2020-07-15: Revisiting Twitter Follower Growth for the 2020 Democratic Candidates

By Nauman Siddique -
Figure 1: Screenshot of 24 of 27 Democratic candidates hopeful of the Presidential nomination on May 25, 2019
(Source: The Washington Post)
In our previous post, Twitter Follower Growth for the 2020 Democratic Candidates, we used the Twitter follower growth between 2019-01-01 and 2019-08-23 as a proxy to measure popularity of the Democratic candidates and categorized the campaign of each candidate based on their absolute Twitter follower growth into four categories: already popular, big winners, nobody noticed, and beneficial. Since our previous post, all the candidates except Joe Biden, the presumptive presidential nominee, have dropped out. In this post, we revisit the Twitter follower growth of all 27 Democratic candidates by extending our study to between 2019-01-01 and 2020-04-18. We plotted an interactive D3 Twitter follower graph to show the trends in the absolute increase of the followers for each candidate.   

Twitter and Web Archives

Forbes, in their article "Campaign 2020: Who Is Winning On Twitter? Revealed In 6 Graphs", details the usage of Twitter by all the Democratic candidates who qualified for the October 2019 Debate and used their absolute increase in Twitter followers for September 2019 as a proxy to measure their campaign success. Lawsuit, in their article "Politics and Fake Social Media Followers", say that "a candidate's social media popularity on social media may not always be what it seems". For supporting their claim, they examined the Twitter and Instagram followers for all of the 2020 Democratic candidates and reported on the percentage of bots or fake accounts identified for each of the candidates. Statista, in their article "Number of Twitter Followers of Democratic candidates for the 2020 U.S. presidential election as of January 2020", shows a bar chart of Twitter follower count for all of the 2020 Democratic candidates. FiveThirtyEight published their two-article series about the NBC and CNN Democratic debates where they analyzed all of the 2020 Democratic candidates based on five criteria. One of the criteria was the Twitter follower growth chart from the night of the debate to the following afternoon. Although the articles used Twitter to measure the social media popularity of the Democratic candidates, they did not perform a longitudinal study of Twitter.

For a better understanding of the Twitter follower growth as a measure of a candidate's popularity, we need to include all the events that have occurred in a candidate's campaign and after their campaign withdrawal, which requires us to study their historical Twitter information. The Twitter API returns current follower count value for a Twitter handle. In order to perform a longitudinal study of Twitter follower count for the Democratic candidates over the past 16 months using the Twitter API, we would have had to actively record the follower count over the entire study time. Since we did not start that process in January 2019, we have to use the web archives for studying the historical Twitter follower count. The technique of using web archives to study historical Twitter follower count was introduced by Miranda Smith in her post "Twitter Follower Count History via the Internet Archive". For this post, we extended the tool created by Miranda Smith to parse the Twitter follower count from the web archives (Historical Twitter Follower Count via Web Archives) and used it to analyze the historical Twitter follower count of Democratic candidates to present a broader view of the Twitter follower growth for each candidate. 

Since the Twitter API doesn't provide historical data, we have to use the web archives containing mementos of the Twitter accounts which need to be scraped for parsing of useful Twitter information. There are challenges involved in scraping the mementos. Figure 2 shows a memento for the Twitter account, @berniesanders, in Bulgarian captured by the Internet Archive. Sawood Alam has explained the role of cookies in archiving of non-English mementos ("Cookie Violations Cause Archived Twitter Pages to Simultaneously Replay in Multiple Languages" and "Cookies Are Why Your Archived Twitter Page Is Not in English"). In order to parse follower count from the Bulgarian language memento, we need to use translators.

Beginning on July 15, 2019, Twitter rolled out a new user interface and on July 1, 2020 stopped supporting the prior user interface. Figure 3 and 4 show that the migration of Twitter user interface from legacy theme to the new user interface had implications on the archival of Twitter mementos  (post by Kritika Garg and Himarsha Jayanetti). Alexander Nwala, in his post "Twitter broke my scrapers", highlights the issues involved with web scraping and impact of the new Twitter user interface.  

Figure 2: Screenshot of a memento for @berniesanders on June 7, 2020 in Bulgarian with the Twitter legacy theme user interface from the Internet Archive

Figure 3: Screenshot of a memento for @berniesanders on June 20, 2020 from the Internet Archive with a "browser not supported" message (Figure 3 and Figure 4 both link to the same URL)

Figure 4: Screenshot of a memento for @berniesanders on June 20, 2020 from the Internet Archive with a "something went wrong" message (Figure 3 and Figure 4 both link to the same URL)

Creating the dataset

We used the 2020 Democratic Party presidential primaries Wikipedia webpage on April 18, 2020 to create our baseline of the candidates. By April 18, 2020, 26 candidates have dropped out of the Democratic primary contest, and Joe Biden is the presumptive presidential nominee. With 25 of 27 candidates announcing their candidacy in 2019, we limited our study between January 1, 2019 and April 18, 2020. In our previous post, Twitter Follower Growth for the 2020 Democratic Candidates, we used the Twitter handle for all 21 candidates mentioned in the FiveThirtyEight article, and for rest of the candidates we used the Twitter accounts that mentioned 2020 Presidential candidacy in their Twitter bio. The only new candidate included in the analysis for the current post is Mike Bloomberg, who announced his candidacy on November 29, 2019.

We collected mementos from multiple web archives using MemGator for all the candidates' Twitter handles between January 1, 2019 and April 18, 2020 and parsed out the follower count from each memento to build our dataset of historical follower count (Historical Twitter Follower Count via Web Archives).  

Follower Graph

We plotted an interactive D3 Twitter follower graph for all 27 Democratic candidates between 2019-01-01 and 2020-04-18 with time on the X-axis and follower count on the Y-axis, which is zoomed in and scrollable because the range of follower count is so large (5k - 12M). 

Figure 5: Screenshot of the 2020 Democratic Candidates Twitter Follower Graph showing the main graph (left) and widgets to control the main graph (right) with minimap widget on the top right, followed by the candidates widget, and  events widget at the bottom.

How to use the graph

  • There is a minimap widget on the top right of the graph, which provides an overview of the graph and allows users to scroll the graph.     
  • Below the minimap widget, we have the candidates widget in sorted order of follower count with checkboxes that allow users to add or remove the line of any candidate. Furthermore, clicking the name of a candidate in the candidates widget brings the line for the candidate into viewport of the main graph. The graph icon beside each candidate links to the Twitter follower growth (absolute and relative) and daily new Twitter follower growth rate (absolute and relative) graphs generated for each candidate. The Twitter icon beside each candidate links to their live Twitter account. 
  • Below the candidate widget, we have the events widget, which allows users to add a vertical line indicating real-world events including debates, primaries and caucuses, and campaign announcement and withdrawal of each candidate onto the graph. Hovering the mouse over any of the debate or primary and caucus lines dims the line of the candidates who did not attend that particular debate or primary and caucus. Hovering the mouse over any of the campaign announcement and withdrawal points shows name of the candidate and date of the campaign announcement or withdrawal. For the two candidates (@andrewyang and @johndelaney) who announced their campaign prior to 2019, hovering the mouse over their campaign announcement point shows a "Pre-2019" message.   
  • The users can view the memento datetime and the Twitter follower count of each memento by hovering over the dots on the line plotted for each candidate.
  • The users can highlight the line of a particular candidate by clicking on their line or the label of the candidate. The candidate highlight can be reversed by clicking again on the label or the line.


Analysis of the Graph

Table 1 is a sortable table that contains the name, Twitter handle, memento date range, start follower count, end follower count, campaign withdrawal date, and follower count growth in absolute numbers and in percentage for all the candidates. Memento date range for a candidate represents the start and the end datetime of all the mementos collected from the web archives for their Twitter account. Increase represents the increase in absolute number of Twitter followers for a candidate from the first memento to the last memento. Increase% represents the percentage increase in the follower count for a candidate with respect to their first memento follower count.


Table 1: List of all the Democratic Candidates with their follower count increase between January 2019 and April 2020.
"*" represents the presumptive presidential candidate.
Name
Twitter Handle		 Memento Date Range Start Follower Count End Follower Count Campaign Withdrawal Increase Increase%
Michael Bennet
@michaelbennet		 2019-01-11 to 2020-03-02 8,461 46,503 2020-02-11 38,042 449.6
Joe Biden*
@joebiden		 2019-01-01 to 2020-04-16 3,173,757 4,967,402 1,793,645 56.5
Cory Booker
@corybooker		 2019-01-01 to 2020-04-07 4,083,516 4,481,248 2020-01-13 397,732 9.74
Pete Buttigieg
@petebuttigieg		 2019-01-14 to 2020--04-17 82,321 1,848,503 2020-03-01 1,766,182 2145.48
Julian Castro
@juliancastro		 2019-01-01 to 2020-03-28 136,161 480,559 2020-01-02 344,398 252.93
Bill de Blasio
@billdeblasio		 2019-01-06 to 2020-03-30 139,007 214,090 2019-09-20 75,083 54
John Delaney
@johndelaney		 2019-03-05 to 2020-02-29 15,156 38,718 2020-01-31 23,562 155.46
Steve Bullock
@governorbullock		 2019-01-01 to 2020-03-31 165,709 192,162 2019-12-02 26,453 15.96
Tulsi Gabbard
@tulsigabbard		 2019-01-01 to 2020-04-09 216,652 815,791 2020-03-19 599,139 276.54
Kamala Harris
@kamalaharris		 2019-01-01 to 2020-04-16 1,988,958 3,567,087 2019-12-03 1,578,129 79.34
Amy Klobuchar
@amyklobuchar		 2019-01-24 to 2020-04-15 550,054 1,037,407 2020-03-02 487,353 88.60
Beto O'Rourke
@betoorourke		 2019-01-01 to 2020-04-13 1,110,617 1,653,745 2019-11-01 543,128 48.9
Tim Ryan
@timryan		 2019-04-05 to 2020-03-25 16,046 38,794 2019-10-24 22,748 141.77
Bernie Sanders
@berniesanders		 2019-01-01 to 2020-04-18 8,942,649 11,809,372 2020-04-08 2,866,723 32
Elizabeth Warren
@ewarren		 2019-01-01 to 2020-04-17 2,156,119 4,099,217 2020-03-05 1,943,098 90.1
Marianne Williamson
@marwilliamson		 2019-02-07 to 2020-03-30 2,599,393 2,787,174 2020-01-10 187,781 7.2
Andrew Yang
@andrewyang		 2019-03-20 to 2020-04-15 184,317 1,399,233 2020-02-11 1,214,916 659.14
Kirsten Gillibrand
@sengillibrand		 2019-01-01 to 2020-04-13 1,296,705 1,511,166 2019-08-28 214,461 16.54
Joe Sestak
@joesestak		 2019-01-11 to 2020-03-01 10,041 13,642 2019-12-01 3,601 35.86
Wayne Messam
@waynemessam		 2020-04-16 to 2020-02-28 5,834 8,410 2019-11-19 2,576 44.15
Eric Swalwell
@ericswalwell		 2019-01-01 to 2020-04-11 23,866 208,163 2019-07-08 184,297 772.22
John Hickenlooper
@hickenlooper		 2019-03-03 to 2020-04-11 128,636 172,360 2019-08-15 43,724 33.99
Jay Inslee
@jayinslee		 2019-01-01 to 2020-04-08 23,865 156,606 2019-08-21 132,741 556.22
Tom Steyer
@tomsteyer		 2019-01-01 to 2020-03-25 211,194 317,552 2020-02-29 106,358 50.36
Mike Gravel
@mikegravel		 2019-03-20 to 2019-11-15 13,737 136,689 2019-08-06 122,952 895
Seth Moulton
@sethmoulton		 2019-01-30 to 2020-04-14 135,567 148,122 2019-08-23 12,555 9.26
Mike Bloomberg
@mikebloomberg		 2019-01-01 to 2020-04-16 2,282,170 2,756,611 2020-03-04 474,441 20.79

Figure 6: Screenshot of Mike Gravel's Twitter account (@mikegravel) captured by Internet Archive on 2019-11-15
  • Mike Gravel's Twitter account, @mikegravel, saw an 895% increase with an absolute increase in the follower count of approximately 123k. After announcing his campaign withdrawal on August 25, 2019, he deleted his Twitter account on November 15, 2019. Figure 6 shows the memento captured by Internet Archive for the Twitter account, @mikegravel, before it was deleted. Figure 7 shows the next memento captured by Internet Archive on November 21, 2019 with the message that the account has been deleted. Mike Gravel is the only candidate to delete his Twitter account after dropping out, thereby negating the use of his increased platform for future endeavors. 
  • There is an 80% similarity between the list of top 10 candidates based on the absolute increase in the Twitter follower count and the final 10 candidates remaining in the nomination race based on their campaign withdrawal date. Of the top 10 candidates based on the absolute increase of the Twitter follower count, Beto O'Rourke and Kamala Harris are the only two candidates who did not appear in the list of the final 10 candidates running for the Democratic Presidential nomination  based on their campaign withdrawal date. 
  • The analysis of daily growth in the new followers of all 26 candidates who have dropped out of the Presidential nomination reveals that: 
  • Based on the absolute increase in the Twitter follower count, nine of the 10 candidates with the least absolute Twitter follower growth have appeared in a maximum of two debates, with the final appearance for any of these candidates being the second debate held by CNN on July 30, 2019.  
  • The article Tulsi Gabbard’s Social Media Stats ≠ Sense highlights the meteoric rise of approximately 135k Twitter followers for Tulsi Gabbard between 2019-10-01 and 2019-10-25 after appearing on the CNN Debate held on October 15, 2019. Using our Twitter follower graph, we narrowed the sharp spike period for Tulsi Gabbard to be between October 1-25, 2019. 

Conclusions

We revisited the Twitter follower growth of all 27 Democratic candidates between 2019-01-01 and 2020-04-18. We found an 80% similarity between the list of top 10 candidates based on absolute increase in the Twitter follower count and the final 10 candidates remaining in the 2020 Democratic Presidential nomination race based on their campaign withdrawal date. Since the campaign withdrawal, we found four candidates (@betoorourke, @tulsigabbard, @waynemessam, and @joesestak) with decreasing follower counts and five candidates (@berniesanders, @ewarren, @amyklobuchar, @tomsteyer, and @billdeblasio) with higher daily new follower growth rate than their campaign duration daily new follower growth rate. We found nine of 10 candidates with the least absolute Twitter follower growth based on absolute increase in the follower count to have appeared in a maximum of two debates, with their last appearance being in the CNN debate held on July 30, 2019. 

Finally, we created an interactive D3 Twitter follower graph for all 27 Democratic candidates, which can be used to analyze the historical Twitter follower counts and the impact of real-world events like debates, primaries, caucuses, and campaign announcement and withdrawal onto the follower count of each candidate.

Interactive D3 Chart: https://oduwsdl.github.io/2020DemFollowerGraph/
D3 Graph Code: https://github.com/oduwsdl/2020DemFollowerGraph

Update

---------
Mohammed Nauman Siddique
(m_nsiddique)

Comments

Post a Comment