Materials and methods

Tweet threads were posted on Twitter via my handle (@EStafne) every Monday morning from 6 Apr. to 8 June 2020. Each thread consisted of 10 tweets that included information as well as a photo in most of them (Supplemental Fig. 1). All threads were first composed in Microsoft Word (Microsoft Corp., Redmond, WA), then edited to fit the 280-character limit of Twitter.

The thread was retweeted on Tuesday morning and Wednesday morning when it was archived on my blog site (Stafne, 2020). A hashtag, #YardFruits, was used to start the thread and make it searchable within Twitter for later reference. For the first nine threads, one species was discussed per thread. These species were, in order of posting, kumquat (Fortunella sp.), loquat (Eriobotrya japonica), blueberry (Vaccinium sp.), blackberry (Rubus subgenus Rubus), grape (Vitis sp.), muscadine (Vitis rotundifolia), passion fruit (Passiflora sp.), fig (Ficus carica), and pear (Pyrus communis). The final thread consisted of single tweets of the following species: mayhaw (Crataegus sp.), pineapple guava (Feijoa sellowiana), false strawberry (Potentilla indica), pomegranate (Punica granatum), southern dewberry (Rubus trivialis), fuzzy raspberry (Rubus moluccanus var. trilobus), poison ivy with Virginia creeper (Toxicodendron radicans, Parthenocissus quinquefolia), and pawpaw (Asimina triloba). The final tweet had a Twitter-generated poll to obtain feedback on the entire thread series by asking if the user gained knowledge from the series (yes or no). Final metrics were taken 1 week after the last thread posted, which included impressions (number of times a tweet shows up in a user’s timeline) and engagements (total number of times a user actively interacted with a tweet, including clicks, retweets, likes, replies, follows, and all other possible activity). Engagement rate is defined as the number of engagements divided by impressions.

Data were obtained from Twitter Analytics (Twitter, Inc.). Analysis was performed in JMP (version 12; SAS Institute, Cary, NC) with the Fit model procedure and means were separated by Tukey’s honestly significant difference test at P ≤ 0.05.

Results and discussion

The average Twitter user has ≈700 followers (Iqbal, 2020), but there is no accepted standard minimum number of followers for a successful account. Interest measured via engagement and steady follower growth would be aspirational targets. With those objectives in mind, my Twitter account saw a 6.5% increase in followers (+137) from April through June, growing from 1980 to 2117. My account grew by using well-targeted, focused tweets and tweet threads, especially those that used hashtags and photos (Barton et al., 2017; Seger et al., 2017). Hutto et al. (2013) found that key methods to attract followers included sharing informative content, encouraging discussion, and providing feedback. The 3 months before the #YardFruits tweets (January, February, and March) averaged ≈46,500 impressions, 17 new followers, and 269 profile visits; whereas April, May, and June averaged 121,000 impressions, 47 new followers, and 897 profile visits. This trend continued into July, after the thread series was over, with 118,000 impressions, 36 new followers, and 1048 profile visits, suggesting the sustained gains were likely a result of the tweet thread series.

Engagement percentage did not differ over time but did differ among the crop species (Table 1). The tweet thread from 8 June listed several different species vs. having only one in the other threads and had the highest engagement percentage (7.5%), possibly due to the broader appeal for my diverse follower audience. Grape, passionfruit, fig, and pear threads had the least engagement and were different from the Other Fruits thread. Kumquat, loquat, blueberry, blackberry, and muscadine were similar to the rest. The engagement percentage shows that tweet threads were primarily viewed as one-way communication (Chung, 2017; Su et al., 2017).

Table 1.

Twitter metrics for #YardFruits tweet thread of fruit crops from 6 Apr. to 8 June 2020.

View Table

Tweets absent a photo (10) received less engagement (2.7%) than those that included a photo [90 (4.7%)], consistent with other reports (Barton et al., 2017; Chung, 2017; Zhang et al., 2020) that show visual content can improve engagement. However, Han et al. (2019) found that audience preference may also play a role in desire for photos, videos, and hashtags. A few of the non-photo tweets (n = 6) included a GIF meme, used as a humorous placeholder when a specific photo was not available. With such a small sample size it was impossible to determine if they were as effective as a photo; however, the data available suggest an actual photo was preferred.

Mean number of likes declined over time (P = 0.01), whereas number of impressions, engagements, and retweets were not significantly different (data not shown). Reasons for this decline could potentially be explained by several factors, including warming weather leading to less computer time, overall COVID-19 news fatigue, and competing news that dominated feeds (e.g., riots, racial inequality).

Most previous studies focused on analyzing data across multiple Twitter users and not a single user with a specific, directed educational content thread, as this study did. As with most extension programs, the goal is transformative life and knowledge changes. Interactions with other Twitter followers and happenstance viewers were positive and they often replied to the tweet thread with questions or comments related to the content. The poll at the end of the final tweet thread had a 100% positive response that something was learned from the series, but there were only 22 respondents.

Growing a Twitter following is work and requires significant time to generate engaging content. However, a researcher or extension worker who wishes to grow their follower number could use the method presented here to do so. Social media has the benefit of quantitative data and thus can be reported as part of job-duty requirements (Gharis and Hightower, 2017). As universities continue to explore and encourage the integration of online learning environments, effective use of Twitter and other platforms will become normative, rather than explorative, activities.