.0 %

87.8 %


11.1 %

1.1 %

12.2 %


89.9 %

10.1 %

To determine if institutional type predicted differences in these factors, we conducted a series of one-way analysis of variance tests (ANOVA) with percentages of overall tweets as the dependent factors and institutional identifiers as independent factors. Results indicated that differences based on institutional level (2-year vs. 4-year+) were significant in both the case of monologic vs. dialogic tweets, F(1) = 96.9, p < .001 and action vs. information tweets, F(1) = 80, p < .001. Mean differences based on this factor revealed that 2-year institutions were slightly more action-oriented in their tweets (MD = 2.5 %) but that 4-year + institutions were slightly more dialogic (MD = 4.4 %; cf. Table 2).
Table 2

Average tweets per account by institutional type









14.7 %

85.3 %

94.5 %

5.5 %



12.2 %

87.8 %

90.0 %

9.9 %



13.5 %

86.5 %

91.9 %

8.1 %

private, non-profit


12.3 %

87.7 %

91.0 %

9.0 %

private, for-profit


12.2 %

87.8 %

89.1 %

10.9 %

Results indicated that differences based on institutional control (public vs. private non-profit vs. private for-profit) were significant in both the case of monologic vs. dialogic tweets, F(2) = 4.1, p < .05 and action vs. information tweets, F(2) = 12.1, p < .001. However, Bonferroni post hoc testing revealed that the only difference between these types rested in a 1.2 % greater action tweet percentage among publics compared to private non-profits (cf. Table 2).

Research Question 2. How Do Institutions Utilize Links on Twitter?

Over one-third of all institutional tweets included links (2.1 million or 37.6 %), consisting of 75.8 thousand unique websites. Anticipating that institutions may link to different types of resources for different reasons, domains were machine coded to identify links pointing to institutional resources (e.g., .edu domains); and the top 250 most popular remaining links were human-coded by type. This list of 250 coded domains only constituted 0.3 % of all domains but represented 62.3 % of all links (meaning that these domains were exponentially more popular than the remaining 99.7 % of links; cf. Fig. 2).
Fig. 2

Tweet frequency count of top 20 domains included as links

Results indicated that 63.5 % of links in the sample of top tweets pointed to prominent social media services such as Facebook, Instagram, Pinterest, or YouTube; 11.8 % to institutional resources (i.e., .edu pages); and 5.0 % to news sources and aggregators (such as Huffington Post or New York Times; cf. Table 3). Of these links, 19.6 % could not be classified because they either used a URL shortener that was no longer valid or were malformed. Resource types were likely referenced for different reasons as links to Facebook were slightly more likely to be action-oriented (16 %) than were other links (12.1 %). The sheer volume of links to specific sites may provide insight into the institution web ecosystem and its information sources (cf. Table 4).
Table 3

Link categorization of top 250 websites linked from tweets


% of overall

% of sample

Explanation or popular examples

Social media

39.5 %

63.5 %,,,


7.4 %

11.8 %,,,

News & Aggregators

3.1 %

5.0 %,,,


12.2 %

19.6 %

unresolvable shorteners, partial URLs


62.3 %

100 %

Table 4

Top 30 domains included as links in institution tweets


% of all links


% of all links


% of all links

17.5 %

0.3 %

0.1 % or

13.8 %

0.2 %

0.1 %

4.3 %

0.2 %

0.1 %

2.9 %

0.2 %

0.1 %

0.5 %

0.2 %

0.1 %

0.4 %

0.2 %

0.1 %

0.3 %

0.2 %

0.1 %

0.3 %

0.2 %

0.1 %

0.3 %

0.1 %

0.1 %

0.3 %

0.1 %

0.1 %

Research Question 3. What Sentiment is Expressed in the Tweets?

We calculated sentiment for each tweet using two parallel methods. The first involved the use of the algorithm presented in Hennessey (2015) to code the entire population of tweets. The second involved the human-coding of a random sample of tweets (n = 2321) in a manner that allowed for redundancy between two coders on a 5-point Likert scale. We then averaged results between coders and assigned a sentiment classification based on the guidelines provided in Table 5. This sample size was large enough to ensure that results were generalizable to the entire population with a 95 % confidence interval and +/- 2 % confidence level. Conducting human coding on such a large sample allowed us to compare results with the machine-coded population and yielded results that would be generalizable even if machine-coding was shown to be different. We utilized both methods to account for the potentially subjective nature of sentiment analysis and to determine whether the two methods produced a similar narrative.
Table 5

Sentiment codes from human coders and accompanying classifications from averages

Individual code



Maybe negative




Maybe positive













Avg Code




Overall, results of the two methods revealed the same pattern: most tweets were neutral (53.1–56.2 %), many tweets were positive (38.7–43.4 %), and a few tweets were negative (0.4–8.1 %). Discrepancies between the two methods seemed minor, with the greatest difference manifested in a 7.7 % higher percentage of tweets classified as negative by machine coding (cf. Fig. 3). From this, we can infer that institutions tend to either express neutral or positive sentiments in their tweets, eschewing tweets with negative sentiment, such as expressing disdain, anger, frustration, or discontent.
Fig. 3

Visual comparison of human-coded and machine-coded sentiment

Slightly more than half of all tweets were neutral in sentiment, meaning that they were worded in a manner that conveyed information without emotion or judgment. These tweets included announcements and solicitations, and some examples included the following:
  • “[Famous author] will give Commencement keynote.”

  • “Designed for people who are interested in learning the fundamentals of Chinese medicine without committing to a…”

  • “Students: Res. Life is going green, so visit the Web site for room selection info. Get your app in before the deadline!”

Positive tweets made up over one-third of tweets and included expressions of gratitude, congratulatory remarks, excitement, self-praise, and words of encouragement. Some examples of tweets with positive sentiment included the following:
  • “The MLK Commemorative Program has begun in [student center]! So excited to have all of our visitors!!”

  • “Today, our board of trustees awarded degrees to 676 March graduates. Congrats to our newest alumni!”

  • “Welcome to campus new students! We’re enjoying the day with these soon to be [team mascots]!”

Negative tweets made up a very small minority of tweets (<10 %) and included mild complaints, reporting of bad news, and social criticism. Some examples of tweets with negative sentiment included the following:
  • “More snow coming? Really?”

  • “It is with great sadness that we can confirm the death of a [university] student. Support avail. via [university’s] Counseling Center.”

  • “The Cross and the Confederate Battle Flag cannot coexist without one setting fire to the other.”

Research Question 4. To What Extent Do Tweeted Images Demonstrate Positive Narratives About College Life?

Previous research focusing on Canadian universities suggested that the skew toward positivity found in institutional Twitter accounts might be reflected in the use of images to draw attention to three themes: campus attractiveness, positive experiences, and successes or celebrations (Veletsianos et al. 2016). To examine the applicability of this result in the U.S. context and further evaluate this finding, this study replicated this examination using the current dataset. By searching for key strings in tweet contents (e.g., .jpg, .png, flickr, instagram), approximately 176,000 tweets were identified as having connected images. A random sample of these images (n = 1027) was coded to allow for generalizability with a 95 % confidence interval and +/- 3 % confidence level.

Human coders used a table consisting of six codes: campus attractiveness (image depicts buildings, facilities, and the natural beauty of the campus), positive experiences (image depicts students, faculty members, or other smiling or engaging in a positive way), faculty and student successes (image depicts convocations, sports victories, dignitaries open buildings, groundbreakings, etc.), text (image depicts flyers or other announcements), other (image depicts something not captured in the rest of the codes), and missing (image is no longer accessible or the link pointing to the image is broken). Missing images represented 5.6 % of all items and presumably were missing due to a variety of factors (e.g., the resource had been moved or deleted since the original tweet creation date). We excluded missing items from further reporting.

Results indicated that images constructing a positive representation of the institution were the most common, accounting for 85.4 % of all images (cf. Table 6). Of these, images depicting positive experiences, such as student activities, were the most common (47.8 %), followed by images depicting campus attractiveness (24.9 %), such as architecture or nature pictures, and successes (12.8 %), such as graduation ceremonies. Only 5.8 % of images were used for textual flyers or announcements. An evaluation of the images categorized as other (8.8 %) suggested that some of these images were used for marketing or selling university products, warning about bad weather, or pointing out oddities of university life; but there was such diversity in these tweets and they represented such a small percentage of overall use that they did not lend themselves to thematic categorization. These results provide further evidence to the overwhelmingly positive picture that institutional Twitter accounts paint for university life.
Table 6

Purpose of communicated images


Image count

Percent of total

Positive experiences


47.8 %

Campus attractiveness


24.9 %

Faculty and student successes


12.8 %

Text / Fliers


5.8 %



8.8 %

Error / Missing


5.6 %


The results of this study provide insight into how institutions use Twitter and also lead to a number of considerations for discussion and implications for future research. First, contrary to predominant narratives and claims in this area which suggest that Twitter can be used to move away from traditional monologic approaches of information dissemination toward a more dialogic, community-responsive paradigm, results indicated that institutional use remains overwhelmingly monologic in nature, regardless of institutional type (89.1–94.5 %). In other words, while the potential exists to engage in dialogue, debate, conversation, and community outreach, Twitter’s verified usage reflects a different reality. In actuality, the educational institutions examined in this study appear to use Twitter primarily to broadcast information and highlight the positive aspects of the institution. This result is significant for both scholarly and practical purposes. The scholarly significance of this result reflects the fact that this study confirmed findings from prior research on a much larger scale. Is this result useful in the real world? There have been numerous calls in recent years for institutions to extend and demonstrate their value to society (e.g., Boyer 1990). Institutional Twitter accounts may serve as vehicles for applying scholarship, for engaging stakeholders in the translation of scholarship, and for helping communities employ scholarship and new knowledge. While there was little evidence of these activities in our research, we hope that our findings will mobilize academic administrators overseeing institutional social media accounts to develop processes enabling more effective use of social media in higher education. It might be worthwhile to consider the development of training programs for social media managers or the development of institutional initiatives that focus on how to use Twitter for the greater common good.

Second, most of institutional Twitter use does not involve links; but, when links are used, almost two-thirds of these links point to other social media sites, such as Facebook or Pinterest, with few pointing to institutional websites (11.8 %) and even fewer pointing to news sites or aggregators (5.0 %). It appears that most of these links are used to support announcements, scheduling, or cross-posting of information across social media. This finding reveals a somewhat limited ecosystem of web participation for institutions, showing that a few websites dominate social media participation. Further, this ecosystem seems to be relatively insular because, when information is shared, it is twice as likely to point to an institutional resource rather than to an external information source. Finally, it appears that this ecosystem lacks diversity as it revolves around social media, institutional websites, and a few select news sources. The evaluation of these links was beyond the scope of this article but may be a fruitful area for future research.

Third, while the majority of tweets posted by institutions are neutral, many are positive; and very few are negative. This result corroborates findings of positivity on Twitter uncovered by other studies (Veletsianos et al. 2016) and does so on a much larger scale. Results suggest that tweets are positively skewed with narratives that frame the institution as a place of success, achievement, and beauty. This result is likely a reflection of the increased attention that branding activities have received in higher education (e.g., Chapleo 2005). Institutional marketing efforts appear to have become domiciled in institutional Twitter accounts. The generalizability of this finding is a cause of concern because the image of the institution represented in these communications is incomplete and potentially misleading (e.g., limited mention is made of the increasing cost of higher education or the myriad of other problems facing higher education). In arriving at the same finding when studying institutional accounts in the Canadian context, we noted:

Perhaps this is the time to pause and reflect. Is this how [institutional Twitter accounts] should be used? Are there alternative, and perhaps more productive, ways that institutional Twitter accounts could be used? And if so, which office should manage them? Is it possible, for example, for these accounts to exist under Lifelong Learning programs and Continuing Education departments, with the mandate to support public learning and enact public scholarship for the betterment of local communities and […] society overall? What should be the role of the main institutional social media accounts (Veletsianos et al. 2016)?

These observations and questions become more significant because our research suggests that the findings reported herein are generalizable and applicable in both Canada and the United States.


We must note several limitations of our study. First, in identifying monologic and dialogic tweets, mentions were used to identify if the tweet was intended for or referenced a specific account. However, not all Twitter communications are public because direct (or private) messages may be sent between users if both users follow one another. It is therefore possible that some dialogic use of Twitter is not discernible from publicly viewable data if institutions and their communities are sending direct messages to one another. However, if institutions are sending direct messages to their communities, then it is doubtful that they are doing so with sufficient frequency to meaningfully impact results. Second, this study focused on Twitter, and readers should be cautioned that the results may not extend to other social media platforms. Future research examining the applicability of these results in other contexts (e.g., Facebook) will be extremely valuable.


In this research, we presented a comprehensive and generalizable picture of the ways higher education institutions in the U.S use Twitter. We analyzed 5.7 million tweets from 2411 college and university Twitter accounts and showed that the preponderance of tweets a) are monologic, b) disseminate information, c) link to a relatively limited and insular ecosystem of web resources, and d) express neutral or positive sentiment. Analysis of the images reveals that they construct an overwhelmingly positive representation of the institution, its faculty, and its environment. Implications of this study suggest that institutional uses of Twitter are not living up to their potential as an educational or communication medium. For these reasons, we suggest that institutions should reconsider how they are using Twitter and explore ways that the medium can be used more meaningfully to affect positive changes within their educational communities and society at large.



George Veletsianos acknowledges funding received from the Canada Research Chairs program.


  1. Barnes, N. G., & Lescault, A. M. (2013). College presidents out-blog and out-tweet corporate CEO’s. UMass Center for Marketing Research. Retrieved from
  2. Bélanger, C. H., Bali, S., & Longden, B. (2014). How Canadian universities use social media to brand themselves. Tertiary Education and Management, 20, 14–29.CrossRefGoogle Scholar
  3. Beverly, J. A. (2013). Public relations models and dialogic communication in the Twitterverse: An analysis of how colleges and universities are engaging their publics through twitter (Doctoral dissertation). The University of Southern Mississippi. Retrieved from
  4. Borysenko, K. (2014). Commit, connect, engage: How college and university presidents are using Twitter. Eduventures. Retrieved from
  5. Boyer, E. (1990). Scholarship reconsidered: Priorities for the professoriate. Princeton, NJ: The Carnegie Foundation for the Advancement of Teaching.Google Scholar
  6. Carnegie Classification of Institutions of Higher Education (2010). The Carnegie Classification of Institutions of Higher Education. Retrieved from
  7. Chapleo, C. (2005). Do universities have “successful” brands? International Journal of Educational Advancement, 6, 54–64.CrossRefGoogle Scholar
  8. Davis, C., Deil-Amen, R., Rios-Aguilar, C., & Canche, M. (2012). Social media in higher education: A literature review and research directions. The Center for the Study of Higher Education at the University of Arizona and Claremont Graduate University. Retrieved from
  9. Gallaugher, J., & Ransbotham, S. (2010). Social media and customer dialog management at Starbucks. MIS Quarterly Executive, 9, 197–212.Google Scholar
  10. Greenhow, C., Robelia, B., & Hughes, J. E. (2009). Learning, teaching, and scholarship in a digital age: Web 2.0 and classroom research: What path should we take now? Educational Researcher, 38, 246–259.CrossRefGoogle Scholar
  11. Hennessey, J. W. (2015). phpInsight – Sentiment analysis in PHP. Retrieved from
  12. Herrmann, M. (2010). Getting a take on Twitter usage. University Business, 13(1). Retrieved from
  13. Jacquemin, S. J., Smelser, L. K., & Bernot, M. J. (2014). Twitter in the higher education classroom: A student and faculty assessment of use and perception. Journal of College Science Teaching, 43(6), 22–27.Google Scholar
  14. Junco, R., Heiberger, G., & Loken, E. (2011). The effect of Twitter on college student engagement and grades. Journal of Computer Assisted Learning, 27, 119–123.CrossRefGoogle Scholar
  15. Junco, R., Elavsky, C. M., & Heiberger, G. (2013). Putting Twitter to the test: Assessing outcomes for student collaboration, engagement and success. British Journal of Educational Technology, 44, 273–287.CrossRefGoogle Scholar
  16. Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of social media. Business Horizons, 53, 59–68.CrossRefGoogle Scholar
  17. Kelly, K. J. (2013). The effectiveness of Twitter as a communication tool in college recruitment (Doctoral dissertation). Texas A&M University – Kingsville. Retrieved from ProQuest LLC. (3566310)Google Scholar
  18. Kimmons, R. (2015a). Online system adoption and K-12 academic outcomes. Journal of Computer Assisted Learning, 31, 378–391.CrossRefGoogle Scholar
  19. Kimmons, R. (2015b). Open online system adoption in K-12 as a democratizing factor. Open Learning: The Journal of Open, Distance and e-Learning, 30, 138–151.CrossRefGoogle Scholar
  20. Kimmons, R., & Veletsianos, G. (2016). Education scholars’ evolving uses of Twitter as a conference backchannel and social commentary platform. British Journal of Educational Technology, 47, 445–464.CrossRefGoogle Scholar
  21. Linvill, D. L., McGee, S. E., & Hicks, L. K. (2012). Colleges’ and universities’ use of Twitter: A content analysis. Public Relations Review, 38, 636–638.CrossRefGoogle Scholar
  22. Palmer, S. (2013). Characterisation of the use of Twitter by Australian universities. Journal of Higher Education Policy and Management, 35, 333–344.CrossRefGoogle Scholar
  23. Pasquini, L. A., Knott, J. E., & Fraiser, J. (2016). Social media for teaching and learning in higher education: A systematic analysis and synthesis of empirical research from 2010-2015. Manuscript submitted for publication.Google Scholar
  24. Perrin, A. (2015). Social networking usage: 2005-2015. Pew Research Center. Retrieved from
  25. Szymańska, A. (2003). Public relations: The process of communication between a higher education institution and its environment: The case of the Wroclław University of Economics. Higher Education in Europe, 28, 471–485.CrossRefGoogle Scholar
  26. Twitter API (n.d.). API overview. Twitter Developers. Retrieved from
  27. University of Texas at Austin (2015). U.S. universities. The University of Texas at Austin. Retrieved from
  28. Veletsianos, G., & Kimmons, R. (2012). Networked participatory scholarship: Emergent techno-cultural pressures toward open and digital scholarship in online networks. Computers & Education, 58, 766–774.CrossRefGoogle Scholar
  29. Veletsianos, G., & Kimmons, R. (2013). Scholars and faculty members’ lived experiences in online social networking systems. The Internet and Higher Education, 16, 43–50.CrossRefGoogle Scholar
  30. Veletsianos, G., & Kimmons, R. (2016). Scholars in an increasingly digital and open world: How do education professors and students use Twitter? The Internet and Higher Education, 30, 1–10.CrossRefGoogle Scholar
  31. Veletsianos, G., Kimmons, R., & French, K. (2013). Instructor experiences with a social networking site in a higher education setting: Expectations, frustrations, appropriation, and compartmentalization. Educational Technology Research & Development, 61, 255–278.CrossRefGoogle Scholar
  32. Veletsianos, G., Kimmons, R., Shaw, A. G., Pasquini, L., & Woodward, S. (2016). Twitter use in Canada’s public universities. Manuscript submitted for publication.Google Scholar
  33. Waters, R. D., Canfield, R. R., Foster, J. M., & Hardy, E. E. (2011). Applying the dialogic theory to social networking sites. Journal of Social Marketing, 1, 211–227. doi:10.1108/20426761111170713 CrossRefGoogle Scholar
  34. Watson, G. (2009). Coaches, colleges explore new frontier. ESPN College Sports. Retrieved from
  35. Welch, B. K., & Bonnan-White, J. (2012). Twittering to increase student engagement in the university classroom. Knowledge Management & E-Learning, 4, 325–345.Google Scholar
  36. Yolcu, Ö. (2013). Twitter usage of universities in Turkey. Turkish Online Journal of Educational Technology, 12, 360–371.Google Scholar

Copyright information

© Springer Science+Business Media New York 2016

Authors and Affiliations

  • Royce Kimmons
    • 1
  • George Veletsianos
    • 2
  • Scott Woodward
    • 1
  1. 1.Instructional Psychology and TechnologyBrigham Young UniversityProvoUSA
  2. 2.Innovative Learning and TechnologyRoyal Roads UniversityVictoriaCanada

Personalised recommendations