Help us improve in just 2 minutes—share your thoughts in our reader survey.
Ballotpedia:Who Runs the States, Part Three, Key Values for Fifty-State Regressions
Praise or blame is extended to political parties for the economic, educational, health and other quality of life outcomes that result from the policies those parties enact into law. To better understand which political party enjoys power in each of the states, Ballotpedia has analyzed state government control from 1992-2013 using the concept of a "partisan trifecta." A partisan trifecta is defined as when a state's governorship and legislative chambers are controlled by the same political party.
The two major political parties claim that their policies will lead to better outcomes. What does the data show?
At Ballotpedia, we explored these issues in a three-part study, Who Runs the States.
This page contains the section of Part Three pertaining to the Fifty-State Regressions.
Fifty-State Regressions
SQLI versus Aggregate Partisan Score, Ordinary Least Squares Regression
This regression analysis showed the correlation between the partisan score of the state summed over the study period, the independent variable, and the aggregated SQLI ranking, the dependent variable. The negative coefficient indicates that a state with a Republican government correlates slightly with a higher ranking, but the P-value indicates that there is a 15 percent chance this relationship is actually zero. The R-squared value indicates that only 4 percent of the variation in the dependent variable can be explained by changes in the independent variable.
The explanatory power of these relationships is very limited, and much of the variation in the data cannot be accounted for by the party variables. This suggests that other, untested variables are more important drivers of these changes in rankings than party control. We also completed a regression comparing the states that had trifectas, either Democratic or Republican, with their change in rankings between the year of the trifecta and two years after the trifecta. This regression also found no statistically significant relationship, and barely any correlation between the two variables.
| Coefficient (direction of relationship) | P-value (statistical significance test) | R-squared (correlation strength) |
|---|---|---|
| -0.14 | 0.17 | 0.04 |
SQLI versus Lagged Trifecta, Ordinary Least Squares Regression
This regression analysis examined the relationship between the performance ranking of a state in a given year and the existence of a trifecta in the state two years before. The existence of a trifecta in the state two years before is measured with a dichotomous variable, with one indicating a trifecta of either major political party and zero indicating any other type of government. This regression allows us to see correlations between whether a state had a trifecta government or not, and its performance relative to other states two years later. The P-value indicates that there is a 92.3 percent chance this relationship is zero. The R-squared value indicates virtually no correlation between these two variables. Similar tests with one and three year lags do not yield statistically significant results.
| Coefficient (direction of relationship) | P-value (statistical significance test) | R-squared (correlation strength) |
|---|---|---|
| 0.761 | 0.419 | 0.00 |
SQLI versus Annual Government Composition Coding, Unlagged Panel Data Regression with Fixed Effects
This regression analysis showed the correlation between the government composition coding for each individual year, the independent variable, and the ranking of that state for each individual year, the dependent variable. This regression analysis identifies each state and follows trends over time, distinguishing the states from one another and controlling for the unchanging characteristics of each state over the time period. In this regression, the positive coefficient indicates that a more Republican state correlates with a higher ranking. The P-value shows the coefficient is statistically significant from zero. The R-squared value shows low correlation strength, with less than 9 percent of the variation of the dependent variable attributable to variation in the independent variable.
| Coefficient (direction of relationship) | P-value (statistical significance test) | R-squared (correlation strength) |
|---|---|---|
| 0.65 | 0.00 | 0.086 |
SQLI versus Annual Government Composition Coding, Lagged Panel Data Regression with Fixed Effects
This regression analysis is constructed identically to its unlagged counterpart, but it shifts the independent variable data back by two years. This lag permits time for the implementation of government policies and seeks to measure their impact after they take effect.
| Coefficient (direction of relationship) | P-value (statistical significance test) | R-squared (correlation strength) |
|---|---|---|
| 0.53 | 0.00 | 0.07 |
“Quality of Life” versus Trifecta Coding, Probit Regression Unlagged
This regression analysis uses a code that identifies either Republican or Democratic trifectas as one (1) and all other types of government as zero (0) as the dependent variable and the state “Quality of Life” ranking in the corresponding year as the independent variable. The probit regression analysis does not account for changes in time and does not identify individual states, but only seeks to identify the likelihood that a state with a higher or lower rank will have a trifecta government in that year. The coefficient and constant can be used to calculate the percent probability that a state will be under the control of a trifecta based on its ranking in a given year. These calculations weakly suggest that a state that ranks first (i.e., has the highest “quality of life”) may be about 7.5 percent less likely to be under a trifecta government than a state that ranks fiftieth (i.e., has the lowest “quality of life”). However, these results are not strictly significant and should not be interpreted as causal, because though they measure both the trifecta governments and the rankings for the same year, they do not control for (typo here?) state characteristics, and take into account additional control variables for economic downturns and other factors.
| Coefficient (direction of relationship) | P-value (statistical significance test) | R-squared (correlation strength) |
|---|---|---|
| 0.0039 | 0.142 | -0.2253 |
SQLI versus Trifecta Coding, Probit Regression Lagged
This regression analysis uses the same variables and framework as the unlagged probit, but shifts the dependent variable back by two periods. This regression allows us to compare the performance ranking of a state in a year with whether or not the state had a trifecta two years before. The relationship was not statistically significant, and the overall correlation was very low.
| Coefficient (direction of relationship) | P-value (statistical significance test) | R-squared (correlation strength) |
|---|---|---|
| 0.0023 | 0.418 | -0.1666 |
For more information on the regression methodologies and outputs, please see Appendices A, B, C, D, E, F and G.
See also
- Ballotpedia:Who runs the states
- Ballotpedia:Who Runs the States, Part One: State Partisanship
- Ballotpedia:Who Runs the States, Part Two: State Quality of life Index (SQLI)
- Ballotpedia:Who Runs the States, Part Three: Overlaying State Partisanship and State Quality of Life (SQLI)
- Part 1 Full report PDF
- Part 2 Full report PDF
- Part 3 Full report PDF
- State government trifectas
Conclusion
Our research shows that partisan control is correlated with economic performance and quality of life, as measured by the SQLI. States generally experience higher economic performance and better quality of life when Republicans control the government, followed closely by states that have divided government. This leaves much room for future research; namely, to what extent did political parties take advantage of their trifecta and enact progressive policies or conservative policies.
External links
