State of the Union addresses and presidential position taking: Do presidents back their rhetoric in the legislative arena?

Abstract

Politicians and presidents are routinely criticized for the mismatch between their policy statements and their legislative actions. While a few studies explore presidents’ support for policy commitments made during their election campaigns, no study systematically examines this relationship for presidents throughout their terms. To determine whether presidents follow through on their policy statements, I examine presidential mentions of three policy areas in State of the Union addresses from 1953 to 2000 and presidents’ subsequent positions on floor votes in Congress. The results indicate that rhetorical attention to economic policy and foreign relations in these speeches increases the chances that presidents will take positions on legislation in these same areas, but there is a disconnect between the rhetoric and their actions on health and social welfare policy. I suggest that this difference can be attributed to their anticipation of success in each policy area because presidents do not want to attach themselves to legislative defeats.

Acknowledgements

A previous version of this research was presented at the 2001 annual meeting of the American Political Science Association. I thank Craig Volden for his guidance and insights. I also thank Jean Schroedel and Gary Segura for helpful comments and suggestions.

Notes

1 Schuh (2006) explores the congruence between values expressed in major speeches, including State of the Union addresses, and the values contained in executive orders. This study differs by looking at policy-related content and presidents’ strategic legislative behavior rather than unilateral actions.

2 Presidential success rates are based on Congressional Quarterly Almanac data compiled by the author.

3 One may question why all roll call votes were included in this measure because this method tends to incorporate amendments to bills which may drive up the number of roll call votes. However, this is also one reason why they are all included. To measure the president's legislative actions and address how important his priorities are, it is necessary to account for all potential roll calls in which he could be active. This includes amendments to bills as well. The amendments represent possible alterations to what the president wants, or could deter some members from voting for the bill. In either case, the president's position on an amendment still represents how important the overall bill is to him since the amendment may kill the bill or alter it considerably. Thus, the inclusion of all roll call votes in each policy area considers these possibilities.

4 The three policy categories contained the following subjects: (1) Foreign relations: arms control, foreign aid, Mutual Security pacts, Peace Corp, Persian Gulf War, state department, treaties, Vietnam War; (2) Economy: agriculture, budget, business regulation, disaster funding, energy policy, federal debt, maritime authorization, natural resources, minimum wage, public works, small business acts, transportation (excluding highway safety); (3) Health and social welfare: child care, child support, drug abuse, health research, highway safety, housing programs, prescription drugs, retirement, social security, unemployment programs, urban development, volunteerism/community service, worker safety. In instances where bills, or in the case of sentences for the Policy Mentions variable, fell into more than one policy category, the sentence or bill was not counted for either category.

5 Previous research has identified several shortcomings with the use of roll call data. The major flaws levied against these data are that they do not account for bills that fail to make it out of committee and do not address the role of presidents in the committee process. Indeed, these are serious defects, but, unfortunately, roll call votes are the most readily available data. As opposed to executive orders, however, it is also the most consequential venue for making substantial policy changes. For research on presidents’ influence in the committee process, see Schroedel (1994). For a more detailed discussion of the various shortcomings of roll call data, see Peterson (1990) and Jones (1994).

6 In transition years to a new presidency (i.e. 1961), presidents do not typically deliver what is called a State of the Union address. In these years of the study, a substitute national speech at the beginning of the year was employed in its place. They are the following: 1969 – Plan for Domestic Legislation (Nixon); 1977 – Economic Stimulus Proposal (Carter); 1981 – Economic Proposals Text (Reagan); 1989 – Budget Plan (Bush Sr.). This avoids the use of any speeches by lame-duck presidents. In 1993, President Clinton delivered what is sometimes referred to as an SOU address despite his incoming status.

7 I do not choose one specification of this variable over the other because a plausible argument can be made to use either one and the robustness of the results across the two different specifications appear to further reinforce the theoretical foundations. In the former case, each variable specification has its advantages. Using the number of sentences devoted to a policy area measures presidents’ rhetorical attention in a similar manner to the construction of the dependent variable since each policy area is only considered in and of itself. That is, with the dependent variable, the president's positions in a given policy area are considered in light of the number of roll calls in that same area, not in consideration of other policy areas. Using the percentage of sentences is advantageous because it accounts for presidents who may give particularly long or short speeches, which thus standardizes the variable. However, it measures their policy statements relative to other policy areas, while the dependent variable does not. With regard to further reinforcement of the theoretical foundations, the robustness of the relationship across the different policy areas varies according to the expectations of the hypothesis. For example, the results for foreign relations are robust across both specifications, while the link between health and social welfare is more tenuous and negative in one of the specifications.

8 Thanks are due to Nicholson et al. (2002) for providing these data through 1996. While Brace and Hinckley (1992) used three original sources for their events variable, Nicholson, Segura, and Woods used two, The World Almanac and Book of Facts and The Encyclopedia of American Facts and Dates for extending the data from 1989 to 1996. The data for 1997–2000 were extended by the author using the same method. Only The World Almanac and Book of Facts was available for the extension of data from 1997 to 2000. For a complete description of the coding methodology, see Brace and Hinckley (1992), Appendix A, pages 183–188.

9 rising unemployment could increase presidents’ attention to social welfare programs since this category includes worker training and education programs. However, since economic performance is usually tied to the tax structure, it is more likely that their attention will turn to tax incentives which fall under the economic policy category.

10 Several additional control variables were tested including presidential approval ratings and a dummy variable for election year, but their inclusion did not improve the explanatory power of the models nor did it substantively change the results.

11 An alternative hypothesis of the rhetoric-substance link in this area could posit an inverse relationship to that discussed here. That is, the negative relationship could be a result of presidents talking less about health and social welfare, but taking a large number of positions against roll calls to defeat Congressional activism in this area, particularly since there were more Republican administrations in the time period of this study. While this is a plausible alternative, it is unlikely both on theoretical and empirical grounds. First, on theoretical grounds, it does not consider that presidents are normally less successful in this area, which makes position taking more costly to their reputation and store of political capital. On empirical grounds, if presidents attempt to suppress Congressional activism in this arena, we should see a higher level of presidential opposition to roll call votes in this area compared to others. However, this is not the case. The percentage of positions where presidents oppose legislation in this area is actually lower (44%) compared to foreign relations (45%) and economic policy (47%). Thus, it appears the negative relationship is due to less overall activism on the part of the president.

12 Both of these cases are chronicled in the 2003 and 1994 Congressional Quarterly Almanac.