Evaluating universities' strategic online communication: how do Shanghai Ranking's top 50 universities grow stakeholder engagement with Facebook posts?

Birte Fähnrich (Berlin-Brandenburgische Akademie der Wissenschaften, Berlin, Germany)
Jens Vogelgesang (Department of Communication, University of Hohenheim, Stuttgart, Germany) (Computational Science Lab (CSL), University of Hohenheim, Stuttgart, Germany)
Michael Scharkow (Department of Communication, Johannes Gutenberg University Mainz, Mainz, Germany)

Journal of Communication Management

ISSN: 1363-254X

Article publication date: 14 May 2020

Issue publication date: 18 August 2020

Abstract

Purpose

This study is dedicated to universities' strategic social media communication and focuses on the fan engagement triggered by Facebook postings. The study contributes to a growing body of knowledge that addresses the strategic communication of universities that have thus far hardly dealt with questions of resonance and evaluation of their social media messages.

Design/methodology/approach

Using the Facebook Graph API, the authors collected posts from the official Facebook fan pages of the universities listed on Shanghai Ranking's Top 50 of 2015. Specifically, the authors retrieved all posts in a three-year range from October 2012 to September 2015. After downloading the Facebook posts, the authors used tools for automated content analysis to investigate the features of the post messages.

Findings

Overall, the median number of likes per 10,000 fans was 4.6, while the number of comments (MD = 0.12) and shares (MD = 0.40) were considerably lower. The average Facebook Like Ratio of universities per 10,000 fans was 17.93%, the average Comment Ratio (CR) was 0.56% and the average Share Ratio (SR) was 2.82%. If we compare the average Like Ratios (17.93%) and Share Ratios (2.82%) of the universities with the respective Like Ratios (5.90%) and Share Ratios (0.45%) of global brands per 10,000 fans, we may find that universities are three times (likes) and six times (shares) as successful as are global brands in triggering engagement among their fan bases.

Research limitations/implications

The content analysis was solely based on the publicly observable Facebook communication of the Top 50 Shanghai Ranking universities. Furthermore, the content analysis was limited to universities listed on the Shanghai Ranking's Top 50. Also, the Facebook posts have been sampled between 2012 and September 2015. Moreover, the authors solely focused on one social media channel (i.e., Facebook), which might restrict the generalizability of the study findings. The limitations notwithstanding, university communicators are invited to take advantage of the study's insights to become more successful in generating fan engagement.

Practical implications

First, posts published on the weekend generate significantly more engagement than those published on workdays. Second, the findings suggest that posts published in the evening generate more engagement than those published during other times of day. Third, research-related posts trigger a certain number of shares, but at the same time these posts tend to lower engagement with regard to liking and commenting.

Originality/value

To the authors’ best knowledge, the automated content analysis of 72,044 Facebook posts of universities listed in the Top 50 of the Shanghai Ranking is the first large scale longitudinal investigation of a social media channel of higher education institutions.

Keywords

Citation

Fähnrich, B., Vogelgesang, J. and Scharkow, M. (2020), "Evaluating universities' strategic online communication: how do Shanghai Ranking's top 50 universities grow stakeholder engagement with Facebook posts?", Journal of Communication Management, Vol. 24 No. 3, pp. 265-283. https://doi.org/10.1108/JCOM-06-2019-0090

Publisher

:

Emerald Publishing Limited

Copyright © 2020, Emerald Publishing Limited


1. Introduction

This study of strategic social media communication in the university sector focuses on fan engagement triggered by universities' Facebook postings. The study contributes to a growing body of knowledge addressing why universities' strategic communications rarely address the evaluation of social media messages and their effectiveness (Piezcka, 2002; Raupp and Osterheider, 2019). In recent years, a growing body of research confirms the increasing importance of strategic communication in the university sector, as evidenced by the increasing relevance attributed to strategic communication by university management and academics (Marcinkowski et al., 2014); rising staff and budgets (Hauser et al., 2019); and increased communication activity (Serong et al., 2017; Metag and Schäfer, 2017). These developments are accompanied by strong demands to evaluate strategic communication and to increase its effectiveness in at least three ways.

First, the intensification and professionalization of universities' strategic communication is a response to rising political and public expectations in the science communication field, where universities are expected to play a key role (Borchelt and Nielsen, 2014). To fulfill their so-called “third mission” (Montesinos et al., 2008), which includes social, enterprise and innovation activities to benefit society, universities are increasingly required to contribute to public understanding of and engagement with science. Academics, politicians and society at large now expect universities to do more to assess the effectiveness of their strategic communications activities as a measure of third mission impact (Nisbet and Markovitz, 2016; Nisbet and Scheufele, 2009).

Second, the need for strategic communication reflects the increasing competition among universities for public attention and image proliferation – as for instance in Germany (Marcinkowski et al., 2014) and in the UK (Bauer and Gregory, 2007) – driven by new modes of governance such as new public management (De Boer et al., 2007). Elite universities listed in international publications such as the Academic Ranking of World Universities – also known as ShanghaiRanking – are more favorably positioned to garner public attention (Hegglin and Schäfer, 2015). This strong positioning in turn demands effective stakeholder management (Rodriguez-Pomeda and Casani, 2016), as rising budgets and claims of professionalism require universities to communicate how they “deliver value for money” (Van Ruler, 2005, p. 163), subjecting the targets, goals and effectiveness of their strategic communication to public scrutiny (Raupp and Osterheider, 2019).

Third, the development of strategic communication in universities has been facilitated by advances in digitalization and a changing media landscape. Previously, university PR was strongly related to and directed toward science journalism (Bauer and Gregory, 2007). In recent years, however, as social media platforms such as Facebook, Twitter, Instagram and YouTube gained in importance (Brossard, 2013), universities have used social media channels to bypass journalistic gatekeepers (Metag and Schäfer, 2017, 2019), tailoring their communication to audience needs and establishing more permanent and dialogic relationships with diverse stakeholder groups (Nisbet and Scheufele, 2009; Lovari and Gigilietto, 2012; Hayes et al., 2009; Davies and Hara, 2017).

Measuring whether and how communication goals are reached is central to evaluating the effectiveness of strategic communication (Nisbet and Scheufele, 2009); to date, however, few studies have examined the effects on target recipients of universities' strategic communication efforts. Against this background, the present study investigates the effects of universities' strategic online communication by considering the Facebook activities of elite universities listed on Shanghai Ranking. In particular, we used quantitative content analysis to evaluate fan engagement with the universities' Facebook posts, conceiving that engagement (in the form of likes, comments and shares) as an effect of what Zerfass et al. (2017) referred to as the output level of corporate communication. The study's founding assumption is that appearing in Shanghai Ranking's Top 50 identifies an institution as among the most reputable and influential universities in the world. On that basis, the study addresses the following overarching question.

How do the Top 50 Shanghai Ranking universities grow their Facebook post engagement?

To begin the paper reviews state-of-the-art knowledge regarding universities' strategic online communication, focusing in particular on social media communication and engagement. We then propose a framework for measuring the impact of Facebook postings and lay the groundwork for the research questions and hypotheses. After describing the study design and methodology, we report the key findings of a content analysis of Facebook postings published by Shanghai Ranking's Top 50 universities. In conclusion, we discuss the practical implications of our findings for universities' social media communication strategies.

2. Strategic communication in the university sector

Strategic communication can be characterized as public and purposeful communication to foster positive relationships with their stakeholders in pursuit of organizational objectives (Holtzhausen and Zerfass, 2013; Raupp, 2017). In the case of universities, strategic communication may encompass science-related communication as well as teaching issues, community relations and university policies (Fähnrich et al., 2019; Roblyer et al., 2010). Today, most universities have established a central communication unit as part of their administration structure. Within that unit, professional communicators implement online and offline communication for the university as a whole. These professionals are responsible for integrating the communication efforts of other university actors, such as chairs or departments; for example, the central communication unit's strategy coordinates the communication activities of all university members and develops guidelines for communicating with diverse stakeholders (Raupp, 2017; Samuel et al., 2017).

In recent years, universities have increasingly used social media platforms such as Facebook, Instagram, Twitter and YouTube for the purposes of strategic communication (Metag and Schäfer, 2019; Hall, 2014; Kelleher and Sweetser, 2012; Wankel and Wankel, 2011). Through social media, universities can direct tailor-made messages to specific stakeholder groups, bypassing intermediaries or gatekeepers and establishing new forms of relationship through the opportunities for dialogue and interaction afforded by these technologies (Kelleher and Sweetser, 2012; Pleil and Zerfass, 2014). Typically, higher education organizations target the following four types of stakeholder (McAllister and Taylor, 2007): (1) prospective and current students; (2) prospective and current employees; (3) external stakeholders (e.g., politicians, business leaders, alumni, potential donors) and (4) journalists. In higher education, social media platforms are considered especially useful for pursuing communication objectives that include fostering dialogue, collaboration and relationship building (e.g., Belluci et al., 2019; Peruta and Shields, 2017; Sandvig, 2016), and these have also been identified as important conditions for effective science communication (Nisbet and Scheufele, 2009).

3. Social media engagement in higher education

According to Machado et al. (2019), user-to-content interactions such as liking, commenting, and sharing are equivalent to word-of-mouth communication because clicking the “like” button and other acts of commenting or sharing automatically publish the post in a user's Facebook newsfeed and subsequently perhaps in the newsfeeds of the user's friends (Swani et al., 2013). Activities such as liking, commenting, and sharing are commonly referred to as “social media engagement” (Khan, 2017), offering users a means of expressing themselves (Wallace et al., 2012). For instance, clicking the “like” button signals that one endorses the message of a Facebook post, while sharing a Facebook post transfers its message directly to one's own newsfeed. By sharing a social media message, users seek to communicate to their contacts that it is in some way significant to them or that it has news value; if shared without any further comments, this usually means that users agree with the message (Valerio et al., 2015). All forms of positive engagement help to strengthen the existing stakeholder-university relationship previously established by becoming a fan of a particular university Facebook page (Clark et al., 2017; Nevzat et al., 2016).

Although universities have embraced the use of various social media platforms such as Facebook, Instagram or YouTube, little is known empirically about the kinds of topic or theme that drive Facebook engagement in this context. To the best of our knowledge, two quantitative content analyses of university Facebook posts have investigated the relationship between post themes or topics and fan engagement. The two studies differ in granularity of coding; while Thelen and Men (2018) distinguished between university reputation, activities, social and political issues, ceremonies and distinctions, sports, celebrations and safety, Peruta and Shields' (2018) coding scheme was more elaborate, comprising 17 content categories. Their results indicate that Facebook posts covering sports and self-promotion, including school spirit or university brand, tend on average to promote higher fan engagement while posts related to research and scholarly or study program content generate lower fan engagement (Peruta and Shields, 2018).

However, the findings from both of these studies should be interpreted with caution. In particular, Thelen and Men's (2018) content analysis was based on a sample of just 146 posts from one university. While Peruta and Shields' (2018) study was based on a much larger sample (N = 5,932 posts from 66 Facebook pages), the number of posts per Facebook page (n = 89) limits the generalizability of their findings because of the high sampling error per sampling unit. To address the issues of heterogeneous coding schemes and relatively small sample size per university, the present study is based on a large-scale content analysis of Shanghai Ranking's Top 50 universities.

4. How the Shanghai Ranking top 50 grow Facebook engagement

The goal of the present study was to develop a fuller understanding of how Shanghai Ranking's Top 50 universities use Facebook to grow fan engagement. According to Belluci et al. (2019), Facebook is among the social media channels most widely used by top-ranked universities. Our content analysis focused on the official Facebook pages of Shanghai Ranking's Top 50 universities for 2015, most of which are located in the US.

Several measures have been proposed to measure Facebook post engagement (Kim, 2016; Park et al., 2016). Here, we draw on the framework developed by Sabate et al. (2014) to assess the effects of different types of social media content in generating user engagement. Themes or topics are based on the tentative findings of Peruta and Shields (2018).

Facebook enables universities to disseminate content through posts on fan pages. Posts may contain news, links to websites, photos, or videos and can be seen by users of the fan page (Entradas and Bauer, 2019; Fähnrich et al., 2019; Ryan, 2019). Unlike ordinary Facebook user profiles, fan pages provide additional functionalities, such as metrics and tools for content and fan administration. For present purposes, the audiences of universities' fan pages are referred to as fans.

Content analysis has often been used to understand the effects of business organizations' social media strategies (Cvijikj and Michahelles, 2013; Jayasingh and Venkatesh, 2015; Kim et al., 2015; Sabate et al., 2014; Swani et al., 2013; Swani et al., 2017; De Vries et al., 2012). Typically, studies investigating Facebook fan pages measure engagement by the number of likes, comments, or shares of Facebook posts. For example, one such analysis of 1,086 posts from the Facebook pages of 92 global brands over a period of one month revealed an average publication rate of 11.4 posts, generating an average of 8,522 likes, 591 shares and 249 comments (Kim et al., 2015) and translating into an average of 748 likes, 52 shares and 22 comments per post. Swani et al. (2017) corroborated these findings in a content analysis of Facebook posts published by 280 Fortune 500 companies. They reported that while posts published by business-to-consumer (B2C) companies generate an average of 621 likes and 78 comments, posts published by business-to-business (B2B) companies receive an average of 19 likes and 2 comments. According to Swani et al. (2017), the average Facebook fan base size of a Fortune 500 B2C company is 1,700,903, as compared to 80,874 for a B2B company. To obtain a normalized value for comparison with Facebook posts for the Shanghai Ranking Top 50, we adopted Cvijikj and Michahelles' (2013) formula. Like ratios (LRs), comment ratios (CRs) and share ratios (SRs) were computed by dividing the number of each by fan base size and multiplying by 10,000. Based on that calculation, Fortune 500 B2C companies have a Facebook LR of 3.65% (per 10,000 fans) as compared to a Facebook LR of 2.35% for B2B companies. These very low ratios are quite common in the business world – even among global brands with large marketing budgets. In an earlier content analysis of 100 Facebook brand pages, Cvijikj and Michahelles (2013) reported similar ratios per 10,000 fans (LR = 5.90%, CR = 1.20%, SR = 0.45%). These findings serve as a benchmark for Facebook engagement with posts published by Shanghai Ranking's Top 50 universities, operationalized as LR, CR and SR.

To formulate the study hypotheses, we reviewed the existing research on Facebook post engagement. De Vries et al. (2012) and Sabate et al. (2014) reported that longer Facebook texts negatively affect fan engagement, and we assume here that this also applies to fans of university Facebook pages. Specifically, as the average Facebook visit lasts about 16 min (Delany and Madigan, 2017), and as social media users generally scroll through their timelines, we formulated the following hypothesis.

H1.

A greater number of words per Facebook post will negatively affect fan engagement.

The independent development of visual studies and strategic communication as academic disciplines (Goransson and Fagerholm, 2018) means that strategic communication research on visual communication remains limited. For that reason, we borrowed from marketing and management research to formulate our next hypothesis regarding post type. A number of studies (e.g., Cvijikj and Michahelles, 2013; Sabate et al., 2014) have reported that photo and video content promote greater fan engagement than text-only messages, indicating that visual content –- or online content's vividness (De Vries et al., 2012) – is effective in triggering fan engagement. As images are thought to play a more important role in communication among students (as digital natives) than among the older population (Delello and McWorther, 2014), we hypothesized that students will be more attracted to visuals than to traditional link-based posts containing a certain amount of text.

H2.

Facebook posts with visual content will receive more fan engagement than posts with text-only content.

As in the case of online journalism, time of publication is likely to influence the effectiveness of a Facebook post in terms of reach and fan responsiveness, as individual Facebook newsfeeds are constantly overloaded with content that originates from multiple sources, especially in daytime hours. However, previous studies of the optimal time of day for generating maximum fan engagement with a Facebook post have reported mixed results. For example, Cvijikj and Michahelles (2013) found that fan engagement is higher when a brand post is published outside business hours (i.e., from 16:00 to 09:00) because when fans check their own Facebook profiles for the first time on a given day, they are more likely to see the post, in turn increasing the probability of engagement. In contrast, Sabate et al.'s (2014) content analysis showed that publishing during business hours does not influence the number of likes but positively affects the number of comments. According to social media agencies, the peak time for student Facebook use is late afternoon and evening (AdEspresso, 2019). On that basis, we formulated the following hypothesis.

H3.

Facebook posts generate greater fan engagement when published during non-business hours as compared to business hours.

Sabate et al. (2014) found no difference in the effect of Facebook brand posts published between Monday and Friday or at the weekend. However, Cvijikj and Michahelles (2013) reported a higher number of comments on workdays, although the number of likes and shares was not affected by day of publication. In contrast to the mixed results from academic studies, commercial audience research indicates that students are highly active on Saturdays (Arens, 2019). Based on the assumption that content overload on workdays makes it less likely that Facebook posts will be liked, shared, or commented, we formulated the following hypothesis.

H4.

Facebook posts generate greater fan engagement when published at the weekend as compared to workdays.

In line with those who believe that social media communication has become more important for higher education organizations in recent years (Kelleher and Sweetser, 2012; McAllister, 2012) because their main target group comprises students born in the digital age, we formulated the following hypothesis.

H5.

Facebook posts now generate greater fan engagement than they did three years ago.

Social media research consistently reports that the overall number of fans positively affects fan engagement by virtue of the so-called “network effect” (Sabate et al., 2014; Jayasingh and Venkatesh, 2015). On that basis, we formulated the following hypothesis.

H6.

The number of fans per Facebook page is positively correlated with degree of fan engagement.

What kinds of post fans see on their Facebook newsfeeds is determined and controlled by an unknown ranking algorithm. While some of the ranking methods used by Facebook are proprietary and not publicly available, we know that the number of likes, shares and comments associated with a previous post strongly influences the visibility of a subsequent post on one's newsfeed (Luckerson, 2015). On that basis, we formulated the following hypothesis.

H7.

The degree of fan engagement with a post is positively correlated with degree of fan engagement with the previous post.

Because the findings reported by Peruta and Shields (2018) are tentative, there is insufficient scientific evidence on which to ground theme-specific hypotheses. For that reason, we instead pose the following general research question:

Among universities' Facebook posts, which themes trigger which kinds of engagement?

5. Method

5.1 Sampling and data collection

Using the Facebook Graph API, which can no longer be used for academic purposes (Bruns, 2019), we collected posts from the official Facebook fan pages of Shanghai Ranking's Top 50 universities for 2015. Specifically, we retrieved all posts in a three-year range from October 2012 to September 2015. On checking the coding material, we found that several universities in non-English-speaking countries posted both in English and in their native language; for simplicity, we removed these universities from the sample (i.e., ETH Zurich, Kyoto U, UPMC Paris, Karolinska Institutet, U Paris-Sud, U Heidelberg, U Tokyo, U Copenhagen). This resulted in a final sample of N = 72,044 posts from 42 university pages. For data collection procedure and all statistical analyses, we used R software (R Core Team, 2015). In particular, the “RFacebook” package (Barbera and Piccirilli, 2015) was used to automate collection of Facebook data, and the “lme4” package (Bates et al., 2015) was employed to estimate a multilevel regression model.

5.2 Measures

In 2015, the Facebook Graph API still provided content and metadata for all objects on the Facebook platform. For the purposes of this study, we collected the metadata for universities' fan pages and posts; profiles indicate the number of fans for each page, and these subsequently served as metadata. Regarding the posts, we included each post's creation time and date (converted to the universities' local times), as well as the message content (including figure captions) and content type (i.e., status, link, photo, video). The creation time and date were then used to identify the time of day (night = 00:00–05:59, morning = 06:00–11:59, afternoon = 12:00–17:59, evening = 18:00–23:59) and day of the week on which the post was published.

The content of each post was automatically coded using regular expressions (Friedl, 2006) containing keywords identified in an initial screening step; this was done by human coders and further refined using the results of a topic model (Blei et al., 2003). Specifically, we exploited the results of a structural topic model (STM) (Roberts et al., 2014) to find additional keywords associated with those identified in the initial screening step. To compute the STM, we used the “stm” package (Roberts et al., 2019).

The combined manual screening and topic modeling yielded six themes: research, teaching, self-promotion, awards and prizes, addressing fans, and events. Research includes posts containing references to scientific findings or faculty members in their roles as researchers or scholars. Teaching includes student-related posts containing keywords such as “program,” “undergraduate,” “course” or “class,” among others. Self-promotion refers to posts containing first-person plural addresses such as “we” or “us.” Awards and prizes includes posts congratulating students and research personnel for scientific achievements, while interaction refers to posts containing second-person singular addresses. While these themes are not mutually exclusive – in other words, they can and do co-occur in the same posts – correlation analysis revealed that they are no more than weakly correlated, indicating little overlap.

In addition, the length of each post was computed by counting the number of words. In each case, number of likes, comments and shares served as indicators of fan engagement. Because of heavy skewness, we log-transformed all dependent variables, as well as the number of Facebook friends per university. A preliminary correlation analysis confirmed that number of likes was strongly associated with number of comments (r = 0.80), with a weaker association between number of comments and shares (r = 0.59). The correlation between number of likes and number of shares was relatively low (r = 0.33), indicating that liking and sharing are different mechanisms of engagement.

Following the statistical approach employed by Swani et al. (2013), we estimated mixed regression models (Raudenbush and Bryk, 2002) – also known as multilevel or hierarchical linear models (HLMs) – to predict the number of likes, shares, and comments. Mixed regression models are especially suitable for exploring the nesting of Facebook posts (Level 1) within Facebook accounts (Level 2).

6. Results

6.1 Characteristics of the Facebook posts

The universities in the final sample published an average of 1,715 Facebook posts between 2012 and 2015 (see Table 1); of these, Yale University (4,356 posts), the University of Pennsylvania (4,117 posts), and Columbia University (3,862 posts) were the most active publishers. The average value of 1,715 posts translates into 11 posts per week; the highest averages were 27 posts per week for Yale University, 26 posts for the University of Pennsylvania, and 24 posts for Columbia University. In contrast, the University of Edinburgh published the fewest Facebook posts (302), representing an average of two posts per week.

The average length of a university Facebook post was 33 words. On average, the University of Colorado Boulder published the longest Facebook messages (M = 69 words), while Columbia University and the University of Minnesota returned the shortest average message length (M = 19 words). Overall, 88% of posts were published on weekdays while the remaining 12% were published at weekends. Most posts were published in the afternoon (48%) or in the morning (38%). The predominant post types were photographs (51%) and links (35%) while other types such as videos (9%) and status updates (5%) were relatively rare. Thematic analysis revealed that 22% of the posts related to fan interaction; 18% related to research or self-promotion; 15% related to teaching, 13% related to events; and 3% related to awards or prizes. Notably, the proportion of posts related to research was highest for the University of Minnesota (55%), the University of Chicago (41%) and Stanford University (34%).

Because of the skewness of the distribution of likes, comments and shares, we report both median and mean values. The overall medians were 109 likes (M = 427), 9 shares (M = 53) and 3 comments (M = 13) per Facebook post (see also Figure 1). Harvard University received the largest number of likes (MD = 1,651), comments (MD = 53), and shares (MD = 147). This result is unsurprising, as Harvard accounts for the most Facebook fans by far (4.2 million). While about half of the Facebook posts published by the University of Michigan, the University of Texas at Austin, and the University of Wisconsin–Madison received more than 500 likes, the combined posts of all other universities totaled less than 500 likes.

Normalizing engagement with each Facebook account in relation to the number of fans presents a somewhat different picture. Overall, the median number of likes per 10,000 fans was 4.6, but this was considerably lower for comments (MD = 0.12) and shares (MD = 0.40). Figure 2 shows that UC San Diego was most effective in triggering fan engagement during the period studied, with a median of 23 likes per 10,000 UC San Diego fans. In other words, while the median number of comments per 10,000 fans was 1, the median number of shares per 10,000 fans was 2.3.

Computing LR, CR, and SR values provides another perspective on engagement among Facebook fans of ShanghaiRanking Top 50 universities. Overall, the average Facebook LR per 10,000 fans was 17.93% (average CR = 0.56%, average SR = 2.82%). The University of Wisconsin–Madison reported the highest LR (70.01%), followed by UC San Diego (48.19%). Interestingly, despite having the largest fan base, Harvard University was among those with the lowest average Facebook LR (5.69%). UC San Diego (1.79%) and the University of Wisconsin–Madison (1.77%) generated most comments per 10,000 Facebook fans, and UC San Francisco reported the highest sharing activity per 10,000 fans (27.91%).

6.2 Assessing the effects of Facebook posts

To predict numbers of likes, comments and shares, we ran three multilevel regression analyses. Inspection of the null model indicated that the amount of variance explained by a Facebook account (Level 2) accounted for up to 42% of the number of likes. More specifically, the null model that predicted the number of comments accounted for 39% of the variance, while the null model that predicted the number of shares accounted for 33% of the variance. This result demonstrates the necessity of estimating an HLM with an intercept that varies for each Facebook account.

Table 2 presents the odds ratios of the estimated regression models. The predictive quality of each model varied between 28 and 47%. H1 (that post length is negatively associated with fan engagement) was confirmed for the number of likes (−4%) and comments (−2%) but not for number of shares. There was partial support for H2 (that visual content increases fan engagement with Facebook messages as compared to text-only posts). Unlike posts in the form of text-only status updates, pictorial content increased the number of likes by 35% and the number of shares by 46% but impacted negatively on the number of comments (−11%). Contrary to our expectation, video posts had a largely negative effect on the number of likes (−30%) and comments (−34%); in line with H2, however, video content increased the number of shares by 27%. Results are somewhat mixed for H3 (that Facebook posts trigger more engagement when published during non-business hours). The number of likes confirmed H3; compared to Facebook posts published in the morning, posting in the afternoon (−5%) or at night (−12%) led to significantly fewer likes, but those published in the evening gained significantly more likes (+17%). A similar pattern was observed for the number of shares. However, the number of comments was higher for all publication times other than the morning, so contradicting H3. H4 (that the likelihood of engagement with Facebook posts is higher at weekends than on workdays) was consistently confirmed for all three types of engagement. Specifically, the number of likes (+24%), comments (+12%) and shares (+10%) was significantly higher when the post was published on either Saturday or Sunday. In Figure 3, which visualizes the combined results for H3 and H4, Facebook posts published between Saturday 18:00 and Sunday 05:59 are clearly more likely to attract likes, comments, and (to some extent) shares.

H5 (that Facebook posts published in 2015 triggered more engagement than those published in previous years) was clearly supported. H6 (that the number of fans of each Facebook account is positively correlated with their degree of engagement with Facebook posts) was also clearly confirmed. H7 (that degree of engagement with a post depends on the degree of engagement generated by the previous post) was consistently confirmed across all three models.

As noted earlier, we made no predictions about theme-specific effects. Posts designated as interaction triggered some comments (+7%) and shares (+3%) but negatively affected the number of likes (−4%). Posts containing research themes negatively affected the number of likes (−12%) and comments (−11%) but positively affected the number of shares (+5%). While self-promotional statements positively triggered likes (+7%) and shares (+6%), posts related to teaching and events received fewer likes, comments and shares on average. Finally, posts related to awards positively impacted the number of likes (+15%) and comments (+4%).

7. Discussion and implications

University communicators' increasing use of social media channels in recent years reflects the growing competition among (elite) universities for public attention and the importance of social media in students' everyday lives, raising the question of how to ensure that such communication is effective. Content analysis reveals that all ShanghaiRanking Top 50 universities maintain a Facebook fan page, and the results clearly indicate that Facebook posts published by these universities trigger fan engagement. This finding corroborates the evidence from global brand studies (Kim et al., 2015) that Facebook posts published by business brands on average generate 748 likes, 52 shares and 22 comments. However, given the much larger fan bases of business brands, the sampled universities have proved very successful in generating Facebook likes (M = 428), shares (M = 53) and comments (M = 13). Comparing the universities' average LR (17.93%) and SR (2.82%) with the global brands' average LR (5.90%) and SR (0.45%) per 10,000 fans, we find that the universities are three times as successful as the global brands in terms of likes and six times as successful in terms of shares in triggering engagement among their fan bases.

Given that global brands also have higher marketing budgets (Hayes, 2014; Martin, 2016), universities can be said to be more effective in triggering likes and shares. Admittedly, fans of university Facebook pages differ from business customers in terms of their higher involvement with and loyalty to their alma mater. In contrast, global brands must invest in large-scale relationship marketing activities to maintain brand loyalty among their customers. Nevertheless, the present findings seem to justify universities' increasing budgets for strategic social media communication (Hauser et al., 2019).

Despite this relative success in generating fan engagement, our findings also highlight several shortcomings in the universities' use of Facebook, and university communicators can exploit these insights to grow and enhance fan Facebook engagement. First, content type is a key consideration for the success of university fan pages. Published photos generate the highest number of likes and shares, and video posts generate more shares than text-only messages. It seems likely that fans tend to share photographs and videos because of their aesthetic value, although it is important to note that video posts also generate significantly fewer likes. This latter finding may relate to the practice of scrolling through the timeline, which is incompatible with watching a video. University communicators must resolve these conflicting objectives by weighing the respective value of likes and shares.

Secondly, while the mixed results reported here make it more difficult to say which themes should be prioritized, it seems clear that posts about awards positively trigger likes and comments while posts related to teaching impact negatively on fan engagement. It is less straightforward to evaluate how other themes influence fan engagement in the present context. For example, while Facebook posts on research themes trigger shares, the lower rates of liking and commenting indicate that the research theme is less appreciated. Our large-scale content analysis aligned with Peruta and Shields' (2018) finding that promotional posts trigger fan engagement while research-related posts have the opposite effect. On that basis, we recommend that universities should carefully review which of their social media channels is best suited to research dissemination.

Finally, one of our more interesting findings is that posts published at the weekend generate significantly more engagement than those published on weekdays. The results also suggest that posts published in the evening generate more engagement than those published at other times of the day. University communicators should review the timing of their social media posts in light of this evidence; for example, it may be advantageous to publish Facebook posts between Saturday evening and Sunday morning rather than on Thursday or Friday because newsfeeds are less overloaded at weekends.

Needless to say, the present study has some limitations that future research should address. First, our content analysis was based entirely on the publicly observable Facebook communications of the ShanghaiRanking Top 50, with no further insight into their reasoning or strategic objectives. For that reason, we cannot exclude the possibility that a given approach (e.g., time of publication) was chosen as an effective means of achieving a particular communication goal. Clearly, then, more research is needed to properly evaluate universities' communication strategies and goals, as well as the professional routines of their social media managers. Secondly, the research design has some limitations; for example, the content analysis was confined to universities listed in the Shanghai Ranking's Top 50, and the sampled posts were confined to the period between October 2012 and September 2015. Additionally, the exclusive focus on a single social media channel (i.e., Facebook) may restrict the generalizability of our findings. Overall, however, we believe that university communicators can learn from the insights provided by this unique dataset, as the underlying mechanisms of fan engagement remain largely the same. Finally, we examined the themes of Facebook posts while neglecting other features such as tone or style, and advanced forms of sentiment analysis such as supervised machine learning (Brynielsson et al., 2014; Scharkow, 2013), may help to enhance existing understanding of the linguistic factors that drive fan engagement.

In conclusion, more research is needed to support practitioners' use of evidence-based strategies that more effectively trigger fans' social media engagement. To that end, universities' should make their strategic communication goals sufficiently transparent to be shared with researchers seeking to analyze the content of their social media posts. By cooperating in this way with academic social media researchers, universities can augment their social media analytics expertise and evaluation resources, which remain limited in comparison to the larger marketing budgets of global brands.

Figures

Fan engagement in Facebook posts published by the Top 50 Shanghai Ranking universities between 2012 and 2015

Figure 1

Fan engagement in Facebook posts published by the Top 50 Shanghai Ranking universities between 2012 and 2015

Normalized fan engagement in Facebook posts published by the Top 50 Shanghai Ranking universities between 2012 and 2015

Figure 2

Normalized fan engagement in Facebook posts published by the Top 50 Shanghai Ranking universities between 2012 and 2015

Fan engagement per day and time of the day in Facebook posts published by the Top 50 Shanghai Ranking universities between 2012 and 2015

Figure 3

Fan engagement per day and time of the day in Facebook posts published by the Top 50 Shanghai Ranking universities between 2012 and 2015

Characteristics of the Facebook accounts and posts published by the Top 50 Shanghai Ranking universities between 2012 and 2015

UniversityTotal number ofAverage number ofPercentage of posts publishedTopicsContent type
FriendsPostsPosts per weekWords per postOn a weekdayIn the morningIn the afternoonIn the eveningInter-actionResearchSelf-promotionTeachingEventsAwardsPhotoLinkVideoStatus update
Columbia U235,4233,862241996295893123333127981
Cornell U272,7592,9201831934648610137141415741565
Duke U290,5122,2191433751539452019151818136218136
Harvard U4,277,3781,4759237780154718811119514441
Imperial College London107,7441,3108459647501362626212123315842
Johns Hopkins U113,6063,3012131864933172617301414133249613
MIT719,8991725112681554311025710107652852
New York U493,5272,76017228542544199129953540718
Northwestern U104,12397763185226116358211414145126148
Princeton U420,4563,505222188324419141214121211514081
Rockefeller U10,7271944122889345510102111151512296272
Stanford U976,207991631983557914349559676225
U Illinois173,8051,636104292165925201913191917533673
U Pennsylvania149,4834,117263282355114252320202023742250
U Toronto253,7736734269831681251731414128345185
U Washington303,4152,46815319945522251343131313533178
U Wisconsin–Madison253,5421,389924913258105013331717137314112
U British Columbia121,0391,404930873053161412132020144730211
U California64,3382,1901433852449251719171717132858112
U Cambridge1,390,3891,47292877403723301117151511563850
U Chicago169,8091,039747100564226415212116672292
U Colorado Boulder154,470173011698829654292228666661974
U Edinburgh138,399302250823938222820161818152952135
U Manchester98,62264572895128622920232828184142124
U Maryland113,3161,183824885244437142416161744361010
U Melbourne174,8661,0527379152444153412161693145813
U Michigan710,5001,07373191544142112810101586482
U Minnesota159,0319686199948481115518554592796
U Oxford2,493,0671,15873674533691922193131294145112
U Southern California203,6871,62910319054406291116141421345196
UC Berkeley354,900187312268240481120611101084438115
UC Irvine74,3252,19214268751436415461515165729103
UC Los Angeles388,46083153575314920221415131311771563
UC San Diego57,7315604529037594271930141410791821
UC San Francisco32,17887453183415361115717178563482
UC Santa Barbara65,4271,17373296463816211613171713434076
UNC Chapel Hill237,7031,500932884446102413151616143257100
University College London126,5107155518675532281320141413592795
UT Austin611,8071,5461038762038414752388125228614
UT Southwestern15,4112,2821424905835230918334266227
Washington U St. Louis43,8741,02563691126720181317191912682093
Yale U1,116,4154,356273479343234111811161616612595
Average435,063171511338838481222181815133513595

Prediction of fan engagement

PredictorsExp(B)
Likes (log)Comments (log)Shares (log)
Theme
Interaction0.96*1.07*1.03*
Research0.88*0.89*1.05*
Self-promotion1.07*1.021.06*
Teaching0.94*0.97*0.87*
Event0.98*0.97*0.96*
Awards1.15*1.04*0.98
Time of publication
Afternoon0.95*1.02*0.95*
Evening1.17*1.20*1.09*
Night0.88*1.12*0.95
Day of publication
Weekend1.24*1.12*1.10*
Content type
Photograph1.35*0.89*1.46*
Link0.61*0.57*0.98
Video0.70*0.66*1.26*
Year of publication
20120.38*0.69*0.55*
20130.44*0.66*0.62*
20140.50*0.63*0.54*
Characteristic of Facebook page
Number of friends (log)1.75*1.57*1.67*
Post characteristic
Length0.96*0.98*1.00
Engagement in previous post (log)1.21*1.16*1.14*
Marginal R20.470.340.28
σ21.130.891.44
τ000.150.140.18

Note(s): Multilevel regression using REML estimation with random intercepts (*p < 0.05, nL1 = 68,804, nL2 = 42), All predictors except post length (number of words/10), number of friends (log) and engagegment in previous posts (log) are dichotomous. The reference group is publication in the morning, status-update (content type) and published in 2015. Reading help: Coefficients can be interpreted as relative percent changes in the outcome variable, e.g. photos generated 1.09 times as many (or 9 percent more) likes than status posts, while videos generated 0.9 times as many (10 percent less)

References

AdEspresso (2019), “The best time to post on Instagram in 2019”, available at: https://adespresso.com/blog/best-times-to-post-on-instagram/.

Arens, E. (2019), “Best times to post on social media for 2019”, available at: https://sproutsocial.com/insights/best-times-to-post-on-social-media/.

Barbera, P. and Piccirilli, M. (2015), “Rfacebook: access to Facebook API via R”, available at: http://CRAN.R-project.org/package=Rfacebook.

Bates, D., Mächler, M., Bolker, B. and Walker, S. (2015), “Fitting linear mixed-effects models using lme4”, Journal of Statistical Software, Vol. 67 No. 1, pp. 1-48, doi: 10.18637/jss.v067.i01.

Bauer, M.W. and Gregory, J. (2007), “From journalism to corporate communication in post-war Britain”, Bauer, M.W. and Bucchi, M. (Eds), Science, Journalism and Society: Science Communication between News and Public Relations, Routledge, London, pp. 33-52.

Bellucci, M., Biagi, S. and Manetti, G. (2019), “Dialogic accounting and stakeholder engagement through social media: the case of top-ranked universities”, The Review of Higher Education, Vol. 42 No. 3, pp. 1145-1184, doi: 10.1353/rhe.2019.0032.

Blei, D.M., Ng, A.Y. and Jordan, M.I. (2003), “Latent dirichlet allocation”, Journal of Machine Learning Research, Vol. 3, pp. 993-1022.

Borchelt, R. and Nielsen, K.H. (2014), “Public relations in science”, Bucchi, M. and Trench, B. (Eds), Handbook of Public Communication of Science and Technology, Routledge, London & New York, pp. 58-69.

Brossard, D. (2013), “Science and new media landscapes”, Proceedings of the National Academy of Sciences, Vol. 110 No. 3, pp. 14096-14101, doi: 10.1073/pnas.1212744110.

Bruns, A. (2019), “After the ‘APIcalypse’: social media platforms and their fight against critical scholarly research”, Information, Communication and Society, Vol. 22 No. 11, pp. 1544-1566.

Brynielsson, J., Johansson, F., Jonsson, C. and Westling, A. (2014), “Emotion classification of social media posts for estimating people's reactions to communicated alert messages during crises”, Security Informatics, Vol. 3 No. 1, pp. 1-11, doi: 10.1186/s13388-014-0007-3.

Clark, M., Fine, M.B. and Scheuer, C.-L. (2017), “Relationship quality in higher education marketing: the role of social media engagement”, Journal of Marketing for Higher Education, Vol. 27 No. 1, pp. 40-58, doi: 10.1080/08841241.2016.1269036.

R Core Team (2015), R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, available at: https://www.R-project.org/.

Cvijikj, I.P. and Michahelles, F. (2013), “Online engagement factors on Facebook brand pages”, Social Network Analysis and Mining, Vol. 3 No. 4, pp. 843-861, doi: 10.1007/s13278-013-0098-8.

Davies, S.R. and Hara, N. (2017), “Public science in a wired world: how online media are shaping science communication”, Science Communication, Vol. 39 No. 5, pp. 563-568, doi: 10.1177/1075547017736892.

De Boer, H., Enders, J. and Schimank, U. (2007), “On the way towards new public management? The governance of university systems in england, The Netherlands, Austria, and Germany”, in Jansen, D. (Ed.), New Forms of Governance in Research Organizations. Disciplinary Approaches, Interfaces and Integration, Springer, Dordrecht, pp. 137-152.

De Vries, L., Gensler, S. and Leeflang, P.S. (2012), “Popularity of brand posts on brand fan pages: an investigation of the effects of social media marketing”, Journal of Interactive Marketing, Vol. 26 No. 2, pp. 83-91, doi: 10.1016/j.intmar.2012.01.003.

Delany, T. and Madigan, T. (2017), Friendship and Happiness: And the Connection between the Two, McFarland & Company, Jefferson, North Carolina.

Delello, J. and McWorther, R. (2014), “New visual social media for the higher education classroom”, in Mallia, G. (Ed.), The Social Classroom: Integrating Social Network Use in Education, IGI Global, Hershey, PA, pp. 368-392.

Entradas, M. and Bauer, M.W. (2019), “Kommunikationsfunktionen im Mehrebenensystem Hochschule [Communication functions within the multilevel system university]”, in Fähnrich, B., Metag, J., Post, S. and Schäfer, M.S. (Eds), Forschungsfeld Hochschulkommunikation [Research Field University Communication], Springer VS, Wiesbaden, pp. 97-122.

Fähnrich, B., Metag, J., Post, S. and Schäfer, M.S. (2019), “Hochschulkommunikation aus kommunikationswissenschaftlicher Perspektive – einleitung [University communication from a communication science perspective]”, in Fähnrich, B., Metag, J., Post, S. and Schäfer, M.S. (Eds), Forschungsfeld Hochschulkommunikation [Research Field University Communication], Springer, Wiesbaden, pp. 1-24.

Friedl, J.E. (2006), Mastering Regular Expressions, O'Reilly, Sebastopol.

Goransson, K. and Fagerholm, A. (2018), “Towards visual strategic communications”, Journal of Communication Management, Vol. 22 No. 1, pp. 46-66, doi: 10.1108/JCOM-12-2016-0098.

Hall, S.J. (2014), “How higher education institutions utilize social media”, (Doctoral dissertation), available at: https://digitalcommons.georgiasouthern.edu/honors-theses/132/.

Hauser, C., Schwetje, T. and Leßmöllmann, A. (2019), “Organisation von Hochschulkommunikation [Organization of university communication]”, Fähnrich, B., Metag, J., Post, S. and Schäfer, M.S. (Eds), Forschungsfeld Hochschulkommunikation [Research Field University Communication], Springer, Wiesbaden, pp. 123-139.

Hayes, T.J., Ruschman, D. and Walker, M.M. (2009), “Social networking as an admission tool: a case study in success”, Journal of Marketing for Higher Education, Vol. 19 No. 2, pp. 109-124, doi: 10.1080/08841240903423042.

Hayes, J.J. (2014), “Increasing enrollment: evaluating college-choice factors at a Midwest Christian University”, (Doctoral Dissertation)”, available at: https://pdfs.semanticscholar.org/692a/398f022496e8c4ed1fececce67f74b9d1da3.pdf.

Hegglin, T. and Schäfer, M.S. (2015), “Der Ranking-Effekt. Zum Einfluss des ‘Shanghai-Rankings’ auf die medial dargestellte Reputation deutscher Universitäten [The ranking effect. How the ‘Shanghai Ranking’ influences the mediated reputation of German Universities]”, Publizistik, Vol. 60 No. 4, pp. 381-402.

Holtzhausen, D.R. and Zerfass, A. (2013), “Strategic communication – pillars and perspectives of an alternative paradigm”, in Zerfass, A., Rademacher, L. and Wehmeier, S. (Eds), Organisationskommunikation und Public Relations [Organizational communication and public relations], Springer VS, Wiesbaden, pp. 73-95.

Jayasingh, S. and Venkatesh, R. (2015), “Customer engagement factors in Facebook brand pages”, Asian Social Science, Vol. 11 No. 26, pp. 19-29, doi: 10.5539/ass.v11n26p19.

Kelleher, T. and Sweetser, K. (2012), “Social media adoption among university communicators”, Journal of Public Relations Research, Vol. 24 No. 2, pp. 105-122.

Khan, M.L. (2017), “Social media engagement: what motivates user participation and consumption on YouTube?”, Computers in Human Behavior, Vol. 66, pp. 236-247, doi: 10.1016/j.chb.2016.09.024.

Kim, D.-H., Spiller, L. and Heche, M. (2015), “Analyzing media types and content orientations in Facebook for global brands”, Journal of Reseach in Interactive Marketing, Vol. 9 No. 1, pp. 4-30, doi: 10.1108/jrim-05-2014-0023.

Kim, H. (2016), “Who likes, shares, and comments on Facebook? A comparative study of Facebook use and engagement between U.S. and South Korean college students”, Asian Journal of Information and Communications, Vol. 8 No. 1, pp. 69-83.

Lovari, A. and Giglietto, F. (2012), “Social media and Italian universities: an empirical study on the adoption and use of Facebook, Twitter and Youtube”, doi: 10.2139/ssrn.1978393.

Luckerson, V. (2015), “Here's how Facebook's news feed actually works”, Time available at: http://time.com/3950525/facebook-news-feed-algorithm/.

Machado, J.C., Vacas-de-Carvalho, L., Azar, S.L., André, A.R. and Pires dos Santos, B. (2019), “Brand gender and consumer-based brand equity on Facebook: the mediating role of consumer-brand engagement and brand love”, Journal of Business Research, Vol. 96, pp. 376-385, doi: 10.1016/j.jbusres.2018.07.016.

Marcinkowski, F., Kohring, M., Fürst, S. and Friedrichsmeier, A. (2014), “Organizational influence on scientists' efforts to go public: an empirical investigation”, Science Communication, Vol. 36 No. 1, pp. 56-80, doi: 10.1177/1075547013494022.

Martin, B.K. (2016), “Resource allocation, marketing strategies, the college choice process: a collective case study analysis of community college marketing administrators”, (Doctoral dissertation), available at: https://ttu-ir.tdl.org/bitstream/handle/2346/67070/MARTIN-DISSERTATION-2016.pdf.

McAllister, S.M. and Taylor, M. (2007), “Community college web sites as tools for fostering dialogue”, Public Relations Review, Vol. 33 No. 2, pp. 230-232, doi: 10.1016/j.pubrev.2007.02.017.

McAllister, S.M. (2012), “How the world's top universities provide dialogic forums for marginalized voices”, Public Relations Review, Vol. 38 No. 2, pp. 319-327, doi: 10.1016/j.pubrev.2011.12.010.

Metag, J. and Schäfer, M.S. (2017), “Hochschulen zwischen Social Media-Spezialisten und Online-Verweigerern. Eine Analyse der Online-Kommunikation promotionsberechtigter Hochschulen in Deutschland, Österreich und der Schweiz [Universities between social media specialists and holdouts. An analysis of universities' online communication in Germany, Austria, and Switzerland]”, SCM Studies in Communication and Media, Vol. 6 No. 2, pp. 160-195.

Metag, J. and Schäfer, M.S. (2019), “Hochschulkommunikation in Online-Medien und social media [University communication in online media and social media]”, in Fähnrich, B., Metag, J., Post, S. and Schäfer, M.S. (Eds), Forschungsfeld Hochschulkommunikation [Research Field University Communication], Springer, Wiesbaden, pp. 363-391.

Montesinos, P., Carot, J.M., Martinez, J.-M. and Mora, F. (2008), “Third mission ranking for world class universities: beyond teaching and research”, Higher Education in Europe, Vol. 33 Nos 2-3, pp. 259-271, doi: 10.1080/03797720802254072.

Nevzat, R., Amca, Y., Tanova, C. and Amca, H. (2016), “Role of social media community in strengthening trust and loyalty for a university”, Computers in Human Behavior, Vol. 65, pp. 550-559, doi: 10.1016/j.chb.2016.09.018.

Nisbet, M.C. and Markowitz, E. (2016), Science Communication Research: Bridging Theory and Practice, American Association for the Advancement of Science, Washington, DC, available at: https://www.aaas.org/sites/default/files/content_files/NisbetMarkowitz_SciCommAnnotatedBibliography_Final.pdf.

Nisbet, M.C. and Scheufele, D.A. (2009), “What's next for science communication? Promising directions and lingering distractions”, American Journal of Botany, Vol. 96 No. 10, pp. 1767-1778, doi: 10.3732/ajb.0900041.

Park, H., Reber, B.H. and Chon, M.-G. (2016), “Tweeting as health communication: health organizations' use of Twitter for health promotionand public engagement”, Journal of Health Communication, Vol. 21 No. 2, pp. 188-198, doi: 10.1080/10810730.2015.1058435.

Peruta, A. and Shields, A.B. (2017), “Social media in higher education: understanding how colleges and universities use facebook”, Journal of Marketing for Higher Education, Vol. 27 No. 1, pp. 131-143, doi: 10.1080/08841241.2016.1212451.

Peruta, A. and Shields, A.B. (2018), “Marketing your university on social media: a content analysis of Facebook post types and formats”, Journal of Marketing for Higher Education, Vol. 28 No. 2, pp. 175-191, doi: 10.1080/08841241.2018.1442896.

Pieczka, M. (2002), “Public relations expertise deconstructed”, Media, Culture and Society, Vol. 24 No. 3, pp. 301-323.

Pleil, T. and Zerfass, A. (2014), “Internet und social media in der Unternehmenskommunikation [Internet and social media in corporate communication]”, in Zerfass, A. and Piwinger, M. (Eds), Handbuch Unternehmenskommunikation [Handbook of Corporate Communication], Springer, Wiesbaden, pp. 731-753.

Raudenbush, S. and Bryk, A. (2002), Hierarchical Linear Models, Sage, Thousand Oaks, CA.

Raupp, J. and Osterheider, A. (2019), “Evaluation von Hochschulkommunikation [Evaluation of university communication]”, in Fähnrich, B., Metag, J., Post, S. and Schäfer, M.S. (Eds), Forschungsfeld Hochschulkommunikation [Research Field University Communication], pp. 181-205, Springer, Wiesbaden.

Raupp, J. (2017), “Strategische wissenschaftskommunikation [Strategic science communication]”, in Bonfadelli, H., Fähnrich, B., Lüthje, C., Milde, J., Rhomberg, M. and Schäfer, M.S. (Eds), Forschungsfeld Wissenschaftskommunikation [Research Field Science Communication], pp. 143-163), Springer VS, Wiesbaden.

Roberts, M., Stewart, B., Tingley, D., Lucas, C., Leder-Luis, J., Gadarian, S., … Rand, D. (2014), “Structural topic models for open-ended survey responses”, American Journal of Political Science, Vol. 58 No. 4, pp. 1064-1082.

Roberts, M.E., Stewart, B.M. and Tingley, D. (2019), “Stm: an R package for structural topic models”, Journal of Statistical Software, Vol. 91 No. 2, pp. 1-40, doi: 10.18637/jss.v091.i02.

Roblyer, M.D., McDaniel, M., Webb, M., Herman, J. and Witty, J.V. (2010), “Findings on Facebook in higher education: a comparison of college faculty and student uses and perceptions of social networking sites”, The Internet and Higher Education, Vol. 13 No. 3, pp. 134-140, doi: 10.1016/j.iheduc.2010.03.002.

Rodriguez-Pomeda, J. and Casani, F. (2016), “Legitimating the world-class university concept through the discourse of elite universities' presidents”, Higher Education Research and Development, Vol. 35 No. 6, pp. 1269-1283, doi: 10.1080/07294360.2016.1160877.

Ryan, T. (2019), “Communication goals of American universities: A social media content analysis”, (Doctoral Dissertation), available at: https://digitalcommons.wku.edu/cgi/viewcontent.cgi?article=1804&context=stu_hon_theses.

Sabate, F., Berbegal-Mirabent, J., Cañabate, A. and Lebherz, P.R. (2014), “Factors influencing popularity of branded content in Facebook fan pages”, European Management Journal, Vol. 32 No. 6, pp. 1001-1011, doi: 10.1016/j.emj.2014.05.001.

Samuel, G., Williams, C. and Gardner, J. (2017), “UK science press officers, professional vision and the generation of expectations”, Public Understanding of Science, Vol. 26 No. 1, pp. 55-69, doi: 10.1177/0963662515597188.

Sandvig, J.C. (2016), “The role of social media in college recruiting”, International Journal of Web Based Communities, Vol. 12 No. 1, pp. 23-34, doi: 10.1504/ijwbc.2016.074273.

Scharkow, M. (2013), “Thematic content analysis using supervised machine learning: an empirical evaluation using German online news”, Quality and Quantity, Vol. 47 No. 2, pp. 761-773, doi: 10.1007/s11135-011-9545-7.

Serong, J., Koppers, L., Luschmann, E., Molina Ramirez, A., Kersting, K., Rahnenführer, J. and Wormer, H. (2017), “Öffentlichkeitsorientierung von Wissenschaftsinstitutionen und Wissenschaftsdisziplinen. Eine Längsschnittanalyse des ,,Informationsdienstes Wissenschaft“ (idw) 1995–2015 [Public orientation of scientific institutions and disciplines. A longitudinal analysis of the ‘Informationsdienst Wissenschaft’ (idw) 1995–2015]”, Publizistik, Vol. 62 No. 2, pp. 153-178, doi: 10.1007/s11616-017-0336-6.

Swani, K., Milne, G. and Brown, B.P. (2013), “Spreading the word through likes on Facebook”, The Journal of Research in Indian Medicine, Vol. 7 No. 4, pp. 269-294, doi: 10.1108/jrim-05-2013-0026.

Swani, K., Milne, G.R., Brown, B.P., Assaf, A.G. and Donthu, N. (2017), “What messages to post? Evaluating the popularity of social media communications in business versus con-sumer markets”, Industrial Marketing Management, Vol. 62, pp. 77-87, doi: 10.1016/j.indmarman.2016.07.006.54.

Thelen, P.D. and Men, R.L. (2018), “Strategic use of Facebook for public engagement in higher education institutions”, Public Relations Journal, Vol. 12 No. 2, pp. 1-27, available at: https://prjournal.instituteforpr.org/wp-content/uploads/Thelen_Men_StrategicUseofFacebook.pdf.

Valerio, G., Herrera-Murillo, D.J., Villanueva-Puente, F., Herrera-Murillo, N. and Rodríguez-Martínez, M.d. C. (2015), “The relationship between post formats and digital engagement: a study of the Facebook pages of Mexican universities”, International Journal of Educational Technology in Higher Education, Vol. 12, pp. 50-63, doi: 10.7238/rusc.v12i1.1887.

Van Ruler, B. (2005), “Commentary: professionals are from Venus, scholars are from Mars”, Public Relations Review, Vol. 31 No. 2, pp. 159-173.

Wallace, E., Buil, I. and de Chernatony, L. (2012), “Facebook ‘friendship’ and brand advocacy”, Journal of Brand Management, Vol. 20 No. 2, pp. 128-146, doi: 10.1057/bm.2012.45.

Wankel, L.A. and Wankel, C. (Eds) (2011), Higher Education Administration with Social Media: Including Applications in Student Affairs, Enrollment Management, Alumni Relations, and Career Centers, Emerald Publishing, Bingely.

Zerfass, A., Verčič, D. and Volk, S.C. (2017), “Communication evaluation and measurement”, Corporate Communications: An International Journal, Vol. 22 No. 1, pp. 2-18.

Corresponding author

Jens Vogelgesang can be contacted at: j.vogelgesang@uni-hohenheim.de