Analyzing the Multidirectional Relationships Between the President, News Media, and the Public: Who Affects Whom?

Abstract

In order to study presidential leadership and responsiveness, this research focuses on the role of the news media and examines the multidirectional relationships between the president, the news media, and the public. One of the purposes of this study is to examine competing theoretical expectations about the causal direction between the three actors by focusing on their issue stances. The potentially reciprocal influences between the three actors are estimated by using vector autoregression and moving average representation simulations. According to the statistical results, the news media significantly interact with the public and the president. In contrast, the direct relationship between the president and the public is weak or insignificant.

Keywords:

• news media
• the president
• public opinion

Notes

1. 1. For instance, the public’s issue stance is measured by utilizing Stimson’s Policy Mood data (1999), a global measure of public policy sentiments.

2. 2. Buchanan (2001) speculates the reciprocal relationship between the three actors but does not offer empirical evidence.

3. 3. According to Graber (2010, p. 231), “even live television and radio broadcasts by the president are not totally devoid of media influence, because camera angles and other photographic techniques that journalists use slant all presentations somewhat.”

4. 4. This reasoning is quite comparable with the argument that popular presidents are less likely to respond to public opinion in order to earn public support (e.g., Canes-Wrone & Shotts, 2004).

5. 5. Compared to the past, today presidents may have better chances to send their messages intact to the public due to the rise of new media. Certainly, it is an interesting argument that presidential influences on the public and the media vary across different times. However, this topic is beyond the scope of the present research. This study first tries to investigate the general relationships between the three actors before delving into the time-varying effects, which will be the next step of the analysis.

6. 6. When elites are more likely to influence news coverage can be an interesting and important question. However, this question is beyond the scope of this article. This study focuses on analyzing the multidirectional relationships between the three actors. The question will be addressed in another study.

7. 7. For instance, the Gallup organization asks respondents “Do you consider the amount of federal income tax which you have to pay as too high, about right, or too low?” Each respondent’s answer can be aggregated as a percentage, and the aggregated respondents’ attitudes form mass public opinion.

8. 8. The nine domestic issues are race, welfare, the environment, crime, education, urban problems, health care, military spending, and size of government.

9. 9. Surveys and more specific rules for classifying liberal-conservative preferences are presented in detail in his study (Stimson, 1999).

10. 10. The Policy Mood series runs from 1958 through 2008, and Stimson offers the data on his Web site (http://www.unc.edu/~jstimson/). The quarterly series is the shortest time interval in the Policy Mood series.

11. 11. According to Wood and Lee (2009), a single coder coded each issue domain. Hence, inter-coder reliability measures are not applicable. Rather, cross-issue coding reliability can be a concern. According to them, the correlation coefficients between the nine issue series are generally high. Furthermore, Wood and Lee (2009, p. 1582) state that while coding error “may be a concern at the sentence level, it becomes less of a concern after the data are aggregated into time series… . Aggregating the coded sentences into time series meant that any lack of precision at the sentence level was swamped by the total count of sentences in each time interval.”

12. 12. The keywords and coding rules are presented in detail in Wood and Lee (2009) and Wood (2009).

13. 13. The Presidential Liberalism series is on http://people.tamu.edu/~b-wood/index_files/replication.

14. 14. Lee (2013) modifies keywords that Wood and Lee (2009) use to construct Presidential Liberalism.

15. 15. For instance, if ten liberal stories and one conservative story regarding a certain issue are observed in a month, the score for the issue in this month is nine (10 – 1).

16. 16. Lee (2013) samples coded news stories and asks another coder to code the sampled stories. According to his report, the percentage of agreement between the two coders is about 86%, and the Krippendorff’s alpha statistic is .72, which is used to check intercoder reliability (Krippendorff, 2004). Because he aggregates coding results, errors at the coding level should not be a major concern. Possible errors at the coding level will become a part of time series noise.

17. 17. The four time series comprising the Coincident Index are payroll employment, personal income, industrial production, and manufacturing and trade sales (the last three are in 1996 dollars) (Conference Board, 2001). Replacing the index with GDP or GNI does not change the main results in this study.

18. 18. In order to consider Granger causality, both the ICS and CICI are lagged one quarter in the VAR system.

19. 19. As Wlezien (1995, p. 987) mentions, “with the exception of defense,” measuring social spending requires some judgment. Generally, previous studies (e.g., Stimson, 1991; Wlezien, 1995) consider “welfare,” “education,” “big cities,” “environment,” and “health” as social or welfare state issues. Generally, this study follows the criteria Wlezien (1995) uses. More specifically, all education spending items under the education categories are considered as education spending. As welfare spending, the items in the “children and family,” “welfare,” and “welfare others” categories are included. “Housing” items are considered as social spending. As health spending, the items under the “health service (senior),” “public health service,” and “health” categories are included. The “pollution abatement” category is also included as social spending.

20. 20. Certainly, there are other ways of measuring government spending, such as using current dollars.

21. 21. All three series are stationary according to various unit-root test results (Dickey & Fuller, 1979, 1981; Kwiatkowski, Phillips, Schmidt, & Shin, 1992; Sims, 1988).

22. 22. “Granger causality” is not equal to “causality” (Granger, 1969). Granger causality shows whether past movements affect current realization.

23. 23. Based on the results of likelihood ratio test and information criteria, such as Akaike’s Information Criterion (Akaike, 1973), four lags of each variable are included in the VAR system. The three series show a similar lag structure.

24. 24. Table 1 presents p values from F tests for the null hypothesis of no Granger causality. Note that the Granger causality test results do not show the direction and magnitude of the relationship between the endogenous variables.

25. 25. The alternative orders of the endogenous variables are “Policy Mood–Macro Media Bias–Presidential Liberalism” and “Macro Media Bias–Presidential Liberalism–Policy Mood.”

26. 26. According to Sims and Zha (1999), the size of the error bands is comparable with the .05 significance level.

27. 27. Note that this study controls the reciprocity between the news media and the public.

28. 28. Lee (2013) shows that the president and the public do not significantly affect the news media. However, to consider Granger causality between the dependent variable and the independent variables, he includes Presidential Liberalism and Policy Mood as lagged variables. In contrast, this study controls the contemporaneous reciprocal relationship between the three actors. The simulation results, in fact, illustrate that the effects of the president and the public on the news media are contemporaneous and temporal.

Additional information

Notes on contributors

Han Soo Lee

Han Soo Lee is a research professor of Kyung Hee Institute for Human Society, Kyung Hee University, Seoul, Republic of Korea.