Map Portfolio

Geography 435: Advanced GIS Final Project

Geography 200: Geographic Techniques

 Geography 200: Geographic Techniques

Capstone Project

Changes in Student Enrollment across Wisconsin School Districts

Introduction

There are many factors that affect changes in student enrollment in school districts across the United States. These factors may include broad demographic variables that vary from the type of school district whether it is urban, suburban, or rural. Many previous studies have focused on a handful of factors to fully explain the change in how many students a school district has over time. Although there are several social and economic factors that are used to analyze these trends, this study will focus on a combination of them to explain the current patterns of school enrollment in Wisconsin school districts. Variables such as median age, property values, and percent Hispanic are important for examining how recent trends of student population in Wisconsin school vary spatially throughout the state.

Not only does the decline of student enrollment pose a threat to a school district in maintaining its resources but it also threatens that institution’s number one goal, which is to provide a quality education to kids in the community it serves. Student success and achievement has always been the aim of a school and without strong communication and cohesiveness, closings or consolidation are likely outcomes. As the study has found the closing down of schools from significant enrollment declines can also happen in growing suburban areas. 

Data Collection

The report includes the collection of enrollment numbers from 1997 to 2011, done by downloading a shapefile of school districts and population variables including median age, race, property values, and income per capita per school district. Birth rate per county is also collected for further analysis as other studies have indicated its importance in the subject matter.

Data is compiled at various scales and can include areas with different boundaries. Depending on the focus of the study, the scale is very important especially for analyzing data about demographics over a large area that contains a significant population. Studies may look at scales including: state, county, municipality, census tracts, block groups, school districts, etc. However, there is overlap between all of these boundaries, which makes it difficult for studies dealing with demographics at multiple scales. This project uses school district boundaries as the scale to analyze their changes among the student population over a period of time.  

Enrollment data for the unified and secondary school districts of Wisconsin are downloaded. This data can be found under the data section of the Wisconsin Department of Instruction (DPI) website. This institution gives enrollment totals in Microsoft Excel files for every school year from 1997 to 2011. The DPI also provides a variety of data about the student population in Wisconsin stratified by gender, race, percent poverty, etc. The years collected for the study include 1997, 2000, 2002, 2003, 2006, 2007, 2008, 2009, 2010, and 2011. From this data, percent change is calculated from the years 1997 and 2011 in Excel. Percent change from 2003 and 2006 to the most current year provided by the DPI (2011) is also calculated to examine how dramatic the changes are within recent years (Wisconsin Department of Public Instruction 2012).

Another necessary set of data in the study is a shapefile from the US Census of unified and secondary school districts, which can be downloaded from the 2010 Census TIGER/Line Shapefile section of their website. Elementary school districts are omitted from the study because of the overlap with secondary districts. Also, since the data for secondary schools covers the same area that contain the elementary districts, data about the population will still be accounted for that space. Analyzing school districts in this way makes it feasible to interpret population variables across the state at a more manageable scale.

The last step in data collection is to gather and download data about the demographics of each school district, which can be retrieved from the National Center for Education Statistics (NCES). There is plenty of data aggregated to the geography of school districts in the United States including median age, property values, and per capita income. All of the data for these variables is collected by the five year 2006-2010 American Community Survey (ACS) provided by the census, which is an ongoing survey that gives communities current information needed to plan services and investments. The ACS randomly samples addresses in every state and continues all year, every year. Many communities use the survey because it has been successful in giving accurate estimates even though there is a sampling error involved. On the other hand, race is retrieved from the 2010 Summary File 1 (SF1), which is based on the Decennial Census and ordered every ten years by the US Constitution. This gives an accurate count of all persons living in the US to reapportion seats in the US House of Representatives. The other variable in the methods looks at race including Hispanic and White. Birth rate is the last variable, which is collected from the Wisconsin Department of Health Services (DHS). The DHS provides birth rates dating back to 1989 at several levels including state and county. This data is collected at the county level to show the spatial pattern of birth rates across the state for 1989 and 2011. 

Methodology

This study utilizes a variety of methods to analyze the patterns of school enrollment change and the relationship with several population variables. The program ArcGIS is used to manipulate and display the data on a map. Examining the data spatially allows the researcher make conclusions about the trends and patterns of school enrollment and several population variables. In order to conduct a spatial analysis of all the Unified and Secondary school districts of Wisconsin this study will include following methods: a correlation matrix between the variables to determine the strength and direction of the relationship between them, areal interpolation, a tool in ArcGIS to determine the percent Hispanic and White for each school district, and division of school districts by type (urban, suburban, and rural) to illustrate how the changes affect certain areas in the state.

The first method is a correlation matrix between the variables, which shows how a number of variables correspond to one another. The objective of the correlation matrix is to find the linear association of the population variables with the percent change and show the strength and direction of the relationship. Correlations are calculated in the program SPSS statistics 19, which also creates a scatterplot for each correlation showing the linear relationship between each variable.

The second method is utilizing the Areal Interpolation tool in ArcGIS. This procedure reaggregates data from a set of polygons to another set of polygons. Since percent Hispanic is provided at the municipal level, the tool allowed for a downscale of data to predict those values for school district boundaries. This method lines up the municipal boundaries with the school district boundaries to estimate percent Hispanic within each polygon. Average or Gaussian data is used in the areal interpolation tool to produce a prediction and standard error surface that can be reaggregated to target polygons. Gaussian data is data averaged over defined polygons, which began as municipal boundaries in Wisconsin. After a prediction surface is created for the value of the Gaussian variable, in this case percent Hispanic, at all points in the data domain, the tool can predict the average value of the variable for the school district boundaries. This tool is new to ArcGIS 10.1, and it is very accurate in its estimations. Percent Hispanic is used because previous studies such as the one done by Sarah Kemp in 2007 show the school districts with a higher Hispanic population tend to increase in student enrollment. The White race variable can also be used with the same tool to show areas with the least diversity.

The third method involves dividing the school districts into three categories: urban, suburban, and rural. This is done to analyze the differences in changes in enrollment for certain areas throughout the state. According the NCES, an urban school district encompasses a large city with a population greater than 50,000. This can be done in ArcGIS through a query, which resulted in 12 school districts containing a population over 50,000. To find suburban school districts a select by location is run and includes schools that are adjacent to the urban districts or are within an urban cluster. The query found 94 suburban schools in the state of Wisconsin and the remaining 272 are classified as rural or exurban. Figure 1 shows a map of this classification of the three different types.

Figure 1: Wisconsin School Districts by Type- NCES Criteria

 Source: NCES & ESRI

Finally, the last method is to do an analysis of all the maps and display the results with the selected variables for a comparison to the map showing percent change. The correlation matrix can also be used in comparing a certain variable that has the strongest relationship with percent decline and draw logical conclusions from.   

Results

After further analysis of the relationship between the population variables and percent change from 1997-2011, three variables showed the strongest association to changes in school district enrollment: property value, median age, and percent Hispanic. Four maps including one for each of the variables were manipulated in ArcGIS to display the findings. Although these three variables are not the only the contributing factors to a change in enrollment, they show that there is a significant linear relationship to both decline and growth. 

Figure 2 shows the percent change in Milwaukee area school districts from 1997 to 2011. Wisconsin's largest metropolitan area differs from others in the state in that there are a number of suburbs that are dealing with significant declining enrollment. While the City of Milwaukee is declining the most in the region, suburban school districts in the north shore such as Shorewood, Nicolet, and Mequon-Thiensville have been losing more students than the other suburban schools. Although some communities such as Mequon have plenty of room to grow, other suburbs such as Glendale and Shorewood have run out of space for new development. A closer look at the demographics for each community should be examined to see what could be triggering this trend. 

Figure 2: Percent Change in Milwaukee Area School Districts 1997-2011

Source: Wisconsin Department of Instruction & ESRI

Figure 3 displays percent change in enrollment of all Wisconsin school districts and illustrates that the majority of the schools are in a declining trend. The shades of blue indicate a positive increase in student enrollment while the lighter color and shades of green reveal a decline. The map shows that the darker the shade the more extreme the change is for each school district.

Wisconsin is predominately a rural state (Figure 1) and the map in Figure 3 shows that the majority of the rural school districts are declining. In fact almost 80 percent of all rural school districts in Wisconsin are experiencing a degree of declination. These areas are primarily located in the northern or western part of the state. Three school districts (Park Falls, Glidden, and Weyerhaeuser) have consolidated to adjacent school districts between the years of 1997 and 2011. The Weyerhaeuser school district in the north consolidated to Chetek in 2006 after a significant decline in student enrollment, which has plagued the district for many years before, became insurmountable. This school district is marked at losing 100 percent of its student population on the map in Figure 3. 

On the other hand, there are some rural school districts that have remained stable or increased in the time period that is covered by the study. These schools are located near the urban centers (Madison, Milwaukee, Green Bay, La Crosse, and the Twin Cities). Rural schools that are increasing can be classified as exurban or just outside of the suburban cluster. They are also located in some of the fastest growing regions of Wisconsin and may soon be classified as a suburban school district by the NCES in the near future.

For urban school districts in Wisconsin, half are increasing and half are decreasing. The Milwaukee school district is experiencing the greatest decline and others include La Crosse, Oshkosh, and Racine as losing the most students out of the group. However, the other urban schools that are increasing tend to be the ones that are still growing as a community and they include Kenosha, Madison, Waukesha, and Appleton. The Kenosha school district located in the southeast corner of the state has grown at 17.6 percent from 1997 to 2011. This district has also added another high school (Indian Trail) in 2011 to capacitate the amount of students in the buildings.

Suburban schools compose most of the blue shades in Figure 3 as they are located around the urban school districts. The study found that 66 percent of suburban schools are stable or increasing in student enrollment. Although the majority of suburban districts are experiencing healthy growth, there are a few that are declining at a significant rate greater than 10 percent and are dispersed throughout the state. These school districts include Two Rivers, Marinette, Nicolet, Mequon-Thiensville, Superior, and Wisconsin Rapids..

Figure 3: Percent Change in Enrollment of all Wisconsin School Districts 1997-2011

Source: Wisconsin Department of Instruction & ESRI

Property values per school district shown in Figure 4 have the strongest correlation with percent change from 1997 to 2011. There is a low positive correlation between the two variables at 0.444 (Figure 5), which indicates that as property values increase, school enrollment is likely to increase in the state of Wisconsin. The study affirms that the rural school districts with higher property values tend to be stable or increasing compared to less affluent rural districts. The map below shows high property value school districts located in the surrounding suburbs of the Milwaukee and Madison urban areas in the south. Most of the low property value school districts are located in the western and parts of north central Wisconsin. Rural school districts tend to be the ones with lower property values as they are isolated away from the cities.

Figure 4: Property Values per School District 

Source: 2006-2010 American Community Survey & ESRI

Figure 5: Percent Change (1997-2011) and Property Value per School District Correlation Output

Calculated in SPSS Statistics 19

Figure 6 shows a map of median age per school district in Wisconsin. The majority of the older populations reside in Northern Wisconsin; however there are a few pockets of high median age in central Wisconsin. Mequon-Thiensville is the major outlier in southeast Wisconsin as this is an aging community outside of Milwaukee. The correlation matrix in Figure 7 shows a low negative correlation with this variable and student enrollment change. In other words, as median age increases, the student population is more likely to decrease. A school district with a high median age has a higher probability of decreasing than one with a young population. The rural school districts in the north have older populations and are experiencing a significant decline. Mequon-Thiensville just north of Milwaukee has a significantly older population than the surrounding districts and is also experiencing a significant decline of almost 15 percent since 1997.

Figure 6: Median Age per School District

Source: 2010 Census & ESRI

Figure 7: Percent Change (1997-2011) and Median Age per School District Correlation Output

Calculated in SPSS Statistics 19

Figure 8 shows the percent Hispanic per school district in Wisconsin. The study done by Sarah Kemp showed that Hispanic families have a higher birth rate and thus have more kids to send to school. The purpose of including this variable with student enrollment is to analyze the relationship between the areas with high Hispanic populations and changes in student enrollment. This map reveals larger Hispanic populations in the southeast and urban parts of the state, however there are rural school districts dispersed in the center and western portions of the state that have a significant percentage of this race. These school districts are showing a positive growth in the number of students, which to name a few include Sauk Prairie, Abbotsford, and Arcadia. The correlation matrix shows that there is a significant linear relationship between percent Hispanic and student enrollment. The Pearson value of 0.255 reveals a low positive correlation indicating that as percent Hispanic increases, student population also increases (Figure 9). However, this is not always the case for every school district such as Milwaukee or Racine. Even though these schools have significant Hispanic populations within their districts, they are experiencing a decline.

Figure 8: Percent Hispanic per School District

Source: 2010 Census & ESRI

Figure 9: Percent Change (1997-2011) and Percent Hispanic per School District Correlation Output

Calculated in SPSS Statistics 19

Figure 10 shows how the areal interpolation tool in ArcMap created a prediction surface by taking the percent Hispanic for each municipality and recording a value over each individual polygon. The next step in this process is to lay down the school district boundaries over this surface to estimate what the percent Hispanic value would be for each one.

Figure 10: Prediction Surface Output of Percent Hispanic using the Areal Interpolation tool in ArcMap

 

Source: US Census & ESRI

Conclusion

There are a number of factors that contribute to the decline of student enrollment in school districts. The majority of schools in the US are experiencing a decline of students because of changes in the population. This study analyzed property value, median age, race, and birth rate data for each school district to explain the reason for decline.

An important variable of the study that has the strongest relationship with enrollment trends is property values per school district. There is evidence showing that more affluent school districts tend to be more stable in their numbers over the last fourteen years. The effect of property values however is severely more felt in rural school districts as the more affluent ones are increasing and the less affluent schools are decreasing. Although there is a higher probability of enrollment to increase with property values increasing, that is not always the case. Mequon-Thiensville school district ranks second in having the highest property values and first for the highest income per capita in the state, but they are declining by almost 15 percent since 1997. There are numerous other factors to consider when explaining this trend for certain school districts.

Percent Hispanic is another variable used in the study that is also related to changes of enrollment. The rural school districts such as Arcadia and Abbotsford have been stable with student populations as they have higher Hispanic population. This shows that Hispanics tend to have more kids than other races and provide stability or increases to a school district. On the other hand, school districts that are predominantly one race and less diverse tend to be declining. When percent white was examined, it showed that schools with high white populations weren’t increasing but also not dramatically decreasing either.

There are a few limitations to the data that was used such as the median age per school district. Although this variable is a strong indicator of explaining school enrollment trends, it is heavily influenced in areas that have a college or university. The median age in these places are slightly skewed due to the young college population residing in them. This brings down the median age for these regions, but does not greatly affect the study’s overall results.

Monitoring the trends of school district enrollment is a complex subject that takes numerous broad demographic variables to fully explain its patterns. Future studies may find other variables useful such as proximity to private or other public schools, school quality by using ACT scores, or the unemployment rate for each district. An open enrollment policy is another variable that is changing schools in the US. Nevertheless, it is important for school administrations to pay close attention to the data and trends from within their own districts. Communication is important for a school district to be cohesive and conduct smart planning goals for the future. Schools that are careful with their budgets will monitor enrollment trends while monitoring their resources to accomplish their number one goal, which is to ensure the success and achievement of the student.