INTERPERSONAL RELATIONSHIPS AND LAWMAKING

Most legislation that passes Congress does so with considerable support from members of both parties. These bills range from trivial measures to significant legislation such as the CARES Act or the Respect for Marriage Act. As Curry and Lee (Reference Curry and Lee2020) demonstrate, the proportion of bills that receive bipartisan support at passage has remained relatively constant over the past few decades, even as our measures of “polarization” between the two major parties has shown an unabated increase. Similarly, Harbridge (Reference Harbridge2015) finds a constant rate of cross-party cosponsorship of legislation across the past three decades, Craig (Reference Craig2023) finds that bipartisan collaboration persists, and Fong (Reference Fong2023) finds that legislators prefer to be cooperative with one another. Legislation that cleanly divides the two parties definitely exists, but purely partisan lawmaking is the exception rather than the rule.

That bipartisan lawmaking persists in the current political environment is a puzzle. Today, both parties are engaged in a never ending battle for majority control of each chamber of Congress as neither party has been able to secure a large or durable majority. This battle has produced a rise in “messaging” bills intended solely to highlight parties’ differences (Gelman Reference Gelman2017; Lee Reference Lee2016). Today’s constant competition, partisan rancor, and focus on messaging has become self-reinforcing. Many ideological moderates have chosen either to retire from Congress or not seek the office at all (Thomsen Reference Thomsen2017), and long-standing norms against campaigning against state delegation members have eroded (Treul Reference Treul2017). Moreover, the centralized nature of modern lawmaking has limited the ability of most rank-and-file members to play an active role in legislating (Chergosky and Roberts Reference Chergosky and Roberts2018; Curry Reference Curry2015), furthering the influence of partisan leaders on Capitol Hill.

One reason for continued bipartisanship are the rules. The lawmaking process requires the building of concurrent majorities across the House and Senate, while also requiring the support of the White House. In periods of divided government, this necessitates bipartisan support to move legislation. But even under unified government, intraparty disagreements and differences in House and Senate apportionment typically produce large, bipartisan winning coalitions (Curry and Lee Reference Curry and Lee2020; Krehbiel Reference Krehbiel2010; Mayhew Reference Mayhew2005). The Senate’s debate rules further reinforce the need for bipartisanship, as support in both parties is typically needed to secure the 60 vote cloture threshold.

While the rules necessitate bipartisanship to legislate, members of Congress still must overcome partisan divides to forge legislative agreements. In prior work, we find that the interpersonal relationships among members of Congress are a key factor in forging bipartisan collaboration across these divides (Curry and Roberts Reference Curry and Roberts2023). We rely on a combination of interviews with senior staff and quantitative analyses of congressional travel and legislative collaboration to demonstrate that members who travel together tend to build relationships that help legislators overcome political differences and collaborate on bipartisan legislation (Alducin et al. Reference Alducin, Kelly, Parker and Theriault2014; Alducin, Parker, and Theriault Reference Alducin, Parker and Theriault2017).

A good example of this occurred in the 117th Congress as Senators Thom Tillis (R–NC) and Chris Murphy (D–CT), who previously had not known one another, were able to forge a relationship on an overseas trip, or “CODEL,” that helped produce a bipartisan deal on a gun control. As Tillis remarked, “[on CODELS] we’re working 12 and 14 hour days, we’re sometimes traveling two or three hours from country to country. And that just gets you into a position where …you build that trust and you build that familiarity, [and] that serves as a basis for getting accomplished what we did.” (Desiderio Reference Desiderio2022).

A substantial body of work demonstrates the difficulties of measuring the presence of relationships among legislators, as well as how those relationships help foster legislative collaboration (Fong Reference Fong2020; Kirkland and Kroeger Reference Kirkland and Kroeger2018). The fracturing of a relationship is even more difficult to measure. Unlike celebrity couples, members of Congress typically do not issue press releases when they “break up” with a colleague. The events of January 6, 2021—though unfortunate in almost every way—provide an observable measure of an event that fractured relationships on Capitol Hill.

JANUARY 6 AND RELATIONSHIPS IN CONGRESS

As shocking and frightening as the storming of the U.S. Capitol was to congressional observers, it was unimaginably worse for those who work inside the chambers. Members and staffs saw their workspaces invaded and ransacked, and some members literally ran for safety as rioters fought their way into the nation’s Capitol. Many members spent hours either hiding in their offices or in a “secure area” while police and eventually the National Guard subdued the rioters. This occurred on a day when many members brought their families to the Capitol to observe the ceremonial counting of electoral votes. Few, if any, thought they would find their physical well-being at risk on this day.

As we noted above, the certification objectors faced outrage from many of their colleagues. It was reported that many Democrats were going to refuse to collaborate legislatively with objectors moving forward. Rep. Connor Lamb (D–PA), a moderate known for his willingness to work across the aisle conceded that, “There are personal relationships at stake…” and that he would find it hard to work with “morally blind” Republicans following the attacks (Calefati Reference Calefati2021). Other Democratic lawmakers were even more specific. For example, Rep. Brad Schneider (D–IL), a member of the bipartisan Problem Solvers Caucus, said he had to cut off relationships with Republicans who had supported de-certification, and reportedly ended legislative efforts he had ongoing with Reps. Jody Hice (R–GA) and Paul Gosar (R–AZ) (see Ann Caldwell Reference Ann Caldwell2021). Rep. Cindy Axne (D–IA), who had previously cosponsored legislation with Rep. Jason Smith (R–MO) expanding kidney disease insurance eligibility, found a new Republican cosponsor—Rep. Jaime Herrera Beutler (R-WA)—who had voted to certify the election results.

The reporting around the fallout of the January 6 riots suggests that the personal relationships that undergird much of what gets done in Congress were severely damaged by perceived Republican support for those who questioned the veracity of the election results and for those who stormed the Capitol. The question that motivates the analyses that follows is: did the damage to relationships on Capitol Hill following January 6 inhibit the patterns of cross-party collaboration and legislative success that we routinely see in Congress?

RESEARCH DESIGN AND ANALYSES

The expectations for our analyses are straight-forward. We expect that those Republicans who voted against certification would find fewer Democratic legislative collaborators beginning in the 117th Congress (2021–22) than had previously been true. Our research design is straightforward, enabling us to take advantage of the shock to relationships on Capitol Hill that occurred via the events of January 6. Obviously, “selection” as a certification objector was not random. Each member of Congress was free to choose whether or not they voted to certify the Electoral College votes from Arizona and Pennsylvania. And those who opposed certification were among the Republicans with the greatest prior support for Donald Trump, both personally and among voters in their districts. Thus, while we do not have random assignment, we do have a sudden shock that affected every member of Congress—the 139 House Republicans who voted to sustain at least of the objections to certification, the 63 House Republicans that supported certification, and every Democratic lawmaker—with many of the latter two groups having their relationships with the former suddenly damaged. We think this uniform shock to relationships on Capitol Hill allows a observation–treatment–observation design.

We conduct several analyses. First, we assess each House member’s bipartisan original cosponsorship rate. This measure, which is taken from Curry and Roberts (Reference Curry and Roberts2023), is the proportion of bills that each member introduces in each Congress that had an original cosponsor from the other party. Original cosponsors are the members who were signed onto a bill as it was introduced, and are typically few in number compared to the total number of cosponsors. Drawing on interview data, we found out that original cosponsorship signals actual hands-on legislative collaboration between members (Curry and Roberts Reference Curry and Roberts2023). While any member can cosponsor any bill, original cosponsorship is viewed akin to coauthorship on an academic paper. Original cosponsors, who are designated on the bill itself, are understood to be the members who collaborated on its development and introduction. If House Democrats punished certification objectors for their role in the events of January 6 by no longer working with them, we would expect to see the bipartisan, original cosponsorship rate for objectors to decline, and decline to a greater degree than for other members, after January 6. We also analyze the raw count of the number of bills for each member that had an original cosponsor of the opposing party, and we again compare before and after January 6, 2021.

Next, we identified the subset of bills introduced in identical form by each House member during the 116th (2019–20) and 117th Congresses (2021–22) to observe changes in bipartisan cosponsorship, while being able to hold the contents of the legislation constant. In other words, while the first two measures could possibly produce results that are confounded by members of Congress introducing systematically different bills before and after January 6, 2021, by looking just at bills that did not change, we can more cleanly observe the effects of objecting to certification on bipartisan, original cosponsorship.

Finally, we employ measures of each members’ legislative effectiveness. Specifically, we analyze three components of Volden and Wiseman (Reference Volden and Wiseman2014)’s Legislative Effectiveness Scores (LES): whether or not each of a member’s bills (1) received action in a committee, (2) received action beyond committee, and (3) passed the House. We use these components, rather than the composite LES measure, for two reasons. First, the composite measure of Legislative Effectiveness includes the number of bills each member introduced. We want to exclude introductions because they do not require cooperation from other members. Second, the composite scores are not directly comparable across congresses given that they are normalized within each Congress (Volden and Wiseman Reference Volden and Wiseman2014). By using the component parts of the LES measure separately, we can make accurate comparisons from Congress to Congress.

Figure 1 displays data on bipartisan collaboration among members of Congress, presenting data on the average bipartisan, original cosponsorship rate for three groups of members—Democrats (Dem), Republicans who voted for certification in Arizona and Pennsylvania (GOP), and Republicans who voted against certification in one or both states (Objectors)—across the three congresses (115th–117th, 2017–22).Footnote ³ These data allow us to observe the three groups of lawmakers for two full congresses before January 6, 2021 (the 115th and 116th), and an entire Congress thereafter (the 117th). Observing two congresses prior to January 6 helps ensure that the 116th Congress was not unusual in some way, and leading us to mistake any changes following January 6 as a regression to the mean. The 115th and 116th Congresses featured different majority parties, and different configurations of party control across government. Importantly, the 116th Congress featured a Democratic Party majority, just like the 117th, so any changes between the 116th and 117th cannot be attributed to a change in party control.

Figure 1.

Bipartisan Original Cosponsorship Rate

The patterns for the first two congresses are similar despite a change from unified to divided party government, and a change in partisan control of the U.S. House from a Republican to Democratic majority. The differences between the groups are slight from 2017 to 2020. Republicans who supported certification were the most collaborative, followed by objectors, and then Democrats. The 117th Congress, however, saw big changes. The overall rate of bipartisan collaborations is reduced for all groups, perhaps reflecting the damages that occurred across the board to relationships in the House. However, the biggest change is among certification objectors. In the 116th Congress, more than half of bills sponsored by objectors had a Democrat as an original cosponsor. In the 117th Congress, that share is less than 25%. This change is particularly impressive if we consider that some objectors, such as Jim Jordan (R–OH) and Mo Brooks (R–AL), never had a track record of securing Democrats as original cosponsors on their bills. Others, however, experienced serious drops. Mike Bost (R–IL) saw his bipartisan cosponsorship rate fall from better than 90% in the 115th and 116th Congresses to 44% in the 117th. Similarly, Bill Posey (R–FL) saw his rate fall from 80% in the 116th Congress to 23% in the 117th Congress.

Moving to a systematic analysis, Table 1 presents four models of bipartisan collaboration. For each, the unit of analysis in each is member-congress, with the dependent variable either each member’s bipartisan original cosponsor rate in each Congress, or the count of each member’s bills with at least one bipartisan cosponsor in each Congress. The rate analyses are OLS regressions, while the count analyses are negative binomial regressions. The first two analyses are these measures from among all of the bills and resolutions each member introduced in each congress. The latter two analyses are these measure from among just the bills (H.R.) each introduced in each congress.

Table 1.

Predicting Bipartisan Collaboration

Note: * $ p\le 0.05 $ ;

** $ p\le 0.01 $ .

Each analysis contains member and Congress fixed effects and one dummy variable, (GOP Objector), which is equal to “1” in the 117th Congress, only, if the member objected to certification (GOP Objector). As such, the results are akin to a simple differences-in-differences analysis in which the coefficients for each dummy variable show us the average within-member change in the pre- and post- January 6 bipartisan collaborative behavior for those in the objector group.Footnote ⁴

The results strongly support our expectations.Footnote ⁵ Members who opposed certification saw a 15 percentage point drop in their bipartisan cosponsorship rate for all bills and resolutions after January 6. This effect was approximately 7 percentage points, if we focus only on bills, which is still statistically and substantively significant. We find a similar effect in count models predicting the number of bills introduced by members with opposite party original cosponsors. Taken together, the results in Table 1 provide strong evidence that certification opponents were punished by their Democratic colleagues as both the rate and number of bills they sponsored with Democratic collaborators fell sharply compared to previous congresses.Footnote ⁶

One limitation of the analyses in Table 1 is that looking at all bills does not preclude the possibility that Republicans who objected to certification also changed their bill introduction behavior after January 6. To account for this potential alternative explanation, we identified all bills that were introduced to the House of Representatives by the same member in identical form in both the 116th and 117th Congresses. Members of Congress frequently reintroduce legislation in a subsequent congress, without making substantive changes, when the bill did not see action in the former. By focusing on this subset of bills, we hold constant the policy content of the bills and can then directly observe the effect of being a certification objector.

We identified 976 bills introduced by Republican members in the 116th Congress and then reintroduced in identical form by the same member in the 117th Congress.Footnote ⁷ Of these, 446 or 45.7% had at least one Democratic original cosponsor in the 116th Congress. In the 117th Congress, we see a sharp difference in the retention rate of Democratic original cosponsors when comparing Republican members who opposed certification and those who supported it. For bills reintroduced by certification objectors, 35.3% lost their Democratic original cosponsor(s) in the 117th Congress. By comparison, only 15.6% of bills reintroduced by Republicans who supported certification lost their Democratic original cosponsor(s).

Table 2 presents a bill-level logistic regression predicting whether or not each of these Republican-sponsored and reintroduced identical bills with a Democratic original cosponsor in the 116th Congress still had a Democratic original cosponsor in the 117th Congress ( $ 1 $ = kept a Democratic original cosponsor; $ 0 $ = lost Democratic original cosponsorship). Our independent variables include a dummy variable for whether or not the sponsor of the bill was a certification objector, the bill sponsor’s first dimension DW-Nominate score, and the change in the Trump vote share in the sponsor’s district between the 2016 and 2020 elections. We cluster standard errors by each member.

Table 2.

Maintaining Bipartisan Original Cosponsorship on Identical GOP Bills

Note:

* $ p\le 0.05 $ ;

** $ p\le 0.01 $ .

The results in Table 2 demonstrate that certification objectors had more difficulty retaining Democratic collaborators in the 117th Congress than did other Republicans. If we hold DW-Nominate and the change in the Trump vote at their mean levels, the probability of a certification supporting Republican losing the Democratic cosponsor(s) on a bill is 19% compared to 32% for a certification objector. Certification objectors saw their odds of retaining Democratic cosponsor(s) in the 117th Congress decline by 41% compared to their other Republican colleagues.Footnote ⁸

Finally, Table 3 presents the results of negative binomial regressions assessing if certification objectors were also less effective advancing their proposals through the House following January 6.Footnote ⁹ The outcome variables are a count of each member’s sponsored bills that received action in a House committee, action beyond a House committee, and that passed the House.

Table 3.

Predicting Legislative Effectiveness

Note:

* $ p\le 0.05 $ ;

** $ p\le 0.01 $ .

The results, once again, meet our expectations. Certification objectors saw a sharp decline in their ability to shepherd proposals through the legislative process in the 117th Congress as compared to previous congresses. The models predict that certification supporters saw, on average, 2.7 of their measures acted on in committee, 2.8 acted on beyond committee, and 2 passed by the House. In contrast, objectors are predicted to have bills acted on at about half those rates across all three measures during the 117th Congress.

These results are particularly noteworthy given that many objectors were not typically productive before January 6. This is simply not a set of members that was known for their effectiveness or collaborative behavior ex ante, yet we still find strong evidence to suggest that their colleagues punished them for their perceived actions on January 6 by not collaborating with them on legislative proposals and by not allowing their legislative proposals to advance through the House.