By Matt Glassman
Last week, two retiring Republican Senators ramped up their criticism of President Trump. On Tuesday, Senator Bob Corker said Trump had “great difficulty with the truth” and that he “debases our country.” Later that afternoon, Senator Jeff Flake lambasted the president, calling him “dangerous to our democracy.” These sharp attacks follow on the heels of similar rebukes from Senator John McCain and former President George W. Bush. Such co-partisan public criticisms have become increasingly common during the first ten months of the Trump administration.
Despite the nearly unprecedented nature of such attacks on a first-year president, many observers have dismissed such critiques as mere rhetoric. Instead, they argue, you should look at congressional roll call vote correlations, where you will find that most GOP members support the president well over 90% of the time. As DNC Chairman Tom Perez noted, “[F]lake voted with Donald Trump 91% of the time” and “Republicans in Congress remain in lockstep with the Trump agenda.”
It’s an open question how much impact public criticism of the president ultimately has on power dynamics in Washington and, in turn, public policy outcomes. Regardless of whether such criticism is a useful tool for senators to influence politics, this is certain: aggregate roll call vote correlations are a poor method of evaluating congressional party support, or individual member support, for a president’s legislative agenda. And consequently, the roll call votes become a poor metric for judging the president’s influence over legislation or his ability to achieve his legislative objectives.
Here I consider five aspects of roll call voting that cause problems for measuring party support and presidential legislative effectiveness: selection bias, ideological overlap, agenda setting, the binary nature of roll calls on complex policy issues, and the dynamic effect legislative votes have on presidential proposals.
First, the universe of roll call votes isn’t random; there is a selection bias as to which bills are brought to the floor. In the modern age, parties have tended toward working out their differences on legislation prior to taking it to the floor. Consequently, parties tend not to bring up bills that are going to fail; the failure of ACA repeal on the Senate floor this year was a very rare exception. In a narrowly divided and partisan Senate, the GOP leadership is mostly only going to allow votes on bills that virtually their entire caucus supports. If the congressional GOP strongly disagrees with the president on legislation, therefore, you’re usually not going to have roll call data on it, because it will not come up for a vote. For example, based on the roll call data, GOP Senators have almost uniformly supported Trump’s executive and judicial nominees. But the data do not include the eleven nominees who have been withdrawn, some of whom almost certainly failed due to lack of party support.
Second, the ideological overlap of the president and congressional co-partisan guarantees significant roll call vote support correlation. Regardless of whether or not they support the president, we wouldn’t expect GOP Senators to suddenly oppose longstanding party policies such as tax cuts simply because the president holds those positions and they want to voice opposition to him. But that highlights the awkwardness of the “support the president” frame for analyzing roll call votes. If both the president and the majority congressional party have the same ideological position on a roll call vote, it’s not clear whether legislators are supporting the president’s policy or if the president is endorsing the party’s policy. You can’t tell who is leading whom.
It’s silly to argue, for example, that ACA repeal is a presidential policy that congressional members are supporting. The congressional GOP was pursuing that policy long before the president entered electoral politics. To say that Flake or Corker “voted with the president” on ACA repeal is simply to say that they both have the same ideological position, which is neither surprising nor useful information on issues where Trump holds a long-held GOP position.
Third, roll call votes tell you nothing about the substance of the policy agenda. Since the votes taken on the floor of the House and Senate are a subset of the overall political agenda, a key decision for congressional leaders is deciding which agenda items will be brought up for a roll call vote, and which will not. One way to not support the president’s agenda is to not consider it. If policies the president would like to pursue are not brought up for votes, they will not show up in the roll call data. One difficulty in arguing that the GOP Senators always support the president’s agenda is that the president’s legislative agenda, to date, has largely been a failure. Little has passed, because little has been voted on.
Similarly, the roll call votes don’t reflect the size of the policy agenda. If you think of the overall Trump agenda as a denominator, the fraction of support the congressional GOP is giving the president drops dramatically. Notably, the 115th Congress has so far declined to take up the immigration restrictions, protectionist trade policies, or infrastructure development that Trump campaigned on. Likewise, his one concrete proposal this year—the FY2018 budget that included deep cuts across many federal non-defense agencies—has been almost completely ignored by congressional appropriators, as were his suggested cuts to the final FY2017 appropriations. On the other hand, both chambers did vote on a Russia sanctions bill that restricted the president’s authority.
Fourth, roll call votes do not tell you who crafted complex legislation or influenced underlying policy choices. Roll call votes are binary choices over legislation that often includes hundreds of provisions. Just as congressional leaders can choose which agenda items to take up, they can also negotiate which underlying provisions to include in the bill. Such decisions may be predominately driven by congressional actors or by administration actors. As with the choice of issues to take up, if underlying policy choices are mostly driven by the priorities of congressional actors, then the president is not leading and influencing outcomes, he is following and acquiescing to them. The president’s apparent disinterest in policy details magnifies this effect.
The congressional GOP and the president will likely both support a final tax reform bill later this Congress. But who will drive the underlying policy choices made in the bill? If the congressional GOP drive the formulation of the bill, that’s not evidence of support for the president, or even evidence of presidential influence in policy. It simply shows that president Trump will support whatever legislation makes it to the floor. Likewise, the final FY2018 appropriations will almost certainly be supported by the president and most GOP Senators. Based on markups and votes taken so far in the House and Senate, the final numbers will look a lot more like the preferences of the congressional party than the president’s FY2018 budget request.
Finally, and perhaps most importantly, the roll call vote correlations do not reflect the dynamic effect that roll call votes have on presidential proposals. This is most clear in regard to nominations. It is true that the GOP Senators have voted almost uniformly for Trump nominations that have reached the Senate floor. But Trump, as with any president, is selecting nominees knowing full well that they need to be confirmed by his party. Again, think about the denominator: who would Trump have nominated to the various executive branch posts if they didn’t require Senate confirmation? Who is working for the president in White House jobs that don’t require confirmation? Then compare that list to the actual list of nominations submitted to the Senate. The latter is a much more traditional set of GOP actors, precisely because the president knew the Senate GOP would never confirm Steve Bannon as Secretary of State or Michael Flynn as Secretary of Defense.
Roll call votes are an appealing metric because they consist of concrete empirical data that is easy to locate and analyze. Their limitations, however, make them largely unsuitable for measuring presidential agenda support. Based solely on roll call data, one might be forgiven for believing President Trump had almost universal consistent backing of his party and was smoothly and powerfully leading the congressional GOP in implementing his legislative agenda. In reality, the president has been largely ineffective in driving legislative outcomes. The congressional GOP has mostly ignored his agenda where it differs from theirs, and the administration has had significant difficulty influencing policy choices, much less getting its major priorities across the finish line in Congress.