Ufdcimages.uflib.ufl.edu



Project Title:A Comparison of Voting Behavior on Environmental Referenda in North and South FloridaBy: Morgan NussbaumAbstract: The objective of this study was to identify similarities and differences in voting behavior between voters in North Florida and voters in South Florida regarding environmental referenda, specifically those that promote forest conservation. The drivers behind voting in favor of policies that promote forest conservation to improve water quality were examined through three ways: referendum, county, and voter characteristics. Twenty eight environmental referenda and the voters as well as general population in the counties selected were the focus of the study. The first step was to collect online data from the U.S. Census Bureau database and The Trust for Public land database in the form of county level data. Individual County Election office websites were looked at as well as direct contact with the offices for more information. The results found were statistically analyzed to determine drivers behind voting in favor of environmental referenda. A key finding was that very few variables were statistically significant at identifying drivers behind voting in favor of environmental referenda. It can be concluded from this that certain characteristics do not make a difference in voting behavior towards environmental referenda. Introduction:Water quality has been and will continue to be a serious concern for our world. If water levels are optimal but the quality is subpar, problems will ensue. With more and more areas being exposed to water pollution, and more substances being created that could end up as contaminants in our water supply, water quality is of grave importance to most of the population. In Florida, water quality has long been a matter of public concern. At the level of public policy, certain measures have been developed to gauge the public’s interest in improving water quality. One of these measures is the environmental referendum. A referendum is a direct popular vote on an issue of public policy, such as a proposed amendment to a state constitution or a proposed law. Referendums, which allow the general population to participate in policymaking, are not used at the national level, but are common at the state and local levels. A referendum is often used to gauge popular approval or rejection of laws recently passed or under consideration by a state legislature. A referendum can also be used to initiate legislative action (American Heritage 2005). The allocation of limited public funds to programs that protect public goods is often determined through referenda placed on ballots during state and local elections (Kline & Wichelns 1994). Through the ballot box, individuals’ voting choices reveal public preferences regarding public goods (Deacon & Schlapfer 2010). Environmental referenda in Florida, then, are designed to reveal citizens’ preferences regarding environmental goods.The state of Florida is diverse, looking at voting behavior within regions can help to explain this diversity. The Northern and Southern regions of Florida have many differences in terms of overall characteristics which may lead to differences in natural resource policies. The amount of vegetation varies throughout the state, there are multiple climates, and population density is greater in the South region of the state. While water quality is a growing concern for Florida as a whole, the two regions’ voting behaviors on environmental referenda can be compared to validate the differences which exist between the two. The North and South regions of Florida, for this research, will be delineated simply by cutting the state in half (see the map of Florida (Figure 1) in the final section of this paper). The state is divided in half by watershed boundaries that the water management districts of the state follow. This map has a detailed layout of all the counties within the state, which is the location parameter used in this research. Comparing characteristics of referenda, the features of different counties, and the demographics of voters within Florida counties may uncover voters’ willingness to pay for this type of policy. The knowledge gained from this study will benefit policy makers in that we will better understand voters’ preferences due to their specific characteristics. Environmental policy makers may choose to use this information to increase favorable votes on environmental protection referenda. The objective of this research is to identify similarities and differences in voting behavior between North and South Florida voters regarding environmental referenda. It will also try to determine the drivers behind voting in favor of policies that promote forest conservation to improve water quality. The four main questions are as follows:In counties throughout North and South Florida,Are there significant differences in voting behavior?Do differences exist in referendum characteristics?Do differences exist in county characteristics?Do differences exist in voter characteristics?It is thought that significant differences exist in preferences for water quality protection programs between the North and the South, which can be uncovered through answering the previous questions. The study will use a three category framework to address these questions: referenda, county, and voter characteristics. Materials and Methods:The main method used was data collection from online databases. Data was retrieved from the U.S. Census Bureau database and from The Trust for Public land database in the form of county level data. Data was compiled directly from these data bases; when data was missing it was estimated using existing data via regression analysis (Statistical Package for the Social Sciences, SPSS). Individual County Election office websites were utilized as well as direct contact with those offices for more information about voting and referendum characteristics. The following provides a list of variables analyzed in the study:Referendum characteristics:Percent of referendums held in North and South countiesPercent of referendums voted on at the county level and municipal levelPercent of referendums that passed or failedIf the time period for raising funds was stated in the referendumIf maximum amount of funds potentially being utilized was stated within referendumIf the referendum stated funds would pay for more than environmentally significant lands (i.e. roads, infrastructure, etc.)If the process used to conserve land was stated in the referendumIf the implementing organization was stated in the referendumIf the referendum was funded through a bond or a taxHow many years since 1988 in terms of when the referendum was heldPercent of referendums that passedNumber of words in referendumDesignated funds for the referendum (2013 dollars)Designated funds per number of residents over the age of 18 County characteristics:Percent of residents over the age of 18Percent of whites in countyPercent of males in countyMedian household income in 2013 dollarsMedian age in countyPercent of residents with over 16 years of educationPercent urban population in countyWater use total withdrawal Water area per square mileVoter characteristics:Percent registered to vote that actually votedPercent of ballots that voted on the referendumPercent that voted in favor of the referendumThe likelihood of a yes votePercent that voted democrat in the last presidential electionOnce the data was compiled into Excel spreadsheets, the software Statistical Package for the Social Sciences (SPSS) was used to conduct multiple statistical tests concerning the significance of each variable. Descriptive statistics were generated (i.e., percent’s, mean, max, min, and standard deviation), box plots were analyzed, and a one way ANOVA was performed on every variable. A simple regression analysis was performed as well. However, none of the variables were significant predictors of a yes vote. This may be because of the small sample size used and available for study. The simple regression analysis added no helpful information, so it was removed from the study. Each variable utilized will be explained in the following section. Results:3.1: Referendum data: Initially, water management districts were considered as a way of viewing the differences in referenda; however, there was not an equal number of referenda in each district. Choosing to view differences in this framework would have resulted in biased outcomes due to a large number of referenda being held in some districts and a small number in other districts. Instead, looking at differences between the North and South provided more helpful information. Sixteen of the twenty eight referenda used in this study were in the South (57% of the total observations) and the other twelve referenda were held in the North (43% of the total observations) (Figure 2). Five of the referenda studied were voted on at the county level (18%) and twenty three at the municipal level (82%) (Figure 3). The year each referendum was voted on was then analyzed in order to see whether or not this would be significant. All referenda used in this study were voted on between the years of 1989 and 2005. Two referenda were voted on in 1989 (7%), thirteen were voted on throughout the 1990’s (46.5%), and thirteen were voted on in the first half of the 2000’s (46.5%) (Table 1). An overwhelming number of referenda included in this study passed (86%). This shows that, overall, voters are concerned about water quality. Only 14% of referenda studied did not pass (Figure 4). Additional variables pertaining to different characteristics of what was included in the language used in the referenda were then assessed. Of the referenda studied, 71% stated the time period in which the funding mechanism used would be financed. For those that would be using a tax it would state how long that tax would be in effect. For those that would be using a bond it would state how long it would take to pay that bond back. Only 25% of referenda did not state the time period in which the funding mechanism used would be financed, and 4% were unknown (Figure 5). Of the referenda studied, 61% stated the maximum amount of funds that would be raised through the proposed payment vehicle within the referendum itself, 36% did not state a max amount, and 3% were unknown (Figure 6). Very few referenda studied would pay for more than environmentally significant lands (36%), such as roads and additional infrastructure, 61% would not, and 3% were unknown (Figure 7). Around 68% of referenda studied stated the process that would be used to conserve the land (e.g., acquisition, outreach) to protect water quality, 25% did not and 7% were unknown (Figure 8). Around 79% of referenda studied did not state the organization that would be implementing the referendum if passed, while 21% did state who would implement the program (Figure 9). The method of funding for each referendum was quite close, 50% would be funded through some sort of bond, 46% through a tax, and 4% unknown (Figure 10). Continuous variables were also assessed in the study. The information below is summarized in Table 1 in section VII. The average number of words per referendum was 71(10.18). This was looked at to see whether or not the length of the referendum would influence voters one way or the other. For the referenda that passed, the designated conservation funds approved in the referendum averaged $93,135,539 (in 2013 dollars). Looked at a little differently, an average of $400.55 per person over 18 would be utilized in that particular referendum. This variable was considered to get an idea of scale based on the population, although this is not actually what each person would pay. 3.2: County data:Descriptive statistics for the variables that characterize the counties studied are described below and summarized in Table 2 of section VII. Averages will be given followed by the variable specific standard deviation in parentheses. On average, 78% (2.99%) of residents in the counties studied were over the age of 18. Eighty-seven percent of the population in the studied counties was Caucasian. However, only 48% of Davie (Broward) county population was Caucasian and nearly 100% of Collier County consisted of Caucasian residents. The average population in the studied counties was half males and half females around 50% each (S.D. 2.9%). The average median age in the studied counties was 41 (5.5). The minimum age of 30 was the median age in Leon County and 50 was maximum median age in Sarasota County. On average, only 25 %( 7.06%) of the residents in the selected counties had over 16 years of education. Polk County had the lowest education level with only 13% of residents with 16 years or more of education and Leon County had the highest with 42%. The median annual household income for the selected counties was $54,463($7,733). The lowest median income was $44,275 in Miami-Dade county and highest was $72,921 in St. Johns County. Next population density was considered. “Urban” was defined as a population of greater than one thousand residents per square mile. On average, the urban population in the selected counties was 87 %( 14.98%). Only 50% of Fernandina Beach County was considered urban population and nearly 100% of Davie (Broward County) was urban. The total water withdrawal per person, per year was looked at for scale as well. This accounts for all the water that goes into the public water supply, agriculture, and energy. For the counties studied the average was 853 Million Gallons per Day (703.83) (MGD). Next, the square mile amount of visible surface water areas such as lakes, rivers, and estuaries in the counties studied was estimated. On average, 0.19(0.10) square mile of the counties studied made up visible surface waters. 3.3: Voter data: Variables related specifically to voter characteristics are discussed below and summarized in Table 3 in Section VII. On average, 63 % (20.9%) of registered voters in a particular county actually voted. However, a large amount of variation occurred among counties, only 13% of registered voters actually voted on these referenda in Davie (Broward County), and 91% voted in Palm Beach (Martin County). The variable explaining percent of ballots voted on looked at who actually voted on the referenda. Registered voters can go and vote, but some may only vote on referendum items that they are interested in or know about. On average, 89 %( 16.9%) of the ballots cast included a vote on the referenda pertaining to forest conservation. There was some variation as only 31% of ballots in St. Augustine beach consisted of a vote on referenda and almost 100% of ballots in Martin (Palm Beach) consisted of a vote on referenda. On average, 60 %( 9.8%) of voters voted in favor of the referendum they were voting on. The likelihood of a yes vote was then assessed. Following (Deacon & Shapiro, 1975) the dependent variable is a logit transformation of the percent of voters approving the referendum: 1)Where ln is the natural logarithm and P(Yesm) is the percent of voters who voted yes to the referendum in county or town m. This variable describes the ratio of the percent voters approving the referendum to the percent rejecting the referendum. Observations less than 0.00 are associated with a referendum that failed and observations equal to or greater than 0.00 are associated with a referendum that passed. With this in mind, the average observation was 0.41(0.43), meaning on average most referenda passed. There were, however, not a lot of near misses. Whether or not voters had voted “Democrat” in the previous presidential election was then looked at, due to studies suggesting that Democrats may be more open to causes such as environmental referenda. On average, 41 % (0.43) of voters had voted Democrat in the previous presidential elections. 3.4: Bar graphs, box plots and ANOVA:All variables were then analyzed for significance using bar graphs, box plots, and a one way Analysis of a Variance (ANOVA). Very few came up to be significant predictors of differences between the North and the South. This section describes the few variables that were significant. The results will be assessed within the question framework stated in the introduction. The ANOVA added answers to all questions and therefore will be mentioned more than once (Table 4).Question 1: A key finding was that the main variables of study: whether or not the referendum passed and the likelihood of a yes vote, were both not significant and had no difference between the North and South. The likelihood of a yes vote had a p value of 0.63, and the percent of referendums that passed had a p value of 0.77 (Table 4). Question 2: Bar charts were used to identify the differences between South and North on whether or not the process used to conserve land is stated in the referendum and whether or not the implementing organization was stated in the referendum (Figure 11 and 12). From figure 11 it can be inferred that there was much more variation in the process being stated in the North than there was in the South. Almost all of the referenda in the North stated the process, where as in the South about half of the referenda stated the process. This variable was significant at the 0.10 level (p = 0.09) (Table 4). From Figure 12 it can be inferred that much more variation occurred in the South on whether or not the implementing organization was stated in the particular referendum. Almost all did not state the implementing organization in the South, where is in the North about half stated the implementing organization. This variable was significant at the 0.05 level (p = 0.02) (Table 4). Question 3: In Figure 13 we see the percent of urban population compared between the North and South counties. The percent of urban areas in voting counties was high in the South and had more variation in percentage in the North. However, the North had more variation in counties with urban population. This variable was significant at the 0.05 level (p = 0.002) (Table 4). Question 4: Figure 14 compares the percent of registered voters that actually voted with the North and South variable and shows that the South has much more variation in percent of residents that voted as compared to the North. The percent registered voters that actually voted on environmental referenda was significant at the 0.05 level (p = 0.03), meaning there is only a 5% chance that the difference between variables can be attributed to random chance. 4. Discussion:From the results above we can conclude that there are significant differences in certain variables compared in the North and in the South, but none played a significant role in determining whether or not the referendum in question passed or the likelihood of the population voting yes on the referendum. Many differences exist between the North and South, and this study shows what factors do not influence voting behavior towards environmental referenda to improve water quality. One of the main limitations of this study was the sample size. The study raises important questions but the number of referenda available for study was not large enough to statistically justify any broad generalizations about differences within the North and South. In answer to Question 1, there was no statistical difference in voting behavior between the North and the South when using these parameters. In answer to Questions 2 through 4, there were statistical differences in some of variables within the three characteristic sections studied. However, these variables did not influence voting behavior. Further studies with a larger sample size can lead to more significant information and inferences about what parameters cause significant differences in voting behavior in the North and in the South. This study provides a solid foundation to continue researching this information. 5. Literature Cited:Deacon, R., & Shapiro, P. (1975). Private Preference for Collective Goods Revealed Through Voting on Referenda. The American Economic Review, 65(5), 943-955. doi: 10.2307/1806631 Kline, J., & Wichelns, D. (1994). Using Referendum Data to Characterize Public Support for Purchasing Development Rights to Farmland. Land Economics, 70(2), 223-233. doi: 10.2307/3146324 The American Heritage New Dictionary of Cultural Literacy. (2005). Referendum definition. . 23 February 2014. Retrieved from Unknown. (2008). County Court. State of Florida court system. 23 February 2014. Retrieved from 6. Tables and Figures:Figure 1: North and South Florida counties defined397860010731500147637510477500Percent of referendums held inNorth and South countiesPercent of referendums voted on at the county and municipal levels43%57%SouthNorth82%18%CountyMunicipalFigure 2. Percent of referendums held in North and South Counties Figure 3. Percent of referendums voted on at the county and municipal levels3954780-571500Figure 4. Percent of referendums that passed or failed Figure 5. If the time period for raising funds was stated in the referendum3963670-635000Figure 6. If maximum amount of funds potentially being utilized was stated in the referendum Figure 7. If the referendum stated funds would pay for more than environmentally significant lands4024630000Figure 8. If the process used to conserve land is stated in the referendum Figure 9. If the implementing organization is stated in the referendumFigure 10. If the referendum was funded through a bond or a tax331152522352000451294522352000625602022733000 Table 1. Descriptive Statistics of continuous variables describing referendum characteristics. # OfStd. obs. Minimum Maximum Mean Deviation Years since 198828117.0010.394.95Percent of referendums that passed280100.0086.0036.00Number of words in referendum264894.0071.4610.18Designated funds (2013 dollars)280746,618,06893,135,539168,280,198Funds per person over 182802,878.25400.56726.17388302518224500491299518605500520700018605500550926018224500 Table 2. Descriptive Statistics: County data: continuous variables # ofStd. obs. Minimum Maximum Mean Deviation Percent of residents over age 182970.3184.1078.572.99Percent white in county A2847.90100.0+86.6611.78Percent male in county2847.2054.8050.362.19Median household income in 2013 dollars2844,27572,92154,4637,733Median age in county2829.5050.1041.175.50Percent of residents with over 16 years of education2813.2041.7024.827.06Percent urban population in county2849.40100.0+86.5714.98Water use total withdrawal (MGD)B28161.003191.00853.89703.83 Water area per square mile 28 0.05 0.54 0.19 0.10 38868355080004516755508000520700050800059448705080006569710508000A Some estimates were calculated to exceed 100% because it used predicted values calculated between census years. B Million gallons per day394081019240500525716519621500611632019240500 Table 3. Descriptive Statistics: Voter Data: continuous variables # of obs. Minimum Maximum Mean Std. Deviation Percent registered that voted2813.2090.9462.6820.91Percent of ballots that voted on referendumA2831.12100.00+88.6916.87Percent that voted in favor of referendum2834.9082.4959.759.77The likelihood of a yes vote28-0.621.550.410.43Percent that voted democrat in the last2418.0064.0041.0013.00presidential electionA Some estimates were calculated to exceed 100% because election office records were incomplete ormissing therefore some values were estimated using data from previous elections.Figure 11. Bar graph of the differences between South and North on whether or not the process used to conserve land is stated in the referendumFigure 12: Bar graph of the differences between South and North on whether or not the implementing organization was stated in the referendumFigure 13: Box plot of the percent urban population within the counties compared between theNorth and the SouthFigure 14: Box plot of percent registered to vote that actually voted compared in the North andSouthTable 4. Analysis of variance table describing differences among voting behavior, registeredvoters and other referendum characteristics for environmental referendums held in north Florida391795017208500 compared to south Florida. Mean Square df F (Between groups) p The likelihood of a yes vote270.240.050.63Percent of referendums that passed270.090.010.77Percent registered that voted275.502061.930.03**If process was stated in referendum253.230.610.09*If implementing organization was stated275.800.860.02**392176017145000421322517145000480885517145000594487017145000 Percent urban population in county 27 11.32 1836.34 0.002** Note: * Equals significant at the p<0.10 level. ** Equals significant at the p<0.05 level. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download