Chatty Cathys and Negative Nancys: Are women more aggressive users of Twitter in House of Representative Campaigns?

Note: This was done in part for a class at Sam Houston State University in the political science department. Full results of the project can be found here:


Women have had a struggle from day one in the United States. They weren’t allowed to vote until the 1920’s. They hold a growing yet small number of seats in Congress. To change this, female candidates have grown in number (Manning and Shogan, 2012) and are taking to new media platforms, like Twitter, to help, in part, spread their message. Because women start out from so far behind, do they act differently on Twitter than their male counterparts? More specifically, are they more aggressive in campaigns in an attempt to catch up to become part of the political elite? This article will investigate the relationship between gender and the style of use of the micro-blogging platform Twitter. Specifically, I expect that women are more aggressive on Twitter than their male peers in these campaigns. To define aggressiveness for the purpose of this article; I will look at the number of tweets in which candidates attack and the number of tweets they publish.




Campaigning, in general, has never been immediate when it comes to technology. Radio didn’t get off the ground in the political realm until the 1930’s after Herbert Hover showed its political potential with a landslide victory in the 1928 election. (D. E. Hamilton (Ed.)) However, the medium was invented commercially a full 10 years earlier when KDKA opened its doors. WEAF radio in New York is said to have broadcast the first paid commercial for an apartment complex. Politically, the radio was still in its infancy though, only broadcasting major speeches and conventions (Spalding, 1963). Eventually, radio became a staple in political campaigns on every level of government. (Willis, 1969) Women were barely allowed to vote and were in no means creating political advertisements in mass quantity.

            Television got off to an equally slow start. Like radio, television was originally considered a novelty early on. Radio was the king of broadcast and the preferred broadcast medium for candidates and listeners (Baran, 2009). It wasn’t until the 1952 campaign between Republican Gen. Dwight Eisenhower and Democrat Adlai Stevenson that television got its big break in political advertising. Eisenhower ran 40 30-second ads of him answering questions about domestic and foreign issues (Smulyan, 1993). Since then television has become a staple of political campaigns and has featured some of the most iconic ads including: the Daisy Girl ad, the Revolving Door, and the Bear in the Woods (West, 2010).

            With the advent of the Internet campaigners took an extraordinarily long time to pick up on its social use. While online communication has increased then, advertising and campaigning were still used in an insignificant number. The 2004 presidential primary campaign changed that though. Democrat Howard Dean raised over $15 million by June 2003. For the most part this was done through small online donations averaging $25 a person, far below the $2,000 limit (Wolf, 2004). Barack Obama and Mitt Romney took full advantage of the Internet during the entire 2012 presidential campaign. Obama and Romney both set up extensive campaign websites to collect donations and used social media websites like Twitter and Facebook to spread their message. But presidential candidates aren’t the only ones who use modern technologies.

Haber says during the 2010 mid-term election, 74 major Senate candidates tweeted 12,008 times, or more than 200 per day (Haber 2011). For the purposes of this article, it is important to understand the way in which the candidates in the Senate elections of 2010 actually tweeted to get in-depth context of the way Twitter can be used. In sum, 26.74% of tweets Haber studied fell into the Campaigning category, which means candidates tweeted where they will be campaigning next. Next was the number of Free Media tweets at 18.65%, which were tweets that shared information someone else wrote about the candidate. The third highest number of tweets in his study were Mobilization tweets at 15.54%, which asked followers to take some type of action. The rest of the categories did not receive above 15% of data. This data, he says, is interesting because it shows candidates use Twitter to mostly advertise campaign stops and campaign life. In addition, the number of Free Media tweets shows candidates use other stories to elevate their own name recognition and stature.

The data Haber collected also showed important trends in the types of tweets winners tweeted. Winners attacked far less than losers, 11% to 13%, and User Interaction tweets 7& to 9%; they tweeted more than losers on Issue tweets 9% to 7%, Mobilization tweets17% to 14.5%. All the percentages were statistically significant. Interestingly, the study also showed the total number of tweets from winners was far less than losers, an average from 142 to 190. This data shows that winners tweeted less than losers but did so efficiently. Haber also compared incumbents and non-incumbents and found that, in short, incumbents tweeted in a manner that made it seem like they weren’t in a competitive race despite the realities. This was to give less credibility to their opponents in the race. Non-incumbents did less attacking and issue tweeting, but more campaign, user interaction, free media, and mobilization tweeting. Non-incumbents typically don’t have much name recognition compared to their elected counterparts that already have a lot of gravitas. This means in campaigns they have to work to establish their name, making them dedicate many resources to tell a lot about themselves. Incumbents have already established their name and can put more resources into other areas of campaigning. This makes it harder for non-incumbents to unseat their opponents (West 2010).

            Haber’s study does have its problems. The way he codes the different tweets left little room for detailed examination as he only used eight types of tweets, one which was a Miscellaneous. His coding of Attack tweets for example left out attacks of other members of Congress, attacks on parties, and attacks of the president. Because of this, some candidates may appear to be less aggressive than they really are. Some candidates would rather criticize the party or president’s actions rather than their opponents. The Campaign category leaves out all material put out by the campaigns. This leaves out the possibility that campaigns put out a lot of their own material like press releases and video advertisements.

            In addition, it completely ignores gender. Until the 1980’s men were the primary office holders and thusly the largest users of political advertisements. The United States saw an explosion of women running and being elected to Congress. Manning and Shogan also show the 2012 election put 98 women in the 113th Congress, 78 in the  House of Representatives and 20 in the Senate, making it the largest number of women in Congress at any one time since the first woman was elected in 1912 to the 64th Congress. Although the number of women is growing, the way they have to campaign is important and drastically different than men due to the different stigmas men and women are under.

            Extensive research has been done into the way that women advertise in political campaigns. Research done on Senatorial and Gubernatorial races show the way men and women advertise are different. In the 1998 campaign females tended to be more negative than men (Robertson and Anderson 2004). In addition they said that female candidates were more likely to call for change and speak more about issues. More importantly, research into gender comes from the warnings female candidates get that negative advertising would violate gender stereotypes and harm the candidate (Godon, 2003). However, research hasn’t proved this to be true.

            The problem with studies discussing gender roles in political advertising is that it only speaks to the traditional broadcast mediums, television and radio. Research has almost completely ignored the way women advertise on the Internet, more specifically campaigning on social media sites like Twitter.




Therefore, I expect that women would be more aggressive users of Twitter in House of Representative campaigns.


As research shows, women are typically behind the eight ball when it comes to campaigning; they historically have been non-incumbents, they have gender stigmas and stereotypes to overcome, and they have a new, free medium in which to convey their message. Because research says non-incumbents are at a disadvantage, they have to campaign with more vigor than their counterparts. Attack tweets, while they give credibility to their opponents, also can make candidates more credible and give more name recognition. In addition, prior research shows women do tend to campaign negatively on broadcast media despite gender stereotypes. Finally, a free medium allows women to campaign freely, a good resource for a non-incumbent, which will allow them, in turn, to publish more messages that give them name recognition like attack tweets.




To accurately measure the way in which candidates use Twitter, a class of undergraduates and two teaching assistants followed all 435 House of Representative races. The coding started 60 days prior to the 2012 general election (Sept. 6 to Nov. 5). For the purposes of this paper, only 422 candidates will be assessed due to time restrictions. Those who were running and did not have a Twitter account were not included in the study. The correlations also excluded those who had Twitter accounts but did not tweet. Each tweet was split into separate categories: Attack, Attack Other, Campaign, Mobilization, Personal, Media, Obama, Romney, Immigration, and User Interaction. Obama, Romney, Immigration, and User Interaction tweets were categories that could be “double dipped”. A tweet could be an Attack Other and Obama, but not Campaign and Issue.

A tweet was categorized as “Attack” when a candidate directly criticized their opponent. “Attack Other” tweets were when the user attacked the opposing party or presidential candidate. “Media” tweets include mentions of media stories or appearances candidates will make. “Campaign” tweets highlighted where the candidate’s campaign has been or campaign material. “Mobilization” tweets asked for followers to take some sort of action related to the campaign, i.e. donate, show up here, etc. “Issue” tweets were those that spoke about any issue a candidate may campaign on, i.e. abortion, immigration, death penalty, etc. “Personal” tweets were unrelated to the campaign. “Obama, and “Romney” were tweets that mentioned the candidates. “ User Interaction” are tweets that are specific to Twitter that mention someone using the @ symbol followed by their Twitter handle used when the candidate directly speaks to someone. “Immigration” were coded when there was any mention of any type of immigration or acts related to it like the DREAM act.





 The candidates total number of tweets in this time period was 26,037 times at an average of 434 tweets per day. Each candidate tweeted an average of 85.8 times during the campaign cycle. In the study, those that had Twitters were 47% Democrats, 46% Republican, and 7% Independent or Third Party candidates (Figure 1). 97 candidates in this dataset did not have a Twitter account. 76.9% were men and 23.76% were female. 47.52% of users were incumbents. Attacks and Attack Other were each used on average 6.59 and 3.63 times respectively per person. Media tweets were used on average 7.27 times per person. Campaign tweets on average 14.59 times per person. Issue tweets on average 12.2 times per person. Mobilization tweets on average 7.18 times per person. Personal tweets on average 25.75 per person. Immigration tweets on average were .83 per person. User Interaction tweets on average were 12.79 per person. There were 3.14 and 1.67 references to Obama and Romney respectively.

Correlation between gender and…


Total Tweets



Attack Other
























User Interaction












In this chart, and for the purposes of this article, women = 0 and men = 1.

In the correlations that were run, there were actually very little correlation between gender and any of the variables coded for. Total number of followers had a result of -.09, which is a relatively weak correlation. Because these two variables are mutually exclusive, meaning you can’t be both a man and a woman, a negative relationship would mean that in whatever category women do more of it. In total number of followers, this means on average a woman is more likely to have more followers than her male counterpart. This, however, may not have any social scientific explanation. There is no evidence or reason as to why women would automatically accrue more followers unless she is either (a) more popular than her opponent, or (b) better at the use of Twitter in getting more followers.

The total number of tweets was of no significance at a correlation of -.0018.

In the number of Issue tweets there was a moderately strong -.11 correlation, meaning women were tweeting more about issues. This may not be surprising because, as seen in the prior research and theory, candidates have to give themselves credibility. Issue tweets are one way to do this. This is surprising, however, because the women in the study only 28 were incumbents. Haber’s study showed that there was a strong correlation between non-incumbents not tweeting about issues.

The Attack and Attack Other categories were relatively surprising. Attack showed a correlation of -.0128, which shows no correlation. Attack Other had somewhat of a correlation, -.08, showing women “attack other” more often than men. However, in the correlation between incumbents and attack, it showed non-incumbents attacked far more than incumbents. This does not translate to gender.

I sought to prove the hypothesis that women will be more aggressive users of Twitter than men. I expected women to tweet more and tweet more and use more Attack and Attack Other tweets. This data has shown that there is no statistical evidence to back up this hypothesis. It shows that women may attack the party and presidential candidates more, they aren’t attacking their opponents or tweeting more than their male counterparts.

            Correlations were also run between winners and losers, which had nine categories of tweets that had correlation. It was also run between competitive and noncompetitive races which had seven instances of statistically significant correlation. Correlation was also run between party identification and the categories which had eight statistically significant correlations. In addition, they were run between incumbents and challengers, which had 10 instances of correlation and five extremely strong correlations. Gender correlations only had six correlations, making it the lowest category with correlations. This is interesting because it could show that a style of tweeting is more prevalent because of campaigning skill than because of inherent differences because of gender identity.




This article sought to answer the question, “Do women in House of Representative campaigns tend to tweet more aggressively?” I expected the results to show that female candidates were more aggressive. The study of the candidates’ tweets showed that there was very little correlation between gender and attacking, or between gender and total number of tweets. This shows that we can’t expect the same style of political campaigning on the Internet as we do on broadcast media. Women and men in these campaigns tend to be about the same when it comes to attacking and number of tweets. This shows that more correlation is present between incumbents and non-incumbents, and winners and losers. These results show that there is a possible “style” in which candidates should be tweeting in and talking about, which is used equally by both men and women candidates. More research should be done in the area of winners and losers to determine, “what is the best strategy to use Twitter in political campaigns?” and focus less on gender demographics.




  1. Manning, J., Shogan, C. Congressional Research Service, (2012). Women in the united states congress: 1917-2012 (RL30261). Washington, D.C.: Government Accountability Office.
  2. Campaigns and elections. In D. E. Hamilton (Ed.), American President: A Reference Resource. Miller Center, University of Virginia. Retrieved from
  3. Spalding, J. (1963). 1928: Radio becomes a mass advertising medium. Journal of broadcasting, 8(1), 31-44.
  4. Willis, E. (1969). Radio and presidential campaigning. Communication Studies, 20(3), 187-193.
  5. Baran, S. (2009). Introduction to mass communication: Media literacy and culture. (6 ed., p. Ch. 9). McGraw-Hill.
  6. Smulyan, S. (1993). Radio advertising to women in twenties america: “a latchkey to every home”. Film, Radio and Television, 13(3), 299-314.
  7. West, D. (2010). Air wars: Television advertising in election campaigns. (5 ed., pp. 2-5). Washington, D.C. : CQ Press.
  8. Wolf, G. (2004, January). How the internet invented howard dean. Wired, (12.01), Retrieved from
  9. Haber, S. (2011). The 2010 us senate election in less than 140 characters: how senate candidates use twitter as a campaign tool, American University, Washington, D.C.
  10. Robertson, T. & Anderson, M. (2004). Gender and politics: Messages by female candidates in political ads in 1998 Senatorial and Gubernatorial elections. North Dakota Journal of Speech and Theatre.
  11. Gordon, A. (2003). Is negative advertising effective for female candidates?. The International Journal of Press/Politics, 8(3), 35-53.

2 thoughts on “Chatty Cathys and Negative Nancys: Are women more aggressive users of Twitter in House of Representative Campaigns?

Leave a Reply

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

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

Google+ photo

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

Twitter picture

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

Facebook photo

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

Connecting to %s