A EPA
United States
Environmental Protection
Aqencv
EPA/600/R-17/376 | Sep 2017 | www.epa.gov/research
Modified HWBI Model(s) Linking
Service Flows to Well-Being
Endpoints:
Accounting for Environmental Quality
Office of Research and Development
NHEERL Gulf Ecology Division
-------
EPA/600/R-17/376
September 2017
Modified HWBI Model(s) Linking Service Flows to Weil-Being
Endpoints:
Accounting for Environmental Quality
Technical Research Report
By
Linda C. Harwell, Lisa M. Smith and J. Kevin Summers
ORD/NHEERL/GED
1 Sabine Island Drive
Gulf Breeze, FL 32561
Gulf Ecology Division
National Health and Environmental Effects Research Laboratory
Office of Research and Development
-------
Notice/Disclaimer Statement
This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and
approved for publication. The information in this article has been funded wholly (or in part) by the U.S.
Environmental Protection Agency. It has been subjected to review by the National Health and
Environmental Effects Research Laboratory and approved for publication. Approval does not signify
that the contents reflect the views of the Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use. All images and copyrights are the
property of the U.S. Environmental Protection Agency.
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the Nation's
land, air, and water resources. Under a mandate of national environmental laws, the Agency strives to
formulate and implement actions leading to a compatible balance between human activities and the
ability of natural systems to support and nurture life. To meet this mandate, EPA's research program is
providing data and technical support for solving environmental problems today and building a science
knowledge base necessary to manage our ecological resources wisely, understand how pollutants
affect our health, and prevent or reduce environmental risks in the future.
The National Health and Environmental Effects Research Laboratory (NHEERL) conducts systems-
based, effects research needed to achieve sustainable health and well-being. Because they are
inextricably linked, NHEERL's research encompasses both human and ecosystem health. To achieve
research goals, the Laboratory research program focuses on:
• Leading innovative research and predictive modeling efforts that link environmental condition
to the health and wellbeing of people and society.
• Advancing research and tools for achieving sustainable and resilient watersheds and water
resources.
• Advancing systems-based research to predict the adverse effects of chemicals and other
stressors across species and biological levels of organization through the development and
quantification of adverse outcomes pathways across multiple scales.
• Using integrated research to identify and characterize modifiable factors that respond to
environmental conditions, and through intervention, improve health and well-being.
• Translating and communicating integrated environmental and health effects science to impact
decisions positively at all levels.
This report presents an approach to modify ORD's Human Weil-Being Index (HWBI) relationship-
function model to increase its utility. Using ORD's Environmental Quality Index (EQI), this research
examines the potential for using existing indicator products in novel ways to add a new facet of
interpretability regarding the linkages between socio-ecological systems and human health and well-
being. The objective of this research is to demonstrate a way to combine existing composite indices for
developing a new layer of information as an extended diagnostic of well-being conditions. Additionally,
descriptions of population-specific HWBI adaptations are presented. These population adaptations
could serve as future EQI-modified HWBI use cases.
William H. Benson, Director
National Health and Environmental Effects Research Laboratory
iii
-------
Table of Contents
Notice/Disclaimer Statement ii
Foreword iii
Acronyms and Abbreviations vii
Acknowledgments viii
Executive Summary ix
1.0 Introduction 12
1.1 Intended Use 14
2.0 The Human Weil-Being Index 15
2.1 About the Index 15
2.1.1 Measuring Weil-Being: Domains 17
2.1.2 Accounting for Community Priorities: Relative Importance Values 18
2.1.3 Drivers of Weil-Being: Services 19
2.1.4 Relationship-Function Equations 21
3.0 Modifying the HWBI: The Environmental Quality Index 22
3.1 Characterizing Environmental Quality: Domains 23
3.2 Calculating the EQI 24
4.0 A Case for Integrating Composite Indices 27
5.0 Approach 30
5.1 HWBI Relationship-Function Suitability and Statistical Independence of EQI 30
5.2 EQI-Modified HWBI Health Domain and HWBI Calculation Method 30
5.4 Significance Testing 31
5.5 Limitations 31
6.0 Results with Discussion 31
6.1 Test of Assumptions 31
6.2 Demonstration 34
iv
-------
7.0 Concluding Remarks 38
8.0 References 40
Appendix A. Applications and Adaptations of the HWBI Framework 43
Appendix B. HWBI Relationship-Function Equation Coefficients 55
Table of Figures
Figure ES-1. Chloropleth maps representing un-modified and EQI modified scores the HWBI
health domain and overall HWBI x
Figure ES-2. Bivariate map depicting the degree and direction of signficant change in HWBI
health domain scored produced by EQI-modification xi
Figure 1. The Human Weil-Being Index (HWBI) conceptual framework 16
Figure 2. Illustration depicting the function of relative importance values (RIVs) within the
HWBI framework. Community RIVs were contributed from Fulford, et. al., 2015 19
Figure 3. Principal component analysis concept for Environmental Quality Index. Analyses
performed for all counties and each rural-urban continuum code (RUCC) 25
Figure 4. Observed versus modeled HWBI health domain score (95% CLIMs) based on HWBI
relationship function equations 32
Figure 5. Observed versus modeled HWBI scores (95% CLIMs) based on HWBI relationship
function equations 33
Figure 6. Normality plot for unmodified HWBI health domain modeled output 33
Figure 7. Normality plot for EQI values 34
Figure 8. Observed, unmodified and EQI-modified health domain and HWBI scores plotted in
context of the remaining HWBI domain scores along a distribution gradient, for all U.S.
counties in relation to the median score of each domain set 35
Figure 9 Spatial distribution of observed, unmodified model and EQI model results for: HWBI
health domain scores (a-c) and the overall HWBI (d-f) 36
Figure 10. Bivariate chloropleth depicting county HWBI health domain scores that exceeded
the 95% confidence threshold 37
Figure A-1. Comparison of the results of using an alternative metric for the Activity
Participation indicator in the Cultural Fulfillment domain 44
Figure A-2. Process for selecting the most robust AIAN and Tribal Group data 45
Figure A-3. Large Tribal Group domain and HWBI scores for the 2000-2010 time-period 46
V
-------
Figure A-4. Map of mean decadal (2000-2010) HWBI for Puerto Rico municipios 48
Figure A-5. Comparison of mean decadal (2000-2010) HWBI for U.S. States & Puerto Rico. 48
Figure A-6. Comparison of mean decadal (2000-2010) indicator scores for Puerto Rico and
U.S. States with the highest and lowest HWBI scores 49
Figure A-7. Decision flow chart for metric adapation in children's well-being index createion. 51
Figure A-8. Children's Weil-Being Index scores fo all US Counties in 2010 53
Tables
Table 1. List of domain indices used to describe the HWBI 17
Table 2. List of service categories or indicators related to domains of the HWBI 20
Table 3. List of domain specific indices used to describe the EQI 23
Table 4. Rural-urban continuum code (RUCC) groups used for stratified calculations of the
EQI 26
Table 5. Comparison of the HWBI and EQI development approaches and the list of steps for
creating a composite index 27
Table A-1. Indicator and metric count per domain. Data representation (HWBI count/CWBI
count) 52
Table B-1. Relationship-function equation coefficients used for the HWBI domain models to
predict well-being from service categories 56
vi
-------
Acronyms and Abbreviations
CUM
Confidence Limits
EPA/USEPA
United States Environmental Protection Agency
EQI
Environmental Quality Index
HWBI
Human Weil-Being Index
NHEERL
National Health and Environmental Effects Research Laboratory
OECD
Organisation for Economic Co-operation and Development
ORD
Office of Research and Development
PCA
Principle Components Analysis
ROE
Report on the Environment
SHC
SHC Sustainable and Healthy Research Program
U.S.
United States of America
vii
-------
Acknowledgments
The authors would like to thank the peer reviewers of this report:
Kyle Buck, US EPA, National Health & Environmental Effects Research Laboratory, Gulf Ecology Division
Tarsha Eason, US EPA National Risk Management Research Laboratory, RTP North Carolina
The authors would also like to acknowledge the efforts of student contractors, Stephen F. Hafner and
Michelle D. McLaughlin, in support of creating this report.
Photo credits: Front Cover (U.S. EPA, Eric Vance); Page 18 - Native American (Microsoft.com); Page 26
-San Juan, Puerto Rico (Creative Commons-Public Domain); Page 29 - Children in field (Microsoft.com)
viii
-------
Executive Summary
The U.S. Environmental Protection Agency (EPA) Office of Research and Development's Sustainable
and Healthy Communities Research Program developed the Human Well-being Index (HWBI) as an
integrative measure of economic, social, and environmental contributions to well-being (Smith et al.,
2014a) and the Environmental Quality Index (EQI) (Lobdell et al., 2014) as an estimate of overall
environmental quality to improve our understanding of the relationship between environmental
conditions and human health. The HWBI is composed of indicators and metrics representing eight
domains of well-being: connection to nature, cultural fulfillment, education, health, leisure time, living
standards, safety and security, and social cohesion. The domains and indicators in the HWBI were
selected to provide a well-being framework that is broadly applicable to many different populations
and communities (Smith et al., 2015; Buck et al., 2017; Orlando et al., 2017). Relationship function
equations have been developed to link HWBI domains to the
provisioning of ecosystem, social and economic services (Summers
et al., 2016). The EQI uses indicators from the chemical, natural,
built and social environment in the construct of five
environmental domains: air, water, land, built and
sociodemographic. EQI can be used as an indicator of ambient
conditions/exposure in environmental health modeling and as
a covariate to adjust for ambient conditions in environmental
models. However, the EQI can be adapted for other uses by
different end users. Both indices have been demonstrated for
all U.S. counties.
This report describes an approach for modifying ORD's Human
Weil-Being Index (HWBI) to increase its utility by introducing a
composite index developed independently of the HWBI effort. Using ORD's Environmental Quality
Index (EQI), this research examines the potential for using existing indicator products in novel ways to
add a new facet of interpretability regarding the linkages between socio-ecological systems and human
health and well-being. This research demonstrates a way to combine existing composite indices for
developing a new layer of information as an extended diagnostic of well-being conditions. HWBI
adaptations and applications are also presented, highlighting population-specific HWBI research. These
adaptations could serve as future EQI-modified HWBI use cases.
ss a:
:::s: :::::
¦ ¦¦¦ HUB tv
Analyses were completed to determine the statistical suitability of synthesizing the HWBI and EQI, to
confirm variable independence and fit within the relationship-equation construct for the HWBI domain
of health. A series of relationship-function equations were developed, linking aspects of the HWBI
(domains) to select economic, ecosystem and social services (Summers et al., 2016). The EQI was
introduced as a modifier within the HWBI health model structure which served as the conduit to affect
an overall EQI-adjusted HWBI. A standard transformation of the modeled HWBI health values made
IX
-------
combining the two indices relatively simple, requiring no further adjustments to either composite
measure to accommodate the integration. Standardized HWBI health values were adjusted by the EQJ
directly then reconstituted to provide a score comparable to the remaining HWBI domains. An
adjusted overall HWBI was calculated for each county to reflect the EQI modification.
Generally, the EQI-adjusted health domain showed little impact on the overall HWBI score. However,
changes in the spatial patterns of modeled health scores were clearly delineated (Figure ES -1). Results
identified approximately 28% of U.S. county HWBI health scores exhibiting significant changes as a
result of the EQI modification, both positive or negative (Figure ES - 2). These results suggest that the
addition of the EQI as a modifying factorto the HWBI relationship-function equation for health
provides an additional layer of diagnostics for understanding well-being.
HWBI Health Domain Model EQI-Adjusted HWBI Health Domain Model
Score
Lower
Higher
HWBI Model
EQI-Adjusted HWBI Model
Figure ES -1. Chloropleth maps representing un-modified and EQl-modified model scores for HWBI Health
Domain (a-b) and Overall HWBI modeled scores (c-d).
x
-------
Significant Change in Modeled HWBI Health Domain Scores
After EQI-Modification
Decrease Increase
Figure ES - 2. Map depicting the degree and direction of significant change (95% CLIM) in HWBI health domain
scores produced by EQI-modified relationship-function equations developed for the HWBI.
Additionally, the integration of the HWBI and EQI provides SHC with an empirical linkage between the
confluence of multiple environmental quality measures and overall human health. This research
contributes to EPA's Sustainable and Healthy Communities Research Program's (USEPA, 2015)
objective of developing research, data, and tools to expand the capabilities of communities to consider
the social, economic, and environmental impacts of decision alternatives on community well-being.
xi
-------
1.0 Introduction
Increasingly, the many valuable attributes that ecosystems provide are being recognized as important
contributors to economic and social vitality. From creating jobs to building greener infrastructure,
many considerations are needed to yield sustainable decision outcomes. As the needs and demands of
a community change, it is often difficult for community stakeholders to consider environmental effects
along with the economic and social impacts associated with development and improvement decisions,
equitably. Sustainability has become a guiding principle in the pursuit of economic growth,
environmental quality, and social equity across communities of all sizes. Sustainability, as a paradigm,
requires finding ways for community stakeholders to assess and track it (USEPA, 2016).
Indicators offer one approach for providing meaningful sustainability measures to inform decision-
making in federal, state and local governing sectors. Using indicators as a means of gauging various
aspects of sustainability is a common practice, as evidenced by the plethora of published indicator
efforts related to the subject (Smith et al., 2012 and 2013b).
Composite indicators, in particular, are gaining popularity as useful
tools in policy analysis and public communication (USEPA,
2016). They offer simple and effective ways to describe
complex, often abstract concepts across a wide-range of fields
(e.g., ecology, sustainability, resilience, economy and society)
(JRC, 2008). When well-constructed, composite indicators
convey a synergistic message to help engage a broad
spectrum of stakeholders in conversations that strengthen
interpretation of indicator results.
For the United States (U.S.) Environmental Protection Agency
(EPA), sustainability is one of four cross-organizational strategies
which emphasizes: (1) advancing sustainability science, indicators, and
tools and (2) using system-based approaches that account for linkages among different environmental
systems (USEPA, 2014). One area of focus within EPA's Office of Research and Development (ORD)
centers on providing data, methods, indicators, models, and tools that can be used to develop
approaches for assessing aspects of community sustainability (USEPA, 2015). As part of ORD's
Sustainable and Healthy Communities (SHC) Research Program, several research efforts identify and
develop environmental indicators and indices to support these goals (Pickard et al., 2015; USEPA, 2008;
Summers et al., 2017). Individually, indicators offer an interpretation of measures in a related scope or
dimension (e.g., economics, living standards). Composite indices summarize indicators to describe a
complete concept (e.g., sustainability, resilience, well-being, environmental quality), quantitatively or
qualitatively. Composite indices provide context for related indicators and are typically more reflective
of actual conditions associated with a concept. Additionally, constituent components of a composite
index are structurally related. Categories are generally grouped together to represent a unique
characteristic of the composite. The relational structure connecting indicators to an index allows for a
Sustainability Goal
Actual
Target
12
-------
The Value of
Composite Indices
and Indicators
The power of composite
indices and indicators is their
ability to synthesize a large
amount of information to
characterize complex systems.
They are multi-dimensional and
independent of time and place-
relevance is limited only by the
availability of data.
The science of developing
composite indicators and indices
is rigorous. Many, sometimes
hundreds, of candidate metrics
represent the data foundation.
At the heart of indicator research
is the index framework. The
framework is the roadmap for
calculating an index, protecting
its integrity and ensuring the
reproducibility of the measures.
Sensitivity, uncertainty and
fidelity analyses are the proving
grounds for the development
approach.
From the literature review to
indicator selection to calculation
methods, the steps for creating
an index are many. The results:
composite measures that serve
many roles—endpoint values,
modifiers, mediators,
interpreters and communicators.
deeper evaluation regarding the contributions of the various
characteristics (as indicators) to composite measure (as an
index) to help identify areas of potential action. Composite
indices are gaining favor in community decision-making
because they are easily understood by technical and non-
technical audiences alike. However, these measures can be
misinterpreted when viewed out of context. To be most
effective, approaches that combine composite indices in
meaningful ways are needed to provide more clarity for
improved interpretation. However, it is critical that each
index selected for integration have clearly defined purpose
and development approach (JRC, 2008).
It is understood that observational data produce more
precise statistical measures. However, assessing an abstract
concept such as well-being, requires a large number of
metrics. The trends, relationships and patterns produced by
models and other statistical analyses can be lost when using
a large volume of raw measurement data. In many fields, it is
desirable to reduce data to include only those measures that
explain the greatest proportion of observed variability. In
community-focused research, measures can be sensitive to
temporal and socioecological shifts on relatively small
scales—what measures are not significant contributors today
may be more significant tomorrow. An equally important
consideration is the possibility that data reduction methods
based on variance accounting alone may eliminate important
indirect relationships that support community characteristics
that people value (Summers et al., 2016).
Combining composite indices has the potential to produce
more holistic information than a single index approach,
whether a reduction or summarization calculation method is
used. This practice is not new. Recognized composite indices
such as the Gross National Income indicator (World
Bank.org) and the Composite Leading Indicator (OECD,
2017), use at least one composite measure to inform the
final index calculation. Although interpretation pitfalls exist,
composite indicators and indices are still one of the more
robust mechanisms for representing data to potentially
13
-------
influence policy (Booysen, 2002). Prospects for synthesizing composite measures to provide more
contextual interpretations of the linkages across complex concepts is compelling, particularly for use in
socio-ecological and sustainability fields.
The purpose of this report is to describe an approach to modify ORD's Human Weil-Being Index
(HWBI), a holistic composite measure, to increase its utility by using another existing composite index,
ORD's Environmental Quality Index (EQI) (Lobdell et al., 2014). The main interest of this research was
three-fold:
• Identify an approach to integrate composite measures that is simple and reproducible;
• Test the capability of the HWBI health model and overall HWBI to reflect directional changes
when using EQI as a modifier; and
• Test for statistical significance should differences manifest.
The authors posit that the EQI, a composite index characterizing overall ambient environmental
conditions, can effectively function as a qualifying variant within the HWBI framework to illustrate and
quantify the impact of socio-ecological conditions on an indicator of overall human health and the
broader measure of well-being. This report presents a brief description about the HWBI and EQI, the
steps for composing a new measure from these indices and a demonstration of results. Additional
published research efforts feature population-specific case studies that explore the relevance of the
HWBI approach for characterizing the well-being of large tribal groups and children of the U.S. and the
U.S. Commonwealth of Puerto Rico. Research highlights describing HWBI modification decisions and
results for each study are available in Appendix A. These adaptations could serve as future
demonstrations of the EQI-modified HWBI approach.
1.1 Intended Use
The information presented in this report represents a first-step approach intended for consideration in
the development of decision-support and communication tools. The demonstration features a method
for combining composite indices to introduce new topic or concept perspectives that may not be
otherwise available. Demonstration results may be used as testing or baseline information for
characterizing the potential influence of natural, built and socio-demographic environments on overall
health and, by extension, well-being. Both the HWBI and EQI provide web-services to promote the use
of these indices, related indicators and the HWBI relationship-function equations when developing
mobile and desktop software applications.
14
-------
2.0 The Human Weil-Being Index
2.1 About the Index
Communities face multiple, often conflicting, decision priorities. Many of these priorities are abstract
(e.g., sustainability, resilience) and frequently lack adequate measures to inform decision-making. To
help fill this information gap, EPA developed the Human Well-being Index (HWBI) (Smith et al., 2012
and 2013a). The HWBI is an index that characterizes well-being based on metrics derived from existing
cultural, economic, and social data. The index is a quantified score (0-100 scale), developed as an
endpoint measure for characterizing well-being outcomes that are responsive to changes in the
provisioning of economic, social and ecological services. The HWBI is part of a larger conceptual
framework that serves as a roadmap for depicting the flow of a common core of community supplied
services that influence the quality of people's lives (Smith et al., 2014a). The HWBI conceptual
framework (Figure 1) is comprised of two groups of indicators, service and well-being, and a suite of
indicator modifiers representing the linkages between services and aspects of human well-being,
common across all communities (Smith et al., 2014b). A nested-indicator design guides the
summarization of data to various aggregate levels until a final composite value is achieved for service
categories, well-being domains and final index. Collectively, the HWBI framework and components
describe an approach for calculating a holistic and quantified human well-being measure to inform
policy decision-making.
COMPONENTS OF THE HUMAN WELL-BEING INDEX
An INDEX is an interpretable and synergistic value or category describing the nature, condition or trend
of a multidimensional concept. An index can be an endpoint or final value as well as one of several values
used to create what is called a composite index.
A DOMAIN represents a summary grouping of characteristics that is based on one or more indicators and
represents a major component of a composite index. A domain and sub-index generally refer to the same
level of information.
An INDICATOR is an interpretable value describing a trend or status for a specific feature or
characteristic. An indicator may be comprised of one or more metrics.
A METRIC is a measurable or observable value - typically referred to as "the data".
15
-------
Quality and Quantity of Capital
Built
Human
Natural ] [ Social
Ecosystem
Air Quality
- Food, Fiber and
Fuel Provisioning
¦ Greenspace
- Water Quality
Water Quantity
Goods and Services
Social
- Activism
-Communication
- Community
Faith-Based Initiatives
- Education
-Emergency
Preparedness
- Family Services
• Healthcare
• Justice
• Labor
Public Works
Good Governance
Capital Investment
Consumption
Employment
Finance
Innovation
Production
Re-distribution
Freedom of Choice and
Opportunity
Domains of Well-being
Connection to Nature
Living Standards
Social Cohesion
Well-being Elements
¦
Societal
) 1
Economic
u
Human Well-being Index
o
o
3
o
ST
£
2.
!
3
IQ
Figure 1. The Human Well-Being Index (HWB1) conceptual framework.
16
-------
2.1.1 Measuring Weil-Being: Domains
Domains are sub-indices that act as proxy measures representing various states of the human
condition and quality of life (Tabl
e 1). Derived from indicators based on standardized data or metrics,
these domain values are summarized to quantify overall well-being. The domains represent
characteristics common across al
1 communities that not only influence people's well-being, but are also
tangibly sensitive to environmental change.
Smith et a I. (2014b) describes the final approach for calculating the HWBI. There are four steps for calculating
the HWBI:
• Indicator scores are calculated as population weighted averages of related standardized metric
values.
• Domain scores are obtained by averaging indicator scores related to a specific domain.
• Relative importance values (RIVs) are optional factors that may be included in HWBI
calculations to represent stakeholder priorities associated with well-being domains.
• The HWBI is calculated as the geometric mean of equally or unequally weighted domain scores.
Table 1. List of domain indices used to describe the HWBI.
Domain
Description
Connection to Nature
Describes how people feel about nature. It is measured by
people's perception of nature and how it affects them.
Cultural Fulfillment
Describes people's cultural involvement. Measures include how
often people participate in the arts and spiritual activities.
Education
Covers basic skills in reading, math and science. Measures of
student safety and health are also included.
Health
Characterizes people's involvement in healthy behaviors,
prevalence of illness, access to healthcare, mortality and life
expectancy.
Leisure Time
Describes how time is spent including: employment, care for
seniors and activities that people partake in for personal
enjoyment. Measures represent work-life balance.
Living Standards
Contains information about lifestyles. It includes measures of
basic necessities, wealth and income.
Safety and Security
Covers information about perceived safety, actual safety and
potential for danger.
Social Cohesion
Describes people's connection to each other and their
community through measures of involvement in family, civic
engagement, and the community as a whole.
17
-------
The HWBI structure allows for metric substitutions while holding the remaining parts of the framework
constant. This design feature allows for the inclusion of data that more closely reflects characteristics
in specific use case applications yet maintains the integrity of the index, allowing for comparisons of
domain and index scores across various spatial, temporal and population scales. Smith etal. (2012 and
2013a) presents a thorough description regarding the HWBI conceptual framework, HWBI indicator
selection and data sources.
The publication, Evaluating the Transferability of a US Human Well-Being Index (HWBI) Framework
to Native-American Populations (Smith et al., 2015), presents the applicability and integrity of the
HWBI framework using metrics scaled to assess well-being for American Indian Alaska Native (AIAN)
and large tribal populations. Potential modifications needed to produce reasonably defensible well-
being assessments were identified and HWBIs were calculated for the AIAN population and large tribal
groups for the time-period covering 2000-2010. Greater than 80% of the data available for a national
AIAN assessment were specific to the target population, while the remaining data were derived from
the general U.S. population. Despite the utilization of non-target data, the AIAN well-being signature
could still be differentiated from the U.S. HWBI, indicating that the HWBI approach is transferable. As
designed, the framework is intended to be used for a variety of spatial scales and demographic
groups; however, the degree to which the structure can be utilized is dependent upon the availability
and quantity of quality data.
See the complete research highlight (pg. 44).
2.1.2 Accounting for Community Priorities: Relative Importance Values
Well-being as a construct often reflects the collective perception of a population or community. The
HWBI framework is designed to accommodate differing viewpoints about the relative importance of
each of the eight domains. Externally supplied priority weighting factors can be applied across the
domains before generating the final index. By weighting how domains contribute to well-being, the
HWBI can better reflect community well-being priorities or values structure. This feature allows the
index to show how a community perceives well-being in terms of possible magnitudes of change.
Information describing the uses of RIVs and HWBI demonstration is available in Smith, et al. (2013b).
Figure 2 provides an example of RIVs utilized in a real community case (Fulford, et al. 2015).
Modifying the HWBI For Characterizing
Well-Being in Native-American
Populations
18
-------
Woodbine, Iowa (Harrison County)
Equal Domain Weighting
Adjusted Domain Weighting
(Percent of Change)
-1%, Social
-5%,
Connection
Cohesion
to Mature
0%, Safety
and
Security
HWBI
55.6
HWBI
57.4
-2%, Living
Standards
Leisure
-5%, Cultural
11%,
Education
Time
5%, Health
Figure 2. Illustration depicting the function of relative importance values (RIVs) within the HWBI
framework. Community RIVs were contributed from Fulford, et. al., 2015.
2.1.3 Drivers of Well-Being: Services
Implementing a community decision typically requires a new investment or redistribution of available
resources. The availability and flow of these resources serve many purposes in a community—
everything from health to employment to clean water. This inventory of resources is often represented
as economic, ecosystem and social services. Within the HWBI framework, twenty-two major service
categories are used to demonstrate the relationships between ecosystem services and human well-
being in the context of economic and social services (Table 2). Service indicators or categories are
derived from population-weighted average metrics associated with each service and represent major
endpoint service functions within communities (e.g. clean air, capital investment, access to healthcare.
Details describing data sources, metric selection and indicator calculations are available in Smith, et.
al., 2014b.
19
-------
Table 2. List of service categories or indicators related to domains of the HWBI.
Service Group Service Indicator Description
E
qj
Greenspace
Water Quality
Natural areas that allow for recreation and aesthetics
(including aquatic spaces)
Services that remove pollutants that enter waterways
Water Quantity
Services that produce, preserve and renew water
resources
20
-------
Service Group
Service Indicator
Description (Continued)
Services provided by individuals or groups acting to
Activism
bring about social, political, economic or environmental
change
Communication
Delivery of information to promote public awareness
Community and
Faith-Based
Initiatives
Spiritual and civic outreach and activities that promote
the betterment of a community
Education
Services provided to improve learning experiences and
to allow for equitable educational opportunities
in
-------
multivariate relationship functions to illustrate likely positive or negative impacts on future well-being
as a results of decision scenarios that disrupt the flow or inventory of select services.
The relationship-function equations are used to evaluate inherent connections between well-being and
the quality and quantity of selected goods and services (Summers et al., 2014; Smith et al., 2014a).
Drawing from similar approaches for developing economic and ecological production functions (Bruins
et al., 2012), the HWBI relationship-function model produces forecasted direction of change in the
domains related to human well-being. The functional relationship equations for each domain were
determined based on results stemming from a step-wise regression process to identify main effects
and primary pair-wise interactions among select economic, environmental and social service indicators
quantified as scores on a 0-100 scale. A brief discussion of this stepwise method follows:
• The method used regression models to identify predictive variables based on adjusted R2 and
sequenced F-tests. The initial model begins with no variables selected and candidate model
variables are evaluated on a pairwise basis including those parameters already selected.
• The f-statistic significance level was the primary determinate for inclusion or exclusion in the
model. After each pass in the stepwise process, the model was refitted with the held variables.
• Entry and exit criteria used for keeping or eliminating effects were based on f-value significance
levels of 0.15 and 0.10, respectively.
• Main effect and two-way interaction terms selected during the step-wise process that exhibited
partial R2 > 0.02 were reviewed for inclusion in the final model.
• Using results from the review of the stepwise selections, a final, multivariate model was created
for each well-being domain by incorporating constructed (fixed) main effects plus interaction
terms identified as significantly improving the explanatory capability of the model.
For each domain, a generalized relationship function equation is as follows:
HWd = f(Se, Ss, Sec))
where a human well-being domain (HWd) is estimated as a function (f) of the combined effects of
economic (Se), social (Ss), and ecosystem (Sec) services. The final HWBI can be calculated from these
individual domain coefficients to present a predicted index score. The primary objective of the HWBI
relationship-function equations is to demonstrate the responsive nature of the HWBI to changes in
services as well as simplifying the integration of HWBI concepts and components in decision-support
tools (Ignatius et al., 2016; Harwell, 2017). Complete details describing relationship-function equation
development methods and demonstration results are available in Summers et al., 2016.
3.0 Modifying the HWBI: The Environmental Quality Index
Data selection, standardization methods and overall intent are decisions that dictate the viewpoint of
indicators and indices. This perspective or presentation of composite measures can exacerbate the
22
-------
potential for misinterpretation or over-simplification of meaning (Booysen, 2002; JRC, 2008). While no
single summary measure can address all possible caveats, combining composite measures can further
reinforce the intended interpretation of results by accounting for effects that were originally outside
the parameters of the original index. For this demonstration, a composite index qualifier was chosen to
modify the HWBI relationship-function equations as a way to test the adaptability of HWBI models as
well as create a new integrated measure.
The Environmental Quality Index (EQI) was developed to address a limitation inherent in current
methods in environmental health research, which tend to focus on a single adverse environmental
exposure at time. Well-designed environmental health studies are expensive and time consuming to
conduct. This often manifests in a research trade-off dilemma—either engage only a few participants
to collect high resolution data or collect data from a larger number of participants at the expense of
detail. This trade-off makes it difficult to account for the plethora of co-occurring environmental
conditions to which study participants may be exposed in addition to the main exposures of interest.
Based on a well-defined framework, the EQI represents a more holistic way to account for overall
environmental quality to inform human health and environmental effects studies. The following
sections briefly describe the approach associated with calculating the EQI (Messer, 2014; Lobdell,
2014).
3.1 Characterizing Environmental Quality: Domains
The EQI is described by domains representing five environmental dimensions: air, land, water, built and
socio-demographic. EPA's Report on the Environment (USEPA, 2008) served as the starting point for
domain identification. Domain identification was further bolstered by expert consultation and a
literature review on environmental exposures and adverse health outcomes. Collectively, these
domains represent an overall measure of ambient environmental exposure conditions. Domain
variables, or constructs, were identified to describe individual domains. These constructs provided the
organizational structure for summarizing and standardizing metrics or data used to quantify the EQI.
Table 3 provides a brief description of the EQI domains, attendant constructs and the primary
environment-to-human health impact association.
Table 3. List of domain specific indices used to describe the EQI.
Domain Description
Air The air domain represents the ambient air environment. Two
traditional air pollutant constructs were considered: (1) criteria
air pollutants and (2) hazardous air pollutants (HAPs). Health
effects linked to air pollutants include death, cancer, heart
disease, respiratory disease, birth outcomes, and neurologic
disorders.
23
-------
Domain
Description (Continued)
Represents the physical environment not covered by air or water. Five
constructs were considered to represent land environmental quality: (1)
agricultural environment, (2) pesticides, (3) facilities, (4) soil
contaminants, and (5) radon potential. Health effects linked to land
constructs include cancer, birth outcomes, birth defects, and asthma.
Represents the overall water environment. Seven constructs were
considered to represent water quality: (1) overall water quality, (2)
general water contamination, (3) recreational water quality, (4)
domestic use, (5) atmospheric deposition, (6) drought, and (7) chemical
contamination. Several studies have demonstrated the association
between water contaminants and pathogens and health outcomes.
Contains five constructs: (1) traffic-related environment, (2) transit
participation and access, (3) pedestrian safety, (4) the various business
environments (such as the food, recreation, health care, and
educational environments), and (5) public housing. Each of these
constructs has both direct and indirect influences human health
including behavioral, physiological and psychological factors.
Socio-demographic Considers the association between socio-demographics and a broad
range of human health factors which are grouped (1) socio-economic
and (2) crime constructs.
3.2 Calculating the EQI
Using data from relevant existing sources, principle component analyses (PCA) were used to determine
the weighting factors of metrics used to calculate the domain-specific indices and final the EQI. PCA, as
a tool for indicator development, is well established in the literature (JRC, 2008; Singh et al., 2009). For
the EQI, the data acquired spanned multiple years for all counties and hundreds of individual measures
(e.g. air pollutants, water and land contaminants, demographics, crime statistics). Each data point was
standardized to reflect the appropriate population and spatial area adjustments. The volume of metrics
used to inform the construction of the EQI necessitated the use of robust methods for identifying the
best information for creating variable constructs (Figure 3).
Land
Water
Built
24
-------
Principal component
analysis (PCA) reduced
multiple variables into
domain-specific
indices for each RUCC
strata and the nation
overall
Domain-specific
indices combined
using PCA to create
EQI for each RUCC
strata and overall
Water
variables
Air
variables
EQI
Built
Indices
Land
Indices
Water
Indices
Socio-
demographic
Indices
Air
Indices
RUCC1 = metropolitan-urbanized
RUCC3 -loss urbanized
RUCC4 =thinly populated
OVERALL
Land I Built
variables I variables
Socio-
demographic
variables
Figure 3. Principal component analysis concept for Environmental Quality Index. Analyses performed for all
counties and each rural-urban continuum code (RUCC).
Domain variable constructs were identified based on the first principle component, representing the
greatest accounting of the total variability in component measures. Individual variable loadings
resulting from the analyses provided the variance "contribution" of each variable within a domain. The
loading factor associated with each variable was multiplied by the variable mean across spatial and
temporal units. Once quantified, variable constructs were summed to produce a domain-specific
county-level dataset (e.g., air, land, water, built and socio-demographic). Each summed domain value
was divided by the square of the PCA eigenvalue to re-scale the value on a zero-point mean
distribution to create the final domain-specific index.
Each domain-specific index was subsequently included in a second PCA procedure, from which the first
principal component served as the basis for the last step in the EQI calculation. Pearson's product
moment correlations were used to eliminate dependent relationships within and between indices and
other county-level variables based on a 0.7 cut-off threshold. The complete process, from PCAs to
correlation analyses, was completed for the Nation overall and four stratified national populations
based on summarized Rural-Urban Continuum Codes (USDA, 2003) groups (Table 4). The results from
these activities produced two different suites of county-scale EQI measures for evaluating health
outcomes. Several health studies, featuring the EQI as an independent measure of ambient
25
-------
conditions/exposures or as a covariate to account for ambient conditions in environmental public
health models, have been published (Jagai et al., 2015; Rappazzo et al., 2015; Grabich et al., 2015).
More complete details about the EQJ development methods are described in (Lobdell et al., 2014).
Table 4. Rural-urban continuum code (RUCC) groups used for stratified calculations of the EQI.
RUCC Group Description
RUCC1
Metropolitan urbanized
RUCC2
Non-metro urbanized
RUCC3
Less urbanized
RUCC4
Thinly populated
Adapting the HWBI for Use Outside the
Fifty U.S. States
The publication, Technical Guidance for Constructing a Human Weil-Being Index (HWBI): A Puerto
Rico Example (Orlando et al., 2017), describes an approach used to adapt the US Human Weil-Being
Index (HWBI) to quantify human well-being for US Commonwealth of Puerto Rico. As a territory of the
U.S., Puerto Rico operates simultaneously as a state-equivalent and as an independent entity. As a
culturally distinct and geographically isolated population, Puerto Rico presented a unique opportunity
for an HWBI application (HWBI-PR). Metric substitutions, data selection and calculated HWBI-PR
measures were compared to U.S. mainland values to evaluate differences. Additionally, the published
adaptation of the HWBI for Puerto Rico provides an example of how the HWBI can be adapted to
different communities and technical guidance on processing data and calculating index using R.
See the complete research highlight (pg. 47).
26
-------
4.0 A Case for Integrating Composite Indices
The HWBI, as an endpoint measure, is generally accepted as a robust composite measure and has
informed research in fields such as ecological economics (Costanza et al., 2014; Kubiszewski, 2013),
environmental policy (Breslow et al., 2016) and public health (Jennings et al., 2016). The index is
reflective of a dynamic and multi-faceted environment in terms of people and environmental changes
that affect their lives. HWBI demonstration results can serve as a foundation in the development of
decision-support tools (Harwell, 2017). Although highly generalized in nature, the quantified
relationship-function model makes the HWBI truly portable (Ignatius et al., 2016) and unique among
other ORD indicator efforts. The relationship-function equations extend the life of the HWBI beyond its
demonstration roots to instill new thoughts and discussions about promoting sustainable decision
outcomes. The integration of the EQI potentially adds another layer of dimensionality to the modeled
HWBI output. Conceptually, the introduction of an external factor to the HWBI framework introduces
conditions not captured in the economic, ecological and social inputs that drive the HWBI relationship-
function model. The combined effect of both indices as factors in the HWBI model could strengthen
interpretation of results that paint a more complete characterization of well-being. Approaches used to
quantify and calculate the two indices were similarly robust, making the HWBI and EQI good candidate
indices for synthesis. Table 5 highlights the strategies or activities used for constructing the HWBI and
EQI compared with the ten development considerations recommended by the Joint Research Centre-
European Commission (2008):
Table 5. Comparison of the HWBI and EQI development approaches and the list of steps for creating a composite
index outlined in "Handbook on constructing composite indicators: Methodology and user guide" (JRC, 2008).
Recommended Step
HWBI/EQI Activity
Variant
Develop theoretical
framework
Frameworks served as road maps
for identifying and quantifying
indicators and indices.
Select Data
Impute Missing Data
Metrics and data were selected
based on analytical soundness,
measurability, spatial coverage,
relevance and relationship to
each other. Proxy measures were
used when data were sparse.
Temporal and spatial imputation
methods (e.g. spatial hierarchy,
single pass forward temporal
imputation, spatial interpolation)
were used to fill data gaps.
The HWBI winsorized extreme
outliers (4x standard deviation of
interquartile range).
27
-------
Recommended Step
HWBI/EQI Activity
Variant (Continued)
Perform Multivariate
Analysis
For the HWBI, Cronbach's o and
logistic, quantile and logistic
regression analyses were
conducted.
Normalize
Data were standardize using the
appropriate population or spatial
stands. All data were normalized
to create normal distribution
metric base for indicators.
The HWBI used a minimum
value/maximum value
normalization method. The EQI
domain variables were normalized
to reflect a 0-mean and 1 standard
deviation.
Weight and aggregate
Data were aggregated based on
framework design.
The EQI used a PCA-based
calculation method with loading
factors as weights. The HWBI used
a mean summary approach with
equally weighted measures. The
HWBI framework does allows for
external weighting contribution
(e.g., RIVs).
Determine robustness and
sensitivity
For the HWBI, Monte-Carlo
uncertainty and sensitivity
analysis were completed to
identify structural and precision
strengths and weaknesses of the
index
Promote transparency
Final indices can be decomposed
into contributing components
Establish linkages
Both HWBI and EQI established
linkages to highlight the relevancy
and value of composite indices
using a variety of methods
Fidelity and response analyses
were conducted using the HWBI.
Visualize and/or Present
Results
Both indices represent
"characterizing" measures. HWBI
and EQI are generally presented
as spatial distributions (e.g. maps)
or synoptic summaries such as
column and pie charts or aster
plots.
HWBI results are used in decision-
support tools. Results are offered
in dashboard format accompanied
by information to put results in
context.
The HWBI and EQI differ in the details of the conceptual frameworks. The HWBI and service categories
represent two ends of a responsive framework design connected by statistical linkages. Based on
"means-ends" heuristics (Newel and Simon, 1972; Hoppe, 2017), the HWBI framework casts services as
28
-------
the means by which change is introduced to improve or sustain aspects of well-being while the HWBI
domains are the response measures or ends that reflect that change. The generalized relationship-
function equations are constant and serve as the conduit between the two sets of measures. Indicators
were chosen and sequestered as either services- or HWBI-related based on the nature of the data—
influencer or responder. All values within the HWBI framework are normalized scores ranging from 0-
100 that suggest better conditions when scores increase. Measures such as the HWBI tend to be easy
to interpret and reproduce, making it highly suitable as an endpoint.
Where HWBI distinguishes between means and ends indicators, the EQI does not. The EQI aims to
collectively represent the totality of known environmental stressors on human health. Indicators
(domain variables) are organized into representative environmental dimensions to inform domain-
specific and subsequent EQI calculations. EQI and associated domains values are distance-from-mean
zero scores, scaled from zero to positive or negative infinity rather than 0-100. Since the EQI is a
descriptor for the potential "risk" of environmental exposures, the descending sequence of scores
represent increasingly better environmental conditions (less adverse exposure risk) that is inversely
related to the HWBI. As a modifying value, the EQI is well-suited.
Modifying the HWBI to Characterize
Children's Weil-Being for the U.S.
The publication, Application of the Human Well-Being Index to Sensitive Population Divisions: A
Children's Well-Being Index Development (Buck et a I,, 2017), presents an adaption of the HWBI for
child populations to test the applicability of the index framework to specific community enclaves.
First, an extensive literature review was completed to ensure the theoretical integrity of metric and
indicator substitutions from the original HWBI framework. Metric data were then collected, refined,
imputed where necessary and evaluated to confirm temporal and spatial availability. Using the same
domains and contextually similar indicators as the original HWBI, a Children's Weil-Being Index (CWBI)
was calculated for the population under age 18 across all US counties for 2011. Implications of this
research point to an effective, holistic and nationally consistent well-being measure for a specific
population that can be tracked overtime. Similarly, there is great potential for the application of the
original HWBI method to other statistical population segments within the greater US population.
These adaptations could help identify and close gaps in equity of resource distribution amongthese
groups.
See the complete research highlight (pg. 51).
29
-------
5.0 Approach
For this effort, both the HWBI and EQI were treated as "found" research—HWBI research supplied the
service measures and relationship-function equations while the EQI results were treated as HWBI
model modifiers. The relationship models for the eight HWBI domains were trained and tested using
11 years of summarized economic, social and ecosystem data and HWBI domain values (2000-2010).
The EQI was derived from a weighted aggregate of pollutant, chemical, built, socio-demographic and
other similar data covering the years 2000-2005. National-scale county and county-equivalent
indicators were chosen as the common spatial unit for this demonstration. For the purposes of this
report, the term "county" encompasses jurisdictional areas labeled county, parish or borough.
5.1 HWBI Relationship-Function Suitability and Statistical Independence of EQI
Since the HWBI relationship-function equations were originally developed using state-level indicators,
observed values were fitted to model outputs to examine fit. Determination of fit was used to test
suitability based on 95% confidence limits and relative percent difference estimates. Correlation and
regression analyses were conducted to confirm variable independence between the EQI and the
relationship-equation constructs for HWBI domain of health and overall composite index. A complete
listing of all HWBI relationship-function equations are available in Appendix B.
5.2 EQI-Modified HWBI Health Domain and HWBI Calculation Method
The EQI is intended to be used as an explanatory variable in human health studies which is a
characteristic more closely aligned with the HBWI service indicators. Since the objective was to
introduce an HWBI modification using the relationship-function equations, the EQI was used as an
adjustment factor applied to the predicted values produced from the HWBI health domain model.
While several methods of EQI-integration were considered (i.e., ordinal regression and multi-level
modeling using EQI as a covariate), simplicity was the final determinant of the approach selected.
While other options may have offered more statistically robust solutions, an easy to understand and
straightforward method will help facilitate the use of research approach on a broader scale. The EQI-
modification described here required the least manipulation of the existing measures while
maintaining the original intent of both indices. The HWBI modeled health domain was modified using
the EQI in the following manner:
• The HWBI health score estimates stemming from the relationship equations were z-score
transformed to standardized modeled output to the EQI scale (zero-mean center point).
• The EQI was subtracted from the transformed health score estimates without further
transformations or modifications. Because the EQI is an inverted scale, subtraction increased a
health domain z-score when the characterization of environmental conditions indicated less
potential for adverse human exposures and decreased it when greater.
• The adjusted health z-scores were reverse calculated—new health value = mean of population
30
-------
of original health score + (z-score * standard deviation—to provide a score comparable to the
remaining HWBI domains.
• Adjusted overall HWBI values re-calculated to reflect the adjusted health domains score-
geometric mean of the 7 original HWBI domain scores and new EQI-adjusted health domain
value.
Where HWBI results are provided, the scores were derived using the domain value estimates from
each of the HWBI services-to-domain relationship-function equations. This step ensured comparability
(model output to modified model output). For this effort, HWBI domains were weighted equally to
more effectively account for any differences in modeled outputs as result of the EQI modification.
5.4 Significance Testing
Patterns of significant differences between original modeled output and EQI modified outputs were
identified using 95% confidence limits of new health domain measures. Z-tests were run to evaluate
mean differences among original and modified health domains and HWBI.
5.5 Limitations
Error is inherent in both empirical and calculated measurements. Secondary-data were used to
calculate the HWBI and EQI. Although this means most error would have been introduced from sources
that are outside of our control, steps were taken to maximize the quality of the indicators produced by
these research efforts. Data management strategies, data review processes, standardization and
imputation methods, and model validation were some of the ways quality was maintained for HWBI
and EQI indicators used this research.
The relationship-function equations used to model the health domain and the HWBI represent a
generalized characterization of the directionality of change in HWBI measures (e.g. increasing or
decreasing) with an overall 95% level of confidence. The model equations were not adjusted to
account for the addition of the EQI as a modifier. The intent of this research was to use an existing
suite of composite measures "as is" and to determine if the HWBI framework and relationship
function-equations were capable of handling modifications without extensive enhancements or
overhaul.
6.0 Results with Discussion
6.1 Test of Assumptions
It was important to establish that the relationship-function equations adequately represented
observed measures for the HWBI and health domain endpoints at county-scale. Collectively, the weight
of evidence from the following evaluations indicated a reasonable confidence in using the EQI and
HWBI for this research effort. A scatter plot of observed health and HWBI values over modeled values
showed good fit using 95% confidence limits (Figures 4 and 5). Observed values that fell outside the
31
-------
confidence limits for model health and HWBI scores were 25.4% and 14.0%, respectively.
0° A e O
E
o
Q
£
Observed O
Modeled
95% CUM
Cumulative Number of Counties
Figure 4. Observed versus modeled HWBI health domain score (95% CLIMs); based on HWBI
relationship function equations.
100"
60-
40-
Observed O
Modeled
95% CUM
1000 2000
Cumulative Number of Counties
32
-------
Figure 5. Observed versus modeled HWBI scores (95% CLIMs); based on HWBI relationship function
equations.
Temporal scale differences associated with metric data used for the HWBI (2000-2010) and the EQI
(2000-2005) were not specifically accounted for since the HWBI models served as the basis of the
evaluation rather than the observed values. However, both HWBI and EQI were calculated based on
metric averages for specific time spans thus reducing the impact of variability introduced by data
collected in years that did not overlap. Of the 3143 counties, parishes and boroughs listed in the 2010
U.S. Census, 3139 could be matched between HWBI and EQI. Independent cities and boundary changes
in Alaska contributed to the < 1% of counties not represented in this analysis.
Modeled HWBI health score and EQI values were tested for normality. Both data sets exhibited
deviations from normality. The HWBI modeled health domain is slightly right skewed with light tailing
while the EQI is slightly left skewed with similar tailing (Figures 6 and 7). The most likely cause of these
deviations relate to imputation or interpolation processing used to fill information gaps where data
were sparse, particularly in counties with low population densities. While not perfectly normal, the
data sets were generally normative and suitable for this demonstration.
Normal Q-Q Plot
Modeled HWBI Health Domain Scores
o
CO
in
in
-3
-2
1
0
1
2
3
Theoretical Quantiles
Figure 6. Normality plot for unmodified HWBI health domain modeled output.
33
-------
Normal Q-Q Plot
EQI Values
o
CM
I
I
CD
I
3
-2
1
0
1
2
3
Theoretical Quantiles
Figure 7. Normality plot for EQI values.
Since existing HWBI relationship-function equations were central to this demonstration, it was
necessary to establish that the EQI, as a potential model modifier, was not a redundant
characterization of either the HWBI or HWBI health domain. Pearson's Product Moment correlation
analysis showed that the EQI was significantly correlated with both the unmodified, modeled HWBI
health domain scores (r = .40, p < .01) and modeled HWBI scores (r = .21, p < .01). The significance of
the correlations was most likely driven by the large number of observations (n=3139); however,
neither correlation result showed a particularly strong relationship. In addition, regression analyses
depicted the EQI as a poor predictor of the health domain and the overall HWBI with R2 = .16 (p < .01
and R2 = .09 (p < .01), respectively. These results showed that the EQI, HWBI and HWBI health domain
scores were reasonably independent measures, suggesting that the EQI could be used as an additional
factor not explicitly included in the original HWBI model considerations.
6.2 Demonstration
34
-------
The most dramatic change in both HWBI health domain scores and the overall HWBI was introduced by
the relationship-function equations ratherthan the EQI modifications. Given the model fit and data
normality, these results were not unexpected. Data imputation methods used by both indices for some
of the more rural counties is the probable explanation. Modeled health domain scores (mean=58.2,
SD=2.4, range=51.6-66.5) exhibited a modest change in the range and mean of values after the EQ.I-
modification (mean=57.4, SD=4.8, range=43.0-75.2). However, modeled HWBI scores (mean=51.9,
SD=3.0, range=41.4 -59.8) remained virtually unchanged (mean=51.S, SD=2.8, range=42.8 - 60.2).
Observed, unmodified and EQI-modified health domain and HWBI scores were plotted in context of
the remaining HWBI domain scores along a distribution gradient. Figure 8 shows how the results for all
U.S. counties align with or deviate from the median (darkest color striations) of each set of scores.
Domains and HWBI
Safety and security
20th 30th
40th
50th 60th 70th 80th
90th
Hea Ith-Observed
Hea Ith-Model
Health-Adjusted Model
Connection to Nature
Leisure Time
Cultural Fulfillment
Living Standards
Education
Social Cohesion
HWBI-Observed
HWBI-Model
HWBI-Adjusted Model
Figure 8. Observed, unmodified and EQI-modified health domain and HWBI scores plotted in context of the
remaining HWBI domain scores along a distribution gradient, for all U.S. counties in relation to the median score
of each domain set.
Spatial-pattern shifts occurred when the modeled health-domain scores were EQI-adjusted. The spatial
distribution changes were more subtle for the overall HWBI scores, calculated using the full
complement of modeled domain scores including the EQI-modified health domain (Figure 9).
35
-------
HWBI Health Domain
HWBI
"O
-Q
o
>•
c
O
-------
Z-test results showed significant differences between the unmodified and EQI-modified health domain
model outputs (z=6.06, pc.OOl), but not for the HWBI (z=1.35, p<0.18). Differences observed between
the two sets of health domain and HWBI scores were evaluated for significance based on 95%
confidence limits, at ±6.77 and ±4.74, respectively. No significant difference was observed for the
modified HWBI. However, the EQI modification in the health domain model produced significant
changes in health scores for some counties (Figure 10).
Significant Change in Modeled HWBI Health Domain Scores
After EQI-Modification
rA
Decrease
Increase
Figure 10. Bivariate chioropleth depicting county HWBI health domain scores that exceeded the 95%
confidence threshold.
Approximately 28% of counties showed a significant effect in modeled HWBI health domain scores
resulting from the EQI modification. While the majority of county health scores were not adjusted
significantly, spatial distribution shifts were detected. As a result of the EQI modification, concentrated
areas of northeastern counties exhibited lower modeled health domain scores, while discreet pockets
of counties in the southeast and southwest showed higher scores.
37
-------
7.0 Concluding Remarks
This demonstration illustrates that by introducing the EQI to the HWBI framework, a new sentinel
indicator of change emerges. This union of indices adds value to the overall HWBI framework,
generating additional drill-down layers which may be used to help identify potential areas of overall
health and well-being disparities in context of economic, ecosystem and social service drivers that
influence quality of life characteristics. By leveraging the HWBI relationship-function equations, the
integration is simple and each index retains its original intent—a characterization of well-being and
environmental quality.
The decision to use the HWBI health domain model construct as the basis for this demonstration was
three-fold:
• Using the HWBI relationship-function equation models makes adapting this approach for other
well-being applications fairly easy, increasing the likelihood that the method will be adopted.
• Concentrating the influence of the EQI on the health domain specifically constrained it use to
the field it was originally intended to serve.
• Using the HWBI model based approach allowed the modified health domain relationship
equations to continue to function in a manner consistent with the other HWBI domain
relationship-function models. The observation is particularly important given the means-ends
nature of the HWBI framework—the value contributed to the HWBI is dependent on the
influence and response sensitivity of the two indices in context of all other HWBI-related
services.
Demonstration results indicate that the EQI-modification of the HWBI model has the capacity to
provide meaningful information related to modeled scores of overall health. While this demonstration
shows promise, some caveats should be considered. As with many indicator-related research efforts,
the lack of higher resolution data contributes to statistical noise, making it more difficult to detect true
patterns. Given that the HWBI relationship function equations were developed using state-level data,
the demonstration performed well but the accuracy of the HWBI domain models could be improved.
These two points alone could become particularly important for using the approach at local-scales.
More spatially and/or population sensitive imputation methods (e.g., Buck, 2017) might alleviate some
of the uncertainty attributed to data sparsity, particularly in more rural locales. Additionally, identifying
and developing easier ways to collect and synthesize existing data could help harmonize temporal
scales across indices. Finer-scale and more current data could help evolve the current HWBI
relationship function model which could lead to more robust and holistic tools that utilize a larger
inventory of existing ORD indicator products.
Individually, both indices contribute to a "cause and effect" dialog, which work best when interpreted
in terms of the characteristics of potential influence or response rather than the actual value itself. The
38
-------
approach described in this report potentially offers a different perspective for evaluating of overall
health characteristics. The EQI-adjusted health domain scores could be used to highlight areas that are
at higher potential risk for poorer overall health in the longer term based on chronic exposures to
adverse environmental conditions, even when current HWBI health domain scores alone suggest
better health characteristics. Conversely, improved HWBI health domain scores resulting from EQI-
modifications could help identify geographical areas where factors other than adverse environmental
exposures (e.g., culture, behaviors, availability of healthcare, age distribution) may be more influential
in overall health conditions.
The approach provides empirical results that can be visualized to identify hotspots of current or
potential future adverse environmental-health relationships. Having a sense of where problems may
exist could help inform environmental improvement decisions that may have the greatest community
impact. Equally important, the EQI-modified HWBI health scores could be used to communicate the
important linkages between a healthy lifestyle and a healthy environment to support that lifestyle and
overall well-being. While the demonstration focused on a county-level application, the approach can
be used at any spatial-scale, provided the appropriate data are available. The resulting suite of
indicators and indices could provide decision-makers and stakeholders alike with measures to assess
and track human health and well-being progress over time in the context of socio-ecological
interactions. The EQI-modified HWBI demonstration may serve as the first step towards building robust
information frameworks to support indicator and index development that promote more integrated
decision-support tools.
39
-------
8.0 References
Booysen, F. (2002). An overview and evaluation of composite indices of development. Social indicators
research, 59(2), 115-151.
Breslow, S. J., Sojka, B., Barnea, R., Basurto, X., Carothers, C., Charnley, S.,... & Hicks, C. C. (2016).
Conceptualizing and operationalizing human wellbeingfor ecosystem assessment and management.
Environmental Science & Policy, 66, 250-259.
Bruins, R.J.F., DeWitt, T.H., Lomnicky, G.A., Wilson, J.K., Specht, D.T. (2012). Developing and Applying
Ecological Production Functions. U.S. Environmental Protection Agency, Washington, DC. [Internal]
Buck, K. D. (2017a). A proposed method for spatial data disaggregation and interpolation. The Professional
Geographer, 69(1), 70-79.
Buck, K. D., Summers, J. K., Smith, L. M., and L.C. Harwell. (2017b). Application of the Human Weil-
Being Index to sensitive population divisions: A Children's Weil-Being Index development. Child
Indicators Research, 1-32.
Costanza, R., Cumberland, J. H., Daly, H., Goodland, R., Norgaard, R. B., Kubiszewski, I., & Franco, C.
(2014). An introduction to ecological economics. CRC Press.
EPA's Report on the Environment (ROE). Office of Research and Development. U.S. Environmental
Protection Agency, Washington D.C.
Fulford, R.S., Smith, L. M., Harwell, M., Dantin, D., Russell, M., and J. Harvey. (2015). Human well-being
differs by community type: Toward reference points in a human well-being indicator useful for decision
support. Ecological Indicators 56: 194-204.
Grabich, S. C., Horney, J., Konrad, C., & Lobdell, D. T. (2015). Measuring the Storm: Methods of
Quantifying Hurricane Exposure with Pregnancy Outcomes. Natural Hazards Review, doi:
10.1061/(ASCE)NH.1527-6996.0000204.
Harwell, L. Community Scale HWBI Tool and User's Guide. (2017). U.S. Environmental Protection
Agency, Washington, USA. EPA/600/R-17/298.
Jagai, J. S., Messer, L. C., Rappazzo, K. M., Gray, C. L., Grabich, S. C., & Lobdell, D. T. (2017). County-level
cumulative environmental quality associated with cancer incidence. Cancer, 123(15), 2901. doi:
10.1002/cncr. 30709.
Hoppe, R. (2017). Heuristics for practitioners of policy design: Rules-of-thumb for structuring
unstructured problems. Public Policy and Administration, 0952076717709338.
Ignatius, A. R., Wolfe, K., Parmar, R., Flaishans, J., Harwell, L., Yee, S., Purucker, T., and M. Galvin.
(2016). Design and Implementation of a REST API for the Human Well Being Index (HWBI).
Jennings, V., Larson, L., & Yun, J. (2016). Advancing sustainability through urban green space: cultural
ecosystem services, equity, and social determinants of health. International journal of environmental
research and public health, 13(2), 196.
40
-------
JRC (Joint Research Centre-European Commission). (2008). Handbook on constructing composite
indicators: Methodology and user guide. OECD publishing.
Kubiszewski, I., Costanza, R., Franco, C., Lawn, P., Talberth, J., Jackson, T., & Aylmer, C. (2013). Beyond
GDP: Measuring and achieving global genuine progress. Ecological Economics, 93, 57-68.
Lobdell, D.,Jagai, J., Messer, L., Rappazzo, K., Messier, K., Smith, G., Pierson, S., Rosenbaum, B., and M.
Murphy. (2014). Creating an Overall Environmental Quality Index - Technical Report. U.S.
Environmental Protection Agency, Washington, DC. EPA Report EPA/600/R-14/304.
Messer, L. C., Jagai, J. S., Rappazzo, K. M., & Lobdell, D. T. (2014). Construction of an environmental
quality index for public health research. Environmental Health, 13(1), 39.
Newell, A., & Simon, H. A. (1972). Human problem solving (Vol. 104, No. 9). Englewood Cliffs, NJ:
Prentice-Hall.
OECD (2017), Composite leading indicator (CLI) (indicator), doi: 10.1787/4al74487-en (Accessed on 22
September 2017)
Orlando, J., Yee, S., Harwell, L., and L. Smith. (2017). Technical Guidance for Constructing a Human
Weil-Being Index (HWBI): A Puerto Rico Example. U.S. Environmental Protection Agency, Washington,
DC. EPA Report EPA/600/R-16/363.
Pickard, B. R., Daniel, J., Mehaffey, M., Jackson, L. E., & Neale, A. 2015. EnviroAtlas: A new geospatial
tool to foster ecosystem services science and resource management. Ecosystem Services, 14, 45-55.
Rappazzo, K. M., Messer, L. C., Jagai, J. S., Gray, C. L., Grabich, S. C., & Lobdell, D. T. (2015). The
associations between environmental quality and preterm birth in the United States, 2000-2005: a
cross-sectional analysis. Environmental Health, 14[ 1), 50.
Singh, R. K., Murty, H. R., Gupta, S. K., & Dikshit, A. K. (2009). An overview of sustainability assessment
methodologies. Ecological indicators, 9(2), 189-212.
Smith, L. M., H. M. Smith, J. L. Case, and L. Harwell. (2012). Indicators and Methods for Constructing a
U.S. Human Well-being Index (HWBI) for Ecosystem Services Research. U.S. Environmental Protection
Agency, Washington, DC, EPA/600/R-12/023.
Smith, L. M., Case, J. L., Smith, H. M., Harwell, L. C., & Summers, J. K. (2013a). Relating ecosystem
services to domains of human well-being: Foundation for a US index. Ecological Indicators, 28, 79-90.
Smith, L. M., Case, J. L., Harwell, L. C., Smith, H. M., & Summers, J. K. (2013b). Development of relative
importance values as contribution weights for evaluating human wellbeing: an ecosystem services
example. Human Ecology, 41[4), 631-641.
Smith, L. M., Harwell, L., Summers, J. K., Smith, H. M., Wade, C. M., Straub, K. R., and J. L. Case. (2014a).
A U.S. Human Well-being Index (HWBI) for multiple scales: linking service provisioning to human well-
being endpoints (2000-2010). U.S. Environmental Protection Agency, Gulf Breeze, FL. EPA Report
EPA/600/R-14/223.
41
-------
Smith, L., C. Wade, K. Straub, L. Harwell, J. Case, M. Harwell, and Kevin Summers. (2014b) Indicators
and Methods for Evaluating Economic, Ecosystem and Social Services Provisioning: A Human Well-
being Index (HWBI). U.S. Environmental Protection Agency, Washington, DC, EPA/600/R-14/184.
Smith, L. M., Wade, C., Case, J., Harwell, L., Straub, K. and J. K. Summers. (2015). Evaluating the
Transferability of a U.S. Human Well-being Index (HWBI) Framework to Native Americans Populations.
Social Indicators Research, 124(1): 157-182.
Summers, J. K., Smith, L. M., Harwell, L. C., Case, J. L., Wade, C. M., Straub, K. R., & Smith, H. M. (2014).
An index of human well-being for the US: a TRIO approach. Sustainability, 6(6), 3915-3935.
Summers, J. K., Harwell, L., and L. M. Smith. (2016). A Model for Change: An Approach for Forecasting
Weil-Being from Service-Based Decisions. Ecological Indicators, 69:295-309.
Summers, J. K., L. M. Smith, L. C. Harwell, and K. D. Buck (2017). Conceptualizing holistic community
resilience to climate events: Foundation for a climate resilience screening index. GeoHealth. 1, 151-
164, doi:10.1002/2016GH000047.
USDA (United States Department of Agriculture). (2003). "Measuring rurality: Rural-urban continuum
codes". Accessed 25 September 2017).
USEPA(U.S. Environmental Protection Agency). (2008). [EPA/600/R-07/045F]. EPA's 2008 Report on
the Environment. Washington, DC: National Center for Environmental Assessment.
USEPA (US Environmental Protection Agency). (2012) Sustainable and healthy communities strategic
research action plan 2012-2016. [EPA 601/R-12/005]. Washington, DC: Office of Research and
Development,
USEPA (US Environmental Protection Agency). (2014). Fiscal Year 2014-2018 EPA strategic plan.
Washington, DC: Office of Science and Technology.
USEPA (US Environmental Protection Agency). (2015) Sustainable and healthy communities strategic
research action plan 2016-2019. [EPA 601/K-15/006]. Washington, DC: Office of Research and
Development, Sustainable and Healthy Communities.
USEPA (US Environmental Protection Agency). State of the Practice for Sustainability Indicators: 2012-
2015. An Internal Report on the Development of Sustainability Indicators in the Sustainable and
Healthy Communities Research Program. Washington, DC: Office of Research and Development.
[Internal]
World Bank. "Gross National Income". (Accessed on 22 September, 2017).
42
-------
Appendix A. Applications and Adaptations of the HWBI
Framework
43
-------
Al. Research Highlight: Transferability of the HWBI Framework to
Native American Populations
The applicability and integrity of the HWBI framework was demonstrated using metrics scaled to assess well-
being for American Indian Alaska Native (AIAN) and large tribal populations. The HWBI approach can be used to
estimate well-being for Native Americans collectively with a reasonable level of confidence. The degree to which
the HWBI structure can be utilized is dependent upon the availability and quantity of quality data. Greater than
80% of the data available for a national AIAN assessment were specific to the target population, while the
remaining data were derived from the general U.S. population. The remaining data were derived from AIAN
population weighted U.S. HWBI values. Despite using roughly 20% non-specific population data, the AIAN well-
being signature could still be differentiated from the U.S. HWBI.
To overcome limitations, data substitution using the described approach, is the most robust method for scaling
the index, but the limited availability of comparable metrics at smaller spatial scales and for specific
demographics may also be problematic. The metrics utilized in the U.S. HWBI range in nature from individuals'
perceptions (survey questions) to rates of occurrences of certain behaviors and outcomes in a population. In
order to maintain index integrity and capture the most holistic and comprehensive picture of a population, it is
sometimes necessary to identify alternative metrics. When choosing alternative metrics, it is imperative that
both the qualitative nature of the information as well as the type of information is as closely matched as
possible. Alternative metrics for AIAN populations were suggested for the HWBI metric Performing Arts
Attendance. This substitution caused a dramatic increase in the metric and the domain score (Figure A-l)
¦Original Metric
]Alternate Metric
Domain: Cultural Fulfillment Indicator: Activity Participation
Figure A-1 Comparison of the results of using an alternative
metric for the Activity Participation indicator in the Cultural
Fulfillment domain
Only data that could be readily identified as
AIAN-related were collected from sources.
Data records were encoded to differentiate
between single ethnic and multi-ethnic
identified information, AIAN and AIAN-
mixed, respectively. For each record, the
collection method was identified as either
random (e.g., exit polls) or complete (e.g.,
vital statistics). Metric categorization was
based upon reported ethnicity, sample size
and temporal scale data availability. All 80
metrics were categorized into one of six
categories. Raw data were organized
hierarchically by population group and temporal
resolution (e.g., AIAN and Tribal grouping by
year and decade). National AIAN and Tribal Group datasets were created by populating metric values from the
most robust data available according to the metric categorization process and from existing U.S. HWBI metric
data (Figure A-2).
44
-------
What is ihc desired
population grwu-p?
National National
AIAN Tribal Grouping
i I
Are annual AlAN-spccific
data available (Category I)?
Arc Tribal-specific population \ Yes
metric data available? J
No
No
No
Arc non-imputed data avai lablc
for counties with >50% of the
population in specific iribc?
Arc decadal AlAN-spccific
data available (Category II)?
No
Use Tribal-specific
populalion metric
data.
Use annual AlAN-spccific
population metric data.
Use decadal AlAN-spccific
population metric data.
Use tribal population
weighted mean of
county metric values.
Use AIAN or Tribal population
percent weighted area values..
Figure A- 2. Process for selecting the most robust AfAN and Tribal Group data avaifabfefor HWBI
assessments
Where, tribal specific data were available, a Tribal Group identifier was included with the data appropriate.
Tribal specific metric values were aggregated into one of 38 Tribal Groups as represented in the tribal
assignments for the U.S. Census (2000). 7 of the 38 Tribal Groups with the greatest percentage of tribal specific
data were selected for HWBI and domain score comparison. The 7 tribal groups with sufficient data include the
Menominee, Navajo, Chippewa, Blackfeet, Alaskan Athabascan, Eskimo, and Sioux. Each of the seven Tribal
groups was compared to the county HWBI scores for which the counties had greater than 50% of the population
identified as tribal-specific.
For the 2000-2010 periodthe iowest ranked indicator scores for the Tribal groups were in the domains of Health,
Living Standards, Safety and Security and Social Cohesion (Fig. 3). Differentiation between tribal domain scores is
dependent upon the specificity of the data included in the assessment. Where AIANAlAN-mixed and U.S. data
comprised the majority of the metrics used to calculate tribal indicators, rarely were differences in domain values
observed. Differences among tribal scores were attributed to tribal specific and county population weighted
metric data which better characterize individual Tribal groups.
AS
-------
The tribal application of the HWBI provides a unique opportunity to examine the performance of the
EQI modified HWBI relationship-equation functions with a population group often underrepresented in
national assessments.
100
¦ 38 Tribal Groups
¦ ALASKAN ATHABASCAN
90
BLACKFEET
¦ CHIPPEWA
80 ¦ ESKIMO
¦ MENOMINEE
70 I ¦ NAVAJO
• T I T T ¦ SIOUX
Connection to Nature Cultural Fulfillment
Education
Health
Leisure Time
Living Standards Safety and Security Social Cohesion
HWBI
Figure A-3. Large Tribal Group domain and HWBI scores for the 2000-2010 time-period.
46
-------
A2. Technical Guidance for Constructing a Human Weil-Being Index (HWBI): A
Puerto Rico Example
The U.S. Environmental Protection Agency (EPA) Office
of Research and Development's Sustainable and Healthy
Communities Research Program (USEPA 2015)
developed the Human Well-being Index (HWBI) as an
integrative measure of economic, social, and
environmental contributions to well-being. The HWBI is
composed of indicators and metrics representing eight
domains of well-being: connection to nature, cultural
fulfillment, education, health, leisure time, living
standards, safety and security, and social cohesion. The
domains and indicators in the HWBI were selected to
provide a well-being framework that is broadly
applicable to many different populations and
communities, and can be customized using community-
specific metrics (Orlando et al. 2017).
A primary purpose of this report was to adapt the U.S.
Human Weil-Being Index (HWBI) to quantify human
well-being for Puerto Rico. This application provided an
example of how the HWBI could be adapted to different communities, especially locations outside the
mainland U.S. Additionally, technical guidance on processing data and calculating index using R is
offered.
Technical Guidance for Constructing
a Human Well-being Index (HWBI):
A Puerto Rico Example
The domains and indicators in the HWBI were selected to provide a well-being framework, which can
be transferred across different populations and communities and customized using more community
specific metrics. The San Juan Puerto Rico case study was a demonstration of how this can be done
utilizing data substitutions more closely link to local environments.
The HWBI adaptations were calculated to compare human well-being both within Puerto Rico and to
the U.S. population. Puerto Rico's population is linked to the U.S. government, economy, and
institutions, yet culturally distinct and geographically isolated from the mainland. Therefore, Puerto
Rico presented an opportunity to explore the transferability of the HWBI to this unique group of
islands. The adaptation of the HWBI to Puerto Rico built upon the existing framework, but considered
data selection options that better informed indicator development within the context of Puerto Rico
communities. Two suites of HWBI indicators and indices were generated, a HWBI using Puerto Rico
specific-data only—dropping metrics where data did not exist—and another using U.S. metrics to fill
information gaps. An HWBI and suite of related domain and indicator values were scaled to the 78
municipios.
47
-------
0.55
0.50
io,5
0.40
0.35
Similarto the approach
used to examine the
transferability of HWBI for
American Indian and
Alaska Native (AIAN)
populations (Smith, et al.
2015), the Puerto Rico
metrics used to calculate
the HWBI were based data
availability and the
appropriate data
substituted for
characterizing HWBI
indicators for Puerto Rico
populations. HWBI scores for
Puerto Rico municipios were calculated based on the range of values within PR and the range of values
for the U.S. (Figure A-4). Additional HWBI comparisons are also presented for Puerto Rico in context of
all U.S. states (Figure A-5)
Figure A-5. Map of mean decadal (2000-2010) HWBI for Puerto Rico
municipios. Higher HWBI scores are indicated with lighter colors; darker blues
indicate lower HWBI scores
2 — ^ZUz^SQ-g 5 0"20^U02uQOZZt-i-fl-
-------
A comparison at the indicator level for Puerto Rico and the U.S. states with highest and lowest HWBI
scores is depicted in Figure A-6. The Puerto Rico application of the HWBI provides a basis for future
consideration of an EQI-modified HWBI demonstration in a geographic area with unique environmental
considerations.
0.8
07'
Puerto Rico
New Hampshire
Louisiana
Biophilia
Basic Educational Knowledge and Skills of Youth
Participation and Attainment
Social, Emotional and Developmental Aspects
Healthcare
Life Expectancy and Mortality
Lifestyle and Behavior
Personal Well-being
Physical and Mental Health Conditions
Cultural Activity Participation
Leisure Activity Participation
Time Spent
Working Age Adults
Basic Necessities
Income
Wealth
Work
Attitude toward Others and the Community
Democratic Engagement
Family Bonding
Social Engagement
Social Support
Actual Safety
Perceived Safety
Risk
Figure A-6. Comparison of mean decadal (2000-2010) indicator scores for Puerto Rico and the U.S. states with
the highest (New Hampshire) and lowest (Louisiana) HWBI scores.
49
-------
References
Orlando, J., Yee, S., Harwell, L., and L. Smith. (2017). Technical Guidance for Constructing a Human
Weil-Being Index (HWBI): A Puerto Rico Example. U.S. Environmental Protection Agency, Washington,
DC. EPA Report EPA/600/R-16/363.
Smith, L. M., Wade, C., Case, J., Harwell, L., Straub, K. and J. K. Summers. (2015). Evaluating the
Transferability of a U.S. Human Well-being Index (HWBI) Framework to Native Americans Populations.
Social Indicators Research, 124(1): 157-182.
USEPA (Environmental Protection Agency). (2015) Sustainable and healthy communities strategic
research action plan 2016-2019. [EPA 601/K-15/006]. Washington, DC: Office of Research and
Development, Sustainable and Healthy Communities.
50
-------
A3. Application of the Human Well-Being Index to Sensitive Population
Divisions: A Children's Well-Being Index Development
Since its development, the Human Weil-Being Index (HWBI) has undergone two adaptations in order to
both assess its applicability and highlight specific populations - U.S. adults (Summers et al. 2014; Smith
et al. 2015). In Puerto Rico, the application of HWBI focused on data and adapting existing metrics and
index structure (county to municipio) to a US territory, whereas the Native American application
focused on distinct populations living within the US boundaries. Results from these studies
demonstrate the adaptability of the HWBI, which allows selected population groups to be highlighted
and compared to the larger US population. However, characteristics such as age and ethnicity also play
large roles in how groups are either benefited or harmed by the access to resources on a larger scale
(Crimmins et al., 2004). While the average citizen may benefit from a community characteristic,
individuals existing outside of the socio-demographic norms may be adversely impacted.
To test this socio-demographic theory, another adaptation of HWBI is undertaken in order to both
conceptually and empirically test a version of the index specifically adapted to children. Children,
defined in the U.S. as those under the age of 18, are considered dependents and under their parent's
care and guidance. The primary objective of this research is to determine whether the HWBI can be
effectively calculated for an age-specified sub-population in the United States (i.e., children). This
requires identification of clear theoretical connections between the original HWBI metrics and the
CWBI in addition to data isolated by age (Figure 1). Success in this age-specific application is defined
by: 1) a clear model for adaptation of the index, 2) availability of data at appropriate scale and
capturing the proper concepts, and 3) a resulting index that is statistically robust and consistent with
other indices of children's health and well-being.
Does the
metric assess
children?
YES
Use Metric
Drop
Metric
NO
YES
NO
Is it a community
measure with
applicability to
children?
Will indicator still
concept without
this metric or
equivalent?
NO
Can comparable
metrics can be
used?
Use new metric
with refined
definition of
connection
NO
YES
YES
Define now connection
Define new connection
Indicator
Metric
Indicator
Figure A- 7. Decision flow chart for metric adaptation in children's well-being index creation for
the original HWBI. Start with metric in order to minimize changes at indicator or domain level.
Move to right indicates metric retention, while move to left indicates metric drop.
51
-------
Benefits of child-specific research are two-fold. First, children are very susceptible to environmental
conditions, whether natural, built or social (Del Carmin Ruis et al., 2016). From pregnancy up to
adulthood, children are developing physically and emotionally and are more vulnerable to poor
environmental conditions than adults (Goldman, 1995; Punch, 2002). Another reason for interest in
child-specific well-being is to serve as a point of reference for children within the larger societal
constructs. Developing a conceptually identical index that examines a specific sub-population
(statistically speaking) allows for direct comparisons between groups. In the case of children, they are
future adults; hence, their development and current well-being can provide a window into adult well-
being ten to twenty years in the future. If the well-being of children is higher compared to the general
population, there is a possibility that the general population's well-being will improve over the next
few generations. Conversely, if children are doing poorly in comparison to the general population, the
future may not look quite as bright. There also exists the possibility that both the present and future
well-being of adults will be positively influenced by investments in family and community centered
issues. While, most communities recognize the importance of investing in children to improve their
well-being, not all investments pay off in predictable ways. A forecasting assessment tool for adult
well-being based on leading indicators of children's well-being in the present would be preferable to
waiting 20 years for the resultant well-being to materialize.
The end result of the adaptation is a set of 8 domains matching those of the original HWBI, 3 changes
of indicator terminology, and the adjustment of 42 metrics to accommodate data availability and
theoretical differences between an index meant to represent an entire population and one specific to
children (Table A-l). While many of the metrics are altered, they are all able to maintain a structure
and premise closely resembling the original HWBI (Summers et al., 2014).
Table A-l. Indicator and metric count per domain. Data representation (HWBI count/CWBI count).
Domain
Indicator Count
Metric Count
Social Cohesion
5/5
17/17
Living Standard
4/4
9/12
Education
3/3
11/12
Connection to Nature
1/1
2/1
Cultural Fulfillment
1/1
2/3
Health
5/5
26/19
Safety and Security
3/3
6/8
Leisure Time
3/3
6/7
Much of what makes the original HWBI, and by extension the CWBI unique is the ability to assess
individual counties across all of the indicators and domains. This allows for the identification of not
only regional trends, but also approaches community level assessment where scores can be compared
given time, monetary and policy investments to determine effectiveness. Every effort is made in the
development of this index to strike the balance of change from the original HWBI and to find metrics
directly representative of children in the community. Such an approach was intended to help to draw
distinct conclusions about their well-being while also remaining rooted in the general well-being of the
52
-------
community. Looking at the overall CWBI scores, there are definite regional clusters apparent (Figure
2). The highest well-being values exist in the much of the upper Midwest and in the Southeast. The
lowest scores are in parts of the deep South, the Southwest, and along areas of the East Coast.
Children's Weil-Being Index
Figure A- 8. Children's Well-Being Index scores for all US Counties in 2010.
A final comparison of CWBI to the original HWBI, both at index and domain levels, reveals a key finding
that involves drivers of well-being between the general population and children. Much of the
influence in a more adult-centric index comes from economic drivers, whereas health and education
tend to drive more of the CWBI. This corresponds to comparisons, where only a few of the domains,
principally Living Standards and Health, have a moderate correlation, while Social Cohesion has a very
weak correlation across the indices. In the case of Living Standards, many of the metrics used to assess
this domain are community level and traits of parents should be very similar to those of the children.
Along the same lines, healthy parents will be more likely to have healthy kids. The remaining domains:
Leisure Time, Safety and Security, Education, and Social Cohesion are more subjective and it is possible
that parent's views do not align with those of their children, or the domain simply measures a very
different conceptual piece when comparing children to the greater community.
The CWBI presents a unique case for identifying the strengths and weakness of the EQI-modified HWBI
relationship function equations.
53
-------
References
Crimmins, E. M., Hayward, M. D., & Seeman, T. E. (2004). Race/Ethnicity, Socioeconomic Status, and
Health. In N. B. Anderson (Ed.), Critical Perspectives on Racial and Ethnic Differences in Health in
Late Life. Washington, DC: National Academies Press.
Del Carmin Ruiz J, Quackenboss JJ, Tulve NS (2016) Contributions of a Child's Built, Natural, and Social
Environments to Their General Cognitive Ability: A Systematic Scoping Review. PLoS ONE 11(2):
e0147741.
Goldman, L. R. (1995). Children - Unique and Vulnerable - Environmental Risks Facing Children and
Recommendation for Response. Environmental Health Perspectives, 103, 13-18,
doi: 10.2307/3432338.
Punch, S. (2002). Research with children - The same or different from research with adults? [Article].
Childhood-a Global Journal of Child Research, 9(3), 321-341,
doi: 10.1177/0907568202009003005.
Smith, L. M., Wade, C. M., Case, J. L., Harwell, L. C., Straub, K. R., & Summers, J. K. (2015 ). Evaluating
the Transferability of a U.S. Human Weil-Being Index (HWBI) Framework to Native American
Populations. Social Indicators Research, 124, 157-182.
Summers, J. K., Smith, L. M., Harwell, L. C., Case, J. L., Wade, C. M., Straub, K. R., et al. (2014). An Index
of Human Weil-Being for the US: A TRIO Approach. [Article]. Sustainability, 6(6), 3915-3935,
doi:10.3390/su6063915.
54
-------
Appendix B. HWBI Relationship-Function Equation Coefficients
55
-------
Table Bl. Relationship-function equation coefficients used for the HWBI domain models to predict well-being
from service categories. The health domain model, which is primarily used in this report, is highlighted in gray.
The remaining domain models are used to calculate the HWBI with EQI-modified health domain values. More
information about this table is available in Summers et al. (2016).
Domain
Service
Coefficient
Partial
R2
Connection to Nature (R2=0.78)
Intercept
+2.4312
Community and Faith-Based Initiatives
+0.5772
0.29
Consumption
+0.4655
0.14
Re-Distribution
-0.3704
0.08
Activism
-1.7559
0.07
Healthcare
-0.1117
0.05
Greenspace
-0.524
0.04
Emergency Preparedness
-2.3885
0.03
Water Quality
+0.0505
0.02
Labor
-1.9341
0.01
Education
+0.2116
0.01
Community and Faith-Based
-1.999
0.01
lnitiatives*Emergency Preparedness
Activism*Emergency Preparedness
+2.1033
0.01
Emergency Preparedness*Labor
+3.2228
0.01
Cultural Fulfillment (R2=0.43)
Intercept
-0.2239
Community and Faith-Based Initiatives
+2.4296
0.21
Air Quality
-0.1007
0.08
Water Quantity
-0.1314
0.03
Emergency Preparedness
+0.0847
0.02
Education
+0.1918
0.02
Innovation
+0.0999
0.02
Communication*Community and Faith-Based
Initiatives
-4.4056
0.02
Communication
+1.2805
0.01
Production
-0.0972
0.01
Air Quality*Community and Faith-Based
Initiatives
+0.2347
< 0.01
Education (R2=0.81)
Intercept
+0.3928
Family Services
+0.3508
0.47
Community and Faith-Based Initiatives
+0.4638
0.14
Consumption
-0.3737
0.06
Production
-0.4887
0.04
Public Works
+0.0782
0.04
Justice
-0.4415
0.03
Activism
+0.5748
0.02
Re-Distribution*Greenspace
+0.3906
0.02
Health (R2=0.54)
Intercept
+0.2311
Family Services
+0.0727
0.25
56
-------
Domain
Service
Coefficient
Partial
R2
Communication
+0.1949
0.08
Labor
+0.0977
0.07
Water Quantity
+0.0204
0.04
Innovation
+0.096
0.02
Emergency Preparedness
+0.0491
0.01
Community and Faith-Based Initiatives
+0.525
0.02
Justice
+0.1491
0.03
Community and Faith-Based lnitiatives*Justice
-0.8663
0.01
Activism*Education
+0.0503
< 0.01
Leisure Time (R2=0.74)
Intercept
+0.5062
Employment
-0.341
0.22
Water Quantity
-0.7197
0.10
Food, Fiber and Fuel Provisioning
+0.6821
0.18
Water Quality
-0.0537
0.07
Water Quantity*Education
+1.6
0.04
Activism
+0.9347
0.03
Greenspace
+0.1382
0.03
Education
-0.5449
0.03
Public Works
+0.5773
0.02
Community and Faith-Based Initiatives
-0.2174
0.01
Finance*Communication
+0.2062
0.01
Activism*Public Works
-1.2947
0.01
Consumption
-0.3924
<0.01
lnnovation*Education
-0.1715
<0.01
Living Standards (R2=0.79)
Intercept
+0.2750
Employment
+0.0923
0.38
Public Works
-0.1462
0.12
Labor
+0.1347
0.10
Activism
+0.3676
0.06
Finance
-0.2594
0.05
Justice
-0.1786
0.03
Water Quantity
+0.0784
0.02
Capital Investment
-0.0249
0.01
Finance*Public Works
+0.7086
0.01
Capital lnvestment*Water Quality
-0.0383
0.01
Food, Fiber and Fuel
Provisioning*Communication
+0.1772
0.01
Safety and Security (R2=0.48)
Intercept
+0.6039
Community and Faith-Based Initiatives
+0.2941
0.18
Water Quality
-0.3806
0.17
Public Works
-0.3853
0.04
Water Quantity
+0.0854
0.01
Activism*Labor
+1.3532
0.02
Finance*Pubiic Works
+0.574
0.02
57
-------
Domain
Social Cohesion (R2=0.77)
Service
Coefficient
Partial
R2
Emergency Preparedness*Justice
+0.8986
0.01
Water Quality*Public Works
+0.6556
0.01
Production*Food
+0.2951
0.01
Production*Healthcare
-0.3043
< 0.01
Justice*Labor
-1.1474
< 0.01
Greenspace*Emergency Preparedness
-0.7423
< 0.01
Finance*Activism
-0.6023
< 0.01
Intercept
-0.8102
Justice
+1.0728
0.41
Air Quality
+0.0425
0.11
Production
-0.3830
0.06
Community and Faith-Based Initiatives
+1.9806
0.04
Capital Investment
+0.1004
0.03
Public Works
+0.0473
0.02
Re-Distribution
+1.2823
0.02
Labor
+0.1207
0.02
Family Services
+0.1529
0.01
Consumption
-0.1481
0.01
Re-Distribution*Community and Faith-Based
Initiatives
-3.5943
0.01
Greenspace*Justice
-2.0480
0.01
Employment*Water Quality
-0.0365
< 0.01
Greenspace
+1.2913
< 0.01
58
-------
-------
vvEPA
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT NO. G-35
Office of Research and
Development (8101R)
Washington, DC 20460
Offal Business
Penalty for Private Use
$300
------- |