Full data for #Covid19uk during UK lockdown

This blog provides access to data on tweets using the #Covid19UK hashtag during the UK lockdown, the first day of which began on 24 March 2020. The data were extracted using TAGS, and then mapped using NodeXL. There were of course other UK-focused hashtags used during this period and some UK-based Covid-19 tweeting that did not use any hashtag. However it was unfeasible to capture all the data. I therefore stuck with one hashtag all the way through. You can see further information, including other search terms and ways of presenting data, in a tweet thread. These data are updated on a weekly basis (except where there is too much data to plot at once, in which case they are charted by day). There’s more about this work on the BMJ Opinion blog.

Details are provided below, but you may want to start with the summary outputs by month between March/ April and December. These summarise the top tweets (by number of retweets received) until end of July 2020, before moving to a different approach from August 2020 onwards where I attempt to capture more diversity in tweeters by making sure that no tweeter has more than 3 tweets in the summaries. Click to go to the Wakelet summary. Note that some tweets may have subsequently been deleted by Twitter or the tweeter, and some users will have left Twitter or been suspended. Accordingly, I have included PDF summaries that capture a permanent record. Links to the PDF summaries are included in the Wakelet summaries for each month:

From March 2022 I have not been producing Wakelet summaries, but the full data are available in NodeXL maps (data at end of each NodeXL report):

I aim to combine outputs from this period into a summary Wakelet when time.

The URLs in the table point to NodeXL reports that describe the data more fully. Note that some of the early days of the UK lockdown had so much tweeting that it was not possible to obtain a complete data set (either because of the 18,000 tweet & retweet limit or because of difficulties accessing the Twitter API; you can only extract Twitter data 7-10 days into the past). Fortunately NodeXL also allows us to look at the tweet data for posts retweeted during the periods where there are data. That means that we can access some of the more popular tweets from the periods that were “missed”. For example, for 20 March, pre-lockdown, there were only 8 hours 21 minutes of data extracted (15:38 to 23:59). However, retweets during this period allowed the identification of tweets going back over 2 weeks. The following graph shows daily data from 3 to 17 March and hourly data thereafter, plotting tweets but not retweets.

Click into the links in the table below to see the NodeXL reports. These are typically weekly reports, starting each Tuesday (as the lockdown began on a Tuesday). Scroll down to the bottom of the NodeXL reports to download the raw data (unless the table below has a link for “data”, in which case the file was too large to upload to the NodeXL Graph Gallery, and needed to be uploaded to Dropbox instead). In the NodeXL reports select the link – “Download the Graph Data as a NodeXL Workbook”. Sometimes there was just too much data to analyse in one go, so I have processed them in days instead. Most of the NodeXL reports were produced in September or October. That means that some of the tweets had been deleted, retweets retracted, sometimes accounts deleted. For example, for 24 March – re-extracted on 1 October – 14,735 tweets and retweets were identified out of 16,192 originally collected. There may be some interesting findings from looking at the posts that have disappeared in the intervening period – perhaps some were from “troll farm” and bot accounts that have been removed. I have a lot of the original TAGS data, which will allow further study of this question.

NodeXL graphs (e.g. the graph below for 24 March) show Twitter accounts that tweeted, retweeted, and/or were mentioned in tweets. These are called “vertices”. You can read more about how to interpret these maps in an article by NodeXL and Pew Research. You can also read a lot more about how to analyse and interpret the raw data on the ScotPublicHealth blog and in my papers published in peer-reviewed medical journals.

The number of Twitter accounts included in the weekly extracts has varied over time, peaking at the start of lockdown, a quiet period over the summer, an increase during September as we contemplated a return of stricter lockdown rules, and again the week before 2nd lockdown rules were introduced in England (2nd national lockdown commenced 5 November), with further restrictions for 26 December and nationwide lockdowns announced 5 January 2021.

Hopefully there will be time over coming months to analyse these data in collaboration with academic teams. Please contact me if you are interested in contributing. However, for the moment my focus is on GP exams and clinical work. Watch this space for further data over coming months.

Week of lockdownDatesVertices (*)URL to access data
Pre lockdown
(4 days)
20-23 March5073720 March  
21 March  
22 March  
23 March
124 – 30 March8634824 March
25 March
26 March
27 March
28 March
29 March
30 March
231 March – 6 April6108431 March
1 April
2 April
3 April
4 April
5 April
6 April
37 – 13 April42330Report
414 – 20 April40443Report
521 – 27 April36964Report
628 April – 4 May35491Report
75– 11 May55688Report
812- 18 May5537712 May
13 May
14 May
15 May
16 May
17 May
18 May
919 – 25 May44803Full week
1026 May – 1 June44714Full week
112– 8 June25091Full week
129 – 15 June21481Full week
1316-22 June18857Full week
1423-29 June30340Full week
1530 June – 6 July23065Full week
167 – 13 July24174Full week
1714 – 20 July14243Full week
1821- 27 July9191Full week
1928 July – 3 August14549Full week
204 – 10 August8357Full week
2111 – 17 August7618Full week
2218 – 24 August6125Full week
2325– 31 August8143Full week
241 – 7 September9774Full week
258 – 14 September35498Full week
2615 – 21 September24634Full week
2722 – 28 September24250Full week
2829 Sep – 5 October13756Full week
 296 – 12 October  17766 Full week 
 3013 – 19 October  19491Full week 
 31 20 – 26 October 21560Full week 
32 27 Oct – 2 Nov31373 Full week 
33 3 – 9 November 13967Full week 
3410 – 16 November8932Full week
3517 – 23 November11577Full week
3624 – 30 November10993Full week
 371 – 7 December 6670  Full week
388 – 14 December5527 Full week
3915 – 21 December21024 Full week
4022 – 28 December13065 Full week
4129 Dec – 4 Jan 202126346 Full week
425 – 11 January30037Full week
4312 – 18 January21996Full week
4419 – 25 January20418Full week
4526 Jan – 1 Feb14400Full week
462 – 8 February7567Full week
479 – 15 February8257Full week
4816 – 22 February9270Full week
4923 Feb – 1 Mar8218Full week
502 – 8 Mar6356Full week
519 – 15 Mar3968Full week
5216 – 22 Mar +3737Full week
#covid19uk data by week from the start of UK lockdown

* (number of tweeters, retweeters & mentioned accounts)

+ Note that there is also an extract for 1-23 March 2021 concluding a full year’s worth of data for the 1st anniversary of lockdown. Data beyond this point are available from the monthly Wakelet summaries listed at the top of this post.

The write up of the first full year (and a day) of tweeting is available in a Lancet Infectious Diseases Media Watch article.

Dr Graham Mackenzie, Edinburgh, Scotland

Updated 3 September 2022

graham mackenzie on twitter

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s