We present the summary statistics for the key variables in our analysis in Table 1. Because each of the dependent variables is over-dispersed, we choose to employ multivariate negative binomial regression analysis. The intraparty caucus data are based upon only when the caucuses were in existence, which in every case is less than the entire span of the data for the other variables.
Our three dependent variables also have an exceptional number of zeros (see Fig. 3). The mere presence of exceptional zeros does not alone mean that a different model must be used, though the zeros in our data are generated in distinct ways. In some cases, members may choose not to travel explicitly for personal or ideological reasons. These members would therefore produce a set of “certain zeros” because they are never going to accept an invitation to travel. Other members on the other hand may have differing probabilities of traveling based on their committee appointments or personal network of relationships in their chamber. If these members do not travel, it would be for a wholly different set of theoretical reasons and therefore these zeros emerge from a separate data generating process (King 1998). Given the distribution of the dependent variables, the data generating processes just outlined, and following other scholars in this area (e.g., Alduncin et al. 2017), we estimate a zero-inflated negative binomial model.
This modeling strategy leads to two sets of estimates per model. First, zero-inflated negative binomial models estimate a logit equation which produces the likelihood of a member falling into the certain zero category (i.e., those members with personal or ideological reasons for never traveling or those members because of their seniority status or committee assignment are never asked to travel abroad). These estimates are interpreted substantively such that a positive coefficient suggests an increased probability of a certain zero. The second set of estimates come from the negative binomial and are conditional on a member not being a certain zero. Therefore, the negative binomial estimates that are positive are associated with more country-days of travel. While one can choose subsets of variables for the logit and negative binomial equations, we argue that the decisions about whether to travel and for how long one travels are impacted by the same set of institutional, personal, and political factors. Therefore, we use the same set of independent variables for both the logit and negative binomial equations.
Our chief independent variable is Ideological
Extremity, which is calculated as the absolute value of the member’s first dimension DW-NOMINATE score. To control for institutional factors, we also include variables indicating whether the member sits on the “power” committees of appropriations, rules, or ways and means, as well as servicing on the armed services or foreign affairs committees. We expect more powerful members (i.e., those on the “big three”) and members whose committees deal with foreign policy concerns to be more likely to travel. Committees dealing with public policy surrounding foreign aid and military activity, especially foreign affairs, have historically been frequent funders of travel (Carnahan 1953). And, as leadership in the House continues to gain power over the past couple of decades (Curry 2015; Theriault 2008), we would expect members of influential committees—who themselves are on the outskirts of leadership if not formally members—to also have disproportionate opportunities to travel.
With an eye toward members’ personal ambitions, chiefly their proximate goal of reelection (Mayhew 1974), we include the member’s vote-share in the election entering the congress. Safer members should feel more comfortable traveling more often. If a member has significant support from their constituents, it is easier to convince them that fact-finding missions abroad are necessary for improving their job performance (McGee and Moniz 2021), or, if not, the punishment that constituents might inflict upon members who excessively travel would be less costly to electorally safe members. We also include the number of terms a member has served in the House of Representatives. More senior members have had more opportunities to travel and also be more likely to have accrued more area-specific policy expertise. Members who choose to pursue good public policy as a goal seek out opportunities to travel to assess the effect of their policymaking abroad. Other scholars have shown evidence that certain types of travel can actually lead to members being more effective lawmakers and that trips can include policy specific information valuable for members who seek additional context or expertise (Moniz and McGee 2021).
Finally, we include a set of demographic variables to attempt to understand how descriptive features of members might impact their travel behavior. Therefore we include binary variables indicating if a member is Black, Latino, or female. It is possible that members might be less inclined to travel if they are nonwhite or female because of concerns about how they might be perceived by their constituents when they travel. We also choose to run separate analyses on Democrat and Republicans as well as include clustered standard errors by member in all models.
While this base model is sufficient for testing both the Ideologically Extreme Hypothesis and the Ideologically Moderate Hypothesis, we must include additional variables to test the Warrior Hypothesis. These models are the same as the base model except they also include binary variables indicating whether a member is part of each faction within their respective party. Members can belong to more than one faction. The Democratic factions include the Congressional Progressive Caucus (liberal), the New Democrat Coalition (moderate), and the Blue Dog Coalition (moderate). The Republican factions include the House Freedom Caucus (conservative), the Tea Party Caucus (conservative), the Republican Study Committee (conservative), and the Main Street Coalition (moderate). Due to limits in faction Rep. Study Committee rosters, these models are restricted only to data from the 107th through the 115th Congresses. Note that, for the Republican factions, multiple models are run because the Tea Party Caucus forms in the 111th Congress and the House Freedom Caucus forms in the 114th Congress, so those models are based on even less data.
Before modeling partisan or bipartisan country-days abroad, it is helpful to see what patterns emerge when we analyze all trips together (see Table 2). By conducting this initial analysis, we are more equipped to understand which factors drive travel in general as well as how those factors impact decisions about taking trips with members from their own, or the other, party. From the start, we see negative and significant logit coefficients for many of the factors that we might expect to drive travel decisions, and in almost every case (across both the Democratic and Republican models), those same variables have significant and positive–negative binomial coefficients. Taken together such a pairing of coefficients indicates both that these variables are associated with members being less likely to reject travel altogether (i.e., not be “certain zeros”) and take more country-days abroad.
The factors we identify here are common culprits in the congressional travel literature; safer members, more senior members, and members associated with powerful committees or those committees with jurisdiction over foreign or defense policy all are associated with increased travel. While none of these findings are surprising, they do produce a useful benchmark as we start to analyze the partisan and bipartisan trips. Besides those variables just identified, we also see Latino Democrat taking more trips abroad and female Democrat taking fewer trips abroad. While the former finding is a bit of a surprise, the finding for female members is in line with our expectations because, perhaps, women fear receiving more scrutiny for engaging in corrupt or questionable behavior. Furthermore, it behooves us to note that ideological extremity demonstrates no significant relationship with travel in general. This finding situates our argument that ideological extremism is a relevant factor mainly when considering with whom members are traveling, rather than as a factor leading them to want to travel at all.
Having established a useful set of baseline findings, we next turn to patterns in partisan member trips abroad (see Table 3). Among both Democrat and Republicans, we see more ideologically extreme members spending more country-days abroad with only copartisans. This finding provides support for our Ideologically Extreme Hypothesis. For Democrat, the most ideologically extreme members travel about five country-days with their copartisans (see Fig. 4). The Democrat with an average ideology score spends just two country-days abroad with copartisans. For Republicans, the effects are very similar with the most extreme member traveling with copartisans for about four country-days, and the average remaining just under two country-days (see Fig. 5). Taken together, these findings suggest that the most extreme members within each party are traveling with only copartisans twice as much as their most moderate colleagues.
We also find a familiar set of negative logit coefficients, indicating a decreased likelihood of rejecting travel altogether, associated with safer members, members on the armed services committee, and members on the foreign affairs committee across both models. Only Democrat on the foreign affairs committee also increased their partisan travel abroad though. It is unclear why there is a differential effect here among Democrat and Republicans. Foreign policy tends to be one of the more conciliatory areas of public policy when it comes to partisanship. Yet, perhaps we are seeing that even that era is starting to fade. More senior Republicans and Black Republicans also travel more with their fellow partisans (the latter contains so few members that caution should be used in interpreting it). It is puzzling to see more senior Republicans embracing more partisan travel since much of the discourse around the contemporary Republican Party suggests that its newer members are to blame for the more raucous behavior exhibited by their caucus (Theriault 2008, 2013; Green 2019).
Next, we examine bipartisan travel behavior. We, again, find safer, more senior, and members from the powerful or travel-inclined committees driving the results. Given that we did not see these variables driving our partisan-only results (except for more senior Republicans and Democrat on the foreign affairs committee), it is not a surprise to see them stand out here. This set of results is therefore consistent with the total travel results from Table 1. We also find support for the Ideologically Moderate Hypothesis, wherein more moderate members are more likely to spend more country-days abroad with members of the opposite party, among Democrat but not Republicans. The most moderate Democrat spend almost 11 country-days abroad with Republicans compared to only about five among their most extreme counterparts (see Fig. 6). This finding indicates that moderate Democrat are traveling about twice as much with Republicans compared to their most extreme colleagues, which is again suggestive of congressional social networks being infected by ideologically charged discontent. This finding is also consistent with the asymmetric polarization of the American party system wherein Republicans having been driven farther to the right by the contemporary conservative ideological movement (Grossmann and Hopkins 2016; Hacker and Pierson 2006; Theriault 2008). Since Democrat are more likely to garner support from diverse interests, it makes more sense to see their more moderate members clinging to bipartisanship and the vestiges of compromise. Republican Latino members appear less likely to reject travel altogether, yet this was also true in the partisan travel model. Democratic Latino members do take more bipartisan trips abroad and this type of travel seems to be driving the finding from the total travel models in Table 1.
Overall, we can also conclude that, with occasional exceptions for Democrat on the foreign affairs committee and more senior Republicans, the traditional factors associated with increased travel among members of Congress seem to be driving bipartisan travel most. Despite this finding, we have also identified a new relationship among ideologues in each party and increased trips abroad with only their copartisans. Yet, we have also proposed hypotheses related to travel that may be shaped in part by ideology but also by the social connections members form with one another. As these intraparty factions are closeddoor and ideologically focused, we expect membership to reinforce the patterns in travel behavior that ideological extremism produces in general. That is, moderate groups like the New Democrat, Blue Dogs, and Main Street Coalition should encourage bipartisan country-days abroad because these groups often note their interest in forming governing coalitions across the aisle (McGee 2020). On the other hand, the ideological extremism of members in the Progressive Caucus, Republican Study Committee, Tea Party Caucus, and Freedom Caucus are likely to produce increases in partisan country-days abroad. In particular, we expect the largest effects to come from the Freedom Caucus, because of its distinct reputation as a home base for true warriors in the House Republican Party (Table 4).
Among Democrat, we find that service on the foreign affairs committee is still oddly driving increased partisan travel (see Table 5). We also see members of the Blue Dog Coalition being less likely to reject partisan travel outright, which is at odds with our expectations. Yet, we do not see the negative binomial coefficient with a significant effect so we cannot also say we are seeing them go on increased travel. In fact, looking at the bipartisan travel model, we see relatively large effects for members in the more moderate factions (i.e., the Blue Dogs and New Democrat) traveling more country-day abroad with members of the opposite party. Members of the Blue Dog Coalition spend about 16 country-days abroad with Republicans compared to only about eight among nonmembers (see Fig. 7).
The effect size among the New Democrat is smaller, but still significant. Members of the New Democrat spend about 12 country-days abroad with Republicans compared to nonmembers who only spend about nine (see Fig. 8). These findings are consistent with our expectations that more moderate members would prefer bipartisan as many of them focus on building coalitions across the aisle in favor of governance. It also makes sense to see the larger effect sizes coming from the Blue Dog Coalition, which is a smaller and more tight-knit faction compared to the New Democrat Coalition. Other factors we have seen before among Democrat pursuing bipartisan travel still appear to be playing a role. That is, we still see safer members, more senior members, Latino Democrat, and members of the foreign affairs committee taking more country-days abroad with Republicans.
Shifting now to the Republicans, we discuss the results as broadly as possible given the unorthodox number of models required due to the different founding dates of the various conservative factions throughout the period of analysis. Looking first at partisan travel (see Table 6), we find that members of the Republican Main Street Coalition are less likely to take trips abroad with their copartisans, at least in one of the models. nature of the finding though it is easily quibbled with. We see some evidence again of more senior Republicans being more likely to travel with copartisans. Generally speaking, the most consistent results relate to those factors related to whether or not a member rejects travel outright, which is not the main focus of this paper nor are the results significantly different from the previous models discussed.
Turning to bipartisan travel (see Table 7), we do present some consistent results in the direction we expect for the conservative factions. As soon as it is founded, we see members of Tea Party Caucus are less likely to take bipartisan trips abroad. In the Tea Party Model, we see members of the Tea Party Caucus spending about seven country-days abroad with Democrat compared to about nine country-days for nonmembers (see Fig. 9). The effect size is smaller in the Freedom Caucus model, with Tea Party members spending just under 10 country-days abroad with Democrat as compared with about 11 for nonmembers (see Fig. 10).
We also find members of the House Freedom Caucus being more likely to outright reject any bipartisan travel. While we do not see the expected increase in partisan travel from these members, the evidence that they are less likely to take bipartisan trips—or even consider them in the case of the HFC—is a step toward evidence of our Warriors Hypothesis. Oddly, we also find a single instance of the RSC being more likely to engage in bipartisan travel, but like the Main Street Coalition finding from the previous set of models it fades and therefore is on flimsy ground to be asserted as a solid finding. The drivers of bipartisan travel we have seen before again show up with safer, more senior, and certain committee service improving the likelihood of taking trips with those across the aisle.
Taken together, the results from highlighting factions largely reinforced our findings from the initial modeling. Moderate Democratic intraparty caucus members spend more time abroad on bipartisan trips and conservative Republican intraparty caucus members tend to stay away from bipartisan trips. While we had hoped to find that factions were also reinforcing the disposition of ideological extremists to travel among their own, we did still uncover that relationship when utilizing our scores derived from DW-NOMINATE. Overall, our results also bolster already existing findings about the institutional, demographic, and district factors that drive the desire of members of Congress to travel. The results suggest that we were accurate in describing the House Freedom Caucus as the easiest case in a difficult test. The substantive results were generally pretty weak for the intraparty caucus Rep. Study Committee, but were strongest for the House Freedom Caucus.