Do Local Roots Impact Washington Behaviors? District Connections and Representation in the U.S. Congress

Abstract

Although commentators often point to the political value of legislators’ geographic ties, less is known about the influence of such connections once in office. Given recent scholarship underscoring the importance of geography as a dimension of identity, we argue that local legislators should behave as descriptive representatives. We collect the hometowns of all members of Congress with known birth locations from 1789 to 2020 to analyze how being born near one’s district impacts legislator behavior. We connect these data to information on a series of behaviors, finding that local legislators emphasize constituency work over policymaking and party-building. Moreover, while local legislators do not demonstrate substantively less partisan unity in roll-call voting, they attract a higher percentage of out-party cosponsors to their bills. Together, our results point to important representational implications regarding the geographic roots of legislators and the role of local connections in the contemporary Congress.

INTRODUCTION

In 2022, Representative John Katko (NY-24) was ranked the third most bipartisan member of the U.S. House and the most bipartisan member of the New York congressional delegation based on his cosponsorship behavior. ¹ In explaining his commitment to bipartisanship amid an increasingly partisan environment, Rep. Katko pointed directly to his commitment to the community he represents, stating “My work across party lines has been deliberate and a point of pride throughout my career…Since coming to Congress, I have aimed to make Congress work again and deliver meaningful results for Central New York.” ² A self-described “Camillus, NY native,” born in the district he now serves, ³ Rep. Katko is a prime example of a legislator with local roots: a legislator with deep geographic connections to their district. At least based on Katko’s telling, such roots can be quite influential over a legislator’s priorities.

Beyond just passing speech, however, both classic and recent political science research suggest that local connections serve as more than mere campaign slogans. According to Fenno (1977, 899), for example, legislators emphasize a wide variety of personal traits—including geographic roots—in an effort to convey “a sense of identification” with one’s constituents, and to emphasize that “[y]ou can trust me because we are like one another.” More recently, Hunt (2022) provides evidence consistent with the idea that local connections can be thought of as a form of descriptive representation, by which members and constituents have a shared place-based identity. Much like other descriptive connections, these local, geographic identities may bolster trust and understanding between a representative and her constituents.

But while such scholarship has investigated the electoral and communications consequences of local reputation (Ansolabehere, Snyder, and Stewart 2000; Fenno 1978; Hunt 2020a; 2022; Kaslovsky 2022), surprisingly little remains known regarding the behavioral implications of local roots when in office. A legislator’s connection to their district lies at the core of the legislator–constituent relationship; indeed, the framers of the U.S. Constitution themselves designed Congress in such a way as to build upon voters’ sympathies for geographic units (e.g., Federalist No. 17). Moreover, previous scholarship underscores numerous in-office behavioral patterns that ought to distinguish descriptive representatives from their nondescriptive counterparts. If local-roots legislators do experience the constituency relationship differently than their nonlocal counterparts, such differences may well redound to their time in office.

Despite the many reasons why such geographic linkages might influence legislative behavior, members of Congress today operate within an environment distinct in many ways from the “textbook Congress” confronted by Fenno. In addition to growing ever more polarized over time, Congress has become engulfed in tightly matched contests over majority control of both the House and the Senate (Lee 2016)—generating pressure for members to commit more resources to partisan fundraising (Heberlig and Larson 2012) and greater effort to messaging. At the same time, scholars have argued that American politics has nationalized (Grumbach 2022; Hopkins 2018), pulling policymaking efforts and legislator attention farther and farther away from local concerns.

In this article, we therefore investigate whether local-roots legislators maintain differences in priorities today and whether and to what extent the “textbook” Congress actually resembled other periods in congressional history. More specifically, we ask: (1) Are local legislators empowered to focus more on policy matters due to their locally based electoral advantages, or do they focus more on helping their communities with local, constituent work as descriptive representatives? (2) Can local roots “release” legislators today—or have they in the past—from partisan pressure in their legislative behavior? To answer these questions, we leverage a new dataset of 11,016 members of Congress with known birth locations (1789–2020) to investigate patterns in how far a legislator was born from their district over time. We combine these data with information on legislator “type” and staff investments from 1993 forward to investigate how local legislators balance partisan considerations relative to policy- and constituent-oriented activities. Thereafter, we use data from 1973 to 2020 to examine the impact of local roots on roll-call voting and cosponsorship behavior. Together, these data enable us to analyze the impact of local roots via the broadest analysis of legislator birthplace and its ramifications to date. Ultimately, we find that local roots operate in a fashion consistent with descriptive representation, leading legislators toward more locally oriented and bipartisan pursuits, rather than enabling them to act more partisan or ideological. At the same time, however, both the numbers of such descriptive local representatives and the influence of those local roots have appeared to abate over time—from a high watermark in the early-1970s.

We proceed as follows. First, we detail how, if local roots function as descriptive representation, local legislators ought to behave differently from their nonlocal counterparts. At the most basic level, while local connections enable a legislator to get into—and stay in—office more consistently, such legislators may feel obligations to commit greater resources to constituency service than their nonlocal counterparts. Thereafter, we extend this logic to show how local legislators’ differential approaches to reelection and sense of identification and obligation encourage them to behave in a less partisan fashion, and possibly even less policy-focused fashion, than similar nonlocal legislators.

We next present a series of tests of these expectations. We first compare local and nonlocal legislators using Bernhard and Sulkin’s (2018) data on legislative style. After establishing that locally born legislators are significantly least likely to adopt party-centric styles, we use a within-district design to show that although local legislators may enjoy greater baseline connections with their constituents, they nevertheless hire a higher percentage of constituency service staff than do their nonlocal counterparts. Finally, we then trace the logic of local and nonlocal difference through to roll-call voting and cosponsorship, finding that while local legislators do not exhibit substantively different partisan unity in their roll-call voting, they do attract a higher percentage of out-party cosponsors to their bills.

Taken together, these results indicate that local connections have important consequences for understanding legislator behavior, the power of parties, and nationalization in American politics. In particular, we agree with recent work that local roots constitute a meaningful and under-explored dimension of descriptive representation, which influences how legislators spend scarce resources and focus their time. In fact, our findings suggest that such legislators are uniquely suited to bypass partisan pressures and work with members across the aisle. And, despite the fact that our longitudinal findings underscore a decline in geographic connections since the days of the “textbook Congress,” they nevertheless indicate that Congress remains a surprisingly local institution. Finally, although this study focuses on the U.S. Congress, the results speak to a core aspect of representation: the need for members to balance partisan, local, and policy pressures, particularly in systems with geographic districts.

LOCAL ROOTS AS DESCRIPTIVE REPRESENTATION

The clear importance of local roots to representation is underscored by the long-standing interest that political strategists and scholars have shown in the topic. Among the earliest treatments is also among the most famous: in his classic work, Key (1949) describes a “friends-and-neighbors” effect among voters whereby voters appear to favor candidates from their local area, even when such candidates may not share the voters’ policy preferences. This effect, Key laments, overwhelms the accountability link between the elector and the elected, enabling representatives to behave in a manner otherwise unacceptable to the voter.

Later, Fenno (1978) expanded on the “friends and neighbors” effect by exploring the intersection of local roots with how members’ present themselves to the constituency. In addition to conveying qualifications, Fenno (1978) suggests that members attempt to convey a sense of identification and empathy with their constituents. Giving the impression that “I am one of you,” empathy has the goal of projecting the image that “I understand your situation and care about it” (Fenno 1977, 899). Personal contact specifically builds trust between the constituent and the legislator, and Fenno (1977, 888) reports that he often found himself “fashioning a highly subjective ‘at homeness index’ to rank the degree to which each congressman seems to have support from and rapport with each group.” According to Fenno, this identification indeed provides voting “leeway” in Washington.

We believe that classic accounts like Key’s and Fenno’s, as well as several more recent lines of research, provide reason to believe that local roots constitute an important dimension of descriptive representation in U.S. politics. Indeed, more than a cliched line in a legislator bio or first line of a campaign ad, local-roots legislators may see themselves as standing in for the perspectives and experiences of people from their geographic units. Moreover, from the voter’s perspective, both the “friends and neighbors” effect and the “empathy” that Fenno underscores are consistent with voters’ conceiving of local roots as a dimension of descriptive representation. As Mansbridge (1999) explains, descriptive representation “forge[s] bonds of trust” and enhances the “feeling of inclusion.” And, as Bianco (1994) argues, identification with one’s representative fosters the belief that the representative shares the values of the constituent.

Empirical research has borne out the prediction that descriptive representation encourages political trust, at least under certain conditions, for identities such as race and gender (e.g., Bowen and Clark 2014; Gay 2002; O’Brien and Piscopo 2019). However, in order for this logic to fully extend to local-roots legislators, local roots themselves must be considered a sort of identity to both legislator and voter. Here, recent research on geographic roots provides quite suggestive evidence in favor of geography-as-identity in politics. In her work on rural resentment, Cramer (2016) provides compelling evidence that place-based identities impact perceptions of relative power and resource allocation, consequently increasing support for candidates who play on these identities. In this context, geography influences political participation in a way that is “complex, many layered, and grounded in fundamental identities,” shaping both trust in elected officials and policy preferences (5). Munis (2022) extends this work by creating a more general “place resentment” scale related to place-based identity. Most recently, in the most comprehensive work on local roots and legislatures to date, Hunt (2020b, 347–8) argues that “local roots offer legislators significantly more opportunities to personally relate to constituents of all partisan persuasions, based on a shared, distinctly local identity that cross-cuts partisanship…” Taken together, several scholars have concluded that geographic identification does indeed constitute a meaningful and consequential identity in politics (Campbell et al. 2019).

If this depiction is correct, it entails a broad series of implications for members of Congress and their approach to representation—both within elections and in the behaviors occurring thereafter. To date, scholarship has focused almost entirely on the electoral implications of local roots. Scholars have demonstrated, for example, that presidential candidates possess an advantage in their home states (Lewis-Beck and Rice 1983), that governors with local connections provide significant coattail effects (Meredith 2013), and that candidate-county ties boost turnout in both primary (Panagopoulos, Leighley, and Hamel 2017) and general elections (Panagopoulos and Bailey 2020). In Congress specifically, Hunt (2020b) uses an original dataset of House incumbents from 2002 to 2018 to show that legislators with local roots receive higher vote shares than their party’s presidential nominee. Similarly, Hunt (2020a) demonstrates that members with local roots are significantly less likely to be challenged in primaries. Of course, scholars have also shown that the strength of such electoral effects varies by context: place-based appeals appear to be especially impactful for rural voters (Jacobs and Munis 2019), and they seem to fade in effectiveness as distance from the legislator’s home county increases (Gimpel et al. 2008). Still, the local-roots advantage has proven so appealing that examinations in a wide variety of other political contexts have uncovered similarly strong preferences for local candidates (Fiva, Halse, and Smith 2020; Jankowski 2016; Tavits 2010)—including in less “personal” systems such as proportional representation (Shugart, Valdini, and Suominen 2005).

Given that we know strategic choices made during the campaign have strong consequences for legislators’ behaviors after assuming office (Sulkin 2005), and given the centrality of descriptive representation to legislative behavior in other contexts, it stands to reason that local-roots legislators should exhibit significantly different legislative behaviors from their nonlocal counterparts. And, indeed, in an analysis of communication styles in e-newsletters, Hunt (2022, 127) reports that local legislators spend more time discussing local places and institutions and less time using partisan terminology than their nonlocal counterparts. Yet little is known about how local roots affect legislative behavior. ⁴

We therefore put this general expectation to the test, extending Hunt’s work on the electoral and communication consequences of local roots to legislative behavior and style. In particular, we investigate whether perceived responsibilities as descriptive representatives lead local-roots legislators to demonstrate a more constituency-focused style of representation, “insulated” from party pressure.

Local Roots, Local Commitment, and Resource Allocation

As Arnold (1990) aptly summarizes, members of Congress face competing pressures when performing their official duties as legislators. We argue that local roots, as the basis for an important dimension of descriptive representation, have far-reaching ramifications for how legislators respond to these pressures.

Existing work suggests that descriptive representatives may feel a greater responsibility to support their constituents in the district. For example, Broockman (2013) demonstrates that Black state legislators are more likely to respond to written requests from Black constituents that offer limited political benefits than are white state legislators. Lowande, Ritchie, and Lauterbach (2019) found that women, racial/ethnic minorities, and veterans are more likely to advocate on behalf of constituents with similar identities to their own by contacting federal agencies. Further still, Grose (2011) provides evidence that descriptive representation enhances both constituency service and project delivery to constituents. If place-based identities function in a similar framework as these identities, then local legislators may feel a similar obligation to allocate important resources toward the district.

From the voter’s perspective, a legislator’s connection to the district may also impact constituent expectations, leading them to expect more help from those with stronger ties to the community. Gay (2002) finds that constituents are more likely to contact descriptive representatives, perhaps because “the shared experience ‘imperfectly captured’ by descriptive representation can form the basis for greater trust in public officials and institutions (Mansbridge 1999, 641).” Even more directly, Campbell et al. (2019, 939) suggest that “Voters may believe that MPs with local ties are more emotionally connected to the constituency and better informed about constituents’ needs” and provide evidence that legislators are rewarded for doing so. If constituents expect locally rooted legislators to provide more “locally focused” representation, there is all the more reason to expect that local-roots legislators will allocate greater resources toward the district than those without a place-based connection to their constituents. For these reasons, we hypothesize that local-roots legislators will commit greater resources to the district—in particular, more of their scarce congressional staff—than will their nonlocal counterparts.

Nevertheless, such investment also implies trade-offs—including allocating fewer resources toward passing desired policies and furthering the party brand. Policymaking is considered to be one of the main goals of members of Congress (Fenno 1973), and experienced staffers have a substantial impact on legislative productivity (Crosson et al. 2020; Ommundsen 2023). We also know that today’s most strident partisan warriors in Congress focus in great part on communications investments (Lee 2016, 112), as broadcasting a clear and consistent message requires investment into communications and public relations staff.

Given these pressures, it is conceivable that local legislators use their electoral leeway to function as policymakers and party leaders. In sharp contrast, we argue—and ultimately provide evidence—that local legislators, as descriptive representatives, instead double down on their local appeals by investing in constituency service. That is, we suspect that they focus their resources more on constituency service, and less on communications, legislation, and party-building activities, than legislators born outside of their districts.

Local Roots, Electoral Insulation, and Party Unity

While we expect representatives with geographic ties to their districts to feel a sense of obligation to “localize” resources, legislators nevertheless have lawmaking responsibilities that they cannot abandon. Indeed, every member must cast hundreds of roll-call votes each legislative session, and nearly every legislator sponsors at least some legislation. As Arnold (1990), Kingdon (1989), and many others have underscored, legislators must take into account not only their own opinions and the opinions of the constituency when undertaking these activities, but also the traceability of the policy and the position of party leaders.

We believe that local-roots legislators, as descriptive representatives, will also weigh these competing pressures differently from their nonlocal counterparts. In particular, we hypothesize that local-roots legislators will undertake policymaking behaviors that are less partisan when they share a local connection with their constituency.

Traditionally, congressional parties have been depicted as exercising “negative” or gatekeeping power (e.g., Cox and McCubbins 1993; 2005), but, as other research has shown, party powers today include a wide variety of tools necessary for more “positive” and disciplinary activities. As partisan competition has grown, members of Congress have ceded greater power and responsibility to party leaders (Lee 2015), who have in turn used that power to exercise greater discipline over the rank-and-file (Rohde 2010). Indeed, a wide variety of recent research has underscored the lengths that party leaders go to guarantee unity on roll-call votes, including but not limited to committee assignments (Leighton and Lopez 2002), campaign contributions (Heberlig and Larson 2012), and even preferential policymaking treatment (Hasecke and Mycoff 2007).

These trends place many legislators in a difficult position. On the one hand, Canes-Wrone, Brady, and Cogan (2002) have shown that legislators who vote with the extreme ends of their party are more likely to be voted out of office. On the other, legislators run the risk of facing serious primary challenges, loss of campaign funds, and public ridicule if they run amok of party orthodoxy. We argue that legislators with local roots can use their status as descriptive representatives to sidestep this predicament in a way that legislators lacking such connections cannot.

According to Hunt (2020b), local incumbents are especially adept at converting local trust into electoral advantage, levering the “cross-partisan” nature of local appeals. Adopting such a cross-partisan appeal could in turn release local legislators from maintaining stringent party discipline. As Fenno (1978) underscores, legislators carefully hone their home styles with one overarching goal in mind: the development of trust between the constituency and the representative. This trust grants the legislator a certain amount of “leeway” in their Washington decision-making, as constituents believe that the legislator is acting faithfully on their behalf. If local-roots legislators can more easily access this kind of “leeway,” they may behave differently inside Congress than nonlocal legislators. In fact, local-roots legislators may well feel more obligated to act in a more localized manner than nonlocal legislators, as a failure to do so would amount to a betrayal of that identification and trust.

Of course, “leeway” can cut multiple ways with respect to partisanship and ideology. That is, while some legislators may use “leeway” to break from the party in a moderate direction, still others may use it to buck their party in an extreme or partisan direction. In the case of leeway generated through local roots, however, we posit that legislators—as descriptive representatives of their constituents—will use their leeway to act in a more bipartisan or cross-partisan fashion. In serving as descriptive representatives, local-roots legislators may feel a responsibility to their constituencies to represent as large a swath of constituents as possible. Inasmuch as acting in an extreme fashion constitutes a more divisive approach to voting and legislating, local-roots legislators will be more apt to use their trust and leeway to pursue more bipartisan behaviors. Here again, we underscore that this logic is consistent with findings from Hunt (2022), who finds that local-roots incumbents issue e-newsletters with more localized, and less partisan, language. In the empirical analyses to come, as well as those in the Supplementary Material, we attempt to adjudicate between these competing accounts.

In sum, we posit that local roots give legislators both a line of electoral support that is not tied directly to party and endows them with a greater sense of responsibility to their constituents. As such, we expect that local roots will not only insulate legislators from party pressures but encourage them to behave in a bipartisan direction. We put this argument to the test by first exploring whether local-roots legislators are more or less likely to be classified as party “soldiers” or “builders” in comparison to nonlocal legislators (based on Bernhard, Sewell, and Sulkin’s [2017] classifications). We then examine two principal policymaking activities: roll-call voting and cosponsoring legislation.

DATA AND MEASUREMENT

In order to examine the effects of local roots on legislative behavior, we collected data on legislator birthplaces from the Biographical Directory of the United States Congress. ⁵ We selected birthplace as our base measure of local roots for several reasons, both theoretical and practical. First, an individual cannot choose where she is born—similar to other identities for which descriptive representatives feel a sense of representational obligation. Other types of local ties, such as post-secondary education or previous political experience, however, are themselves (to varying degrees) a function of the candidates’ own choices. Thus, not only do such factors differ qualitatively in relation to dimensions of descriptive representation like race and gender, but they are also themselves the result of strategic choices on the part of prospective congressional candidates. This is not to say that such factors lack value for a candidate: indeed, Hunt (2022) demonstrates that deeper connections to the district—including those forged later in life—contribute to the electoral advantage enjoyed by local-roots candidates, as well as to the local focus of their communications strategies. However, with respect to descriptive representation in office—our primary focus in this article—we aim to capture the dimensions of local roots that tap into the “symbolic” value of local connections, as opposed to their “instrumental” benefits (Hunt 2022, 12). ⁶

We proceeded by scraping and cleaning the biographies in the Directory, leaving us with the birth places of 11,016 unique legislators over 116 congresses. Using this information, we first determined whether a member was born directly in their current congressional district. To do so, we used ggmaps (Kahle and Wickham 2013) in R to obtain the latitude and longitude of these locations and overlaid them on district maps from Lewis et al. (2013). This provides us with a binary measure of local roots; overall, about 40.9% of legislators were born directly in their congressional districts. ⁷ We then validate our binary measure of birth-in-district by replacing it with Hunt’s (2022) measure (see Supplementary Table A.2). Despite some tactical differences in underlying approach for locating birthplaces, we were encouraged by the high level of agreement between the results (see Appendix A of the Supplementary Material for additional details). ⁸

Nevertheless, there is naturally some measurement error in geocoding. In fact, in collecting our data, we identified several sorts of issues that we believe are unaddressable using a binary measure alone. Among the trickiest of these issues occurs when members are simply listed as being born inside large cities. For example, if a member is listed as being born in “Chicago,” Google Maps (the engine underlying ggmaps) may provide an incorrect latitude and longitude within Chicago, simply due to a lack of specificity regarding a legislator’s first childhood residence or hospital of birth. ^{9 , 10}

As a result, we next took members not born directly in their districts and calculated the shortest geodesic distance from their birth place to any coordinate along the border of their congressional district. We believe this measure ameliorates the aforementioned ambiguity error, because the hypothetical member from “Chicago” would still be listed as being from within a few miles of their actual birth place—as opposed to simply being marked as “not born in the district.” Thus, while we provide results for binary measurements for comparison to previous research, we also operationalize our main independent variable as logged distance born from the district. ¹¹ The end result is a dataset with 39,489 observations at the legislator-congress level, spanning from 1789 to 2020 (Crosson and Kaslovsky 2024).

Figure 1 displays the distribution of the unlogged independent variable, demonstrating the heavy right skew in the data. The left panel zooms in on the distribution of members born within 100 miles from the district, while the right panel displays the rest of the distribution. About 68% of the data fall into the left panel. This figure underscores what years of received wisdom about U.S. politics has underscored: the U.S. House of Representatives exhibits considerable localism, at least in terms of the type of legislator who serves. According to our estimates, about 52% of all observations and 48% of unique legislator-districts are either zero (legislator was born in the district) or under 10 miles from the nearest district border. Recast slightly, about 49% of legislators in the House of Representatives have represented a district fewer than 10 miles from their birthplace, and 62% have served in their state of birth. Although directly comparable data from other countries are (to the best of our knowledge) not available, the 63% of the 114th U.S. House representing their state of birth easily outpaces the estimated 47% of British MPs representing their region of birth in 2015. ¹²

Figure 1. Count of Legislators by Distance from the District, 1789–2020

These figures, however, have ebbed and flowed considerably over time. Median distances (in miles) that members are born from their districts, 1st to the 116th Congress, are presented in Figure 2. We present these data along with dates for important transportation innovations, to provide additional context to the evolving meaning of distance over time. ¹³ As the figure shows, there is an upward trend in median distance until about the 40th Congress (shortly after the Civil War), when there is a stark decrease in member birth distance from their districts. The decrease continues for over 100 years, peaking during the 92nd Congress (the early 1970s).

Figure 2. Average and Median Distance from District, 1789–2020

Overall, the distance members are born from their districts has increased over time—though not monotonically by any means. In general, the House has remained astoundingly local. Despite the fact that members of Congress are wealthier and better educated than the average American (which demographers have shown to positively predict geographic mobility), they nevertheless are more likely to serve in their birth state (60%) than the average American is to work in theirs (58%) as of 2019. ¹⁴ In the peak of House localism in the 1970s, 61% of House members were born within 10 miles of the districts they were representing, and 74% served in their birth state. In comparison, the 1970 U.S. Census reports that about 65% of Americans were born in their state of residence. ¹⁵

At least two trends in these raw data seem noteworthy. First, although transportation advances certainly made living and moving farther from one’s district more feasible, such advancements’ association with our distance-from-district metric are quite mixed. Certainly, the introduction of passenger rail, and its growth throughout the 1800s, coincides with farther legislator distance from birthplace. Such figures also generally coincide with the nation’s move westward and with major influxes of immigrant populations. However, other technologies, such as the automobile and even the airplane have, at best, extremely delayed effects. We do not wish to draw anything close to causal conclusions from these high-level trends. Rather, we note only that, although members of Congress are no doubt representing districts farther from their birthplaces today than in most periods of American history, this trend is hardly monolithic.

Second, we note that the most “at home” period of congressional representation is not a hundred or more years ago but during the late 1960s and early 1970s. That is, the “textbook” Congress upon which Fenno built his study of homestyle—and upon which many foundational texts on Congress were written—appears far more extraordinary than ordinary, at least in terms of political geography. Indeed, even the Congresses of the 1940s, during which Key observed his “friends and neighbors effect,” were notably more “local” than most of congressional history. Given how many foundational works of representation build upon empirical observations from this period, these data suggest that studies of representation today may well confront a Congress that differs considerably in basic dyadic linkages between members and their districts.

These broad trends notwithstanding, we do not want to overinterpret over-time differences. Indeed, the meaning of distance has varied considerably over time. As the ease of travel has improved, being born 50 miles from the district today likely means something very different than it did in the 1800s. Due to such factors as well as data constraints on our independent variables, we limit our main analyses to 1973 onward. Below, we connect these data with a wide variety of information on legislative behavior in order to assess the ways in which local-roots legislators differ in office from their nonlocal counterparts.

FINDINGS

Legislative Behavior and Local Roots

Before examining the specific behaviors we highlight above, we first investigate differences between local and nonlocal legislators at a broader level. More specifically, we assess our baseline claim that local-roots legislators approach representation with a stronger focus on matters of constituency service, relative to partisan and policy pursuits. Measuring one’s status as a party soldier, policy expert, or constituent servant is difficult both conceptually and methodologically, so we draw upon recent advances in the measurement of legislative “style” to examine whether local-roots legislators behave distinctly from their nonlocal counterparts.

In their book on legislative style in Congress, Bernhard and Sulkin (2018) introduce a useful measure of legislator “type” that combines a wide variety of resource-allotment, fundraising, and other behavioral outcomes. These categories range from party-focused roles, such as “Party Soldier” and “Party Builder,” to more policy-focused styles such as “Ambitious Entrepreneur” and “Policy Specialist.” Importantly, the method also classifies legislators as “District Advocates.” Using these categorical data, we estimate a series of multinomial logistic models, introducing our two measures of local roots as predictors for legislator type.

As with the remainder of our models, we include a variety of secondary variables known to influence legislative behavior. We include indicators for each congress—and, later, district-level fixed effects (which are not included in the legislator-type models due to the outcome variable exhibiting very little within-unit variation). ¹⁶ The specification of these models, and those thereafter, take on the following form:

$$ \begin{array}{rll}{Y}_{idc}={\beta}_1{\mathbf{LocalRoots}}_{id}+\Phi {\mathbf{Z}}_{id}+{\alpha}_d+{\tau}_c+{\epsilon}_{idc},& & \end{array} $$

where i indexes legislators, d indexes districts, and c indexes congresses. $ {\mathbf{Z}}_{id} $ represents the vector of controls, including party, seniority, majority party status, membership on powerful committees, chairmanship status, legislator gender, and same party presidential vote share. ¹⁷ Institutional position likely impacts a legislator’s ability to act independently, and gender has been shown to influence the extent to which legislators feel pressure to behave in district-oriented ways (Lazarus and Steigerwalt 2018). Same Party Presidential Vote Share is used to capture district safety, which may influence how much leeway legislators perceive they have to focus less on the party and more on the district. ¹⁸ $ {\alpha}_d $ encodes district fixed effects, which address the possibility that variation in preferences over legislative style may be due to inherent, unchanging district characteristics. These fixed effects hold such time-invariant characteristics constant. ¹⁹ $ {\tau}_c $ encodes congress fixed effects, ²⁰ and standard errors are clustered on districts. ²¹

For ease of presentation and interpretation, we collapse the five legislator-type categories into three: “District Focused” (District Advocates), “Policy Focused” (Ambitious Entrepreneur and Policy Specialist), and “Party Focused” (Party Soldier and Party Builder). The results from these models are presented in Table 1, comparing “Party Focused” to “District Focused” and “Policy Focused,” respectively. Columns 1 and 2 use the binary measure of local roots as the independent variable, and columns 3 and 4 use the logged measure.

Table 1. Legislator Type and Local Roots: Party versus Policy and District Focus, 1989–2008

Note: Standard errors in parentheses. * $ p<0.10 $ , ** $ p<0.05 $ .

As the table clearly underscores, local-roots legislators are significantly least likely to be classified as Party Focused based on their behaviors in office. Indeed, in comparison to both Policy and District Focus, local-roots legislators exhibit a lower propensity to adopt a Party Focus. Further, this result persists regardless of which operationalization of local roots we use. Same Party Presidential Vote Share also stands out as a large, negative, and significant predictor of District Focus, indicating that electoral safety is associated with legislators being more likely to adopt a Party Focus in comparison. However, our empirical strategy does not account for potential confounders on this control variable, so we interpret it with caution (Keele, Stevenson, and Elwert 2020).

Interestingly, our results also indicate that local-roots legislators are significantly more likely to adopt a District Focus over a Policy Focus. As we discuss in the results and discussion to come, this constitutes an interesting representational trade-off that voters may make when confronted with local versus nonlocal candidates. On the one hand, local-roots legislators keep the district first in their behaviors, demonstrating a propensity for District Focus above all other types. However, they also may forgo influence over major national policy, and within the party—which controls much of Congress’s substantive and political outputs today.

Staff Allocation and Local Roots

What, then, do these results mean for local-roots legislators with respect to specific behaviors of importance to legislative scholars? To capture the priorities of local and nonlocal legislators, we first investigate an important “investment” decision (Salisbury and Shepsle 1981) that members of Congress make in office: choices over hiring and assigning staff for constituency service functions. Using staff allocation data from Crosson et al. (2021), we examine the percentage of a legislator’s office that is made up of constituency staffers, from 1993 to 2014. Over this time period, the average member allocates 38.5% of their staff toward constituency service. ²² $ {}^{,} $ ²³

To examine whether staff variation is explained by changes in the preponderance of local-roots representatives, Table 2 presents regressions of constituency-service staff size on the same measures of local roots from above. The coefficients on both measures of local roots are all in the expected direction, although only the coefficient on the logged distance measure reaches conventional levels of statistical significance.

Table 2. Local Roots and Percentage of Staff That Focus on Constituency Service, 1993–2014

Note: Estimates are from OLS regressions with district-clustered standard errors in parentheses. The unit of observation is the legislator-Congress. **p<0.05, *p<0.10.

When all variables are set at their median values, moving from the 25th to the 75th percentile of log(Miles Born from the District + 1) is associated with about a 1.77 percentage point decrease in the proportion of staffers dedicated toward constituency service (from a baseline of about 40%). Although the magnitude of this effect may seem small, its size is in part due to the large amount of variation being controlled for with district fixed effects. Further, the maximum staff size in the House of Representatives is currently set at 18 full-time workers. As Crosson et al. (2021) underscore with respect to legislative staff, the small size of House offices (particularly in relation to Senate offices) forces members of Congress into acute trade-offs when building their personal staff. These trade-offs have grown even more difficult in recent years, due to Republicans’ dramatic cuts to Member Representational Allowances (MRAs) in 2010. These results suggest that, in spite of such budget cuts, local-roots legislators nevertheless maintain their propensity for hiring more constituency-service staffers. In fact, when this analysis is broken down by decade (as shown in Supplementary Table G.1), the results appear to be concentrated in the time period after the 2000s, although all coefficients remain appropriately signed. ²⁴

In addition to these results, we again underscore that a focus on constituency service is not costless in terms of the activities a member is able to pursue (Hall 1998). Indeed, local legislators may be “locked” into a more casework-heavy style of representation (cf. Harden 2015; McCrain 2021), at the expense of policy specialization. Supplementary Table I.1 underscores such consequences: as noted above, both measures of local roots indicate that legislators born in and around the district adopt a District Focus at a higher probability than they do a Policy Focus.

Local Roots and Party Unity on Roll Calls

Whether measured through staff investments or broader “style” categorizations, our results indicate that local-roots legislators behave in a localized manner consistent with a perceived role as a descriptive representative. But while local focus constitutes an important component of the in-office representation, such a focus likely also affects the substantive policy representation these legislators provide. Local representatives may focus less on policy overall, but does the less-partisan nature of their overall behavior redound to the character of the substantive activities that they do pursue?

As Fenno argues, local-roots legislators are likely to enjoy increased “leeway” from their constituencies. However, it is less clear what they will do with such leeway, particularly with respect to roll-call voting and bill writing. We contend that, as descriptive representatives, local-roots legislators will feel an obligation to act in a more bipartisan fashion, representing a larger contingent of the constituency. However, it is also possible legislators could use leeway to act in a more partisan fashion, particularly in a hyper-partisan era in American politics. In the following two sections, we put these expectations to the test.

We first analyze the impact of birthplace on party unity in roll-call voting. Ideally, we would capture partisan roll-call loyalty using a measure based on party leaders’ definition of loyalty (Meinke 2024). Absent such a measure for our time period of study, we instead use standard party unity scores, which Meinke (2024) shows are reasonably well-correlated with his more fine-tuned measure. We pair our data with covariates from Volden and Wiseman (2020), described above. Doing so sets the time frame of this analysis from the 93rd to the 116th Congress, or 1973 to 2020. Here, we regress party unity scores on our two measures of local roots, using an OLS regression with both district and congressional fixed effects.Table 3 presents the results, with column 1 using the binary measure of local roots as the independent variable and column 2 using the logged measure. Both measures are signed as expected, negative on the binary measure and positive on the distance measure, indicating that legislators with local roots vote in line with the party less. They also reach some level of statistical significance (though column 1 reaches only at p < 0.1). Still, this is a quite modest association. Thus, overall, it does not appear that local legislators vote differently in a meaningful way than their nonlocal counterparts, in spite of their propensity toward less partisan legislative styles.

Table 3. Member Birth Place and Party Unity, 1973–2020

Note: Estimates are from OLS regressions with district-clustered standard errors in parentheses. The unit of observation is the legislator-Congress. **p<0.05, *p<0.10.

Generally speaking, this is not consistent with either of the aforementioned usages of representational leeway. However, given that legislators could conceivably use leeway to behave in either a partisan or bipartisan direction, it is possible that the above results mask important heterogeneity. To more fully examine this possibility, we split our sample into legislators who we posit are more ideologically extreme and less ideologically extreme than the districts they represent, and then rerun our analyses within these groups. While we detail this process and full results at greater length in Appendix J of the Supplementary Material, our approach compares legislators’ actual CFscores (from Bonica 2019) to their predicted CFscore based on the partisanship of their district. Regardless of whether we examine more-extreme or more-moderate legislators, our results remain unchanged. That is, local roots do not appear to be especially associated with party unity in roll-call voting.

This lack of association persists even after we explored other potential sources of heterogeneity. Given that polarization increased over the time of this analysis, it is possible that small or null results are driven by lower levels of available variation in the latter portion of our time period. However, as our decade-specific results show in Supplementary Table G.2, although almost all of the coefficients remain appropriately signed, log(Miles Born from District +1) only reaches traditional levels of statistical significance in the 1970s. Although these results are consistent with those presented in Table 3, they do not indicate the presence of meaningful patterns over time. Second, it is possible that our main results mask heterogeneity by district safety. While all of our analyses control for Same Party Presidential Vote Share, legislators in safe districts may be uniquely suited to behave individualistic ways. However, as Supplementary Table J.3 shows, the only coefficient to reach significance at p < 0.1 occurs in the regression subset to unsafe districts. ²⁵ Thus, there does not appear to be a concentrated effect among safe members.

Bipartisanship among Cosponsors

In some ways, the small associations between local roots and party unity make sense: given that party leaders aggressively avoid roll-call votes that fracture their caucus, legislators sometimes face few opportunities to break from the party. However, roll-call voting is by no means the only important policymaking behavior: sponsorship and cosponsorships are both core to a representative’s legislative duties. In addition to being a key activity for members of Congress, cosponsorship “happens before efforts by leaders or influential members to persuade members to change their positions and before any potential source of selection bias, including agenda control, implying that it fully captures the underlying issue space and might be a better indicator of ideology than floor votes” (Desposato, Kearney, and Crisp 2011, 532). As a result, cosponsorship is a particularly valuable measure of partisanship, as individual legislators have significantly more control over this behavior.

As we have argued previously, inasmuch as local-roots legislators think of themselves as descriptive representatives, they may feel an obligation toward behaving in an especially bipartisan manner with respect to bill cosponsorship and the attraction of cosponsors. However, legislators with local roots may also be especially well-positioned to transcend partisan animus in cosponsorship for additional, practical reasons. First, their local connections may allow them to more credibly reach out to out-partisans within their own state or region. Second, given the above results that legislators with local roots appear to focus more on the district and less on policy and party related activities, their legislative work may be less controversial to members of the out-party. Thus, we ask: to what extent are local legislators’ sponsorship activities more or less bipartisan than their nonlocal counterparts’?

To examine this question, we generated a measure of bipartisanship that captures how much a legislator’s sponsored bills attract both in-partisans and out-partisans. We build our measure, which is substantively similar to the Lugar Center’s “Bipartisanship Index” ²⁶ and the measure used by Volden and Wiseman (2016), by drawing cosponsorship and sponsorship information from several sources. The first source comes from Fowler’s (2006) study of cosponsorship in Congress. For more recent congresses, however, we compiled bill-level data from ProPublica’s Bulk Data on Congressional Bills. ²⁷ After compiling this information, we then tabulated the number of Republican and Democratic cosponsors that a legislator was able to attract to her sponsored bills. Using this information, we constructed a ratio of in-party to out-party cosponsors, such that higher levels of the ratio signify more “partisan” coalition-building strategies.

With this variable, we again estimated the OLS models from our analysis of party unity, including the congress and district-level fixed effects. ²⁸ We present the results of these regressions for both the binary and continuous measure of birthplace locality in Table 4. Here again, we also run a series of supplemental analyses in Appendix J of the Supplementary Material designed to test our descriptive-representation expectations based on differential usage of local-roots leeway. As with our analysis of party unity, we again split legislators into “more moderate” and “more extreme” than excepted subgroups, based on district partisanship and presidential vote share. We then rerun the analyses found in Table 4 on these subgroups. As we summarize in the Supplementary Material, these results are not consistent with a differential response to leeway based on candidates’ pre-existing preferences.

Table 4. Member Birth Place and Cosponsorship Attraction, 1973–2020

Note: Estimates are from OLS regressions with district-clustered standard errors in parentheses. The unit of observation is the legislator-Congress. **p<0.05, *p<0.10.

Here, the results are consistent with the less-partisan nature of our earlier legislative style findings. Indeed, regardless of whether we adopt the binary or continuous measure of local roots, we find that local-roots legislators are significantly less likely to attract cosponsorships that are solely from their own party. Moreover, these results do not appear to be driven solely by a legislator’s ideology: no matter whether we control for the partisan tilt of the district, the roll-call extremity of the legislator, ²⁹ or both (the latter two of which we display in Supplementary Tables K.2 and K.3), local legislators consistently attract a greater proportion of out-party cosponsors than do their nonlocal counterparts. What does this result mean substantively? Holding all control variables at their medians, moving from the 25th to the 75th percentile in log(Miles Born from the District + 1) is associated with a 2.44 percentage point increase in in-party cosponsor attraction. Once again, this result points to a behavioral pattern of local-roots members acting in a less partisan fashion.

While the results do not appear to be driven by a legislator’s ideology, there once again still may be heterogeneity by district safety and time period. To investigate these possibilities, we rerun the results subset by whether or not the legislator is in a safe or unsafe district. The results from this analysis are presented in Supplementary Table J.4. Interestingly, although all coefficients are signed as expected, the results appear to be concentrated among legislators from safe districts, or districts where the incumbent received at least 57.5% in the previous presidential election. This indicates that local legislators are more likely to interact with the opposing party when they are already electorally comfortable. This is particularly noteworthy given that safe districts are simultaneously also the most partisan-homogeneous. Yet especially in these districts, we observe the most bipartisan behavior. Second, we break down the analysis by decade in Supplementary Table G.3. While almost all coefficients are signed as expected, the effects are all concentrated in the pre-2000s.

These findings have noteworthy implications for representation in the modern Congress. In particular, they suggest that local-roots legislators are uniquely well positioned to transcend nationalization. For example, the aforementioned Representative Katko stated the following in his press release announcing his bipartisan score from the Lugar Center: “I have consistently worked with both Democrats and Republicans to better Central New York and our nation. Doing so has yielded significant results, passing meaningful legislation that protects drinking water, invests in infrastructure, addresses the opioid epidemic, and strengthens our national security.” ³⁰ By relying on shared conceptions of place-based identity, local legislators are able to credibly claim they are working with the other party in order to do what is best for constituents, as opposed to betraying their own party. However, the results subset by decade indicate that this credibility may be fading over time; rising levels of partisan polarization may simply make such displays of bipartisanship less feasible in the modern Congress.

What about the District?

To this point, we have examined geographic descriptive representation largely from the legislator’s point of view, or the “supply” side of representation. Across several representational behaviors, we find evidence consistent with the idea that local-roots legislators do behave as though they view themselves as descriptive representatives. However, representation obviously involves linkages between a representative and her constituency, and different constituencies are apt to “demand” different styles of representation (Fenno 1978; Harden 2015; McCrain 2021). Lacking direct measures of demand for locally rooted legislators and different sorts of representation, we sought to hold such demand-side variation constant in the preceding analysis by including district-level fixed effects. However, if the claim of local-roots-as-descriptive-representation holds, it should be the case that local legislators are especially likely to behave as descriptive representatives when they perceive that constituents demand localized descriptive representation.

What kinds of constituents are likely to demand localized representation? Using a conjoint experiment, Vivyan and Wagner (2016, 85) provide evidence that constituent preferences over legislator resource allocation may be driven by “the degree to which voters themselves have a local or national/cosmopolitan outlook.” They explain that this orientation “captures the extent to which individuals care about and pay attention to developments in their local area over and above national and international developments (Jennings 1967, 293). It reflects an individual’s degree of attachment, sense of belonging and general affect towards his or her local area…” Essentially, the more a constituent cares about their local community in comparison to the nation as a whole, the stronger their demand for the legislator to focus on the district. Vivyan and Wagner (2016, 85) go on to explain that “Potential antecedents of local-cosmopolitan disposition include factors such as personality traits, education and social class.”

This argument is also supported by observational work at varying levels in the United States. According to research by Harden (2015) at the state legislative level and by McCrain (2021) at the federal level, factors like education, income, and geography can predict demand for district-focused representation. High-education individuals are more likely to migrate from their childhood homes—and move in general—compared to those with lower education, who often remain geographically rooted for generations. Indeed, even if the college-educated return home, their time in school frequently has drawn them away from home and introduced them to other places, geographies, and people with whom they may identify. Inasmuch as this rootedness is correlated with a developed sense of local or geographic identity, one might expect education, then, to mediate the relationship between a legislator’s local roots and the style of representation she provides to the district.

Finally, strong place-based identities may also have an ideological dimension. For example, in her study of the politics of resentment, Cramer (2016, 55) argues that rural consciousness is composed of perceptions of power, perceptions of resources, and perceptions of values and lifestyles. Moreover, as Jacobs and Munis (2022) find, voters’ feelings toward particular kinds of geographies can predict their eventual vote choices, particularly in the case of rural areas. Thus, it may be the case that district demand for locally rooted legislators is based on an underlying value orientation.

Unfortunately, as Supplementary Table L.1 underscores, an extreme lack of within-district variation in district-specific characteristics precluded our inclusion of such covariates in our district fixed-effects models. ³¹ Indeed, among the district-specific characteristics we examine, an average of only 1.8% of the variation in these factors is within-district—compared to an average of 14.3% for the eight variables included in the previous analyses. Consequently, when we simply include these variables as covariates in our regressions, they exhibit very few results and large standard errors. ³² However, we do explore district-level dynamics in alternative ways. First, we estimate a series of models without district fixed effects—but including a series of district-level control variables. Although the results provide some baseline insights into possible confounds between local roots and our three dependent variables, the associations are primarily null, so we present them in the Supplementary Material (Tables M.1–M.3). ³³

Second, we examine whether local-representation demand enhances the trends we have uncovered above. Even maintaining district fixed-effects, we are able to examine whether local-roots legislators behave distinctly in districts that may value localized descriptive representation the most. While a full theory and empirical exploration of constituent demand for local representation lies beyond the scope of the work here, we nevertheless build upon previous work on representation to provide a series of preliminary results in Tables 5 and 6. In the first of these tables, we regress all three of our primary dependent variables—staff allocation, cosponsorship, and party unity—on subsets of our data, based on a district’s education levels. In the second, we endeavor to measure physical ties to geography by running the same models, but with percentage of residents living in a different state in the previous year as our mediating variable. In both tables, Panel A depicts districts that are below median in percentage with a bachelor’s degree over the course of our dependent variable (Table 5) and previous out-of-state residency (Table 6), while Panel B presents districts that are above median in those variables. For both variables (though to a greater extent for education than previous residence), the results are consistent with expectations. In fact, we find some evidence that our main results are driven by particularly noteworthy associations between local roots and legislative behavior in below-median districts. Indeed, effect sizes were as much as 50% and 300% larger in cosponsor attraction and staff allocation, respectively, in below-median districts compared to our full-sample results.

Table 5. Member Birth Place and Differential Findings by % Bachelor’s Degree

Note: Estimates are from OLS regressions with district-clustered standard errors in parentheses. The unit of observation is the legislator-Congress. **p<0.05, *p<0.10.

Table 6. Member Birth Place and Differential Findings by Interstate Constituent Mobility

Note: Estimates are from OLS regressions with district-clustered standard errors in parentheses. The unit of observation is the legislator-Congress. Note that the data begin in 2007. **p<0.05, *p<0.10.

Overall, while we take these results to be at least consistent with our belief that local roots constitutes a dimension of descriptive representation, they are certainly not dispositive. Future research should further examine the demand for localized and geographic representation. In addition to constituents’ physical ties to their places of residence, it seems plausible that other factors—including psychological, sociological, and even ideological factors examined in relation to other forms of descriptive representation—could influence demand (and, consequently, supply) of localized descriptive representation. Indeed, our hope is that future research will explicitly connect Hunt’s work and our own to constituent-level research on place-based identity and resentment (e.g., Jacobs and Munis 2022).

DISCUSSION AND CONCLUSION

A legislator’s connection to their district is central to representation, particularly in a polity with geographic representation like the United States. Nevertheless, a legislator’s relationship with her party also lies at the heart of modern congressional behavior. As a result, members of Congress face conflicts between district and party that frame campaign messaging to research allocation to policymaking behavior and beyond. Yet, particularly in the modern Congress, members also face sharp conflicts between localism and the growing nationalization of American politics (e.g., Grumbach 2022). In this article, we analyze how legislators’ connections to their district via birthplace shape their legislative behavior in general—and how those connections may influence their relationship to the party specifically.

Generally speaking, we find that local-roots legislators behave in a manner consistent with descriptive representation, systematically focusing on the district more than nonlocal-roots legislators. Although local-roots legislators do not exhibit large differences in their roll-call records, we find that they build cosponsorship coalitions in a more bipartisan fashion and that they are less likely to adopt party “soldier” or “builder” legislative styles. In service of this style, we also find that these legislators commit more staff to constituency-service function, consistent with Hunt’s depiction of local-roots legislators’ valence appeals in campaigns. Thus, rather than enabling legislators to pursue national policy influence, local connections appear to encourage legislators to maintain a focus on district advocacy. However, we also find evidence that these relationships are weakening, both in strength and in the raw frequency of local-roots connections.

What might these findings mean for the quality of representation that constituents enjoy when they elect a candidate with strong ties to the district? First, we believe that there are mixed consequences for dyadic representation. On the one hand, such representation may enable a legislator to act in a manner more broadly appealing within the district, rather than maintaining stringent allegiance to a single party. On the other, the district-heavy focus of local-roots legislators could prevent them from influencing the most important national policy issues of the day. It is not clear on its face which type of representation is “best” for constituents, and it is beyond our scope here to provide arguments for one type of representation over the other. Nevertheless, we do contend that changes over the past 50 years could carry important implications for how scholars understand fundamental concepts such as homestyle.

Second, these findings also have consequences for the changing nature of collective representation in the United States. Our data indicate that localism has declined in modern times, with about 43% of the 116th Congress (2020) being born fewer than 10 miles away from their district in comparison to about 50% in the 106th Congress (2000). Based on our results, this indicates that modern members may be more policy- and party-oriented than their predecessors, potentially altering the nature of policy outputs and national discourse (Grimmer 2013, 624). Further, it also indicates that the pipeline to Congress may be changing, as the diversity of members increases and the potential pathways to election expand. Thus, while Congress continues to be remarkably local, we believe that these recent trends could indicate that candidates may now be running based on party networks or ties to national interest groups. We hope that future research will investigate this possibility.

Given our findings and the questions they generate, future research should also continue to investigate the relationship between local roots and other important member activities. For instance, one could investigate local roots and the procurement of federal grants. Additionally, it would be useful to understand whether local legislators are more likely to work with other legislators from their communities—perhaps helping to explain their greater tendency toward bipartisanship in their cosponsor attraction. Finally, scholars could also investigate whether similar relationships are demonstrated in the Senate. Although our alternative results (Supplementary Table A.1) indicate that state roots display similar impacts on legislative behaviors as local roots, the larger and more heterogeneous constituencies may lead to important differences in Senator and House member behavior. These questions represent only a small number of consequences of local roots in office, a feature of representation quite intentionally incorporated in the U.S. constitutional design but that, as recent scholarship has forcefully underscored, has gone under-explored in the study of Congress.