The Impact of Bicycle Facilities on Single Family Residential Property Values

MISSING IMAGE

Material Information

Title:
The Impact of Bicycle Facilities on Single Family Residential Property Values Evidence from Alachua County, Florida
Physical Description:
1 online resource (76 p.)
Language:
english
Creator:
Szatmary, Kevin
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.A.U.R.P.)
Degree Grantor:
University of Florida
Degree Disciplines:
Urban and Regional Planning
Committee Chair:
BEJLERI,ILIR
Committee Co-Chair:
ALAKSHENDRA,ABHINAV

Subjects

Subjects / Keywords:
bicycle -- lanes -- property -- trails -- values
Urban and Regional Planning -- Dissertations, Academic -- UF
Genre:
Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Bicycle lanes and trails are an important part of the transportation infrastructure of a city. They provide recreational and health benefits, as well as an alternate means of transportation for commuters. In some cities in the United States, where bicycle culture is growing, the value of properties with close proximity to these amenities has increased. The objective of this study was to assess the impact that bicycle lanes and trails provide to single family residential property values. This was accomplished by using ordinary least squares regression to create a hedonic pricing method, followed by a locational study using geographically weighted regression. The study was performed in Alachua County, Florida, using geographic information systems (GIS). The analysis showed that the impact of bicycle lanes and trails varies by location. The use of ordinary least squares does not take location into account, therefore geographically weighted regression was employed to consider spatial variation. It was found that in some locations, proximity to a bicycle facility decreased the value of homes, possibly stemming from issues regarding decreased privacy or increased crime. In other locations, property values were increased at homes located near bicycle facilities, as a result of commuting ease or recreational benefit.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Kevin Szatmary.
Thesis:
Thesis (M.A.U.R.P.)--University of Florida, 2014.
Local:
Adviser: BEJLERI,ILIR.
Local:
Co-adviser: ALAKSHENDRA,ABHINAV.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2014
System ID:
UFE0046834:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 THE IMPACT OF BICYCLE FACILITIES ON SINGLE FAMILY RESIDENTIAL PROPERTY VALUES: EVIDENCE FROM ALACHUA COUNTY, FLORIDA By KEVIN SZATMARY A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS IN URBAN AND REGIONAL PLANNING UNIVERSITY OF FLORIDA 201 4

PAGE 2

2 201 4 Kevin Szatmary

PAGE 3

3 To my cats, Wheezy and Mr. Jigglesworth, for supporting my academic endeavors and only accidentally turning off my computer once

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my Chair, Ilir Bejleri, and my Co Chair, Abhinav Alakshendra, for their guidance and support. I learned a considerable amount as a result of t heir assistance. I would also like to thank my teachers who have supported me and given me advice; I never would have made it without their help. Additionally, I would like to thank my classmates, especially Sarah Benton, for reminding me of due dates Las tly, I would like to thank my amazing girlfriend, Katie Skorpinski, for putting up with my endless shenanigans.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURE S ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 Problem ................................ ................................ ................................ .................. 12 Research Questions and Objectives ................................ ................................ ....... 12 Method ................................ ................................ ................................ .................... 13 2 LITERATURE REVIEW ................................ ................................ .......................... 15 Bicycle Lane and Trail Impacts ................................ ................................ ............... 15 Bicycle Facilities in Alachua County ................................ ................................ ....... 17 Valuing Residential Properties ................................ ................................ ................ 18 Variables ................................ ................................ ................................ .......... 18 Hedonic Pricing Model ................................ ................................ ..................... 20 Geographically Weighted Regression ................................ .............................. 21 Study Limitations ................................ ................................ .............................. 21 Case Studies ................................ ................................ ................................ .......... 22 Opinion Based Studies ................................ ................................ ..................... 22 Hedonic Pricing Technique Studies ................................ ................................ .. 23 Variables Used ................................ ................................ ................................ 27 3 METHODOLOGY ................................ ................................ ................................ ... 28 Case Study ................................ ................................ ................................ ............. 28 Data Collection ................................ ................................ ................................ ....... 31 Parcels ................................ ................................ ................................ ............. 31 CAMA Tables ................................ ................................ ................................ ... 31 Bicycle Facilities ................................ ................................ ............................... 32 Census Data ................................ ................................ ................................ ..... 32 Combined Dataset ................................ ................................ ............................ 33 Data Limitations ................................ ................................ ................................ 33 Regression A nalysis ................................ ................................ ............................... 34

PAGE 6

6 4 FINDINGS AND RESULTS ................................ ................................ ..................... 38 Ordinary Least Squares (OLS) ................................ ................................ ............... 38 Alachua County ................................ ................................ ................................ 40 500 feet ................................ ................................ ................................ ............ 40 1,000 feet ................................ ................................ ................................ ......... 41 2,000 feet ................................ ................................ ................................ ......... 41 mile ................................ ................................ ................................ ............... 42 mile ................................ ................................ ................................ ............... 43 Outcome of OLS ................................ ................................ ............................... 43 Geographically Weighted Regression (GWR) ................................ ......................... 44 University of Florida ................................ ................................ .......................... 48 Hawtho rne Trail ................................ ................................ ................................ 50 Millhopper Road ................................ ................................ ............................... 52 SW Archer Road & SW 91 st Street ................................ ................................ ... 54 5 DISCUSSION AND CONCLUSION ................................ ................................ ........ 56 Discussion ................................ ................................ ................................ .............. 56 Limitations ................................ ................................ ................................ ............... 57 Conclusion ................................ ................................ ................................ .............. 58 APPENDI X A RESULTS OF ORDINARY LEAST SQUARES ................................ ....................... 59 B RESULTS OF GEOGRAPHICALLY WEIGHTED REGRESSION .......................... 68 C LIST OF BICYCLE FACILITIES IN ALACHUA COUNTY, FL ................................ 69 LIST OF REFERENCES ................................ ................................ ............................... 73 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 76

PAGE 7

7 LIST OF TABLES Table page 3 1 Attributes available in final dataset. ................................ ................................ .... 33 3 2 Variables used in regression analysis. ................................ ............................... 36 4 1 Results of ordinary least squares. ................................ ................................ ...... 39 A 1 Descriptive statistics for all single family properties within Alachua County. ...... 59 A 2 Descriptive statistics for all single family properties within 500 feet of a bicycle facility in Alachua County. ................................ ................................ ....... 59 A 3 Descriptive statistics for all single fa mily properties within 1,000 feet of a bicycle facility in Alachua County. ................................ ................................ ....... 60 A 4 Descriptive statistics for all single family properties within 2,000 feet of a bicycle facility in Alachua County. ................................ ................................ ....... 60 A 5 Descriptive statistics for all single family propert ies within mile of a bicycle facility in Alachua County. ................................ ................................ .................. 61 A 6 Descriptive statistics for all single family properties within mile of a bicycle facility in Alachua County. ................................ ................................ .................. 61 B 1 Descriptive statistics for all single family properties in Alachua County us ed in geographically weighted regression. ................................ ................................ .. 68 C 1 List of bicycle facilities in Alachua County, Florida. ................................ ............ 69

PAGE 8

8 LIST OF FIGURES Figure page 3 1 Map of study area and bicycle facilities. ................................ ............................. 30 4 1 Coefficients for distance from bike lane variable for all properties in GWR dataset. ................................ ................................ ................................ ............... 46 4 2 Coefficients for distance from bike lane variable for all properties within one mile of the University of Florida. ................................ ................................ ......... 49 4 3 Coefficients for distance from bike lane variable for all properties within one mile of the Hawthorne Trail. ................................ ................................ ................ 51 4 4 Coefficients for distance from bike lane variable for all properties within one mile of Millhopper Road. ................................ ................................ ..................... 53 4 5 Coefficients for distance from bike lane variable for all properties within one mile of the intersection of SW Archer Road and SW 91 st Street. ........................ 55 A 1 Ordinary least squares results for all single family properties in Alachua County. ................................ ................................ ................................ ............... 62 A 2 Ordinary least squares results for all single family properties within 500 feet of a bicycle facility. ................................ ................................ .............................. 63 A 3 Ordinary least squares results for all single family properties within 1,000 feet of a bicycle facility. ................................ ................................ .............................. 64 A 4 Ordinary least squares results for all single family properties within 2,000 feet of a bicycle facility. ................................ ................................ .............................. 65 A 5 Ordinary least squares results for all single family properties within mile of a bicycle facility. ................................ ................................ ................................ .. 66 A 6 Ordinary least squares results for all single family properties within mile of a bicycle facility. ................................ ................................ ................................ .. 67 B 1 Properties of geographically weighted regression variables. .............................. 68

PAGE 9

9 LIST OF ABBREVIATIONS CAMA C OMPUTER A SSISTED M ASS A PPRAISAL FGDL F LORIDA G EOGRAPHIC D ATA L IBRARY GIS G EOGRAPHIC I NFORMATION S YSTEMS GWR G EOGRAPHICALLY W EIGHTED R EGRESSION OLS O RDINARY L EAST S QUARES SW S OUTHWEST UF U NIVERSITY OF F LORIDA

PAGE 10

10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts in Urban and Regional Planning THE IMPACT OF BICYCLE FACILITIES ON SINGLE FAMILY RESIDENTIAL PROPERT Y VALUES: EVIDENCE FROM ALACHUA COUNTY, FLORIDA By Kevin Szatmary May 201 4 Chair: Ilir Bejleri Cochair: Abhinav Alakshendra Major: Urban and Regional Planning Bicycle lanes and trails are an important part of the transportation infrastructure of a city They provide recreational and health benefit s as well as an alternate means of transportation for commuters In some cities in the United States where bicycle culture is growing, the value of properties w ith close proximity to these amenities has increased. The objective of this study was to assess the impact that bicycle lanes and trails provide to single family residential property values. This was accomplished by using ordinary least squares regressi on to create a hedonic pricing method, followed by a locational study using geographically weighted regression. The study was performed in Alachua County, Florida, using geographic information systems (GIS ) The analysis showed that the impact of bicycle lanes and trails varies by location. The use of ordinary least squares does not take location into account, therefore geographically weighted regression was employed to consider spatial variation. It was found that i n some locations, proximity to a bicycle facility decreased the value of homes, possibly stemming from issues regarding decreased privacy or increased crime.

PAGE 11

11 In other locations, property values were increased at homes located near bicycle facilities, as a result of commuting ease or recreational benefit.

PAGE 12

12 CHAPTER 1 INTRODUCTION Problem The city of Gainesville, F lorida, touts a ranking of 37 th on the list of bike friendly cities in the United States (UF Office of Sustainability, 2013) Gainesville also ranks seventh for metro areas in which commuters bike to work In 2009, 3.3% of workers aged 16 or older used their bicycle as their primary source of transportation to work (McKenzie & Rapino, September 2011) Academic research supports that bicycle paths, trails, and greenways can provide increa sed recreation, more flexible transportation, lower crime, and as the key point of this research increased property value. How are residential property values and local economy affected by the p resence of bicycle facilities? This thesis seeks to examin e residential properties and their values near bicycle lanes and trails to determine whether there is a positive impact, negative impact, or no impact for access to this amenity. Research Questions and Objectives The purpose of the research is to determine whether or not the presence of a bicycle l ane or trail has an impact on the just (market) value of a home. The research is based on the following questions: 1. an off street facilit y, such as the Gainesville Hawthorne Trail) have an impact on its just (market) value? 2. If so, what is the estimated impact of the facility on the just value of said properties? Is it positive or negative? 3. How far from a bicycle lane or trail is the impact significant? (For example, in some studies (Campbell & Munroe, 2007; Racca & Dhanju, 2006)

PAGE 13

13 value will increase by a certain dollar amount for every foot closer to the amenity, but is insignificant past a c ertain distance.) To study the impacts, I used geographic information systems (GIS) in combination wit h ordinary least squares regression to create a hedonic pricing method and geographically weighted regression using Alachua County as a case study. The hedonic pricing model is a method which isolates characteristics of housing which internally impact its price (such as number of bedrooms, number of bathroom, square footage, etc.) or externally (such as accessib ility to parks, schools, etc.) (Lindsey, Man, Payton, & Dickson, 2004) Hedonic pricing models can be used to infer a premium that a homebuyer would be willing to pay to purchase a home near an amenity in this case, a bicycle lane or trail or away from a disamenity (Racca & Dhanju, 2006) such as pollution or an industrial park. The hedonic pricing method will reveal how the dependent variable just v alue, is related to the independent variables such as the number of bedrooms, square footage, or distance from bicycle facility) Independent variables which are found to be statistically relevant will then be used to determine the impact on housing prices. Geographically weighted regression will be used to determine locations at which the impact is significant. Method The value of a home can be attributed to many different factors. The use of a hedonic pricing method will allow one to estimate how different factors contribute to the overall value of the property. The analysis was performed using ordinary least squares (OLS) at various specified distances from the bicycle lanes and tr ails in Alachua County. The rel ationship between the independent variable (just value) and several independent

PAGE 14

14 locational structural, and neighborhood variables was estimated. Geographically weighted regression (GWR) analysis was then used, making use of only the variables from OLS which were determined to be statistically significant. The GWR analysis seeks to improve upon the results by taking into account spatially varying relationship Utilizing GWR allows for analysis at specific locations, as the imp acts may vary between locations.

PAGE 15

15 CHAPTER 2 LITERATURE REVIEW Bicycle Lane and Trail Impacts Besides being a source of transportation, bicycle lanes and trails can have a both children and adults, and help to influence good health and improved fitness by way of exercise. This can, in turn, also lead to lower medical bills (Karadeniz, 2008) In addition to health benefits, bicycle facilities may provide environmental benefits by decreasing the number of trips made by car. Trails also can provide aesthetic benefits as they may be incorporated into a larger greenspace, such as a park (Lindsey, Wilson, Yang, & Alexa, 2008) The Gainesville Hawthorne Trail, somet imes referred to as simply the Hawthorne Trail, is one such example. Rail trails in particular may also help to preserve a historical sense. Rail trails are multiuse trails built upon the former rights of way of railroad lines. As some railroad lines becam e unprofitable to operate throughout the 1970 s, railroad companies sought to abandon the lines in order to reduce costs. This was enabled in 1983 with National Trails System Act. This act allowed railroad companies to donate, sell, or lease the right of w ay for use as a trail to various qualified public or private entities, such as the Rails to Trails Conservancy. (Rails to Trails Conservancy, 2005) The Hawthorne Trail is an example of a rail trail, conve rted from the former corridor of the Seaboard Coast Railroad (Florida Division of Recreation and Parks, n.d.) Gainesville historically was a central meeting point of many railroads; rail trails have been con structed along some of these now abandoned rail lines. These routes may also provide access to historic buildings, such as the Gainesville Depot ( which will

PAGE 16

16 be incorporated into Depot Park ) and to bridges, such as the 13th Street pedestrian overpass bridg e, a former railroad bridge which is now part of the Depot Avenue Rail Trail. (Gainesville Community Redevelopment Agency, 2013) This bridge provides easy access to Shands Hospital from homes located acro ss SW 13th Street. Bicycle lanes and trails may also be a source of economic benefits to a community. The Gainesville Hawthorne State Trail, a 16 mile paved rail trail which connects Gainesville to Hawthorne to the east, passes through the Paynes Prairie P reserve State Park (Florida Division of Recreation and Parks, n.d.) Paynes Prairie attracted 176,134 visitors in 2011 2012 to Alachua County (Florida Department of Environmental Protection, 2012) While the park itself does not charge admission, the influx of visitor to the county has economic potential. Despite the positive ramifications, the expansion of bicycle networks is not without o pponents. Opposition is typically brought by property owners who are adjacent to bicycle facilities. Some property owners believe that off street bicycle trails in particular can decrease property value by increasing the potential for crime and reducing th eir privacy (Hawthorne, Krygier, & Kwan, 2008) Others believe that the addition of bicycle lanes within a city is a luxury, leaving transportation planners to justify their construction (Krizek, 2007) The construction of trails in Gainesville has been subject to opposition from residents In September of 2012, the Alachua County Commission decided in a 3 2 vote tha t the Archer Braid Trail, a multi use trail, would not extend through the Haile Plantation development located west of Gainesville (McClenny, 2012) The 2.2 mile section of the trail was to traverse SW 91st Street and SW 46th B oulevard as p art of the Archer Braid Trail. The trail, when completed, would

PAGE 17

17 have linked the Kanapaha Park near Haile Plantation to the city of Archer in southwest part of Alachua County (Florida Department of Transportation, 2012) For now, the Archer Braid Trail runs along the south side of Archer Road, but terminates to the east at SW 91st Street, where it connects with an existing off street facility that merges onto an on street bike lane near Tower R oad. Bicycle Facilities in Alachua County There is a total of 1,185 miles of roadways and trails in Alachua County. The on road portion of the network is 823 miles, while 229 miles of the road portion includes paved shoulders or striped bike lanes. Of the entire 1,185 mile network, 362 miles are trails, however only 58 miles are existing. (Sprinkle Consulting, Inc., 2001) There are several classifications of bicycle facilities in Alachua County. They are a s follows: Alternate street oriented route Dedicated street oriented route with bike lane Dedicated street oriented route with paved shoulder Future off road trail No street facility without sidewalk No street facility with sidewalk Off street facility Sha red street oriented route These types of facilities differ based on their relative location in the right of way or how the facilities are marked on the ground. (City of Gainesville, 2011) The least accessible facilities are those which include no lane or division from the road. Both types of lanes indicated with no street facility fit this description. There is no curb lane, no bicycle lane, and may or may not have a sidewalk. This is the l east desirable type of facility. (City of Gainesville, 2011)

PAGE 18

18 Moderately accessible facilities are those which are located on roads that are not busy or have lower speed limits, such as through residential neighborhoods. These facilities may have a wide parking lane that may or may not be in use. Alternate street oriented routes and shared street oriented routes are of this type. (City of Gainesville, 2011) The most desirable routes are those which include facilities specifically intended for use by cyclists. Off street facilities are paved routes that are separated from the road, such as rail trails Dedicated street oriented routes are those which include a lane adjacent to the road which is marked for use by cyclists with pavements markings or signage. (City of Gainesville, 2011) Valuing Residential Properties In order to quantify the impact of an independent variable has on the value of residential properties, one must understand how the variables can impact the housing values. The value of residential properties is a function of several variables, and can be estimated using regression models su ch as the hedonic pricing model. Variables Estimation of the impact of bicycle lanes and trails on property values first requires an understanding of the variables which impact sale values. There are many diff erent variables which contribute to the sale price of residential properties. The variables used by studies vary, but there are many concept s that apply to all of them. According to Freeman (2003) the independent varia bles fall into three different categories: environmental, structural, and neighborhood. Environmental variables include those which are specific to a location, such as air quality or water quality. Structural variables include those which define certain ph ysical aspects of the house

PAGE 19

19 (such as number of bedrooms or number of bathrooms), characteristics of the lot (such as acreage), presence of a particular amenity (such as a fireplace or garage), and miscellaneous property characteristics (such as the age of assessed value). Neighborhood variables include items such as quality of nearby schools, racial makeup of the neighborhood, or the household incomes of the neighborhood. Nicholls & Compton (2005) expand the above categories into a six category model. In addition to environmental, structural, and neighborhood variables, the property value is said to be impacted by community variables (such as school district) and s The property value is also said to be impacted by locational variables. Locational variables are those whose proximity and accessibility may or may not be desirable, and can inc lude such disamenities (such as landfills or power lines) or amenities such as churches or parks. Lastly, the property value can be impacts by time related variables, and typically involve attributes such as available interest rates or the number of days a property is on the housing market. The estimated impact of a bicycle lane or trail on the resale price of a house can be determined by using proximity to the amenity as an explanatory variable. Explanatory variables, or independent variables, are those wh ich contribute to the value of the dependent variable. In a simple linear regression, the regression equation contains just one explanatory variable. In a multiple linear regression, there are multiple explanatory variables. Using the hedonic pricing metho d, the dependent variable is the

PAGE 20

20 value of the home, and the structural, neighborhood, and environmental characteristics make up the explanatory variables. (Karadeniz, 2008) Hedonic Pricing Model The hedonic pricing model, as represented by Freeman (2003) can be represented using the following equation: Rh j = Rh (S j N j Q j ) (2 1) Where: Rh j = the price of the jth residential location S j = structural characteri stics N j = neighborhood characteristics Q j = environmental characteristics. Nicholls and Crompton (2005) provide a linearized model which includes the expanded six categories: P 1 + S X S + N X N + C X C + L X L + E X E + T X T + (2 2) Where: P = observed property prices X S = structural attributes X N = neighborhood attributes X C = community attributes X L = locational attributes X E = environmental attributes X T = time attributes 1 = constant term x = estimates of relevant attributes implicit marginal prices after differentiation = stochastic disturbance term. In both cases, the sale value of the residential property is represented by the dependent variable on the left side of the equation and the explanatory variables, represented by various attributes, is on the right side of the equation.

PAGE 21

21 Geographically Weighted Regression Geographically weighted regression is a form of linear regression which is used to model relationships which may di ffer spatially. This tool creates a separate regression model for each individual record in the provided dataset. The principle idea of GWR is that there is variation across a given spatial area which ordinarily least squares does not take into account. Th e availability of coefficients for individual records addresses the issue that the average coefficient over a large geographic area may not be representative of the sample dataset due to said variation across the landscape (Legg & Bowe, 2009) Study Limitations These models are not without some limitations. One common problem among regression models is that the explanatory variables may not be independent of one another. For example, a home which has more bedrooms may also have more square footage. This phenomenon, called collinearity, or multicollinearity for three or more variables (Freeman, 2003), can cause the parameters of the hedonic pricing model to be exceptionally large or to have the inco rrect sign, which may cause difficulty in determining which variables impact sale prices (Karadeniz, 2008) Some variables may not be statistically significant to the study. To determine which variables a re statistically significant, a backwards elimination technique can be used. This method handles issues stemming from multicollinearity. The backwards elimination technique begins by placing all of the independent variables into a multiple linear regressio n. The t statistic value can then be used to determine the relative strength of prediction of each of the independent variables. The independent variable with the lowest significance is removed, and the process is repeated with additional

PAGE 22

22 variables being r emoved, one at a time, until all the t statistics of the model are significant. The remaining variables can then be incorporated into the hedonic price model. (Karadeniz, 2008) Case Studies A number of st udies have been completed for various municipalities to study the impact of bicycle lanes and bicycle trails on property values. Some have concluded that there is a positive impact, which others have concluded that there is a negative impact or no signific ant impact at all. Several studies have used opinion surveys to determine if however there are some which have studied the impact using the hedonic pricing technique. The first subsection discusses perceptions that homeowners and homebuyers have regarding their proximity to trails. These opinion based studies largely use surveys or questionnaires to determine if property owners felt that living near a bicycle lane or bi cycle trail has positive impacts, such as increased aesthetics or increased property values, or negative impacts, such as increased crime or decreased privacy. The second subsection discusses studies which use the hedonic pricing method to determine which attributes contribute to the resale value of a home. These attributes are largely the independent variables, which are discussed in the third subsection. Opinion Based Studies A study by Greer (2000) of recreational trai ls and their effect on property values and public safety was conducted in Omaha, Nebraska. This study surveyed 149 households along three separate trail segments from the Omaha trail network. Questions related to the economic impact of the trails found the following statistics: 81%

PAGE 23

23 effect or no effect on the ease of sale of their homes. The clear majority of residents (63.8%) who bought their homes after the construction of the trails reported that the trail had positively influenced their purchase decision, and residents living along the trails appear to perceive there to be a positive relationship between the trails and quality of life (75%). Another study of home sales near two rail trails in Massachusetts found data related to the average time it took to sell homes in the area surrounding the trails. While the average sale price of the homes was similar whether or not they were located near a trail, the difference of the average time on the market was significant. Specifically, homes near rail trails sold in an average of 29.3 days as compared to 50.4 days for homes not near rail trails. (Penna, 2006) While this does not reflect on the value of the homes, it does suggest that properties located near rail trails may be more desirable than those that are not. Hedonic Pricing Technique Studies A study was performed by the Unive rsity of Delaware in 2006 for the Delaware Center for Transportation and the State of Delaware Department of Transportation. In the paper, researchers sought to find how proximity to bicycle lanes and trails impacted residential property values in New Cast le County, Delaware. GIS data was used in combination with a hedonic pricing model to calculate these values. In this study, a positive significance of approximately $8,800 was found for properties located within 50 meters of a trail. The study concludes t hat the effect of the nearby bike paths, while small, is positive. There is approximately a 4 percent increase in the value of homes located near trails. (Racca & Dhanju, 2006)

PAGE 24

24 Nicholls and Crompton (2005) used the hedonic pricing method to study the impact of greenways in three different neighborhoods near the Barton Creek Greenbelt and Wilderness Park in Austin, Texas. The results were fo und to be insignificant in one neighborhood, Lost Creek, however the impact was substantial in the other two. In the Barton neighborhood, it was found that 12.20% ($44,332) of the resale value of a home was attributed to being adjacent to the greenway. In the Travis neighborhood, it was found that 5.70% ($14,774) of the resale value was attributed to greenway proximity. In all cases, only properties which were directly adjacent to the greenbelt showed a significant impact; those which had a view but were no t adjacent had no significant positive impact on the sales price. (Nicholls & Crompton, 2005) Lindsey, Man, Payton, and Dickson (2004) also used the hedonic method to study the Monon Trail, a rail trail which extends north out of Indianapolis. Results were positive and significant; homes within one half mile of the trail were found to have 14% ($13,056) of their sale price attributed to proximity to this flagship trail. (Lindsey, Man, Payton, & Dickson, 2004) A study completed by Campbell and Munroe (2007) showed similar results, but included results for multifamily and commercial properties in addition to single family homes located in Raleigh, North Carolina. In all cases, the value of the property increased as one moved closer to the trail. In single family homes, prope rty values decreased 0.0312% for every 1% increase in distance from the greenway. In multifamily homes, the impact was just 0.0013%, and in commercial properties, the impact was 0.0172%. The study found that the highest increases in single family values we re within 1,000 feet of the trail, however the impact was significant enough that homes within

PAGE 25

25 5,000 feet (approximately one mile) had a higher value of $3,245 as compared to similar homes located farther away. For multifamily homes, the increased value wa s found to be $231. For commercial properties, the increased value was found to be $4,472. (Campbell & Munroe, 2007) The impact of bicycle trails may not always be a positive one. The creation of bicycle trails in the United States has been controversial. Hawthorne, Krygier, & Kwan (2008) and Krizek (2007) discuss that trail opponents often believe trails will decrease their property values due to increased crime and negative effects on their privacy. This same literature also suggests that these concerns may be based on mistaken perceptions, which are the real reason behind the hostility and conflict surrounding the development of trails (Hawthorne, Krygier, & Kwan, 2008) Some consider spending money on bicycle facilities to be a luxury (Krizek, 2007) given that such a low percentage of people actually use them. Just 0.8% of trips in North America are completed by bicycle (Krizek, 2006) (2006) the author uses the hedonic pricing method to study homes in the Minneapolis instance used a different regression to compare the inner city versus the suburbs. In his research, the author hypothesized that greenways in suburbs were valued differently than those in the city because suburbanites were less likely to use bicycles as a source of transportation. The study confirmed this: properties within the inner city which are located within 400 meters of bicycle lane or within 400 meters of an off street trail sold for $2,272 more and $510 less, respectively, suggesting that urban homeowners highly value access to bicycle lanes, but no t access to trails. In the suburbs, homes within 400

PAGE 26

26 meters of a bicycle lane sold for $1,059 more, indicating that suburbanites do place value on bicycle lanes but less so than their urban counterparts. However, the study also found the suburban homes wi th 400 meters of an off street trail sold for $240 more, suggesting that suburban homeowners value bicycle trails slightly. (Krizek, 2006) A study by Netusil (2005) was performed in Portland, Oregon largely regarded as a bicycle friendly community which used the hedonic pricing method to examine the impact of environmental zoning on single family housing prices. The study found that a trail within 200 feet of a s ingle family home would decrease its sale price by approximately 5.54%. The study suggests that this impact may be caused by noise and congestion stemming from proximity to a trail. The study goes on to state that the impact of the trail itself may not nec essarily be the cause, as the negative impact may be caused by the proximity to industrial areas located near railroad rights of way which were later converted to trails. Another study performed by Correll, Lill ydahl, and Singell (1978) makes use of a similar model. This study involved three different neighborhoods within the city of Boulder, Colorado. Two of the three neighborhoods were found to have decreasing resale values in homes as one moves away from the trail. The third was found to have increasing values, indicating the living near the trail decrease property values. The article makes an interesting note that increased property values gene rate more tax income for the local government; as such, one can infer that a higher number of such (Correll, Lillydahl, & Singell, 1978)

PAGE 27

27 Variables Used The hedonic studies did not explain why the variables used were chosen. Of the hedonic studies reviewed, the most used variables were the acreage or lot size, the square footage of the home, and the age of the home at the most recent sale. The lot size was used in six studies, and found to be significant in each of them. The square footage was also used in six studies, although Campbell & Munroe (2007) used this variable only in evaluating commercial properties, not on single family or multifamily homes. The age of the sale was used in six studies as well. In the seven studies, the structural variables used to study single family properties a re as follows: acreage/lot size (6), house size/square footage (6), age at sale (6), number of bathrooms (5), numbers of bedrooms (4), number of garage bays (4), number of stories (3), number of fireplaces (3), number of rooms (2), building value (2), year s since last sale (1), air conditioning (1), porch (1), and land value (1). The neighborhood variables used also vary by study. Each study used proximity to a trail, however some used an increasing distance to study the impact (Campbell & Munroe, 2007; Correll, Lillydahl, & Singell, 1978; Krizek, 2006; Netusil, 2005) in order to determine the change as distance increases from the trail. The other three studies used direct adjacency to the trail as a dummy variable (Lindsey, Man, Payton, & Dickson, 2004; Nicholls & Crompton, 2005; Racca & Dhanju, 2006) The othe r neighborhood variables used are as follows: racial makeup of neighborhood (3), school quality (2), and household incomes (2).

PAGE 28

28 CHAPTER 3 METHODOLOGY The goal of this case study is to determine whether or not there is an increased value of a residential property within a reasonable distance of a bicycle facility. The hypothesis is that there will be a small negative correlation between the property value and the distance to a bicycle facility; that is, as the distance increases from a bicycle facility, the just value of a home will decrease. In other words, properties nearest the facilities will be valued more due to their proximity. The relationship will be impacted by other structural and neighborhood variables, as pr eviously discussed. T wo popular methods for valuing this premium include the use of opinion based studies and hedonic pricing studies. While there are many opinions to assess value, these studies only show perception. They are useful for indicating qualities of a neighborhood which may impact value, and thus suggestions for variables in analysis, however the results are qualitati ve in nature. The goal of the study is to quantify the value by studying market values as determined by the property appraiser ; as such, the hedonic pricing method is used for the purpose of this analysis. The backwards elimination technique was first use d to examine the significance of the available variables. Ordinary least squares was then used to determine the impact of each individual variable. Case Study Alachua County is located in North Central Florida. It has an area of 969.12 square miles. As of the 2010 Census, the population was 247,336 with 255.22 persons per square mile. There were 96,544 households and 113,194 housing units. (U. S. Census Bureau, 2014) Gainesville is home to the Unive rsity of Florida, the largest

PAGE 29

29 university in the state of Florida, as well as Santa Fe College. These two academic institutions both have a large student population. The University of Florida had 50,086 students (UF Office of Institu tional Planning and Research, 2013) and Santa Fe College had 6,659 students (Santa Fe Office of Planning and Institutional Effectiveness, 2013) in 2013. The University of Florida also employs more than 41,000 people (University of Florida, 2013) than 12,000 (Un iversity of Florida, 2014) The University of Florida generated a substantial amount of concentrated bicycle traffic. Approximately 12% of the students, faculty, and staff commute to the university campus by way of bicycle. This number exceeds all oth er universities in the State University System in Florida combined. (Sprinkle Consulting, Inc., 2001) Figure 3 1 shows the study area and bicycle facilities for study.

PAGE 30

30 Figure 3 1. Map of study area a nd bicycle facilities

PAGE 31

31 Data Collection There are a few important datasets needed for this study: a list of all of the single family residential parcels in Alachua County, and a list of all of the bicycle facilities within the county as well as data pertaining to neighborhood attributes Parcel data was obtained from the Florida Geographic Data Library (FGDL) Property data was obtained computer assisted mass ap praisal (CAMA) database, included several attributes pertaining the buildings and parcels located within the county. The bicycle facilities data was obtained from the City of Gainesville Public Works Department. This data included a list of every bicycle f acility in the county, both existing and proposed. Data pertaining to neighborhood attributes was obtained from the 2010 United States Census. In order to simplify the regression, all datasets would require editing. Parcels The parcels data, provided by t he Florida Department of Revenue and obtained from FGDL was use d for ease in determining which properties were zoned for single family residential use. This dataset, PARCELS_10, inclu ded the land use of each parcel located in the county Because the study is focused only on single family property values, o land uses were removed. Additionally, only parcels which were greater than 0.10 acres were used. CAMA Tables The CAMA datase t included an excess of information. Much of it was unnecessary for this study. From the CAMA dataset, only the GISbldgpub (building information) and GISsale (sale information) tables were used. These two tables include

PAGE 32

32 all of the variables used in the pre viously mentioned hedonic studies such as those found in Lindsey, Man, Payton & Dickson (2004) study of Minne a polis St. Paul, Minnesota (2006) Data pertaining to the structure and land acreage, square footage, age of home at most recent sale, number of bedrooms, number of bathrooms, and number of stores were extracted from the GISbldgpub table. Data pertaining to the just value was e xtracte d from the GISsale table Only properties which included one or more bedroom, one or more bathroom, one or more story, and a properly formatted year built were used. Bicycle Facilities The bicycle facilities data is necessary to determin e which properties lie within a reasonable distance of an appropriate facility. While the table includes data on every bicycle lane or trail in Alachua County, some of this data could not be used. Future off road trails are not feasible for study as they have not yet been co nstructed and the study is intended to assess value based on existing trails Additionally, routes with the do not have facilities intended only for use by cyclists. Only off street facilities (trails) and dedicated street oriented routes were used (those with paved curb lanes or marked bicycle lanes) ; all other types were removed. Census Data D ata from 2010 Uni ted States Census was obtained for neighborhood variables. The Census provides information about the population, race, and income on a census block, census block group, census tract, city, county, and state level. Percent minority and population density ar e available at a census block level, however the smallest

PAGE 33

33 denomination that income data is available at is the block group level. This data was joined using ArcMap to estimate the population density and percent minority in neighborhoods. Combined Dataset The edited parcel dataset and CAMA tables were combined using ArcMap using the parcel number to identify unique records. The Near analysis tool of ArcMap was then used to determine the distance from each parcel to its nearest off street or dedicated stree t oriented facility. The combined dataset contains 50,309 single family properties. Table 3 1 lists the attributes available for use in the regression analysis. Table 3 1. Attributes available in final dataset. Attribute Description ParcelID Parcel number as assigned by Alachua County Property Appraiser Acres Area of parcel in acres LndSqFoot Area of parcel in square feet Amount Amount of most recent sale YearSale Year of most recent sale Beds Number of bedrooms Baths Number of bathrooms ActYr Year structure was constructed AC _code Presence (1) or absence (0) of air conditioning Heat _code Presence (1) or absence (0) of heating Stories Number of stories in structure TotArea Total floor area of structure Near_Dist Distance to nearest applicable bicycle facility in meters FtFromBL Distance to nearest applicable bicycle facility in feet Data Limitations price as the dependent variable would present significant challenges due to questionable data quality, such as home sales with incorrectly formatted dates, or very large or ve ry small sale prices. Additionally, housing prices may fluctuate greatly over

PAGE 34

34 time because of economic externalities. For these reasons, the just (market) value of the property was used. The just value is calculated by the Alachua County Property Appraiser using a computer assisted mass appraisal (CAMA) method. The CAMA reproducing property, and the ability of the property to earn income (Alachua C ounty Property Appraiser, n.d.) Regression Analysis T o answer the research questions, t wo different types of regression analyses were utilized: ordinary least squares (OLS) and geographically weighted regression (GWR). The first, OLS, was utilized using the methods set forth in the literature. The basis for analysis is derived from the following model: V i = f ( L i S i N i ) (3 1) Where: V i = the just value of the ith residential location L i = locational variables S i = structural variables N i = neighborhood variables. The dependent variable in this model is the just value of the property. The locational, structural, and neighborhood variables are all independent variables. The model can be expanded into the following equation: V i = X i + ActYrBlt i + Acres i + Beds i + Baths i +Stories i + TotArea i + AC_code i + FtFromBL i + DenPop2010 i + Pct_Mnrty i (3 2) Where: V i = the just value of the ith residential location ActyrBlt i = Actual year built Acres i = Acreage of parcel

PAGE 35

35 Beds i = Number of bedrooms in property Baths i = Number of bathrooms in property Stories i = Number of stories in property TotArea i = Total covered area of property AC_code i = Presence or absence of air conditioning FtFromBL i = Free to nearest bicycle facility DenPop2010 i = Population density (persons/acre) Pct_Mnrty i = Just (market) value The locational variable in the model is the distance of a property from the bicycle facility (FtFromBL) After running ordinary least squares, th e expected coefficient of this attribute is negative, a bicycle facility increases. The distance was calculated using the Near function in ArcMap, which calculates the distance from each parcel to the nearest fe ature (in this case, bicycle facilities). The structural variables in the model are the parcel size, total area, number of stories, number of bedrooms, number of bathrooms, and presence of air conditioning. The expected value of the parcel size, total area number of bedrooms, and number of bathrooms is positive, indicating an increased property value as the number of each increases. The expected value of the presence of air conditioning and the presence of heating system is also positive, indicating an inc reased property value for each of these systems being present in the property. The expected value of the number of stories is negative, indicating a decreased property value as the height of the structure increases. The neighborhood variables in the model are the percent minority and population density. The expected value of percent minority is negative, indicating a decreased property value as the percent minority increases. The expected value of population density is positive, indicating an increased prop erty value as population density increases. This stems from the desire to be closer to the city where the population

PAGE 36

36 density is higher. The average population density of the city of Gainesville is 1993.17 persons per square mile, as compared to 255.22 pers ons per square mile for the whole of Alachua County. Table 3 2 shows the variables to be used in the regression analysis and their expected values. Table 3 2. Variables used in regression analysis Variable Name Description Type Expected Value ActYrBlt Actual year built Independent + Acres Acreage of parcel Independent + Bed s Number of bedrooms in property Independent + Baths Number of bathrooms in property Independent + Stories Number of stories in property Independent + TotArea Total covered area of property Independent + AC_code Presence or absence of air conditioning Independent + FtFromBL Feet to near est bicycle facility Independent DenPop2010 Population density (persons/acre) Independent + Pct_Mnrty Percent minority Independent JV Just (market) value Dependent n/a The study seeks to address changing value as distance increases from bicycle facilities. Properties were chosen at different distances from the facilities in Alachua County: 500 feet, 1000 feet, 2000 feet, mile, and mile. Additionally, all properties within the county were used to determine overall value. Using ordinary least squares to study spatial relationship s can create an issue because of spatial heterogeneity. OLS assigns a coefficient for each variable; the c oefficient is in essence, an average of the coefficients for each record in the data. Geographically weighted regression was used to account for this problem. GWR calculates a coefficient for each record in the data, and allows for the study to be

PAGE 37

37 analyze d at specific locations. In other words, the value of a property may be impacted in one part of the county different than in another part of the county. Several different locations were chosen for additional analysis: parcels near the Hawthorne Trail, par cels near Millhopper Road, and parcels near the University of Florida, parcels near the intersection of SW Archer Road and SW 91 st Street. Since they are both heavily travelled for recreational use, th e expected value for parcels near the Hawthorne Trail a nd near Millhopper Road is negative, indicating decreasing value as one moves away from these facilities. The expected value for parcels near the University of Florida is negative, indicating decreasing value as one moves away from the university. The valu e is expected to be negative because of the concentration of bicycle lanes near the university, facilitating ease in commuting by bicycle for students. The expected value for parcels near the intersection of SW Archer Road and SW 91 st Street is positive, indicating an increasing value as one moves away from bicycle facilities. This value is expected to be positive due to strong opposition for extension and expansion of the bicycle trails in this area.

PAGE 38

38 CHAPTER 4 FINDINGS AND RESULTS Ordinary Least Squares (OLS) Ordinary least squares was employed first, utilizing just value as the dependent variable, and ten explanatory variables. Regressions were run for the entire county, and at distances of a 50 0 feet, 1,000 feet, 2,000 feet, mile, and mile away from a bicycle facility. Table 4 1 shows the results for each regression.

PAGE 39

39 Table 4 1. Results of ordinary least squares. Model 1 Alachua County Model 2 500 feet Model 3 1,000 feet Model 4 2,000 feet Model 5 1/4 mile Model 6 1/2 mile Variable Name Coefficient (T Statistic) Coefficient (T Statistic) Coefficient (T Statistic) Coefficient (T Statistic) Coefficient (T Statistic) Coefficient (T Statistic) ActYrBlt 401.19 333.99 357.01 353.14 355.58 361.82 ( 41.39) (14.50) (23.10) (30.48) (26.10) (33.09) Acres 2400.77 2965.07 2857.54 2903.44 2940.26 2799.54 (3 9.14) (15.74) (18.19) (24.01) (19.89) (24.41) Beds 3682.95 964.51 1990.16 3534.26 2825.55 3756.01 ( 10.00) (1.11) ( 3.32) ( 7.97) ( 5.36) ( 9.10) Baths 17672.42 17150.00 764.39 14932.34 14259.55 15513.82 (472.56) (15.55) (18.06) (26.28) (21.21) (29.46) Sto ries 565.91 4057.42 3200.62 4031.34 3677.31 2084.95 (483.01) (3.18) (3.67) (6.48) (4.90) (3.66) Tot Area 65.82 53.80 61.40 62.48 61.91 64.25 (219.34) (73.49) 119.62 (164.51) (136.84) (183.86) AC_ code 5666.05 239.80 1034.10 2568.30 2496.98 3777.14 ( 6.59) ( 0.14) ( 0.84) ( 2.70) ( 2.25) ( 4.17) FtF romBL 0.15 4.45 4.20 1.62 3.66 1.30 (2.15) (1.30) (3.58) (3.56) (4.65) (3.93) Den Pop2010 1293.41 1921.87 1653.47 1535.49 1586.09 1310.00 (2 1.74) (15.51) (18.71) (22.56) (20.29) (21.46) Pct_Mnrty 154.41 198.11 167.39 185.23 179.96 168.54 ( 18.78) ( 11.33) ( 13.75) ( 20.25) ( 16.66) ( 19.68) R Squared 0.80 0.72 0.76 0.78 0.76 0.79 n 50,309 8,905 17,354 31,061 22,698 36,729 *An asterisk next to a number indicates a statistically significant p value (p < 0.05).

PAGE 40

40 Alachua County Using all of Alachua County, excluding the properties which lack proper attributes (regarding the number of stories, number of bedrooms, and number of bathrooms), resulted in a dataset containing 50,309 properties. The average just value for these properties was $147,277.34, and the average distance from a bicycle faci lity was 2,105.14 feet. After running ordinary least squares, a multiple R squared value of 0.80 was returned, indicating that 80% of the just value of the included properties could be explained by the variables used. All variables, with the exception of the number of stories, we re found to be statistically significant at the 0.05 probability level. Based on the output, value of properties increases by $0.15 per foot as one moves away from a bicycle facility. The presence of air conditioning yielded a negative coefficient, indicat ing a lower value for properties with air conditioning, a value which is contrary to the expected value. 500 feet Next, regressions were run at various distances. Selecting only properties within 500 feet of a bicycle facility resulted in a dataset with 8,905 properties. These properties had an average just value of $ 124,920.48 with an average distance from a bicycle facility of 204.9 4 feet The OLS results for the 500 foot data set yielded a multiple R squared value of 0.72, indicating that 72% of the just value of the included properties could be explained by the variables used. The number of bedrooms, presence of air conditioning, and feet away from a bike lane were all found to not be statistically significant below the 0.05

PAGE 41

41 level. The t statistic for total area, a value of 73.49, indicates that this variable has the greatest influence. 1 000 feet A distance of 1,000 feet was selected next to include additional nearby properties which may impacted by their location relative to a bicycle facility. In the 1,000 foot dataset, there were a total of 17,534 properties. These properties had an average just value of $128,238.63. The average distance to the nearest bicycle facility was 424.89 feet After running ordinary least squares on the 1,000 foot dataset, all variables were found to be statistically significant at the 0.05 level, with the exception of the presence of air conditioning. The multiple R squared value for this dataset was 0.76, indi cating that 76% of the just value could be attributed to these variables. Furthermore, it was found that the coefficient for the number of bedrooms was 1990.16, indicating that was the number of bedrooms increases, the just value of the home decreases. Th is is contrary to the expected value, suggesting that there may be an issue with this model. A t statistic of 119.62 was found for the total area, once again indicating that this was the more influential variable. The coefficient for distance from bike lan e was found to be positive and significant; the value of a property increases by $4.20 for every foot one moves away from the facility. 2 000 feet A dataset for properties within 2,000 feet of a bicycle facility was selected next. Within this dataset, there was 31,061 single family properties, more than half of those within the entire county. The average just value for these properties was $ 132 905.48 with an average distance of 787.40 feet to the nearest bicycle facility.

PAGE 42

42 Once again, the results of ordinary least squares were questionable. All variables were found to be statistically significant, and the R squared value was found to be 0.78, signifying that 78% of the just value could be explained by these variables. Howeve r, despite being statistically significant, the coefficient for the number of bedrooms was negative a value of 3,534.26 suggesting that the just value of homes within this dataset decreases as the number of bedrooms increases. Additionally, the coeffi cient for the presence of air conditioning was negative, suggesting that homes with air conditioning are valued less than those without. Both of these statistics do not conform to expected values. The coefficient for distance from bike lane was found to be positive and significant; the value of a property increases by $1.62 for every foot one moves away from the facility. Once again, the t statistic for the total area indicated that this was the most influential variable. mile A distance of mile was selected next, to attempt to explain the difference between the results of the 1,000 foot and 2,000 foot datasets. Within mile of a bicycle facility, 22,698 properties are located. These properties have an average just value of $ 130 149.09 and are located an average distance of 556.52 feet from a bike lane. The ordinary least squares results for the mile dataset were found to be similar to those of the 2,000 foot dataset. A multiple R squared value of 0.76 was found, indicating that 76% of the just value could be explained by these variables. All ten variab les were found to be statistically significant. The large negative coefficient for the number of bedrooms raised concern once again, as did the large negative coefficient for the presence of a bicycle facility. OLS yielded a coefficient of 3.66 for the dis tance from

PAGE 43

43 bike lane variable; this indicated that properties increase in value by $3.66 for every foot one moves away from the bicycle lane. mile Lastly, a distance of mile was used. This value was chosen because it was the furthest distance at which other studies were able to produce consistent results with negative coefficient for the distance from bike lane variable Within the mile dataset are 36,729 single family properties. These properties have an average just value of $136,191.68. The average distance from a bicycle facility for these propertie s is 974.24 feet. The OLS results for the mile dataset were similar to the mile and 2,000 foot datasets as well as the whole Alachua County dataset As before, all variables were found to be statistically significant. The variable with the most influence was the total area once again, with a t statistic of 183.36 and a coefficient of 64.25 This result is similar to the result found in the dataset for the whole county. The coefficient of 1.30 for the distance from bike lane suggests once again that the value of properties increases as one moves away from the facility. In this case, that value is $1.30 per foot. Outcome of OLS The results of these datasets may not be reliable for a variety of reasons. The expected value of the coefficient for some variable is contrary to what one would rationally expect. For example, in Florida, the presence of an air conditioning system would be expected increase property v alue rather than decrease it. The same could be said for the number of bedrooms. There is additional concern regarding the number of bedrooms, as its value is related to the total area; that is, properties with more bedrooms tend to have larger square foot age. Given that no model provided by the

PAGE 44

44 literature indicated a negative relationship, the reliability of these models is questionable. While the coefficient for distance from bike lane was positive and significant in most results, the reliability of these results is questionable as well. It is hypothesized that the value may vary, depending upon the location in the county. For example, the value of a property near the University of Florida may be found to be influence d greatly by its proximity to a bicycle facility. Given that there is insufficient parking on and around the campus, the use of alternative transportation such as bicycles can be expected for commuters. Another example where location may factor in is the residential areas around the Hawthorne T rail. The 16 mile trail allows for recreational bicycling; it is possible that properties located adjacent or near the trail will be influenced by their proximity. Complete descriptive statistics for ordinary least squares regressions can be fou nd in Appendix A. Geographically Weighted Regression (GWR) Since the ordinary least squares results include properties throughout all of Alachua County, geographically weighted regression was employed to address spatial differences. While ordinary least sq uares disregards the possibility of spatial variants, geographically weighted regression, a function found in the ArcMap GIS software creates a regression for each individual feature (in this case, each parcel) located in the dataset. After the GWR was executed, 14,015 of the 50,309 parcels (27.9%) returned coefficients of 1.797698 x 10 308 According to the manufacturer of the ArcMap software, ESRI, features ( parcels) which return this value should not be trusted due to local

PAGE 45

45 multicollinearity (ESRI, 2013) ; thus, these records were removed, leaving 36,294 parcels to evaluate. Descriptive statistics for these remaining parcels as used for geographically weighted regression can be found in Appendix B When using properties within the whole county, all coefficients showed their expected sign, with the exception of the distance from bike lane variable. This suggests that over all, the value of homes is not positively impacted by bicycle lanes and is supported by the results of the ordinarily least squares analysis. H owever the location at which a home is located may produce different results. For the 3 6 ,294 properties in the county, 15,318 (about 42%) showed a negative coefficient for the distance to bike lane variable. This means that the just value for about 42% of the properties in the GWR dataset is increased by their proximity to a bicycle facility. Figure 4 1 show s the locations of the negatively and positively correlated properties within the w hole county.

PAGE 46

46 Figure 4 1. Coefficients for distance from bike lane variable for all properties in GWR dataset.

PAGE 47

47 The mean just value for the GWR dataset is $138,383.99. The mean just value for the properties with a negative correlation coefficient for the f eet from bike variable is $141,693.93 ; in other words, the avera ge value added by bicycle lanes can be estimated at $3,309.94 ( a 2.9% increase ) in locations where there is an added value in Alachua County The averages for th e entire county may not necessarily be reliable, therefore specific locations would need to be chosen for additional analysis. Four locations were selected within the county for additional analysis The first location includes single family h omes a round the University of Florida. This location was chosen because of proximity to the university campus, facilitating ease in commuting by bicycle and shown in the data supporting bicycle ridership It is hypothesized that this area will exhibi t a negative correlation as one moves away from the university campus; that is, properties located further from the university will be valued less than those located near it The second and third locations include single family properties near t he Hawthorn e Trail and single family properties near Millhopper Road The locations were chosen because of their popularity among recreational cyclists. The Hawthorne Trail is an off street facility, while Millhopper Road has a striped bicycle lane with signage. It i s hypothesized that these two areas will exhibit a negative correlation to increasing distance as well. Lastly, single family homes near the intersection of SW Archer Road and SW 91 st Street (near Haile Plantation) were selected for the fourth location Th is area was chosen because of recent strong opposition by residents to expand bicycle facilities through this area. I t is therefore expected that there will be a positive correlation at this location indicating that

PAGE 48

48 properties further away from the trail will be valued more than those properties which are closest to it. For each of these locations, single family homes were evaluated within one mile of the facility, or in the case of the university campus, within one mile of the border. This distance was c hosen because it was the furthest distance in the literature at which all variables were statistically significant, regardless of their sign. University of Florida There are 3,212 single family properties in the GWR dataset which are within one mile of th e University of Florida boundary. Of these, 2,492 (about 78%) exhibited a negative coefficient for the feet from bike lane variable. The mean coefficient for this location is 2.92, suggesting that, on average, the just value of homes within one mile of th e University of Florida increase by $2.92 for every foot one moves closer to the university. The average distance from a bicycle facility for this location is 1102.40 feet (a little less than of a mile). Th e values in this area conform to the expected va lue. Figure 4 2 shows the homes within one mile of the University of Florida and whether they are positive or negatively impacted.

PAGE 49

49 Figure 4 2. Coefficients for distance from bike lane variable for all properties within one mile of the University of Florida.

PAGE 50

50 Hawthorne Trail There are 895 single family properties in the GWR dataset within one mile of the Hawthorne Trail. Of these, only 283 produced a negative coefficient for the feet from bike lane variable. The mean coefficient for these homes is 2.36 suggesting that the average just value increases by $2.36 per foot as one moves away from the trail. The average distance from the trail is 1027.16 feet. The values in this area do not conform to the expected value. Figure 4 3 shows the homes within one mile of the Hawthorne Trail and how their just value is impacted by proximity.

PAGE 51

51 Figure 4 3. Coefficients for distance from bike lane variable for all properties within one mile of the Hawthorne Trail.

PAGE 52

52 Millhopper Road Within one mile of Hawthorne Road, the re are 313 single family parcels in the GWR dataset Of those, just 53 (about 17%) produced a negative coefficient for the feet from bike line variable. The mean coefficient for these properties is 4.67, indicating that the average just value increases by $4.6 7 for each foot further from the trail. The average distance to Millhopper Road is 900.03 feet. The values in this area do not conform to the expected value. Figure 4 4 shows the properties within one mile of Millhopper Road and whether they are positi vely or negatively impacted.

PAGE 53

53 Figure 4 4. Coefficients for distance from bike lane variable for all properties within one mile of Millhopper Road.

PAGE 54

54 SW Archer Road & SW 91 st Street There are 1,135 single family properties within one mile of the intersection of SW Archer Road and SW 91 st Street in the GWR dataset Within this area, not a single property showed produced a negative correlation for the feet from bike lane variable. The mean coefficient for this location was 5.95, indicating an increa se of $5.95 for each foot one moves away from a bicycle facility. This conforms to the expected value. Figure 4 5 shows the properties within one mile of the intersection of SW Archer Road and SW 91 st Street and how their proximity is impacted.

PAGE 55

55 Figure 4 5. Coefficients for distance from bike lane variable for all properties within one mile of the intersection of SW Archer Road and SW 91 st Street.

PAGE 56

56 CHAPTER 5 DISCUSSION AND CONCLUSION Discussion As discussed in the literature, the value of a home can be attributed to many factors. The value differs depending upon the attributes of the property, the location, as well as the distance from an available facility. Environmental factors may also play a It has been shown that the structural attributes of a single family properly produce the most impact on value. This is reflected in the t statistic of the total area of the property repeated ly producing a high v alue in different ordinary least squares regressions. The total value is further contributed to by the locational and neighborhood characteristics, albeit to a lesser extent. When reviewing the locational variable, feet from bike lane (FtFromBL) the d ista nce was found to be important, however accessibility is important to consider as well. For example, properties along the Hawthorne Trail were mostly found to not benefit from their proximity to this amenity. This does not fit the expected outcome; however the values of properties closest to the east trailhead near the city of Hawthorne did show an increased just value. The properties along the trail may not have direct access because of other neighboring properties, fences, or terrain. Trail users may need to travel a distance more than the studied distance in order to access the trail. The issue of accessibility can also be applied to homes in neighborhoods. While a property may lie within a linear distance of a facility, it may not be directly accessible w ithin that distance because of the design of the neighborhood. Depending upon that design, a user of a particular facility may be required to travel a distance further than the

PAGE 57

57 studied distances. To study the issue of accessibility, f urther analysis utiliz ing the Network Analyst feature of ArcMap would be informative. The type of facility may also have an impact. Casual cyclists may see less value in striped bicycle lanes because of perceived safety, particularly of those along major or busy roads They may only see recreational value in off street facilities such as the Hawthorne Trail. However, t he impact of off street facilities may also produce a negative impact. This stems properties. Whil e positive impacts are present such as recreational benefits some residents view the nearby trails as a potential source of crime. Others view the trails as negatively impacting privacy. These perceptions can lead to decreased value, as seen in the neigh borhoods near the intersection of SW Archer Road and SW 91 st Street an area in which the expansion of a bicycle trail was contested, and eventually blocked Limitations The study was developed to analyze single family properties, therefore the results ca nnot be applied to multi family properties or commercial properties. T hese types of properties may be impacted in a similar manner as nearby single family properties. Further research into the effect of bicycle facilities on the values of mutli family resi dences and commercial properties would be useful. The use of ordinary least squares produced a best case R squared value of 0.80 indicating that the ten variables used explained 80% of the variation in just values among the single family properties studied. The other 20% (or greater at some distances ) of the variation may be explained by variables which were not included in the model. For example, the value of properties may be impacted by a waterfront view or

PAGE 58

58 proximity to a park. There are many pos sibilities to consider all of which cannot simply be covered in this study. The just value of a home may also not be a truly accurate representation of value to an individual. The property appraiser determines this value, however the actual price of the p roperty when sold may be determined by other factors, such as the amount time on the market or structural characteristics that are not available from the property appraiser, such as a home being fenced in or having been remodeled or upgraded In regards to the distance from a bicycle lane, a homebuyer may search for a home and value it more because of its proximity and accessibility if the homebuyer is choosing the property explicitly for those reasons. A study of the actual sale prices of homes would be us eful, however inconsistent data from the property appraiser and lack of data concept. Quantifying this value is difficult, however with quality data, the use of regression can help to estimate this value. Conclusion This study examined how bicycle lanes in Alachua County impact property values of single family homes. Locational, structural, and neighborhood values were employed using ordinary least squares regression to cre ate a hedonic pricing method G eographically weighted regression was then used to improve upon the results by studying specific locations Results showed that the impact varies greatly depending upon location. Further examinat ion of specific bicycle facilities in Alachua County confirmed that the impact can be positive or negative and varies based on that location. One may draw the conclusion that when studying locational variables, a case by case approach should be taken give n the impact of location on the relationship.

PAGE 59

59 APPENDIX A RESULTS OF ORDINARY LEAST SQUARES Table A 1. Descriptive statistics for all single family properties within Alachua County. Variable Name Minimum Maximum Mean Standard Deviation ActYrBlt 1853 2011 1978.18 21.19 Acres 0.10 159.55 1.01 2.48 Beds 1 5 3.11 0.67 Baths 1 10 2.00 0.68 Stories 1 5 1.14 0.42 TotArea 180 22685 2375.78 1082.24 AC_code 0 1 0.94 0.24 FtFromBL 0 34324.10 2105.14 2895.89 DenPop2010 0 146.63 3.77 3.50 Pct_Mnrty 0 100 29.86 25.17 n = 8,905 Table A 2 Descriptive statistics for all single family properties within 500 feet of a bicycle facility in Alachua County. Variable Name Minimum Maximum Mean Standard Deviation ActYrBlt 1853 2011 1971.82 23.41 Acres 0.10 159.55 0.85 2.46 Beds 1 5 3.00 0.65 Baths 1 8 1.88 0.64 Stories 1 3 1.10 0.37 TotArea 252 22685 2127.00 963.99 AC_code 0 1 0.91 0.28 FtFromBL 0 434.02 204.95 131.38 DenPop2010 0 57.39 4.10 3.78 Pct_Mnrty 0 100 32.95 26.93 n = 8,905

PAGE 60

60 Table A 3 Descriptive statistics for all single family properties within 1,000 feet of a bicycle facility in Alachua County. Variable Name Minimum Maximum Mean Standard Deviation ActYrBlt 1853 2011 1973.42 22.43 Acres 0.10 159.55 0.71 2.00 Beds 1 5 3.03 0.65 Baths 1 8 1.90 0.64 Stories 1 4 1.11 0.37 TotArea 252 22685 2160.86 961.95 AC_code 0 1 0.92 0.27 FtFromBL 0 868.30 424.89 257.47 DenPop2010 0 70.61 4.26 3.60 Pct_Mnrty 0 100 32.41 26.54 n = 17,534 Table A 4 Descriptive statistics for all single family properties within 2,000 feet of a bicycle facility in Alachua County. Variable Name Minimum Maximum Mean Standard Deviation ActYrBlt 1853 2011 1974.71 21.91 Acres 0.10 159.55 0.69 1.90 Beds 1 5 3.07 0.65 Baths 1 8 1.93 0.64 Stories 1 4 1.11 0.38 TotArea 252 22685 2214.40 981.98 AC_code 0 1 0.93 0.25 FtFromBL 0 1735.78 787.40 483.86 DenPop2010 0 70.60 4.25 3.46 Pct_Mnrty 0 100 32.08 26.19 n = 31,061

PAGE 61

61 Table A 5 Descriptive statistics for all single family properties within mile of a bicycle facility in Alachua County. Variable Name Minimum Maximum Mean Standard Deviation ActYrBlt 1853 2011 1974.06 22.17 Acres 0.10 159.55 0.69 1.86 Beds 1 5 3.04 0.64 Baths 1 8 1.91 0.63 Stories 1 4 1.11 0.37 TotArea 252 22685 2178.89 962.14 AC_code 0 1 0.93 0.26 FtFromBL 0 1146.52 556.52 333.90 DenPop2010 0 70.61 4.27 3.56 Pct_Mnrty 0 100 31.90 26.15 n = 22,698 Table A 6 Descriptive statistics for all single family properties within mile of a bicycle facility in Alachua County. Variable Name Minimum Maximum Mean Standard Deviation ActYrBlt 1853 2011 1975.49 21.69 Acres 0.10 159.55 0.70 1.87 Beds 1 5 3.08 0.66 Baths 1 8 1.95 0.66 Stories 1 4 1.12 0.39 TotArea 180 22685 2253.95 1016.17 AC_code 0 1 0.93 0.25 FtFromBL 0 2292.51 974.24 627.08 DenPop2010 0 146.63 4.24 3.61 Pct_Mnrty 0 100 32.13 26.27 n = 36,729

PAGE 62

62 Figure A 1. Ordinary least squares results for all single family properties in Alachua County

PAGE 63

63 Figure A 2. Ordinary least squares results for all single family properties within 500 feet of a bicycle facility.

PAGE 64

64 Figure A 3. Ordinary least squares results for all single family properties within 1,000 feet of a bicycle facility.

PAGE 65

65 Figure A 4. Ord inary least squares results for all single family properties within 2,000 feet of a bicycle facility.

PAGE 66

66 Figure A 5. Ordinary least squares results for all single family properties within mile of a bicycle facility.

PAGE 67

67 Figure A 6. Ordinary least squares results for all single family properties within mile of a bicycle facility.

PAGE 68

68 APPENDIX B RESULTS OF GEOGRAPHICALLY WEIGHTED REGRESSION Table B 1 Descriptive statistics for all single family properties in Alachua County used in geographically weighted regression. Variable Name Minimum Maximum Mean Standard Deviation ActYrBlt 1853 2011 1975.29 23.37 Acres 0.10 159.55 1.24 2.83 Beds 1 5 3.06 0.68 Baths 1 8 1.93 0.70 Stories 1 4 1.14 0.42 TotArea 180 22685 2306.64 1097.55 AC_code 0 1 0.92 0.28 FtFromBL 0 34324.10 2211.44 3188.35 DenPop2010 0 146.63 3.40 3.65 Pct_Mnrty 0 100 31.07 27.62 n = 36,294 Figure B 1. Properties of geographically weighted regression variables.

PAGE 69

69 APPENDIX C LIST OF BICYCLE FACILITIES IN ALACHUA COUNTY, FL Table C 1. List of bicycle facilities in Alachua County, Florida. Name Type 2300 NW 53rd Ave to NW 45th Ave OSF 6th Street Rail Trail OSF Bledsoe Dr DSORBL Buckman Dr DSORBL Center Dr DSORBL Depot Ave Rail Trail OSF Downtown Connector Trail OSF Fletcher Dr DSORBL Fraternity Row DSORBL Gainesville Hawthorne Trail OSF Gale Lemerand Dr DSORBL Hull Rd DSORBL Inner Rd DSORBL Kermit Sigmon/SR 24 Trail OSF Millhopper Rd DSORBL Mowry Dr DSORBL Museum Rd DSORBL N Main St DSORBL N SR 121 DSORPV N State Road 121 DSORPV N US HWY 441 DSORPV NE 12th Ave DSORBL NE 15th St DSORBL NE 16th Ave DSORBL NE 16th Ave DSORPV NE 19th Ter DSORBL NE 2nd St DSORBL NE 39th Ave DSORPV NE 39th Ave DSORBL NE 53rd Ave DSORPV NE 55th Blvd DSORPV NE 55th Blvd DSORBL NE 9th St DSORBL NE SR 26 DSORPV NE US HWY 301 DSORPV NE Waldo Rd DSORPV Newell Dr DSORBL Newell Dr OSF

PAGE 70

70 Name Type NW 13th St DSORBL NW 13th St DSORPV NW 143rd St DSORBL NW 143rd St DSORPV NW 14th Ave DSORBL NW 16th Ter DSORBL NW 173rd St DSORPV NW 17th St DSORBL NW 22nd St DSORBL NW 23rd Ave DSORBL NW 23rd Ave DSORPV NW 27th Ave DSORPV NW 31st Blvd DSORPV NW 34th St DSORBL NW 38th St DSORBL NW 39 Ave DSORBL NW 39th Ave DSORBL NW 39th Rd DSORBL NW 39th Rd OSF NW 43rd St DSORBL NW 45th Ave DSORBL NW 45th Ave DSORPV NW 45th Ave OSF NW 51st St DSORBL NW 53rd Ave DSORBL NW 53rd Ave OSF NW 53rd Ave DSORPV NW 55th St DSORBL NW 64th Blvd OSF NW 6th St DSORPV NW 75th St DSORPV NW 75th St OSF NW 91st St OSF NW US HWY 441 DSORPV Old Archer Road Trail OSF Radio Rd DSORBL S Main St DSORBL S US HIGHWAY 441 DSORBL S US HWY 441 DSORBL SE 11th St DSORBL SE 16th Ave DSORPV SE 24th St DSORBL

PAGE 71

71 Name Type SE 2nd Ave DSORBL SE 2nd Ave OSF SE 35th St DSORPV SE 3rd Ave DSORBL SE 4th Ave DSORBL SE 7th Ave DSORBL SE 8th Ave DSORBL SE Blvd OSF SE Depot Ave DSORBL SE Hawthorne Rd DSORBL SE Hawthorne Rd DSORPV SE US HWY 301 DSORBL SE US HWY 301 DSORPV SE Williston Rd DSORBL SR 235 DSORPV SR 24 DSORPV SR 45 DSORPV Stadium Rd DSORBL SW 10th Ter OSF SW 122nd St DSORPV SW 12th St DSORBL SW 13 ST US 441 DSORBL SW 13th St DSORBL SW 16th Ave DSORBL SW 16th St DSORBL SW 20th Ave DSORBL SW 23rd Dr DSORBL SW 23rd St DSORBL SW 23rd Ter Trail OSF SW 24th Ave DSORBL SW 24th Ave OSF SW 2nd Ave DSORBL SW 2nd Ave SR 26A OSF SW 33rd Pl DSORPV SW 34th St DSORBL SW 35th Pl DSORBL SW 37th Blvd DSORPV SW 38th Ter OSF SW 43rd St DSORBL SW 62nd Blvd Sidewalk OSF SW 62nd St Sidewalk OSF SW 75th St DSORBL

PAGE 72

72 Name Type SW 75th St DSORPV SW 8th Ave DSORPV SW 91st St OSF SW Archer Rd DSORBL SW Archer Rd DSORPV SW Archer Rd OSF SW SR 45 DSORPV SW Williston Rd DSORBL SW Williston Rd OSF Terwilliger Path OSF Union Dr DSORBL US Hwy 441 DSORPV W Newberry Rd DSORBL W Newberry Rd DSORPV W University Ave DSORBL W US HWY 27 DSORPV Waldo Rd OSF Waldo Rd DSORBL Waldo Rd DSORPV Waldo Rd Greenway Trail OSF Waldo Rd Greenway Depot Ave Rail Trail OSF OSF = Off street facility DSORBL = Dedicated street oriented route with bike lane DSORPB = Dedicated street oriented route with paved shoulder

PAGE 73

73 LIST OF REFERENCES Alachua County Property Appraiser. (n.d.). Frequently asked questions. Retrieved February 14, 2014, from Alachua County Property Appraiser: http://www.acpafl.org/faq.html Campbell, H. S., & Munroe, D. K. (2007, May). Greenways and greenbacks: The impact of the Catawba Regional Trail on property values in Charlotte, North Carolina. Southeastern Geographer, 47(1), 118 137. City of Gainesville. (2011, August 30). Bicycle & pedestrian program. Retrieved October 17, 2011, from City of Gainesville: http://www.cityofgainesville.org/GOVERNMENT/CityDepartmentsNZ/PublicWork s/TransportationServices/BicyclePedestrianProgram/tabid /274/Default.aspx Correll, M. R., Lillydahl, J. H., & Singell, L. D. (1978, May). The effects of greenbelts on residential property values: Some findings on the political economy of open space. Land Economics, 54(2), 207 217. ESRI. (2013, April 18). Geogra phically weighted regression. Retrieved February 14, 2014, from ArcGIS Help 10.1: http://resources.arcgis.com/en/help/main/10.1/index.html#//005p0000002100000 0 Florida Department of Environmental Protection. (2012). Final balances fiscal year 2011 2012. Re trieved January 21, 2014, from Devision of Recreation and Parks: http://www.dep.state.fl.us/parks/files/linked/final_balances_fy11_12.pdf Florida Department of Transportation. (2012). Alachua projection information. Retrieved October 10, 2012, from Florida Department of Transportation: http://www.dot.state.fl.us/publicinformationoffice/construc/pdf%20files/ALACHUA. pdf Florida Division of Recreation and Parks. (n.d.). Florida State Parks. Retrieved from Gainesville Hawthorne State Trail: http://www.floridast ateparks.org/gainesville hawthorne/ Freeman, A. M. (2003). The measurement of environmental an resource values: theory and methods. Washington, DC: Resources for the Future. Gainesville Community Redevelopment Agency. (2013). CRA projects. Retrieved 22 201 4, January, from Gainesville Community Redevelopment Agency: http://www.gainesvillecra.com/about_cra_projects.php Greer, D. L. (2000). Omaha recreational trails: Their effect on property values and public safety. National Park Service.

PAGE 74

74 Hawthorne, T., Krygi er, J., & Kwan, M. P. (2008, March). Mapping ambivalence: Exploring the geographies of community change and rails to trails development using photo based Q method and PPGIS. Geoforum, 39(2), 1058 1078. Karadeniz, D. (2008). The impact of the Little Miami S cenic Trail on single family residential property values. University of Cincinnati, Cincinnati, OH. Krizek, K. J. (2006). Two approaches to valuing some of bicycle facilities' presumed benefits. Journal of the American Planning Association, 72(3), 309 320. Krizek, K. J. (2007). Estimating the cconomic benefits of bicycling and bicycle facilities: An interpretive review and proposed methods. In P. Coto Milln, & V. Inglada, Essays on Transport Economics (pp. 219 248). Physica Verlag HD. Legg, R., & Bowe, T. (2009, Spring). Applying geographically weighted regression to a real estate problem. ArcUser, pp. 44 45. Lindsey, G., Man, J., Payton, S., & Dickson, K. (2004, Fall). Property values, recreation values, and urban greenways. Journal of Park and Recreation Administration, 22(3), 69 90. Lindsey, G., Wilson, J., Yang, J. A., & Alexa, C. (2008, February). Urban greenways, trail characteristics and trail use: Implications for design. Journal of Urban Design, 13(1), 53 79. McClenny, B. (2012, September 25). County: No trail through Haile. The Gainesville Sun. Retrieved from http://www.gainesville.com/article/ 20120925/ARTICLES/120929730 McKenzie, B., & Rapino, M. (September 2011). Commuting in the United States: 2009. United States Census Bureau. Netusil, N. R (2005, May 1). The effect of environmental zoning and amenities on property values: Portland, Oregon. Land Economics, 81(2), 227 246. Nicholls, S., & Crompton, J. L. (2005). The impact of greenways on property values: Evidence from Austin, Texas. Journal of Leisure Research, 37(3), 321 341. Penna, C. D. (2006). Home sales near two Massachusetts rail trails. Racca, D. P., & Dhanju, A. (2006). Property value/desirability effects of bike paths adjacent to residential areas. University of Delaware, College o f Human Services, Education, and Public Policy, Newark, DE. Rails to Trails Conservancy. (2005, March). Railbanking and rail trails: A legacy for the future. Retrieved from Rails to Trails Conservancy: http://www.railstotrails.org/resources/documents/resou rce_docs/RailbankingHisto ry.pdf

PAGE 75

75 Santa Fe Office of Planning and Institutional Effectiveness. (2013). Fast facts for SFCC 2013. Retrieved January 22, 2014, from http://www.sfcc.edu/opie/fast_facts_2013 Sprinkle Consulting, Inc. (2001). Alachua countywide bi cycle master plan. U. S. Census Bureau. (2014, January 6). State & county quickfacts: Alachua County, Florida. Retrieved January 21, 2014, from http://quickfacts.census.gov/qfd/states/12/12001.html UF Office of Institutional Planning and Research. (2013, February 27). Enrollment by class level. Retrieved January 22, 2014, from University of Florida Factbook: Enrollment: http://www.ir.ufl.edu/factbook/enroll.htm UF Office of Sustainability. (2013). Transportation. Retrieved June 1, 2013, from Sustainable UF : http://sustainable.ufl.edu/transportation/ University of Florida. (2013, August 2). Facts and rankings. Retrieved February 18, 2014, from About UF: http://www.ufl.edu/about uf/facts and rankings/ University of Florida. (2014, February 18). Health system facts and figures. Retrieved February 18, 2014, from University of Florida Health: https://ufhealth.org/health system facts and figures

PAGE 76

76 BIOGRAPHICAL SKETCH Kevin Szatmary was born in Fort Lauderdale, Florida and has lived in Florida for his entire life. After graduating from high school in 2001, he attended community college before attending the University of F lorida in Gainesville, Florida, to pursue a Bachelor of Science in Geomatics. He has served as an intern for the Army Corps of Engine ers and the National Park Service as a land surveying technician and a GIS technician. He graduated in 2009, and decided to remain at the University of Florida to pursue a Master of Arts in Urban and Regional Planning. Kevin enjoys college athletics, old m aps, and riding bicycles.