Jeffrey Cohen’s The President on Capitol Hill: A Theory of Institutional Influence is a book about presidential agency. That is to say, do presidents have the ability to influence their political environment and the behavior of congressmen? Much of the prior literature, he argues, has focused on the power of individual presidents, concluding that the office itself holds limited sway. Cohen concludes otherwise.

Cohen argues for an institutional-strategic understanding of presidential influence that runs counter to the personalistic view of presidential influence and the “Two Presidencies” thesis. “This theory leads us away from the dumb luck or deterministic explanations of presidential-congressional relations” in which they are “mere hostages or captives of the larger environment” and “away from personal-based accounts to more institutionally based understanding of presidential influence” (p. 245). Cohen conceives of presidents strategically using the trappings of the office to set an agenda that manipulates Congress. Influence is not, he proposes, how often a president succeeds within Congress but rather the ability to change an individual legislator’s votes, which leads to more strategic victories.

The president’s attempted influence for Cohen is measured in his public statements in support of an issue position. By taking a position, presidents “convert a floor vote from a nonpresidential into a presidential roll call. This alters the voting calculus of members, as greater weight is attached to the presidential consideration” (p. 143). The president’s position becomes, to the public, the party’s position. How responsive members of the House will be to the president’s position is context specific: this is what Cohen considers direct influence. The president’s position on an issue would matter more to House members when the president’s party is in the majority, when they are in the same party, when the president is popular, when the issue is salient, and when the issue is foreign-policy related: all these are examples, conversely, of indirect influence.

Cohen uses this big-picture view of measuring attempted influence—taking a public position—to try to incorporate the many types of public lobbying. Lobbying is “any action that the president takes to try to change a member's expected votes” (p. 17). The total presidential influence is not the focus; rather, it is on the influence of public presidential influence-seeking. These actions include many that the president could do on behalf of a bill, from inviting legislators to White House events, to visiting districts, to rallying the public for that position.

To study the president’s influence, Cohen examines all House roll-call votes from 1953 (the start of Congressional Quarterly) to 2012, nearly 27,000 votes. An additional 9,000 votes from 1877 to 1952 are also included in some analyses to assess whether modern presidents (1952–2012) are more influential than their premodern (1877–1932) and early-modern (1932–52) predecessors. Salient issues are identified using Gallup’s Most Important Problem Series, and presidential approval is based on that month’s rating in Gallup. To measure levels of lobbying as a mediator for influence, he considers the amount of time the president devotes to an issue in the State of the Union address. Cohen assumes this is an indication of how much time and resources will be devoted to a policy goal.

Cohen finds a significant, if small, level of presidential influence. Although only a few legislators’ votes are changed, he predicts it leads to five additional wins each year that would not otherwise have occurred. Averaging over their years in office, Nixon, Bush Sr., and Eisenhower were the most influential presidents. Ford, Clinton, Obama, Kennedy, and Bush Jr. “lack any measurable influence” (p. 67).

When engaging in low to medium levels of lobbying, the president is most effective. High levels of lobbying have tapered influence. This could reflect the fact that these were harder cases, which is why more time and resource-intensive support were deemed necessary. It could also reflect out-partisans pulling away from the president’s position; the more important the win seems to be for his party, the more that lobbying is a form of negative influence.

Presidents are more influential when their party controls the House and are more influential with co-partisans when their party is in the minority. The effect of being in the majority is stronger when polarization increases, but if their party is in the majority, polarization undermines their influence. Countering the “Two Presidencies” thesis, Cohen does not find that the president is more influential for foreign policy bills. He does find that the president has become less influential since the end of the Cold War, though he attributes this effect to increasing polarization.

A notable strength of this book is its interest in the counterfactual case. Cohen recognizes that to understand the president’s influence, scholars must consider both cases in which this treatment was applied and cases in which it was not. He also recognizes that we cannot know how House members would have acted had they been in the other condition: with or without presidential position taking. As such, his inclusion of nonpresidential roll calls, which too often are omitted in the literature, is an important methodological feature of this work.

Cohen’s work also benefits from his recognition that presidents, as rational and strategic actors, do not weigh in at random. The removal of lopsided votes and the use of propensity score weighting techniques help him account for the fact that presidents consider Congress and the public when deciding to act. Cohen balances cases on treatment propensity based on whether there is a Democratic majority in the House, whether it is a foreign policy vote, the presidential approval rating for that month, the level of party polarization in the House, whether the president’s party is in the majority, and whether the president is a Democrat. This system allows him to compare treated and untreated cases that are more similar, creating a more realistic approximation of the counterfactual case.

This is not to say the book is without weaknesses, which Cohen acknowledges in his writing. The study is limited to overt presidential positions; concealed agendas and secret lobbying could have additional influence not incorporated here. House members could also modify their behavior with respect to the president whether or not the president states a position by individual inference in an attempt to curry favor. Presidents may also have reasons for weighing in other than a desire to influence votes, such as an election promises, credit claiming, or wanting an easy win. The models and theory also cannot predict why some presidents were more influential than others nor provide a rationale for delimiting the modern, early modern, and premodern presidencies as the literature does.

Two weaknesses Cohen does not consider are mathematical. First, rather than presenting a conservative to liberal index of presidential positions, these positions would have been better rendered as factor variables, with a reference category of no position taken. This would have more readily aligned with his discussion of presidential behavior and of the results.

Additionally, some of the hypotheses Cohen identifies in the text are stated in the form of finding no effect. A regression model can have a not significant result, but this does not demonstrate necessarily that there is truly no relationship or indicate how confident one can be in the absence of an effect. In fact, in some such cases, the regressions find a significant effect counter to Cohen’s predictions, such as the relationship between majority/minority status and foreign policy influence.

Overall, Cohen presents an interesting new view of presidents’ agency and their capacity to change congressional behavior. His findings counter the expectations of the “Two Presidencies” thesis and of the personalistic presidency. The President on Capitol Hill will be of interest to scholars of American politics, especially those focused on the presidency and congressional voting behavior. It also presents an invitation for presidential comparativists to study this relationship in other contexts. A companion piece examining presidential influence in the Senate is a future possibility.