Production of Significant Legislation in the American States Nathaniel A. Birkhead Kansas State University

Production of Significant Legislation in the American States Nathaniel A. Birkhead Kansas State University birkhead@ksu.edu Abstract: In this paper, I evaluate the relationship between legislative characteristics and legislative productivity. Whereas previous analyses of legislative productivity have relied on counts of all legislation enacted, these measures do not allow one to account for the significance of a particular piece of legislation, thus conflating the trivial bills with the consequential. I am building a new dataset of legislative productivity (from 2007-2014), relying on data from Project Vote Smart. Using this new measure of legislative output, I show that production of significant legislation is not influenced divided government, but rather bicameral distances and the ideological differences between the parties. Preliminary Draft (More importantly, preliminary data). Paper Prepared for the 16 th Annual State Politics and Policy Conference May 19-21, 2016 University of Texas at Dallas 1

Introduction Although scholars have long been interested in analyzing state legislative output (cites TK), researchers thus far have been hampered by the need to rely upon aggregate counts of enacted bills, which conflates landmark legislation with trivial legislation. Thus far, there has been a disjuncture between what we know to be theoretically appropriate that in the eyes of the parties, legislators, and voters, some legislation is more important than other legislation and what we are able to empirically evaluate. At the national level, for example, Howell, Adler, Cameron, and Riemann (2000) showed that divided government disproportionately reduced the likelihood of Congress passing landmark enactments but actually increased the number of minor enactments. State legislative researchers have not had such a resource to test theories of lawmaking. This project seeks to fill this gap by developing a database of significant legislation passed by the states from 2007-2014. By identifying the legislation that is significant - such as the authorization of financial aid for undocumented students 1 or banning sexual orientation discrimination in the workplace 2 this database offers the opportunity to evaluate how legislators treat important legislation differently than routine legislation such as the decision to allow convenience stores to sell growlers of draft beer. 3 This is not to suggest that one standard is more valuable than the other indeed, the ability to acquire growlers at a convenience store very nearly defines the word convenient but that this database offers the ability to expand the theoretical and analytic approach to studying legislative output. In this paper, I begin by developing the theoretical value of a more nuanced version of legislative output, showing that this new dataset will help state politics scholars develop and test new theories 1 OR Senate Bill 932 Relating to cost of attending public universities 2 CO Senate Bill 07-025 Amending Colorado Revised Statute 24-34-402 3 MO Senate Bill 919 Modifies provisions relating to intoxicating liquor 2

of legislative activity. I then describe the two-pronged method I am using to develop this database: contemporaneous newspaper analyses of legislative sessions and the Key Votes archive from Project Vote Smart (PVS). In a manner similar to Mayhew s Divided We Govern, I am using state newspaper accounts of the legislative session to identify significant bills passed in the session. The Key Votes from PVS list the bills that Vote Smart deems to be notable, which I have then coded for legislative significance ranging from landmark to minor. I then posit a theory of significant legislative output, drawing on the programmatic goals and institutional resources of the political parties. Specifically, I argue that significant legislation is more likely to be passed under unified government when the parties are polarized. I test this theory using the significant Vote Smart data, showing that while political polarization boosts the likelihood of passing significant legislation, unified government has no effect. I conclude by discussing my goals for this project, while also offering suggestions for how the database can be used by other researchers. The Importance of Legislative Significance At the same time that the Pennsylvania state legislature enacted House Bill 605 which renamed a bridge on US Route 219 over US Route 422 in Ebensburg Borough, Cambria County, as the Alexander Miller Abercrombie Memorial Bridge, the legislature also enacted a program authorizing the use of medical marijuana (Senate Bill 3). As Cameron (2000) writes of Congress, The vast bulk of legislation produced by that august body is stunningly banal. Moreover, we understand that legislators enact legislation to satisfy a host of priorities from passing landmark legislation that serve policy goals, to commemorative bills that serve re-election goals (Kingdon 1989). 3

If we are interested in how legislators or parties operate in general forms, it makes sense to focus on all legislative outcomes. On the other hand, as Clinton and Lapinski observe, testing theories of lawmaking, as well as building new ones, on trivial legislation seems to be suspect (2006, pg. 233). Political parties have been the key focus for scholars seeking to understand legislative productivity. Moreover, the parties, and party leadership, are likely to emphasize some legislative bills more than others (Synder and Groseclose 2000). Mayhew (1991) argued that at the national level, legislative output was more or less constant whether we were under divided or unified government. By contrast, Howell et al (2000) show that when we more carefully characterize legislative output based on its importance, different patterns of parties and partisanship emerge. They show that landmark enactments are far less likely to pass under divided government than under unified government. On the other hand, legislative output of minor and commemorative legislation actually passes more often under divided government, as legislators are unable or unwilling to compromise on more important bills. The refinement of this theory was facilitated by more refined measures. Research of legislative productivity at the state level has largely been restricted to aggregate measures of output (Gray and Lowry 1995; Squire 1998, Bowling and Ferguson 2001, Rogers 2005, Hicks 2015). When it has been disaggregated, it has been done so at the policy level (e.g. Gray and Lowry 1995; Bowling and Ferguson 2001). Hicks (2015) shows that partisan influence on aggregate the legislative output is conditional - the legislature is far more productive when the parties are polarized and dominated by one party - than it is when the legislature is polarized and only a small margin separates the two parties. Thus, when the parties have programmatic goals, as they are most likely to during an era of polarization, it simultaneously raises the potential for productivity and gridlock. The potential for gridlock is exacerbated by the presence of divided government, when the minority party can rely on a sympathetic governor to obstruct the majority's agenda. Nonetheless, even when 4

polarized or under divided government, the political parties in the states are as unlikely as the national parties are to treat all legislation the same. To determine whether or not this actually true however, requires a different data source. Constructing the Significant Legislation Dataset State legislatures enacted roughly 25,000 pieces of legislation in 2014, and the scale alone makes it difficult to determine which pieces of legislation are banal and which pieces are significant. Fortunately, Project Vote Smart regularly updates what they call Key Votes for each legislative session. Four PVS staff members monitor news on state legislatures, identifying when important votes are occurring. Along with about 20 interns, they use four criteria to guide their selection of key votes: 1) Is this a contentious vote? 2) Does this represent a clear issue position? 3) Is there substantial media coverage? 4) Is this issue easy for voters to understand? The staffers exercise some state-specific discretion, as what it means to be a contentious vote in a legislature with a 51/49 party breakdown may differ from a state with a 70/30 party split. While PVS interns rotate regularly (there are about 20 interns per semester) staffers are more stable: there have been only 16 staff members working on the Key Votes department since 2011. Accordingly, year-to-year fluctuations in standards should be relatively modest. After the staffers compile their selection of key votes, they are then passed to a state specific board of advisors mostly academics, but also a few journalists - for approval. Relying upon the PVS data serves as a valuable first step in the process, as it provides a manageable list of legislative output for each legislature s session. Even so, it potentially induces two forms of 5

errors. On the one hand, PVS may include data in their Key Votes that fulfills their mission, but does not actually constitute significant legislation (Type II error). For example, they include votes on bills that are important, but do not actually pass. This is important for voters to better understand how their elected officials are voting, but does not fit our needs to developing a dataset of significant legislation. Alternatively, though related, PVS may include legislation that does pass, but is nonetheless relatively inconsequential. 4 On the other hand, PVS may miss otherwise important legislation because it didn t fit their criteria (Type I error). Remedying this potential Type I error is a difficult task. It requires finding some other way of culling the approximately 25,000 bills enacted by state legislatures every year to determine what significant legislation PVS is leaving out. Following Mayhew s approach in Divided we Govern, I am compiling newspaper accounts and session wrap-ups on the state legislatures by state and local newspapers to help augment the PVS data, and also corroborate what the RAs and I have called significant. Unfortunately, this portion of the project is quite slow while some state papers have dedicated beat writers who write end-of-session wrap ups like the New York Times or Washington Post do for Congress many papers do not, or do not consistently over time. 5 This portion of the project is ongoing, and the data will not be included in this analysis. By contrast, both facets of the Type II errors can be remedied easily enough. The first is done by culling the data, pulling out bills that failed to become law. The second is similarly simple, though more onerous: it requires going over the data and identifying whether a bill is significant, ordinary, or minor. Table One, below, shows a few examples of each type of legislation. (Table One about here) 4 For example, one Key Vote in 2013 was Idaho s SB 1117, which authorized permits for heavy trucks on certain routes. 5 I would love to hear suggestions on possible resources that serve this goal 6

Two RAs have performed this work, completing data from 2007-2014 for 38 states. The inter-rater reliability for determining whether or not a piece of legislation is significant is modest: 0.81. A third RA will go over the data this summer and resolve any discrepancies. In all, we categorized 662 of the 1396 Key Votes that became law as significant (47%). (Table Two about here) Table Two, above, shows the number of significant bills in the PVS data for each state and each year. A number of patterns stand out. First, some states appear to be passing more significant legislation than other states. California serves as an extreme outlier in this case, passing 14% of the significant legislation in the dataset. 6 By contrast, West Virginia only passed 4 significant bills in the time period. The second pattern apparent in Figure One is that it is rare for a state to routinely pass significant legislation. Rather, significant legislation seems to ebb and flow, with a session with little (if any) significant legislation followed by a session with considerably more significant output. Third, some years appear to have more significant output than others whereas there were only 65 significant bills passed in 2007, there were 105 in 2013. Nonetheless, despite the year-to-year variance, a unit root test confirms the stationarity of the data. 7 When Should We Expect Legislative Activity? All Enacted Bills I begin by developing a simple model of legislative output, based on Hicks (2015) and Rogers (2005). Using data from the various Books of the States, I set my dependent variable as the number of Enacted Bills in a calendar year. I set this as a function of Average Polarization in the 6 There are a number of plausible explanations for this, largely centering on the high level of professionalism and relative richness in media coverage. Nonetheless, all results are replicated, and almost completely unchanged, after dropping CA. These are presented in the Appendix. 7 When the number of panels is large and the time dimension is limited, the Harris-Tzavalis test is appropriate (Harris 1999). ρ= -0.10, z=-13.9, p<0.01 7

legislature which is taken as the average of the distance between Democratic and Republican medians in the upper chamber, and the distance between party medians in the lower chamber. These values come from the Shor-McCarty (2011) common-space ideal points dataset, and range from 0.465 to 3.13 with a mean of 1.55. As indicated above, I expect productivity, when measured on aggregate legislation, to be unaffected by polarization. Following Hicks (2015) I account for the party seat distribution in the legislature by calculating the Folded Partisan Seat Margin (FPSM) using membership data from the Books of the States. FPSM accounts for the percentage of seats held by one party in both the House and Senate, and theoretically ranges from 50, where one party dominates both chambers, to 100 where there is a perfectly divided balance of power. In practice, FPSM ranges from 58.8 to 99.9, with a mean of 85. I expect that as the seat margin is small (higher values of FPSM), productivity will drop. By contrast, when the seat margin is large (FPSM small), productivity will increase. I expect that FPSM will be particularly influenced by the ideological differences between the parties. That is, when party polarization is low, I expect FPSM to have a smaller impact on number of enactments than when polarization is high. Accordingly, I estimate an interaction between Party Polarization and FPSM. I account for the ideological difference between the House and Senate by adopting the same approach as Binder (1999) and Hicks (2015): calculating the difference between the chamber medians, again using the Shor-McCarty scores. As Hicks (2015) points out, Shor and McCarty s approach of using NPAT scores to identify the common space facilitates direct comparisons between chambers, thus allowing for an ideal measure of bicameral distance that is consistent across space and time. This Bicameral Distance measure ranges from 0 to 1.15, with a mean of 0.2. I expect that as the bicameral distance increases, the number of enacted bills will decrease. 8

If the majority in at least one chamber of the legislature is of a different party than the governor, the measure Divided Government takes a value of 1. If the upper chamber, lower chamber, and governor s mansion are all controlled by the same party Divided Government takes a value of 0. Data on control of the chambers and governor s mansion comes from the Books of the States. 8 Consistent with Rogers (2005) I expect that fewer bills will be enacted under divided government than under unified government. I also estimate an interaction term between divided government and FPSM. I expect that divided government will lead to a bigger decrease on legislative output when the legislature is wholly dominated by the opposition party. By contrast, when the governor has a large (though minority) set of supporters in the legislature to work with, I expect productivity to get a modest boost. To make some control for the supply of bills in the legislature that can become enacted (Squire 1998), I account for the Number of Bill Introductions (in 100s). This data also comes from the Books of the States. Values range from 1.40 to 182.96 with a mean of 21. I expect that as the number of introductions increases, so too will the number of enactments. I use Squire s 2007 index of Professionalism to account for salary, staff, and session length. There are mixed results for professionalism in the literature Rogers finds professionalism to be positively associated with output, Squire finds it to be negatively associated with output, and Hicks finds no result. Consequently, I do not have any expectations for the relationship professionalism should have with output. 8 In a number of states, there has been a partisan tie in the legislature. Often, these ties are solved in some form of power-sharing agreements (McGrath and Ryan 2014). For the purposes of determining control, I mark a legislature as controlled by the party based on the partisanship of the majority leader. 9

I use the NCSL s data to determine if Term Limits have been adopted and implemented in the state legislature, with values of 1 if they have been and 0 if otherwise. I expect to find that term limits boost legislative output, consistent with Hicks (2015). To evaluate these relationships, I use a regression model with state and year random effects. The results are presented in Table Three, below, and they provide mixed support for my expectations. I will hold off on interpreting any concepts associated with interactions, which I will do in detail below. (Table Three About Here) Going over the other coefficients, bicameral distance has no effect on the aggregate number of bills enacted by the legislature, contrary to my expectations. On the other hand, the number of bill introductions is positively associated with bill enactments, as are the presence of term limits. Legislators in term limited legislatures seem to be spurred by the ticking clock to pass as many bills as possible. My null results of professionalism support Hicks null results, as well. First, with respect to the Polarization X Seat Margin interaction, which shows predicted values of enactments in Figure One below, we see that the direction of the relationship is opposite what I hypothesized. That is, as the seat margin becomes more competitive, the polarized legislatures become more, rather than less, productive. By contrast, the less polarized legislatures are more productive when one party is firmly in control than when the party balance is close. (Figure One About Here) Figure Two, which shows the relationship between divided government and seat margin, offers more support for my hypothesis. Here, I present the marginal effects of the interaction terms. As expected, divided government has a negative effect on productivity when one party 10

dominates the legislature. By contrast, seat margin does not influence productivity under unified government. (Figure Two About Here) Some of my results differ from those of Hicks (2015) who offers the most similar analysis. I attribute these differences to the time period under inquiry (his data runs from 1991-2009, whereas mine is from 2007-14) and differences in model specification. The broader point in this exercise is not to re-evaluate what others have done with similar models. Rather, the goal is to outline a number of the measures used in the analysis, and to establish a baseline set of expectations for how the legislatures in this sample operate. To summarize the important points, we find that highly polarized legislatures tend to be more productive as the seat margin gets more competitive, less polarized legislatures have the opposite relationship, and that divided government decreases productivity. Significant Legislation Now, we finally turn to the significant legislation data. The model is similar, with the only difference that I omit the overall number of bill introductions. I expect legislatures with polarized parties to be more likely to pass significant legislation than those with less polarized parties. I also expect divided government to depress the number of significant acts passed by the legislature, particularly when the legislature is dominated by the opposition party. Lastly, I expect that as the distance between upper and lower chambers increases, the number of significant acts should decrease. I fit a negative binomial regression model, again with state and year random effects. The results are presented in Table Four, below. (Table Four About Here) 11

Here, we see that professionalized legislatures are no more likely to pass significant legislation than unprofessionalized legislatures, and that legislatures with term limits are no more likely to pass significant bills than otherwise. Contrary to expectations, I also find that there is no support for the expectation that divided government had any effect on output of significant legislation either directly or indirectly. The marginal effects (presented in the Appendix) show that across the range of the data, divided government did not have an impact on FPSM s influence on significant legislation, nor did FPSM have an impact on divided government s influence. To get a handle on the other relationship in the data, and because many of my expectations cannot be easily evaluated based on the regression coefficients alone, I present the following figures to better evaluate them. Figure Three, below, presents the marginal effects of the interaction between party polarization and seat margin. The left panel shows that seat margin has a very modestly depressive effect on significant productivity when polarization is high, but not when it is low. By contrast, polarization has a significant impact on productivity across the range of partisan seat margins. (Figure Three about Here) To put these results into context, Figure Four shows the predicted probability that the legislature will pass at least one significant piece of legislation in the session. Across the range of the data, we see that highly polarized legislatures are quite a bit more likely to pass a significant bill than a less polarized legislature. The slopes differ, however, and the difference between more polarized and less polarized legislatures diminishes as the seat margin shifts into balance. (Figure Four About Here) Though I am now running the risk of this paper only being a series of annotated graphs, I show in Figure Five the predicted probability that the legislature will pass at least one significant bill for 12

legislatures with high and low levels of polarization across the ranges of bicameral distance. Again, we see that highly polarized legislatures are wholly more productive than less polarized legislatures, but that as the distance the chambers increases, the likelihood of passing significant legislation drops considerably. (Figure Five about Here) These findings underscore a number of important differences and similarities between legislative characteristics when considering all enacted bills from significant bills. On the whole, polarized legislatures are more productive whether we are measuring significant bills or all enacted bills than less polarized legislatures. The relationship between seat margin and polarization runs different ways depending on what the measure of legislative output is, however. While polarized legislatures are less likely to produce significant legislation when the seat margin is tight, they are more likely to pass other bills. Moreover, whereas divided government decreases the overall output of enacted bills by the legislature, the partisanship of the governor was inconsequential in predicting the passage of significant legislation. By contrast, what was more influential for significant legislation is ideological differences between the lower and upper chambers. Discussion and Extensions One important preliminary finding in this research is that significant legislation is very likely to pass when the parties are polarized, irrespective of the partisan seat margin. This shows that when parties offer clearly defined programmatic goals, they are more likely to pass legislation that dramatically changes the status quo. The fact that these parties are able to do so, even when the seat margin is small, suggests that perhaps the minority party is unwilling to wholly obstruct the majority party s legislative agenda. Rather, perhaps these represent the party leadership identifying the open space in the gridlock interval where there is room for negotiation (Krehbiel 1998). In the future, I 13

will explore the effect that shifts in the gridlock interval have on the likelihood of passing all legislation, and of passing only significant legislation. Regarding the null effect of legislative professionalism on productivity, there are a few ways forward. The first is to follow Bowen and Greene (2014), who suggest disaggregating the components of professionalism, and using the individual components separately. In particular, whereas legislative salary may not predict production of significant legislation, number of staffers might as Gray and Lowry (1995) found. The second is to consider the sources of these significant bills. Perhaps professional legislatures are writing their own items of significant legislation, but that the more citizen legislatures are relying on the support of interest groups (Balla 2001, Garrett and Jansa 2015). In particular, the American Legislative Exchange Council (ALEC) has been very successful at providing model legislation often reflecting particularly extreme shifts to the status quo - to unprofessionalized legislatures (Hartel-Fernandez 2014). Thus perhaps two different processes may be occurring simultaneously, obscuring the relationship between professionalism and legislative production. Lastly, the null effect of divided government may be best explained by the importance of bicameral distance. As the ideological differences between the lower and upper chambers grow, irrespective of partisan control, we should expect to find less and less agreement on major policy issues (Binder 1999). Again, this would be better explored by focusing on the key pivot points in the legislature. Nonetheless, this finding underscores the importance of ideological, rather than partisan, characteristics of the key actors in lawmaking. Going forward, I will continue to refine this dataset, both by resolving the discrepancies by the coders in the PVS Key Votes dataset, and by folding in newspaper coverage as it is available. My hope is that when it s complete, this dataset will help researchers develop and test new theories of 14

legislative production. Beyond exploring the relationship between legislative organization and significant bills, this dataset could enable researchers to identify co-sponsorship networks among significant bills, or perhaps help determine the extent to which women and minorities or workingclass legislators pass major legislation on par with affluent white men. Ultimately, the ideal measure of legislative significance would incorporate the methodology as outlined by Clinton and Lapinski (2006). They draw on a wide number of raters, and use both contemporaneous and retrospective sources, and their measures are perhaps the most robust. Their model maximizes the benefits of local knowledge of the legislation itself deferring the act of rating to experts on legislation - while diminishing the likelihood that idiosyncratic reactions will drive any results. While there is no empirical reason not to extend their approach to the state legislative arena, the scope of the task is considerable. In this manner, PVS and legislative roundups would only be a few data point of many, included with a number of other sources. 15

Works Cited Aldrich, John H. 1995 Why Parties? Chicago, IL: Chicago University Press Balla, Steven J. 2001. Interstate Professional Associations and the Diffusion of Policy Innovations. American Politics Research 29(3):221-45. Binder, Sarah. 1999. The Dynamics of Legislative Gridlock, 1946-1996. American Political Science Review 93(3):519-33. Bowen, Daniel C. and Zachary Greene. 2014. Should We Measure Professionalism with an Index? State Politics and Policy Quarterly 14(3):277-296. Bowling, Cynthia. J. and Ferguson, Margaret. R. 2001. Divided government, interest representation, and policy differences: Competing explanations of gridlock in the fifty states. The Journal of Politics 63: 182-206. Cameron, Charles. 2000. Veto Bargaining: Presidents and the Politics of Negative Power. New York: Columbia University Press Chiou, Fang-Yi and Lawrence S. Rothenberg 2008 Comparing Legislators and Legislatures: The Dynamics of Legislative Gridlock Reconsidered. Political Analysis 16:197-212. Clinton, Joshua and John S. Lapinski. 2006. Measuring Legislative Accomplishment. American Journal of Political Science 50(1): 232-249. Coleman, John C. Unified Government, Divided Government, and Party Responsiveness American Political Science Review 93:821-835. Cox, Gary W. and Mathew D. McCubbins. 2005. Setting the Agenda: Responsible Party Government in the U.S. House of Representatives New York, NY: Cambridge University Press Garret, Kristin N. and Joshua M. Jansa. 2015. Interest Group Influence in Policy Diffusion Networks State Politics and Policy Quarterly 15(3):387-417. Gray, Virginia and David Lowery. 1995. Interest Representation and Democratic Gridlock. Legislative Studies Quarterly 20: 531-52. Harris, Richard D. F. and Elias Tzavalis. 1999. Inference for unit roots in dynamic panels where the time dimension is fixed Journal of Econometrics 91(2):201-226. Hertel-Fernandez, Alexander. 2014. Who Passes Business Model Bills? Policy Capacity and Corporate Influence in U.S. State Politics Perspectives on Politics 12(3):582-602. Hicks, William D. 2015. Partisan Competition and the Efficiency of Lawmaking in American State Legislatures, 1991-2009. American Politics Research 43: 743-70. Howell, William and E. Scott Adler, Charles Cameron, and Charles Riemann. 2000 Divided Government and the Legislative Productivity of Congress, 1945-94 Legislative Studies Quarterly 25:285-312 Krehbiel, Keith. 1998. Pivotal Politics Chicago: University of Chicago Press. Mayhew, David. 1991. Divided We Govern. New Haven: Yale University Press. 16

McGrath, Robert J. and Joshua Ryan. 2014. Even Money: Tied Chambers, Power Sharing, and Party Effects in State Legislatures Paper presented at 14 th Annual State Politics and Policy Conference, Bloomington IN. Rohde, David W. 1991. Parties and Leaders in the Post reform House. Chicago, IL: Chicago University Press. Rogers, James R. 2003. The Impact of Bicameralism on Legislative Production. Legislative Studies Quarterly 28: 509-528. Rogers, James R. 2005. The Impact of Divided Government on Legislative Production. Public Choice 123: 217-33. Rosenthal, Alan and Rod Forth. 1978. The Assembly Line: Law Production in the American States. Legislative Studies Quarterly 3: 265-91. Shor, Boris and Nolan McCarty. 2011. The Ideological Mapping of American Legislatures. American Political Science Review 105(3): 530-51. Snyder, James M and Tim Groseclose. 2000. Estimating Party Influence in Congressional Roll Call Voting. American Journal of Political Science 44(2):193-211 Squire, Peverill. 1998. Membership Turnover and the Efficient Processing of Legislation. Legislative Studies Quarterly 23: 23-32. Squire, Peverill. 2007. Measuring State Legislative Professionalism: The Squire Index Revisited. State Politics and Policy Quarterly 7(2): 211-27. Squire, Peverill and Gary Moncrief. 2010. State Legislatures Today: Politics Under the Domes. Boston, MA: Longman. 17

Table One: Identifying Significant Legislation State Year Bill Number Topic Significant Legislation CO 2013 SB 33 Authorizes In-State Tuition for Undocumented Immigrants MN 2014 HF 2091 Increases the Minimum Wage CT 2012 SB 280 Repeals the Death Penalty Ordinary Legislation MN 2013 SF 523 Prohibits Employers From Asking Applicants for Criminal History CT 2011 SB 1152 Funding for Construction at the University of Connecticut Health Center ID 2012 H 660 Amends the Judicial Retirement Fund Minor Legislation SD 2013 SB 52 Authorizes Gamblers to Bet $0.01 on Lottery Games CO 2014 SB 29 Requires Paint Manufacturers to Establish Recycling Programs CT 2007 Sub HB 7350 In-School Suspensions 18

Table Two: Significant Legislation by State and Year 2007 2008 2009 2010 2011 2012 2013 2014 AK 3 1 1 1 1 1 5 AL 0 0 2 2 3 2 4 3 AR 1 1 0 2 11 0 AZ 1 6 3 6 9 10 3 4 CA 7 9 16 11 15 12 10 12 CO 5 6 7 14 2 1 8 2 CT 2 2 5 1 6 4 5 2 DE 3 0 4 0 4 2 3 3 FL 2 0 1 0 4 5 2 2 GA 0 2 3 3 1 2 5 HI 0 0 2 1 2 0 2 1 IA 4 3 4 5 1 2 3 0 ID 1 4 2 5 5 7 3 1 IN 4 2 1 0 3 1 2 5 KS 1 1 3 2 4 2 8 5 KY 2 2 1 2 0 0 3 1 MA 1 1 1 1 1 2 ME 2 4 2 1 1 1 MN 3 4 0 0 1 2 6 4 MS 1 3 0 2 1 1 2 2 MT 1 1 4 3 ND 3 1 2 4 NE 3 2 3 2 1 2 0 NM 3 3 2 4 1 1 NV 0 0 4 0 5 3 NY 0 1 3 2 1 2 1 1 OH 1 1 0 0 3 6 0 2 OK 1 1 2 2 2 1 3 5 PA 0 1 2 4 2 5 1 0 RI 1 1 1 3 4 2 3 SC 3 1 0 1 2 1 1 SD 3 1 0 1 0 2 2 2 TX 1 1 3 4 VT 1 1 1 1 2 1 WA 4 3 3 3 2 1 1 0 WV 0 0 2 2 0 0 0 WY 1 2 0 5 2 1 1 19

Table Three: Model of Legislative Enactments with State and Year Random Effects 20

Figure One Enacted Legislation by Polarization and Folded Seat Margin 21

Figure Two: Marginal Effects of Divided Government and Partisan Seat Margin 22

Table Four: Count Model of Significant Legislation with State and Year Random Effects 23

Figure Three: Marginal Effects of Polarization and Seat Margin Significant Legislation 24

Figure Four: Probability of Passing Significant Legislation by FPSM and Polarization 25

Figure Five: Probability of Passing Significant Legislation by Bicameral Distance and Polarization 26

Appendix: Analysis without CA 27

28

29

30