Provides valuable benchmarks for science communication on the platforms within the

Provides valuable benchmarks for science communication on the platforms within the study as well as for future platforms. Also, the items do not represent a randomly distributed, year-round sample. “Throwback Thursday” items, for example, were posted only on Thursdays, “Guess What It Is” items were posted only on Mondays, and “Wow” items were posted only on Tuesdays. The 8 weeks included in the sample were not randomly distributed but represent only the end of thePLOS ONE | DOI:10.1371/journal.pone.0156409 May 27,14 /Engagement with Particle Physics on CERN’s Social Media Platformscalendar year. These temporal characteristics of the sample add further possible confounding factors to the study. Next, while the number of users for some platforms was high, two platforms had relatively few followers during the time of the study: Twitter French and Instagram. Hence, when outliers appeared on these platforms, it could be a result of chance. Additionally, when focusing on visit duration and retention rate, often the numbers that clicked-through in the first place were so small as to be only the hardened fans determined to spend time to find out more. Future work could include setting thresholds for inclusion in datasets, or devising measures that integrate numbers of users who clicked through and the time they spent on site. Finally, another limitation stems from the reliance on datasets provided by for-profit corporations such as Engagor. Engagor does not divulge the algorithms used to generate the data. Future work could focus on developing open, free tools for generating social media analytics for research purposes.Evidence-Based Insights for PracticeHow does our research inform practice? Specifically the practice of using social media for science communication. Here, we provide some evidence-based insights with examples (Table 9).NewsWith the increased prevalence of social media, news stories from an organisation are no longer confined to traditional media. Many people, especially younger generations, get their news directly from social media [48], either Vesatolimod web receiving it directly from the organisation or via a share from their social network. With such a wealth of news on social media, audiences react by liking, commenting, sharing and clicking-through, but stay very little time on the webpage, quickly consuming the content and moving on. This type of post is therefore more focused on marketing and engagement and less on education in terms of the social media strategy.Images and AnimationUsers respond more readily to images than text on social media [49,50]. In order to control for this variable, all posts in the study contained an image. Furthermore, Facebook algorithms are configured to promote image-based posts to a wider audience. In this study, animations were also used on Twitter and Google+, receiving a relatively strong reaction in terms of likes and shares [51,52]. Of the 35 high engagement items, more than half were not news related but involved beautiful images (e.g. [53]) or surprising images (e.g. [54]). Meaning that an organisation can use imagery on social media for all three strategic themes: marketing, engagement andTable 9. Content characteristics and related user behaviour on social media. Likes News Image Animation Video/Virtual tour on webpage P144 Peptide chemical information Discussion Clickbait Tailored content Human story doi:10.1371/journal.pone.0156409.t009 Comments Shares Click-throughs Visit duration Retention ratePLOS ONE | DOI.Provides valuable benchmarks for science communication on the platforms within the study as well as for future platforms. Also, the items do not represent a randomly distributed, year-round sample. “Throwback Thursday” items, for example, were posted only on Thursdays, “Guess What It Is” items were posted only on Mondays, and “Wow” items were posted only on Tuesdays. The 8 weeks included in the sample were not randomly distributed but represent only the end of thePLOS ONE | DOI:10.1371/journal.pone.0156409 May 27,14 /Engagement with Particle Physics on CERN’s Social Media Platformscalendar year. These temporal characteristics of the sample add further possible confounding factors to the study. Next, while the number of users for some platforms was high, two platforms had relatively few followers during the time of the study: Twitter French and Instagram. Hence, when outliers appeared on these platforms, it could be a result of chance. Additionally, when focusing on visit duration and retention rate, often the numbers that clicked-through in the first place were so small as to be only the hardened fans determined to spend time to find out more. Future work could include setting thresholds for inclusion in datasets, or devising measures that integrate numbers of users who clicked through and the time they spent on site. Finally, another limitation stems from the reliance on datasets provided by for-profit corporations such as Engagor. Engagor does not divulge the algorithms used to generate the data. Future work could focus on developing open, free tools for generating social media analytics for research purposes.Evidence-Based Insights for PracticeHow does our research inform practice? Specifically the practice of using social media for science communication. Here, we provide some evidence-based insights with examples (Table 9).NewsWith the increased prevalence of social media, news stories from an organisation are no longer confined to traditional media. Many people, especially younger generations, get their news directly from social media [48], either receiving it directly from the organisation or via a share from their social network. With such a wealth of news on social media, audiences react by liking, commenting, sharing and clicking-through, but stay very little time on the webpage, quickly consuming the content and moving on. This type of post is therefore more focused on marketing and engagement and less on education in terms of the social media strategy.Images and AnimationUsers respond more readily to images than text on social media [49,50]. In order to control for this variable, all posts in the study contained an image. Furthermore, Facebook algorithms are configured to promote image-based posts to a wider audience. In this study, animations were also used on Twitter and Google+, receiving a relatively strong reaction in terms of likes and shares [51,52]. Of the 35 high engagement items, more than half were not news related but involved beautiful images (e.g. [53]) or surprising images (e.g. [54]). Meaning that an organisation can use imagery on social media for all three strategic themes: marketing, engagement andTable 9. Content characteristics and related user behaviour on social media. Likes News Image Animation Video/Virtual tour on webpage Discussion Clickbait Tailored content Human story doi:10.1371/journal.pone.0156409.t009 Comments Shares Click-throughs Visit duration Retention ratePLOS ONE | DOI.