Genetic and environmental influences on residential location in the US

ArticleinHealth & Place 18(3):515-9 · February 2012with36 Reads
Impact Factor: 2.81 · DOI: 10.1016/j.healthplace.2012.02.003 · Source: PubMed
Abstract

We used a classical twin design and measures of neighborhood walkability and social deprivation, using each twin's street address, to examine genetic and environmental influences on the residential location of 1389 same-sex pairs from a US community-based twin registry. Within-pair correlations and structural equation models estimated these influences on walkability among younger (ages 18-24.9) and older (ages 25+) twins. Adjusting for social deprivation, walkability of residential location was primarily influenced by common environment with lesser contributions of unique environment and genetic factors among younger twins, while unique environment most strongly influenced walkability, with small genetic and common environment effects, among older twins. Thus, minimal variance in walkability was explained by shared genetic effects in younger and older twins, and confirms the importance of environmental factors in walkability of residential locations.

Full-text

Available from: Eric Strachan
Genetic and environmental influences on residential location in the US
Glen E. Duncan
a,b,
n
, Elizabeth J. Dansie
c
, Eric Strachan
d
, Melissa Munsell
e
, Ruizhu Huang
e
,
Anne Vernez Moudon
a,e
, Jack Goldberg
a,f
, Dedra Buchwald
c
a
Department of Epidemiology, University of Washington, Raitt Hall, Seattle, WA 98195–3410, USA
b
Department of Epidemiology, Interdisciplinary Graduate Program in Nutritional Sciences, University of Washington, Raitt Hall, Seattle, WA 98195–3410, USA
c
Department of Medicine, University of Washington, Twin Registry, Raitt Hall, Seattle, WA 98195–3410, USA
d
Department of Psychiatry, University of Washington, Raitt Hall, Seattle, WA 98195–3410, USA
e
Department of Urban Design and Planning, University of Washington, Raitt Hall, Seattle, WA 98195–3410, USA
f
Vietnam Era Twin Registry, Seattle VA Epidemiologic Research and Information Center, Seattle, WA 98195–3410, USA
article info
Article history:
Received 7 September 2011
Received in revised form
30 January 2012
Accepted 8 February 2012
Available online 17 February 2012
Keywords:
Environment
Genetics
Neighborhood
Twins
Walkability
abstract
We used a classical twin design and measures of neighborhood walkability and social deprivation, using
each twin’s street address, to examine genetic and environmental influences on the residential location
of 1389 same-sex pairs from a US community-based twin registry. Within-pair correlations and
structural equation models estimated these influences on walkability among younger (ages 18–24.9)
and older (ages 25þ ) twins. Adjusting for social deprivation, walkability of residential location was
primarily influenced by common environment with lesser contributions of unique environment and
genetic factors among younger twins, while unique environment most strongly influenced walkability,
with small genetic and common environment effects, among older twins. Thus, minimal variance in
walkability was explained by shared genetic effects in younger and older twins, and confirms the
importance of environmental factors in walkability of residential locations.
& 2012 Elsevier Ltd. All rights reserved.
1. Introduction
The role of the physical or built environment in supporting
healthy lifestyles has gained increased attention over the last
decade. However, causal conclusions about the effects of the built
environment on health are tenuous because the allocation of
people into residential locations is neither random nor simply
based on individual choices. People are born in certain environ-
ments with certain genetic predispositions that may be associated
with important health outcomes. In other words, residential
selection and health may exhibit some familial transmission.
Moreover, irrespective of shared environment and genes, not
everyone can live in the environment or ‘‘neighborhood’’ they
prefer based on their own experiences and expectations. The
differential allocation of people to locations due to social forces
and other selection processes (i.e., social and economic stratifica-
tion) is addressed in statistical models by adjusting for individual
characteristics such as age, sex, race, income, education, and,
more recently, housing and land use preferences (Frank et al.,
2007; Handy et al., 2006). Even so, the problems of structural and
possible familial confounding still exist (Messer, 2007; Oakes,
2004, 2006; Oakes and Church, 2007). Statistical models can
adjust for known correlates, but omit important factors that are
difficult or impossible to measure or unrecognized.
Twin studies provide one method to examine the host of
factors that drive the selection of residential locations because
the degree of similarity in the choices made within monozygotic
(MZ) and dizygotic (DZ) pairs can be partitioned into genetic, as
well as shared and non-shared environmental factors. For exam-
ple, in one co-twin control study, MZ twins lived spatially closer
to one another than DZ twins (Neyer, 2002) suggesting that
choice of residential location may be partially heritable. Similarly,
at least two classical twin studies have investigated genetic and
environmental influences on residential location selection. This
design typically requires the use of population-based twin regis-
tries that compare a phenotype of interest in MZ and DZ pairs. The
Australian Twin Registry found strong effects for both genetic and
shared environmental factors on residential location, measured as
the distance from major urban centers, with heritability influ-
enced by both the sex and age of the twins (Whitfield et al., 2005).
As shared environmental effects decreased with age, additive
genetic factors increased. In contrast, the Netherlands Twin
Register observed both common and unique environmental
factors determined residential location of Dutch twins, suggesting
differences in choice of residential location by urbanization level
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/healthplace
Health & Place
1353-8292/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.healthplace.2012.02.003
n
Correspondence to: Box 353410, 305 Raitt Hall University of Washington,
Seattle, WA 98195–3410, USA. Tel.: þ 1 206 616 2680; fax: þ 1 206 685 1696.
E-mail address: duncag@u.washington.edu (G.E. Duncan).
Health & Place 18 (2012) 515–519
Page 1
were not genetic in origin (Willemsen et al., 2005). Despite their
differences, age and geography appear to influence genetic and
environmental factors on selection into residential locations.
Notably, the Australian and Dutch studies defined residential
location using a categorical description of the urbanization level.
One measure that may have implications for public health is an
indicator of environmental walkability (Frank et al., 2005). Highly
walkable environments can exist across the spectrum of urbani-
zation. Walkability indices are continuous measures that capture
how walkable a location is independent of urbanization, applic-
able in cities, suburbs, and even rural areas. We, therefore,
conducted a classical twin study of the genetic and non-genetic
environmental factors influencing residential location using a
walkability index in a community-based US twin registry.
The aims of our study were to: (1) estimate the contribution
of common and unique environmental factors to walkability;
(2) determine if the common and unique environmental influ-
ences on walkability vary by age; and, (3) determine if common
and unique environmental influences on walkability are indepen-
dent of neighborhood socioeconomic status.
2. Methods
Participants and setting. The study sample consisted of 1,389
same-sex twin pairs from the University of Washington Twin
Registry (UWTR). The UWTR is a community-based sample of
adult twin pairs assembled from Washington State Department of
Licensing records. Details of the construction and characteristics
of the Registry are described elsewhere (Afari et al., 2006). Briefly,
in Washington State, all new driver license applicants and
identification card applicants are asked if they are a member of
a twin. Since 1999, data on all twin applicants have been
transmitted to the UWTR from the Department of Licensing.
Twins were mailed a brief survey that included items on zygosity,
socio-demographics, height and weight, general health and com-
mon medical conditions, and lifestyle behaviors. We only used
data from twins who lived in Washington State and for whom had
residential street addresses from surveys completed between
2006 the present. A waiver of written consent was approved for
the initial survey by the university’s institutional review board.
Measures. Demographic information used in the analysis con-
sisted of age, sex, zygosity, race (dichotomized as white/non-
white), education (categorized into less than high school, high
school graduate, or at least some college), and marital status
(categorized into single, married, widowed/divorced/separated, or
living with someone). Zygosity was determined using standard
questions on childhood similarity that are highly accurate (Eisen
et al., 1989; Torgersen, 1979).
We constructed a neighborhood walkability index that is
commonly cited in the physical activity and built environment
literature (Frank et al., 2005). The index combines measures of
land use mix, residential density, and street connectivity taken
within a 1-km distance from a location of interestin this case,
the participants’ home. Participants’ home addresses were geo-
coded with a minimum match score of 100%. Addresses that failed
automatic geocoding were reviewed individually and matched
manually. A 1-km buffer was delineated along the street network
around each twin’s residential location using ESRI StreetMap
Premium (North America NAVTEQ 2009 Release 1, 2009). The
neighborhood walkability index was calculated for each network
buffer, and included measures of land use mix, as areas occupied
by residential, commercial, or office land uses, obtained from the
Washington State parcel database (the University of Washington
Geographic Information Service, Parcels Working Group, 2009);
residential density from the 2000 US Census block groups; and
intersection density. All spatial analyses were performed within
ArcGIS 9.3.1 (ESRI, Redlands 2008). The final walkability index
was derived from the z-score of each measure, and entered into
the following equation: (6* z-score of land use mix)þ (z-score of
net residential density)þ (z-score of intersection density). The
final output is a single continuous value representing the walk-
ability for each participant, with higher values representing more
walkable environments.
We also constructed a composite measure of the neighborhood
socioeconomic status as described by Singh (Singh, 2003). This
composite area-based deprivation index consists of 17 socio-
economic indicators that are related to the material and social
conditions, and relative socioeconomic disadvantage, in a given
community. The indicators comprising the index for our twins
were calculated at the census tract level, using data from the 2000
Washington State Census and standardized to the means and
standard deviations of all of the twins in the state. The range of
the Singh index for Washington state is 2.9 toþ 4.6, represent-
ing the least to the most deprived areas.
Analyses. Descriptive statistics were calculated as means and
standard deviations for continuous variables and percentages for
categorical variables. Descriptive statistics were also calculated
after stratifying the twins into a younger (ages 18–24) and older
(ages 25þ ) group. Because the UWTR has a lower age limit of 18
and is relatively young, these categories were designed to
separate twins who may still be living at their parental home or
in transient housing such as a college dormitory or apartment,
from twins who have left the household of origin. Independent
samples t-tests and
w
2
tests assessed differences on continuous
and categorical demographic characteristics, respectively.
We estimated the genetic and environmental influences on
walkability in two ways. First, we calculated twin correlations
with 95% confidence intervals for the walkability score stratified
by zygosity. Comparing the within-pair twin correlations provides
an initial indication of the genetic and environmental influence on
walkability. An MZ correlation that is larger than the DZ correla-
tion suggests a genetic influence on walkability, and a correlation
that is similar in MZ and DZ pairs indicates a role for common
family environment. The absence of correlations in either MZ or
DZ pairs suggests that the variation in walkability is due primarily
to unique environmental influences.
To help put the within-pair correlations into perspective, we
constructed a separate sample of ‘‘faux pairs’’ to estimate the
correlation of walkability scores for two unrelated individuals to
serve as a comparison. To construct this group, we randomly
selected 400 individual twins from our sample. Then, we ran-
domly selected another unrelated twin for each of the 400 twins
originally selected matched to the reference twin on age within
five years and sex. Thus, the faux pairs measured the correlation
in walkability for two unrelated individuals of the same age and
sex living in Washington state. For the descriptive and correlation
analyses, po 0.05 defined statistical significance.
To formally estimate genetic and environmental effects on
walkability we used structural equation modeling (SEM). A model, fit
to the raw data and stratified by age group, estimated the percentage
of phenotypic variance due to additive genetic (A), common envir-
onmental (C), and unique environmental (E) components. The full
ACE model, as well as reduced models (i.e., AE, CE, E), were t to the
data. The best-fitting model was determined by a likelihood ratio
w
2
test comparing the full ACE model to reduced models that did not
include all A, C, and E effects (e.g., CE). A non-statistically significant
w
2
test between the full and reduced models indicates that the
reduced model is a better, more parsimonious fit to the data. Akaike’s
information criterion was used as a global measure of goodness of
model fit, with lower values being an indication of a better-fitting
model (Akaike, 1987). To determine if geneti c and environmental
G.E. Duncan et al. / Health & Place 18 (2012) 515–519516
Page 2
influences on walkability were independent of neighborhood socio-
economic status, we refit our models after adjusting for Singh Index
scores. All SEM analyses were conducted using the program MPlus.
3. Results
Complete physical street addresses were available and geo-
coded for 1317 of 1389 twin pairs. As shown in Table 1, the two
groups differed significantly on all demographic characteristics
examined (all po 0.05). The older age group had a higher mean
walkability index than their younger counterparts (0.44 versus
0.76, p o 0.05). Mean walkability for all twins was 0.36
(range, 8.47–26.68), and the corresponding quartiles of walk-
ability were 6.76 (25th percentile), 1.88 (50th percentile), and
4.00 (75th percentile), respectively. Thus, the mean walkability
index of the older twins’ residences was in the third and younger
twins in the first quartile of the walkability distribution.
Among all twins, strong correlations were found for walkabil-
ity (po 0.05), while within-pair correlations were stronger for
MZs than DZs (Table 2). Within-pair correlations for older twins
were attenuated substantially compared with younger twins,
although all associations remained significant (po 0.05). As
expected the within-pair correlations for faux pairs were small
and non-significant (r¼ 0.03, 95% CI¼0.07–0.16; p¼ 0.57).
In younger twins, the best-fitting, most parsimonious model
for residential walkability index was the reduced CE model. In the
older twins the model-fit differences between the reduced AE and
CE models were trivial (i.e., did not provide a reliable means of
choosing between them) as were the differences between those
models and the full ACE model. Given that situation, the full ACE
model likely provides the best description of the data (Table 3).
In the younger age group (top panel), 65% of the phenotypic
variance in walkability was accounted for by the common
environment (C) and 35% by the unique environment
(E) components, respectively. In contrast, among older pairs
(bottom panel), the unique environment (E) component was
dominant, explaining 62% of the total variance in walkability
with smaller and equal (19%) contributions of additive genetic
(A) and common environment (C) factors.
Table 4 displays the genetic analyses controlling for the Singh
Index. In both age groups, the Singh Index was significantly
associated with walkability, but it only accounted for a small
portion ( 10%) of the total variance. In both younger and older
twins, the full ACE model was the best fit (for the older twins, we
observed the same trivial differences between the reduced AE and
CE models as were observed in the unadjusted models). In the
adjusted models, the greatest proportion of variance was
explained by the C component in the younger cohort (48%), and
Table 1
Demographic characteristics of individual twins by younger (18–24 years) and
older (25 þ years) age groups.
Characteristic All twins Younger
twins
Older
twins
n
(n¼ 2634) (n¼ 1763) (n¼ 871)
Age (years) 27.5
(14.3)
19.4 (1.3) 43.9 (14.5)
Sex (% Male) 39 42 34
Zygosity (% Monozygotic) 56 56 57
Race (% White) 85 82 92
Education (%)
Less than high school 10 13 6
High school graduate 34 44 18
Some college/College
graduate
55 43 76
Marital status (%)
Single 69 93 21
Married 18 2 51
Widowed/Divorced/
Separated
7 0.2 20
Living with someone 65 8
n
Twins less than 25 years of age differed significantly from twins 25 years of
age and older with respect to all demographic characteristics examined (po 0.05).
Age presented as the mean (standard deviation) while all other variables
presented as percentages.
Table 2
Correlations for neighborhood walkability among twin pairs by younger (18–24
years) and older (25 þ years) age groups and zygosity.
Zygosity Younger twins Older twins
(n¼ 878 pairs) (n¼ 432 pairs)
All 0.65 (0.61–0.69)
n
0.34 (0.04–0.42)
n
Monozygotic 0.68 (0.63–0.73)
n
0.39 (0.28–0.49)
n
Dizygotic 0.62 (0.55–0.68)
n
0.28 (0.14–0.40)
n
n
Indicates significant Pearson’s correlation (p o 0.05); coefficients presented
with 95% confidence interval in parentheses.
Table 3
Estimated genetic and environmental parameters for neighborhood walkability by age group.
Estimates of variance components
a
Test of model fit
Model
b
Additive
genetics (A)
Common
environment (C)
Unique
environment (E)
D
X
2
df
D
p
D
AIC
c
Younger twins (18–24 years)
ACE 0.12 (0.04, 0.26) 0.56 (0.47, 0.66) 0.32 (0.28, 0.36)
AE 0.70 (0.67, 0.73) 0.30 (0.27, 0.33) 53.23 1 o 0.001 51.23
CE 0.65 (0.62, 0.68) 0.35 (0.31, 0.38) 3.56 1 ¼ 0.06 1.56
E 1.00 492.81 2 o 0.001 488.81
0.5
Older twins (25–87 years)
ACE 0.19 (0.01,0.56) 0.19 (0.03, 0.50) 0.62 (0.54, 0.71)
AE 0.40 (0.33, 0.48) 0.60 (0.52, 0.68) 1.68 1 0.60 0.32
CE 0.34 (0.28, 0.42) 0.66 (0.59, 0.73) 1.32 1 0.25 0.68
E 1.00 55.38 2 o 0.001 51.38
a
Proportion of variance explained by additive genetics, common environment, and unique environment according to each model.
b
ACE refers to the model including additive genetics (A), common environment (C), and unique environment (E). AE includes only additive genetics and unique
environment, CE common and unique environment, and E unique environment; reduced models are compared to ACE.
c
Akaike’s information criterion (AIC) is a global measure of goodness of model fit; best-fitting and most parsimonious models are shown in bold.
G.E. Duncan et al. / Health & Place 18 (2012) 515–519 517
Page 3
the E component in the older cohort (51%), with a relatively small
additive genetic component in both age groups (13% and 20% in
younger and older twins, respectively).
4. Discussion
In unadjusted models, our results demonstrate that common
environment explains the greatest proportion of phenotypic
variation in neighborhood walkability among younger twins, with
no additive genetic and a small unique environmental influence.
In contrast, among older twins, unique environmental influences
dominate with smaller and equal contributions of additive
genetic and common environmental factors. Our results also
suggest that environmental factors are still strongly related to
neighborhood walkability even after controlling for the level of
neighborhood social deprivation, with a small contribution from
additive genetics. In fact, among younger twins, the effect of
additive genetics on residential selection appears non-significant
until social deprivation is accounted for in the model. However,
although the additive genetic effect is small and largely stable
over time, the relative contribution of unique environmental
experience to variability in residential selection increases drama-
tically over time compared to the common environment. Speci-
fically, among younger twins, common environment is most
important with a lesser contribution of unique environmental
factors and a small contribution of additive genetics. Among older
twins, walkability is more strongly influenced by environmental
factors that are unique to individuals with a smaller effect of
common environment and additive genetics.
The differential patterns of effects according to age are not
entirely surprising. The younger cohort (18–24 years) represents
the traditional college-age student in the US. Individuals in this age
group usually gain greater independence from their parents upon
graduation from high school, but have limited financial resources
and are much less likely to have started a family. Thus, younger
individuals still depend to some extent on the shared financial and
housing security provided by their parents. Twins at this age are
also more likely to live together (compared to older twins) which
could contribute to a greater shared environment component in
residential selection models. However, we repeated our analyses
excluding twins that shared the same address and obtained results
consistent with those presented in Tables 3 and 4. This suggests
that the stronger common environmental influence on neighbor-
hood walkability among younger twins is not due merely to an
abundance of twins living together. In contrast, individuals in the
older group (over 25 years) have typically graduated from college
and are beginning their careers and family. In this presumably
more settled, older group, unique environment is more powerful,
with only modest genetic and common environmental effect.
Differences between age groups in key demographic variables
support this notion; younger twins were much less likely to have
graduated from college and much more likely to be single com-
pared to older twins who were more likely to be married. In
addition, younger twins more often lived together than their older
counterparts, possibly explaining why the relative contribution of
genetic and environmental factors was so highly dependent on age.
Although the higher within-pair correlations in MZ than DZ
twins in both younger and older age groups suggest a genetic
influence, the magnitude of this difference was not significant.
The substantial attenuation of the correlations with age further
supports a diminishing genetic influence. Interestingly, the
within-pair correlations for walkability among faux pairs, who
by definition were unrelated and thus shared no genetic and
common environmental influence, was essentially zero and non-
significanteven though the faux pairs could be living in differ-
ent neighborhoods with the same environmental characteristics
(i.e., unique environment). In other words, we tested these age
and sex-matched faux pairs based on the possibility that age and
sex-matched people, at least in Washington State, might system-
atically end up in similar neighborhoods for reasons entirely
outside of familial confounding. However, because there was no
correlation among the faux pairs we can assume that similarities
in age and sex alone cannot explain the pattern of phenotypic
variation in neighborhood walkability we found among younger
and older twin pairs in the present study.
Our findings can be compared with those reported in the
Netherlands Twin Study (Willemsen et al., 2005). The Netherlan ds
study, which used an urbanization phenotype, found that a CE
model fit their data best in both younger and older age groups.
However, our results differ with respect to the magnitude of the
shift in C and E components with age; C was twice as strong as E
among younger twins, and E was twice as strong among older twins.
In contrast, the shifts were much more subtle among both men and
women over the three birth cohorts in the Netherlands sample.
Interestingly, the full ACE model instead of the reduced CE model
was the best fit among younger twins when we controlled for the
Singh index, providing evidence for a small genetic component.
This finding is similar to that observed in the youngest cohort of
Australian twins (Whitfield et al., 2005) which characterized
Table 4
Estimated genetic and environmental parameters for neighborhood walkability by age group, controlling for social deprivation score.
Estimates of variance components
a
Test of model fit
Model
b
Additive
genetics (A)
Common
environment (C)
Unique
environment (E)
b
Singh
D
X
2
df
D
p
D
AIC
c
Younger twins (18–24 years)
ACE 0.13 (0.04, 0.26) 0.48 (0.39, 0.58) 0.31 (0.27, 0.35) 0.08 (0.10, 0.27)
AE 0.63 (0.59, 0.66) 0.29 (0.26, 0.32) 0.08 (0.10, 0.27) 44.19 1 o 0.001 42.19
CE 0.58 (0.55, 0.62) 0.34 (0.30, 0.37) 0.08 (0.10, 0.27) 4.15 1 ¼ 0.04 2.15
E 0.90 (0.88, 0.93) 0.10 (0.07, 0.12) 451.05 2 o 0.001 447.05
0.5
Older twins (25–87 years)
ACE 0.20 (0.03, 0.50) 0.19 (0.04 0.44) 0.51 (0.44, 0.59) 0.10 (0.08, 0.13)
AE 0.40 (0.34,0.47) 0.50 (0.43, 0.57) 0.10 (0.08, 0.13) 2.07 1 0.15 0.07
CE 0.35 (0.28, 0.41) 0.55 (0.49, 0.59) 0.10 (0.08, 0.13) 2.00 1 0.16 0.001
E 0.91 (0.88, 0.94) 0.09 (0.06, 0.12) 69.21 2 o 0.001 65.21
a
Proportion of variance explained by additive genetics, common environment, and unique environment according to each model.
b
ACE refers to the model including additive genetics (A), common environment (C), and unique environment (E). AE includes only additive genetics and unique
environment, CE common and unique environment, and E unique environment; reduced models are compared to ACE.
c
Akaike’s information criterion (AIC) is a global measure of goodness of model fit; best-fitting and most parsimonious models are shown in bold.
G.E. Duncan et al. / Health & Place 18 (2012) 515–519518
Page 4
residential location using a measure of urbanization. The ACE model
was still the best fit among older twins even after controlling for the
Singh index, resulting in identical and equal contributions of
additive genetic and common environmental influences on neigh-
borhood walkability between the unadjusted and adjusted models.
This highlights that walkability is not simply a surrogate for
neighborhood material and social conditions and relative socio-
economic disadvantage but rather a unique phenotype.
This study adds to the existing literature in many respects.
Similar to the Australian study, our twins resided across a large
geographic area. Washington State covers over 71,000 square miles,
and has a total population over 6.7 million concentrated in a few
large and many small cities. Residential location categories used in
previous studies defined locations as urban, suburban, and non-
urban. We used an alternative measure of residential location
defined as neighborhood walkability, which characterizes the prox-
imal environment, including its residential density and ease of
access to local retail and businesses. Low walkability scores indicate
that the person lives away from neighbors and from commercial
activity, in an area that could be suburban, exurban or even rural,
while high walkability scores apply to urban or suburban neighbor-
hoods, or even small towns. In contrast, the Netherlands twin study
drew from a sample of comparatively highly urbanized twins. Even
the lowest of the five urbanization levels, which ranged from ‘‘very
heavy’’ to ‘‘none’’, likely would correspond to a suburban, but not
exurban, location in Washington state or Australia.
Understanding the factors influencing why people live in
different neighborhoods is a challenging task and, like others,
our study has limitations. For example, our SEM models estimated
genetic and environmental sources of variation in walkability but
cannot elucidate mechanisms. Thus, determining the exact nature
of the common and unique environmental contributions to
walkability and evaluating how these factors may be amenable
to intervention is difficult. Nonetheless, our results suggest that
environmental factors underlie the walkability of residential
location among twins living in the US. The relative contribution
of common or familial and unique environmental factors to
walkability depends on age, while shared genetic factors account
for minimal variance in walkability regardless of age.
Implications of these findings extend beyond merely explaining
proportions of variation in neighborhood walkability. For example,
multiple surveys and market research reports indicate that many
US households want housing with improved accessibility, land-use
mix, and diverse transportation options (Carnoske et al., 2010;
Russonello and Stewart: Research and Communications, 2004;
Levine and Frank, 2007; Litman, 2009), all of which contribute to
greater neighborhood walkability. In fact, the stock of such housing
may be far short of demand (Russonello and Stewart: Research and
Communications, 2004; Levine and Frank, 2007; Litman, 2009),
suggesting that governmental and political barriers to building
such communities need to be overcome (Carnoske et al., 2010;
Russonello and Stewart: Research and Communications, 2004).
Acknowledgment
This work was supported by RC2HL103416-02 (Buchwald). The
funding source had no role in the study design; in the collection,
analysis and interpretation of data; in the writing of the report;
and in the decision to submit the paper for publication.
References
Afari, N., Noonan, C., Goldberg, J., Edwards, K., Gadepalli, K., Osterman, B., Evanoff,
C., Buchwald, D., 2006. University of Washington Twin Registry: construction
and characteristics of a community-based twin registry. Twin Research And
Human Genetics 9, 1023–1029.
Akaike, H., 1987. Factor analysis and AIC. Psychometrika 52, 317–332.
Belden Russonello and Stewart: Research and Communications, 2004. 2004
National Community Preference Survey, Washington, D.C., pp. 1-55.
Carnoske, C., Hoehner, C., Ruthmann, N., Frank, L., Handy, S., Hill, J., Ryan, S., Sallis,
J., Glanz, K., Brownson, R., 2010. Developer and realtor perspectives on factors
that influence development, sale, and perceived demand for activity-friendly
communities. Journal of Physical Activity and Health 7 (Suppl 1), S48–59.
Eisen, S., Neuman, R., Goldberg, J., Rice, J., True, W., 1989. Determining zygosity in
the Vietnam Era Twin Registry: an approach using questionnaires. Clinical
Genetics 35, 423–432.
Frank, L.D., Saelens, B.E., Powell, K.E., Chapman, J.E., 2007. Stepping towards
causation: do built environments or neighborhood and travel preferences
explain physical activity, driving, and obesity? Social Science and Medicine 65,
1898–1914.
Frank, L.D., Schmid, T.L., Sallis, J.F., Chapman, J., Saelens, B.E., 2005. Linking
objectively measured physical activity with objectively measured urban form:
findings from SMARTRAQ. American Journal of Preventive Medicine 28,
117–125.
Handy, S., Cao, X.Y., Mokhtarian, P.L., 2006. Self-selection in the relationship
between the built environment and walkingEmpirical evidence from north-
ern California. Journal of the American Planning Association 72, 55–74.
Levine, J., Frank, L.D., 2007. Transportation and land-use preferences and residents’
neighborhood choices: the sufficiency of compact development in the Atlanta
region. Transportation 34, 255–274.
Litman, T.A., 2009. Where we want to be. Home location preferences and their
implications for smart growth. Victoria Transport Policy Institute, Victoria, B.C,
pp. 1-40.
Messer, L.C., 2007. Invited commentary: beyond the metrics for measuring
neighborhood effects. American Journal of Epidemiology 165, 868–871.
Neyer, F.J., 2002. Twin relationships in old age: a developmental perspective.
Journal of Social and Personal Relationships 19, 155–177.
Oakes, J.M., 2004. The (mis)estimation of neighborhood effects: causal inference
for a practicable social epidemiology. Social Science and Medicine 58,
1929–1952.
Oakes, J.M., 2006. Commentary: advancing neighbourhood-effects research–selec-
tion, inferential support, and structural confounding. International Journal of
Epidemiology 35, 643–647.
Oakes, J.M., Church, T.R., 2007. Invited commentary: advancing propensity score
methods in epidemiology. American Journal of Epidemiology 165, 1119–1121.
Singh, G.K., 2003. Area deprivation and widening inequalities in US mortality,
1969–1998. American Journal of Public Health 93, 1137–1143.
Torgersen, S., 1979. The determination of twin zygosity by means of a mailed
questionnaire. Acta Geneticae Medicae et Gemellologiae (Roma) 28, 225–236.
Whitfield, J.B., Zhu, G., Heath, A.C., Martin, N.G., 2005. Choice of residential
location: Chance, family influences, or genes? Twin Research and Human
Genetics 8, 22–26.
Willemsen, G., Posthuma, D., Boomsma, D.I., 2005. Environmental factors deter-
mine where the Dutch live: Results from the Netherlands Twin Register. Twin
Research and Human Genetics 8, 312–317.
G.E. Duncan et al. / Health & Place 18 (2012) 515–519 519
Page 5
    • "samples have documented the influence of shared environmental factors on residence (Whitfield et al., 2005; Willemsen et al., 2005), other studies have revealed subtleties in the data. For example, a more recent report (Duncan et al., 2012) using a continuous walkability index employing specific information about neighborhood urban form (Frank et al., 2005) observed that shared environmental factors were the largest contributor to residential selection among younger twins (ages 18–24.9) whereas unique environment was largest in older twins (ages 25+). "
    [Show abstract] [Hide abstract] ABSTRACT: Physical activity, neighborhood walkability, and body mass index (BMI, kg/m(2)) associations were tested using quasi-experimental twin methods. We hypothesized that physical activity and walkability were independently associated with BMI within twin pairs, controlling for genetic and environmental background shared between them. Data were from 6,376 (64% female; 58% identical) same-sex pairs, University of Washington Twin Registry, 2008-2013. Neighborhood walking, moderate-to-vigorous physical activity (MVPA), and BMI were self-reported. Residential address was used to calculate walkability. Phenotypic (non-genetically informed) and biometric (genetically informed) regression was employed, controlling for age, sex, and race. Walking and MVPA were associated with BMI in phenotypic analyses; associations were attenuated but significant in biometric analyses (Ps<0.05). Walkability was not associated with BMI, however, was associated with walking (but not MVPA) in both phenotypic and biometric analyses (Ps<0.05), with no attenuation accounting for shared genetic and environmental background. The association between activity and BMI is largely due to shared genetic and environmental factors, but a significant causal relationship remains accounting for shared background. Although walkability is not associated with BMI, it is associated with neighborhood walking (but not MVPA) accounting for shared background, suggesting a causal relationship between them. Copyright © 2014. Published by Elsevier Inc.
    Full-text · Article · Dec 2014 · Preventive Medicine
    0Comments 2Citations
    • "This, in a nutshell, is the selection problem. Selection problem can also refer to the possibility that individuals select residential environments based on genetics or family upbringing (Duncan et al., 2012; Whitfield et al., 2005; Willemsen et al., 2005). The selection variables, rather than any putative environmental effects, may be responsible for findings that link environmental characteristics to health outcomes. "
    [Show abstract] [Hide abstract] ABSTRACT: No causal evidence is available to translate associations between neighborhood characteristics and health outcomes into beneficial changes to built environments. Observed associations may be causal or result from uncontrolled confounds related to family upbringing. Twin designs can help neighborhood effects studies overcome selection and reverse causation problems in specifying causal mechanisms. Beyond quantifying genetic effects (i.e., heritability coefficients), we provide examples of innovative measures and analytic methods that use twins as quasi-experimental controls for confounding by environmental effects. We conclude that collaboration among investigators from multiple fields can move the field forward by designing studies that step toward causation.
    Full-text · Article · May 2014 · Health & Place
    0Comments 5Citations
  • [Show abstract] [Hide abstract] ABSTRACT: Socioeconomic status and other socio-demographic factors have been associated with selective residential mobility across rural and urban areas, but the role of psychological characteristics in selective migration has been studied less. The current study used 16-year longitudinal data from the U.S. National Longitudinal Survey of Youth 1979 (NLSY79) to examine whether cognitive ability assessed at age 15–23 predicted subsequent urban/rural migration between ages 15 and 39 (n = 11,481). Higher cognitive ability was associated with selective rural-to-urban migration (12 percentile points higher ability among those moving from rural areas to central cities compared to those staying in rural areas) but also with higher probability of moving away from central cities to suburban and rural areas (4 percentile points higher ability among those moving from central cities to suburban areas compared to those staying in central cities). The mobility patterns associated with cognitive ability were largely but not completely mediated by adult educational attainment and income. The findings suggest that selective migration contributes to differential flow of cognitive ability levels across urban and rural areas in the United States.
    No preview · Article · Sep 2014 · Intelligence
    0Comments 4Citations
Show more

Discover cutting-edge research

ResearchGate is where you can find and access the latest publications from your field of research.

Discover more