Results and Discussion: Results
The present study rendered a quantitative analysis of available data for 328 total person-vote observations for the leaders of 36 congressional committees and subcommittees most related to the energy sector and the development of Russian sanctions and for one congressional representative who played a key role in moving a sanction bill through the Senate using the unanimous consent procedure. The quantitative analysis relied on four primary regression models as discussed in the Methodology section. These models were run in Stata using panels for both the congressional representative (R1, R2, and R4) and the sanction (R3). Three regression models (univariate, interaction term with no controls, and interaction term with full controls) were used to measure the relationship that energy interests held on the three dependent variables (person-vote spectrum, person-votes weighted by chamber, and person-votes weighted by the full Congress).
Results
The results of the regression models suggest that the presence of energy interests, with an emphasis on the energy index, influenced the person-vote decision for the sample set. Increases in the level of energy interest suggest an increase in the probability of voting in favor of the Russian sanctions in the sample period, which is denoted by the positive statistical significance of the energy interest dummy variable, χ1. However, the presence of energy interests suggests an increase in the probability of voting against Russian sanctions that specify the energy sector, which is denoted by the negative statistical significance of the interaction term, β7*(χ2*χ1). The aggregate results for the regression models are found in Table 21 and are expanded on for each hypothesis, the index regression, and the influence of political party affiliation with the sitting president in the respective sections. Specific Stata coding for the regression models is provided in Appendix B.
Note. This table presents the fixed effects panel regressions of the impact of energy interests on person-vote voting behaviors. The regression models used are:
● Vote/Voteweighted = α + α1*χ1 + ε (1)
● Vote/Voteweighted = α + α1*χ1 + β7*(χ2*χ1) + ε (2)
● Vote/Voteweighted = α + α1*χ1 + β1*χ2 + β4*χ3 + β5*χ4 + β6*χ5 + β7*(χ2*χ1) + ε (3)
The univariate model measured the influence of the energy interest dummy variable, χ1, on voting behavior. The interaction term model measured the influence of the energy interest dummy variable and energy sanction weight, χ2, on voting behavior. The controlled interaction term model measured the influence of the energy interest dummy variable and energy sanction weight on voting behavior using presidential party affiliation, χ3, and congressional chamber affiliation, χ5, as controls. Because of fixed effects, political affiliation was unable to be measured in Models 1–3.
a Due to fixed effects, modeling was unable to be completed using person-vote panels. Instead, R3 utilized sanctions panels, which resulted in the inability to explore an interaction term. Thus, the third regression model for R3 only included the univariate regression model and the control regression model without an interaction term. However, R3 included the political party affiliation variable, which was unable to be measured in R1, R2, and R4 due to person-vote fixed effects. b R1 has a lower number of observations, n = 325, than R4 and R2, where n = 328, due to the incomplete PFD data from deceased congressional representatives. c R3 included only the first four sanctions due to the unavailable data from the UCSB, resulting in a lower number of observations, n = 267, compared to R1, R2, and R4, where n = 325 or n = 328.
*p < .05. **p < .01. ***p < .001.
Hypothesis 1
Hypothesis 1 measured the relationship between personal financial holdings and the person-vote decision in the sample set. This was modeled using R1. The hypothesis would be supported should the interaction term dummy variable, β7, which measured the interaction between personal financial holdings and Russian energy-specific sanctions, render a value greater than zero. This would be expressed in Table 21 as a positive value with statistical significance at the 90% or higher confidence level. A stronger confidence level would correlate to a stronger relationship between personal financial holdings and the person-vote decision. This did not occur as no model rendered a confidence level at or above 90%.
While there was statistical significance at the 95% confidence level for personal financial holdings on person-vote behavior for Russian sanctions in general when measured against person-votes weighted by chamber, there was no statistical significant correlation from the results to suggest that personal financial holdings influenced person-vote behavior for energy-specific sanctions, either positively or negatively. This held true across Models 1, 2, and 3 for those dependent variables. The influence of affiliation with the political party of the sitting president on person-vote behavior was observed in the person-vote spectrum results at the 95% confidence level and in the person-votes weighted by chamber at the 90% confidence level. Because the results for R1 modeling do not render positive statistical significance for the interaction term dummy variable, the results do not support Hypothesis 1.
Hypothesis 2
Hypothesis 2 measured the relationship between campaign contributions and the person-vote decision in the sample set. This was modeled using R2. The hypothesis would be supported should the interaction term dummy variable, β7, which measured the interaction between campaign contributions and Russian energy-specific sanctions, render a value greater than zero. This would be expressed in Table 21 as a positive value with statistical significance at the 90% or higher confidence level. A stronger confidence level would correlate to a stronger relationship between campaign contributions and the person-vote decision.
While there was no positive influence reported in the results, statistically significant negative influence was reported at the 99% confidence level for the person-vote spectrum and person-votes weighted by the full-Congress-dependent variables and at the 95% confidence level for the person-votes weighted by chamber-dependent variable. This negative influence indicates that the presence of campaign contributions correlated to a Nay vote on energy-specific sanctions. This is supported by the statistically significant correlation of the presence of campaign contributions to voting for Russian energy sanctions in general at the 99% confidence level for the person-vote spectrum, the 90% confidence level for person-votes weighted by chamber, and the 95% confidence level for person-votes weighted by the full Congress. The influence of affiliation with the political party of the sitting president on person-vote behavior was observed in the person-vote spectrum results at the 90% confidence level and in the person-votes weighted by the full Congress at the 95% confidence level. Because the results for R2 modeling do not render positive statistical significance for the interaction term dummy variable, the results do not support Hypothesis 2.
Hypothesis 3
Hypothesis 3 measured the relationship between constituency stake in the energy sector and the person-vote decision in the sample set. This was modeled using R3. The hypothesis would be supported should the interaction term dummy variable, β7, which measured the interaction between constituency stake in the energy sector and Russian energy-specific sanctions, render a value greater than zero. Because of the use of fixed effects with panels by sanction, no interaction term was measured in R3. Instead, the statistical significance at the 90% or higher confidence level for the energy interest dummy variable would determine support for Hypothesis 3. A stronger confidence level would correlate to a stronger relationship between constituency stake in the energy sector and the person-vote decision.
R3 used Models 1 and 3, and no statistically significant values were rendered for energy interest. Only the chamber affiliation was statistically significant at the 99% confidence level when observed in the person-votes weighted by chamber. This result is biased by the model, however, and is therefore not applicable to the present study. The influence of affiliation with the political party of the sitting president on person-vote behavior was not observed in the models. Because the results for R3 modeling do not render positive statistical significance for the energy interest dummy variable, the results do not support Hypothesis 3.
Index Regression
The index regression measured the relationship among the energy interest index, the aggregate relationship between general energy interests, and the person-vote decision in the sample set. This was modeled using R4. Identification of a relationship between the energy interest index would be supported should the interaction term dummy variable, β7, which measured the interaction between the energy interest index and Russian energy-specific sanctions, render a value greater than zero. This would be expressed in Table 21 as a positive value with statistical significance at the 90% or higher confidence level. A stronger confidence level would correlate to a stronger relationship between the energy interest index and the person-vote decision.
While there was no positive influence reported in the results, statistically significant negative influence was reported at the 95% confidence level for the person-vote spectrum for both Models 2 and 3. This negative influence indicates that the presence of campaign contributions correlated to a Nay vote on energy-specific sanctions. This is supported by the statistically significant correlation of the presence of the energy index to voting for energy-specific sanctions in general at the 99% confidence level for the same dependent variable. Additionally, the presence of the energy index correlated to positive person-vote behavior on Russian sanctions in general at the 90% confidence level for Model 1 and the 95% confidence level for Models 2 and 3 for the person-vote spectrum dependent variable. The influence of affiliation with the political party of the sitting president on person-vote behavior was observed in the person-vote spectrum results at the 95% confidence level.
Changing the dependent variable to person-votes weighted by chamber and by the full Congress dramatically influenced the results of the modeling, with the presence of the energy index correlated to person-vote behavior for energy-specific sanctions not rendering in the modeling for both dependent variables. The presence of the energy index correlating to positive person-vote behavior on Russian sanctions in general was observed at the 90% confidence level for Model 1 and the 95% confidence level for Models 2 and 3 for the person-votes weighted by chamber-dependent variable but was only observed at the 90% confidence level for Model 3 for the person-votes weighted by the full-Congress-dependent variable. Similarly, a statistically significant value for the interaction term was only rendered through Model 2, using the person-votes weighted by chamber-dependent variable at the 90% confidence level and was negative. The influence of affiliation with the political party of the sitting president on person-vote behavior was observed in the person-votes weighted by chamber results at the 90% confidence level. Because the results for the R4 modeling render statistical significance for the interaction term dummy variable, the results support the development of a theoretical framework for additional studies into the relationship between energy interests and public policy development that concerns the energy sector.
Political Party Affiliation with the Sitting President
The present study posited that any relationship between energy interests and person-vote behavior would be more pronounced should the congressional representative be of the same political party as the sitting president. This was posited due to the president’s role as the chief executor of foreign policy for the United States and the functional party leader. Identification of this relationship would be supported should the presidential political party affiliation dummy variable, β4, render a value greater than zero. This would be expressed in Table 21 as a positive value with statistical significance at the 90% or higher confidence level. A stronger confidence level would correlate to a stronger relationship between affiliation with the political party of the sitting president and the person-vote behavior of congressional representatives.
Positive statistical significance was observed for R1, R2, and R4 at the 95%, 90%, and 95% confidence levels, respectively, for the person-vote spectrum dependent variable. Changing the dependent variable to person-votes weighted by chamber reduced the confidence level to 90% for both R1 and R4 while removing statistical significance from R2. However, for person-votes weighted by the full Congress, political party affiliation with the sitting president was statistically significant at the 95% confidence level for R2 while not significant for R1 and R4. This relationship was not statistically significant for any dependent variable modeled using R3. Additionally, the presence of statistical significance was not correlated to statistical significance, positive or negative, for the interaction term dummy or person-vote behavior for sanctions, whether energy-specific or not. Because the results for this relationship only rendered positive statistical significance for six of the 12 regression models, and because no correlation was observed between this relationship and the interaction term dummy or person-vote behavior for sanctions, the results do not support the conjecture that political party affiliation with the sitting president would render a stronger positive person-vote behavior concerning the Russian sanctions, energy-specific or not.
