When a Man Meets a Woman: Comparing the Use of Negativity of Male Candidates in Single- and Mixed-Gender Televised Debates

Abstract

Much work has been done to analyze the consequences of the increasing representation of women in politics. Usually, this research compares male and female politicians from a female perspective. For instance, many studies in political communication investigate how and why female candidates show campaign styles similar to or different from their male colleagues. In contrast to this, few studies are interested in how men change their behavior when women enter the political arena. Some of these studies have demonstrated that men limit their negativity when confronted with female candidates. Unfortunately, these analyses focus predominantly on (a) the United States and (b) gender differences in campaign advertising. We seek to provide empirical evidence for non-U.S. campaigns from the most important single campaign events: televised debates. To do so, we analyze data covering all German televised debates broadcast since 1997 where male politicians participated in a two-candidate single- or mixed-gender debate. Our results indicate that the gender of the political opponent affects incivility but not the use of attacks in general. In mixed-gender debates there is less incivility. However, differences in the treatment of male and female opponents tend to decline over time. We conclude that female candidates transform campaign communication—not only because they communicate differently from men, but also because they influence male politicians’ political communication.

Keywords:

• attacks
• incivility
• gender
• televised debates
• content analysis

Notes

1. There is a lively and long ongoing debate on the definition of negative campaigning and its core element (i.e., attacks). In line with most scholars, we support the so-called directional approach, which excludes any further evaluations of an attack. For a discussion of the different perspectives, see, for example, Lipsitz and Geer (2017); Nai and Walter (2015).

2. Before the introduction of American-style televised debates, there was a tradition of so-called elephant rounds. They were aired in the run-up to every national election between 1972 and 1987 and included the leaders of all parties represented in the national parliament (see, e.g., Schrott & Lanoue, 1992). This format was (and still is) used at the state level. Because elephant rounds do not fit our definition of a televised debate—that is, live discussion between the (typically two) top candidates who have the best chances of being elected as the head of the government (national level: chancellor; state level: prime minister, respectively the governing mayor in Berlin, Bremen, or Hamburg)—we do not include them in our sample.

3. Most transcripts can be downloaded via https://www.uni-koblenz-landau.de/de/komepol/publikationen/working_paper_series

working_paper_series. For transcripts, also see Maier and Faas (2003a, 2003b), Maier, Maier, Reinemann, and Maurer (2006) as well as www.gles.eu.

4. The identification of coding units was done by two coders, while the coding of the strategy was carried out by five coders. In general, the codebook of the 2009 German televised debate was used as a reference guide to code the debates. The codebook and the data of the 2009 debate can be downloaded via the homepage of the German Longitudinal Election Study (GLES; see www.gles.eu).

5. This is in line with the definition of “idea units” as a “phrase, clause, or sentence that carries a claim” (Jamieson, Waldman, & Sheer, 2000, p. 51). Hatfield and Weider-Hatfield (1978, p. 46) define a thought unit as “the minimum meaningful utterance having a beginning and end, typically operationalized as a simple sentence.”

6. All non-candidate statements (i.e., statements from journalists, questions from the audience, video segments) are excluded from this analysis (N = 6,756). Nonfunctional candidate statements (N = 618) are units that were so incomplete that they could not be understood. For example, statements were coded as non-functional when a candidate was interrupted and consequently what she or he said so far made no sense or when the candidate mumbled and it was impossible to understand hear his or her statement.

7. Survey data were retrieved from www.wahlrecht.de. This variable considers the difference in the share of votes between a candidate’s party and the party of his or her opponent.

8. To assess intercoder reliability, about 25% of all functional candidate messages were selected and coded independently by two coders. On average, intercoder reliability for strategy was .901 (Holsti’s formula).

9. The average intercoder reliability was .823 (Holsti’s formula).

10. The share of negativity was calculated as follows: × 100.

11. The share of uncivil attacks was calculated as follows: × 100.

12. Most content analyses of televised debates distinguish between incumbents and challengers. This does not make much sense in parliamentary democracies where coalitions often have to be formed after an election. As a consequence, candidates joining a coalition may debate with each other at the end of a legislative period. This was the case in Germany, for instance, when Chancellor Angela Merkel and her Minister of Foreign Affairs, Frank-Walter Steinmeier, participated in a debate in the run-up to the 2009 national election. Steinmeier could not criticize Merkel very much because he was also responsible for the previous government’s performance. As a consequence, the proper distinction for parliamentary political systems is between members of the government and members of the opposition instead of incumbents and challengers (for this approach, see also Maier & Jansen, 2017).

13. There are, of course, more variables that can potentially influence the share of (uncivil) attacks (e.g., age, political experience, party affiliation, or measures of debate format). Due to the limited number of cases, we only consider variables for which studies have found a significant impact on use of attacks (see Maier & Jansen, 2017).

14. Most recent research on negative campaigning analyzes those interactions by rearranging the data into a dyadic structure (see, e.g., Dolezal, Ennser-Jedenastik, & Müller, 2015). The question is then whether, for example, a negative press release can be understood as a direct retaliation (i.e., a counterattack) of a former attack. Although this is a very interesting and promising approach, we are interested in the factors influencing a candidates’ debate strategy (and not so much in explaining why a particular candidate statement during the debate turned out as an attack or not). As we know that candidates prepare themselves very well before a debate (see, e.g., Schroeder, 2016), we can assume that at least the decision to attack the political opponent is not so much the result of specific interactions but rather has been made beforehand. However, in order to measure the effects of the behavior of the political opponent, we control for the opponent’s level of (uncivil) attacks.

Additional information

Notes on contributors

Jürgen Maier

Jürgen Maier is professor of Political Communication at the Department of Political Science, University of Koblenz-Landau, Germany

Anna-Maria Renner

Anna-Maria Renner is PhD candidate at the Graduate School for Gender Research at the University of Koblenz-Landau, Germany.