Tweeting the pandemic

6th January 2022 by Timo Hannay [link]

Another year, another SchoolDash social-media review (see our previous posts for coverage of 2020, 2019 and 2018).

That said, 2021 was far from a normal year. In recognition of these remarkable times, this post looks back over the last three years to that distant time when wearing a face mask meant you were a surgeon or a bank robber, and getting 'pinged' didn't usually involve cancelling your holiday plans. What has the intervening period looked like through the agitated, all-seeing eye of Twitter?

 

Viral spikes

Figure 1 shows daily numbers of education-related tweets from the UK during the three-year period from January 2019 to December 2021. These are based on a random sample of just under 1.2 million tweets that mention one or more education-related topics (such as 'schools' or 'students') and have been geolocated to the UK. Activity levels are normalised so that January 2019 has a mean value of 100.

At the finest-grained level, there is a weekly cycle with weekend dips and mid-week peaks. There are also regular lulls during the winter, summer and (to a lesser extent) spring holidays. But overlaid on these relatively predictable weekly and annual rhythms are some huge aberrant peaks that correspond to appropriately seismic events in the national education system, particularly in England: the first lockdown and accompanying school closures in March 2020, the A-level and GCSE results debacle in August 2020, the announcement of England's second national lockdown at the end of October 2020 (though schools remained open), the closing again of schools in January 2021, their eventual re-opening in March 2021 and a second consecutive non-exam season in August 2021.

It is also interesting to see that the longest-term trend of all is one of gradual decline, albeit interspersed with the series of Twitter-spasms described above. By late 2021, the rate of tweeting about education seems to have dropped to about half of that seen in early 2019. (This may be no bad thing: fond though we are of Twitter as a platform for data analysis, we're less sure about its value as medium for human communication.)

 

Watchwords

Figure 2 provides an initial glimpse into what all those tweets were about. It shows the words that characterised each month relative to the same months in 2018 and 2019, thus highlighting themes that were particular to 2020 and 2021.

During January 2021, discussion focused on school closures, while in February attention shifted towards the vaccine rollout and by March Twitter users were celebrating the fact that schools had re-opened – only to return to less encouraging subjects by April. In May, an orchestrated campaign drew attention to the government's "catastrophic" cuts in arts education budgets, but by June and July we were back to the all-too-familiar mixture of face masks, COVID tests and self-isolation.

The withdrawal of troops from Afghanistan in August threw a fleeting spotlight on the even greater trials faced by another country's education system, but in September it returned once more to the UK, where England found itself with a new education secretary, though mostly the same old problems – which continued into October. In November, the COP26 climate conference made a brief appearance and in December university lecturers' strikes hit the news, but the real obsession throughout remained COVID.

This unremitting viral gloom almost makes one yearn for the halcyon days of January 2020, when coronavirus barely registered and the main topic of discussion was the effects of Brexit on the Erasmus programme (which enabled British students to study across Europe), along with various petitions on sign language, sanitary products for girls and halal meat. Those were the days.

(Use the menu below to select a month to view.)

 

Subject examinations

Figure 3 provides a more quantitative view of the way in which interest in certain topics has waxed and waned over time. This works by identifying and searching for indicative words.

The start of the pandemic in March 2020 is very clearly visible on this timeline and although the level of tweeting has returned to pre-pandemic norms (or lower, as seen in Figure 1), the mix of topics certainly hasn't. A few were one-offs, such as the furore over racism in June 2020 following the murder of George Floyd in Minneapolis and the calls in May 2021 for the UK government to reverse cuts to arts education budgets. Other concerns came and went several times over, including school closures, school re-openings, COVID-related rules, tests and vaccinations, key workers and food poverty. Others still were more or less perennial points of discussion, especially COVID itself and distance learning (which predates the pandemic, since homework was a thing even in the days before COVID-19).

It's also interesting to note certain missing topics. For example, the COP26 climate conference, which took place in November 2021 and showed up briefly in Figure 2, hardly registered in Figure 3 and is therefore omitted.

(Click here to show all topics again. You can also click on the figure legend below to turn individual data sets on or off; double-click to show one data set on its own. Hover over the graph to see corresponding data values.)

 

If you're happy and you know it, send a tweet 🎶

We can also tell something about how happy or otherwise these Twitter users felt. As in our previous study, we do this by comparing the prevalence of emojis associated with positive (eg, 😀), neutral (eg, 😐) or negative emotions (eg, 😩), allowing us to calculate a 'Net Emoji Score'. This is basically the percentage excess of positive emojis over negative ones.

Figure 4 shows a rolling 7-day-average for all education-related tweets during the period from 2019 to 2021. During 'normal' times, in 2019, the overall score was postive – in other words, Twitter users were more likely to use happy emojis than unhappy ones. During the course of 2021, however, sentiment took a dive, reaching previously uncharted depths when schools were closed to in-person teaching in March 2020 and January 2021.

But interestingly, and perhaps surprisingly, positive emojis were in the ascendant during 2021, reaching a peak at the end of May. As far as we can tell, this wasn't due to any single topic, but rather a general sense of delight at things returning to more or less normal, with schools open again to in-person teaching and events such as school trips back on the agenda. True, things crashed back to earth at the end of the summer term (all those sad goodbyes) and at the end of the calendar year (on very low levels of tweeting, and possibly related to the Omicron variant), but throughout the latter part of 2021, levels were mostly in the normal pre-pandemic range of +10 to +30. If anything, they were a bit higher than the corresponding period in 2019.

Finally, let's look at how these sentiments correlated with particular topics of discussion. Figure 5 shows the average Net Emoji Score of tweets by COVID-related topic. The mean across all education-related tweets was +13.2 (ie, positive emojis were 13.2% more prevalent than negative ones). Topics to do with COVID invariably ranked lower than this, though exams, distance learning and school openings were all in positive territory. COVID rules, as well as tests and vaccinations, were somewhat negative, as were key workers (stemming from sympathy for their plight rather than any animosity). But the main targets of Twitter's rage were government policy, school closures and coronavirus itself.

So 2021 was an unusual year on Twitter too. Everyone was still banging on about COVID-19 and its effects on children and education, but there was less tweeting and what messages there were contained much more positive sentiment. For now, it looks like the new normal still involves talking about the pandemic, but also appreciating those things that we used to take for granted: in-person teaching, school trips and lessons well learned.

As ever, we welcome your thoughts: [email protected].