Radhika Mcdonald 7505 NW 142 nd Ave Alachua FL 32615 Phone: 352 682 2171 E Mail: email@example.com When to Tow the Party Line: H.R. 1938 April 2013 HONORS THESIS
April 2013 H.R. 1938 1 Acknowledg ement I would like to thank Dr. Lawrence Kenny, whose invaluable guidance made this research possible. Table of Contents I. INTRODUCTION 2 II. METHOD 3 III. SAMPLE 4 IV. DEPENDENT VARIABLE 4 V. INDEPENDENT VARIABLE S 4 VI. RESULTS 7 R EGRESSION 1 8 R EGRESSION 2 11 VII. CONCLUSION 14 BIBLIOGRAPHY 15
April 2013 H.R. 1938 2 I. Introduction The proposed Keystone XL p ipeline, an exte nsion to the existing Keystone p ipeline has drawn a lot of attention nationwi de. The issue of whether or not to construct the extension has been highly politicized and is controversial because of some of the environmental implications involved in its construction, particularly regarding its location to the Sand Hills and Ogallala Aquifer in Nebraska (two environmentally sensitive areas). The pipeline, a project of the TransCanada company will effectively connect the tar sands area in Alberta to refineries on the Gulf of Mexico in Texas. On May 23 rd 2011 Republican House represent ative Lee Terry (of Nebraska's 2 nd District) introduced the bill H.R. 1938 (also known as the "North American Made Energy Security Act") The bill is a directive that orders the president to "hurry up" and decide on the construction of the Keystone XL Pipe line and urges him to approve it The bill was co sponsored by 36 representatives from both parties in the House. It was approved in the House on July 26 th 2011 by a vote of 279 to 147, with 5 representatives "not voting An official summary of H.R. 193 8 from the Library of Congress is provided in Figure 1. North American Made Energy Security Act Directs the President, acting through the Secr etary of Energy (DOE), to coordinate with each federal agency responsible for coordinating or considering an aspect of the President's National Interest Determination and Presidential Permit decision regarding construction and operation of the Keystone XL pipeline (from Hardisty, Alberta, to Steele City, Nebraska, and then on to the U.S. Gulf Coast through Cushing, Oklahoma) to ensure that all necessary actions are taken on an expedited schedule. Requires each such agency to comply with any deadline the Sec retary establishes. Directs the President, within 30 days after the final environmental impact statement, but not later than November 1, 2011, to issue a final order granting or denying the Presidential Permit f or the Keystone XL pipeline. States that no a ction by the Secretary pursuant to this Act shall affect any duty or responsibility to comply with any requirement to conduct environmental review. Declares the sense of Congress that: (1) the United States must decrease its dependence on oil from countrie s hostile to its interests, and (2) increased access to Canadian energy resources will create jobs in the United States. Figure 1
April 2013 H.R. 1938 3 In the 112 th Congress there were 193 Democrats and 239 Republicans. The Republican Party had the clear majority therefore according to public choice theory, the existence of logrolling ( the act of vote trading) between Democrats and Republicans is unlikely, and will no t be considered for the purposes of this study. Logrolling within the House Democratic Party is also unlikely in this case and will not be investigated in this study. Of the 279 representatives that voted "AYE" on H.R. 1938, 47 were from the Democratic Party (Figure 2) Since 144 Democrats voted against the bill, the 47 representatives have voted against party lines (approximately 24.5% of the party). The fact that 24.5% of the Democratic Party voted against party lines prompts the question of why? This study examines the possible determinants of Democratic House of Representative party dissent in voting on H.R. 1938 Figure 2 II. Method Linear Regression has been used to test the various hypotheses (independent variables) and their level of significance and impact on the dependent variable. Both dummy variables and continuous variables were used as independent variables in this study.
April 2013 H.R. 1938 4 III. Sample The sample use d in this study consists of the 191 Democratic representatives in the 112 th Congress in 2011 that voted on H.R.1938 (2 representatives did not vote) The Republican representatives were not included because the focus is on deviation within the Democratic P arty. Data was collected at the congressional district level where available. IV. Dependent Variable Vote AYE' The Dependent Variable Vote AYE' is a dummy variable ( qualitative variable) that explains how a representative voted on H.R. 1938. A value of 1 i s assigned to all AYE' votes, and a value of 0 is assigned to all NO' votes. This was chosen as the dependent variable because the study is examining a voting decision, and the possible reasons behind that decision. Table 1 Topic Dummy Dependent Variable Description House Vote on H.R. 1938 Vote AYE' 0 if voted NO' 1 if voted AYE' V. Independent Variables In this study four variables were chosen that were thought to have a significant impact on whether a representative voted AYE' or NO' on the bill. Other variables like industry, state level oil or natural gas production, and age, were considered but upon further research and data analysis did not ap pear to correlate significantly or appropriately with the dependent variable.
April 2013 H.R. 1938 5 1. L CV Score LCV Score (League of Conservation Voter Score) is an ideological measure of how "pro environment" a member of congress is. The scoring system ranges from 0 to 100, 0 being "anti environment and 100 being "pro environment." Experts decide which env ironmental issues members voted on are the most important, and scores are calculated based the number of "pro environment" votes cast divided by the total number of votes counted For Example if the experts decide that there were 20 environmentally imp or tant votes in 2011, a representative with a LCV score of 80 voted "pro environm ent" on 16 of the 20 votes counted (or 80%) LCV scores for individual representatives are available on the League of Conservation Voter Website 1 Data Collected for this variab le is from the year 2011 ; included in the score is the representative s vote on H.R. 1938 The LCV Score variable is a continuous variable where the exact score each representative received is used. The official position of the League of Conservation Voter s on H.R. 1938 is "NO" therefore it is hypothesized that the higher the LCV score, the more likely a representative is to vote "NO on the bill. The expected relationship between the independent variable LCV Score and dependent variable Vote AYE' is a str ong negative correlation; therefore the coefficient on LCV Score should have a negative sign. 2. % Bachelors Degree or Higher % Bachelors Degree or Higher is a continuous variable that measures the percentage of people that have a bachelors degree or highe r per congressional district. The basic assumption is that people with higher education tend to be better informed about environmental and political 1 League of Conservation Voter Website: http://scorecard.lcv.org
April 2013 H.R. 1938 6 issues. It is also assumed that because universit ies and university professors tend to be more liberal 2 co ngressional districts with a high percentage of university graduates may be more liberal as well. Because democrats are historically "pro environment" and it is in a representative's interest to satisfy their constituents, it is hypothesized that the highe r the % Bachelors Degree or Higher in a congressional district the more likely a representative is to vote "NO on H.R. 1938. The coefficient is predicted to be negative, and have a negative correlation with the dependent variable. 3. % Unemployed on Pipeli ne % Unemployed on Pipeline measures the unemployment rate in congressional districts located along the pipeline. It was calculated by multiplying t he percent unemployed per congressional district with a dummy variable of whether or not a congressional dis trict was located along the keystone pipeline. This was done to make the unemployment variable more accurate as unemployment is assumed to be most affected by the creation of jobs in areas located around the pipeline, and least affected in areas located aw ay from the pipeline It is hypothesized that the greater the level of unemployment along the pipeline ( % Unemployed on Pipeline) the more likely a representative will be to vote AYE' on H.R. 1938. The coefficient is expected to be positive. 4. Male Male is a dummy variable that describes the gender of the representative voting 3 A value of "0" was given to females that voted, and a value of "1" was given to males that voted (Table 2) The basic assumption is that males would be more likely to deviate fr om their party's position, 2 "Research from the 1950s onward shows that in terms of values, issue specific attitudes, ideological self identification party affiliation, and voting behavior professors can typically be classified as liberals and more often than not are Democrats (Ladd and Lipset 1976; Lazarsfeld and Thielens 1958). 3 The representatives gender was taken from the Almanac of American Po litics 2012
April 2013 H.R. 1938 7 and would be less averse to the confrontation that follows when they don't "tow the party line." Therefore it is hypothesized that males are more likely to vote "AYE" than females and the coefficient will be positive. Table 2 VI. Results Summary statistics are reported in Table 3. A correlation coefficients matrix is provided in Table 4. Two sep arate linear regressions were run on the data, one with the independent variable LCV Score, and one without it. Table 3 Summary Statistics Minimum Maximum Mean ( Standard deviation) Dependent Variable: Vote 'AYE' 0 1 0 .24607 (0.43185) Independent Variables: LCV Score 20 100 87.87958 (16.24228) % Unemployed on Pipeline 0 11.4 0.39005 (2.00805) % Bachelors Degree or Higher 8.1 66 29.46021 (11.26254) Male 0 1 0.74346 (0.43787) N 191 Topic Dumm y Dependent Variable Description Gender of Representative Voting on H.R. 1938 Male 0 if Female 1 if Male
April 2013 H.R. 1938 8 Table 4 Correlation Coefficients Matrix Sample size 191 Critical value (2%) 2.34624 LCV Score % Unemployed x Location on Pipeline % Bachelors Degree or Higher Male LCV Score 1. % Unemployed x Location on Pipeline 0.19605 1. % Bachelors Degree or Higher 0.38195 0.13972 1. Male 0.19011 0.04617 0.05555 1. Regression 1 Regression 1 mea sures the effects of the independent variables, LCV Score, % Bachelors Degree or Higher, % Unemployed on Pipeline, and Male on the likelihood of a vote of AYE' on H.R. 1938 See Table 5. The formula used is: !"#$ !"# ! !"#" ! !"#$ !"#$ ! !"#! ! !"#$%&'(#) !"#$%&"' !" !"#$%"&$ ! !!"# ! !"# !"#$% !"#$"" !" !"# !" ! !"#$ !"# The r squared value (R 2 ), which measures the percentag e of variation in Vote AYE' (how a representative voted) explained by this model, is 0.48945. This means that the model explains approximately 48.945% of the variation
April 2013 H.R. 1938 9 Significant Variables In the first regression only the independent variable LCV Score was found to be statistically significant. The coefficient was negative as predicted ( 0.0165). The t statistic for LCV Score is 1 0.5414 The variable has a p value of 0.E+ 0 which is extremely low and as close to zero as the analysis program StatPl us permits This p value means that the null hypothesis that no relationship exits between the two variables can be rejected. The t stat value ( 10.5414) is significant at the 1% level or the 99 % confidence level. An example of the impact of this variable is illustrated in the following calculation: ! !"#$"%&' !"#$$%!%#&' !"#$%#&% !"#$%&$'( ! !"#$ !" !"#$ !" !"##$%&'() !"## ! !"#$ !" !"!!# ! ! !"#$$ This means that an average fluctuation from the mean would result in a 0.2679 lower probability that the representative would vote AYE' on the bill. The hypothesis that the higher the LCV score the less likely a representative is to vote AYE' was proven to be correct in this case. Insignificant Variables The remaining variables, % Bachelors Degree or Higher, % Unemployed on Pipeline, and Male were found to be stat istically insignificant at the 2 % level in regression 1. However, with a 1 tail test the variable Male would be significant at the 5% level, and the variables % Bachelors Degree or Higher and % Unemployed on Pipeline would be marginally significant at the 10% level. % Bachelors Degree or Higher had a t stat of 1.559 and a p value of 0.12, % Unemployed on Pipeline had a t stat of 1. 474 and a p value of 0.14, and Male had a t stat of 1.8507 and a p value of 0.0658. The null hypothesis could not be rejected under regression 1 because the alpha value for
April 2013 H.R. 1938 10 significance was at least 2 %. Alt hough the variables were insignificant, the hypothesized signs on the coefficien ts for all of the variables were proven to be correct. Table 5 Linear Regression 1 Regression Statistics R 0.6996 R Square 0.48945 Adjusted R Square 0.47847 Standard Error 0.31187 Total Number Of Cases 191 Voted = 1.7040 + 0.0974 Male + 0.0170 % Unemployed x Location on Pipeline 0.0034 % Bachelors Degree or Higher 0.0164 LCV ANOVA d.f. SS MS F p level Regression 4. 17.34329 4.33582 44.5775 0.E+0 Res idual 186. 18.09126 0.09726 Total 190. 35.43455 Coefficients Standard Error t Stat p level H0 (2 %) rejected? Intercept 1.70402 0.14302 11.91469 0.E+0 Yes Male 0.09743 0.05264 1.8507 0.0658 No % Unemployed x Location on Pipeline 0.01698 0.01152 1.4741 0.14214 No % Bachelors Degree or Higher 0.0034 0.00218 1.55909 0.12068 No LCV 0.01635 0.00155 10.5414 0.E+0 Yes
April 2013 H.R. 1938 11 Regression 2 A second regression was conducted to test the level of significance the variables % Bachelors Degree or Higher, % Unemployed on Pipeline, and Male would have if the LCV Score variable were not included. See Table 6. The formula for this regression is: !"#$% ! !"#$ ! !"#" !"#$ ! !"#$ ! !"#$%&'(#) !"#$%&"' !" !"# !"#$! ! !""# ! !"# !"#$% !"#$"" !" !"# !" The second regression had a R 2 of 0.1844 which means that 18.44% of the variation is explained by this model. Although the R 2 is not as high as the first regression it explains enough of the variation to merit reporting. Significant Variables All three variables were found to be significant at the 2 % level. % Bachelors Degree or Higher had a t statistic of 4.58 and a p value of 0.0001 The null hypothesis can be rejected in this ca se, and there is a strong negative relationship between % Bachelors Degree or Higher and Vote AYE' These results imply that the higher the percentage of bachelors degrees within a congressional district, the lower the probably that a representative will vote AYE'. The impact of the percentage of bachelor degrees in a district is exemplified in the following calculation: !""#$ ! !! !"!#$ ! !"#! This illustrates that an average fluctuation from the mean would result in a 0.1321 (lower) probabili ty that the representative would vote AYE'.
April 2013 H.R. 1938 12 % Unemployed on Pipeline had a t stati stic of 2.46472 and a p value of 0.01461 which means that it is significant at the 98% confidence level. The null hypothesis can be rejected in this case. An example of the impact of an average fluctuation from the mean unemployment in a congressional district located on the pipeline on a representative s decision is given in the following calculation: !"#"$ ! !!"!# ! !"#!$ This means that an average fluctuation wou ld result in a 0. 07104 higher probability that the representative would vote AYE' Although the impact of this variable is smaller than the others, t he hypothesis that an increase in unemployment along the pipeline would correspond with a greater likelihoo d of voting AYE' was proven to be correct, as the variable is significant and positively correlated with Vote AYE' The variable Male had a t statistic of 3.01661 and a p value of 0.0029 1, which means that it is significant at the 98% confidence level. The null hypothesis can be rejected for this variable. The hypothesis that males would be more likely to vote AYE' on H.R. was proven to be correct as the variable is significant and the coefficient is positive. The impact of Male is shown in the followin g calculation: !"#"$ ! ! !"#"$ This means that if you are a male (designated with the binary code 1) than the probability that you will vote AYE' on the bill is increased by 0.19693. Although there were more males than females in the 112 th House Democratic party (142 males and 49 females), 43 of the 142 males voted AYE' on the bill while only 4 of the 49 females voted AYE'. It is interesting that approximately 30% of male representatives voted in favor of the bill, while only 8% of female repre sentatives voted in favor.
April 2013 H.R. 1938 13 Table 6 Linear Regression 2 Regression Statistics R 0.42945 R Square 0.18443 Adjusted R Square 0.17134 Standard Error 0.39312 Total Number Of Cases 191 Voted = 0.4315 + 0.1969 Male + 0.0354 % Unemployed x Location on Pipeline 0.0117 % Bachelors Degree or Higher ANOVA d.f. SS MS F p level Regression 3. 6.53512 2.17837 14.09563 0. Residual 187. 28.89944 0.15454 Total 190. 35.43455 Coefficients Standard Error t Stat p level H0 (2%) rejected? Intercept 0.43148 0.09667 4.4635 0.00001 Yes Male 0.19693 0.06528 3.01661 0.00291 Yes % Unemployed x Location on Pipeline 0.03538 0.01435 2.46472 0.01461 Yes % Ba chelors Degree or Higher 0.01173 0.00256 4.58174 0.00001 Yes T (2%) 2.34645
April 2013 H.R. 1938 14 VII. Conclusion There may be many factors that influence whether a representative of the House decides to "tow the party line." This study highlights a few of those factors like a representative s ideology, the employment demographics of their congressional district, and the percentage of the district's population that h as a bachelor degree or higher, or whether they were male or female. In deciding whether or not to tow the line, a representative's decision involves considering their own personal beliefs along with the demands of their constituents. In the bill H.R. 1938 this study found that one of the more influential factors was a representative's LCV Score, how p ro environmental a representative was. If a representative has a low score and has historically voted against the "environmental choice" then they were likely to vote AYE' on H.R. 1938. The LCV score also explains why some of the representatives located in the Atlantic and Pacific coastal states would vote for the Keystone XL pipeline when they did not directly stand to gain from the "job creation" that proponents claim it will produce. In the statistical analysis, the hypotheses predicting the direction of impact (sign of the coefficient) that independent variables would have on the dependent variable was correct in all instances. In the first regression one of the four variables had statistical significance at the 99% confidence level ( LCV Score ). In th e second regression all three of the tested variables had statistical significance at the 98% confidence level. Further research could uncover more possibilities as to why Democratic representatives in the House would choose not to "tow the party line."
April 2013 H.R. 1938 15 Bibliography A Vote Smart Guide to the Keystone XL Pipeline Project Vote Smart (2013, March 29). Retrieved April 2013, from http://votesmart.org/blog/2013/mar/29/vote smart guide keystone xl pipeline/#.UWGevKXhtk9 Barone, M., McCutcheon, C., & National Journal Group Inc (2011). The almanac of American politics 2012: The senators, the representatives and the governors : their records and election results, their states and districts Chicago: University of Chicago Press. Davies, S. (n.d.). Interpreting R egression Output Retrieved 2013, from http://www.simon davies.org.uk/Interpreting_OLS_Output(cross sectional).pdf Gross, N. (2011, March 3). Political Bias in the Graduate Admissions Process: A Field Experiment Retrieved April 2013, from http://www.soc i.ubc.ca/fileadmin/template/main/images/departments/soci/faculty/gross/audit_pa per_march_3.pdf H.R. 1938 (112th): North American Made Energy Security Act (On Passage of the Bill) -GovTrack.us (n.d.). Retrieved 2013, from http://www.govtrack.us/congres s/votes/112 2011/h650 Lemieux, P. (2004). The Public Choice Revolution Retrieved 2013, from http://www.cato.org/sites/cato.org/files/serials/files/regulation/2004/10/v27n3 2.pdf North American Made Energy Security Act (2011; 112th Congress H.R. 1938) GovTrack.us (n.d.). Retrieved from http://www.govtrack.us/congress/bills/112/hr1938#summary/libraryofcongress StatPlus 2007 Professional Help Elementary Concepts (n.d.). Retrieved from http://www.analystsoft.com/en/products/statplus/content/help/src/ analysis_basic_statistics_elem entary_concepts_statistics.html Stratmann, T., & Garrett, D. (1995, July). Logrolling in the U.S. Congress. Free Online Library Retrieved 2013, from http://www.thefreelibrary.com/Logrolling+in+the+U.S.+Congress. a01729485 9 TransCanada | Keystone XL Pipeline (n.d.). Retrieved from http://keystone xl.com/home/transcanada/ U.S. States Metropolitan Areas MSAs Cities Counties Congressional Districts ACS 2011 Social Characteristics (2012). Retrieved 2013, from http://proxi mityone.com/acs2011dp2.htm