University of Southern California Dissertations and Theses

A geospatial analysis of income level, food deserts and urban agriculture hot spots

(USC Thesis Other)

A geospatial analysis of income level, food deserts and urban agriculture hot spots

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

A Geospatial Analysis of Income Level, Food Deserts and
Urban Agriculture Hot Spots

by

Hildemar Cruz

A Thesis Presented to the
Faculty of the USC Graduate School
University of Southern California
In Partial Fulfillment of the
Requirements for the Degree
Master of Science
(Geographic Information Science and Technology)

May 2016

Copyright ® 2015 by Hildemar Cruz

To those individuals that dare to challenge the collective way of life in an effort to live true to
their beliefs.

iv

Table of Contents
List of Figures ............................................................................................................................... vii
List of Tables ................................................................................................................................. ix
Acknowledgments........................................................................................................................... x
Abstract .......................................................................................................................................... xi
Chapter 1: Introduction ................................................................................................................... 1
1.1 Food Environment ............................................................................................................ 1
1.2 Food Deserts ..................................................................................................................... 2
1.3 Urban Agriculture (UA) as an Alternative ........................................................................ 3
1.4 Objective of this study ...................................................................................................... 4
Chapter 2: Background and Literature Review .............................................................................. 6
2.1 Urban Agriculture ............................................................................................................. 6
2.1.1 Defining UA........................................................................................................... 7
2.1.2 Perspectives on UA ................................................................................................ 8
2.1.3 GIS Prior Study Shortcomings and Current Study Implications ......................... 10
2.2 Food Deserts ................................................................................................................... 13
2.2.1 Accessibility ......................................................................................................... 13
2.2.2 Travel Time and Access to Transportation .......................................................... 14
2.2.3 Criteria of Income Level ...................................................................................... 15
2.3 Food Justice .................................................................................................................... 15
2.3.1 Issues with Discrimination ................................................................................... 16
2.4 UA and Food Desert Research in Los Angeles County .................................................. 18

v

Chapter 3: Methodology ............................................................................................................... 21
3.1 Study Area and Scale of Analysis ................................................................................... 21
3.2 Data and Sources............................................................................................................. 23
3.3 Methodology ................................................................................................................... 27
3.3.1 Spatial Autocorrelation and Hot Spot Analysis ................................................... 27
3.3.2 Buffers & Directional Distribution Analysis ....................................................... 30
3.4 Regression Modeling ...................................................................................................... 30
Chapter 4: Results ......................................................................................................................... 33
4.1 Hot Spot Analysis Urban Agriculture and Poverty ......................................................... 33
4.2 Hot Spot Analysis Food Deserts and Poverty ................................................................. 39
4.3 Buffer and Directional Distribution ................................................................................ 46
4.4 Regression Modeling ...................................................................................................... 52
4.4.1 Outcome of Exploratory Regression Model ........................................................ 52
4.4.2 OLS Regression Model ........................................................................................ 55
4.5 Review of Findings ......................................................................................................... 58
4.5.1 Overlap in Hottest Spots for UA and Food Deserts ............................................. 60
Chapter 5: Discussion and Conclusion ......................................................................................... 63
5.1 Summary of Findings ...................................................................................................... 63
5.2 Significance of Findings ................................................................................................. 64
5.3 Study Limitations and Future Research .......................................................................... 66
5.3.1 Limitations ........................................................................................................... 66
5.3.2 Future Research ................................................................................................... 67
Appendix A: Maps of Demographic Data Utilized in Analysis ................................................... 73

vi

Appendix B: Exploratory Regression Models – Raw Results ...................................................... 76
Appendix C: Ordinary Least Squares (OLS) Results ................................................................... 83

vii

List of Figures
Figure 1 Map of UA Sites Courtesy of CultivateLA .................................................................... 10
Figure 2 CLF Infographic Detailing the Growing Green Report for Boston ............................... 11
Figure 3 Map of Study Area: Los Angeles County ...................................................................... 22
Figure 4 Map with Consolidated Point Features of UA Sites: Los Angeles County .................... 26
Figure 5 Summary of Workflow ................................................................................................... 27
Figure 6 Hot Spot Analysis of UA Sites ....................................................................................... 35
Figure 7 Population Density for LA County................................................................................. 36
Figure 8 Comparison of UA Site Hot Spots & Percentage of Poverty for LA County ................ 37
Figure 9 Antelope Valley Region Comparison of UA Site Hot Spots & Percentage of Poverty . 38
Figure 10 Low Income & Low Access to Food Source 1-10 Miles ............................................. 40
Figure 11 Low Income & Low Access to Food Source 0.5-10 Miles .......................................... 41
Figure 12 Food Desert Hot Spot Analysis .................................................................................... 43
Figure 13 Low Income & Low Access to Food Source & Percentage Below Poverty ................ 45
Figure 14 Antelope Valley Region Comparison of Food Desert & Percentage Below Poverty .. 46
Figure 15 Buffer of Hottest Sites in Antelope Valley Region Based on Food Desert Hot Spots . 47
Figure 16 Buffer of UA Sites in Antelope Valley Region Based on Food Desert Hot Spots ...... 48
Figure 17 Directional Distribution of UA Hot Spots .................................................................... 50
Figure 18 Directional Distribution of Food Deserts Hot Spots .................................................... 51
Figure 19 Map of OLS Residuals of Selected Variable from Exploratory Regression Model ..... 57
Figure 20 Combined Hot Spots for UA site & Food Deserts ....................................................... 59
Figure 21 Northern County Hottest UA Sites in Relationship to Food Desert Hot Spots ............ 60
Figure 22 Southern County Hottest UA Sites in Relationship to Food Desert Hot Spots ............ 61

viii

Figure 23 Northern County UA Hot Spots in Relationship to Food Desert Locations ................ 62
Figures 24 & 25 Census Block Group Demographic Data for Employment & Income Ratios ... 73
Figures 26 & 27 Census Block Group Demographic Data for Public Assistance & Health
Insurance ....................................................................................................................................... 74
Figures 28 & 29 Census Block Group Demographic Data for Poverty Ratios & Percentage
Below Poverty ............................................................................................................................... 75
Figure 30 OLS Model Diagnositc Results ................................................................................... 84
Figure 31 Histograms & Scatterplots for explanatory variable & dependent variable ................. 85
Figure 32 Histograms of Residuals for OLS Model ..................................................................... 86
Figure 33 Graph of Residuals in Relation to Predicted Dependent Variable Values for OLS
Model ............................................................................................................................................ 87

ix

List of Tables
Table 1 Summary of Required Spatial Dataset ............................................................................. 23
Table 2 Summary of Required Software ...................................................................................... 24
Table 3 Summary of Explanatory Variables ................................................................................. 31
Table 4 Z-score and P-value Confidence levels ........................................................................... 34
Table 5 Highest Adjusted R-Squared Results ............................................................................... 52
Table 6 Summary of Variable Significance .................................................................................. 54
Table 7 Highest Adjusted R-Squared Results ............................................................................... 56
Table 8 Summary of OLS Results - Model Variables .................................................................. 83

x

Acknowledgments
I will be forever grateful to my family for their endless support of all of my adventures.

xi

Abstract
Since the turn of the twenty-first century, concerns with disparities in food access and food
consumption have been a growing topic for scholars and activists alike (Reisig and Hobbiss
2000; Whelan et al. 2002). The incorporation of agriculture in urban settings is one possible
remedy to sustain population growth and increasingly high demands for food. Green spaces can
help high-risk communities gain access to fresh, organic produce and reduce the presence of food
deserts. However, within the spectrum of sustainability socioeconomic factors play a critical role
in a community’s access to healthy organic foods. Although various studies associate an increase
in access to food with the implementation of urban agricultural practices (LeClair and Aksan
2014), social exclusion remains a dominant obstacle in the successful integration of Urban
Agriculture (henceforth: UA) in communities facing food insecurities (Meenar and Hoover 2012;
Tiarachristie 2013). By expanding on the research and data collected by CultivateLA, this study
assesses the relationship between clusters of different types of UA practices in LA County based
on income levels to determine possible overlaps with food deserts in underserved communities.
Using the geospatial analysis methods of Hot Spot Analysis, Buffers, and Directional
Distribution to test the bivariate hypotheses, the pattern demonstrated by each of these
phenomena, UA sites and food deserts, reveals that there is a significant statistical difference
between them based on income levels within LA County. The findings indicate that a higher
number of UA sites are located in neighborhoods with low percentages living under poverty,
while 85% of neighborhoods with high percentages living below poverty are designed as food
deserts. These results provide spatial statistical evidence of how these phenomena overlap,
providing a platform for further exploration by city planners and other policy makers to remedy
limited access to healthy foods in high-risk areas.

1

Chapter 1: Introduction
Over a decade into the new millennium, humanity continues to face the dilemma of sustaining
high demands for food as population grows. Concerns with disparities in food access and
consumption have been a growing topic for research and development within a variety of
academic and professional fields, as well as governmental agencies efforts (Reisig and Hobbiss
2000; Whelan et al. 2002). Within the spectrum of sustainability, socioeconomic factors play a
critical role in a community’s access to healthy organic foods. One remedy to this issue is the
incorporation of agriculture into densely populated urban settings. Although various studies
associate an increase in access to food with the implementation of urban agricultural practices
(LeClair and Aksan 2014), social exclusion remains a dominant obstacle in the successful
integration of Urban Agriculture (henceforth: UA) in communities facing food insecurities
(Meenar and Hoover 2012; Tiarachristie 2013). The following chapter presents the existing
conditions, problems and objectives to food access addressed in this thesis.
1.1 Food Environment
The Center for Disease Control (CDC) defines the food environment as “the physical presence of
food that affects a person’s diet; a person’s proximity to food store locations; the distribution of
food stores, food service, and any physical entity by which food may be obtained; or a connected
system that allows access to food” (Center for Disease Control 2015). Moreover, the CDC
further explains that the term food environment is also used to describe a communities’
collective local food landscape as well as the nutritional quality. The reference to a
neighborhood’s food environment is useful when describing the retail or physical aspects of food
(presence and accessibility to food stores and markets) and the consumer impact (healthiness and
affordability). Understanding the full scope of a neighborhood’s food environment enables a

2

thorough analysis of the conditions that affect the way communities feed themselves. More so,
the food communities select can be influenced by a full range of other factors, like taste, price,
convenience, knowledge and availability (Glanz et al. 1998). When deficiencies emerge in one of
the components of the food environment, other aspects are affected like overall public health
which in turn can have an economic impact (Bader et al. 2010).
1.2 Food Deserts
11.5 million people, or 4.1 percent of the total U.S. population, live in low-income areas more
than 1 mile from a supermarket. Neighborhoods with low access to affordable fresh food sources
that make up a healthy full diet are considered food deserts (CDC 2010). Alternatively, these
areas have an increase access to unhealthy cheap food. This phenomenon has been linked to
obesity and diet related health problems which pose a risk in a communities’ overall public
health as well as impacting the economic stability on both the micro and macro level (USDA
2009).
As more public resources and attention are given to the identification and assessment of
food desert, the way the qualifying variables are defined have a determining factor in the
outcome of the analysis. Two methods of assessment are primarily implemented: information
obtained by geographic information systems (GIS) and surveying/observation (LeClair and
Aksan 2014). Research on this topic shows that there still remains disparity in determining all the
available resources for food access in high poverty neighborhoods (Raja, Ma, and Yadav 2008;
LeClair and Aksan 2014; Short, Guthman, and Raskin 2007).
Geographic Information Systems (GIS) technology is already widely used in the daily
lives of most urban city dwellers (Li 2004). In regards to the methods of measurement of food
access, as suggested by LeClair and Aksan (2014), there is a great need to rethink the methods

3

employed to define areas that lack nutritious and affordable food which are classified as food
deserts. Between navigating streets to locating the nearest resource, basic user-friendly
geospatial tools are just a smart phone away. In performing a geospatial analysis and establishing
the nature of the relationship between food desert hot spots and urban agricultural hot spots
based on income level, municipalities can allocate resources to remedy food access issues in high
risk areas.
1.3 Urban Agriculture (UA) as an Alternative
In support of this growing movement, Olivier De Schutter (2014), the Special Rapporteur on the
right to food for the United Nations (UN), states that the push to focus food production towards
rebuilding local food systems making them decentralized and flexible benefits both local
producers as well as consumers. According to a study conducted by the non-profit Conservation
Law Foundation (CLF), urban agriculture (UA) can positively affect a community in multiple
ways, including reducing carbon footprints, producing micro businesses, and serving
communities (CLF 2012). Initially, UA alleviated the environmental strain on dense urban cities.
The CLF study explains, UA does so by reducing the demand for imported produce, improving
domestic water use through gray water systems, and reducing pollutants in the atmosphere with
the establishment of roof gardens (CLF 2012).
By creating open green spaces, communities can also better identify with their
surroundings, producing a greater desire to care for the land. Green spaces can help high-risk
communities gain access to organic, fresh produce, dismantling food deserts. However, within
the spectrum of self-sustainability, socioeconomic factors play a critical role in a community’s
access to healthy organic foods. Although various study associate an increase in access to food
with the implementation of urban agricultural practices (LeClair and Aksan 2014), social

4

exclusion remains a dominate influence in the successful integration of UA in communities
facing food insecurity (Meenar and Hoover 2012; Tiarachristie 2013).
1.4 Objective of this study
The purpose of this thesis is to conduct an analysis examining the relationship among income
levels, food desert hot spots, and urban agricultural hot spots in Los Angeles (LA) County,
California by expanding existing studies of each topic. As an emerging social movement, urban
community-based agriculture such as Community Supported Agriculture (CSA), farmer’s
markets, and community gardens have the potential to remedy Food Deserts (Meenar, Hoover
2012). Cultivate Los Angeles (http://cultivatelosangeles.org/) published a study highlighting the
state of LA’s UA practices in LA County. Although they were able to collect data and categorize
existing practices, the scope of their analysis is limited. There is an opportunity to expand on this
research and find the relationship between socioeconomic levels and food justice through the
participation in UA. This study aims to monitor the accessibility of fresh produce within dense
urban communities based on their income level and highlight disparities in areas indicated as
food deserts, which may inform policy and accommodate lack of access.
This thesis examines the relationship between clusters of different types of urban
agricultural practices in LA County based on income levels. By expanding on the research and
data collected by Cultivate LA, this research investigates possible overlaps with food deserts in
underserved communities. Initially, it is important to explain why UA is a relevant research area
in relations to food security by analyzing the positive effects on the community level. Using the
economic datasets provided by the US Census Bureau, the study established which communities
are under served due to economic hardship. The study then compares proximity to food retailers,
which provide the criteria for a food desert. Lastly, it is beneficial to understand the relationship

5

between dense urban populations and the concentration of urban agricultural practices when
outlining their utility, which in turn can inform policy to remedy limited access in high risk
areas. The study aims to show how income levels directly affect the implementation of urban
agriculture while highlighting the disparity in high risk urban demographics which are largely
surrounded by food deserts and have limited access to affordable healthy foods.
The subsequent chapters of this thesis is as follows. A review of existing research and
studies related to food access, UA, and food deserts is covered in Chapter Two. The same
chapter will examine variables utilized in previous studies to classify food deserts and their
relevance to this study. Chapter Three outlines the study area, data sources for this analysis and
any modifications applied to the datasets, and the methodologies implemented in order to assess
the relationship between clusters of different types of urban agricultural practices in LA County
based on income levels to determine possible overlaps with food deserts in underserved
communities. An analysis of the results of the methodologies used is examined in Chapter Four
including their shortcomings. Lastly, Chapter Five reviews the findings of this thesis and
includes recommendations for future research.

6

Chapter 2: Background and Literature Review
The importance of understanding the dynamics of food environments, as expanded in Chapter 1
of this thesis, determines the conditions of a community’s overall food choices and diet quality
(USDA Food Environment Atlas 2015). Concentrations of distinct occurrences such as UA and
food deserts within a demographic area can serve as an indicator of the food environment for that
neighborhood. This chapter expands on existing research regarding urban agricultural practices,
criteria for determining food deserts, and remaining obstacles for high-risk populations to food
access relative to disparities based on income, availability of resources and inequality (Bader et
al. 2010; Meenar and Hoover 2012; Tiarachristie 2013; Cohen and Reynolds 2014; Reynolds
2014). The different areas of research outlined in this chapter set the criteria for this study and
establish the parameters for this research. The studies mentioned in this chapter investigate how
these different phenomena affect selected demographics, but miss to connect and examine the
spatial statistical relationship between income and these food environment occurrences.
2.1 Urban Agriculture
UA is much more than a farmer’s market or the distribution of fresh produce by Community
Supported Agriculture (CSA). UA is the roof garden with a chicken coop that help supply fresh
eggs and vegetables to residents in apartment complexes. It is the school garden that teaches
students photosynthesis and how things can grow with care and maintenance. Moreover, it is an
opportunity to reduce the environmental impact on the already limited resources on the planet
while providing a chance of economic growth through established micro-businesses (Rogus and
Dimitri 2014; Vitiello and Wolf-Powers 2014; Ackerman et al. 2014). This trend of growing
food locally is not new, but as Schutter (2014) from the United Nations stated, it has the potential
to remedy food scarcity. This thesis aims to analyze the correlation of the increasing popularity

7

of growing food in dense yet diverse urban settings based on income levels, and provide an
analysis to delineate relationships between high concentrations of food deserts and a lack of
implemented urban agricultural practices in LA County.
2.1.1 Defining UA
As defined by Bailkey and Nasr (2000), UA involves the growing, cultivating and distribution of
food locally in and around a village, town, or city. There are two types of places UA sites
develop in: intra-urban areas, which are within a city, and peri-urban areas, which are rural
communities in the outskirts of cities, towns or villages (RUAF Foundation 2015). Schutter
(2014) mentioned in his report that the high demands for imports of goods by wealthy countries
is a driving force for the poverty around the world. He expressed that humanitarian relief should
shift into supporting impoverished countries to develop the ability to be self-sustaining and
revert to a locally invested production of resources. In order to remedy the effects of
globalization, countries must revert to local resources as well as a local mindset.
Although UA offers the potential for strengthening the social ties of a community, it
dominantly facilitates two major roles for the communities involved; food security and the
potential for economic stimulation by creating new job opportunities (Ackerman et al. 2014; US
EPA 2013; Heumann 2013). Food security means having both adequate quantity and quality of
food for a household. If either factor is compromised this can lead to health issues and is an
indication of economic difficulties. The conditions for low access to healthy foods may vary, for
example low access in rural areas consists of a different set of conditions than low access in
urban areas. The implementation of UA in low access areas has shown to be a viable method to
improve the availability of healthy foods to these communities (LeClair and Aksan 2014).

8

Several studies show that UA has proven to be a staple income for developing countries
(Zezza and Tasciotti 210). However, UA does not contribute strongly to job creation in the
United States (Vitiello and Wolf-Powers 2014; Cohen and Reynolds 2014). Issues with land use,
local food policy and the seasonal nature of UA limit the capacity for steady income flow,
although it can serve as a supplemental income in some areas (Angotti 2014). Nonetheless, UA
can economically impact a community by increasing the availability of staple foods for a
household which in turn alleviates some of the strain on resources for other expenses (Ackerman
et al. 2014).
2.1.2 Perspectives on UA
Several analyses have emerged regarding the benefits and curation of UA throughout the world.
The book by Mougeot (2005) is one of the first accounts of analysis for UA across multiple
countries with diverse socio-political and economic systems. The book concentrates on strategies
to incorporate urban farming through urban planning. The countries reviewed include Argentina,
Botswana, Côte d’Ivoire, Cuba, France, Togo, Tunisia, the UK, and Zimbabwe. There is a
growing interest in the United States to participate and implement UA, however, there still
remains a large deficit of analysis of this phenomenon, especially regarding the socioeconomic
component of participation. Mougeot’s study provides examples of case studies and examine
existing research to formulate evidence for the relationship between income and food
environments. Moreover, this association highlights that areas like food deserts dominate in low-
income communities and UA practices dominate in high-income communities in developing
countries, which is the aim of this thesis to investigate.
When observing international examples of urban agricultural implementations in a
community, Australia serves as a great site to explore, as it is socially and economically similar

9

to areas within the United States. In a study by Mason and Knowd (2010) they investigate the
development and effects of UA specifically in Sydney, Australia. The article explains how a
population’s health is affected by urban sprawl, large corporate supermarket food dominance,
obesity, and globalization. The study shows how UA can diminish those effects in the developed
world and reflects upon the increasing demand for locally grown agri-food. However, the study
does point out the challenge that most cities face in the ability to consolidate the high demand
that industrialization provides versus the growth capacity of UA practices.
Changing perspectives from a global scale to the United States, California has
considerable qualities for analysis. The state produces the most amount of food in the United
States and at the same time has two of the top five most densely populated cities in the country,
San Francisco and Los Angeles (US Census Bureau 2010). Interests in UA within these dense
cities has increased over the years reaching households through farmers markets, community
gardens, CSAs and even farm to table restaurants (Surls et al. 2015). CultivateLA is a collection
of UA sites throughout Los Angeles County. Each site was confirmed, mapped and classified as
a community garden, farm, nursery, or school garden. This data is focused on Los Angeles
County and not the whole state of California, making it a good basis to start gathering urban
agriculture data for a targeted study area. The data collected does incorporate an Agriculture
Density Index, which measures the concentration of agriculture in various cities throughout the
county (Cultivate, 2013). Their findings include:
 761 School Gardens
 211 Nurseries
 171 Farms
 118 Community Gardens
Total: 1,261

10

Figure 1 Map of UA Sites Courtesy of CultivateLA
2.1.3 GIS Prior Study Shortcomings and Current Study Implications
The 2012 case study for UA in Boston conducted by The Conservation Law Foundation and
CLF Ventures, Inc. (henceforth: CLF) creates a tangible analysis of the multi-dimensional
impact of UA in a high population, low open space city. This case study analyzes job creation,
economic benefits, environmental impacts, and health benefits for establishing 50 acres of UA in
the city of Boston. It is a very thorough investigation and provides the logistical procedures
needed to implement a citywide program, including policy barriers and opportunities. It is an
excellent account of how a city can establish a program that can holistically collect and assess
the impact of UA. Currently, city officials have not picked up this program and UA remains
random and scattered throughout the city. This study serves as an example of research being
invested in the creation of UA practices within cities but there is a lack of analysis of social
exclusion and other dominant obstacle in the successful integration of UA in the communities
within these cities facing food insecurities (Meenar and Hoover 2012; Tiarachristie 2013).

11

Figure 2 CLF Infographic Detailing the Growing Green Report for Boston
Similar to the assessment conducted for the city of Boston, the case study and report
conducted in collaboration by Urban Design Lab, The Earth Institute, and Columbia University
gives a comprehensive analysis of the potential of establishing a citywide UA program in New
York City (Ackerman, 2012). However, unlike the Boston case study, this report provides
geospatial representation of waste management, potential roof top gardens, and water
conservation through storm water collection. It is an extensive working model that incorporates
the full logistical life cycle of a citywide UA implementation.
Currently a separate organization, although heavily influenced by the previous study,
Five Boroughs (http://www.fiveboroughfarm.org) is executing its Phase III of UA throughout 5
boroughs in New York City (NYC). The program works independently to establish a citywide

12

plan to enhance the sustainability of NYC. Phase I was the developmental stage of policies and
matrices to boost and expand UA in NYC. Phase II brought about a partnership with NYC
Department of Parks & Recreation to implement and measure the impact of UA practices in the
city. This includes a 28% increase of food producing farms and gardens in the last 2 years.
Currently in Phase III, the project aims to serve as an adaptable model for UA implementation in
cities by releasing a Data Collection Toolkit. This document is made available online and
provides instructions on how to collect data from UA sites and directs registered urban farmers
to the affiliated website: http://farmingconcrete.org/barn/ (2015) to input their results. The results
are then visualized through a web map. This project is creating a platform to adapt UA practices
within multiple cities and even incorporates community development and empowerment as one
of its goals. However, the project lacks investigation of the relationship between socioeconomic
obstacles that emerge due to the range of income and varied poverty levels within the city
(Cohen and Reynolds 2015).
Chiara Tornaghi from the University of Leeds, UK (2014) published an article calling for
a critical geography of UA. As a growing trend with positive implications, Tornaghi claims that
there is a need to increase investigation on the topic. In doing so, areas of inequality can be
addressed. For example, food cultures and consumption habits in urban areas can be mapped and
analyzed to determine desired foods as well as bring awareness to possible health risks.
Currently there is a lack of investigation of the full life cycle of UA. The areas affected by UA
are connected, influenced and dependent of each other, creating a life cycle of the practice.
Buying locally grown food not only has a socioeconomic impact by creating micro businesses,
but the reduction of importing food to a region has environmental consequences as well.
Likewise, it affects the public health of a community.

13

2.2 Food Deserts
As more public resources and attention is given to the identification and assessment of food
desert, the way the qualifying variables are defined have a determining factor in the outcome of
the analysis. Two methods of assessment are primarily implemented; information obtained by
geographic information systems (GIS) and surveying/observation (LeClair and Aksan 2014).
Research on this topic shows that there still remains disparity in determining all the available
resources for food access in high poverty neighborhoods (Raja, Ma, and Yadav 2008; LeClair
and Aksan 2014; Short, Guthman, and Raskin 2007). Further research implies that small food
retailers, bodegas and corner stores may be easier to reach and cater to distinct food cultures.
However, issues of exclusivity of ethnicities and affordability still limits neighborhood’s access
to healthy food. The following sections of this chapter examine the methodology used to assess
food deserts in previously published articles, and define the following variables for this study:
access to healthy foods, travel time, access to a vehicle, and methods to outline income levels.
2.2.1 Accessibility
Accessibility to fresh, healthy and affordable foods is a dictating factor in determining if a
neighborhood is a food desert. A study conducted by Shaffer (2002) indicates 2.3 times more
supermarkets per household in Los Angeles County in high-income neighborhoods when
compared to low-income neighborhoods. The disproportion is further highlighted by ethnicity;
largely white neighborhoods have 3.2 times as many supermarkets as black neighborhoods and
1.7 times as many as Latino neighborhoods. (Shaffer 2002). Although this study is over ten years
old and demographic changes are possible to have taken place, it does indicate a measurable
disparity of access to affordable healthy foods, specifically for low-income demographics.

14

The research conducted in this thesis will match the criteria established by the United
States Department of Agriculture (USDA) Economic Research Report Number (ERR) 140 to
define which areas within Los Angeles County are considered food deserts. The report considers
an area as having low access to food sources when at least 500 people and/or 33 percent of the
tract population resides more than 1 mile from a supermarket or large grocery store in urban
areas, and more than 10 miles in rural areas (USDA 2012). Data extracted from the USDA’s
Food Access Research Atlas provides aggregated figures to expand the degree of limited access
based on availability of food sources. The data expands the criteria defined by ERR140 areas to
20 miles away from a large food store. Unfortunately, since this data is aggregated and is at a
larger unit scale, it does not enable a detailed analysis of affected demographics.
2.2.2 Travel Time and Access to Transportation
A study conducted by Inagami’s et al. (2006) on the body mass index (BMI) of low-income
neighborhoods and the locations of healthy food supplies confirmed that the longer the distance
traveled to reach a grocery store, the higher BMI in high poverty neighborhoods within Los
Angeles County. Individuals that traveled more than 1.75 miles to a market weighted about 5
more pounds then those who had shorter travel times. Access to a vehicle is an important factor
and a potential barrier for households to obtain healthy affordable foods. An alternative method
of reaching supermarkets or large grocery stores is public transportation. Using public transit to
buy food supplies, especially for demographics that can only afford to go once a month to make
purchases, can be difficult considering the amount of time and load it requires (USDA 2012).
Having access to a private vehicle alleviates the potential of community members within a food
desert to purchase low quality foods at a nearby vendor.

15

Since travel time is an important factor for access to healthy foods, the parameters of
vehicle accessibility used in the methodology of this thesis are based on the research conducted
by the USDA. The criteria established by the USDA’s Food Access Research Atlas (2015)
regarding the percentage of vehicle availability within a community, classify the variable low
vehicle access if:
 at least 100 households are more than ½ mile from the nearest supermarket and have no
access to a vehicle; or
 at least 500 people or 33 percent of the population live more than 20 miles from the
nearest supermarket, regardless of vehicle access (Food Access Research Atlas 2015).
2.2.3 Criteria of Income Level
The USDA ERR 140 report characterizes poverty levels as low-income tracts within the US
Census block groups based on two criteria; a poverty rate equal to or greater than 20 percent, or a
median family income that is 80 percent or less of the metropolitan area and/or statewide median
family income (USDA 2012). This criteria is identical to the process used by the Food Access
Research Atlas.
2.3 Food Justice
One of the positive outcomes of UA, which has been touched upon repeatedly by the previously
mentioned studies, is the nutritional benefit of growing food locally. In 2013, Assembly Speaker
John A. Perez (D-Los Angeles) delivered an editorial regarding his invested interest for his
district to develop and incorporate UA. He provides statistical support for UA in Los Angeles
County, as well as highlighting the economic benefits to Angelino communities in deflating food
deserts. Overall, this article serves as a reference point for the legislative climate in support of or
against the use of public open spaces for the cultivation of food (Perez, 2013).

16

A new social movement has emerged to tackle scarcity and access to food, it is called
Food Justice (FJ). In an effort to fight for the right to healthy fresh food, the FJ movement uses
active participation techniques to ensure that the responsibility as well as the benefits of food
systems is shared equitably. This includes how food is grown, processed, transported,
distributed, and consumed (Gottlieb and Anupama 2010). The FJ movement covers a wide range
of food inequalities ranging from farmer’s rights to transparency of labeling food. Through
activism and grassroots efforts the over-industrialized food system, which has reached a global
capacity, can increase cultural awareness of food rights. UA practices are a possible alternative
to defend FJ, however, issues of discrimination and relevance still dominate in low-income areas
when establishing UA sites.
2.3.1 Issues with Discrimination
While city planner and government agencies may be on board to implement UA practices in their
communities, broader social and economic issues must be address prior to executing a plan of
action (Surls et al. 2015). In order to fully understand the social and cultural context of food and
avoid exclusion, open dialogue with the community must be take place before implementing a
solution (Short, Guthman, and Raskin 2007; Raja, Ma, and Yadav 2008; Hu et al. 2011; LeClair
and Aksan 2014). For example, there are certain foods that are forbidden for one ethnic group,
while for another the way food is prepared and served may hold a cultural significance. Each
restriction or guideline is a key component to the way communities consume food.
Social exclusion or marginalization is a term used to describe groups within a society that
are systematically prevented from full access to the rights, opportunities and benefits that are
normally available to other groups within the society. These rights are fundamental parts of
society assimilation and include housing, employment, healthcare, civic engagement, education,

17

and more. When inequality can stunt progress and stability social exclusion not only affects the
individuals being excluded, but the society as a whole (Silver 1994). One can conclude that
social exclusion is a form of discrimination, since it constitutes the unfair treatment of a group
versus another group. However, intention plays a role in regards to the type of discrimination
that social exclusion falls into. Unintentional discrimination may still be considered unlawful
behavior. One form of unintentional discrimination is owned as disparate impact discrimination,
which is when an employer or other agent creates practices that have an inequitable unfavorable
effect on persons in a protected class (Civil Rights Act of 1964).
Social exclusion based on income level, race and ethnicity are contributing factors to the
limitations for access to healthy affordable food for underserved communities. The same study
conducted by Inagami’s et al. (2006) confirmed that Supermarkets in Los Angeles County
located in low-income neighborhoods are less likely to stock healthy foods than stores in higher-
income areas. The study collected data by performing random interviews of individuals residing
in high poverty neighborhoods and census tract data to determine the location of supermarkets
versus high poverty neighborhoods. They then performed statistical analysis using multilevel
linear regression models that resulted in this disparity. Additionally, a 2003 study by Sloane et al.
conducted a comparative analysis of available healthy affordable food in dominantly African
American neighborhoods versus wealthier neighborhoods with low concentration of African
Americans in the Los Angeles Metropolitan area. The study results show that in a high poverty
predominantly African American community in Los Angeles, 3 out of 10 food stores lacked
fruits and vegetables, while nearly all of the stores in predominantly white high income areas
sold fresh produce.

18

As criticized in the 2013 report by Giovania Tiarachristie, UA sites are glorified as a tool
for empowerment in underserved communities; however, her research shows through qualitative
analysis that lingering racism and race-class issues still remain. She conducted a study
investigating an UA revitalization project in a low-income neighborhood in Harrisburg,
Pennsylvania. The article highlights a lack of communication and knowledge base of the
demographics prior to carrying out the revitalization plans, creating conflict with the existing
community. The project also failed to take into consideration the food culture of the
neighborhood in question, creating more waste than healthy food access. Tiarachristie’s article
reinforces the need to analyze and quantify emerging patterns and relationships between the
popularity of UA practices, the reality of food deserts and how income plays a deciding factor of
participation.
2.4 UA and Food Desert Research in Los Angeles County
In the fall of 2011 the U.S. Department of Agriculture (USDA) awarded a $29,000.00 grant to
the Los Angeles Neighborhood Land Trust (LANLT) in an effort to address health issue related
to access to healthy food and support local food system. The funding expands the People’s
Garden Initiative by developing educational resources and programs related to UA by supporting
and establishing new community gardens in underserved areas (Marketing Weekly News 2011).
Prior research for Los Angeles County devoted to investigating the topics of food security and
improving access to healthy foods sources for neighborhoods designated as food deserts focuses
primarily on the criteria of food deserts, and explores the potential value of UA to improve
conditions (Los Angeles Food Policy Council 2012; Hingorani and Chau 2013; Jackson et al.
2013). However, there is a lack of investigation on the spatial statistical relationship between
income levels and these two existing component of the food environment in the county.

19

The 2011 research report by Longcore et al. addresses the issue of a lack of citywide
coordination for the implementation of community gardens as a method to remedy issues of food
access in Los Angeles County. The report documents a project to develop a municipal strategy to
guide decision makers on prioritizing which high need neighborhoods would benefit the most in
fostering community gardens. The strategies include identifying the “landscape of need” which
catalogues the neighborhoods with the greatest need for healthy affordable foods; “potential
siting considerations” or areas that are ill advised for the overall health of those involved to
establish new community gardens; and “landscape of opportunity” which maps the most
favorable areas to establish a new community garden. Each map is made available for public use
as a .kmz file and accessible to view for free through Googles Earth (Longcore et al. 2011).
The criteria selected for the exclusion and inclusion of potential areas to establish new
community gardens are of particular interest. The categories selected to avoid establishment of a
new garden include: transportation infrastructure, like freeways and rail lines; gasoline service
stations; and areas designated as contaminated with hazardous substances and pollutants, like
Superfund sites. Overall health and safety is the major consideration for excluding areas, which
from a policy and planning perspective is critical. Likewise, favorable areas for establishing new
gardens are largely community centered, such as schools, parks, places of worship, and publicly
owned vacant parcels (Longcore et al. 2011). Although this study creates a great starting point to
analyze optimal land use to identify areas of critical needs and where to establish UA sites to
remedy this need, a broader analysis is needed to fully understand the socioeconomic dynamics
of these areas. Moreover, using the same methodology to expand the analysis with other types of
UA sites like farmers markets, farms and nurseries can prove to be an essential tool for policy
makers when faced with decision on implementing services.

20

A study conducted by Ruelas et al (2011) highlights the effects of farmer’s markets in
low income urban communities in East and South Los Angeles from 2007-2009. The study
collected anonymous qualitative information for a period of two years to examine and track the
use of farmers markets and develop a demographic profile. The dominant demographic for each
market studied were Hispanic women with an income level less than $15,000 a year. The
majority of responders lived within a 4 mile radius of the market and expressed a satisfaction
with the access to healthy food options. This study highlights the potential of UA sites, farmers
markets in particular to stimulate, to reach underserved demographic groups although still at a
disadvantage regarding distance. The study is limited to measurements of market utilization
impact and satisfaction and lack quantitative analysis of the role of farmers markets for these
communities. This thesis addresses the quantitative analysis on a broader scale by statistically
examining the concentrations of incidents within a geographical area that appear over time, and
therefore providing valuable data regarding which demographics gain access to these healthy
food sources.

21

Chapter 3: Methodology
This chapter explains the selection of the study area, the data sources for this study, and the
methods used to test the bivariate hypotheses; the relationship between established UA sites and
food deserts in LA County based on poverty levels. The primary geoprocessing functions of
ArcGIS Desktop used to analyze the bivariate hypotheses are explored through the use of Spatial
Autocorrelation, Hot Spot Analysis, Buffers, and Directional Distribution Analysis to examine if
there is a relationship between the mean incomes of each phenomena. Once the data is prepared,
consolidated and preliminary analysis is conducted, then the statistical significance can be
determined by performing a Hot Spot Analysis of these features. An analysis of the pattern
demonstrated by each of these phenomena; UA sites and food deserts, can reveal if there is a
significant statistical difference between income levels for these neighborhoods.
3.1 Study Area and Scale of Analysis
Los Angeles County was selected for this analysis due to its size; estimated population in the
county as of 2014 was 10,116,705 which is about a quarter of the total population of the whole
state of California (United States Census Bureau 2015). It is an urban area with a diverse range
of ethnicities and incomes, which enables a large enough study area to uncover patterns but still
serve as a controlled variable. Due to the range and flexibility of food cultures within the county,
there is a higher chance to identify multiple clusters or patterns of food access inequality based
on the criteria outlined by the USAD’s (2009) report on food access. Some of the factors
highlighted in the study include travel time to affordable food suppliers and overall cost of food.
Figure 3 is a map displaying the block groups of the selected study area of Los Angeles County.

22

Figure 3 Map of Study Area: Los Angeles County

23
3.2 Data and Sources
Table 1 Summary of Required Spatial Dataset

Dataset File type
Data
type
Details Source
Temporal
resolution of
the dataset
Urban
Agricultural
site in LA
County
Excel
.xlsx
Point
feature
class
All captured locations of
school gardens,
nurseries, farms and
community gardens
CultivateLA
Data up to date
through July
2013
USDA
Farmers
Market
Directory
Excel
.xlsx
Point
feature
class
All registered locations
of farmers markets in the
US
United State
Department
of
Agriculture
Data up to date
through July
2015
Demographics
profile
Shapefile
Point
feature
class
Demographic data of US
and Puerto Rico
including commuting,
poverty, and income.

US Census
Bureau
Based on 2010
Census
TIGER/Line
Shapefiles and
the 2010 Census
Summary
Food Access
Research Atlas
Excel
.xlsx

Polygon
feature
class
Accessibility to sources
of healthy food.
Individual-level
resources that may affect
accessibility.
Neighborhood-level
indicators of resources.
United State
Department
of
Agriculture
Based on 2010
census tract
polygon
Census block
groups
Shapefile
polygon
feature
class
All block groups units
within California
US Census
Bureau
Boundaries
published 2010
and ACS
estimations
valid through
2013
TIGER/line
street network
files
Shapefile
and .dbf
polyline
feature
class
Street network within
California
US Census
Bureau
Published
January 12,
2014
Los Angeles
Urban Area
Shapefile
polygon
feature
class
Case study area
US Census
Bureau
Boundaries
valid as of 2010
Prevalence of
Childhood
Obesity, 2008
Shapefile
polygon
feature
class
Concentration of child
obesity
LA County
Enterprise
GIS
Based on 2008
data figures

24
Table 2 Summary of Required Software

Los Angeles County has a robust collection of diverse datasets that are readily available for
public and academic use made available by the LA County GIS Data Portal, Los Angeles
County Department of Regional Planning and academic institutions which serve as a reservoir
for GIST data. In addition, private entities have gathered and prepared a series of datasets on a
large range of topics that are available for a minimal cost. For the sake of continuity and
efficiency, the majority of data sources implemented in this study are provided by the US Census
Bureau and other governmental agencies, with the exception of data provided by CultivateLA.
The latter dataset is a research study conducted in association with an academic institution
(UCLA 2013) and therefore reassured the integrity and accuracy of the information.
The datasets utilized for this analysis provide geocoded point features of UA sites within
the county. These points include locations of farms, school gardens, and community gardens.
This data was supplied by CultivateLA and its usage has been authorized, including the
expansion of the existing dataset. In addition to the data provided by CultivateLA, point features
provided by the United States Department of Agriculture (USDA) Farmer’s Market Directory for
each registered market in LA County were extracted from the dataset and combined with the
layer from CultivateLA to create a single layer. Since both layers are projected using
Software Manufacturer Function Access
ArcGIS Desktop 10.3.1 Esri  Data Manipulation and Analysis
 Geoprocessing Functions
 Overlay analysis
 Proximity analysis
 Table analysis and
management
 Surface creation and
analysis
 Statistical analysis
 Selecting and
Extracting data
USC GIST
Server

25
GCS_North_American_1983 XY coordinate systems, their geodatabases were combined using
Microsoft Excel and a feature class was generated using the XY table tool in ArcGIS, as Figure 4
demonstrates. This process was executed without any issues.
If UA is going to serve as a remedy to food access disparities, then the value of these
designated sites must be taken into account in order to measure the impact they have on the
surrounding demographics. Not all types of UA sites have the same value in regards to
productions and accessibility. Based on the data provided by CultivateLA, the most prevalent
type of UA site throughout LA County is school gardens which total 761 out of 1,261 sites or
60%. School gardens may produce some amount of food which may supplement the diet of the
students who tend to them. As published by the research conducted by CultivateLA (2013) there
is a string of benefits for the children involved with school gardens ranging from better behavior
to improved test scores. However, school gardens remain small in scale and restrict access to the
general public.
This presents a problem when conducting exploratory analysis on accessibility of
resources. The other sites captured by CultivateLA may have some forms of restrictions as well,
like membership fees for community gardens. The data provided by CultivateLA does not
confirm if the captured sites are open to the general public nor any additional restrictions. Since
the possibility of restrictions are not confirmed for any of the sites, this study will include school
gardens within the analysis. Further research is recommended in order to fully assess the extent
of accessibility for all types of UA sites.

26

Figure 4 Map with Consolidated Point Features of UA Sites: Los Angeles County

27
3.3 Methodology

Figure 5 Summary of Workflow
3.3.1 Spatial Autocorrelation and Hot Spot Analysis
Once the layers are combined, a Spatial Autocorrelation analysis is generated in order to
establish the nature of the pattern expressed with the set of features and the associated attributes.
Establishing the spatial correlation of these features confirms if there is a significant statistical
pattern. This in turn can provide important information to policy makers or interested agencies
when implementing a new program to address issues of access to healthy foods. The Spatial
Autocorrelation tool from the Spatial Statistics toolbox uses the Global Moran’s I function to
calculate the Z score value for the consolidated UA dataset to determine if the null hypothesis is
either accepted or rejected. Patrick Alfred Pierce Moran defines Moran's I equation as (1950):

28

(1)

where is the number of spatial units indexed by and ; is the variable of interest ; is the
mean of ; and is an element of a matrix of spatial weights.

Secondly, the Hot Spot Analysis function is run from the Spatial Statistics toolbox to
reflect hot and cold clustering of UA point features. Our eyes and minds naturally try to find
patterns, regardless if they exist or not. A hot spot analysis tool can provide a statistical
confirmation of concentrations of incidents within a limited geographical area that appear over
time. Therefore, quantifying the spatial pattern of UA sites and food deserts in Los Angeles
County by running a hot spot analysis can provide valuable data regarding which demographics
gain access to these healthy food sources. Before generating the analysis, a spatial weights
matrix needs to be created. Providing a weight for each feature is required to establish an
accurate statistical measure of the data. This study used an inverse distance weighted strategy for
neighboring features to reflect the variation of their influence. Additionally, in an effort to further
explore the pattern created by this dataset, it is important to highlight an Average Nearest
Neighbor summary of the combined UA sites layers which shows significant amount of
clustering with a negative z-score of -50.468. This indicates low values clustered in the study
area.
The spatial weighted matrix (SWM) file generated was applied to account for the
conceptualization of spatial relationships and the distance method implemented is Euclidean
distance, which is calculated with the following equation:
D = sq root [(x1–x2)**2.0 + (y1–y2)**2.0] (2)
Concentration levels, or hot spots, are highest in the south and southwest regions of the county,
which are the most densely populated regions of the county.

29
The next set of data to incorporate into the analysis is demographic information for LA
County from the US Census Bureau. The unit of analysis for this layer is census block groups.
This dataset provides a larger range of demographic information for the entire USA for the last 5
years, including information on commuting, poverty, and income in Geodatabase table format.
The information provided within this dataset will later be combined with a selected region with
the hottest concentrations of UA and food desert to serve as explanatory variables when
conducting an Exploratory Regression analysis and further regression modeling. This layer
delineates the median income levels for both hot spots of UA sites as well as food deserts.
The feature attributes extracted for Los Angeles County from the US Census Bureau data
table include income, commuting and housing characteristics. This data was then joint to the
dataset from the USDA’s Food Access Research Atlas. The USDA provided this dataset for
download on their website which provides an analysis of food deserts throughout the US (2015).
The tables are easily joined since they shared the same GEOIDs, although a new field for each
table was created and the integers of the GEOID fields were copied over. The study characterizes
low-income tracts within the US Census block groups based on two criteria: a poverty rate equal
to or greater than 20 percent, or a median family income that is 80 percent or less of the
metropolitan area and/or statewide median family income (USDA 2015). This study defines low
access to food sources or living far from a market where ½ mile distance was used in urban areas
and 10 miles was used in rural areas. Additionally, the parameters used by the Food Access
Research Atlas will be utilized, henceforth defining low vehicle access if at least 100 households
are more than ½ mile from the nearest supermarket and have no access to a vehicle.

30
3.3.2 Buffers & Directional Distribution Analysis
To understand the spatial extent and the regional movement of local food systems in Los
Angeles County, Proximity toolsets were implemented to determine the contiguity of features.
The Buffer tool is frequently used in studies utilizing geographical information systems (GIS) to
measure accessibility in Food Environments (Charreire et al. 2010). This study used a series of
buffers to delineate categories of Low Access to food sources as outlined by the USDA’s Food
Access Research Atlas; within ½ -10 miles. Additionally, a Directional Distribution Analysis
tool from the Spatial Statistics toolbox is applied to both the dataset for UA and food deserts to
determine if there is a relationship to any particular feature by highlighting their distributional
trends. In order to ensure that the desynchronization of UA and food deserts is represented in a
clear scale appropriate to the analysis conducted by this research, the County level will not be the
scale of analysis. Rather, smaller unites of analysis and study areas will be selected based on the
results of the hot spot analysis. This will therefore take into account the mountainous divide
within the geography of the county, which accounts for the limited population.
Prior to executing both analyses mentioned above, the data from the Food Access
Research Atlas was examined to explore the validity of the comparative analysis. The
frequencies of populations living far away from affordable food sources by ½-10 miles in LA
County totaled to 12.8% of the total population in 2010 census. Low income neighborhoods with
low access to food total 6.5 percentage of the population and low income tracts for the county
total 48%. The results of the analysis will be discussed in the next chapter.
3.4 Regression Modeling
Regression modeling is the first step to further understand what factors may lead to the spatial
patterns of UA and food deserts and inform decisions to better equip underserved communities

31
with fresh and affordable food sources. Based on the previously conducted analysis, one region
was identified for further exploration. The block with the “hottest” collection of both US sites
and classified as a food desert area is selected for an Exploratory Regression analysis. Once
selected, the data associated with the block group is extracted and combined with the
demographic data from the US Census Bureau. The dependent variable selected for the analysis
is neighborhoods with Low Access to food sources within ½-10 miles, as previously used
throughout this study. 9 explanatory variables were tested and transformed to a continuous 0-1
scale.
Table 3 Summary of Explanatory Variables
Explanatory Variables
Population Density
Percentage below Poverty
Percentage under 17
Percentage over 65
Access to vehicle
Median Income
Employment Status
Access to Health Insurance
Food Sources/UA

The following method was used to determine the weight for the population density,
population below poverty, age and obesity features. The highest and lowest values for the
following features in the selected block group were identified and given a scaled value of 0 for
the lowest and 1 for the highest. All other values were adjusted to fall within the 0-1 scale.
Access to vehicle, Employment status and access to health insurance were valued as 0 = no and 1
= yes. Household with income levels at or below poverty ($42,420 per year) received a score of
1 while incomes above received a score of 0. Lastly, areas within 1+ mile of a food source and

32
UA sites receive a score of 1 and areas closer to a food source are scored 0. The results of the
variables and parameters tested will be discussed in the following chapter.
The results of the exploratory analysis will then be used to determine what combination
of variables can yield a viable Ordinary Least Squares (OLS) model. If the exploratory analysis
does not yield a viable model, the variables with the highest significance and the model with the
highest adjusted R
2
(Adj R
2
) values will be modeled using the OLS Regression tool.

33
Chapter 4: Results
Chapter 4 documents the results of the spatial analysis conducted to test the bivariate hypotheses
to examine if there is a relationship between UA sites and food deserts in LA County based on
poverty levels through the use of Spatial Autocorrelation, Hot Spot Analysis, Buffers,
Directional Distribution Analysis and Regression Modeling. There exists limited studies and
analysis for LA County on how both phenomena affect each other. The data utilized in this
analysis were described in the previous chapter, including how they were obtained, prepared, and
the defined criteria for analysis. An analysis of the pattern demonstrated by UA sites and food
deserts can reveal if there is a significant statistical difference between income levels for these
neighborhoods.
This chapter highlights the spatial patterns or autocorrelation and examines which block
groups in LA County have the highest or lowest concentration of UA and food deserts. Section
4.1 reviews the results of the hot spot analysis of Urban Agriculture sites in the county as well as
making a comparison with areas within the county of high levels of poverty. Food desert hot
spots are examined in section 4.2 as well. Buffers and the directional distribution for selected
areas where each of these phenomena intersect are further explored in section 4.3. An
exploratory regression model is executed for the dependent variable of block groups that are
identified as low income and low access to healthy food resources within 0.5-10 miles contained
by the county. The results are reviewed in section 4.4. Lastly the collective results of these
exploratory analysis are reviewed in section 4.5.
4.1 Hot Spot Analysis Urban Agriculture and Poverty
The results from the hot spot analysis of UA sites indicate the statistically significant clusters of
these occurrences. A total of 1,438 weighted features were analyzed. Figure 6 shows

34
concentration levels, or hot spots, are highest in the south and southwest regions (Metro or
Central LA, West Side, and parts of San Fernando) of the county with a small clustering in the
north east region (Antelope Valley). These areas are the most densely populated regions of LA
County, as Figure 7 confirms, therefore justifying the results of a high concentration or “hottest"
incidents of urban agricultural practices. The resulting map in Figure 6 classifies the sites using
the GI Z-scores, separating each by the confidence percentage. The table below illustrates the
criteria of the z-score and p-values used to determine the confidence level in this analysis.
Table 4 Z-score and P-value Confidence levels

The coldest sites are ten in total and are shown in the map of Figure 6. There are three
sites in the South Bay area and seven between the Metro and San Fernando Valley region of the
county. When comparing poverty levels for the neighborhoods these sites are located, the areas
are close in proximity to neighborhoods considered below poverty levels. This outcome show
that there is a low probability that UA sites will emerging in low income neighborhoods. There a
total of 198 hottest UA sites with very minimal overlap in areas living below poverty, which
again reinforces that UA sites are less likely to emerge in low income neighborhoods.

z-score (Standard Deviations) p-value (Probability) Confidence level
< -1.65 or > +1.65 < 0.10 90%
< -1.96 or > +1.96 < 0.05 95%
< -2.58 or > +2.58 < 0.01 99%

35

Figure 6 Hot Spot Analysis of UA Sites

36

Figure 7 Population Density for LA County

37

Figure 8 Comparison of UA Site Hot Spots & Percentage of Poverty for LA County

38
In an effort to understand how the pattern of UA sites affects neighborhoods with the
greatest needs, a layer depicting percentage of poverty within LA County was added. Figure 8
shows the resulting map. The layer represents census tracts with population that falls below
poverty levels by a range of percentage starting from 0%-7.7% and scaling up to 79%. This map
shows the overlap between poverty levels and the weight of probability of UA sites within the
county. Figure 9 below enlarges the north east, Antelope Valley region, to highlight the
dynamics of these patterns and shows the relationship between both layers. Several of the hottest
UA sites fall within regions above poverty levels with very minimal sites within the highest
indicated tracts. The results of further analysis exploring the nature of the relationship between
these two factors is reviewed in the sections below.

Figure 9 Antelope Valley Region Comparison of UA Site Hot Spots & Percentage of Poverty

39
4.2 Hot Spot Analysis Food Deserts and Poverty
As explained in Chapter 3 the parameters of the data provided by the Food Research Atlas
utilized in this study are based on dense urban neighborhoods, therefore two possible
classification were tested in this analysis. Figure 10 shows the census tracts that are classified as
Low Income and Low Access to healthy food sources by 1-10 miles. Due to the scale of this
analysis and the population density in certain regions of LA County, census tract classification
fails to fully capture the nature of how these demographics interact with this space. The second
classification, represented in Figure 11 and utilized for the remainder of this analysis, is Low
Income and Low Access to healthy food sources by 0.5-10 miles.

40

Figure 10 Low Income & Low Access to Food Source 1-10 Miles

41

Figure 11 Low Income & Low Access to Food Source 0.5-10 Miles

42
The results from the hot spot analysis of the census tracts classified as food deserts also
represent the statistically significant clusters of this phenomena. Figure 12 indicates that
concentration levels, or hot spots, are highest in the south and south central regions (Metro or
Central LA, East Side, South Central, and parts of San Gabriel Valley) of the county with a small
clustering in the north east region (Antelope Valley). Once again, as Figure 7 shows, these areas
are the most densely populated regions of LA County, and justifying the results of a high
concentration or “hottest" potential for food deserts to emerge. The resulting map in Figure 12
uses the same classification parameters as used in the hot spot analysis for UA sites.

43

Figure 12 Food Desert Hot Spot Analysis

44
As previously applied to the hot spot analysis of UA sites, the layer with the
neighborhoods with percentages of below poverty neighborhoods within LA County was
compared to the layer representing low income and low access tracts within .5-10 miles of
healthy food sources. Figure 13 is the resulting map representing the overlap between these two
layers. The south and south central regions of the county have the greatest quantity of overlap
between food desert neighborhoods and high percentages living below poverty. Out of the 176
neighborhoods with high percentages of demographics living below poverty, 151 are also
classified as food deserts, which is 86% of the total. The north east, Antelope Valley region was
enlarged in Figure 14 to show the relationship between both layers as it was done with UA sites.
There are several food desert areas that overlap with the layer below poverty. Considering that
low income is a criteria for establishing regions considered as food deserts in this study, it is
expected for areas to overlap. However, it is worth mentioning that the areas overlapping did not
have the highest percentage below poverty as classified by the layer. Further analysis was
conducted and will be reviewed in the sections below.

45

Figure 13 Low Income & Low Access to Food Source & Percentage Below Poverty

46
Figure 14 Antelope Valley Region Comparison of Food Desert & Percentage Below Poverty

4.3 Buffer and Directional Distribution
The Antelope Valley region was selected for further analysis. A multi-ring buffer was applied to
the hottest UA sites. Four distances were selected to emulate the ranges associated with the
criteria for food deserts established by the USDA’s Food Access Research Atlas; 0.5, 1, 5, and
10 miles. These buffers assist in outlining the ease in access to healthy foods based on the food
desert hot spots. Figure 15 shows the results of the buffers. Three out of the nine sites selected
for the buffers are within 1 mile or less of the food desert hot spots, with the majority at a
distance of 5 miles or more. It is worth noting that within that same figure, several warm UA
sites are actually within the hottest food desert regions. As Figure 16 shows, the majority of

47
those sites are school gardens and may have limited accessibility by the surrounding community,
which is not a component provided by this study. Further analysis is recommended in order to
establish which UA sites actually allow the surrounding neighborhoods to access their resources.

Figure 15 Buffer of Hottest Sites in Antelope Valley Region Based on Food Desert Hot Spots

48

Figure 16 Buffer of UA Sites in Antelope Valley Region Based on Food Desert Hot Spots

Three new feature classes were created for both the hot spot analysis (UA sites and Food
deserts) showing the directional distribution of the mean center for each based on the results of
the hot spot analysis results. These layers summarizes the spatial trends within each feature to
further reveal possible relationships. The study areas selected for UA sites include Antelope
Valley, San Fernando Valley and part of West LA, and the remaining south east regions of the
county. Once the data for each study areas was exported and added to the map, a directional
distribution analysis was executed for each. Figure 17 shows the three directional ellipses for UA
sites. The ellipses confirms a south and east bias for the Antelope Valley region, a south and
west bias for the San Fernando and West LA region, and a deeper south and east bias for the
remaining regions of the county. The directional distribution for food deserts shown in Figure 18

49
highlights a south and west bias for the Antelope Valley and San Fernando/West LA regions,
while the remaining regions of the county show a similar bias like the UA features of south and
east.

50

Figure 17 Directional Distribution of UA Hot Spots

51

Figure 18 Directional Distribution of Food Deserts Hot Spots

52
4.4 Regression Modeling
4.4.1 Outcome of Exploratory Regression Model
Although a wide range of demographic variables were provided by US census block groups, the
results of the exploratory regression models did not provide a single passing model. The
maximum number of explanatory variables indicated in the analysis was 9. The full raw report
for the model outcome is in Appendix B and shows the results of the ninth tested potential
summary from the output report provided by ArcGIS. The highest Adj R
2
tested was 0.73. This
figure represents the amount of correlations between the dependent and independent variables
ranging between values of 0-1. The Adjust R
2
value of 0.73 is not substantially low and provides
the basis for further exploration through OLS modeling. Appendix B also shows a summary for
each section, which confirms that there is no viable model.
Table 5 shows the figures of the Exploratory Regression Global Summary, which list the
results of the five diagnostic tests for a passing model. Further examination of this section of the
report reveals low passing percentages. All the models have a value of 0.00 for both the Jarque
Bera p-value (JB) and Spatial Autocorrelation p-value (AS). This indicates non-normally
distributed model residues and significant spatial autocorrelation impacting the results. Further
exploration of these outcomes will be reviewed in the results of the OLS regression modeling.
Table 5 Highest Adjusted R-Squared Results

Percentage of Search Criteria Passed
Search Criterion Cutoff Trials # Passed % Passed
Min Adjusted R-Squared > 0.50 1,010,894 220,731 21.84
Max Coefficient p-value < 0.05 1,010,894 169,872 16.80
Max VIF Value < 7.50 1,010,894 371,174 36.72
Min Jarque-Bera p-value > 0.10 1,010,894 0 0.00
Min Spatial Autocorrelation p-value > 0.10 30 0 0.00

53
The most revealing section of this report is the Summary of Variable Significance. This
section highlights the consistency of variable relationships by confirming the percentage of
statistically significant for each candidate explanatory variable. Table 6 presents the significance
of each variable explored. The variables with the highest and most consistent % of significance
were derived from the same dataset provided by the USDA’s Food Research Atlas. The % of
significance varied thereafter from data obtained from the US Census demographics tracts. Not
having health insurance for individuals 18-64 years of age had a high % of statistical significance
throughout the analysis but was not a consistently strong predictor. Therefore, not having health
insurance can be considered a factor when combined with additional variables to explain the
phenomena of food deserts. Income per capita had a high statistical significance and maintained
a stable negative variable relationship. Similarly, households that received public assistance had
a high statistical significance and maintained a stable positive variable relationship. Overall these
figures assist in evaluating which demographic has a higher possibility of explaining or
predicting areas that are food deserts.

54
Table 6 Summary of Variable Significance
Variable % Significance % Negative % Positive
LA1_20MILES.LATRACTS_HALF
Low Access Tracts within 0.5 miles
100.00 0.00 100.00
LOW INCOMETRACTS
Low Income Tracts
100.00 0.00 100.00
LA1_20MILES.LAHUNVHALF
Vehicle access, housing units without and
low access at 0.5 mile
100.00 0.00 100.00
X27_HEALTH_INSURANCE.B27010E50
No health insurance coverage age: 35-64
99.27 34.93 65.07
X27_HEALTH_INSURANCE.B27010E33
No health insurance coverage age: 18-34
94.59 34.19 65.81
X19_INCOME.B19001E1
Per Capita Income
83.08 86.44 13.56
LA1_20MILES.POP2010
Population, tract total
82.48 46.59 53.41
X22_FOOD_STAMPS.B22010E2
Household Received Food Stamps/Snap
In The Past 12 Months
82.25 10.05 89.95
X19_INCOME.B19001E5
Household Income In The Past 12 Month
$20,000 To $24,999
80.22 40.78 59.22
X23_EMPLOYMENT_STATUS.B23025E4
Employment Status For The Population
16 Years And Over: Employed
79.90 89.74 10.26
X17_POVERTY.C17002E1
Ratio Of Income To Poverty Level In The
Past 12 Months: total
79.01 27.48 72.52
X19_INCOME.B19001E3
Household Income In The Past 12 Month
$10,000 To $14,999
77.77 40.42 59.58
X23_EMPLOYMENT_STATUS.B23025E2
Employment Status For The Population
16 Years And Over: In Labor Force
76.96 65.10 34.90
X19_INCOME.B19001E4
Household Income In The Past 12 Month
$15,000 To $19,999
73.70 51.39 48.61
X19_INCOME.B19001E2 72.90 49.34 50.66

55

The summary of Multicollinearity between the independent variables shows that there is
a significant similarity in the poverty, income, employment status, and age and sex data from the
US census. These results indicate redundancy of the explanatory variables which in turn can
indicate an over counting bias within the model, creating an unreliable model. Although the
report shows that the types of explanatory variables provided in this analysis are not strong
enough to create a viable model through this method of regression modeling, further analysis
through OLS can examine a global model to identify and measure the relationship of factors that
lead to food disparity in the county.
4.4.2 OLS Regression Model
The model selected for further regression analysis had the highest Adj R2 and the lowest Akaike
Information Criteria (AICc). Table 7 lists the nine variables for the model and the results of the
Household Income In The Past 12 Month
$20,000 To $24,999
X27_HEALTH_INSURANCE.B27010E17
No health insurance coverage age: < 18
71.61 12.79 87.21
X01_AGE_AND_SEX.B01001E1
Sex By Age: Total
71.07 35.69 64.31
X23_EMPLOYMENT_STATUS.B23025E7
Employment Status For The Population
16 Years And Over: Not in Labor Force
68.57 65.02 34.98
X23_EMPLOYMENT_STATUS.B23025E5
Employment Status For The Population
16 Years And Over: Unemployed
67.49 10.94 89.06
X01_AGE_AND_SEX.B01001E26
Sex By Age: Female
65.29 21.7 78.63
X01_AGE_AND_SEX.B01001E2
Sex By Age: Male
59.81 57.51 42.49
X27_HEALTH_INSURANCE.B27010E66
No health insurance coverage age: 65 +
26.49 60.39 39.61

56
corrected AICc, JB, Koenker’s studentized Breusch-Pagan p-value (K(BP)), the Variance
Inflation Factor (VIF), and the residual (SA).
Table 7 Highest Adjusted R-Squared Results

Figure 19 displays the output map of residuals from the OLS analysis of the above
selected variables. Issues arise when comparing the results from this map with the results to the
map of Low Income & Low Access to Food Source 0.5-10 Miles in Figure 11. The areas in blue
indicated locations whose actual value are lower than the model estimates, those areas are not
designated as food deserts in Figure 11. Neutral areas are regions with low population which
have little statistical relevance for this analysis. However, If the residuals in the red areas are
locations with actual values higher than the model estimated but are in actuality the locations
designed as food deserts by the Food Research Atlas dataset, then there seems to be a disconnect
with the explanatory variables in explaining their relationship between Low Access and Low
Income neighborhoods 0.5miles to 10 miles from healthy food sources.
AdjR2 AICc JB K(BP) VIF SA Model
0.73 -1150.51 0.00 0.00 4.29 0.00 +LA1_20MILES.LATRACTS_HALF
+LA1_20MILES.LOWINCOMETRACTS
-LA1_20MILES.POP2010
+LA1_20MILES.LAHUNVHALF
-X19_INCOME.B19001E4
-X19_INCOME.B19001E5
-X27_HEALTH_INSURANCE.B27010E50
+X01_AGE_AND_SEX.B01001E1
- X23_EMPLOYMENT_STATUS.B23025E7

57

Figure 19 Map of OLS Residuals of Selected Variable from Exploratory Regression Model

58
Appendix C has the full output report of the OLS regression model. Included in the
results of this report is a summary of the model variables. The results show that the model
explains 73% of how these variables predict or influence food deserts. All variables proved to be
statistically significant and had a VIF below 7.5 showing a low redundancy. The model the
residuals proved not to be normally distributed, indicating a biased model.
Additionally, several graphs (Figure 30) are made available for each explanatory variable
and the dependent variable. The graphs support the results from report by visualizing the issues
within the model. The graphs of the variables distribution and relationships highlight the issues
with outlines in the data. The histograms show skewed distributions by several variables. Lastly,
scatterplots of the variable distribution and relationships are linear but are not diagonal, so they
do not to represent a positive or negative relationship. Figure 31 in Appendix C shows the
histograms graph of residuals for OLS model’s over- and under predictions and confirms that the
model is bias due to the fact that residuals are not normally distributed. Overall the OLS
regression modeling enables further understanding of the influence of income, population
density, access to a vehicle, health insurance coverage, age and gender plus employment status to
the emergence of food deserts.
4.5 Review of Findings
When combining both hot spot layers for UA sites and food desert neighborhoods, the data
suggests that their patterns match. This means that there is an overlap between hot spots of UA
sites and hot spots of food deserts, both can be found within heavily populated regions of Los
Angeles County. Figure 28 maps both layers. Although the concentration of hot and cold features
seem similar, different information is revealed after further analysis of the datasets.

59

Figure 20 Combined Hot Spots for UA site & Food Deserts

60
4.5.1 Overlap in Hottest Spots for UA and Food Deserts
After selecting and exporting the hottest features from the UA hot spot analysis, all features from
the food desert hot spot layer that contain the specified UA sites were selected and exported as
well. Figures 29 and 30 show a majority of low value food desert features that contain the
“hottest” UA sites. Statistically, the highest values of UA sites and lowest values of the food
deserts fall within the same spatial location. This finding, ultimately indicates a stronger
likelihood that a UA site will emerge in neighborhoods that are not food deserts.

Figure 21 Northern County Hottest UA Sites in Relationship to Food Desert Hot Spots

61

Figure 22 Southern County Hottest UA Sites in Relationship to Food Desert Hot Spots
However, the last figure in this chapter, Figure 31, reveals that further examination of this
data is required in order to fully understand the nature of the relationship between these
phenomena. When combining the hot spot analysis of UA sites with the Low Income Low
Access .5-10 miles food desert layer, there are UA sites that geographically fall within the
neighborhoods that are classified as food deserts. These sites are not the “hottest” UA sites, but
have a 95% confidence rating. This results show that there are some healthy food resources in
high need areas. However, as indicated previously in section 4.3 of this chapter, the types of UA
sites can determine if the surrounding neighbors have access to those resources or if they are
limited to a selected group.

62

Figure 23 Northern County UA Hot Spots in Relationship to Food Desert Locations

63
Chapter 5: Discussion and Conclusion
The concluding chapter of this study provides a concise summary of the findings regarding the
hot spot analysis of UA sites and food deserts within Los Angeles County, in what direction
these patterns are distributed, and the results of the exploratory regression modeling. In addition,
the significance of these findings are discussed in reference to this study and to the topic of food
justice at large. The study concludes by reviewing the limitations of this research and suggest
future research to enhance the comprehension of phenomena and patterns related to this field.
5.1 Summary of Findings
The hot spots analysis conducted on the UA sites shows high concentration of UA sites in
the north east region or Antelope Valley region and in the south and south western regions of the
county. These areas hold the highest population density in the county, so frequency of these sites
will be more common than in less populated regions. Although the hot spot analysis for food
deserts is based on the same conditions of population density, the highest concentration of food
deserts resulted in the south, south central, central LA and San Gabriel Valley areas with some
weight given to the Antelope Valley region. Further analysis of these findings indicated that a
higher number of UA sites are located in neighborhoods with low percentages living under
poverty. However, 85% of neighborhoods with high percentages of the demographic living
below poverty are designed as food deserts.
The directional distribution of the UA sites hot spot analysis show the three directional
ellipses with a south and east bias for the Antelope Valley region, a south and west bias for the
San Fernando and West LA region, and a deeper south and east bias for the remaining regions of
the county. The directional distribution for the food desert hot spot analysis, shows a south and
west bias for the Antelope Valley and San Fernando/West LA regions while the remaining

64
regions of the county show a similar bias like the UA features of south and east. Further analysis
of these features demonstrated, after comparing the overlap of the hottest UA sites with the food
desert features that UA sites are more likely to emerge in areas not designated as food deserts.
The Antelope valley region was selected as a study area to further explore the potential of
UA sites to serve as healthy food sources for neighborhoods designated as food desert. Using the
distances established by the USDA’s Food Access Research Atlas; .5, 1, 5, and 10 miles, a buffer
was applied to the UA site features in that region. The majority of these sites are at a distance of
5 miles or more from neighborhoods designated as food deserts, while a third of the sites are
within 1 mile or less.
The results of the exploratory regression analysis did not designate a viable model due to
high instances of multicollinearity between the independent variables. The highest
multicollinearity occurred with data on poverty, income, employment status and age/sex. The
highest resulting Adj R2 of this exploration yielded 0.73. The model with the highest Adj R2 and
lowest AICc was selected for further exploration through OLS regression modeling. The results
confirmed that the all the explanatory variables are statistically significant but the model’s
residuals proved not to be normally distributed, indicating a biased model. Overall the OLS
regression modeling enables further understanding of the influence of a variety of variables to
the emergence of food deserts.
5.2 Significance of Findings
Los Angeles County has a wide range of demographics living within the region. Within the
county the broad spectrum varies from extremely affluent neighborhoods to areas housing
populations living below poverty, without access to resources like healthy affordable foods. Food
environments and food cultures emerge, as dominant groups take root in an area. Understanding

65
the full scope of a neighborhood’s food environment includes defining what kind of food sources
surround these regions, how close are these food sources, and how much time it takes to travel to
and from as well as which mode of transportation is used like public transit or an owned vehicle.
These details can enable a thorough analysis of the conditions that affect the way communities
feed themselves. Urban agriculture is propelled as a way to supplement the access to healthy
food options in dense urban settings by creating pockets of resources grown locally by the
groups that are at the highest risk.
The findings in this study confirm that the UA sites are more prominent in areas that are
not considered food deserts, are above poverty level, and already have access through different
means to healthy affordable food sources. These results provide spatial statistical evidence of
how these phenomena overlap with each other, providing a platform for further exploration.
Additionally, the results from this exploratory analysis confirm the existence of a disparity in the
successful integration of UA in communities facing food insecurities due to socioeconomic
exclusion. In order for UA to serve as a remedy against disparities in food access, then it is
critical to understanding why these sites are prominent in areas that already have access to
healthy food sources and not in neighborhoods classified as food deserts. Existing literature on
each of these topics; UA and food deserts, continue to explore the individual impact of these
food environment phenomena. However, the exploration conducted in this thesis examines the
relationship shared by each element and how income influences their development. This study
hopes to encourage city planners and other policy makers at different government levels to
further assess the relationship of how healthy food practices are being applied and understand
why certain practices are not taking shape in areas that need it the most.

66
5.3 Study Limitations and Future Research
5.3.1 Limitations
It is worth noting that this analysis contained instances of areas with low income and low access
food desert neighborhoods with UA sites within their block group in LA County. In order to
fully understand the nature of access within UA sites in the county, further information is
required. Different types of UA sites may allow or restrict access to participate, collect and
benefit from the food sources they grow. For example, a school garden may restrict access of
their harvest to students and their families. A community garden may require its participants to
rent a lot within their boundaries, creating an economic barrier for individuals who already
experience financial hardships. Weights should be given to different types of UA sites to clearly
outline if they are indeed “accessible” to neighboring communities and to which degree. This
study did not have access to those details based on the information provided by CultivateLA or
the USDA’s Farmers Market locator.
As mentioned previous in this chapter, the exploratory regression analysis did not
produce a reliable model for the dependent variable of low access, low income food desert
neighborhoods within .5-10 miles distance, due to the diversity and quality of the data provided
for this analysis. Expanding the data collected in this analysis can provide additional factors to
help determine independent variables that can serve as indicators to this occurrence. Lastly, one
challenge faced in this study includes the scale of the data utilized in the variety of analyses.
Although the majority of the demographic data was made available in the smallest scale possible;
block groups, the majority of the datasets for food deserts were in the census tract scale. This
hindered the process of analysis by limiting the details in scale for specific neighborhoods
affected. The information provided by these datasets had to be outsourced through different
sources causing inconsistencies and potential for errors.

67
5.3.2 Future Research
This study provided a quantitative analysis to illustrate the patterns of UA sites based on income
and the influence that food deserts have on the emergence of these sites. However, future
analysis to understand the reasoning why these sites are more dominant in certain demographics
over others requires qualitative research. For example, surveying and interviewing of sites where
UA programs failed can reveal missing components and provide avenues to previous mistakes.
Another beneficial area to expand on is a qualitative exploration of food sources in the region in
particular. Some neighborhoods where large grocery stores chains are not readily accessible
actually have small independent convenience or corner stores that may provide a limited amount
of culturally relevant food sources. This factor may provide further information on food cultures
for high risk demographics. Likewise, it may play a crucial preliminary role before establishing
an UA site to determine which type is the best fit for the neighborhood it will serve.
In addition to assessing and providing weight to alternative food sources, specific areas of
Los Angeles County have high numbers of mobile healthy food choices such as certain food
trucks and fruit vendors. These small scale, isolated, and sometimes moving sources are not
calculated into this analysis but do provide potential healthy food choices. An exploration of the
demographic scale these features supply as well as the quantity of individuals that benefit from
them is worth exploring, with the expectation of some hybrid implementation of these practices
to remedy the growing dilemma of scarcity of access to healthy food sources.

68
REFERENCES
Ackerman, K., R. Plunz, M. Conard, R. Katz, E. Dahlgren, and P. Culligan. 2012. “The Potential
for Urban Agriculture in New York City: Growing Capacity, Food Security, and Green
Infrastructure.” Urban Design Lab, The Earth Institute, Colombia University. Accessed
on September 19, 2015.
http://www.urbandesignlab.columbia.edu/sitefiles/file/urban_agriculture_nyc.pdf

Ackerman, K., Conard, M., Culligan, P., Plunz, R., Sutto, M., & Whittinghill, L. 2014.
“Sustainable food systems for future cities: The potential of urban agriculture.” The
Economic and Social Review. 45(2): 189-206. Accessed on September 19, 2015.
http://search.proquest.com.libproxy2.usc.edu/docview/1545343321?accountid=14749

Bader, M. D. M., Purciel, M., Yousefzadeh, P. and Neckerman, K. M. 2010. “Disparities in
Neighborhood Food Environments: Implications of Measurement Strategies.” Economic
Geography. 86: 409–430. Accessed on September 19, 2015. doi: 10.1111/j.1944-
8287.2010.010

Bailkey, M., and J. Nasr. 2000. “From Brownfields to Greenfields: Producing Food in North
American Cities.” Community Food Security News. Fall 1999/Winter 2000:6

Cohen, Nevin and Kristin Reynolds. 2015. “Resource needs for a socially just and sustainable
urban agriculture system: Lessons from New York City.” Renewable Agriculture and Food
Systems. 30:103-114. Accessed on September 19, 2015.
doi:10.1017/S1742170514000210.

The Conservation Law Foundation and CLF Ventures, Inc. 2012. “Growing Green: Measuring
Benefits, Overcoming Barriers, and Nurturing Opportunities for Urban Agriculture in
Boston.” Accessed on September 19, 2015. http://www.clf.org/static-growing-
green/GrowingGreen.Report.July2012.pdf

Conway, Delores, Christina Q. Li, Jennifer Wolch, Christopher Kahle, and Michael Jerrett. 2010.
"A Spatial Autocorrelation Approach for Examining the Effects of Urban Greenspace on
Residential Property Values." Journal of Real Estate Finance and Economics. 41 (2):
150-169. Accessed on September 19, 2015.
doi:http://dx.doi.org.libproxy2.usc.edu/10.1007/s11146-008-9159-6.
http://search.proquest.com.libproxy2.usc.edu/docview/734374136?accountid=14749.

Five Boroughs Farm. 2015. Design Trust for Public Space. Accessed on September 19, 2015.
http://www.fiveboroughfarm.org/

Five Borough Farm Data Collection Toolkit. 2015. Farming Concrete Barn. Accessed on
September 19, 2015. http://farmingconcrete.org/barn/

Gottlieb, Robert, and Anupama Joshi. 2010. Food justice. Cambridge, Mass: MIT Press.

69

Heumann, L. 2013. “Urban agriculture for food security: Good but not enough.” The Urban
Fringe, A Practitioners Blog from the Berkeley Planning Journal. Accessed on September
19, 2015. http://ced.berkeley.edu/bpj/2013/03/urban-agriculture-for-food-security-good-
but-not-enough/

Hingorani, Anisha and Haan-Fawn Chau. 2013. “Los Angeles Food System Snapshot 2013
A baseline report of the Los Angeles regional foodshed.” Los Angeles Food Policy
Council. 64 pp. Accessed on July 18, 2015. http://goodfoodla.org/wp-
content/uploads/2013/11/LA-Food-System-Snapshot-Oct-2013-small.pdf

Hu, Alice, Angela Acosta, Abigail McDaniel, and Joel Gittelsohn. 2011. “Community
Perspectives on Barriers and Strategies for Promoting Locally Grown Produce from an
Urban Agriculture Farm.” Health Promotion Practice. 14(1): 69-74. Accessed on
September 19, 2015. doi: 10.1177/1524839911405849

Inagami, S., Cohen, D., Finch K. B., and Asch, S. 2006. “You are Where you Shop: Grocery
Store Locations, Weight, and Neighborhoods.” American Journal of Preventive Medicine
31(1): 10-17.

Jackson, Jaemi, Kelly Rytel, India Brookover, Nicholas Efron, Genevieve Hernandez, Erik
Johnson,Youngjun Kim, Wenchong Lai, Mayra Navarro, Alba Pena, Zachary Rehm,
Hyejeong Yoo, and Zachary Zabel. 2013. “Cultivate L.A.: An assessment of urban
agriculture in Los Angeles County.” Master Capstone. University of California
Cooperative Extension - Los Angeles.

Jiao, Junfeng, PhD., Anne V. Moudon DrSc, Jared Ulmer M.P.H., Philip M. Hurvitz PhD., and
Adam Drewnowski PhD. 2012. “How to Identify Food Deserts: Measuring Physical and
Economic Access to Supermarkets in King County, Washington.” American Journal of
Public Health 102, (10): E32-E39. Accessed on September 19, 2015.
http://search.proquest.com.libproxy1.usc.edu/docview/1222692866?accountid=14749.

LeClair, Mark S., and Anna-Maria Aksan. 2014. “Redefining the food desert: Combining GIS
with direct observation to measure food access.” Agriculture and Human Values 31, (4):
537-547

Li, C. 2004. “Spatial Ability, Urban Wayfinding and Location-Based Service: A review and first
results.” Centre for Advanced Spatial Analysis, University College London. Working
Paper Series. Paper 77-March. ISSN 1467-1298. Accessed on September 19, 2015.
https://www.bartlett.ucl.ac.uk/casa/pdf/paper77.pdf

Longcore, Travis, Christine Lam, Mona Seymour, and Alina Bokde. 2011. “LA Gardens:
Mapping to Support a Municipal Strategy for Community Gardens.” University of
Southern California GIS Research Laboratory Research Report. 9 pp. Accessed on
October 19, 2015.
http://spatial.usc.edu/wp-content/uploads/2014/03/LAGardens-Research-Report.pdf

70

Los Angeles Food Policy Council. 2012. “Building a Healthy Food System for Los Angeles:
Strategic Priorities 2012 2013.” Good Food Office City of Los Angeles. 40 pp. Accessed
on July 18, 2015. http://goodfoodla.org/wp-content/uploads/2013/02/Good-Food-Office-
Strategic-Priorities-3-8-13-1203pm.pdf
Mapping. 2013. Cultivate LA. Accessed on September 19, 2015.
http://cultivatelosangeles.org/2013/07/16/mapping-3/

Mason, D., and I. Knowd. 2010. “The emergence of urban agriculture: Sydney, Australia.”
International Journal of Agricultural Sustainability 8(1): 62-71

Meenar, Mahbubur R., and Brandon M. Hoover. 2012. “Community food security via urban
agriculture: Understanding people, place, economy, and accessibility from a food justice
perspective.” Journal of Agriculture, Food Systems, and Community Development 3, (1):
143

Mougeot, L. J. A. Ed. 2005. Agropolis: The Social, Political and Environmental Dimensions of
Urban Agriculture, Ottawa, Ontario, Canada: International Development Research
Centre.

Perez, J. 2013. “Urban agriculture is a gateway to healthy foods.” California Agriculture
67(040): 192. Accessed on September 19, 2015.
http://go.galegroup.com.libproxy.usc.edu/ps/i.do?id=GALE%7CA352599039andv=2.1an
du=usocal_mainandit=randp=AONEandsw=wandasid=82fa0006064bebcecbdf215a6ab5
8f22

Raja, Samina, Changxing Ma, and Pavan Yadav. 2008. “Beyond food deserts: Measuring and
mapping racial disparities in neighborhood food environments.” Journal of Planning
Education and Research 27, (4): 469-482

Reynolds, K. 2015. “Disparity Despite Diversity: Social Injustice in New York City's Urban
Agriculture System.” Antipode. 47: 240–259. Accessed on September 19, 2015.
doi: 10.1111/anti.12098.

Rogus, Stephanie and Carolyn Dimitri. 2015. “Agriculture in urban and peri-urban areas in the
United States: Highlights from the Census of Agriculture.” Renewable Agriculture and
Food Systems. 30: 64-78. Accessed on September 19, 2015.
doi:10.1017/S1742170514000040.

RUAF Foundation: Resource Centres on Urban Agriculture & Food Security. 2015. “Urban
Agriculture: What and Why.” Available at: http://www.ruaf.org/urban-agriculture-what-
and-why

Ruelas, Valerie, Ellen Iverson, Preston Kiekel, and Anne Peters. 2012. “The role of farmers’
markets in two low income, urban communities.” Journal of Community Health 37 (3):
554-62.

71
Sloane, D., Diamount, A., Lewis, L., et al. 2003. Improving the Nutritional Resource
Environment for Healthy Living Through Community-Based Participatory Research. The
Journal of General Internal Medicine 18 (2003): 568–575.

Shaffer, A. 2002. “The Persistence of L.A.’s Grocery Gap: The Need for a New Food Policy and
Approach to Market Development.” Center for Food and Justice, Urban and
Environmental Policy Institute. Occidental College. Accessed on September 19, 2015.
www.departments.oxy.edu/uepi/publications/the_persistence_of.htm

Short, Anne, Julie Guthman, and Samuel Raskin. 2007. “Food deserts, oases, or mirages?”
Journal of Planning Education and Research 26, (3): 352-364

Silver, Hilary. 1994. "Social exclusion and social solidarity: Three paradigms." International
Labour Review 133 (5-6): 531-78. Accessed on January 12, 2016.
http://search.proquest.com.libproxy2.usc.edu/docview/1299072328?accountid=14749

Stopka, Thomas J,PhD., M.H.S., Christopher Krawczyk PhD., Gradziel, Pat,PhD., R.D., and
Geraghty, Estella M,M.D., M.P.H. 2014. "Use of Spatial Epidemiology and Hot Spot
Analysis to Target Women Eligible for Prenatal Women, Infants, and Children Services."
American Journal of Public Health 104 (1): S183-S189. Accessed on September 19, 2015.
http://search.proquest.com.libproxy2.usc.edu/docview/1499838500?accountid=14749.

Surls, Rachel, Gail Feenstra, Sheila Golden, Ryan Galt, Shermain Hardesty, Claire Napawan and
Cheryl Wilen. 2015. “Gearing up to support urban farming in California: Preliminary
results of a needs assessment.” Renewable Agriculture and Food Systems. 30: 33-42.
Accessed on September 19, 2015. doi:10.1017/S1742170514000052.

Taylor, John R. and Sarah Taylor Lovell. 2012. “Mapping public and private spaces of urban
agriculture in Chicago through the analysis of high-resolution aerial images in Google
Earth.” Landscape and Urban Planning. 108(1): 57-70. Accessed on September 19,
2015. http://dx.doi.org/10.1016/j.landurbplan.2012.08.001

Tiarachristie, Giovania Genevieve. 2013. “Race, Class, and Food Justice in South Allison Hill,
Pa.” Dickinson College Honors Theses. Paper 53

United Nations Human Rights: Office of the High Commissioner for Human Rights. 2014.
“Local and small-scale farming: a solution to hunger and malnutrition.” Accessed on
September 19, 2015.
http://www.ohchr.org/EN/NewsEvents/Pages/LocalAndSmallScaleFarming.aspx

United States Department of Agriculture. 2012. “Characteristics and Influential Factors of Food
Deserts.” Economic Research Service. 2012 ASI 1506-9.131; economic research rpt. no.
140.

72
United States Department of Agriculture. 2009. “Access to Affordable and Nutritious Food:
Measuring and Understanding Food Deserts and Their Consequences.” Economic
Research Service. Report to Congress. Accessed on September 19, 2015.
http://www.ers.usda.gov/media/242675/ap036_1_.pdf

United States Department of Agriculture, Economic Research Service. 2015. “Food Environment
Atlas.” Accessed on September 19, 2015. http://www.ers.usda.gov/data-products/food-
environment-atlas/.aspx

United States Department of Agriculture, Economic Research Service. 2015. “Food Access
Research Atlas.” Accessed on September 19, 2015. http://www.ers.usda.gov/data-
products/food-access-research-atlas.aspx

United States Environmental Protection Agency. 2013. “Urban agriculture, basic information.”
Available at Web site Accessed on September 19, 2015.
http://www.epa.gov/brownfields/urbanag/basic.htm

“USDA expands people's garden initiative to sow seeds for community-based agriculture in
underserved areas of Los Angeles.” 2011. Marketing Weekly News: 785.

Vallianatos, M., Azuma, A. M., Gilliland, S., & Gottlieb, R. 2010. Food Access, Availability,
and Affordability in 3 Los Angeles Communities, Project CAFE, 2004-2006. Preventing
Chronic Disease, 7(2), A27.

Voigt, K. A. 2011. “Pigs in the backyard or the barnyard: Removing zoning impediments to
urban agriculture.” Boston College Environmental Affairs Law Review, 38(2), 537-
566,213.

West's Encyclopedia of American Law, edition 2. S.v. "Civil Rights Act of 1964." Accessed on
January 12, 2016
http://legal-dictionary.thefreedictionary.com/Civil+Rights+Act+of+1964

Wilson, Steven G., David A. Plane, Paul J. Mackun, Thomas R. Fischetti, and Justyna
Goworowska. 2012. “Patterns of Metropolitan and Micropolitan Population Change:
2000 to 2010.” U.S. Census Bureau. Report Number: C2010SR-01. Accessed on
September 19, 2015. http://www.census.gov/population/metro/data/c2010sr-
01patterns.html

Zezza, A. And L. Tasciotti. 2010. “Urban agriculture, poverty, and food security: Empirical
evidence from a sample of developing countries.” Food Policy 35(4): 265–273

73
Appendix A: Maps of Demographic Data Utilized in Analysis

Figures 24 & 25 Census Block Group Demographic Data for Employment & Income Ratios

74

Figures 26 & 27 Census Block Group Demographic Data for Public Assistance & Health Insurance

75

Figures 28 & 29 Census Block Group Demographic Data for Poverty Ratios & Percentage Below Poverty

76

Appendix B: Exploratory Regression Models – Raw Results
******************************************************************************
Choose 1 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model
0.38 4196.06 0.00 0.00 1.00 0.00 +LA1_20MILES.LOWINCOMETRACTS***
0.20 5840.77 0.00 0.00 1.00 0.00 +LA1_20MILES.LAHUNVHALF***
0.16 6195.65 0.00 0.00 1.00 0.00 +LA1_20MILES.LATRACTS_HALF***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
Choose 2 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model
0.72 -802.78 0.00 0.00 1.07 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS***
0.52 2527.47 0.00 0.00 1.02 0.00 +LA1_20MILES.LOWINCOMETRACTS***
+LA1_20MILES.LAHUNVHALF***
0.39 4097.32 0.00 0.00 1.28 0.00 +LA1_20MILES.LOWINCOMETRACTS*** -
X27_HEALTH_INSURANCE.B27010E50***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
Choose 3 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model
0.73 -1077.07 0.00 0.00 1.41 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF***
0.72 -843.66 0.00 0.00 1.33 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** -
X27_HEALTH_INSURANCE.B27010E50***
0.72 -827.86 0.00 0.00 1.08 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** -
X23_EMPLOYMENT_STATUS.B23025E7***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
Choose 4 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model
0.73 -1114.69 0.00 0.00 1.43 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X27_HEALTH_INSURANCE.B27010E50***

77
0.73 -1112.14 0.00 0.00 1.45 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E4***
0.73 -1106.81 0.00 0.00 1.42 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E5***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
Choose 5 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model
0.73 -1128.46 0.00 0.00 1.52 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E4*** -X27_HEALTH_INSURANCE.B27010E50***
0.73 -1124.94 0.00 0.00 1.45 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E4*** -X19_INCOME.B19001E5***
0.73 -1124.93 0.00 0.00 1.45 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E2*** -X27_HEALTH_INSURANCE.B27010E50***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
Choose 6 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model
0.73 -1133.64 0.00 0.00 1.68 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E4*** -X19_INCOME.B19001E5*** -
X27_HEALTH_INSURANCE.B27010E50***
0.73 -1132.91 0.00 0.00 1.55 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E2** -X19_INCOME.B19001E4*** -
X27_HEALTH_INSURANCE.B27010E50***
0.73 -1131.48 0.00 0.00 1.56 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E3** -X19_INCOME.B19001E4*** -
X27_HEALTH_INSURANCE.B27010E50***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
Choose 7 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model

78
0.73 -1139.93 0.00 0.00 4.10 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E4*** -X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E1*** -
X23_EMPLOYMENT_STATUS.B23025E7***
0.73 -1139.29 0.00 0.00 2.91 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E4*** -X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E26*** -
X23_EMPLOYMENT_STATUS.B23025E7***
0.73 -1138.60 0.00 0.00 1.91 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E2*** -X19_INCOME.B19001E4*** -
X27_HEALTH_INSURANCE.B27010E50***
+X23_EMPLOYMENT_STATUS.B23025E5***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
Choose 8 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model
0.73 -1147.07 0.00 0.00 4.22 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E4*** -X19_INCOME.B19001E5*** -
X27_HEALTH_INSURANCE.B27010E50*** +X01_AGE_AND_SEX.B01001E1***
-X23_EMPLOYMENT_STATUS.B23025E7***
0.73 -1146.28 0.00 0.00 2.99 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E4*** -X19_INCOME.B19001E5*** -
X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E26*** -
X23_EMPLOYMENT_STATUS.B23025E7***
0.73 -1144.17 0.00 0.00 4.15 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E2** -X19_INCOME.B19001E4*** -
X27_HEALTH_INSURANCE.B27010E50*** +X01_AGE_AND_SEX.B01001E1***
-X23_EMPLOYMENT_STATUS.B23025E7***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
Choose 9 of 22 Summary
Highest Adjusted R-Squared Results
AdjR2 AICc JB K(BP) VIF SA Model
0.73 -1150.51 0.00 0.00 4.29 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** -LA1_20MILES.POP2010**
+LA1_20MILES.LAHUNVHALF*** -X19_INCOME.B19001E4*** -

79
X19_INCOME.B19001E5*** -X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E1*** -
X23_EMPLOYMENT_STATUS.B23025E7***
0.73 -1149.60 0.00 0.00 3.04 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** -LA1_20MILES.POP2010**
+LA1_20MILES.LAHUNVHALF*** -X19_INCOME.B19001E4*** -
X19_INCOME.B19001E5*** -X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E26*** -
X23_EMPLOYMENT_STATUS.B23025E7***
0.73 -1149.52 0.00 0.00 4.24 0.00 +LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E2** -X19_INCOME.B19001E4*** -
X19_INCOME.B19001E5*** -X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E1*** -
X23_EMPLOYMENT_STATUS.B23025E7***
Passing Models
AdjR2 AICc JB K(BP) VIF SA Model
******************************************************************************
** Exploratory Regression Global Summary (LA1_20MILES.LILATRACTS_HALFAND10) **

Percentage of Search Criteria Passed
Search Criterion Cutoff Trials # Passed % Passed
Min Adjusted R-Squared > 0.50 1010894 220731 21.84
Max Coefficient p-value < 0.05 1010894 169872 16.80
Max VIF Value < 7.50 1010894 371174 36.72
Min Jarque-Bera p-value > 0.10 1010894 0 0.00
Min Spatial Autocorrelation p-value > 0.10 30 0 0.00

------------------------------------------------------------------------------

Summary of Variable Significance
Variable % Significant % Negative % Positive
LA1_20MILES.LATRACTS_HALF 100.00 0.00 100.00
LA1_20MILES.LOWINCOMETRACTS 100.00 0.00 100.00
LA1_20MILES.LAHUNVHALF 100.00 0.00 100.00
X27_HEALTH_INSURANCE.B27010E50 99.27 34.93 65.07
X27_HEALTH_INSURANCE.B27010E33 94.59 34.19 65.81
X19_INCOME.B19001E1 83.08 86.44 13.56
LA1_20MILES.POP2010 82.48 46.59 53.41
X22_FOOD_STAMPS.B22010E2 82.25 10.05 89.95
X19_INCOME.B19001E5 80.22 40.78 59.22
X23_EMPLOYMENT_STATUS.B23025E4 79.90 89.74 10.26
X17_POVERTY.C17002E1 79.01 27.48 72.52
X19_INCOME.B19001E3 77.77 40.42 59.58
X23_EMPLOYMENT_STATUS.B23025E2 76.96 65.10 34.90
X19_INCOME.B19001E4 73.70 51.39 48.61

80
X19_INCOME.B19001E2 72.90 49.34 50.66
X27_HEALTH_INSURANCE.B27010E17 71.61 12.79 87.21
X01_AGE_AND_SEX.B01001E1 71.07 35.69 64.31
X23_EMPLOYMENT_STATUS.B23025E7 68.57 65.02 34.98
X23_EMPLOYMENT_STATUS.B23025E5 67.49 10.94 89.06
X01_AGE_AND_SEX.B01001E26 65.29 21.37 78.63
X01_AGE_AND_SEX.B01001E2 59.81 57.51 42.49
X27_HEALTH_INSURANCE.B27010E66 26.49 60.39 39.61

------------------------------------------------------------------------------

Summary of Multicollinearity*
Variable VIF Violations Covariates
LA1_20MILES.LATRACTS_HALF 1.55 0 --------
LA1_20MILES.LOWINCOMETRACTS 1.86 0 --------
LA1_20MILES.POP2010 1.11 0 --------
LA1_20MILES.LAHUNVHALF 1.44 0 --------
X17_POVERTY.C17002E1 32.73 313540 X01_AGE_AND_SEX.B01001E26 (88.61),
X01_AGE_AND_SEX.B01001E1 (88.61), X23_EMPLOYMENT_STATUS.B23025E2
(75.83), X23_EMPLOYMENT_STATUS.B23025E4 (56.50),
X01_AGE_AND_SEX.B01001E2 (39.67), X23_EMPLOYMENT_STATUS.B23025E7
(17.69), X19_INCOME.B19001E1 (8.94)
X19_INCOME.B19001E1 10.99 36006 X23_EMPLOYMENT_STATUS.B23025E4
(15.10), X23_EMPLOYMENT_STATUS.B23025E2 (13.06),
X17_POVERTY.C17002E1 (8.94), X01_AGE_AND_SEX.B01001E1 (5.31),
X01_AGE_AND_SEX.B01001E26 (3.62), X01_AGE_AND_SEX.B01001E2 (2.27),
X23_EMPLOYMENT_STATUS.B23025E7 (1.30)
X19_INCOME.B19001E2 1.94 0 --------
X19_INCOME.B19001E3 1.96 0 --------
X19_INCOME.B19001E4 1.81 0 --------
X19_INCOME.B19001E5 1.67 0 --------
X27_HEALTH_INSURANCE.B27010E17 1.76 0 --------
X27_HEALTH_INSURANCE.B27010E33 3.35 0 --------
X27_HEALTH_INSURANCE.B27010E50 3.21 0 --------
X27_HEALTH_INSURANCE.B27010E66 1.08 0 --------
X01_AGE_AND_SEX.B01001E1 121.78 282385 X17_POVERTY.C17002E1 (88.61),
X01_AGE_AND_SEX.B01001E26 (66.43), X01_AGE_AND_SEX.B01001E2 (66.43),
X23_EMPLOYMENT_STATUS.B23025E2 (59.39),
X23_EMPLOYMENT_STATUS.B23025E4 (51.54),
X23_EMPLOYMENT_STATUS.B23025E7 (14.56), X19_INCOME.B19001E1 (5.31)
X01_AGE_AND_SEX.B01001E2 30.02 159997 X01_AGE_AND_SEX.B01001E1
(66.43), X17_POVERTY.C17002E1 (39.67),
X23_EMPLOYMENT_STATUS.B23025E2 (32.38),
X23_EMPLOYMENT_STATUS.B23025E4 (28.87),
X01_AGE_AND_SEX.B01001E26 (14.92),
X23_EMPLOYMENT_STATUS.B23025E7 (6.33), X19_INCOME.B19001E1 (2.27)

81
X01_AGE_AND_SEX.B01001E26 32.13 217910 X17_POVERTY.C17002E1 (88.61),
X01_AGE_AND_SEX.B01001E1 (66.43), X23_EMPLOYMENT_STATUS.B23025E2
(44.18), X23_EMPLOYMENT_STATUS.B23025E4 (37.64),
X01_AGE_AND_SEX.B01001E2 (14.92), X23_EMPLOYMENT_STATUS.B23025E7
(6.33), X19_INCOME.B19001E1 (3.62)
X23_EMPLOYMENT_STATUS.B23025E2 87.55 241171 X17_POVERTY.C17002E1
(75.83), X23_EMPLOYMENT_STATUS.B23025E4 (66.43),
X01_AGE_AND_SEX.B01001E1 (59.39), X01_AGE_AND_SEX.B01001E26 (44.18),
X01_AGE_AND_SEX.B01001E2 (32.38), X19_INCOME.B19001E1 (13.06),
X23_EMPLOYMENT_STATUS.B23025E7 (9.92)
X23_EMPLOYMENT_STATUS.B23025E4 67.18 207017
X23_EMPLOYMENT_STATUS.B23025E2 (66.43), X17_POVERTY.C17002E1
(56.50), X01_AGE_AND_SEX.B01001E1 (51.54), X01_AGE_AND_SEX.B01001E26
(37.64), X01_AGE_AND_SEX.B01001E2 (28.87), X19_INCOME.B19001E1 (15.10),
X23_EMPLOYMENT_STATUS.B23025E7 (9.90)
X23_EMPLOYMENT_STATUS.B23025E5 2.37 0 --------
X23_EMPLOYMENT_STATUS.B23025E7 12.58 24457 X17_POVERTY.C17002E1
(17.69), X01_AGE_AND_SEX.B01001E1 (14.56),
X23_EMPLOYMENT_STATUS.B23025E2 (9.92),
X23_EMPLOYMENT_STATUS.B23025E4 (9.90), X01_AGE_AND_SEX.B01001E26
(6.33), X01_AGE_AND_SEX.B01001E2 (6.33), X19_INCOME.B19001E1 (1.30)
X22_FOOD_STAMPS.B22010E2 2.70 0 --------
* At least one model failed to solve due to perfect multicollinearity.
Please review the warning messages for further information.
------------------------------------------------------------------------------
Summary of Residual Normality (JB)
JB AdjR2 AICc K(BP) VIF SA Model
0.000203 0.549768 2168.540247 0.000000 2.693563 0.000000
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF***
+X17_POVERTY.C17002E1*** -X19_INCOME.B19001E3*** -
X19_INCOME.B19001E4*** -X19_INCOME.B19001E5***
+X27_HEALTH_INSURANCE.B27010E17** -
X27_HEALTH_INSURANCE.B27010E50*** -
X27_HEALTH_INSURANCE.B27010E66
0.000189 0.549274 2175.574646 0.000000 2.471783 0.000000
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF***
+X17_POVERTY.C17002E1*** -X19_INCOME.B19001E3*** -
X19_INCOME.B19001E4*** -X19_INCOME.B19001E5*** -
X27_HEALTH_INSURANCE.B27010E50*** -
X27_HEALTH_INSURANCE.B27010E66
+X23_EMPLOYMENT_STATUS.B23025E5
0.000184 0.549273 2175.592594 0.000000 12.962272 0.000000
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF***
+X17_POVERTY.C17002E1*** -X19_INCOME.B19001E3*** -
X19_INCOME.B19001E4*** -X19_INCOME.B19001E5*** -

82
X27_HEALTH_INSURANCE.B27010E50*** -
X27_HEALTH_INSURANCE.B27010E66 +X01_AGE_AND_SEX.B01001E1

------------------------------------------------------------------------------
Summary of Residual Spatial Autocorrelation (SA)
SA AdjR2 AICc JB K(BP) VIF Model
0.000000 0.731520 -1150.509021 0.000000 0.000000 4.290658
+LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** -LA1_20MILES.POP2010**
+LA1_20MILES.LAHUNVHALF*** -X19_INCOME.B19001E4*** -
X19_INCOME.B19001E5*** -X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E1*** -
X23_EMPLOYMENT_STATUS.B23025E7***
0.000000 0.731482 -1149.602202 0.000000 0.000000 3.041721
+LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** -LA1_20MILES.POP2010**
+LA1_20MILES.LAHUNVHALF*** -X19_INCOME.B19001E4*** -
X19_INCOME.B19001E5*** -X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E26*** -
X23_EMPLOYMENT_STATUS.B23025E7***
0.000000 0.731478 -1149.518998 0.000000 0.000000 4.244403
+LA1_20MILES.LATRACTS_HALF***
+LA1_20MILES.LOWINCOMETRACTS*** +LA1_20MILES.LAHUNVHALF*** -
X19_INCOME.B19001E2** -X19_INCOME.B19001E4*** -
X19_INCOME.B19001E5*** -X27_HEALTH_INSURANCE.B27010E50***
+X01_AGE_AND_SEX.B01001E1*** -
X23_EMPLOYMENT_STATUS.B23025E7***
------------------------------------------------------------------------------
Table Abbreviations
AdjR2 Adjusted R-Squared
AICc Akaike's Information Criterion
JB Jarque-Bera p-value
K(BP) Koenker (BP) Statistic p-value
VIF Max Variance Inflation Factor
SA Global Moran's I p-value
Model Variable sign (+/-)
Model Variable significance (* = 0.10, ** = 0.05, *** = 0.01)
------------------------------------------------------------------------------

83
Appendix C: Ordinary Least Squares (OLS) Results
Table 8 Summary of OLS Results - Model Variables

Variable Coefficient [a] StdError t-Statistic Probability [b] Robust_SE Robust_t Robust_Pr [b] VIF [c]
Intercept -0.372731 0.010955 -34.023140 0.000000* 0.013365 -27.887779 0.000000* --------
LATRACTS
_HALF
0.478818 0.007152 66.952908 0.000000* 0.009610 49.825228 0.000000* 1.490116
LOWINCO
METRACTS
0.662076 0.006884 96.172641 0.000000* 0.007981 82.958325 0.000000* 1.542347
POP2010 -0.000005 0.000002 -2.332716 0.019678* 0.000002 -2.349286 0.018824* 1.096390
LAHUNVH
ALF
0.001052 0.000060 17.508036 0.000000* 0.000059 17.776475 0.000000* 1.389980
INCOME.B1
9001E4
-0.000376 0.000109 -3.460513 0.000559* 0.000114 -3.296599 0.001001* 1.481767
INCOME.B1
9001E5
-0.000357 0.000117 -3.053649 0.002283* 0.000117 -3.047284 0.002331* 1.439334
HEALTH_I
NSURANCE
.B27010E50
-0.000130 0.000031 -4.163121 0.000037* 0.000030 -4.326989 0.000019* 2.361647
AGE_AND_
SEX.B01001
E1
0.000029 0.000007 4.042121 0.000061* 0.000007 4.246849 0.000027* 4.290658
EMPLOYM
ENT_STAT
US.B23025E
7
-0.000066 * 0.000016 -4.042439 0.000061 0.000015 -4.477739 0.000010* 2.916849

84

Figure 30 OLS Model Diagnositc Results

85

Figure 31 Histograms & Scatterplots for explanatory variable & dependent variable

86

Figure 32 Histograms of Residuals for OLS Model

87

Figure 33 Graph of Residuals in Relation to Predicted Dependent Variable Values for OLS Model

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Urban green space accessibility and environmental justice: a GIS-based analysis in the city of Phoenix, Arizona

PDF

The wildland-urban interface in Lassen County, California: a change analysis 2000–2015

PDF

Redefining urban food systems to identify optimal rooftop community garden locations: a site suitability analysis in Seattle, Washington

PDF

Using pedestrian accessibility indicators to locate schools: a site suitability analysis in Greenville County, South Carolina

PDF

Finding food deserts: a study of food access measures in the Phoenix-Mesa urban area

PDF

A spatial and temporal exploration of how satellite communication devices impact mountain search and rescue missions in California’s Sierra Nevada mountain range

PDF

A spatial analysis of violent crime cold spots: testing the capable guardian component of the routine activity theory

PDF

Distribution and correlates of feral cat trapping permits in Los Angeles, California

PDF

Does the Bay Area have a social center? Delimiting the postmodern urban center of the San Francisco Bay Area

PDF

Spatial narrative of the invasive lionfish in the Western Atlantic and Caribbean oceans: a GIS story map

PDF

Residential housing code violation prediction: a study in Victorville, CA using geographically weighted logistic regression

PDF

Geographic information systems eelgrass (Zostera marina) habitat restoration suitability model, Long Island Sound, USA - a ‘sound-wide’ model

PDF

A spatiotemporal analysis of environmental risk factors of Lyme disease in the Northeastern United States

PDF

Spatial distribution of the endangered Pacific pocket mouse (Perognathus ssp. pacificus) within coastal sage scrub habitat at Dana Point Headlands Conservation Area

PDF

Scenario-based site suitability analysis and framework for biodiversity conservation: agricultural zone, Galapagos Archipelago, Ecuador

PDF

Using geospatial technology to establish marsh bird monitoring sites for a pilot study in Maine in accordance with the North American Marsh Bird Monitoring Protocol

PDF

A comparison of GLM, GAM, and GWR modeling of fish distribution and abundance in Lake Ontario

PDF

Use of GIS for analysis of community health worker patient registries from Chongwe district, a rural low-resource setting, in Lusaka Province, Zambia

PDF

GIS analysis of helicopter rescue in San Bernardino County, California

PDF

Measuring seasonal variation in food access: a case study of Everett, Washington

Asset Metadata

Creator Cruz, Hildemar (author)

Core Title A geospatial analysis of income level, food deserts and urban agriculture hot spots

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 02/26/2016

Defense Date 01/11/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag food deserts,food justice,geospatial analysis,hot spot analysis,Income,Los Angeles County,OAI-PMH Harvest,regression modeling,Social exclusion,spatial statistical analysis,urban agriculture

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Warshawsky, Daniel (committee chair), Longcore, Travis (committee member), Sedano, Elisabeth (committee member)

Creator Email hildemac@usc.edu,hildemarcruz@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-216793

Unique identifier UC11277464

Identifier etd-CruzHildem-4170.pdf (filename),usctheses-c40-216793 (legacy record id)

Legacy Identifier etd-CruzHildem-4170-0.pdf

Dmrecord 216793

Document Type Thesis

Format application/pdf (imt)

Rights Cruz, Hildemar

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA