Gender (and Race) in the 2020 Primaries: No, the Playing Field Isn’t Level.
Classifying twitter conversations
A recent column by Margaret Sullivan about the extent of sexism in the current coverage of 2020 POTUS candidates concludes that many storylines about female candidates would never have been written about men. In Marvelous AI’s analysis of Twitter conversations, we see evidence for the same. In this post, we will quantify and dissect differences between how male and female candidates are being discussed.
In a post last week, we discussed how we clustered tweets mentioning Kamala Harris to detect common themes. For the week ending February 23, we applied similar clustering techniques to the Twitter conversations about a couple dozen potential 2020 Democratic presidential candidates. We then manually categorized each cluster into the following categories (examples for each category below).
- Policy: a discussion of the candidate’s views on issues, either as candidates or as currently serving public officials.
- Campaign: horse-race discussion about campaign events or candidate rankings.
- Character: discussions about a candidate’s behavior, racial / ethnic identity, sexuality, appearance, hypocritical behavior. Although some of the discussion calls out the sexist storylines, we included those in the same category because, as George Lakoff would tell you, negating a frame is activating a frame, and it takes space away from more productive conversations.
- Corruption: a particular line of right-wing attack, mostly on former Obama or Clinton administration officials.
- Culture: discussion of culture wars hot-spots like Jussie Smollett or Covington High School (and candidates’ reactions to these developing stories).
For each cluster type, we computed the total volume of tweets it covers and the average left-right bias of the twitter users posting the tweets (see earlier post discussing how author bias is determined). We visualize the cluster types for a particular candidate as follows. Each color-coded bar represents a cluster type, with height representing the proportion of Twitter traffic covered by the cluster. The position of the bars on the x-axis represents the average political bias for that cluster, with the x-axis color-coded to indicate left (blue) / right (red) bias.
As you can see, over 60% of the clustered coverage for Klobuchar is character-related; 38% is policy-related, and less than 2% is campaign related. All of the coverage is slightly left-of-center. In addition to the character example above, here are some others:
Quantifying conversation types
Is this amount of character-focused coverage typical for presidential candidates? Let’s take a look at the field as a whole. We created the same sort of graph for all of the top- and middle-tier candidates and placed them in two columns. Male candidates are on the left, female candidates are on the right. The y-axis position of a candidate’s chart is proportional to that candidate’s Twitter mentions, approximating popularity (though not necessarily standing in the polls).
The difference between male and female candidate coverage is striking. First impressions:
- Female candidates, on the whole, receive much more character-focused coverage than male candidates, with a significant outlier in each categories, as discussed below.
- Top-tier female candidates receive more character-focused coverage than lower-tier ones.
- Looking at the political bias of character coverage, for most candidates it is fairly centrist, confirming the intuition that mainstream media still can’t stop itself from promoting sexist storylines. The two exceptions are Harris and Booker, whose character attacks are much more right-wing.
- Policy-oriented discussion for top-tier candidates is scant.
What do “character” conversations actually talk about?
The details of character-focused discussions differ for female vs. male candidates. As discussed above, main themes for Klobuchar are her temperament and vengefulness. For Warren, it’s her identity and honesty.
Kamala Harris gets hit for supposedly misrepresenting her family background, and for her appearance / fashion.
The previous week, we also saw attacks related to her sexual history.
All of these storylines are common in media coverage of female politicians, as shown in prior studies1.
As Lucina di Meco says in her upcoming paper,
“Coverage of female politicians seems to be particularly negative as they try to break the highest glass ceiling, running for Prime Minister or President against a male candidate, therefore openly defying societal expectations of women as supportive more than competitive, best fit for the role of the great woman behind every great man (Kittilson and Fridkin 20082; Meeks 20123).”
Tulsi Gabbard: exception that proves the rule
Gabbard is the only female candidate who does not receive scores of character-driven attacks. In fact, on the face of it, most conversations mentioning her are concerned with policy. However, she is widely thought not to be a credible candidate, and in fact her policy positions are attacked as borderline treasonous from all sides of the spectrum in the mainstream media (e.g. NBC News story). On Twitter, however, most of the Gabbard-related conversation is oddly positive but coming from suspicious characters like the possibly fictitious Sarah Abdallah, who we highlighted as a prominent backer in a post two weeks ago.
Gabbard policy tweet
The lack of character attacks on Gabbard is consistent with observations that women are much more likely to get attacked when they have a realistic shot at winning, whereas long-shot female candidates in some ways get treated more fairly. Kirsten Gillibrand, another long-shot candidate, is also covered more in terms of her policy positions than character.
Cory Booker, Race, and Culture Wars
The sole male candidate with a significant amount of character coverage is Cory Booker, who is mainly ridiculed for his vegan lifestyle:
Booker is also the only black male candidate, although the tweets aren’t explicitly racist.
The three top declared candidates who are not white males (Harris, Warren, and Booker) all have significant amounts of coverage focused on “culture wars” issues (see examples above), mostly from the right-of-center tweeters. This is yet another indication of the race and gender-based divisions that we’re sadly likely to see much more of over the course of the campaign.
Does anyone care about policy?
More on policy discussions in future posts. For now, we’re seeing little substantive discussion for top-tier candidates; what shows up is largely critical of the candidates; and there’s even less discussion of actual substantive differences between the candidates.
We’re working on automating the clustering and theme discovery, and will be keeping careful track of how the conversations develop. We also hope that quantifying the differences in coverage between male and female candidates will call attention to the discrepancies and make the media more conscious of their storylines.
- e.g. Hayes, Danny and Jennifer L. Lawless. 2016. Women on the Run: Gender, Media, and Political Campaigns in a Polarized Era. New York: Cambridge University Press. 2016.
- Kittilson, M., & Fridkin, K. (2008). Gender, Candidate Portrayals and Election Campaigns: A Comparative Perspective. Politics & Gender, 4(3), 371-392. doi:10.1017/S1743923X08000330
- Meeks, L. (2012), Is She “Man Enough”? Women Candidates, Executive Political Offices, and News Coverage. Journal of Communication, 62: 175-193