When Losers Work Together

When Losers Work Together: Opposition Party Unity in the U.S. House Across Time Jason Seitz University of Georgia Paper prepared for the Annual Conference of the Southern Political Science Association January 8-10, 2003 Abstract: What are the specific factors affecting the level of unity found within a given opposition party in Congress? This paper seeks answers to this question. The role of political parties in the legislative process has received much attention in the scholarly literature in recent years. One important debate deals with the ability of political parties to achieve their partisan goals through the manipulation of party voting in Congress. While many have suggested that political party affiliation greatly influences the voting behavior of members of Congress, others have maintained that party has only limited influence. In this paper I argue that this divergence in opinions stems from the two key deficiencies: one conceptual, and one analytical. Conceptually, researchers have falsely dichotomized the relationship between ideology and party when discussing the voting behavior of members of Congress. Analytically, scholars have failed to pay attention to the dynamics of the minority party. While much has been offered as to what causes the majority party in Congress to vary in its level of voting coordination, the specific dynamics of partisanship with regard to the opposition (or minority) party in Congress is ignored by most working in the area of congressional partisanship. The extent to which the minority party opposes the majority in the policy-making process contributes to the moderation of final policy outcomes, which consequently reflect more closely the preferences of the floor as a whole, and ultimately the electorate as a whole. Thus, the factors that promote and prohibit the minority party’s role in the legislative process are essential for understanding how public policy is formed. To this end, this project seeks to understand the dynamics of opposition party unity. I will argue that there are several key factors central to the varying levels of opposition party unity on roll-call voting. These include the presence of divided government, the unity of the majority party, interparty heterogeneity, the amount of time out of majority and the presence of a one chamber majority. Each of these factors is shown to significantly impact the level of opposition unity. “The opposition should so conduct itself in Parliament as to persuade the people of the country that it could be an improvement on the government of the day. No one will deny that our system works best when there is a change of government at reasonable interval” (Knowles 1957) The role of political parties in the legislative process has received much attention in the scholarly literature in recent years. One important area of debate deals with the ability of political parties to achieve their partisan goals through the manipulation of party voting in Congress. Whether using the framework of “Cartel Theory” or “Conditional Party Government,” many have suggested that presence of political parties in Congress significantly influences the legislative process (see Cox and McCubbins 1993, 2002; Rohde 1991; Aldrich and Rohde 1998).1 Others have maintained that party has only limited influence. Instead, these scholars hold that ideological preferences rather than partisan manipulation guide legislators’ vote choices during the process of forming a bill into law (Influential works include Brady et al, 1979; Krehbiel, 1993, 2000; Levitt, 1996; McCarty et al, 2001; Schickler and Rich, 1997). This discrepancy has led to little agreement among political scholars as to the role that political parties play in the legislative process. This divergence of opinions stems from the neglect of two key deficiencies: one conceptual, and one analytical. I will discuss these two limitations in greater detail below, but, briefly, the conceptual limitation in the literature stems from the false dichotomization of the relationship between ideology and party when discussing the voting behavior of members of Congress. Furthermore, there is an analytical limitation in the preponderance of this body of literature. This deficiency stems from the fact that when discussing the notion of political parties in Congress, scholars have systematically neglected the presence of the minority, or opposition, party. The minority (opposition) in Congress is the group of legislators within the political party that holds a minority of the seats for each chamber. Current theories of congressional parties focus on the inner workings of the majority party and fail to examine the dynamics of minorities. 1 Binder, 1996; Binder et al, 1999; Crook and Hibbing, 1985; Epstein, 1999; Jenkins, 1999; Jones, 1968; Kerr, 2000; Nokken, 2000; Patterson and Caldeira, 1988; and Snyder and Groseclose, 2000 -1- Figure 1. Party Unity Across Time 90 88 86 84 Party Unity 82 80 78 76 74 72 70 10 1 10 3 10 5 63 65 67 69 71 73 75 77 79 81 83 85 87 89 93 91 95 97 99 Congressional Session Minority Unity Poly. (Minority Unity) Note: The measure of minority party unity used here from a dataset compiled by Cooper and Young (2002a). Keeping these two points in mind, we can now turn to the framework of this paper. I will argue that the role that political parties play in the legislative process must be found in the aggregate rather than the individual because parties are aggregate entities. Thus, party unity rather than party influence on individual votes is central to understanding the weight of political parties in Congress. At the same time, my concern in this paper will center on the minority party in Congress because of the role of oppositions in the process. That there has been a neglect of minority parties in theories of congressional party influence adds to the import of this work as well. As the introductory quote by Stanley Knowles concerning the role of oppositions affirms, minority parties do have a significant responsibility in legislative institutions – they offer a clear choice and unified alternative to the present administration. -2- The level at which the minority party acts coherently, as a party, then is important for the functioning of representative democracy. Without choice, democracy does not exist. Thus, the central concern of this paper is exploring the variation of opposition party unity across time. We have reason to believe that minority party unity varies across time. Indeed, Figure 1 above shows the party unity scores for the minority party in the U.S. House of Representatives from the 63rd through the 106th Congress (1913-2000). Certainly, we can see considerable variation in these scores. There are various important implications that can be attributed to this variation. Certainly, the extent to which the opposition party is able to demonstrate its desires and act coherently is an important link in understanding the legislative process. To this end, this project seeks to understand the exact dynamics of opposition party unity. This paper will argue that there are five significant factors that influence the behavior of the opposition party. These are the presence of divided government, the unity of the majority party, interparty heterogeneity, the amount of time out of majority and the presence of a one chamber majority. Each of these factors is expected to significantly impact the level of opposition party unity on roll call voting. I will begin with a review of the pertinent literature followed by an exposition of the major hypotheses and theoretical structure. I will then proceed to a discussion of the operationalization and methodological design and finally, to a summary discussion of the significant findings. Review of the Literature Allow me to structure this review in light of the two major concerns expressed above that serve as the primary impetus for this paper: the conceptual limitations of the party/ideology dichotomy and the analytical absence of studies concerning minority parties. First, let us address the literature concerning individual voting behavior in Congress. Various scholars -3- have argued that political parties do play an important role in shaping how legislators vote2. Nokken (2000) looks at the voting records of party switchers and argues that, "if parties do significantly constrain members' voting behavior, drastic changes in roll-call behavior should coincide with changes in party affiliation" (423). Furthermore, against the ideological camp, he argues, "If ideology is the primary determinant of roll-call voting, one should not observe drastic changes in roll-call behavior when members switch parties" (422). His analysis supports his theoretical claims, and he concludes that, "Party switchers apparently do not 'die in their ideological boots.' Instead, members migrate to the other party where, after pledging their support to that party, they die in a different set of boots, partisan boots" (438). Elsewhere, Crook and Hibbing (1985) looking at the effects of changes in the seniority system of committee chairs on party loyalty conclude that, "the removal of seniority from the list of universalistic rules of congressional procedure has led several key Democrats to toe the party line to a much greater degree than they would have if adherence to the seniority system had continued unabated" (225). As evidenced by these few examples, the thrust of this strand of the literature holds that partisanship plays a significant role in the voting behavior of individual legislators. Note, however, that no specific attention is paid to those factors that lead political parties to be unified in the aggregate. Others have argued that there is no partisan effect on legislator's voting behavior3. One important example is the work of Schickler and Rich (1997) who argue that, "the majority party can credibly threaten to punish its dissidents precisely when the party least needs to do so: when it has a substantial majority and is relatively unified" (1342). To this end, they find that any sanctions seen, "had far more to do with deterring open disrespect for public symbols of the party, than with safeguarding the majority party's procedural control of the 2 As previously mentioned these include Binder, 1996; Binder et al, 1999; Cox and McCubbins, 1993; Crook and Hibbing, 1985; Epstein, 1999; Jenkins, 1999; Jones, 1968; Jones, 1970; Kerr, 2000; Nokken, 2000; Patterson and Caldeira, 1988; and Snyder and Groseclose, 2000. 3 Some important examples include see Brady et al, 1979; Krehbiel, 1993, 2000; Levitt, 1996; McCarty et al, 2001; Schickler and Rich, 1997 -4- House" (1371). So, they would conclude that the inability of the party to discipline its members would lead legislators to simply vote their preferences. Party influence only exists when it is not needed – when the majority party is large and homogeneous. Another work arguing this same notion is Levitt (1996). Levitt looks at the U.S. Senate and finds that while party does play a role in legislators’ vote choices; it is "ideology [that] is the primary determinant" (438). In addition, Krehbiel (2000) is a prominent advocate for the preferencebased theory of legislative voting. He maintains, "The well-known measure of party voting is incapable of discriminating between nonpartisan and partisan behavior… [Measures of partisanship] seem to be artifacts of preferences rather than evidence of party discipline, party cohesion, party strength, or party government" (225). This alternate strand of the literature maintains that the individual preferences of legislators are largely responsible for understanding the voting behavior of legislators. Clearly, we can see that there is no consensus as to whether legislators vote because of their partisanship or because of their ideological preferences. Why might this be the case? As mentioned above, it is my position that this is largely due to the misconceptualization of the nature of party in the literature. Whereas many would argue that individual congressional voting stems from either partisan loyalty or personal preference, this researcher sees this division as largely enigmatic. What should be made clear is that when addressing the notion of political parties, we are dealing with aggregations of individual preferences, and thus, this distinction between party and ideology need not be made. Preferences play a role at the individual decision level, and political parties are important at the aggregate level. The place to look for the importance of political parties in the legislative process is not in the individual voting record but in the extent to which the collective individuals making up the party are able to sufficiently coordinate their actions to affect political outcomes. Stated differently, the party cannot be found in the individual, however, the aggregation of these individual preferences results in partisan effects. By thinking of partisan affects in this manner, we are -5- better able to recognize how political parties might be important in the legislative process. Thus, we can reconceptualize the central question that arises when addressing how political parties are active in the formation of public policy. More fittingly, we should ask, what factors affect the extent to which members of political parties are able to act in a coherent manner in Congress? In order to understand how political parties might affect the legislative process, we must first fully understand those factors that account for the varying degree to which these members of legislative parties are able to coordinate their individual preferences into coherent partisan policy positions. Given this new framework, we can now address the second weakness of much of the literature. The second, analytical limitation resulting in the observed inconsistency of the scholarly literature stems from the fact that the research question has only been partially addressed. Surprisingly, a puzzling omission is found in this literature. While much has been offered as to what causes the majority party in Congress to vary in its level of voting coordination, the specific dynamics of partisanship with regard to the opposition or minority party in Congress is ignored by most of the scholars working in the area of congressional partisanship. While not in a position to dictate the policy-making process, the minority party does play an important role in opposition. We can think of the opposition’s role as a mediating one. The extent to which the minority party opposes the majority in the policy-making process has the ability to moderate final policy outcomes. Moderate policy, consequently, has the potential to more closely reflect the desires of the electorate as a whole because it reflects more closely the preferences of the floor as a whole. It seems quite reasonable, then, to suggest that the role of opposition parties in democracy institutions is an important one. Thus, the factors that promote and prohibit the minority party’s role in the legislative process are essential for understanding how public policy is formed. -6- Except for two works by Charles O. Jones in 1968 and 1970, there has been an utter neglect4 of any distinction between the majority and minority parties in Congress. It is the genuine impression of this researcher, however, that there are tangible distinctions between the party dynamics of a political party controlling a majority of the floor and one existing in the minority. Surely, the diminished status held by the opposition party in the legislative process would lead one to believe that there are distinct factors that shape the ability of the opposition party to act cohesively and vote as one. Jones (1968, 1970) has been the lone voice discussing the merits and workings of the minority party in Congress. However, the attention that he has paid is unquestionably outdated. Moreover, his goal is not specifically to understand the dynamics of party voting. Rather, his purpose is to show that, "the minority party's role in the House of Representative is by no means a consistent role over time" (1968, 492). He points to various factors including minority party strength, minority party unity, minority party leadership, presidential power, and majority party leadership which "determine the range of strategies available to the minority in congressional policy making" (492). While discussing how the minority party applies party unity, he does not make it clear as to what factors lead to the varying levels of party unity over time, and that is our goal in this research project. Two of the dominant theories of congressional parties, “Cartel Theory” and “Conditional Party Government” center on the ability of the majority party to control its members and get what it wants – partisan policy outcomes. Cox and McCubbins’ (1993, 2002) “Cartel Theory” holds that “The majority party acts as a structuring coalition, stacking the deck in its own favor – both on the floor and in committee – to create a kind of ‘legislative cartel’ that dominates the legislative agenda” (270). 4 There have been some recent works concerning the rights of minorities in Congress. Binder (1997) develops a partisan theory of institutional development that, in her opinion, explains the past history of minority rights in Congress. Krehbiel and Meirowitz (2002) argue that the development of minority (and majority) rights are a function of individual preferences rather than partisan conflict. These works do not however address the dynamics of minority party unity. -7- The theory Conditional Party Government was developed by Rohde (1991) and extended by Rohde and Aldrich in several other papers including 2002. Their argument suggests that party government is conditional on three main factors: ideologically homogenous party memberships, institutional structures and rules designed to facilitate majority rule, and leaders willing to exercise authority as the instrument of their party caucus. What is distinctive about these theories, at least for the purposes of this paper, is that neither discusses the potential for the opposition party to have any influence in the political process. Whether the majority party uses its leadership to run the floor like a “cartel” or conditionally controls the floor based on the relative homogeneity of the party and the willingness of party leadership to maintain the unity of the party, there is clearly no place for the party with a minority of the seats in Congress in these theories. They simply respond to what the majority party does. What should be clear from this brief discussion of the relevant literature is that there are real limitations with previous research. The question of party influence in Congress has been met with discrepancies because of the failure to recognize that party influence is an aggregate phenomenon and not an individual one. Additionally, the relevant research regarding parties in Congress is limited to the majority party because of the systematic neglect of opposition parties. Because of these limitations, a new theory capable of answering the question, “What factors affect the extent to which the opposition party is able to act in a coherent manner in Congress?” It is answers to this question that this paper seeks. Theory and Hypotheses Now that I have discussed what others have said about party discipline in Congress, I will briefly address the expected relationships we propose and the corresponding theories behind each. The dependent variable that we are concerned with is the level of unified voting within the opposition party on roll call votes. The exact nature of this variable will be further -8- explained below, but in short this variable is the “average percentage of the minority party who voted with their party majority” (Cooper and Young 2002b, p. 2). Party votes are “votes where at least 50% of Democrats opposed at least 50% of the Republicans (p. 2). Again, the question that we are asking is “what are the specific factors that led to the varying degree of unified voting on roll call votes by members of the opposition party.” It is the position of this paper that there are five significant factors related to the legislative process that affect the extent to which the minority party acts in unison. Briefly, these are the presence of divided government, the unity of the majority party, interparty heterogeneity, the amount of time out of majority and the presence of a one chamber majority. What is important to think about in each case is its relationship to the proposed pieces of legislation that the opposition party will be voting on (The argument for the Time Out of Majority variable follows a different logic). Each of these factors, it is argued, either moderates the shape of these proposed bills or generates policy proposals that are more partisan in nature. Depending on which dynamic is at play, we would expect them to either decrease or increase, respectively, the level of unity in the voting behavior of the opposition party. With this in mind let us turn to a discussion of these factors. Hypothesis 1: The presence of divided government (the executive and majority party in at least one chamber of Congress occupied by conflicting parties) will lead to a decrease in the level of opposition party unity. Divided government, the phenomenon of split party control of the executive and at least one branch of Congress, is expected to have a negative impact on the level of opposition party unity. During periods of divided partisan control of government, the majority party in Congress must propose legislation that anticipates how the executive will respond, passage or veto. Thus, we would expect policy proposals to be more moderate (centrist) in nature. -9- Figure 2 – Unified Government Policy Shift OD SQ MR ER Policy Shift OD SQ ER MR Figure 3 – Divided Government OD ED SQ MR ED OD SQ MR -10- Without going into great detail, Figures 2 and 3 below present two simple cases of divided and unified government that should make this dynamic clearer. In this one-dimensional policy space, I assume that legislators have single-peaked, symmetrical preferences5 allowing us to equate distance with utility. That is, given the executive in our example (ER), as policy proposals move further away from his ideal policy as designated on Figure 2, say to the majority’s ideal point (MR), utility decreases. Thus, for our executive, the status quo (SQ) policy is preferred to the opposition’s ideal policy proposal (OD) because it is closer to the executive’s ideal point. Certainly, these simplistic spatial models are ultimately caricatures of the real legislative environment, but they do allow me to shed light on the notion that policy proposals will be more “moderate” at least as the opposition party is concerned during periods of divided government as opposed to unified government. Please allow me to elaborate. In Figure 2, Unified Government, we see two examples of unified government with two arrangements of the ideal points of the Majority Party (MR) and the Executive (ER). Also included are the ideal point of the opposition party (OD) and the status quo policy (SQ). The two possibilities here account for the possibility that the executive could be either closer to the opposition party or further away than the majority party. In this simple model, we can see that in each instance policy proposals made by the Republican majority can approach its on policy ideal point to the extent that it is a policy proposal that is preferred by the executive to the status quo policy. Thus, any policy that is closer to the majority and the executive than the current policy (SQ) will be proposed because it will have both the majority in Congress and the executive preferring it to the old policy. So, in the first case of unified government, the majority party has the potential to shift policy to its ideal point because it is a point that will be preferred by the executive to the status quo. Alternatively, the majority party has the potential to shift policy almost to its ideal point. Remember that the policy shift is constrained 5 Additionally, I am assuming that legislators have a unique ideal point in the policy space and that our major actors ER, MR, and OD are unitary actors. That is, MR represents the ideal point of the majority as whole. -11- by the ideal point of the executive as well (because of the executive veto) so any policy shift must be closer to the executive than the status quo policy or else it would be turned down by the executive. We can conclude then that policy proposals during unified government have the potential to be more partisan (and less moderate) in nature because the executive is of the same party. Figure 3, Divided Government, provides an alternate illustration. Here the same actors are aligned in a policy space with the executive now being of the same party as the opposition in Congress. The two cases represent the two possibilities that exist6 – either the executive being closer ideologically to the majority than the opposition party ideal or vice versa. In this instance, there can be no legitimate policy shift away from the executive towards the majority party because any policy proposal further away from the executive than the current policy would be vetoed. Therefore, we would expect that policy proposals by the majority party will tend to be more centrist in nature lest they chance an executive veto subsequent to passage. How then, will this affect the party unity of the opposition party? Under unified government, the executive and majority party are of the same party and very similar ideologically, therefore we would expect policy proposals to reflect a very partisan preference. Fewer members of the opposition party would find favor in a proposed bill that largely reflects the majority party’s preferences. Consequently, we would expect to see greater levels of opposition party unity during times of unified government. Alternately, during periods of greater division of the partisan control of government (divided government), policy proposals will be more moderate in nature because of the balancing affect the executive has on the majority party. In these instances, we would expect unity of the opposition party to decrease 6 Certainly, the possibility exists that the opposition party and the executive could have identical ideal points, but the insight of this model would be the same. -12- because proposed bills would be moderate enough to sway some opposition party members away from voting with their own party. Therefore, during times of divided government, we would expect there to be less coordination among the members of the minority party and lower levels of opposition party unity. Hypothesis 2: As the unity of the majority party increases, the level of opposition party unity will increase. Another factor that should affect the relative level of opposition party unity is the corresponding unity of the majority party. Again, it is important to note how proposed bills will look as the majority party becomes more unified in its actions. Certainly, we would expect that the closer ideologically the majority party becomes the more partisan (reflective of the majority party’s ideal point) policy proposals will be. The logic behind this belief is that as the majority party leadership becomes more certain about the guaranteed number of votes they have within their own party, proposed legislation can move away from the floor ideal point and towards the party ideal point because they need to sway fewer members of the opposite party to ensure that the proposal will succeed. Additionally, a unified majority party will not actively seek opposition party members to support majority party legislation (at least not to the same degree as a less unified majority) because they already have the number of votes within their own party. Because proposals will be both more partisan and opposition party members will be less likely to be courted in order to pass legislation, ideologically unified majority parties will in turn promote greater unity within the opposition party. Faced with no adequate alternative the minority party will collectively vote in opposition to the majority party’s proposed legislation. As a result, I expect that as the unity of the majority party increases, the unity of the minority party will also increase. -13- Hypothesis 3: As the majority and minority party become more ideologically divergent, the level of opposition party unity will increase. The ideological distance between the majority and opposition should also positively affect the level of opposition party unity. Again, we must think of this dynamic as one centered on the characteristics of the proposals offered to the minority party. As the majority and minority become more ideologically distinct, we would expect that bills proposed by a more distant majority to be less advantageous for opposition party members and thus, defection to be less likely. Stated differently, the greater the ideological distance between majority party and opposition, the more divergent their opinions about public policy will be. Therefore, what becomes more advantageous to the majority party becomes increasingly objectionable to the opposition party. Hence, we should expect that as the two parties increasingly differ ideologically, the opposition party would increasingly contest proposed legislation in a cohesive manner. Hypothesis 4: During periods of a split-party control in Congress (when the majority party in the Senate is not the same as the majority in the House of Representatives), I expect there to be lower levels of opposition party unity. Periods of a split-party control in Congress7 (when the majority party in the Senate is not the same as the majority party in the House) should also impact relative levels of opposition party unity. This hypothesis is understandably related to the argument above about divided government. The thought is that the check on policy proposals that the other chamber presents to the majority party of the initial chamber provides a moderating effect as the two become more ideologically divergent. During periods of split-party control, we would expect the two chambers to be quite divergent concerning ideal policy. As has persistently been the case thus far, what is important to think about is the nature of the policy proposals presented to the opposition party. Because legislation must pass, in identical form, both chambers of 7 Like the 107th Congress and the first three congressional sessions of the Reagan administration. -14- Congress, we would expect that the majority party in either chamber would anticipate the eventual scrutiny a proposed bill would receive in the other chamber. Coincidentally, legislation put forward by the majority party in one chamber must fall within the range of suitable possibilities. When the other chamber is controlled by the opposition party, we would expect to see a moderation of bill proposals. This moderation pulls legislation towards the opposition party’s own preferences and thus coordination for the minority becomes more convoluted. And so, we would expect that during periods of split-party control in Congress, the opposition party would also lessen the intensity with which it collectively opposed the majority party. Hypothesis 5: As the length of time a given opposition party has been in the minority (as opposed to the majority) increases, I expect that that party’s unity to increase. The logic behind the inclusion of this variable is somewhat different from the other four concepts I have discussed so far. I would argue that as the length of time a given minority has been out of majority increases, the desire of the party to get back into the majority will also increase. Following the logic of Cox and McCubbins (1993) and Aldrich (1995), a coherent party label capable of providing significant information is one way to increase those chances. As Aldrich puts it, “Parties formulate policies to win elections rather than winning elections to promulgate policies” (13). Additionally, Cox and McCubbins argue that there are incentives for legislative parties to act cohesively because “the party’s reputation, based on its record is a public good for all legislators in the party” (123). Aldrich concurs when he says “Affiliation with a party provides a candidate with, among other things, a ‘brand name.’ In advertising, successful brand names convey a great deal of information cheaply: they cue an established reputation” (49). So, we can think of the opposition party’s label as a collective good that is used to provide information to voters. The extent to which this label provides consistent information should -15- affect the electoral success of the opposition party. If, on the one hand, the opposition party cohesively opposes the majority party in Congress, the party label sends a clear message to voters and provides dependable information about what that party represents. Accordingly, voters will reward such a party because they represent a distinct choice from the majority party. If, on the other hand, an opposition party does not exhibit a high level of unity, then no clear message is sent. If a significant proportion of the minority party time and again side with the majority party on policy issues, then the information conveyed by the party labeled becomes inconsistent and illogical. Thus, as the time a minority party has been out of the majority increases, the desire of the members in that party to regain the majority will increase as well. So, as this desire to regain the majority increases over time, we would expect the cohesion of the minority party to also increase because members will want to maximize the good information provided to voters by the party’s “brand name.” In summary, I argue that there are five highly important factors that account for the varying degree of unified voting found in the opposition party in Congress. These are the presence of divided government, the corresponding unity of the majority party, the ideological divergence of the two parties within the House and those periods when there is partisan conflict across the House and Senate, and the length of time the minority has been out of the majority. Now that we have laid out the main hypotheses of this analysis, let us discuss how we might meticulously evaluate these theoretical claims. Operationalization and Methods We can now turn to the operationalization of the concepts theoretically linked to the varying levels of opposition party unity followed by a discussion of the statistical analysis. For the testing of these hypotheses, I will be using the Prais-Winsten transformed regression estimator (1954) found in the Stata 7.0 software package because of the time-series nature of our variables. This estimator corrects for first-order serially correlated residuals which we -16- would expect to find in time-series data8. The dependent variable, Opposition Party Unity, will be quantified using party unity scores compiled from the Cooper and Young (2002b) dataset. Their work provides the percentage of members voting with a majority of their party on Party Votes. As I mentioned before, Party Votes are defined as “roll call votes where at least 50% of the Democrats opposed at least 50% of the Republicans” (2). The party unity score for each opposition party, by congressional session, from the 63rd to 106th congressional session will be used making the sample size 44 cases. This variable suits our theoretical analysis particularly well because it captures the extent to which the opposition party collectively opposes the majority party. Remember that much of the argument I have presented rests on the notion that as the policy proposals presented to the minority party becomes more skewed (from their collective policy preferences), so to will the unified nature of the opposition increase. Now let us look at the construction of the independent variables. For the Divided Government variable, I have used a simple dichotomous variable with 1 representing periods of divided government and 0 otherwise. The Majority Party Unity score was also available in the Cooper and Young (2002b) dataset and is calculated the same as opposition party unity above. The Interparty Distance variable was calculated, using DWNOMINATE scores (Poole and Rosenthal 1997), by finding the party median for each party and calculating the absolute distance between the two for each congressional session. The One House variable is a dichotomous variable as well with the variable equaling 1 during those periods when the Senate was controlled by a different party than the House and equaling a 0 otherwise. The Time Out variable is simply a count of the number of congressional sessions a given opposition party has been in the minority. Now that the variables have been laid out, we can turn to the statistical analysis. 8 The Breusch (1978)-Godfrey (1978) Lagrange multiplier test for nonindependence in the error distribution found in Stata 7.0 yields a Breusch-Godfrey Chi-square under the null that the errors are independent. An analysis of our data yields a significant Chi-square of 6.85 (p=0.0089). Thus, we must reject the null that the errors are independent. The errors in this analysis would be first-order serially correlated, and so the Prais-Winsten estimator is a better alternative than OLS. -17- Table 1. Prais-Winsten Regression with Robust Standard Errors Variable B S.E. t-statistic Divided Government Interparty Distance Majority Party Unity Time Our of Majority One House Majority R2 Durbin-Watson (Original) Durbin-Watson (Adjusted) N= -3.23 9.56 0.27 0.01 -0.71 .80 1.20 2.24 44 0.998 7.612 0.131 0.100 1.326 -3.23** 1.269 2.04* 0.08 Note: **Denotes significance beyond the 0.01 level (two-tailed), * Denotes significance beyond the 0.05 level (two-tailed). Analysis Table 1 above presents the results of the Prais-Winsten Regression with robust standard errors. At first glance, the model performs very well accounting for 80 percent of the change in the unity of the minority party voting. The adjusted Durbin-Watson of 2.24 suggests that there is no problematic level of autocorrelation in this transformed Prais-Winsten analysis. Multicollinearity diagnostics were performed and no problematic levels were found10. All of the variables in the analysis have the correct sign. The Divided Government variable has a noteworthy effect on the dependent variable with a t-value of –3.23, which is statistically significant beyond the 0.01 level. During periods of divided government the level of opposition unity is decreased by around 3 percentage points. Majority Party Unity also has a statistically significant impact with a t-value of 2.04. While Interparty Distance, Time Out, and One House all have the correct sign, none of their effects are statistically discernible from zero. Overall, the model performs very well and explains a substantial proportion of the variation in opposition party unity with a very parsimonious model containing five variables. Because Interparty Distance carries the correct sign, we can say that this variable has a statistically significant effect at the 0.05 level (one-tailed). 10 An OLS regression was run to assess multicollinearity diagnostics and no problems were found. The highest VIF was 1.55, well below any problematic level. 9 -18- Conclusion The focus of this paper has attempted to answer the question: What factors affect the extent to which political parties are able to act in a coherent manner in Congress? For this project I was specifically concerned with the dynamics of the opposition or minority party in Congress. The goal has been to make clear the specific reasons as to why one might see variations in the levels of party unity in opposition parties in Congress. This paper has argued that there are five significant factors in analyzing the roll call behavior of the opposition party. These include the presence of divided government, the unity of the majority party, interparty heterogeneity, the amount of time out of majority and the presence of a one chamber majority. Each of these concepts was expected to significantly impact the level of opposition party unity on roll call voting. The statistical models performed very well, and we are able to gain substantial insight as to how these factors affect the voting behavior of the opposition party. Each variable displayed the theorized sign in our analysis suggesting that the logical framework these variables were based on shows some validity. That is, the opposition party responds in part to the type of legislative proposals offered to it by the majority party. If those proposals are highly objectionable, the opposition party will be unified in its opposition to the bill. If those proposals are moderated by the majority’s recognition of other actors in the process (the president in periods of divided government and the Senate during split-party control of Congress) then the minority party will be less resolved in its opposition of these proposed bills. Therefore, we can begin to understand the aggregate dynamics of party unity in the U.S. House. The value of this research comes from the enhanced understanding we have gained through peering systematically at those factors, which affect the aggregate behavior of political parties. Remember that I have argued that it is the aggregate behavior of political parties and not individual affects they have on members of Congress that is most important for understanding the role of parties in the legislative process. While our specific purpose was -19- the minority party, understanding fully what contributes to the majority party’s cohesiveness is also an important question. Future research will broaden this project to include the majority party. The insight that we have gained allows us to more fully understand those factors that facilitate members of the minority party in Congress to coordinate their voting behavior and act collectively as a party. If, as I suggested above, the extent to which the minority party opposes the majority in the policy-making process contributes to the moderation of final policy outcomes, which consequently reflect more closely the preferences of the floor as a whole, and ultimately the electorate as a whole, then the ability of the opposition party to oppose immoderate legislation collectively plays a very important part in sustaining the representative nature of public policy. -20- Bibliography Aldrich, John H. and David W. Rohde. 1998. “Measuring Conditional Party Government.” Paper presented at the Annual Conference of the Midwest Political Science Association. Binder, Sarah A. 1996. "The Partisan Basis of Procedural Choice: Allocating Parliamentary Rights in the House, 1789-1990. American Political Science Review. 90: 8-20. _____. 1997. Minority Rights, Majority Rule: Partisanship and the Development of Congress. New York: Cambridge University Press. Binder, Sarah A. et al. 1999. "Uncovering the Hidden Effect of Party." Journal of Politics. 61: 815-31. Brady, David W. and Phillip Althoff. 1974. "Party Voting in the U.S. House of Representatives, 1890-1910: Elements of a Responsible Party System." Journal of Politics. 36: 753775. Brady, David W. et al. 1979. "The Decline of Party in the U.S. House of Representatives, 1887-1968." Legislative Studies Quarterly. 3: 381-407. Cooper, Joseph and Garry Young. 2002a. “Party and Preference in Congressional Decision Making: Roll Call Voting in the House of Representatives, 1889-1999” in Brady and McCubbins (eds.) Party, Process and Policy: Studies of the History of Congress. Stanford: Stanford University Press. Cooper, Joseph and Garry Young. 2002b. “Codebook for the Party Voting Dataset.” Internet publication found at: http://jhunix.hcf.jhu.edu/~jcooper/Codebook.pdf Cox, Gary, and Matthew D. McCubbins. 1993. Legislative Leviathan: Party Government in the House. Berkeley: University of California Press. Crook, Sara B and John R. Hibbing. 1985. "Congressional Reform and Party Discipline: The Effects of Changes in the Senority System on Party Loyalty in the U.S. House of Representatives." British Journal of Political Science 15: 207-26. Epstein, David. 1999. "Legislating from Both Sides of the Aisle: Information and the Value of Bipartisan Consensus. Public Choice. 101: 1-22. Jenkins, Jeffery A. 1999. "Examining the Bonding Effect of Party: A Comparative Analysis of Roll-Call Voting Jones, Charles O. 1968. "The Minority Party and Policy-Making in the House of Representatives." American Political Science Review. 62(2): 481-93. Jones, Charles O. 1970. The Minority Party In Congress. Boston: Little, Brown and Co. Kerr, Brinck. 2000. "Party Voting in the U.S. Senate, 1877-1990: An Assessment of Longand Short-Term Effects." Congress & The Presidency. 27:189-206. Knowles, Stanley. 1957. "The Role of the Opposition in Parliament." Address to the Empire Club of Canada, 21 March 1957, Ontario Woodsworth Memorial Foundation, Toronto. -21- Krehbiel, Keith. 1993. "Where's the Party?" British Journal of Political Science. 23: 235-266. Krehbiel, Keith. 2000. "Party Discipline and Measure of Partisanship." American Journal of Political Science. 44:206-221. Krehbiel, Keith and Adam Meirowitz. 2002. “Minority Rights and Majority Power: Theoretical Consequences of the Motion to Recommit,” Legislative Studies Quarterly. 27: 191217. Levitt, Steven D. 1996. "How Do Senators Vote? Disentangling the Role of Voter Preferences, Party Affiliation, and Senator Ideology." American Economic Review. 86:425-41. McCarty, Nolan. 2001. "The Hunt for Party Discipline in Congress." American Political Science Review. 95: 673-87 Nokken, Timothy P. 2000. "Dynamics of Congressional Loyalty: Party Defection and Roll Call Behavior, 1947-1997." Legislative Studies Quarterly. 25 (3): 417-44. Patterson, Samuel C. and Gregory A. Caldeira. 1988. "Party Voting in the United States Congress." Poole, Keith and Howard Rosenthal. 1997. Congress: A Political-Economic History of Roll Call Voting. Oxford University Press. Prais, S.J. and C.B. Winsten. 1954. “Trend Estimators and Serial Correlation.” Cowles Commission Discussion Paper No 383, Chicago. Rohde, David W. 1991. Parties and Leaders in the Postreform House. Chicago: University of Chicago Press. Schickler, Eric and Andrew Rich. 1997. "Controlling the Floor: Parties as Procedural Coalitions in the House." American Journal of Political Science. 41: 1340-1375. Snyder, James M. and Tim Groseclose. 2000. "Estimating Party Influence in Congressional Roll-Call Voting." American Journal of Political Science. 44: 187-205. -22-