Knowledge, Expertise, and Committee Power in Congress. James M. Curry Department of Political Science University of Utah

Transcription

1 Knowledge, Expertise, and Committee Power in Congress James M. Curry Department of Political Science University of Utah Prepared for presentation at the 2016 Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 7, 2016 Thanks to Scott Adler, Rick Hall, Keith Krehbiel, Frances Lee, Timothy Nokken, Jason Roberts, Josh Ryan, Alan Wiseman, the participants in the Utah Political Science Colloquium for their helpful comments on earlier drafts of this paper and to Matt Haydon for his research assistance. This research was supported by a Congressional Research Grant from the Dirksen Center.

2 ABSTRACT The institutional features and processes political scientists believe to underpin committee power, including gatekeeping and proposal rights and an ex-post veto, have seemingly declined as regular order has become a thing of the past in the House and Senate. Rather than conclude that committees have lost influence in Congress, I argue that these developments uncover that committee power is instead based in the knowledge and expertise advantages enjoyed by committees and their members on issues within their jurisdictions. Drawing on a mixed methodological approach utilizing both interviews with congressional staff and an original dataset of every amendment offered under open amending procedures in the House of Representatives from , I find committees can leverage their knowledge and expertise of policy issues into influence. This new explanation of committee power has implications for how we understand policymaking and power in the modern Congress.

3 Committee power is presumed by political scientists to be grounded in institutional features, such as gatekeeping rights, the use of restrictive rules, the nature of bill introduction referral, monopoly agenda setting and proposal abilities, and the conference committee process (see, e.g., Denzau and Mckay 1983; Ripley 1983; Shepsle and Weingast 1987; Snyder 1992; Weingast and Marshall 1988). This presumption is overstated. These institutional features are not formal, codified rules of congressional policymaking, but are better understood as behavioral norms and standard procedural practices that typically have been adhered to in the development, consideration, and passage of legislation in the House and Senate. More importantly, in the contemporary Congress most of these standard practices have declined as the use of more unorthodox approaches to legislating have increased. Today, formal committee stages of the legislative process are frequently bypassed, and major legislation is typically negotiated and drafted among key legislators in informal, behind-the-scenes meetings. These changes have two possible consequences for our understanding of congressional committees. Either the conventional wisdom about committee power holds and committee influence over the behavior of lawmakers, and the outcomes of policies, has waned as a result, or committees remain important and influential, suggesting committee power must rest on other foundations. I argue for the latter. Specifically, I argue that the superior knowledge and expertise that committee members, particularly senior committee members and committee staff, develop and possess on issues related to their committees jurisdictions provide a durable and important source of influence for committees and their members. Because the typical rank-and-file lawmaker is pressed for time (Hall 1996; Kingdon 1989), lacks deep knowledge of most policy areas (Curry 2015), and has fewer staff resources than in the past (Drutman 2016), possessing knowledge and expertise on a policy issue not only empowers members in policymaking activity, 1

4 but makes them influential sources of cues and information for their colleagues, allowing them to shape the roll-call voting behavior of other lawmakers on the floor. Knowledge asymmetries between committee and non-committee members have been acknowledged for decades (Austen-Smith and Riker 1987; Curry 2015; Fenno 1966; 1973; Gilligan and Krehbiel 1989; Kiewiet and McCubbins 1991; Kingdon 1989; Krehbiel 1991; MacNeil 1963; Manley 1970), but most scholarship argues that majoritarian legislative processes restrain committees from using their superior knowledge for influence (Gillian and Krehbiel 1989; Kiewiet and McCubbins 1991; Krehbiel 1990; 1991; Krehbiel and Rivers 1988). I place knowledge and expertise front and center with committee power, highlighting reasons committees can use these advantages to influence the behavior of their colleagues. Drawing on a mixed methodological approach that includes interviews with congressional staff and an original dataset of floor amendments to committee bills, I provide evidence for this theory and discuss its implications for our understanding of committee power and policymaking in Congress. PROCEDURES AND PERCEPTIONS OF COMMITTEE POWER The most prominent theories of committee power in Congress have emphasized the role institutional features and procedural norms. The roots of this tradition can be traced to Woodrow Wilson s description of committees as decisive gate-keepers, able to lock up bills in dim dungeons of silence (1885, 69) never to be heard from again. The ability of the 19th Century committees Wilson studied to kill legislation by refusing to report it back to the floor provided them with significant power. In more recent decades, institutional features provided a compelling explanation for the deference lawmakers appear to afford other committees. Scholars of the 1970s and 1980s, dissatisfied with explanations of committee power that emphasized deference and reciprocity as norms explaining this behavior, wished to develop explanations more 2

5 consistent with emerging new institutionalist and rational choice approaches (Evans 2011, 401). Scholars since then have identified various procedural rights and standard processes that are argued to be central to committee influence in Congress, including the bill referral process, committee gatekeeping rights, monopoly agenda setting (or proposal) rights, the use of restrictive rules, and the conference committee process. Gatekeeping is perhaps the most lauded of these (e.g., Denzau and Mckay 1983; Ripley 1983; Snyder 1992; Wilson 1885), and is typically viewed as, the most important tool possessed by committees, (Maltzman 1997, 65). Gatekeeping is argued to be a strong source of negative power, providing committees a right to stop the consideration of any legislation under its jurisdiction with which it does not agree. While gatekeeping rights have never been absolute (Shickler and Pearson 2009), for the most part they have been presented as rarely upended. Monopoly proposal or agenda setting abilities build on gatekeeping and have likewise been asserted as a major factor in committee influence (e.g., Baron and Ferejohn 1989; Denzau and Mckay 1983). In addition to the discretion to act or not on legislation referred to it, committees have customarily had significant authority to draft and re-draft legislation and essentially shape the nature of policy proposals considered on the floor. Combined, gatekeeping and proposal abilities are argued provide committees with both positive and negative powers. Finally, Shepsle and Weingast (1987) assert that committees enjoy an ex-post veto over legislation via the conference committee process, which in turn reinforces power gained from gatekeeping and proposal rights. Because senior committee members typically dominate the memberships of the conference committees appointed to resolve bicameral differences on a bill, they have a second chance to amend or kill legislation before the final vote in each chamber. This ensures committee members can re-address whatever the floor has done to their bills. 3

6 These and other processes have dominated contemporary scholarly discussions of committee power, but since the 1970s, congressional processes have evolved in ways that should lead us to reexamine their relationships to committee power. For instance, both the gatekeeping and proposal rights have been undermined by an increased frequency in legislation bypassing committee consideration altogether. As shown in Figure 1, during the 1970s and into the 1980s (through the 100th Congress), better than 80-90% of the bills passing either the House or the Senate had been previously considered and reported by a committee in that chamber. However, by the end of the time-series those rates are less than 60% in the Senate, and around or below 50% in the House. Gatekeeping and proposal rights have been further challenged by an increase in the referral of bills to more than one committee, at least in the House. Figure 2 shows the percent of bills referred to more than one committee, and demonstrates that multiple referrals have been on a steady rise in the House. Both of these changes suggest modern committees cannot derive significant influence from blocking legislation or enjoying monopoly proposal abilities on policy issues under their jurisdictions. 1 Bills are routinely considered on the floor without a committee report, and on many bills more than one committee has an opportunity to shape and report legislation. [FIGURE 1 about here] [FIGURE 2 about here] An additional change undermines the logic of Shepsle and Weingast s (1987) ex-post veto. As shown in Figure 3, the use of conference committees has been on the decline for decades. Conferences were used frequently in the 1970s and 1980s, with hundreds of conference 1 Sinclair (2012) notes that the multiple referrals of bills empowers party leaders as they get to lead the process of resolving differences among different committee passed versions of the same bill. 4

7 reports considered each Congress, and as many as 25% of the bills passing both the House and Senate utilizing a conference for bicameral resolution. However, at the end of the time-series conferences are rare, used just seven times in the 111th Congress and three times in the 112th. Not since 1998 have conferences been used to reconcile bicameral differences on more than 10% of bills. According to Shepsle and Weingast, the conference committee process is central to the ex-post veto because conference committees, by tradition, are primarily composed of lawmakers from committees of jurisdiction. However, as less formal processes for bicameral negotiations have become more common, party leaders have become the lead negotiators in bicameral resolution, and the formal involvement of committee members has declined (Sinclair 2012). [FIGURE 3 about here] Contemporary wisdom suggests that these and other changes to standard procedures should indicate waning committee influence in Congress. Without well-heeded rights to unilaterally block and advance legislation, committee influence in congressional policymaking may be weakening. But another possibility is that committee power has largely persisted because it stems from other sources. Indeed, recent scholarship has continued to find committees to be important and powerful institutions in Congress, highlighting their influence over public policymaking (Adler and Wilkerson 2013; Woon and Anderson 2012) and policy implementation (Aberbach 2001; Ainsworth, Harward, and Moffett 2012; Shipan 2004). Committees also remain influential in the allocation of federal funds (Clemens, Crespin, and Finocchiaro 2015), and continue to be the target of lobbying efforts (Hojnacki and Kimball 1998; 1999; 2001; Esterling 2007). Further, committees remain important to lawmakers careers. Members work hard to obtain ideal committee assignments (Frisch and Kelly 2006), find their assignments to be valuable (Grimmer and Powell 2013), and find that their legislative lives are 5

8 structured by committee activity (Lipinski 2009). In short, scholars continue to find committees to be important and influential, and so we should consider alternate explanations of their power. KNOWLEDGE AND EXPERTISE AS SOURCES OF COMMITTEE POWER An alternative explanation for committee power is that committees and their members are able to draw on their superior knowledge and expertise about the issues under their jurisdictions, and the respect their colleagues accord them because of their superior knowledge and expertise, to influence the behavior of their colleagues, including the positions they take on roll-call votes on bills and amendments. This argument is based in three aspects of congressional politics. First, knowledge and expertise are valuable, and needed, commodities in Congress. Bills have to be written, and for lawmakers to draft quality legislation they either need to possess a certain level of policy and political expertise on the issue being addressed, or they need to be able to draw on others who possess this expertise. While some bills are clearly intended for messaging purposes, and do not need to be meticulously crafted, other are intended to influence public policy, and do require knowledge and care to draft effectively. Votes also have to be taken, and on every vote members of Congress need to know enough to cast a good vote one that either will help with their reelection, or at least not hurt. This dynamic means that those in-the-know on any issue will be empowered. Knowledgeable and expert lawmakers, or those who have access to knowledgeable and expert staff, especially when they are in the majority, will be involved in drafting major legislation on that issue. Just as important, these expert and knowledgeable lawmakers will become sources of information and cues for their colleagues. Members of Congress do not have the time or resources to become expertly informed, or even relatively well-informed, on every issue and policy, and so turn to others for help when votes on those issues occur on the floor. Kingdon 6

9 (1989) finds that lawmakers draw on various sources of information and cues when making voting decisions, but notes that they are especially likely to turn to fellow lawmakers they trust and whom they view as knowledgeable and expert on the issue, a finding echoed in several other studies (Box-Steffensmeier, Ryan, and Sokhey 2015; Curry 2015; Matthews and Stimson 1975; Sullivan et al. 1993). Further, we know that the cues lawmakers take from their colleagues influence their votes and that this is true for members of the House (Curry 2015) and the Senate (Box-Steffensmeier, Ryan, and Sokhey 2015). In short, knowledge and expertise is valuable in Congress. Members are often in need of it, and those who have it are empowered to influence their colleagues. Second, knowledge and expertise are asymmetrically distributed in Congress, with committee and committee members typically more knowledgeable and expert than noncommittee members on the relevant issues, and senior committee members typically the most knowledgeable. This asymmetry exists, first, because lawmakers deal with the inability to be master-of-all issues by specializing their legislative efforts and their development of policy knowledge on a small number of issues. Typically, the issues they focus on relate to issues important to their constituents and those related to their committee assignments two factors that often align (Adler and Lapinski 1997; Hall 1996; Hall and Grofman 1990; Parker, et al 2004). 2 Staff hiring patterns further skews this distribution of knowledge. Figure 4 shows the average number of staffers employed by members of the House and Senate, and by the standing committees in each chamber, from 1995 to Across the time series, the typical House committee employed three to four times as many staffers as the typical member. Although 2 Of course, not all members of a committee are experts as Kaiser (2013) attests about the lack of knowledge among several members of the House Financial Services Committee about financial sector issues during the drafting and consideration of the Dodd-Frank reforms. However, where expertise does exist, it is almost always among the members of the relevant committee. 7

10 senators are provided with more generous staff and the difference is less stark, committee staffs seem to be growing relative to personal staffs in recent years. It is important to note that these raw numbers understate the gap as typically only half of a member s staff work on policy (the other half focus on communications and constituent service), while most committee staff focus their efforts on policy. [FIGURE 4 about here] Third, for several reasons, committees and their members are likely to often, though not always, push for policies that differ from what might be pushed by rest of their party or the rest of the chamber under other circumstances. The knowledge-asymmetry between committee and non-committee members has been acknowledged for decades (Austen-Smith and Riker 1987; Curry 2015; Fenno 1966; 1973; Gilligan and Krehbiel 1989; Kiewiet and McCubbins 1991; Kingdon 1989; Krehbiel 1991; MacNeil 1963; Manley 1970), but most scholarly focus has been on how the knowledge- and expertise-advantages of committees are restrained by majoritarian processes for deciding committee assignments, chairmanships, and floor consideration of bills (see, e.g., Krehbiel 1991; Kiewiet and McCubbins 1991). Yet there are good reasons to expect that not only do committee push for policies that differ from the majority party or chamber median, but that they can influence other lawmakers to accept these proposals. For one, balance of the evidence suggests that the committee assignment process results in committees that are not representatives of the chamber s preferences on its issues (e.g., Hall and Grofman 1990; Adler and Lapinski 1997; Londregan and Snyder 1994; Groseclose 1994; Parker, et al 2004). Further, this preference bias among committee members is exacerbated by an intensity bias as actual involvement in the policymaking on any issue is dominated by just a small number of committee members (Hall 1996), likely skewing the preferences of those 8

11 involved even further from the rest of the chamber. Moreover, the expertise held by some committee members may itself cause them to see and understand policy areas differently, developing opinions and preferences that differ from their colleagues who are otherwise ideologically similar. For instance, Kaiser (2013) notes that Barney Frank (D-MA) differed from many of his similarly-liberal Democratic colleagues regarding proposed financial reform policies because of his familiarity and knowledge of financial sector policy. Furthermore, majoritarian processes thought to restrain committees are probably not very effective. The conventional logic is that because the decisions to consider a bill, amend it, and pass it occur via majority floor votes, chamber majorities will block bills they do not like, or else alter them before passage. But the ability of the chamber or the floor to assess a committee s policy proposal, and then amend it appropriately or kill it, requires that members understand the policy proposal well enough and understand how it relates to their preferences. Indeed, scholarship asserting floor-based checks on committee power routinely assume that each legislator has full knowledge and information about what a bill will do and how that relates to their preferences and the preferences of every other legislator in the chamber. But the lack of knowledge and expertise possessed by most members on policies beyond their specialties undermines this assumption and this possibility. If lawmakers often cannot independently understand the implications of a policy proposal or relate it to their preferences, they cannot make these assessments and take these actions without assistance. This is why cue-giving (and cue-taking) has been so powerfully demonstrated in Congress (Box-Steffensmeier, Ryan, and Sokhey 2015; Curry 2015; Kingdon 1989; Matthews and Stimson 1975; Sullivan et al. 1993). Lawmakers often cannot assess the work of committees on the floor because they rely on information from the committees to make these assessments, and this provides opportunities for 9

12 committees to influence the behavior of other lawmakers. Combined, lawmakers needs for information, the superior knowledge and expertise of committees and their members, the differences in policy preferences between committees and the floor, and the doubt cast on the utility of majoritarian processes for limiting the influence of committees, there are good reasons to expect that knowledge and expertise are important sources of committee power and influence in Congress. OBSERVING COMMITTEE INFLUENCE Knowledge- and expertise-based committee power should manifest itself many observable ways. For instance, it should ensure committee members are brought in and involved in drafting legislation on their issues, even when the committee stages of the legislative process are formally bypassed, because their political and policy clout and expertise will be needed to write quality legislation, and to build legitimacy for policy proposals. 3 It should also result in committees and committee members being able to influence rates of support or opposition for policy proposals on the floor through the provision of cues and information. It is this latter manifestation that is examined here. Specifically, I employ a mixed methodological approach of both elite interviews with congressional staff and quantitative analyses of a unique dataset of amendments offered under open amending procedures during the 109th and 110th congresses ( ) to assess whether members floor votes are influenced by cues and information provided by committees and senior committee members. 3 For instance, with the negotiating and drafting of omnibus spending packages in recent years, various news reports noted Appropriations chairman Hal Rogers s (R-KY) central position in the negotiations, and the role of Appropriations committee staff in the drafting process. From late 2015, for instance, see, Mike Lillis and Scott Wong, Leaders inch toward funding deal, The Hill, December 9, 2015; Lindsey McPherson, Negotiators Hope to Unveil Omnibus on Monday, Roll Call December 10, 2015; Lindsey McPherson and Tamar Hallerman, House Passes Five-Day Extension of Government Funding, Roll Call, December 11, 2015; Roll Call Staff, The Snail s Pace of Omnibus Negotiations, Roll Call, December 11,

13 Interview evidence comes from five in-depth, semi-structured interviews with staffers who work for and advise members of the House of Representatives. Each interview took place over the phone and was audio recorded. The interviews lasted between 30 and 40 minutes and discussed how staffers help their lawmakers become informed about bills and floor amendments that are outside of their lawmakers issue specialties. In particular, the discussions focused on how they went about becoming informed enough to make vote recommendations to their bosses under limited time constraints. More information on the interviews and the interviewees can be found in the Appendix. The interviews establish that committees are prime sources of information and cues drawn on by lawmakers to make roll-call voting decisions Building on the interview evidence, quantitative analyses employing a dataset of every amendment offered under open amending procedures on the floor of the House between 2005 and 2008 (the 109th and 110th congresses) assess the degree to which the cues conveyed by senior committee members during floor debate influence the likelihood that other lawmakers support or oppose those amendments. In total, data on 1,188 amendments were collected, including whether each was adopted, if it received a roll-call vote, the vote cast by each member of Congress, and the characteristics of each amendment sponsor. To identify the cues conveyed by senior members of the relevant committee on each amendment, the floor debate was read. With each amendment, the statements made by the committee chair, the minority s ranking member, the relevant subcommittee chair, and subcommittee s ranking member were coded as either in favor of the amendment, opposed to the amendment, or neutral/no statement. The assumption made (and confirmed in the interviews) is that these statements reflect the cues and information provided by committee staff behind the scenes, and represent the positions of the committee majority and minority, respectively, on each 11

14 amendment, if a position existed. A focus is placed on amendments offered under open amending procedures because it avoids the potentially confounding influence of the Rules Committee s decisions over which amendments to allow to be considered on the floor and which to block. When the Rules Committee grants a bill a restrictive rule, it limits the number and type of amendments offered on the floor, and decides which amendments are in order and which are not. This process allows the Rules Committee to often filter out amendments that the majority leadership and the proponents of the bill do not like, but might pass or cause tough votes for majority lawmakers. In contrast, under open amending procedures, any germane and otherwise permissible amendment is in order to be offered and considered, including amendments from both sides of the aisle that might garner support, but that the committee of jurisdiction for the bill does not like. Committees abilities to influence votes on amendments under these conditions cannot be attributed to institutional features or procedural advantages awarded to committees, but instead the ability of committees to draw on their knowledge and expertise to convince their colleagues to support or oppose an amendment, or the deference other lawmakers afford committees for these reasons. The expectations for these combined qualitative and quantitative analyses are clear. For a knowledge and expertise basis of committee power to be supported, interviews should demonstrate that lawmakers and their staff see committee members and staff as important sources of knowledge, expertise, and cues, and that they turn to these actors when they need information. The quantitative analyses should demonstrate that when senior committee members convey a position on the floor it influences the likelihood an amendment is adopted and the likelihood lawmakers support or oppose an amendment. 12

15 INTERVIEW FINDINGS Previous studies demonstrate that committees and their members serve as important sources of knowledge and cues (e.g., Box-Steffensmeier, Ryan, and Sokhey 2015; Curry 2015; Kingdon 1989). The interviews I conducted reiterate this finding, but more importantly clarify how and when lawmakers turn to committee members and committee staff for information to help them make decisions, and the role that these cues play in the decision making processes of lawmakers, especially regarding floor amendments. Each interview was in-depth, discussing how member offices seek and obtain the information they need to make decisions on floor amendments. The results provide context to how these decisions are made and the factors and actors that influence these decisions. That the number of interviewees is small and non-representative of the full body of Congress should not be a concern given the large-n quantitative analyses below. The results of these interviews provide important insights for interpreting the quantitative findings. Each interview started with a discussion of how the staffer would gather information to help their lawmakers make decisions on votes if they had no time constraints. Each was then asked how this hypothetical process changes when there is very limited time such as under open amending procedures and decisions must be made quickly. Each staffer was also asked about specific amendments they advised their member on during recently considered fiscal year 2016 spending bills to provide some context to the discussion. With unlimited time, staffers would identify all the relevant and interested parties [in the bill or amendment] and touch base with them and their groups to understand their position on it. 4 Such relevant parties include trade groups, constituent groups, stakeholders, and others in the public, as well as other lawmakers involved in the bill or close to the member, the party 4 Interview 2 13

16 leadership, caucuses like the Republican Study Committee or the Tuesday Group, and other key players. However, the unrealistic nature of this hypothetical scenario was underscored by the laughter with which each staffer initially replied. Lawmakers and staffers never have all the time in the world. Rather they make decisions under compressed time schedules. This fundamentally changes their pursuit of information and decision-making processes. With limited time, staffers narrow their search. Generally, they first reach out to an amendment s sponsor to get an overview of its intent: My first call is usually to the amendment s sponsor. 5 There are a couple of things I need to understand. The first is the author s intent. What the sponsoring office intended the amendment to do, why they intended it to do that, if there is a problem they are trying to fix, or if the current [bill] language causes a problem or something. 6 For lawmakers who do not specialize in an issue, deciphering an amendment s purpose is not always easy. As one staffer put it, A lot of time these amendments are offered very quickly and it s not entirely clear what the meaning is. As you know, they are often written in parliamentary language by legislative counsel and it can be hard to decipher. 7 So staffers try to get all the information they can from the office of the sponsor. In general, these staffers professed knowing little about most issues on which they have to advise their bosses. As one put it, Many House staffers who aren t committee staffers tend to cover a lot of issues for their boss. You might be covering nine different issue areas. So, maybe you re pretty deep in one of those things. But a lot of them you re skating on the surface. 8 This is a major driver of the need for information and cues. Four of the five staffers indicated that committee members and staffers were among the most important sources for this information. 5 Interview 3 6 Interview 4 7 Interview 1 8 Interview 3 14

17 For at least one staffer the committee served as a check on what the sponsor s office told him, as he was confident the committee staff were well versed and informed: I ll go to the committee to double check what the sponsor s office told me. A lot of times the committee has a lot more information. They have vetted all of these amendments and they have a lot of background information. They have probably done some studies on the issue that might be relevant for us to know. The committees have a pretty good wealth of knowledge on these things. 9 Another noted that committee staff actively aggregate useful information and have it at hand: [Committee staff] are often the point of aggregating support or opposition to a bill or amendment. They will have at the ready support or opposition letters for whatever they are working on which is nice to have in one place. 10 Another noted that committee staff are especially helpful on complex amendments: Especially if the amendment text is really long or if the way it s written is not very comprehensible on first read we really depend on committee staff to give us the summary or a one-pager. 11 For these staffers, the reason to turn to committees is that they are knowledgeable and expert on the policy issue, and could quickly provide the information needed. These two staffers note that they work to develop relationships with staffers in other offices whose members focus on different issues and sit on different committees: For example, I ve called people I think are very knowledgeable on transportation questions, even if they work in an out-of-state office, say, because they ve been on the committee and maybe they know some of the background, and the question doesn t affect their state at all but they are just someone who knows the distinction between the different programs being affected. 12 So if it s a tech issue, he [the lawmaker] knows of a certain member that he is friends with who has taken the lead on a lot of tech issues. That s the office we reach out to. Depending on the issue he has different members that he works with a lot and he knows that they are very strong in certain issues 13 9 Interview 2 10 Interview 4 11 Interview 4 12 Interview 3 13 Interview 2 15

18 The opinions of committee chairs (and ranking members for lawmakers in the minority) were noted as especially important as cues for how their members might want to vote on floor amendments. This was especially true if a lot of amendments are being offered, as is often the case under open amending, on big spending bills, and on major reauthorizations: Yeah, let s say it s the NDAA [annual Defense bill] and there are 400 amendments and you re really not sure and you have to make suggestions to your boss. Trying to get any recommendation out of the chairman becomes top priority. 14 The way staffers described the weight placed on the recommendations of chairs and ranking members sounded a lot like deference. Committee leaders knew more, and cared more, about these issues, so if they recommended support or opposition it weighed heavily. As this staffer put it, the recommendations of the sponsor of the bill being amended (who typically a committee member), and the chair, are very important: Whatever specific bill, you want to know what the bill s sponsors think because that helps weigh your decision. Maybe in an ideal world you wouldn t support this amendment. Maybe you re relatively neutral. Maybe it s not a big deal for you. But if it s a big deal to the bill s sponsor that will weigh into your decision making, and if it s a big deal to the chairman. 15 These staffers also discussed how they obtained information from committee members, staffers, and chairs. Their responses noted that floor debate was often important: Yeah, the floor debate is very important. Maybe it s not something that your boss has time to watch but usually that s the easiest place to get the chairman s recommendation because the chairman, or whoever is sponsoring the bill, has to stand there and engage in floor debate when someone offers an amendment. 16 But the interviews also suggested statements made on the floor by senior committee members reflected the information being offered by committee staff behind the scenes. 14 Interview 3 15 Interview 3 16 Interview 3 Because for the most part these [floor statements] are pre-prepared statements that the staff have written talking to legislative staff can get you just as much information. I d say 16

19 the time floor debate is most important is during an open rule because you might not always have a description of the amendment handy. So if you look at the amendment text and it says, strike line 15 on page 2 that s not going to tell you a whole lot very quickly about why they are doing what they are doing. So we try to watch. 17 Sometimes this information is by request, but often it is actively distributed by committee staff: If we do want this information, it comes through a dear colleague or an that they ve sent out ahead of time discussing the amendment. 18 Generally, the interviews indicate that member offices are often in need of information, turn to committee members and staffers for this information because they view these actors as knowledgeable, and pay special attention to the positions taken by senior committee leaders. The next section builds on these qualitative findings to quantitatively assess the impact of the cues communicated by senior committee members on support for floor amendments. QUANTITATIVE FINDINGS First, Table 1 presents the results of logistic regression analyses predicting the likelihood that each amendment was adopted, first for all amendments in the dataset, and second for just those receiving a roll-call vote. The primary explanatory variables are indicators of positions expressed by the four committee leaders on the floor the committee chair and relevant subcommittee chair in the majority, and the ranking member and subcommittee ranking member in the minority. Each is coded to 1 if the member expressed support during floor debate, -1 if the statement expressed opposition, and 0 if no statement was made or if the statement was neutral. Included with these indicators are several control variables, the most important of which is the amendment sponsor s ideological extremity. This measure is each amendment s sponsor s first dimension DW-NOMINATE score difference from the chamber median. Sponsors that are perceived as more extreme and likely are introducing amendments that are more extreme. More 17 Interview 4 18 Interview 2 17

20 extreme amendments, all else equal, should have lower likelihoods of adoption than more moderate amendments. 19 Additionally, each amendment is coded for whether the sponsor was in the majority or minority party, the seniority of the sponsor (in years served in the House), and whether or not the sponsor was on the committee of jurisdiction for the bill. Additionally, a variable indicates if an amendment was offered by Representative Jeff Flake (R-AZ) who offered 91 amendments during the four years analyzed here, more than double that of any other lawmaker, and saw just two of his amendments adopted. 20 Finally, dummy variables are included in the analyses for the committee of jurisdiction for each bill that had amendments offered to it, with the Appropriations Committee serving as the excluded category. 21 [TABLE 1 about here] 19 One downside to using this measure to assess the extremity of amendments is that it does not allow for strategic behavior that the sponsor may be typically extreme, but is proposing something more centrist or moderate to try to get it passed. Nor does it allow for the possibility that the sponsor may be often extremist on a number of issues, but not on the particular issue the amendment addresses. Or that the amendment is compromise work between an extreme member and other less extreme members, or even a bipartisan set of members (indeed some amendments have co-sponsors, but they are not formally listed in the congressional record making it hard to universally identify them). Indeed, these considerations are important. Nevertheless, the extremity of the sponsor, generally speaking, which is reflected in their DW-NOMINATE score, does color the reactions other lawmakers will have to their amendments. Members of Congress carry reputations, which may be ported to their proposals. For instance, if Steve King (R-IA) proposes an amendment, other lawmakers may expect that it may serve some extreme, conservative purpose, just as they might expect an amendment offered by Barbara Lee (D-CA) or Jim McDermott (D-WA) will serve distinctly liberal purposes. That the measure so well explains member s votes on their colleagues amendments suggests that these reputations, if not the actual content of the amendments, meaningfully influence floor votes on amendments. 20 All of Representative Flake s amendments proposed to eliminate specific earmarks from bills. This was done largely to make a political point, favored by Flake and others, that earmarks were bad and that the earmarking process was in need of reform. Flake himself did not expect his amendments to be adopted, which is clear from many of his floor statement, and they often were rejected without much debate or without committee leaders expressing opposition as such opposition was assumed after a while. For this reason, a dummy controlling for these amendments is included because they hold something of a special status in the data. 21 Of the 1,187 amendments, 1,081 were offered to Appropriations Committee bills, 61 to Financial Services Committee bills, 25 to Science, Space, and Technology Committee bills, 19 to Transportation and Infrastructure Committee bills, and 1 to a Natural Resources Committee bills. 18

21 The first column of results shows the impact of committee cues on the likelihood of adoption for all amendments offered. The results show that the cues given by the chair, subcommittee chair, and subcommittee ranking member impact the likelihood of adoption, with statements in favor increasing the chances of adoption, and statements opposed decreasing the chances. 22 The second column shows the results for the subset of amendments that received a roll-call vote. Most amendments, nearly 60%, were not subject to a roll-call vote but either received a voice vote, were withdrawn by the sponsor, or subjected to a point of order. Many of these amendments were clearly acceptable or unacceptable to members of the chamber, as well as the committee of jurisdiction for the bill. Committee cues were likely both unnecessary and unlikely to influence meaningful levels of support on most of these amendments. However, amendments receiving roll-call votes are generally more controversial. 23 Roll-call votes are typically only requested when there is sizeable support and opposition to the amendment. Under these conditions, the committee will need to provide cues and information to try to sway votes and influence an outcome. The results in column 2 suggest these cues do appear to make a difference with the coefficients for the committee chair, subcommittee chair, and subcommittee 22 Notably, throughout these analyses the minority ranking member of the full committee of jurisdiction does not appear to have an impact via the statements made on the floor. This is, in large part, because Representative Jerry Lewis (R-CA), who was the Republican ranking member on the Appropriations Committee during the 110th Congress, largely refrained from speaking on the floor during the consideration of appropriations bills, and he was the ranking member for 40% of the amendments in this dataset. His limited participation on the floor accounts for the lack of an impact, however, these results shouldn t be read that the ranking member does not have an impact. In all likelihood, Lewis was active behind the scenes directly and indirectly through his staff, providing cues and information, and leveraging his knowledgeand expertise into influence. 23 Floor amendments offered under open amending procedures are initially subjected to a voice vote at the end of debate. Any member may subsequently request that a recorded vote be taken that would trump the results of the voice vote, so long as one-fifth of a quorum of members (25 members in most cases) supports the request. However, in practice, a recorded vote is scheduled whenever a member requests it, even if no other members support the request. 19

22 ranking member again statistically significant. Notably, these findings hold controlling for the amendment sponsor s ideological extremity, which is a strong predictor of the fate of each amendment in both analyses, with amendments offered by more extreme sponsors faring less well. Figure 5 presents predicted likelihoods of amendment adoption from both analyses, and shows that the influence of committee cues is noticeable regardless of the extremity of the sponsor. In each graph, the likelihood that an amendment is adopted decreases as the sponsor becomes more extreme. However, in every graph, across the measure of extremity, amendments on which committee leaders vocalized support were more likely to be adopted than those on which committee leaders noted opposition. These findings indicate that committees have significant abilities to influence support and opposition for floor amendments to their bills. [FIGURE 5 about here] A counter explanation for the above results is that committee leaders, strategically or otherwise, primarily note their support for amendments already likely to pass, and only note their opposition to amendments already likely to fail. While this explanation cannot adequately account for the limited knowledge most rank-and-file non-committee lawmakers bring to the consideration of floor amendments, or for the above interview evidence that most lawmakers profess the need for information and cues to making voting decisions and frequently turn to committees and their members for this information, it is worth assessing with the quantitative data as well. Figure 6 visualizes the relationship between the extremity of an amendment s sponsor and the positions taken by committee leaders on the floor. As shown, each committee leader is slightly more likely to take a supportive position on amendments sponsored by less extreme lawmakers and slightly more likely to oppose amendments sponsored by more extreme 20

23 lawmakers, But the relationships are indeed very slight with the variables never correlated more strongly than r= Committee leaders are not taking their positions at random, but the weak correlations presented in Figure 6 suggest this cannot fully account for the results above. [FIGURE 6 about here] While the above results show that the cues senior committee leaders provide during floor debate influence the likelihood amendments are adopted or rejected, they cannot show if leaders on different sides of the aisle have differing influence, or if these cues influence some lawmakers more than others. Table 2 presents the results of logistic regression analyses predicting the impact of committee cues on the votes taken by each member on each amendment receiving a roll-call vote. The primary independent variables of interest are the same as in Table 1, as are many of the control variables, including the party, seniority, committee assignments, and DW- NOMINATE extremity of the amendment sponsor. In addition, two important control variable can be added because the data are of each individual representative s vote on each amendment. The first is a measure of the absolute differences between the first dimension DW-NOMINATE scores of the voting member of Congress and the sponsor of the amendment. All else equal, members should be more likely to support amendments offered by sponsors more similar to themselves than more different. Second, a measure of the partisan favorability of each member s district is included, measured as the percent of the two-party vote won by the presidential candidate of their party in their district in the 2004 election. Larger values indicate safer and more partisan districts for a lawmaker, and may induce more partisan behavior on the floor. [TABLE 2 about here] The results in column 1 are for lawmakers in the majority party during each Congress. These results show that the statements made by the committee chair, the subcommittee chair, and 21

24 the subcommittee ranking member significantly influence a lawmaker s likelihood of supporting or opposing an amendment. When committee leaders from the member s party the chair or the subcommittee chair express support or opposition to an amendment, a member is more or less likely to vote for it, respectively. The statements made by the subcommittee ranking member also appear to have an effect, indicating that some circumstances that it may not just be partisan cues that help lawmakers decide how to vote. In fact, these results may indicate that when the subcommittee chair and ranking member indicate provide the same cue, committee influence over legislator behavior is especially pronounced. The results are similar for minority lawmakers (column 2) with the statements made by the committee and subcommittee ranking members, and the subcommittee chair, significantly influencing the likelihood a lawmaker supports or opposes an amendment. Importantly, these findings hold controlling the difference between the voting member and the amendment sponsor. The DW-NOMINATE difference measure should be a strong predictor of a lawmaker s vote on any amendment, especially given the interview findings above that member s offices reach out directly to amendment sponsors for insight and information. That committee cues have an influence controlling for this difference measure, this suggests a very strong role for committee cues. Figure 7 explores this more fully. Each panel of this figure plots predicted likelihoods that a lawmaker votes in favor of an amendment as their difference from the amendment sponsor increases, with the results split by the cues provided by each leader. [FIGURE 7 about here] The results are striking. Among lawmakers in the majority, the cues expressed by committee and subcommittee chairs strongly influence the likelihood of support whether the 22

25 lawmaker is very similar to or very different from the amendment sponsor in terms of their typical policy views. The effect appears especially strong on amendments offered by colleagues who are very similar to the voting lawmaker. In cases where the amendment sponsor and voting lawmakers in the majority party have nearly identical DW-NOMINATE scores, for instance, the likelihood that the lawmaker will vote to adopt the amendment is around 75% if the committee chair does not express a position. However, when the chair indicates support, the lawmaker s likelihood of voting for adoption is better than 90%; when the chair expresses opposition that likelihood drops below 50%. This is a substantively large effect. Similar, though slightly smaller, effects are found for the cues expressed by other committee leaders, when the party of the leader and the voting member align. Most important is that the effects described here hold even when the voting lawmaker and the amendment sponsor are more different. While the overall likelihood of the lawmaker voting to adopt drops as the sponsor becomes more different, the lawmaker is always more likely to support the amendment if senior committee leaders in their party express support, and less likely to support the amendment is senior committee leaders in their party express opposition. While the coefficients cross-party subcommittee leaders are significant in the models above, the figures show the substantive effect is very small when controlling for the difference between the voting member and the amendment sponsor. Seemingly, same-party committee cues are the most influential. IMPLICATIONS FOR COMMITTEES, CONGRESS, AND POLICYMAKING These findings have important implications for how we understand committees and policymaking in Congress. First, these findings should direct political scientists to rethink our understanding of committee influence. Political scientists have considered long institutional features and legislative processes to be the primary underpinnings of committee influence. But 23