Original research from the team at Buffer, the social media sharing tool, unleashed a huge number of tweets and retweets yesterday with their article The Biggest Social Media Science Study: What 4.8 Million Tweets Say About the Best Time to Tweet. The team at Buffer analyzed the data of their 2 million users to find correlations between tweet engagement – clicks, retweets, favorites, & replies – and the time of day those tweets were posted.
You can read the full post here for the details of the study, which spanned users across the globe and in all time zones. There are so many interesting takeaways from the results the Buffer team found that we had trouble picking among them, but here are a few:
- Tweets sent between 2-3 AM earn the most clicks on average.
- Tweets sent between 2-3 AM earn the most total engagement on average.
- The most popular time to tweet is 12-1 PM.
- In the US, tweets sent at 9 PM earn the most retweets and favorites on average.
By total engagement, they’re including retweets and favorites, although most engagement actions were clicks. Be sure to check out their time-zone-specific breakdown on “the specifics on each of the best time to tweet for clicks”. (Hint: it probably won’t affect your lunch plans very much.)
Buffer’s analysis yielded a number of interesting findings, including the fairly large discrepancy between the most popular times to send a tweet and the most effective times to send a tweet you want retweeted (or otherwise actively engaged). Their study inspired us to look around ACI for other Twitter studies being done, and not surprisingly, it’s a really popular topic. Read on for a smattering of posts on current Twitter research with sometimes-surprising results.
We’ll start with the R-Statistics Blog by Tal Galili, which has numerous posts analyzing data from Twitter. In one post, Data Science Tweet Analysis – What tools are people talking about?, R-Statistics contributor Chris Musselle at Mango looks at hashtag counts and co-occurence with terminology and references to data science and big data. Through his research, he found that “Tweets mentioning data science most commonly include hashtags for general analytics concepts and ‘buzzwords’, with specific technologies only occasionally mentioned.”
Galili himself takes on Twitter usage and behavior during conferences in the post Comparing #rstats and #pdf15 intraday hashtag streams. In the post, he writes: “What we see is that tweet activity is greater during sessions than during lunch and breaks. This implies that 1) people were actively tweeting about presentations, and 2) people were actually mingling as opposed to being glued to their phones during networking breaks. The exception is at the end of a session, where it appears people take a moment to tweet about the session before heading out.”
Galili also publishes the R-Bloggers blog. Because each blog post is followed by a corresponding tweet, he used that data to look at how data on retweets and favorites might differ by day of week and posted his findings in Analyzing R-Bloggers’ posts via Twitter in an analysis similar to that of the Buffer team. The weekday that yielded the highest tweets per post was actually tied, with both Mondays and Fridays receiving the highest number of retweets, and Thursdays at 3rd place for highest retweets. For the highest number of favorites per post, Mondays and Thursdays tied for first place, with Fridays just trailing behind. We say: between this one and the Buffer post, you can safely begin to hash out your Twitter posting/sleep schedule with complete confidence. Well, maybe complete isn’t the right word … in the post URL Originality Analysis back in R-Statistics, he analyzed the distribution of initial posts and shares of links shared via Twitter and found that discoverability and retweeting don’t necessarily go hand-in-hand.
The Research Newsroom blog at Macquarie University also posted some pretty fascinating Twitter studies, this time going in a very different direction with the post New study shows how the echo chamber effect amplifies misinformation about HPV vaccines online. Led by Dr. Adam Dunn from Macquarie University’s Australian Institute of Health Innovation, this study found that vaccine-related bias could be predicted just from social connections alone, without even needing the tweets’ links or written content. According to the blog, Dr. Dunn stated, “We think these results are less likely the consequence of opinion contagion and more likely due to an echo chamber effect – where users are preferentially connected to other users who share their views. This has important implications for trying to deal with polarisation that can shape and hold public opinion about vaccines.”
Hmm… it looks like Wired’s Zoologic might agree with you, Dr. Dunn. While Macquarie’s study focused on HPV vaccines, Zoologic’s post Anti-Vaxxers Are Using Twitter to Manipulate a Vaccine Bill points out a similar pattern with measles vaccines, and performed their own hashtag analysis with some pretty fascinating findings.
In What is the half-life of a tweet? in her blog The Culture of Chemistry, Michelle Francl-Donnay did a little exploring of her own data after noticing a similarity between a Twitter analytics graph and a half-life graph from her chemical kinetics courses. It inspired her to analyze five data sets from a month of her own tweets to check out her tweet’s average half-life, which turned out to be about the same amount of time the Apple Watch battery might last under heavy use.
Heather Doran is tackling the ultimate #scicomm takedown in her research; in posts like Why scientists use social media, she details her research into the use (and non-use) of social media among scientists, much like the ongoing research from Paige Brown Jarreau in From the Lab Bench. In her post, Doran notes, “Many people reported opening a Twitter account and then not knowing what to do with it, ignoring it for a few weeks/months and then slowly going back to it after finding a community that interested them or a person that they were interested in following.”
The student learning perspective is covered in the post Study examines impact of texting and tweeting on academic performance, where the Inside Higher Ed Technology and Learning Blog describes an article by Jeffrey Kuznekoff published in Communication Education. In that study, Kuznekoff investigates the effects of mobile phone usage in class on student learning and performance. The study found that students who did not use mobile phones – or who engaged in “relevant texting” – fared better overall than those engaged in Twitter or non-relevant texting. The results might be a bit of a bummer for the most dedicated Tweeps, but surely there is class-relevant and non-class-relevant tweeting as well – so maybe it’s a matter of finding more effective Twitter activities for the classroom. Be sure to check out the full text of that study here; we think it will inspire some great future research into a plethora of Twitter engagement ideas for instructors.
Feeling inspired yet to perform your own Twitter research? Check out the referenced Twitter profiles below to get a head start and even more inspiration – and be sure to share your Twitter research ideas and projects in the comments below.