Increasingly members of the U.S. House of Representatives rely upon funds provided by campaign contributors who live outside of their districts and states (Baker 2016a; Gimpel et al. 2008). More than half of the House depends upon donor contributions for half of their campaign revenue and a quarter of these members are highly dependent upon out-of-the-district contributions (Baker 2016a). And the share of members who are highly dependent is growing.Footnote 1 This trend is potentially problematic if it interferes with representation. Members of the House are intended to be the most locally focused of all federal officeholders, so any incentive to redirect their focus beyond their districts could detract from the representation they provide to constituents. And there is evidence that a greater dependency upon out-of-the-district funds leads dependent House members to be less ideologically responsive and more ideologically distant from their constituencies (Baker 2016a). In light of these trends, a better understanding of donors’ motivations is paramount in revealing the ways they ideologically pull members away from their districts. The donors who provide these contributions are thought to be seeking “surrogate representation” (Gimpel et al. 2008, p. 373; Mansbridge 2003)—that is, seeking representation from congressional members outside of their respective House districts rather than simply relying upon the representation provided by their House member.

Yet precisely why this set of donors chooses to contribute in this way remains unclear. Mansbridge (2003) theorizes that the utility of geographic representation has declined because it no longer reflects the voter’s “significant interests” (Mansbridge 2003, pp. 522–523). By giving to House members on a national basis, donors recover or gain additional representation on issues that perhaps do not have a geographic basis to them. In addition to policy representation (Barber et al. 2017; Francia et al. 2003), previous work suggests political donors may also be motivated by their ideological leanings and partisan ties to extend financial support to multiple members or candidates for Congress (Bramlett et al. 2011; Francia et al. 2005; Francia, et al. 2003; Rhodes et al. 2016; Baker 2016b).

This study examines the partisan and policy preferences that lead donors to contribute to House races nationwide. Using data from the Cooperative Congressional Election Studies (CCES) 2008–2014 in which respondents’ self-report whether they have made contributions to congressional candidates running outside of the donor’s home state, I uncover evidence that those giving outside of their states do so not only to recover partisan representation that they fail to receive from their own House member but to gain additional forms of representation. Both donors’ ideological identification and donors’ ties to their party directly increase their likelihood of giving to House races nationwide. Policy motivations across a variety of issues, such as stances on the national budget and climate change, also drive the choice to contribute, with conservative positions more frequently predicting out-of-state giving than liberal policy positions. Additionally, while these policy preferences motivate out of state contributing, I find they are not connected to contributing within the district—a result which suggests donors who give to House races out of state have distinct motivations.

Congressional Donors: Giving Patterns and Preferences

Although, on the whole, an understanding of the donor population’s demographics and preferences is quickly advancing in political science (see Francia et al. 2003, 2005; La Raja and Schaffner 2015; Hill and Huber 2017; Bramlett et al. 2011; Rhodes et al. 2016), more studies need to focus on the donors who contribute to U.S. House races nationwide. Outside influence from non-constituent donors is more likely to lead to distortions in representation in the U.S. House of Representatives. Indeed, previous work uncovers ideological distortions in the 2006–2010 congresses stemming from members responsiveness to House donors (Baker 2016a). Short 2-year terms along with rising election costs make House candidates even more susceptible to donors’ influence. Individual donors are also the largest source of campaign money, particularly for non-incumbent House candidates, further underlining the importance of studying donors to House races (Bonica et al. 2013). While congressional donors received ample attention in Francia et al. (2003) seminal study, their survey and supporting interviews were conducted in 1996. As congressional polarization increases, particularly in the House (Bonica et al. 2013), it is reasonable to assume that new issues or partisan motives might be driving donors to contribute.

Nonetheless, Francia et al.’s (2003) study lays important groundwork for expectations about congressional donors’ current preferences. They find most congressional donors are conservative on “economic, social welfare, and foreign-policy and defense issues, but moderate to liberal on social issues” (62). However, partisan donors generally hold opposing positions on cutting taxes, environmental protection, and poverty reduction (60–63). Francia et al. (2003) also note “consistent conservative Republicans” comprise “30 percent of all donors and more than 60 percent of all Republican donors” (66) whereas “traditional Democrats” comprise “22 percent of all donors and more than two-thirds of” Democratic donors (66). These findings denote the importance of examining both donors’ economic and social policy preferences and they support the idea that congressional donors have partisan motivations for contributing.

More recent work by Gimpel et al. (2008) examines the population of House donors of interest here, but their study does not directly address surrogate seekers’ policy or partisan motivations for giving. They uncover evidence that inflows of non-constituent funds are strongly tied to the competitiveness of the race and the ideological extremism of the member, but they find more limited evidence of access-seeking giving patterns. They attribute a major role to party networks in channeling funds to important races and suggest “strategic partisanship” drives the contributing choices of the donors who contribute out of their home districts (Gimpel et al. 2008, p. 390). Their findings suggest surrogate seekers are able to target their contributions for greatest impact. Using the CCES data, Rhodes et al. (2016) also examine the self-reported giving behavior of out of state congressional donors—as one of four classes of donors they identify—but, like Gimpel et al. (2008), they do not explore their policy preferences. That said, their study helps establish that “strategic investors”—their name for congressional donors who give out of state—are distinct from other classes of donors, including party and interest group donors. They find “strategic investors” are more likely to be medium to large donors, to be ideologically extreme, and to be very politically engaged. Importantly, these findings suggest surrogate seekers have the capacity to identify the candidates who align with their partisan and policy preferences.

Three additional studies focus on the motivations of donors who contribute to Senate races and can provide some insight about what to expect from potential surrogate seekers in House races. Drawing upon an original survey dataset from 2012, Barber (2016a) constructs ideal point estimates of Senate donors’ ideological positions. These estimates are based upon their responses to various policy questions, although Barber (2016a) does not report the responses or use separate policy responses in predictive models. He finds ideologically extreme donors rate the recipient candidate’s ideology, their objections to the opponent’s ideology, and their likelihood of “affecting the outcome of the race in favor of their preferred candidate” as important reasons to contribute (153). Using the same ideal point approach from the same survey in a second study and in combination with DW-NOMINATE scores, Barber (2016b) examines the ideological congruence between donors and senators. He finds senators are ideologically closer to donors than to the partisans or voters in their states. While precise causality cannot be established, the results suggest senators are responsive to out of state financial constituencies as Barber shows that is the source of the bulk of their campaign money. In a final study using the same survey data, Barber et al. (2017) explicitly examine policy agreement between out of state donors and senators’ voting records. They find policy agreement is an extremely significant factor in predicting out of state donors’ contributions—this provides reasonable grounds to assume policy preferences might also predict contributions from House donors. However, they also find the donor’s ideological extremity decreases the likelihood of contributing to a Senate candidate in general as well as to a Senate candidate outside of the donor’s state. This suggests Senate surrogate seekers might be distinct in some ways from House surrogate seekers as other studies find House contributions are tied to the ideological extremity of the donor, the candidate, or both (Ensley 2009; Hill and Huber 2017; Rhodes et al. 2016).

What Might Motivate Surrogate Seeking?

Political donors who make contributions to congressional campaigns are known to be strategic givers in that they give to multiple political entities and they are thought to be loyal and ideologically driven partisans who are highly politically engaged (Rhodes et al. 2016; Francia et al. 2003). There is also evidence that party donors in particular are likely to make contributions to congressional candidates who represent their party’s best chances of gaining or retaining seats in the House (Baker 2016b). Thus, previous work lends support to the hypothesis that congressional donors may give outside of their district or state for a variety of partisan reasons. First, it might be the case that the donor’s partisanship does not match the partisanship of the member of Congress who represents the donor’s district. As a consequence, the donor contributes to another congressional candidate in order to gain substitute representation. It could also be the case that donors might support congressional candidates outside of their home-states and districts to gain additional representation.

Specifically, as implied by Mansbridge (2003), the donor would gain additional partisan representation stemming from their party’s legislative and electoral successes leading to majority control of the government. The distinction between these two goals is important because substitute representation is less likely to introduce major political inequalities than the purchase of additional representation, which expands the power of the individual donor beyond a dyadic relationship with a given member of Congress. In other words, the donor is no longer simply seeking to recover a dyadic relationship with a different representative in their party, the donor wants influence beyond a one to one relationship which all partisan voters receive if their House member shares their partisanship. Of course, it might also be the case that a given donor seeks both substitute and additional representation simultaneously. To test whether donors are seeking, at a minimum, substitute partisan representation I compare the donors’ partisan identification with their incumbents’ to see if a mismatch between the two leads the donor to contribute out of state. To test whether the donor has partisan motivations that might more closely approximate the quest for additional representation, the relationship between contributing to the party’s national committees and contributing to House races nationwide is also examined.

Similarly, donors might also give money to congressional candidates outside of their home states or districts in order to gain either substitute or additional policy representation in specific issue areas. Policies that are more solidly in the domain of the federal government rather than state governments should be priorities for those seeking to make contributions to congressional candidates outside of their districts and states. Fortunately, the Cooperative Congressional Election Studies (CCES) surveys provide several examples of federally focused policies, such as questions on gun control, the environment, and affirmative action. The assumption is that the person contributing outside of his or her state and/or district hopes to have a broader impact on national policies. Second, evidence of surrogate seeking might also be evident if donors who are concerned about the national budget appear to be targeting House races rather than Senate races. The CCES question on the national budget is utilized to test institutional-specific surrogate seeking. As a final way to determine whether donors who contribute out of state are seeking additional policy representation, they are compared to those who only contribute to House races in their home districts. If the same national policies do not significantly predict in-district versus out-of-state giving, then at a minimum this finding would suggest these two groups of donors have different motivations for giving and lends support to the idea that out of state donors want additional influence of some kind.

Unfortunately, the CCES survey does not ask respondents whether they contributed to House races outside of their home districts. Rather, the survey asks whether they gave out of state. This methodological issue helpfully raises one more layer of variation to consider: out of state versus in-state surrogate seeking. In a state with one congressional district, an out of state contribution is the same as an out of district contribution. As the number of congressional districts in the state increases, the donor’s potential number of in-state surrogate options also increases. Both in theory and in practice, in-state surrogate seeking should lead to fewer distortions in representation than out of state surrogate seeking. House candidates running in the same state are more likely to be similar to one in some ways because they are constrained by several constant variables, such as key industries that fuel the state’s economy, and geographic-specific problems, such as the presence of a coast-line and the accompanying threat of hurricanes, that members of both parties must care about to be successful representatives in the state.Footnote 2 Furthermore, donors who give within the state are still state constituents and insofar as House delegations from the same state represent the state’s needs, the donor is entitled to at least a share of this state-based representation even if the donor should not gain added influence over district-specific representation provided to other districts in the state where the donor does not live. Thus, the dependent variable which captures out of state giving is helpful in that it focuses attention on the type of surrogate seeking that is most likely to lead to distortions in representation. To control for in-state surrogate seeking, I include the number of congressional districts in all of the models described below.

Finally, there is good reason to believe that finding surrogate matches among all of the House candidates running in a given election cycle might be easier for conservative donors than liberal donors due to changes in party politics since the 1980s. As Grossman and Hopkins (2016) theorize: “The Democratic Party has long encouraged party leaders to assemble a policy agenda from the aggregated preferences of the party’s numerous constituencies, courting the mass electorate with a large assortment of concrete benefits favoring targeted populations” (3)—making the Democratic party a group-focused institution. Conversely, they argue the Republican party “can be most accurately characterized as the vehicle of an ideological movement.” Further, “most Republican voters—and nearly all of the party’s activists, financial supporters, candidates, and officeholders—identify as conservative and voice support for the abstract values of small government and American cultural traditionalism” (3). In their comparison of the two parties’ donor bases, Grossman and Hopkins (2016) find “in choosing which candidates to support, Republican donors were more likely to say that (conservative) ideology is always important; they were also much more likely to agree that donors are motivated by ideological goals” (115). In contrast, Democratic donors “are more interested than Republican donors in demographic representation and less concerned with ideological fidelity” (115). Similarly, La Raja and Schaffner (2015) argue “conservative donors are pulling the Republican Party further to the right by focusing their largesse on like-minded candidates and groups, while ignoring the party organization, which tends to invest in more moderate candidates” (51).

In effect, these two studies suggest that for conservative donors the search for surrogate partisan and policy representation is more likely to be satisfied by the same set of House candidates whereas for liberal donors the quest for partisan and policy surrogate representation might: require a larger number of candidates to represent the different positions and/or group-orientations held by the donor; entail a different set of contribution strategies; and/or be satisfied by contributing to the party organization as much as giving to candidates because liberals see less of a distinction between their set of candidates and the party organization’s ideological orientation (La Raja and Schaffner 2015, pp. 51–53). Conversely, conservative policy preferences should be more readily satisfied by contributing to House candidates out of state than liberal policy preferences simply because it is easier for conservatives to find a suitable surrogate House representative than for liberal donors to do so. Responses to national issue questions on the CCES are expected to reveal these differences in the likelihood of contributing to House races out of state. Conservative donors may also have benefited from the emergence of the Tea Party. Candidates who identified with or were endorsed by the Tea Party are more easily identified by donors as true conservatives. In effect, the Tea Party label lowers the cost of identifying surrogates for conservative donors. To test for this effect, I examine the relationship between Tea Party support and out of state contributing to House races.

Data and Methods

Data from the Cooperative Congressional Election Studies (CCES) of 2008, 2010, 2012, and 2014 are used in a series of logistic models.Footnote 3 The dependent variable in the first set of models is a dummy variable indicating whether the respondent self-reported making a contribution to a House candidate outside of her state whereas in the second set of models the dummy indicates whether the donor only made a contribution to House candidates within his or her district. The entire series of logistic regression models that follow utilize the subpopulation of political donors in the survey sample,Footnote 4 in addition to the application of the usual survey weights.

Three partisan models are utilized. The first model uses a dummy variable to identify when the respondent’s self-reported partisanship does not match the respondent’s perception of the partisan identification of the incumbent member who represents the respondent’s district as a Democrat or Republican (MisMatch). Responses indicating the member of Congress is an Independent and those in which the respondent was not sure or never heard of their incumbent House member are excluded. Respondents who identified as either “strongly Democratic” or “not very strongly Democratic” are coded as Democrats and respondents who identified as either “strongly Republican” or “not very strongly Republican” are coded as Republicans. Self-reported leaners and independents are coded as the reference category. A second partisan model tests whether respondents are more likely to give to House candidates outside of the state if they report having made a contribution to one of their party’s committees. A third model includes an interaction term between this variable and the partisanship of the donor to see if Democratic party donors are less likely to contribute to out of state House races. All three partisan models utilize data from the CCES surveys 2008–2014.

Six different policy models are employed using data from different CCES surveys. The policy variables for climate change, gun control, abortion, and the national budget have been specified as factor variables that correspond to the liberal or conservative response for each question with moderate responses coded as the reference category. This strategy was followed partly because the response options for these questions are not true ordinal scales. For climate change, respondents who said “global climate change has been established as a serious problem, and immediate action is necessary” are coded 2 and climate change skeptics are coded 3 if they responded that “global climate change is not occurring; this not a real issue,” 1 otherwise. On the issue of gun control, those who responded that “laws covering the sale of firearms should be made “more strict” are coded 2, those favoring fewer restrictions are coded 3, and those responding the laws should be “kept as they are” are coded 1. Both of these questions only appeared in the 2010 and 2012 CCESs. On the issue of abortion those who responded that “by law, abortion should never be permitted” are coded 3 for being pro-life and pro-choice respondents are coded 2 for responding: “by law, a woman should always be able to obtain an abortion as a matter of personal choice.” All other responses to this question are coded 1. The response options to a question about support for affirmative action programs are on an ordinal scale ranging from “strongly support” to “strongly oppose”—as a consequence, this variable is not transformed. Both the questions on abortion and affirmative action are posed in the 2008–2012 CCESs.Footnote 5

For the budgetary questions, respondents are asked their most preferred solution to balancing the federal budget. Responses indicating domestic spending cuts are the most preferred option are coded as 3 for the conservative position and those responding defense spending cuts are the most preferred option are coded as 2. The reference category for this variable corresponds to those who favored raising taxes as their most preferred option and are coded 1. This question appeared in 2008–2012 surveys whereas the question about Tea Party support appears in 2010–2014 surveys. The question asks respondents “what is your view of the Tea Party” with five categories ranging from “very positive” to “very negative.” I flip this scale so that positive and higher values represent more positive views of the Tea Party.

In addition to the primary explanatory variables above, I also employ a number of control variables in all of the models. One of the strongest predictors of whether someone is able to make a campaign contribution is household income. The CCES of 2008 and 2010 use the same twelve-point family income scale. However, the 2012 and 2014 surveys featured an extended version of the scale with five additional income categories at the top of the scale. I recoded these scales to mirror the 2008/2010 12-point scale as closely as possible so that the highest income category represents households making $150,000 or more (Family Income). Another important predictor of political contributing is political engagement. I utilized a two-parameter logistic item response model to create my measure of political engagement. I include four component terms indicating whether the respondent: (1) attended a political meeting in the past year or not; (2) volunteered for a campaign over the past year; (3) displayed a political sign during the past year; and (4) is interested in the news most of the time (Political Engagement).Footnote 6 The higher the respondent’s latent score the more politically engaged the respondent is. In addition to these two variables, I also include a dummy variable indicating whether the respondent is a woman and another indicating whether the respondent is a minority. Both women and minorities might be more likely to make contributions to House members outside their districts in order to gain descriptive representation in Congress (Female, Minority). I include the five-point ideology scale that ranges from “very liberal” to “very conservative” in all of the models (Ideology). A series of goodness of fit tests revealed this variable has curvilinear effects on the dependent variable. Thus, a squared version of the variable is included in all of the models. I also added a count of the number of congressional districts in the donor’s state—the logic of this control variable is explained above. Finally, I add the Democratic presidential percentage share of the two-party vote in the district as a measure of district ideology (Democratic Vote Share).

Predictors of Outside Giving

Before examining specific partisan and policy motivations for giving to out of state congressional races, I ran a baseline logistic model without any of the primary explanatory variables of interest to show the direct effect of ideology on this particular type of contributing behavior.Footnote 7 Very liberal and very conservative donors in the sample have roughly the same overall probability of giving outside of the state, respectively 13% and 14% (see Fig. 1). Very liberal donors are 4% points more likely than moderate donors to contribute out of state and, thus, have a 31% higher probability of giving, whereas very conservative donors are 5% points more likely than moderate donors to do so, and thus, have a 36% higher probability of giving than moderates (see Fig. 1). Clearly, ideology is an important factor driving political donors to contribute out of their home states.

Fig. 1
figure 1

Outside contributing by ideological position

Full size image

The results of the partisan logistic regressions can be found in Table 1 and the average marginal effects are displayed in Fig. 2 alongside those from the other models. The dummy variable represents a mismatch between the respondent’s self-reported partisanship and the respondent’s perception of the incumbent representative’s partisanship. When a mismatch arises, the average marginal effect of the probability of giving out of state is 2% points higher (see Fig. 2). However, this effect varies significantly by the ideology of the donor when adjusted predications at those representative values are examined for the subpopulation of donors in the sample. When represented by an incumbent of the opposite party, donors who are very liberal have a 7% point higher predicted probability of giving out of state than those who are moderate, meaning they are 44% more likely to contribute in this way than moderates. Those who are very conservative have a 6% point higher predicted probability of giving out of state than moderates, and, thus, are 40% more likely to contribute than moderates. Similarly, contributing to a party committee significantly increases the respondent’s average marginal probability of contributing to congressional candidates outside of the state by 4% points (see Fig. 2) but once again the respondent’s ideology is an important factor. Very liberal and very conservative party donors have a 6% point higher probability of giving out of state than moderate party donors and, respectively, a 5% point higher probability of giving out of state than the very liberal or very conservative respondents who did not report making contributions to political parties. Overall, very liberal and very conservative party donors are 29% more likely to give out of state than ideologues who are not party donors. Although, in general, party contributors are more likely to give out of state, there are differences depending upon the partisanship of the donor. The positive interaction term in Table 1 suggests Democratic donors who contributed to the party committees are slightly more likely to give to out of state House races than Republican donors who contributed to the party committees. They have an overall 14% point probability of giving versus a 12% point probability for Republicans.

Table 1 Partisan models
Full size table
Fig. 2
figure 2

Average marginal effects

Full size image

The results of the policy logistic regression models can be found in Table 2. Figure 2 shows the average marginal effects for each liberal and conservative response item to each policy question with the exception of affirmative action. What is immediately apparent is that conservative policy positions more strongly predict out of state giving to congressional races than liberal policy positions (see Fig. 2) with some variation in the magnitude of the difference when adjusted predictions at representative values are examined for the subpopulation of donors in the sample (see Fig. 3).Footnote 8 With respect to climate change, those who agree that climate change is a “serious problem” are 3% points more likely and those who agree that climate change is “not occurring” are 4% points more likely than those responding to the moderate response option, to make contributions to congressional candidates outside of the state (see Fig. 3). Thus, climate change believers have a 21% higher probability and climate change skeptics have a 27% higher probability of giving out of the state than those claiming “some action should be taken” or “more research” needs to be conducted. Once the donor’s ideology is considered, the differences magnify. Those who are very conservative and are skeptical about climate change have an overall 22% point probability of giving outside of the state whereas those who are very liberal and feel climate change is a major problem have an overall 16% point probability of giving out of state. Those who agree that “laws governing the sale of firearms should be made less strict” are 4% points more likely to give outside of the state than those who believe the laws should be “kept as they are”—thus, overall, they are 27% more likely than moderates to surrogate seek (see Fig. 3). The more liberal stance that gun laws should be “more strict” does not significantly impact the probability of contributing out of state (see Table 2). However, those responding as conservatives and very conservatives who favor less regulation respectively have a 14% point probability and 19% point probability of contributing out of state overall. Neither the pro-life position—abortion should “never be permitted”—nor the pro-choice position—abortion should “always be an option”—significantly predicted giving to House races outside of the state (see Table 2). However, the coefficient for the conservative position is positive whereas the coefficient for the liberal position is negative in keeping with the previous results (see Table 2, Fig. 2).

Table 2 Policy models
Full size table
Fig. 3
figure 3

Predicted probabilities using adjusted predictions at representative values

Full size image

After several model specifications and consultations with goodness of fit statistics, I find affirmative action has a curvilinear rather than a linear direct effect on the dependent variable (see Table 2). For those donors who “strongly support” affirmative action, the probability of giving outside of the state is 3% points higher than those who “strongly oppose” affirmative action and, as such, they are 23% more likely to contribute than those who “strongly oppose” affirmative action. Strong supporters are also 3% points more likely to contribute than those who simply “support” affirmative action. This policy question is the exception to the overall pattern. In this instance, the liberal position more strongly predicts out of state giving than the conservative position.

The last two policy questions respectively pertain to balancing the national budget and support for the Tea Party. The results are displayed in Table 3, the average marginal effects are presented in Fig. 2, and the predicted probabilities for the sample at representative values are displayed in Fig. 3. Those donors who cite defense spending cuts as their most preferred method of balancing the national budget are not significantly motivated to contribute out of the state whereas those who cite domestic spending cuts as their most preferred solution are significantly likely to contribute out of state (see Table 3, Fig. 3). Donors taking the more conservative stance have a 2% point higher probability of contributing out of state than those who prefer defense spending cuts (see Fig. 3). Overall, the percentage difference in their likelihood of giving is 18% higher than those selecting either of the other response categories. To test whether those donors are targeting House races nationwide because the House controls the budgetary appropriations process, I use responses to two CCES questions respectively asking whether the respondent made contributions to Senate races in their home state or in other states. Those who prefer domestic cuts are not significantly likely to make a contribution to either their own Senator or a Senate candidate outside of their state but they are more likely to give to House races nationwide (results not shown; p < 0.05). Thus, those preferring domestic cuts seem motivated to contribute in an institutional-specific way.Footnote 9

Table 3 Fiscal policy models
Full size table

The average marginal effect for each degree of Tea Party support is 2% points (see Fig. 2); however, the probability of giving varies greatly between the lowest end of the scale representing those with “very negative” views of the Tea Party versus those at the high end of the scale representing “very positive” views of the Tea Party (see Fig. 3, Table 3). The difference in the probability of giving between these two groups of donors is 8% points with those holding “very positive” views of the Tea Party having an overall 17% probability of giving out of the state—this represents a percentage difference of 47% in the overall probability of giving between those holding either extremely negative versus extremely positive views of the Tea Party in the sample (see Fig. 3).

Importantly, the district count control variable is both positive and significant in all of the models suggesting that the more congressional districts the state has, the more likely donors are to contribute out of state (see Tables 1, 2, 3). Although states with higher numbers of congressional districts have larger populations of donors—a feature that drives up the probability of instances of out of state giving—the significance of this variable could also be an indication that those giving out of state are also giving out of their districts. If likely out of state donors were led to contribute in-state, the probability of giving out of state should diminish with an increase in the number of districts. Put another way, if the district count coefficient was negative, this would imply that with more in-state surrogate options, surrogate seekers would be less likely to give out of state. However, the reverse relationship is uncovered.

To determine whether out of state donors differ from in-district contributors, I utilize the same set of policy models with a different dependent variable indicating that the donor only contributed to a House race within her home district (see Table 4). The first major difference between these groups of donors is that ideology is not a significant predictor for those who only contribute to congressional races within their districts. Additionally, none of the policy variables are significant and this finding does not vary by liberal or conservative responses to the various policy questions. Political engagement is a positive and significant predictor of in-district giving in all of the models (see Table 4). Clearly, although these donors are politically sophisticated, their reasons for contributing differ from those who contribute out of state. Specific policy motivations do not appear to be important drivers of in-district giving.

Table 4 Policy and fiscal models of in-district donors to house campaigns
Full size table

Seeking Additional Representation

The results suggest donors who decide to contribute to House candidates running in districts nationwide are seeking both substitute representation as well as additional representation. With respect to substitute representation, donors who do not share partisanship with their incumbent are more likely to contribute to House candidates in other states. The more ideological the donor is, the more likely a mismatch leads to an out of state contribution. Yet there is also evidence that the decision to give to congressional races around the country represents an effort by donors to gain additional influence beyond the dyadic representation provided by to them by their particular member of Congress. Donors who are already connected to a party network, as indicated by the fact that they contributed to a party committee, have a 29% higher chance of giving to House races out of state than similarly extreme non-party donors and they have a 59% higher probability of giving than moderate donors in the sample who did not contribute to a party committee. This finding is in keeping with previous research suggesting party donors are major supporters of congressional candidates and are responsive to party seat-maximization goals (see Baker 2016b; Francia et al. 2003; Rhodes et al. 2016, and as predicted by Mansbridge, 2003). This suggests donors desire the added influence that stems from their party’s majority control of the government. However, interestingly, the results suggest Democratic donors who contribute to their party organizations are slightly more likely to give to House candidates than Republican donors. As La Raja and Schaffner (2015) point out, this may relate to the fact that they see fewer differences between the ideological leanings of Democratic candidates and the Democratic party organizations. Finally, the fact that both sets of party donors give to multiple candidates is further evidence that they are seeking additional influence rather than simply seeking to replace the influence they lost as a consequence of being represented by someone who does not share their partisanship. Between 2006 and 2012, on average, Democratic party donors contributed to 4.9 House candidates and Republican party donors contributed to 3.5 House candidates (calculated by author using Bonica’s 2013b DIME data).

Moreover, the additional influence donors are seeking is not merely partisan influence but also policy-specific influence, although the degree to which policy stances motivate out of state giving varies by ideology and by issue. Two pieces of evidence suggest donors might be specifically seeking additional policy representation. First, conservative donors who favor domestic spending cuts contributed to House races outside of their states but they did not contribute to Senate races. This suggests they seek institutional-specific surrogate representation when it comes to the national budget. Second, policy stances did not significantly predict contributing to House races for donors who only contribute within their districts. This further suggests these two sets of contributors have different motivations and implies out of state contributors’ representative preferences are not satisfied in some way within their districts.

The results also suggest surrogate seeking may be more challenging for liberal donors than it is for conservative donors because of asymmetric polarization in the party system (La Raja and Schaffner 2015; Grossman and Hopkins 2016). Grossman and Hopkins (2016) argue “the Republican donor base is made up of ideological conservatives, whereas the Democratic donor base is a collection of policy-issue activists, often motivated by social identity group concerns” (115)—while they argue this is especially true for donors to the party organizations, they also present evidence that Republican campaign donors in general are more concerned about ideological purity than Democratic campaign donors. Thus, one reason conservative policy stances predict out of state contributions to House candidates more frequently than liberal positions is that conservative donors have an easier time finding surrogates who represent both their ideological stances and policy preferences. This was the case for those denying that climate change is occurring, those who are against stricter gun control laws, and those who support domestic spending cuts as their most preferred method of balancing the federal budget. In contrast, the liberal positions on gun control, abortion, and the federal budget do not significantly predict out of state contributing at all and the liberal position on affirmative action was the only significant instance in which liberal respondents have a higher probability of giving to House races out of state than those taking the conservative stance. It is worth noting that affirmative action is a group-oriented policy area and thus, more likely to be a concern for liberals, in keeping with Grossman and Hopkins (2016) theory.

Further evidence that liberals might have more difficulty finding potential surrogate representatives among all House candidates lies in the fact that many of the aforementioned policy preferences are not significantly tied to out of state contributing even though liberals contribute to more House candidates on average than conservatives (see above). Rhodes et al. (2016) also uncover higher rates of out of state giving for Democratic donors in 2014. Based upon their higher frequency of contributing out of state, the likelihood that their policy preferences would be significant should be higher but that is not the case. Additionally, Democratic donors in this set of data and in other studies have been shown to be more likely to contribute to the party organizations suggesting they pursue alternative means to gain more representation (La Raja and Schaffner 2015). Conservative donors may also have benefited from the emergence of the Tea Party movement in response to the Republicans’ electoral defeat in the 2008 election (Williamson et al. 2011; Arceneaux and Nicholson 2012)—candidates running under the Tea Party label are easy to recognize as true conservatives thus lowering the donor’s cost of identifying ideologically pure surrogates. Recall Tea Party supporters have a significant and positive likelihood of giving to House races in other states—those holding “very positive” views of the Tea Party have an overall 17% point probability of giving to House races outside of the state where the donor lives and, in comparison to those holding “very negative” views of the Tea Party, they have a 47% higher probability of surrogate seeking in House races.

In light of these results, surrogate seekers in House elections appear to have a distinct calculus of giving that is politically sophisticated and strategic, and extends beyond recovering a dyadic representative relationship to gaining additional representation from multiple candidates and from party networks, along multiple policy dimensions. Understanding the motivations of political donors remains a critical undertaking for future assessments of the health of democracy. House members and Senators who depend upon contributions from outside of their districts and states have already been shown to be less responsive to constituents’ ideological preferences (Baker 2016a; Barber 2016b). As the dependency for contributions outside of the district grows in House races (Baker 2016a; Gimpel et al. 2008), surrogate representation may become more dominant than geographic representation as a determinant of House members’ behavior in office. Efforts to gauge whether congressional members are more responsive to donors’ partisan and policy preferences than the preferences of constituents depend upon knowledge of what those preferences are in the first place. Thus, the results of this study lay important groundwork for future investigations of the distortions in representation that might stem from surrogate seeking.