Revisiting “Who Influences Whom?” Agenda Setting on Biofuels

Abstract

Although numerous advancements have been made in the study of agenda setting among the media, the president, and Congress, scholars have struggled to develop a cohesive theory about who influences whom. To address this problem and augment the agenda-setting literature, I integrate exogenous agenda-setting variables from the American politics and policy literatures into traditional agenda-setting models. In contrast to prior research, which has tended to favor broad policy domains and foreign policy issues, I focus on one specific domestic policy area–biofuels. The results indicate that exogenous agenda-setting variables should be included in models of dynamic agenda setting, because they play an important role in the relationship among the media, the president, and Congress. In the case of biofuels, economic indicators, public opinion, and election year politics affected the agendas' of the media, the president, and Congress, and they drowned out the influence that these actors had on one another.

Notes

1. To clarify, X is said to be a Granger cause of Y; when controlling for lagged values of Y, lagged values of X are significant predictors of Y. Moreover, if X is found to be a Granger cause of Y that does not mean it is the cause of Y, because an unmodeled Z could be driving both X and Y.

2. Kingdon (1995) categorizes these as variables in the “political stream.”

3. For example, see “Sebelius: Mammogram Recommendations Won't Set Policy” http://www.cbsnews.com/8301-503544_162-5698251-503544.html

4. The subsidy was first introduced through the US Energy Tax Act of 1978 and it was most recently augmented in the 2008 Farm Bill (Yacobucci 2008a,b).

5. Consumption mandates are part of the renewable fuels standard (RFS), which was established in the 2005 Energy Policy Act (EPA). Although the RFS does not mandate production of ethanol, it requires that a minimum amount of fuel come from renewable sources, and thus far the RFS has primarily been satisfied by corn-based ethanol. The Energy Independence and Security Act (EISA) of 2007 also revised the RFS to increase the required volume of specific renewable fuels, including “cellulosic ethanol,” “advanced biofuels,” and “unspecified” sources, all of which could include ethanol (Yacobucci 2008a).

6. I also went through each media article, presidential speech, and bill to ensure that they were focused on the topic at hand, biofuels, and did not merely mention one of my search terms outside of this context.

7. This data is available online at http://people-press.org/dataarchive/. The Pew Research Center for the People and the Press bears no responsibility for the interpretations presented or conclusions reached based on analysis of the data.

8. Data on corn prices are from the United States Department of Agriculture (USDA) online at http://www.ers.usda.gov/Data/FeedGrains/CustomQuery/Default.aspx. For the “custom query” I specified corn prices at market, monthly for all years. Data on food prices are from the Bureau of Labor Statistics consumer price index (CPI), online at http://data.bls.gov/cgi-bin/dsrv.

9. Negative binomial models were most appropriate given the characteristics of my data. My data were inappropriate for Poisson regression as they violated the assumption of equidispersion. Ordinary least squares analysis was also inappropriate because my data do not conform to a normal distribution. Likewise, an OLS model produced highly disparate results from that of the negative binomial analysis, causing changes in levels of significance and the direction of the relationship in several cases. Finally, given the relatively small-N (120) in this study, I opted for a set negative binomial models rather than the Vector Autoregression (VAR) models used in prior studies, because they have the same basic structure, but are less demanding on the data.

10. I tested these models using lags of other lengths as well, but that did not add anything significant to the models.

11. I found there is not an appreciable difference between models with basic standard errors and those with HAC standard errors.

12. Preliminary results from a study of media framing of biofuels by the author suggest that the media has focused on the negative aspects of biofuels.

13. It is important to note that all variables in these graphs are measured at yearly, rather than monthly, intervals; hence they provide a less precise representation of the agenda-setting power of the exogenous variables than do the negative binomial models I discuss next.