Median Income, Median Housing Cost, and Vote Choice for Florida’s $15 Minimum Wage Amendment
Mario Villegas
2023-12-11
Introduction
In a democracy, voting is one of the most important tools for citizens to voice their thoughts, concerns, and opinions—especially during election season. Given the importance of voting in a democracy, it attracts the attention of many scholars in the political science discipline. Whether people vote “correctly” is a perennial question scholars have attempted to answer. Of course, scholars can define a “correct vote” in different ways, inevitably leading to different findings depending on the interpretation.
In this paper, I will examine the relationship between regional financial security and vote choice for Florida’s $15 Minimum Wage Amendment. I hope to shed light on the following question: How does median income and median housing cost impact the vote for Florida’s $15 Minimum Wage Amendment? I will analyze precinct and census data to see if there is a relationship between median income, median housing cost, and vote choice for the minimum wage amendment. I will use OLS analysis to interpret this relationship.
My analysis will particularly shed light on rational choice theory, which states that people vote to increase their economic and political utility (Downs, 1957). Examining median income allows me to see whether wealthier regions are willing to vote for a policy that is ostensibly not for them. Furthermore, examining median housing cost allows me to see whether high-cost areas will vote for a policy that could obviate their increasing costs. Finally, regions with higher median incomes and housing costs will elucidate how the interaction between the two variables affects support for this policy.
This analysis will also speak to relative power theory and conflict theory, albeit not as much as rational choice theory. If wealthier populations are the ones supporting the minimum wage amendment, then this could indicate the presence of an income bias in the electorate (Solt, 2010; Stockemer, 2017). However, if poorer populations are the ones supporting the minimum wage amendment, then this could indicate that there is tension between the poorer and wealthier populations, leading to increased mobilization for this policy (Brady, 2004; Stockemer, 2017; Stockemer & Parent, 2014).
The importance of this analysis lies in the predictive power of vote choice. If we are able to predict how people will vote based on descriptors or demographics, such as income, then political actors can be more efficient in their efforts. For example, if campaigns could accurately predict vote choice, then they would know which areas to target in order to mobilize support for their policy or candidate. Moreover, if legislators could accurately predict vote choice, then they could put forth policies that help the majority of their constituents. In either case, researching this topic engenders a more informed and efficient set of political actors.
The layout of the paper is as follows: I will first give context to my analysis by going over rational choice theory, relative power theory, and conflict theory. Next, I will discuss my expectations for this analysis. Then, I will go over the data, methods, and procedure of the analysis. Finally, I will analyze and discuss the results.
Literature Review and Hypotheses
Rational Choice Theory
Anthony Downs (1957) defined a rational voter as someone who votes to increase their economic or political utility. In this framework, the individual votes cognitively since their decision is made with a cost-and-benefit approach. How a person conceptualizes their political and economic utility is subject to many variables, such as elite rhetoric, agenda setting, interpersonal networks, priming, and framing (Iyengar, 1987; Iyengar & Simon, 1993; Lyons et al., 2016; Sides et al., 2022; Zaller, 1992). All of these factors can impact how an individual interprets a vote choice affecting them. For example, if the media highlights the Democratic candidate as a benefit to the economy over the Republican candidate, then people may believe that the Democratic candidate will increase their economic utility. Furthermore, if an individual has family members who consistently state that the Republican candidate is better for the economy, then that person might believe that voting for the Republican candidate will benefit their economic utility. These examples underscore an important characteristic of rational choice theory: voting rationally is difficult. Deciding what will or will not benefit the individual takes time, effort, and resources. The vast amount of political stimuli in everyday life makes rational voting harder than utilizing cues or shortcuts.
Scholars have tested rational choice theory in various studies. Blais & Young (1999) found that participants in their experiment did not vote in the way rational choice theorists would expect. Instead, participants relied on their sense of civic duty to make their way to the ballot box (Blais & Young, 1999). Another experiment by Blais et al. (2014) found that participants’ vote choice did not maximize their payoff. Even though some participants did exhibit rational voting, Blais et al. (2014) found too many inconsistencies with rational choice theory. Brians & Grofman (1999) also come to mixed conclusions on the traditional rational choice model. Knowing that people of higher socioeconomic status tend to vote more often, one would presume that removing barriers to voting—such as voter registration—would benefit people of lower socioeconomic status. However, Brians & Grofman (1999) found that areas allowing election day registration benefited the middle class more than the lower class—contrary to what we might expect of the rational choice model.
However, there is also literature to suggest that people can vote rationally. For instance, Van der Straeten et al. (2010) found that people do behave according to the rational choice model if the cognitive workload is minimal. If the cognitive workload is too large, then voters will rely on heuristics (Van der Straeten et al., 2010). Similarly, Hsu & Sung (2002) found that people do vote instrumentally when offered an incentive to do so; however, the authors also state that instrumental voting cannot be used to explain most of the voting in real elections. Finally, Henderson et al. (2010) examined Democratic voters in the 2008 election. They found that Clinton voters in the primary election were more likely to vote for Obama in the general election if they resided in a battleground state—indicating that people understood when their vote “mattered” more (Henderson et al., 2010). Even among those who did not vote for Obama, it was issue positions and ideology that caused this shift—further supporting a more rationally-oriented approach to voting (Henderson et al., 2010).
In the interest of this paper, rational choice theory would suggest that lower-income populations would be more likely to vote if the ballot initiative benefits them. However, this logic assumes that lower-income populations are willing to participate in politics. In the next section, I will take a look at scholarship on income inequality and political participation.
Income Inequality and Political Participation: Relative Power Theory and Conflict Theory
The literature on income inequality and political participation is inconclusive. Two competing theories seek to explain the effect of income inequality on political participation. The first theory is relative power theory, which states that higher levels of income inequality depress political participation (Solt, 2010; Stockemer, 2017). The second theory is conflict theory, which states that higher levels of income inequality have no effect or even increase political participation (Brady, 2004; Stockemer, 2017; Stockemer & Parent, 2014). Stockemer’s (2017) meta-analysis on voter turnout explicates these competing views and finds that the literature slightly favors relative power theory at 54%. As such, Stockemer’s (2017) meta-analysis reinforces how ambivalent the findings are on the relationship between income inequality and political participation.
Solt’s (2008) analysis examined the relationship between a country’s economic inequality and the citizens’ political engagement. Solt’s (2008) findings fall in line with relative power theory: Economic inequality discourages political interest, discussion, and participation. In particular, economic inequality affects lower-income populations more, corroborating relative power theory (Solt, 2008). Additionally, Solt (2010) examined Schattschneider’s hypothesis at the gubernatorial level in the United States. Once more, Solt (2010) found that income inequality negatively impacts political participation, and income inequality promotes an income bias in the electorate. Anderson & Beramendi (2008) also examine the effect of income inequality on voter turnout. They found that countries with more income inequality experience less electoral participation, and as income inequality increases over time, participation should continue to decrease (Anderson & Beramendi, 2008). Finally, one of Schattschneider’s (1960) hypotheses in his book, The Semisovereign People, states that the electoral gap between lower and higher-income citizens was a result of increasing income inequality (Solt, 2010). As such, Schattschneider’s (1960) hypothesis encapsulates a core idea of relative power theory: income inequality depresses turnout and engenders an income bias in the electorate (Solt, 2010).
Other scholars have found that the relationship between income inequality and political participation is ambivalent or even positive, thereby supporting conflict theory. For instance, Kim et al. (2023) speak on the relative power theory and conflict theory debate by examining social mobility. Kim et al. (2023) state that social mobility moderates the effects of income inequality, such as class conflict and frustration. As such, social mobility is the mechanism by which income inequality promotes or depresses turnout (Kim et al., 2023). In addition, Polacko et al. (2021) examine how polarization affects the relationship between income inequality and political participation. In depolarized systems, income inequality can have a negative effect on turnout; however, as polarization increases, this negative effect is attenuated, lending more credence to conflict theory (Polacko et al., 2021). Lastly, Wilford (2020) finds income inequality has only a small effect on voter turnout. Instead, Wilford (2020) finds that economic hardship has a strong negative effect on political participation, even when income inequality is also featured in their model.
Evidently, the literature on rational choice theory, relative power theory, and conflict theory is inconclusive. This analysis hopes to accompany the rational choice, income inequality, and political participation literature by examining how median income and median housing cost affect vote choice for a minimum wage amendment. Intuitively, one would expect lower-income populations to vote for a minimum wage increase. However, given the rational choice and income inequality literature has found mixed results, further examination is needed to understand this relationship.
I hypothesize the following:
H1: Regions with a higher median income will be less likely to vote for the $15 minimum wage.
H2: Regions with a higher median housing cost will be more likely to vote for the $15 minimum wage.
H3: Regions with higher median incomes and housing costs will be less likely to vote for the $15 minimum wage.
In line with the expectations from rational choice theory, I expect areas with a higher median income to vote against the minimum wage increase. These regions will be less likely to benefit from a minimum wage increase, thereby materializing in a vote against this amendment. However, areas with a higher median housing cost may benefit from a minimum wage increase, given that an increase in housing cost does not necessarily mean an increase in income. As such, I expect regions with a higher median housing cost to vote for the minimum wage amendment. Finally, I expect areas with high median incomes and median housing cost to vote against the minimum wage increase. These regions will host populations who make more money and have high housing costs; theoretically, these populations should be able to afford the higher costs since they also have higher incomes. Therefore, these regions will be less likely to benefit from a minimum wage increase, resulting in a vote against the amendment.
Data, Method, and Procedure
I am using precinct and US Census data for this analysis. The precinct data stems from the Florida Division of Elections (FDoE), and the Voting and Elections Science Team (VEST) at the University of Florida. I will proceed with the analysis in the following way: First, I will clean and organize the FDoE and VEST data. Second, I will perform a spatial join between the newly organized precinct data and US Census data in QGIS. Finally, I will import the spatially-merged dataset into R in order to see the relationship between median income, median housing cost, and vote choice for the $15 minimum wage amendment. I will use OLS analysis and create visualizations to accompany my findings.
I will begin by importing Florida precinct data from the FDoE and VEST datasets. The FDoE dataset has the total vote count for the $15 minimum wage amendment, whereas the VEST dataset includes a shape file that has the coordinates of each precinct. In order to make use of the vote count and the precinct coordinates from each respective dataset, I need to create a new dataset that combines the FDoE and the VEST data. I will start by cleaning and organizing the “No for Rejection” votes; then, I will do the same with the “Yes for Approval” votes. The code down below provides a brief example of the data wrangling in this analysis. I am creating a data frame for each county and precinct that was included in the FDoE dataset but was not present in the VEST dataset. The example shows that Brevard County and precinct 134 were missing in the VEST dataset, so I created a new data frame that includes the county, precinct, vote choice, vote total, and name of the contest in order to match the VEST dataset. This process continues until the new dataset matches the VEST dataset.
#adding BRE134
new_row_state_BRE_134 <- data.frame(
county = "BRE",
precinct = 134,
Vote_Choice = "No for Rejection",
Vote_Total = 0,
Contest_Name = "Amendment No. 2: Raising Florida’s Minimum Wage"
)
statewide_data <-rbind(statewide_data, new_row_state_BRE_134)After completing this step, I can now turn to QGIS. Here, I will perform a spatial join between the newly organized precinct data and US Census data. A spatial join will allow me to aggregate the precinct vote count up to the tract level; this will let me see the coordinates of the vote choice at the tract and precinct level. The spatial join is critical in order to analyze the relationship between US Census variables and vote choice for the $15 minimum wage amendment. The code down below shows the process of loading in the spatially-joined file, and the explanatory variables of interest: median income and median housing cost.
#load in updated dataset
fl_tract_combined <- read.dbf("/Users/mariovillegas/Documents/Grad School/UF/Election Data Science/Combined Shape File Final/Combined Shape File Final.dbf")
#tidycensus variables
census_api_key("46084314d5b1f9a307f6e12da8bf375b1acc6ff3")
acs_vars <- load_variables(2021, "acs1", cache = TRUE)
median_income <- get_acs(geography = "tract",
state = "FL",
variables = "B06011_001",
year = 2020) %>%
select(GEOID, estimate) %>%
rename(median_income = estimate) %>%
rename(GEOID20 = GEOID)
median_housing_costs <- get_acs(geography = "tract",
state = "FL",
variables = "B25105_001",
year = 2020) %>%
select(GEOID, estimate) %>%
rename(median_housing_costs = estimate) %>%
rename(GEOID20 = GEOID)Next, I will combine the spatially-joined file and census explanatory variables together in one large dataset. The code down below shows the process of creating the new dataset: First, I calculate the percentage of “yes” votes in the spatially-joined dataset. Then, I add the explanatory variables to the dataset using two inner joins. Next, I scale the variables for readability in my regression tables. Lastly, I filter a minimum of 25 votes per tract in the dataset. With this new dataset, I can use OLS analysis to see the effect of median income and median housing cost on vote choice for the $15 minimum wage amendment.
#getting percentage of 'yes' votes
fl_tract_combined_tract_votes <- fl_tract_combined %>%
group_by(GEOID20) %>%
summarise(total_vote = sum(G20AME2YES, G20AME2NO),
yes_vote = sum(G20AME2YES),
no_vote = sum(G20AME2NO),
pct_yes = (yes_vote / total_vote) * 100,
count = n())
#join census dataset(s)
fl_tract_census_median_housing <- inner_join(fl_tract_combined_tract_votes, median_housing_costs, by = "GEOID20")
fl_tract_census_median_housing <- inner_join(fl_tract_census_median_housing, median_income, by = "GEOID20")
#scaling median income and housing cost
fl_tract_census_median_housing$scaled_median_housing_costs <- fl_tract_census_median_housing$median_housing_costs / 1000
fl_tract_census_median_housing$scaled_median_income <- fl_tract_census_median_housing$median_income / 10000
#filtering for minimum vote
fl_tract_census_median_housing <- fl_tract_census_median_housing %>%
filter(yes_vote > 25)Results and Graphs
The results of this analysis are seen down below. I will begin by interpreting the regression model. First, we can see that an increase in median housing cost is associated with a greater chance of voting in favor of the $15 minimum wage amendment. This finding supports my hypothesis, as I expected high-cost areas to see the benefit of a minimum wage increase, which would have resulted in a vote for this amendment. Second, an increase in median income is associated with a greater chance of voting in favor of the $15 minimum wage amendment. This finding goes against my original hypothesis, as I expected areas with a higher median income to see little benefit in a minimum wage increase, which would have resulted in a vote against this amendment. Finally, areas that have a high median income and median housing cost have a greater chance of voting against the $15 minimum wage amendment. This finding supports my hypothesis, as I expected areas with a high median income and median housing cost to see little benefit from a minimum wage increase, which would have resulted in a vote against this amendment.
Three scatterplots can also be seen down below. First, I plotted the relationship between median housing cost and the percentage of “yes” votes. As reflected in the regression table, we can see a positive relationship between median housing cost and voting for the minimum wage amendment. The second scatterplot shows the relationship between median income and the percentage of “yes” votes. Once more, we can see the positive relationship between both variables. Finally, the last scatterplot shows the relationship between median income and the percentage of “yes” votes contingent on the median housing cost. If the median housing cost is above $1200, then there is a positive relationship between median income and voting for the minimum wage amendment. However, if the median housing cost is below $1200, then there is a negative relationship between median income and voting for the minimum wage amendment. I will further dissect these findings in the next section.
| Dependent variable: | ||
| Percentage ‘Yes’ Vote Choice for Minimum Wage | ||
| (1) | (2) | |
| Median Housing Cost | 22.1*** | 41.5*** |
| (1.1) | (0.8) | |
| Median Income | 9.5*** | 16.0*** |
| (0.4) | (0.3) | |
| Median Housing Cost X Median Income | -9.5*** | |
| (0.2) | ||
| Observations | 3,547 | 3,547 |
| R2 | 0.9 | 0.9 |
| Adjusted R2 | 0.9 | 0.9 |
| Residual Std. Error | 22.7 (df = 3545) | 15.6 (df = 3544) |
| F Statistic | 12,336.3*** (df = 2; 3545) | 18,698.7*** (df = 3; 3544) |
| Note: | p<0.1; p<0.05; p<0.01 | |
Discussion and Conclusion
In pursuit of this analysis, I sought to add to the extant literature that asks the following question: Do people vote “correctly?” I examined how median income and median housing cost affect vote choice for Florida’s $15 Minimum Wage Amendment. The results clarify and obfuscate this question.
The analysis showed that areas with a higher median housing cost are more likely to vote for the minimum wage amendment. This makes intuitive sense, given that people who live in high-cost areas could benefit from a minimum wage increase. The analysis also indicated that areas with higher median incomes and median housing costs are less likely to support the minimum wage amendment. This, too, makes intuitive sense. People who have a high median income and live in high-cost areas are more likely to afford the increased expenditures; therefore, these areas will be less likely to benefit from a minimum wage increase. However, the analysis also showed that areas with a higher median income are more likely to support the minimum wage amendment. This does not make as much intuitive sense. One would think that areas with a higher median income would vote against the minimum wage increase, given that they are least likely to benefit from it. This also obfuscates the rational choice theory claim that people vote for their own economic utility (Downs, 1957). In this analysis, some areas are voting for other people’s economic utility.
I also gathered some insight into the income inequality and political participation literature. This analysis highlighted that wealthier populations supported the minimum wage amendment more than poorer populations. This finding may indicate wealthier populations were more active on this issue than poorer populations, even though poorer populations would benefit more from this amendment. If this were the case, then we may be seeing an income bias among the electorate, with wealthier populations participating more than poorer populations. This finding would support relative power theory, as relative power theory suggests the presence of an income bias with increasing income inequality (Solt, 2010; Stockemer, 2017). In contrast, this finding would not support conflict theory, as conflict theory suggests more participation from wealthier and poorer populations with increasing income inequality (Brady, 2004; Stockemer, 2017; Stockemer & Parent, 2014). However, it is important to note that I did not directly test for this relationship; the results from my analysis are simply suggestive of this relationship. Future scholars interested in income inequality and political participation should directly test for this relationship.
There are some limitations in this analysis. Potentially the biggest limitation is the ecological inference problem. The ecological inference problem states that it is not possible to make claims about individual behavior based on regional, state, or national-level data. For instance, it would be erroneous to state that every Hispanic supports the Democratic party since the Hispanic community largely votes for the Democratic party. There will always be some level of variation, and this analysis is no exception. While this analysis found that high-cost areas were more likely to support the minimum wage amendment, this does not necessarily mean that everyone in that region will support the amendment. As such, I cannot make any declarations about individual behavior from this analysis. Another limitation is the spatial join performed in this analysis. The spatial join only aggregated precinct-level data up to the tract level; I did not disaggregate the precinct data down to the block level and then reaggregate it back to the tract level. This procedure would have garnered more accurate coordinates and potentially affected the outcome of the analysis. Future scholars should take note of this when engaging in a similar project. Finally, this analysis only looks at Florida, which makes it difficult to generalize these findings to other states. Another problem of analyzing only one state is the limited variation in the variables, which also lowers the explanatory power of my findings. Future scholars should explore a similar analysis in other states to build on the findings in this project.
Despite the mixed results, this analysis still added to the extant literature in the voter behavior domain. It seems that regions experiencing higher housing costs are willing to vote to improve their economic utility, which is to be expected. However, it seems that there is also a case to be made for expressive voting (Carter & Guerette, 1992), in which areas that stand to benefit least from a minimum wage increase are willing to help others who would. This is a welcoming finding that scholars should explore in the political and voter behavior disciplines.
References
Anderson, C. J., & Beramendi, P. (2008). Income, inequality, and electoral participation. In C. J. Anderson & P Beramendi (Eds.) Democracy, inequality, and representation: A comparative perspective, 278-311. Russell Sage Foundation.
Blais, A., & Young, R. (1999). Why do people vote? An experiment in rationality. Public Choice, 99(1-2), 39-55.
Blais, A., Pilet, J. B., Van der Straeten, K., Laslier, J. F., & Héroux-Legault, M. (2014). To vote or to abstain? An experimental test of rational calculus in first past the post and PR elections. Electoral Studies, 36, 39-50.
Brady, H. E. (2004). An analytical perspective on participatory inequality and income inequality. Social Inequality, 667-702.
Brians, C. L., & Grofman, B. (1999). When registration barriers fall, who votes?: An empirical test of a rational choice model. Public Choice, 99(1-2), 161-176.
Carter, J. R., & Guerette, S. D. (1992). An experimental study of expressive voting. Public Choice, 73(3), 251-260.
Downs, A. (1957). An economic theory of democracy. Harper and Row.
Henderson, M., Hillygus, D. S., & Tompson, T. (2010). “Sour grapes” or rational voting? Voter decision making among thwarted primary voters in 2008. Public Opinion Quarterly, 74(3), 499-529.
Hsu, L. C., & Sung, Y. (2002). Experimental evidence on voting rationality and decision framing. Taiwan Economic Review, 30(2), 247-272.
Iyengar, S. (1987). Television news and citizens’ explanations of national affairs. American Political Science Review, 81(3), 815-831.
Iyengar, S., & Simon, A. (1993). News coverage of the Gulf crisis and public opinion: A study of agenda-setting, priming, and framing. Communication Research, 20(3), 365-383.
Kim, D., Kim, M. S., & Lee, S. J. (2023). Income inequality, social mobility, and electoral participation in the US counties: Revisiting the inequality-participation nexus. Political Studies, 71(4), 1129-1150.
Lyons, J., Sokhey, A. E., McClurg, S. D., & Seib, D. (2016). Personality, interpersonal disagreement, and electoral information. The Journal of Politics, 78(3), 806-821.
Polacko, M., Heath, O., Lewis-Beck, M. S., & Dassonneville, R. (2021). Policy polarization, income inequality and turnout. Political Studies, 69(2), 455-477.
Riker, W. H., & Ordeshook, P. C. (1968). A Theory of the calculus of voting. American Political Science Review, 62(1), 25-42.
Schattschneider, E. (1960). The semisovereign people: A realist’s view of democracy in america. Cengage Learning.
Sides, J., Vavreck, L., & Warshaw, C. (2022). The effect of television advertising in united states elections. American Political Science Review, 116(2), 702-718.
Solt, F. (2008). Economic inequality and democratic political engagement. American Journal of Political Science, 52(1), 48-60.
Solt, F. (2010). Does economic inequality depress electoral participation? Testing the Schattschneider hypothesis. Political Behavior, 32, 285-301.
Stockemer, D. (2017). What affects voter turnout? A review article/meta-analysis of aggregate research. Government and Opposition, 52(4), 698-722.
Stockemer, D., & Parent, S. (2014). The inequality turnout nexus: New evidence from presidential elections. Politics & Policy, 42(2), 221-245.
Van der Straeten, K., Laslier, J. F., Sauger, N., & Blais, A. (2010). Strategic, sincere, and heuristic voting under four election rules: An experimental study. Social Choice and Welfare, 35, 435-472.
Wilford, A. M. (2020). Understanding the competing effects of economic hardship and income inequality on voter turnout. Politics & Policy, 48(2), 314-338.
Zaller, J. (1992). The nature and origins of mass opinion. Cambridge University Press.