Conference Committee Structure and Majority Party Bias in U.S. States

Abstract How representative are conference delegations in state legislative chambers? I argue that differing conference rules across state legislative chambers influence majority party control over conference delegations. With an original data set encompassing all state-level conference committees from 2005 to 2016, I compare the observed policy preferences between conference delegation and majority party medians when the majority party unilaterally appoints and when the minority party has influence over conferee selection. My results show that in state legislative chambers where the minority can influence conference appointments, delegations are ideologically biased away from the majority party. These findings underscore how majority parties are limited when minorities have procedural rights.


Introduction
In every bicameral state in the United States, both chambers must agree on legislation before sending the bill to the executive. If passed bills differ between chambers, either one chamber can concur with the other's amendments, or they can form a conference committee. Conference committees are ad hoc joint committees composed of members from both legislative chambers tasked with unifying differences. They routinely handle the most politically contentious and salient legislation and are expeditious when passing comprehensive legislation by preventing bills from moving back-and-forth between chambers. 1 They are often the last actors to substantively modify bills, with conferees having the substantial latitude to alter legislation toward their preferences (Oleszek et al. 2015;Sinclair 2016). Conferences' powers are amplified due to the fact that their choices cannot be modified; adoption of the conference report is subject to a single yay-or-nay vote. 2 The unamendable nature of the conference report has led some scholars to remark that conferees' influence is so great that many laws are actually written in conference (Clapp 1963; Van Beek 1995).
Conference committees' last-mover advantage allows conferees to not only change legislation but also alter the behavior of other actors in the legislative process. This independent influence conflicts with contemporary party-dominated theories of legislative organization. Partisan models view conference conferees as agents of the majority leadership, pursuing goals valuable to the majority party (e.g., Cox and McCubbins 2005;Cox and McCubbins 1993;Kiewiet and McCubbins 1991;Rohde 1991). Conferees appointed by the majority leadership are expected to serve majority party priorities, lest they risk losing credit-claiming opportunities associated with the conference (Mayhew 1974). Nagler (1989) summarizes the situation accordingly: "The conference does indeed convey influence to the conferees. However, there is reason to believe that the conferee… represent [s] only what is acceptable to a majority of the majority party on the floor" (76).
To date, most of the conference committee literature has focused solely on the U.S. Congress. In this context, conference committees serve as a vehicle for majority party agenda control as the majority leadership unilaterally appoints conferees. However, Congress' lack of institutional variation inhibits the theoretical testing of conditions when conferees may deviate from majority party goals. I overcome this static condition by switching the unit of analysis from the national legislature to state legislatures. Conference rules vary substantially on the sub-national level, yet few scholars have explored how differences in conference rules affect state-level legislatures (Emrich 2022a;Ryan 2014). Overall, scholars' understanding of conference committees in state legislatures is largely speculative, despite nearly every state using conference committees to resolve bicameral differences (Gross 1980). 3 This article advances our understanding by exploring how differing conference appointers influence majority party control over conference delegations. Given legislative scholars relative lack of knowledge surrounding conference committee behavior (Longley and Oleszek 1989), analyzing how varying conference rules influence majority party power over has substantive importance in explaining how policies are modified when minorities have procedural rights in resolving bicameral differences (Clark and Linzer 2015).
Conference appointer rules differ across states and within state legislative chambers, providing important variation to investigate how majority leaderships may be limited by institutional features. In contrast to earlier work (Gross 1980;1983;Lauth 1990;Ryan 2014) that focuses on a few states or a single year, I use an original data set encompassing all state-level conference delegations from 2005 to 2016 across 41 states to analyze the degree to which conference delegations reflect systematic 2 Of course, the median legislator can reject the conference report in anticipation of a better future bill, though this is an uncertain strategy because the future bill would need to be reconsidered and passed in both chambers. 3 Some states (e.g., Maine) call panels between chambers to resolve differences in legislation committees of conference instead of conference committees. These terms have identical meanings, and committee of conference is used when referring to applicable states. majority party bias. 4 I find that on average, conference delegations do not exhibit a pro-majority slant. Instead, their ideological distribution resembles the chamber's, in line with the predictions of Krehbiel (1991;1993).
Second, I compare the observed policy preferences between conference delegation and majority party medians when the majority party unilaterally appoints and when the minority party has influence over conferee selection. My findings indicate that the presence of minority appointer rights in the chamber widens the difference between conference delegation and majority party medians, pushing the average conference delegation away from the majority party towards the chamber median. These results highlight how majority parties are limited when institutional designs favor the minority party (Ryan 2014).

Conference Committees and Conference Appointers
Prior work has examined the independent influence that conference committees have on legislative outcomes, finding that conferences maximize their policy benefits when receiving majority support in both chambers (Tsebelis and Money 1997;Vander Wielen 2010). Conferees commonly receive support from their chambers due to the quality of information they use to construct stable outcomes and reduce the likelihood of stalemate (Rogers 2001;Rybicki 2003). Conference committees broker deals on legislation, bargaining across chambers to limit the risk of failure, elucidating why the parent chambers delegate authority to conferees (Vander Wielen 2013).
The unamendable nature of conference reports aligns prima facie with partisan models of legislative organization, which suggest that institutional characteristics are an important source of party power (Cox and McCubbins 2005). Spatially, party leaders can utilize conference reports to promote extreme legislative policies if the conferee proposal is preferable to the median than the status quo. In Congress, the Speaker of the House and Senate Majority Leader control conferee appointments, and can therefore select party loyalists to broker a deal ideologically consistent with their preferences (Lazarus and Monroe 2007). 5 Accordingly, congressional conferees largely abide by majority party preferences (Vander Wielen and Smith 2011). In state legislatures, the majority party in every chamber holds a numerical advantage in conference appointments. This serves the majority party in seeking their favored policy outcome, which in a unidimensional setting is the majority party median. Unless the state legislative chamber is perfectly homogeneous ideologically (which is unlikely e.g., Shor and McCarty 2011), the majority party median will differ from both the chamber and minority party medians. Like congressional conference delegations, I expect a baseline level of majority party influence over state legislative conference delegations (Nagler 1989;Vander Wielen and Smith 2011). This unconditional effect should lead to conference delegations ideologically resembling the majority party median, resulting in delegation compositions favoring the majority party: Hypothesis 1: Ideological medians of conference delegations reflect the majority party median in state legislative chambers.
Hypothesis 1 runs counter to Krehbiel (1991;1993), who challenges the partisan nature of conference committees. In an informational context (Krehbiel 1991), partisanship has no bearing on the composition of conference delegations. Instead, policy expertise derived from expert members of relevant standing committees dominates bicameral negotiations. Consider Massachusetts' appointment of Republican State Representative Kimberly Ferguson to a conference delegation in 2019 tasked with coalescing various proposed funding increases for education. 6 With experiences as Ranking Minority Member on the Education Joint Committee, Ferguson indicated upon appointment to the delegation the existing "bipartisan commitment to education" (Sentinel and Enterprise 2019). Other legislators echoed similar sentiments, with House Minority Leader Bradley H. Jones noting that "As a member of the Foundation Budget Review Commission whose 2015 report provided the impetus to revisit the state's education funding formula, Representative Ferguson understands the key issues that need to be addressed… to ensure that all Massachusetts students have access to a quality education" (Massachusetts House Republican Caucus 2019). Broadly, informational conditions suggest that conferees selected are more representative of the chamber's median voter (cf. Hall and Wayman 1990). This leads to my second hypothesis: Hypothesis 2: Ideological medians of conference delegations reflect the chamber median in state legislative chambers.
Following Krehbiel, the conditions under which state legislative conferees follow the majority party's preferences are unclear. First, while party leaders may instruct conferees when they go to conference, these instructions are not binding in any chamber. Second, although congressional conferees mostly follow majority party desires, this may be a result of the majority party leadership having total discretion over who serves on the conference committee. Most state legislatures follow Congress by unilaterally bestowing the majority leadership conference appointing rights. 7 That said, several chambers' legislative rules endow the lieutenant governor or minority party leadership with appointing rights whereby these actors can assign a single or multiple legislators to the conference delegation. 8 For example, Illinois' House of Representatives Rule 73 stipulates that "Each conference committee shall be comprised of five members of the House, three appointed by the Speaker and two appointed by the Minority Leader…" (Illinois House of Representatives 2015, 46). Figure 1a, b provide visualizations of conference appointing rules by state legislative 6 The conference committee resolved differences between H.B. 4145 and S.B. 2365. The bills' main purpose was the addition of 1.5 billion dollars in extra funding for Massachusetts public schools over a seven-year period as suggested by Massachusetts' Foundation Budget Review Commission report in 2015. 7 An exception to this norm is Oklahoma's House of Representatives, which automatically designates conferees based on legislation type (e.g., Conference Committee on Banking, Financial Services, and Pensions). 8 If a state's lieutenant governor has the same partisan identification as the minority party of the legislative chamber, I anticipate that they behave as minority leaders within the chamber do. chamber across the United States. In total, 14 (4 lower and 10 upper) state chambers have minority appointing rights. These rules were coded from the National Conference of State Legislatures (National Conference of State Legislatures 1998) and further confirmed by examining chamber rules. Appendix B provides a detailed listing of state conference committee appointer rules.
When majority party leaderships decide to engage in post-passage bargaining in legislative chambers where the minority party influences conferee selection, they are likely aware that minority appointers will pursue their preferences. As a result, the majority leadership likely cedes some control over conference delegation composition when forced to collaborate with minority leaders. This expectation is similar to how relevant majority party standing committee members can interfere with majority leadership goals in conference, as common practice is to appoint members from the standing committees or subcommittees with jurisdiction over the bill (Longley and Oleszek 1989;Smith 1988). 9 Once in conference, standing committee conferees can use their agenda-setting abilities to shift legislation towards their preferences (Vander Wielen 2010), while also utilizing their "ex post veto" rights to revert changes made on the chamber floor (Shepsle and Weingast 1987). Often, rogue standing committee conferees are high-demanding legislators seeking to modify legislation to better satisfy their (and their constituencies') preferences for electoral gain (Weingast and Marshall 1988).
For minority party leaders, they likely share similar goals to individualist majority legislators, seeking to have their ideological preferences satisfied for electoral purposes. Additionally, minority leaders possess a collective political incentive when obstructing the majority party leadership's goals in conference (Lee 2009). The minority party's linked electoral fate encourages its members to cooperate against the majority to discredit the majority leadership's agenda by being less willing to collaborate with the majority party in pursuit of partisan collective gain (Koger and Lebo 2017;Lee 2016). Mainly, compromise on average is a negative outcome for the minority party, because if they cooperate with the majority, they undercut their future electability (Gilmour 1995). Even if they receive concessions from the majority in conference to form the winning coalition, the majority still executes its mandate and receives the lion's share of the rewards (Balla et al. 2002;Groseclose and Snyder 1996). Because of this, I expect minority leaders to appoint conferees who are ideologically dissimilar from the majority leadership's preferences. 10 Substantively, I anticipate that in legislatures where the minority has the ability to influence conference appointments, conference delegations resemble the chamber median more so than the majority party median: Hypothesis 3: If chamber rules permit the minority party to appoint its own members to conference committees, the ideological median of conference delegations will be closer to the chamber median than the majority party median. 9 Lazarus and Monroe (2007) describe such in the context of U.S. House of Representatives conference appointments: "sometimes appointing members of a jurisdictional committee gets in the way of another of the Speaker's goals as a selected agent of the majority party: engineering the passage of legislation that is beneficial to (a majority of) the party" (595). 10 I expect this effect to be unconditional. As Lee (2016) suggests, even if the parties' preferences on a bill to not diverge substantially, minority parties still benefit by denying the majority party a policy victory for which they could claim credit.

Data and Methods
To analyze how variation in conference appointer affects majority party control over delegation composition, I created a new time-series cross-sectional data set on conference delegations from 2005 to 2016. The data collection starts in 2005; however, several states only report conference committee actions in recent years. For example, Florida and West Virginia only maintains conference delegation information from 2011 to present, whereas Kentucky starts in 2008. 11 Iowa, Louisiana, and Oregon all report from 2007 onward. Conference delegation information was scraped from state legislature web archives, as most states provide detailed bill histories of when conferences and conferees were appointed. For those which did not, web archives were searched for written and audio documentation of conference committees. Conference delegation information was then transcribed from these materials. 12 This data set contains all conference delegations for both lower and upper chambers. In situations where one chamber appoints conferees and the other does not, no conference committee convenes. However, given that I am focused on the appointing rights of majority parties, these cases are retained for analytical leverage in testing how varying institutional features impact majority control over legislative outcomes. 13 Despite most states using conference committees, several do not. Most known is Nebraska's unicameral legislature. Additionally, New Jersey and Rhode Island do not have conference committees to prevent legislative gridlock. Other states have codified conference committees or similar institutions, but they serve specific purposes or are seldom utilized. For example, Delaware only uses a joint appropriations committee for budget bills, whereas Arkansas, Connecticut, and New York have rarely used conference committees in the past century, with none occurring throughout the scope of my analysis. 14 Maine and Ohio have bill histories with insufficient documentation to discern who served on the conference committee. 15 Additionally, Oklahoma's legislature uses conference committees, but is idiosyncratic in its use of permanent standing conference committees for legislation. 16 Broadly, I exclude from the analysis those delegations where I was unable to recover conference membership. The resulting data 11 Kentucky and Florida list the conference committees formed prior to 2008 and 2011, respectively, but do not provide conferee membership information.
12 Data for many states from 2011 to 2016 was taken from Open States, which is a nonprofit organization that uses crowd-sourcing and web scraping to compile data on legislators' and legislatures' activities for all 50 states. This data is accessible from Openstates.org. Appendix A illustrates the requisite steps to acquire the conference delegation information. 13 The conference delegations in my data set are those which are initially appointed. I do not account for conferences in which a single conferee may be removed or added. This seldom occurs across chambers, and the conditions under which legislators are removed from or added to conference delegations are unclear.
14 Connecticut's session journals make reference to committees of conference, though there is no information about their formation. 15 Maine and Ohio report when their committee of conferences form, but do not name individual legislators serving on the committees. 16 Oklahoma's Senate authorized 21 additional permanent standing conference committees in 2011 to handle all contentious legislation. These standing conference committees do not report individual membership, so Oklahoma's Senate data spans from 2005 to 2010. For example, bills with fiscal impacts are often referred to Oklahoma's General Conference Committee on Appropriations. Moreover, Oklahoma's General comes from 41 states and 82 legislative chambers across 942 chamber-years. In total, I tally 16,541 lower and 16,783 upper chamber conference delegations, resulting in 33,324 total groupings. Figure 2 shows the within-chamber changes in conference delegation trends for noncommemorative bills for all states which appointed named conference delegations between 2005 and 2016. 17 Notably, states have employed conference committees at consistent rates over time. 18 As expected, conferencing trends for lower and upper chambers within a state are quite similar over time. In most circumstances, the opposite chamber appoints a conference delegation when the originating chamber does. Additionally, states have employed conference delegations at consistent rates over time, with few states appointing conference delegations regularly. Hawaii employs conference delegations most often, with over 86% of their passed legislation going to conference. 19 Hawaiian legislators regularly serve on several conference committees concurrently, overwhelming legislators who have been quoted as needing "to be at four different places right now" during an ordinary legislative day (Blair 2011). In contrast, most states appoint conference delegations for less than 10% of passed bills. Appendix B offers information on the number of conference delegations appointed for each chamber.
To examine the relationship between conference committee appointment rights and chamber ideology for evidence of majority party bias, I focus on the median legislator in each delegation. However, standard difference-of-median tests for conference delegations are an insufficient method given that most delegations have small memberships (typically between three and five per state legislative chamber), inhibiting non-parametric difference-of-median tests (Freidlin and Gastwirth 2000). To circumvent this complication, I follow the empirical strategy of Vander Wielen and Smith (2011). They note that "if the number of conference delegations included in the analysis is large, it follows that the distribution of differences between the conference delegation and chamber medians… would have a mean of zero if conferees are indeed representative of the parent chamber" (279). By constructing a distribution of differences across medians, I can aggregate the delegations and plot them for a descriptive visualization of majority party influence in conference delegation composition. The resulting visualizations permit an analysis of whether the distribution of conference delegations deviates in a statistically significant manner from the expectations of my hypotheses. There would be support for hypothesis 1 if the mean of the delegation distribution equals the majority party median; evidence for hypothesis 2 would exist if the mean equals the chamber median. 20 Conference Committee on Appropriations preceded the additional standing conference committees, and spans both chambers. Thus, House conference committees were also omitted when necessary.
17 Data for noncommemorative bills between 2005 and 2016 come from Emrich (2022a) who explores the conditions under which legislative leaders leverage conference committees to reconcile bicameral differences. 18 These consistent rates of conference committees diverge from congressional studies which show that conferences are an increasingly rare method used to resolve bicameral differences (Oleszek 2010;Ryan 2011). 19 Hawaii has 8,476 conference groupings in the data set (4,169 House and 4,307 Senate). 20 This method assumes that conferences are randomly drawn from the parent chamber, which is unlikely to be true given majority party powers of appointing conferees and norms of deference to standing committee members during the selection process. Therefore, I use Wilcoxon signed-rank tests to nonparametrically assess whether conference delegation medians diverge from majority party and chamber medians. They demonstrate that the conference delegation and majority party medians are nonidentical populations (p < 0.001) as well as conference delegation and chamber medians (p < 0.001).
Similarly, I plot the distribution of differences between majority leadership appointed conferees and minority party influenced conferees to illustrate variation in majority party control conditional on conference appointer. To reiterate hypothesis 3, I expect conference delegations appointed solely by the majority party to be closer ideologically to the majority party median than minority influenced delegations. That is, minority party appointments limit majority party control over conference committees by widening the ideological distance between the conference delegation and majority party medians. Twelve state legislative chambers possess minority appointing rights throughout my data set, totaling 3,056 conference delegations across 127 chamber-years.  Poole and Rosenthal 1985). Mapping state legislators onto a single dimension allows for clear comparisons of the conference delegations relative to both the majority party and chamber medians. Additionally, higher order dimensions commonly pertain to parochial interests as opposed to partisan conflict (Miller and Schofield 2003). My analysis uses Shor and McCarty's (2011) common-space scores for state legislatures and legislators. These ideal points are generated from the fusion state-level roll call voting data and surveys of state legislative candidates, allowing scholars to make comparisons of interparty heterogeneity. For ease of interpretation, I rescale the Shor-McCarty scores so that the majority party's ideology within the chamber (whether Republican or Democrat) corresponds to positive values.
I supplement the distribution of ideological differences by modeling the contrasts between conference delegation and majority party medians. I test hypothesis 3 with the following generalized equation, estimating ordinary least squares models with two-way random effects of the following form (Smithson and Merkle 2013):

Conference IdeologyÀMajority Party Ideology
where Conference Ideology À Majority Party Ideology it is a continuous variable computed as the difference in Shor and McCarty (2011) common-space scores between the median conferee's ideology and the median majority party legislator's ideology. γ i represents state random effects to account for unobserved heterogeneity across states and δ t represents session random effects to sweep out differences common to a legislative session. ϵ it is the error term. By incorporating these random effects, conference delegations are treated as being nested within states and legislative sessions. 22 21 Idaho's and Texas' Senates also allow for the lieutenant governor to appoint conferees, but they have perfect partisan symmetry with their Senate majority parties from 2005 to 2016, making those years majority appointed.
22 Appendix E shows models with year random effects instead of session random effects. The results are unchanged across specifications.
My main independent variable β Minority Conference Appointer it is a dummy variable equal to one if the minority party in a state legislative chamber has influence over the conference appointing process in a given year, and 0 if the state chamber's majority unilaterally selects conferees. A negative value from this variable would suggest that minority party appointments limit majority party control over conference committees by broadening the ideological gap between conference delegation and majority party medians. Importantly, conference appointing rules do not vary within-state across my data set. That is, there are no state legislative chambers from 2005 to 2016 that switch their rules from the majority party unilaterally appointing conferees to minority influence, or vice versa. 23 ρ it controls for chamber-level factors that might influence the relationship between parties and conferees. First, I control for the heterogeneity of preferences across and within political parties from chamber to chamber. I measure the ideological distance between parties in state legislative chambers by using the Shor and McCarty (2011) interparty heterogeneity measure which is equal to the average distance between the median ideal points of Republican and Democratic legislators in each chamber. I gauge the diversity of preferences within the chamber by using Shor and McCarty's (2011) intraparty heterogeneity indicator, which is the standard deviation of majority party legislator's ideal points. Second, I account for the level of professionalism in the state legislature. There is likely a direct correlation between the professionalism of the legislature and the partisanship of legislators, whereby highly professionalized legislatures likely have legislators with stable partisan preferences (Battista and Richman 2011;Fiorina 1999;Ryan 2014). I capture each state's professionalism with the Squire Index (Squire 2017), which is a weighted combination of salary, days in session, and staff per legislator relative to members of Congress. 24 Lastly, I control for political factors like intercameral agreement and disagreement by estimating separate models with binary variables equal to one for when the lower and upper chambers are unified or divided. Party leaders undoubtedly consider the preferences of the opposing chamber when deciding on delegation composition since a majority of both chambers must agree on the conference report. Figure 3 shows the distribution of differences between conference delegation and majority party medians for all state legislative chambers between 2005 and 2016. Likewise, Figure 4 provides a density curve of the distribution of differences between conference and chamber medians. Both diagrams contain vertical dashed lines to indicate the aggregated conference delegation ideology. Figure 3 demonstrates a statistically significant divergence between conference delegation and majority party medians. Substantively, the average conference delegation is 0.262 points away from the majority party median, over one-sixth the mean ideological distance between the majority and minority parties. 25 This result shows that the average state chamber conference delegation does not mirror its majority party median, offering no support for hypothesis 1.  Figure 3 shows the distribution of differences between the conference delegation and majority party medians on the Shor-McCarty common space scale. Mean = À0.262, t = À102.9, p < 0.001, 95% CI: (À0.268, À0.258). Note. Figure 4 shows the distribution of differences between the conference delegation and chamber medians on the Shor-McCarty common space scale. Mean = À0.01, t = 4.39, p < 0.001, 95% CI: (À0.015, À0.006). 25 The average interparty heterogeneity between majority and minority parties from 2005 to 2016 was 1.479 points on the Shor-McCarty common space scale.

Results
In contrast, Figure 4 shows that on aggregate there is barely any difference between conference delegation and chamber medians. Although significant, the average distance between conference and chamber medians is a trivial 0.01 points on the ideological scale, and this gap is in favor of the minority party. Therefore, descriptive evidence suggests that state-level conference delegations holistically reflect Krehbiel's (1993) claim that conferees reflect the chamber's median voter, providing evidence for hypothesis 2.
It is possible that the results from Figures 3 and 4 may be warped by those conference delegations unrepresentative of the majority party and chamber (e.g., the outliers with Shor-McCarty differences less than negative two). It is unclear why these delegations stray so significantly from the majority party and chamber. For example, Colorado possesses 19 of the 21 most outlying delegations with respect to the majority party median, despite its majority leadership unilaterally appointing conferees. As a robustness check, Appendix C replicates Figures 3 and 4 using ideological differences between conference delegation and majority party/chamber medians restricted to plus/minus 0.5 on the Shor-McCarty scale. Substantive results regarding distance between average conference and majority party medians are unchanged regardless of ideological threshold.
An additional robustness test accounts for how frequently a state legislative chamber forms conference delegations and potential overweighting from those few states that disproportionately use conference committees. Appendix D reproduces Figures 3 and 4 with the removal of Hawaii's House and Senate as well as the removal of Hawaii and Mississippi's respective legislatures. Results are unchanged across all specifications.
To test hypothesis 3, Figures 5 and 6 show distributions of the differences between conference delegation and chamber/majority party medians, conditional on conference appointer. Like Figures 3 and 4, the density curves contain vertical dashed lines to indicate the average conference delegation ideology. 26 The top panels for Figures 5  and 6 refer to the distribution of conference delegations' median legislator when the majority has unilateral conference appointing rights, and the bottom panels show the distributions when the minority can influence delegation composition.
Both figures support hypothesis 3, as the average conference delegation is distanced from the majority party's preferences when the minority party possesses conference appointing rights. Figure 5 provides clear evidence for hypothesis 3, as minority conferee appointing rights shifts the average conference delegation in the direction of the minority party nearly one-third of the mean ideological distance between the majority and minority parties. 27 Figure 6 provides more muted effects for how minority appointing rights influence conference delegations relative to chamber medians. Substantively their presence shifts the conference delegation less than 5% of the mean ideological distance between majority and minority parties.
To further examine hypothesis 3, the results of the random effects models are provided in Table 1. 28 Overall, the models corroborate that minority conference appointer rights limit majority control over conference delegations. Model 1 shows the bivariate relationship between minority appointer rights and conference delegation ideological composition, illustrating that minority influence has a statistically significant effect on reducing majority control. The presence of minority appointer rights has a similar substantive effect to the difference shown in Figure 5, corresponding to a nearly one-third shift in interparty heterogeneity between the majority and Figure 5. Ideological distance between conference delegation and majority medians conditional on appointer rights. Note. Figure 5 shows the distribution of differences between the conference delegation and majority party medians on the Shor-McCarty common space scale conditional on conference appointer. Majority only mean = À0.228, Minority rights mean = À0.574, t = 36.78, p < 0.001, 95% CI: (0.328, 0.365). Figure 6. Ideological distance between conference delegation and chamber medians conditional on appointer rights. Note. Figure 6 shows the distribution of differences between the conference delegation and chamber medians on the Shor-McCarty common space scale conditional on conference appointer. Majority only mean = À0.007, Minority rights mean = À0.039, t = 4.39, p < 0.001, 95% CI: (0.018, 0.046). minority parties. 29 This finding supports hypothesis 3 and is shown in Figure 7, which presents model 1's estimates of the average marginal effect of minority appointing rights on conference delegation ideological composition. Models 2 and 3 mitigate against confounding omitted variables by incorporating controls to capture state-and chamber-level political conditions. Both include the interparty heterogeneity, intraparty heterogeneity, and legislative professionalism variables, but differ with respect to partisan control of the legislature. Model 2 tests for the effects of a split legislature, whereas model 3 examines unified legislatures.
Several control variables are uniformly significant in predicting differences between conference and majority party medians for all alternative model specifications. Increases in interparty heterogeneity widen differences between conference delegation and majority party medians, consistent with the notion that interchamber differences stimulate moderate behavior. When parties are farther apart, conference delegations must be sufficiently broad to forge compromise and attract a larger coalition within the chamber for approving the conference report (Ryan 2014). Conversely, rises in intraparty heterogeneity within the majority party corresponds to conference delegations more similar to the majority party. While a full account of conference appointment strategies conditional on preferences within the majority party is beyond the scope of this article, the majority leadership likely seeks to maximize its policy goals by pursuing partisan advantage in its conferee selection when faced with heterogeneous coalitions within its party (Lazarus and Monroe 2007). The level of professionalism has no impact on the difference between conference and majority party medians, as its effects do not reach statistical significance in models 2 or 3. Partisan control of the legislature matters slightly in determining the distance between conference and majority medians. The presence of a split or unified legislature slightly shifts conference delegation medians away from majority party medians.

Discussion
Using an original data set of state legislative conference committees over a decade, my results suggest that on aggregate there is no pro-majority slant in state legislative conference delegations. Instead, conference groupings are ideologically centered on the chamber median, providing evidence that conferences serve as an informational tool for the chamber (Krehbiel 1991;1993). This finding contrasts from congressional-level research which demonstrated a clear partisan bent in the representativeness of conference committees (Vander Wielen and Smith 2011).
Moreover, I leverage variation in institutional rules across U.S. state legislatures to assess the significance of minority conferee appointing rights on how representative conference delegations are with respect to the majority party. When examining chambers that permit minority influence over conference composition, I find that the majority party is limited in the conference appointing process. If conference delegations independently pursue their interests (Vander Wielen 2010), these findings speak to the policy implications of minority influence over the conference stage, consistent with prior work that demonstrates how majority parties are hindered when legislative rules advantage the minority party (Ryan 2014). This article focuses on a single institutional rule associated with conference committees, building upon recent research which has investigated the conditions under which state legislatures go to conference (Emrich 2022a). Unlike Congress which seldom goes to conference in recent sessions due to surging polarization (McCarty, Poole, and Rosenthal 2016;Ryan 2011;Shor and McCarty 2011), thousands of conference committees are appointed annually in the states. Further research could explore why many of these conference committees fail. This phenomenon is well-studied by scholars of the U.S. Congress, yet little is known about the circumstances leading to failure in the states. Nearly 30% of state legislative conference committee reports fail to pass the legislature. Moreover, over a third of conferenced bills fail to become law (Emrich 2022a). Although conference committees continue to be appointed, these failures underscore the difficulties of legislating in the modern, polarized context.

A. Appendices Appendix A-Open States Data Appendix
Most of the data from 2011 through 2016 for this article comes from Open States, a nonprofit organization that leverages crowd-sourcing and web scraping to compile data on legislators' and legislatures' activities for all 50 states, Washington, D.C., and Puerto Rico. The code responsible for scraping bills and votes from Open States can be found on the Open States Github.
Broadly speaking, the Open States repository tracks bills, reviews upcoming legislation, and provides billlevel information on how state-level representatives are voting. For the purposes of this article, I leverage Open States' public domain bulk data. Specifically, I utilize Open States' Legacy CSV data which draws from the Open States API v1 which was last updated on November 3rd, 2018. The Legacy CSV files are a CSV transformation of the data available on Open States' Legacy JSON archives. The Legacy CSV files can be found here.
Within the Legacy CSV files are several .csv folders for each of the 50 states: legislators.csv, legislator_roles. csv, committees.csv, bills.csv, bill_actions.csv, bill_sponsors.csv, bill_votes.csv, and bill_legislator_votes.csv. I leverage the bill_actions.csv to provide information on conference committee delegations. Each row within a bill_actions.csv provides an additional step in the legislative process for a bill. As such, conference committee delegations were uniquely identified by legislative session using the "session," "chamber," "bill_id" and "action" columns. Specifically, the "action" column often indicated the specific members of a conference delegation.         A.1.7. Table results using year random effects