The Partisan Content of Candidate Messages in US Senate Elections

‡ Wisconsin Advertising Project (WAP) data are used for this project and can be found at http://wiscadproject.wisc.edu/.

Abstract

This paper analyzes the partisan messages of US Senate campaign advertisements. Using data from 2000 to 2004, we measure the levels of partisan cues and “owned” issue mentions in US Senate campaigns. We hypothesize that campaigns are likely to use campaign communications involving partisan cues and owned issues when national and local conditions favor one of the parties. Our results recognize some variation in campaign strategies, providing strong evidence that candidates, political parties, and outside groups respond differently to local and national conditions when cueing party labels and party-owned issues.

Acknowledgements

We thank Joseph Giammo and Carly Schmitt for their helpful comments and suggestions along with Amy Lore for her research assistance.

Supplementary Material

Supplemental data for this article can be accessed at http://dx.doi.org/10.1080/17457289.2014.997240.

Notes

‡ Wisconsin Advertising Project (WAP) data are used for this project and can be found at http://wiscadproject.wisc.edu/.

1. Some challengers campaign so that their name is recognizable in future elections or that they will be better connected to the party establishment. The latter complicates our study because such candidates would be more likely to cue the party label not to win votes, but for alternative motives. Given the high-profile nature of US Senate races, such circumstances ought to be rare.

2. Ads from 2000 are taken from the top 75 markets, which represent over 80% of the US population.

3. Elections in Montana (2000, 2002), North Dakota (2000, 2004), Hawaii (2000), South Dakota (2002, 2004), Wyoming (2000, 2002), and Vermont (2000) are omitted because these states are not in a top 75–100 media market.

4. Seven ads produced (206 airings) are excluded because they are in support of third party candidates. These data only represent 0.0003% of all the sampled ads aired.

5. We expect this will remain true, in part, because the advent of Internet-based campaigning has not produced a dramatic effect on congressional candidate messaging or campaign strategies (Druckman, Kifer, and Parkin 2014).

6. See Appendix A for our classification scheme.

7. For our analyses in Table 2, this n is reduced to 791,977 because sponsorship of an ad in Minnesota (2002) and another Florida (2000) could not be determined. The Minnesota ad ran 153 times while the Florida ad ran 112 times.

8. See Appendix B for descriptive statistics on the partisan content of the sampled ads.

9. Paired t-test = 95.38, p < .01.

10. In a contingency table, χ² = 33,000, p < .01.

11. In a contingency table, χ² = 61,000, p < .01. See Appendix C for graphical reproduction of the average mention of party label by US Senate ad sponsor.

12. Independent group t-test = 225.59, p < .01.

13. In a contingency table considering party mentions by year, χ² = 1300, p. < .01 for 2000 data, χ² = 1100, p. < .01 for 2002 data, and χ² = 15,000, p. < .01 for 2004 data. See Appendix D for graphical reproduction of the average mention of party label by US Senate campaigns per election.

14. A separate content analysis of the data was conducted to provide some greater validity to the number of party mentions for the 2000 ads. Of the 620 ads analyzed, 35 mention a party label (5.6%) with 51 storyboards missing.

15. In 2002, there is a 1% difference between Republican and Democrat-supporting ads mentioning a party label. Independent group t-test = 3.99, p < .01.

16. As seen in Appendix A, ads mentioning party labels are more likely to contain Republican party-owned issue appeals.

17. See Appendix E for graphical reproduction of the average number of issue appeals mentioned in US Senate ads by sponsor.

18. See Appendix F for graphical reproduction of the average number of issue appeals per US Senate campaign year.

19. See Appendix G for graphical reproduction of the average number of appeals on salient issues per US Senate campaign year.

20. The economy was a top concern for voters during these years as well.

21. We use the gllamm's logit link option along with adaptive quadratures to estimate our model.

22. No original conclusions are clearly contradicted from these alternative tests.

23. Specifically, around 4% (n = 1603) of interest group ads aired included a party label. From an ad production view, only 14 sampled ads across the three election years included a party label.

24. Issue Salience is measured as a count of the number of salient topics mentioned in the ad. The issues considered salient are those seen in Appendix G and are based on Gallup poll questions. This variable excludes economy-based appeals, as the WAP does not code generic appeals about the economy, but only related issue mentions like taxes, minimum wage, and the deficit.

25. This perspective is challenged frequently (see, for example, Lewis-Beck et al., 2008).