Content uploaded by Minkyung Koo
Author content
All content in this area was uploaded by Minkyung Koo on Jun 11, 2019
Content may be subject to copyright.
American Psychologist
The Socioecological Psychology of Upward Social
Mobility
Shigehiro Oishi, Minkyung Koo, and Nicholas R. Buttrick
Online First Publication, December 17, 2018. http://dx.doi.org/10.1037/amp0000422
CITATION
Oishi, S., Koo, M., & Buttrick, N. R. (2018, December 17). The Socioecological Ps ychology of
Upward Social Mobility. American Psychologist. Advance online publication.
http://dx.doi.org/10.1037/amp0000422
The Socioecological Psychology of Upward Social Mobility
Shigehiro Oishi
Columbia University
Minkyung Koo
University of Illinois at Urbana–Champaign
Nicholas R. Buttrick
University of Virginia
Intergenerational upward economic mobility—the opportunity for children from poorer
households to pull themselves up the economic ladder in adulthood—is a hallmark of a just
society. In the United States, there are large regional differences in upward social mobility.
The present research examined why it is easier to get ahead in some cities and harder in
others. We identified the “walkability” of a city, how easy it is to get things done without a
car, as a key factor in determining the upward social mobility of its residents. We 1st
identified the relationship between walkability and upward mobility using tax data from
approximately 10 million Americans born between 1980 and 1982. We found that this
relationship is linked to both economic and psychological factors. Using data from the
American Community Survey from over 3.66 million Americans, we showed that residents
of walkable cities are less reliant on car ownership for employment and wages, significantly
reducing 1 barrier to upward mobility. Additionally, in 2 studies, including 1 preregistered
study (1,827 Americans; 1,466 Koreans), we found that people living in more walkable
neighborhoods felt a greater sense of belonging to their communities, which is associated
with actual changes in individual social class.
Keywords: upward mobility, walkability, social ecology
Supplemental materials: http://dx.doi.org/10.1037/amp0000422.supp
The United States has always been “the land of opportu-
nity,” the place where, if people work hard and play by the
rules, they’ll get ahead (Hochschild, 1996). Upward eco-
nomic mobility is a valued goal shared widely among
Americans (Davidai & Gilovich, 2015; Kraus & Tan, 2015).
However, American optimism appears to be in decline
(Aaronson & Mazumder, 2008; Stephens, Markus, & Phil-
lips, 2014). A recent New York Times poll, for instance,
showed that over 30% of Americans now feel that the
American dream is out of reach—the most pessimism since
The New York Times started asking the question in 1996
(Sorkin & Thee-Brenan, 2014). Previous psychological re-
search on upward mobility has centered on the importance
of internal individual factors such as intelligence, skills, and
motivation, generally finding that being smart and moti-
vated helps people climb up the economic ladder (e.g.,
Deary et al., 2005; Snarey & Vaillant, 1985). The present
research, by contrast, takes a socioecological approach,
which explores the interrelationship between people and
their lived ecologies (Oishi, 2014; Stokols, 1992; Yamagi-
shi, 2011); here we investigate linkages between the built
environment and actual upward mobility.
Regional Variations in Upward Social Mobility in
the United States
Although upward social mobility is generally in decline in
the United States (Chetty et al., 2017), it is easier to get
ahead in some parts of the United States than in others.
Using comprehensive tax return data, Chetty, Hendren,
Kline, and Saez (2014) found that parts of the country are
still fluid—in some areas, such as Pittsburgh, Pennsylvania,
Shigehiro Oishi, Department of Psychology, Columbia University;
Minkyung Koo, Department of Business Administration, University of
Illinois at Urbana–Champaign; Nicholas R. Buttrick, Department of Psy-
chology, University of Virginia.
We thank Hyewon Choi for providing us with various individual differ-
ence measures used in Studies 3 and 4 and Jordan Axt, Hyewon Choi,
Samantha Heintzelman, Kostadin Kushlev, Brandon Ng, Lorien Rice, Jane
Tucker, Erin Westgate, Tim Wilson, and Nadia Danienta for their assis-
tance with data collection, comments, and feedback.
Correspondence concerning this article should be addressed to Shigehiro
Oishi, Department of Psychology, Columbia University, 406 Schermer-
horn Hall, 1190 Amsterdam Avenue, New York, NY 10027. E-mail:
shigeoishi@gmail.com
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
American Psychologist
© 2018 American Psychological Association 2018, Vol. 1, No. 999, 000
0003-066X/18/$12.00 http://dx.doi.org/10.1037/amp0000422
1
the odds of reaching the top fifth of income in young
adulthood (around age 30) for those growing up in house-
holds from the bottom income quintile is equal to that in the
most mobile countries in the world. However, in the least
fluid areas, such as Charlotte, North Carolina, the odds of
rising are three times worse, lower than in any developed
country that the authors have data for. Chetty et al. (2014)
proposed five factors to explain these regional differences:
the area’s racial makeup, the level of income inequality, the
quality of the K–12 school system, the strength of social
capital (measured by voter turnout, the percentage of people
who returned their census forms, and various measures of
community participation), and the percentage of children
living in homes with single parents. These five factors
account for a substantial amount of regional variations in
upward mobility.
In this article, we identify a new predictor of economic
mobility: the way in which cities are organized. We propose
that the walkability of one’s area is an important predictor
of intergenerational upward mobility. We define walkability
as how easily people can live their lives on foot or using
public transportation—in highly walkable areas, people can
go to work, for example, or to their local grocery store
without needing a car. In contrast, in less walkable areas,
cars are needed for practically every task. In urban planning,
geography, and transportation research, walkability is typi-
cally measured by physical characteristics such as intersec-
tion density and street connectivity, as well as land use (e.g.,
mixed residential and commercial use) and dwelling density
(see Frank et al., 2006). Walkability is associated with urban
vibrancy and recreational opportunities (Forsyth, 2015), and
walkable cities tend to have better public transportation than
do less walkable cities.
Why Walkability Matters
Walkability may be associated with higher upward mo-
bility for several reasons. The requirement of car ownership
in less walkable cities is a major barrier to the job market for
anyone without the means to afford one (Ong & Blumen-
berg, 1998; Raphael & Rice, 2002). By reducing the need
for a car, a more walkable city opens its employment
possibilities up to a far wider range of prospective employ-
ees than in a less walkable city. Thus, the first reason why
we think walkability is associated with upward social mo-
bility is increased access to jobs.
Another pathway to increased upward mobility may run
through improved physical health. People living in a walk-
able city tend to be healthier (Frank et al., 2006; Sallis et al.,
2009; Todd et al., 2016; Van Cauwenberg, Van Holle, De
Bourdeaudhuij, Van Dyck, & Deforche, 2016). Healthy
people are able to work longer hours or on multiple jobs and
are more likely to move up the economic ladder over time
than are those less healthy (Power, Matthews, & Manor,
1996). In addition, walking is associated with better aca-
demic achievement (e.g., Hillman et al., 2009). To the
extent that the walkability of a city is associated with the
mean level of physical fitness of its residents and walking is
associated with better academic achievement, walkable cit-
ies might have higher levels of upward social mobility due
in part to physical fitness and academic achievement.
A third pathway is more psychological. In a walkable
city, one expects that people from lower socioeconomic
strata are more likely to feel a sense of belonging and a
sense of place than they would in an unwalkable city. In an
unwalkable city, people with lesser means, including re-
duced access to transportation, find it harder to get around
and thus have more limited access to the city as a whole.
Unable to reach the totality of the city, they might not
necessarily feel a sense of belonging to the city at large or
feel that the whole city is their city. A limited sense of
belonging might preclude people from lower socioeconomic
status (SES) from applying to jobs in certain parts of the
city. In contrast, people in a walkable city, able to get
wherever they wish to go, might feel a broader sense of
place and a stronger sense of belonging and therefore apply
to jobs in most parts of the city. Feeling like one belongs in
a place has been shown to have positive effects on motiva-
tion and accomplishment (e.g., Allen, Kern, Vella-Brodrick,
Hattie, & Waters, 2018; Baumeister & Leary, 1995; Yeager
et al., 2016). A sense of belonging is motivating especially
among those in disadvantaged circumstances, because it
makes people feel that they fit in a community, that their
struggle is fairly common to others, and that there are
people who will support their efforts (e.g., Shnabel, Purdie-
Vaughns, Cook, Garcia, & Cohen, 2013; Walton & Cohen,
2011). If people feel a great sense of belonging to their city,
they are likely to apply for jobs outside their immediate
neighborhoods and feel that they could work there. In con-
trast, if people do not feel a sense of belonging to the city,
they might feel that they do not fit in and might not apply to
jobs, for instance, in the downtown.
The Current Research
The present article tested whether walkability was posi-
tively related to upward social mobility and investigates
potential mechanisms. Consistent with our hypotheses, a
recent study found that higher degrees of urban sprawl (e.g.,
percentage of the population living in low-density suburban
developments) was negatively associated with upward so-
cial mobility across 122 American commuting zones (CZs;
Ewing, Hamidi, Grace, & Wei, 2016). We expanded on that
work in our Study 1. Using a broader data set of 389
commuting zones, with a more focused definition of walk-
ability and with tighter controls, we tested the role of
walkability above and beyond Chetty et al.’s (2014) five
factors, as well as other related variables. Study 2 used
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
2OISHI, KOO, AND BUTTRICK
individual-level data from the American Community Sur-
vey (ACS) to examine whether car ownership plays a role in
the link between walkability and upward mobility. Follow-
ing the research strategies of socioecological psychology
(Oishi, 2014), the next two studies looked to individual
psychology as a pathway between the environment and its
related outcomes. We investigated one’s sense of belonging
as a key translational mechanism between the ease of get-
ting around and one’s ability to climb the socioeconomic
ladder.
Study 1
In Study 1, we established the link between walkability
and upward social mobility using the earning records from
all American citizens born between 1980 and 1982 whose
parents filed taxes (Chetty et al., 2014).
Method
Income, demographic, and city-level covariate data for
Study 1 largely came from a data set put together by Chetty
et al. (2014), available at Equality-of-Opportunity.org. Data
on commuting-zone-area walkability came from www
.walkscore.com. Commuting-zone-level voting data were
adapted from Leip (2012). Commuting-zone level longevity
at age 40 for the lowest income quarter was taken from
https://healthinequality.org/data/.
Intergenerational upward economic mobility was opera-
tionalized as the probability of reaching the top fifth of
income in young adulthood (at age 30) for those coming
from the households in the bottom fifth of the income
distribution. Following Chetty et al. (2014), we used the
commuting zone of each individual as the grouping variable
of choice. Commuting zones (CZs) are “geographical ag-
gregations of counties that are similar to metro areas but
cover the entire United States” (Chetty et al., 2014, p. 1555).
Because CZs ranged in population size in 2000 from 1,193
(Murdo, South Dakota) to 16,393,360 (Greater Los Ange-
les, California), we weighted each of the 741 commuting
zones in our regression analyses by population, to account
for the spread of populations within the sample: 62 zones
had less than 10,000 residents, whereas 62 zones had more
than 1 million residents. Ordinary least-squares regression
(without weighting) would weigh observations with less
than 10,000 residents and those with more than 1 million
residents equally. To infer totals for the United States as a
whole, it is therefore important to weigh observations by
their populations.
Our walkability data came from www.walkscore.com, a
well-validated measure of the walkability of an area (Carr,
Dunsiger, & Marcus, 2011; Duncan, Aldstadt, Whalen,
Melly, & Gortmaker, 2011). A Walk Score is computed
based on access to various amenities (e.g., restaurant, bank,
post office) and physical factors such as population density,
block length, and intersection density. Walk Scores range
from 0 to 100. We were able to obtain walkability scores for
389 commuting zones, which contained 8.98 million indi-
viduals for whom we had intergenerational mobility infor-
mation.
Results and Discussion
First, the population weighted simple regression showed
that upward social mobility was substantially higher in more
walkable commuting zones than in less walkable commut-
ing zones (b⫽.00050, SE ⫽.000060, ⫽.39), t(387) ⫽
8.33, p⬍.001, effect size r⫽.390.
Furthermore, a weighted multiple regression showed that
the association between walkability and upward mobility
remained significant even after entry of the factors previ-
ously found to impact upward social mobility (Chetty et al.,
2014)—percentage of African Americans, degree of income
inequality, quality of K–12 education, social capital, and
percentage of children with single mothers (the Five Fac-
tors). The effect size for walkability was substantial: Walk-
ability explained 11% of additional variance uniquely be-
yond the previously identified Five Factors (R
2
with
walkability ⫽.52, R
2
without walkability ⫽.41; b⫽
.00049, SE ⫽.00053, ⫽.389), t(376) ⫽9.278, p⬍.001,
⌬R
2
⫽.11.
Alternate explanations. There are several potential al-
ternate explanations for our primary findings just discussed.
First, walkable cities might be more politically liberal than
are less walkable cities, and liberal policies (e.g., more
generous welfare) in walkable cities might be responsible
for the association between walkability and upward social
mobility. Thus, we ran another weighted regression predict-
ing upward mobility from walkability, the Five Factors, and
the percentage of voters in a commuting zone who voted for
the Democratic candidate in the 1996 presidential election
(when these participants were in their teens, because the
political climates when they were growing up would be
more relevant to their ultimate economic mobility as adults
than would the current political climate
1
). Walkability re-
mained a significant predictor above and beyond the Five
Factors and the percentage of Clinton voters in 1996 (⫽
.258), t(375) ⫽4.873, p⬍.001, ⌬R
2
⫽.029.
Walkable cities are healthier. Indeed, longevity estimates
for the poorest quarter of the population was longer in more
walkable cities, r(386) ⫽.335, p⬍.001. The older that
1
It should be noted that walkability effect remained significant even
when we used election data from different years. Controlling for the
percentage of Democratic votes, we found these results for the following
years: in 2012 (⫽.369), t(375) ⫽6.505, p⬍.001; in 2008 (⫽.393),
t(375) ⫽7.048, p⬍.001; in 2004 (⫽.338), t(375) ⫽5.946, p⬍.001;
in 2000 (⫽.311), t(375) ⫽5.499, p⬍.001; in 1992 (⫽.291), t(375) ⫽
6.225, p⬍.001; in 1988 (⫽.263), t(375) ⫽5.789, p⬍.001; and in 1984
(⫽.285), t(375) ⫽6.055, p⬍.001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
3
WALKABILITY AND UPWARD MOBILITY
residents from the poorest quarter of a city were expected to
live, the more likely that children of the poor parents moved
up the economic ladder as adults, r(368) ⫽.348, p⬍.001.
Thus, we next ran another weighted regression predicting
upward mobility from walkability, the Five Factors, and
longevity. Walkability remained a significant predictor
above and beyond the Five Factors and longevity (⫽
.200), t(374) ⫽4.275, p⬍.001, ⌬R
2
⫽.020.
Walkable cities are also different from less walkable
cities in terms of economic conditions, labor structure,
education spending, religiosity, and various other factors.
To test robustness of the walkability findings, we conducted
a series of additional weighted multiple regression analyses,
controlling for the Five Factors plus household median
income, percentage of religious people, share of manufac-
turing as a source of employment, state income tax progres-
sivity, local government expenditure per capita, and violent
crime rate. Controlling for all these variables simultane-
ously, walkability still remained a significant predictor of
upward social mobility (b⫽.000314, SE ⫽.000066, ⫽
.239), t(348) ⫽4,767, p⬍.001, ⌬R
2
⫽.023. Overall, the
walkability of an area was a robust predictor of upward
social mobility beyond factors previously used to explain
upward mobility as well as other potential third variables
such as political culture, economic conditions, and labor
structure.
Propensity score matching analysis. In addition to
these analyses, we conducted a propensity score matching
analysis, an econometric method to strengthen the possible
causal inferences from observational data by accounting for
potential systematic differences in selected baseline charac-
teristics between groups. Propensity score analyses use a
procedure that identifies pairs of cases (one treated and one
untreated) in the data with otherwise matched baseline
characteristics (Rubin & Thomas, 1996). These paired sets
are then subjected to a ttest to determine whether the
treatment (in this case, walkability) is related to the outcome
(in this case, upward intergenerational economic mobility)
even after the baseline characteristics are matched away.
We chose to match our commuting zones on their popula-
tion, urbanity (coded as urban or not urban; taken from
Chetty et al., 2014), and number of historic buildings (be-
cause cities with more historic properties [defined as those
registered as historic sites by the National Park Service,
n.d.] are likely older and have a denser urban core) as
proxies for baseline urban structure to better isolate the
effect of walkability specifically. We first created a propen-
sity score using a logistic regression analysis, in which the
dichotomized walkability score was regressed on popula-
tion, urbanity, and the number of historic buildings. Out of
377 CZs that had the data on walkability and three matching
variables, we were able to find 125 pairs of CZs (i.e., 250
CZs) that were matched on our three variables (propensity
score threshold ⬍.10) but differed in terms of walkability.
As predicted, a paired ttest showed that walkable cities
(M⫽.0978, SE ⫽.0313) had higher upward social mobility
than did matched unwalkable cities (M⫽.0837, SE ⫽
.0404), t(124) ⫽3.539, p⫽.001, d⫽.321. Furthermore, a
general linear model analysis, in which upward social mo-
bility was the within-factor and the Five Factors were co-
variates, showed that upward social mobility was higher for
walkable than for matched unwalkable cities, additionally
controlling for the Five Factors, F(1, 116) ⫽4.143, p⫽
.044, ⌬R
2
⫽.034. Thus, the propensity score matching
analyses also showed that upward social mobility is higher
in walkable than in less walkable cities.
2
Using tax data from almost nine million Americans born
between 1980 and 1982, Study 1 demonstrates that upward
social mobility is substantially higher in more walkable
areas (r⫽.390). The more walkable an area is (as indexed
by Walkscore.com), the more likely Americans whose par-
ents were in the lowest income quintile are to have reached
the highest income quintile by their 30s. This relationship
holds above and beyond factors previously used to explain
upward mobility, factors such as income inequality and
social capital, and is robust to various political, economic,
and demographic controls; to alternate specifications of
upward mobility; and to potentially unspecified third vari-
ables.
Study 2
In Study 2, using data from the American Community
Survey (ACS) on over 3.66 million Americans, we exam-
ined one potential mechanism for the association between
walkability and upward social mobility: the possibility that,
in more walkable cities, a car is less important for finding a
good job.
Method
We used data from the 2009 –2013 ACS, a product of the
U.S. Census Bureau that tracks various demographic, housing,
economic, and social indicators from a broad, representational
sample of the American population (https://www.census.gov/
programs-surveys/acs/), supplemented with data from the U.S.
Department of Labor, Bureau of Labor Statistics (2009-2013).
We have complete data for over 3.66 million Americans from
305 metro areas (walkability scores for each metro area come
from www.walkscore.com).
Results and Discussion
To test our hypothesis that employment is less dependent
on car ownership in walkable cities than in less walkable
2
We additionally conducted an instrumental variable analysis, using the
number of historical buildings in a commuting zone as our instrument of
choice. The results are consistent with the other analyses here and can be
found in the online supplemental materials.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
4OISHI, KOO, AND BUTTRICK
cities, we conducted a multilevel analysis (Level 1 ⫽re-
spondents; Level 2 ⫽cities), using HLM 6.04 (Rauden-
bush, Bryk, & Congdon, 2004). On average, car ownership
was associated with a 1.16 increase in log (p/[1 ⫺p])
employment, t(5,402,596) ⫽161.85, p⬍.001, odds ratio
(OR)⫽3.17, 95% confidence interval (CI) [3.13, 3.22], or
a 26.04% employment advantage over those without a car
(see Table 1). However, as predicted, the association be-
tween car ownership and employment status was signifi-
cantly smaller in walkable cities than in less walkable cities
(b⫽⫺.011, SE ⫽.00025), t(5,402,596) ⫽⫺43.34, p⬍
.001, OR ⫽.989, 95% CI [.988, .989]. In a city with a
walkability score 1 SD above the mean, car ownership was
associated with a 1.00 increase in log (p/[1 ⫺p]) employ-
ment (a 23.03% employment advantage), whereas in a city
with a walkability score 1 SD below the mean, car owner-
ship was associated with a much larger 1.31 increase in log
(p/[1 ⫺p]) employment (a 28.82% employment advantage).
Employment status, however, is also associated with var-
ious individual factors (e.g., age, years of education), and
the link between car ownership and employment may plau-
sibly be moderated by city-level factors such as (a) the
city’s cost of living (indexed by the median income of the
city), in that it may be more expensive for everyone to own
a car, thus dampening the relationship between car owner-
ship and employment, or (b) the citywide employment rate:
Where there is less competition for any given job, it may be
easier to get a job without a car. Thus, in the next analysis,
we statistically controlled, at the individual level, for gen-
der, race, age, years of education, student status, and the
presence of an infant at home, as well as for commuting-
zone-level population size, median income, and unemploy-
ment rate. The results were largely unchanged: Car owner-
ship was associated with a .81 increase in log (p/[1 ⫺p])
employment, a 19.15% employment advantage (SE ⫽
.0107), t(3,664,221) ⫽74.97, p⬍.001, OR ⫽2.24, 95% CI
[2.19, 2.29], which, again, was moderated by walkability
(b⫽⫺.0083, SE ⫽.00046), t(3,664,221) ⫽⫺17.80, p⬍
.001, OR ⫽.992, 95% CI [991, .993]. That is, controlling
for a host of variables, car ownership was associated with a
21.60% employment advantage in a less walkable city (⫺1
SD), whereas car ownership was associated with a 16.59%
employment advantage in a walkable city (⫹1SD).
Employment itself, though necessary for economic mo-
bility, is far from sufficient; one must get a well-paying job,
not just any job. Thus, we next tested whether wages are
Table 1
Employment Status Predicted From Car Ownership and Other Individual-Level Controls, as
Well as Walkability and Other City-Level Controls in Study 2
Variable b(SE)tdfp
For INTRCPT1, B0
INTRCPT2, G00 1.428832 (.014313) 99.825 297 .000
POP, G01 ⫺.000482 (.000381) ⫺1.265 297 .207
WALK, G02 .009716 (.000858) 11.323 297 .000
UNEMPLOY, G03 ⫺.054288 (.005503) ⫺9.864 297 .000
MED, G04 .008440 (.001634) 5.165 297 .000
For FEMALE slope, B1
INTRCPT2, G10 .079304 (.003868) 20.503 3664221 .000
For MARRIED slope, B2
INTRCPT2, G20 .574766 (.004418) 130.089 3664221 .000
For BLACK slope, B
INTRCPT2, G30 ⫺.620181 (.005222) ⫺118.752 3664221 .000
For HISPANIC slope, B4
INTRCPT2, G40 .010814 (.005439) 1.988 3664221 .047
For EDUCATION slope, B5
INTRCPT2, G50 .102306 (.000498) 205.377 3664221 .000
For IN SCHOOL slope, B6
INTRCPT2, G60 ⫺.300917 (.005121) ⫺58.758 3664221 .000
For AGE slope, B7
INTRCPT2, G70 .020626 (.000198) 104.365 3664221 .000
For INFANT at Home slope, B8
INTRCPT2, G80 ⫺.109492 (.007150) ⫺15.314 3664221 .000
CAR slope, B9
INTRCPT2, G90 .806842 (.010762) 74.971 3664221 .000
POP, G91 .000406 (.000291) 1.395 3664221 .163
WALK, G92 ⫺.008248 (.000463) ⫺17.802 3664221 .000
UNEMPLOY, G93 ⫺.028971 (.004061) ⫺7.134 3664221 .000
MED2010I, G94 ⫺.008017 (.001098) ⫺7.303 3664221 .000
Note. Individual level predictors of employment status (B1 ⫽gender; B2 ⫽marital status; B3 ⫽being African
American; B4 ⫽being Hispanic American; B5 ⫽the highest level of education; B6 ⫽being currently in school;
B7 ⫽age; B8 ⫽having an infant at home; B9 ⫽owning a car. INTRCPT ⫽Intercept; G ⫽commuting zone
level predictors; POP ⫽population; UNEMPLOY ⫽Unemployment rate; MED ⫽Median income in 2010.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
5
WALKABILITY AND UPWARD MOBILITY
less dependent on car ownership in walkable than in un-
walkable cities. We further analyzed the data from respon-
dents who reported having a job and received a wage (we
excluded respondents who said they had a job but reported
having zero wages).
First, we ran the simplest model, in which one’s wage
(log-transformed) was predicted from car ownership at
Level 1 (within city), and the intercept and slope for car
ownership were predicted from walkability at Level 2 (be-
tween cities). Not surprisingly, car ownership was associ-
ated with a higher wage (b⫽.59, SE ⫽.012),
t(4,637,163) ⫽155.17, p⬍.001 (see Table 2). Given the
intercept was 9.63, the coefficient of .59 here translates into
car owners’ annual wage being roughly $27,406 as opposed
to non-car owners’ annual wage of roughly $15,214, an
approximately $12,000 advantage. It is important to note
that the car ownership advantage was significantly smaller
in walkable than in less walkable cities (b⫽⫺.0049, SE ⫽
.00012), t(4,637,163) ⫽⫺40.81, p⬍.001. In a less walk-
able city (⫺1SD), the wage advantage for car owners was
roughly $14,000, whereas in a walkable city (⫹1SD), the
wage advantage for car owners was smaller, roughly only
$10,000. Using the same set of controls as in the analysis of
employment, the moderation role of walkability remained
significant (b⫽⫺.0038, SE ⫽.00018), t(3,161,012) ⫽
⫺21.01, p⬍.001.
Using data from approximately 3.66 million Americans,
we found that car ownership was a higher barrier to entry
into the job market in less walkable cities. Even after
controlling for a host of demographics, we found that own-
ing a car was more important for employment in a less
walkable city than in a more walkable one and that non-car
owners in more walkable cities were less disadvantaged in
their average wages than were non-car owners in less walk-
able cities. Study 2’s findings suggest that a reason why
children from low-income families living in walkable cities
had a better chance of moving up an economic ladder as
adults is that they did not have to rely on a car as much as
did those living in less walkable cities.
Study 3
In the first two studies, we showed that walkability pre-
dicts upward social mobility and that the employment status
and wages of residents are less dependent on car ownership
in walkable than in less walkable cities. In the next two
Table 2
Annual Wage (Log-Transformed) Predicted From Car Ownership and Other Individual-Level
Controls, as Well as Walkability and Other City-Level Controls in Study 2
Variable b(SE)tdfp
For INTRCPT1, B0
INTRCPT2, G00 10.097962 (.006583) 1533.864 297 .000
POP, G01 .000605 (.000174) 3.469 297 .001
WALK, G02 .004800 (.000412) 11.651 297 .000
UNEMPLOY, G03 ⫺.000220 (.002595) ⫺.085 297 .933
MED, G04 .011275 (.000757) 14.892 297 .000
For CAR slope, B1
INTRCPT2, G10 .209253 (.004639) 45.104 3161012 .000
POP, G11 ⫺.000710 (.000119) ⫺5.988 3161012 .000
WALK, G12 ⫺.003759 (.000179) ⫺21.006 3161012 .000
UNEMPLOY, G13 .000039 (.001731) .022 3161012 .982
MED, G14 .000244 (.000419) .583 3161012 .560
For FEMALE slope, B2
INTRCPT2, G20 ⫺.380611 (.001040) ⫺365.820 3161012 .000
For MARRIED slope, B3
INTRCPT2, G30 .255015 (.001187) 214.865 3161012 .000
For BLACK slope, B4
INTRCPT2, G40 ⫺.114753 (.001724) ⫺66.582 3161012 .000
For HISPANIC slope, B5
INTRCPT2, G50 ⫺.065245 (.001581) ⫺41.264 3161012 .000
For EDUCATION slope, B6
INTRCPT2, G60 .110821 (.000162) 684.708 3161012 .000
For IN SCHOOL slope, B7
INTRCPT2, G70 ⫺.693620 (.001622) ⫺427.594 3161012 .000
For AGE slope, B8
INTRCPT2, G80 .024229 (.000053) 456.878 3161012 .000
For INFANT at Home slope, B9
INTRCPT2, G90 .045585 (.002060) 22.133 3161012 .000
Note. Individual level predictors of employment status (B1 ⫽gender; B2 ⫽marital status; B3 ⫽being African
American; B4 ⫽being Hispanic American; B5 ⫽the highest level of education; B6 ⫽being currently in school;
B7 ⫽age; B8 ⫽having an infant at home; B9 ⫽owning a car. INTRCPT ⫽Intercept; G ⫽commuter zone
level predictors; POP ⫽population; UNEMPLOY ⫽unemployment rate; MED ⫽median income.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
6OISHI, KOO, AND BUTTRICK
studies, we looked at psychological differences between
people living in more and less walkable neighborhoods and
at how those differences were associated with upward social
mobility. In Study 3, we examined whether a sense of
belonging is such a mechanism.
Method
Participants. This study was presented to participants in
a package with other studies; however, all relevant materials
are reported here and in the online supplemental materials.
This study, and the study following, were approved by the
University of Illinois Institutional Review Board (Protocol No.
16,188). Although we originally aimed to recruit 750 partici-
pants, an opportunity arose to double our collection, and we
took advantage of that opportunity before looking at our data.
Our final sample was 1,827 participants
3
(53% female; M
age
⫽
42.24, SD ⫽12.29). Participants were recruited from a nation-
ally representative panel of Americans maintained by Light-
speed GMI (the preregistration for this study can be found at
https://osf.io/4mv3t/register/565fb3678c5e4a66b5582f67?
view_only⫽d4bd7df3065b42679a57db30b57f18ea).
Materials and procedure. We measured participants’
perceived walkability of their current place of residence and
the surrounding areas using one dimension (land-use mix
diversity) of a widely used self-report measure of walkabil-
ity, the Neighborhood Environment Walkability Scale—
Abbreviated (NEWS–A; Cerin, Saelens, Sallis, & Frank,
2006). Assessment of land-use mix diversity was chosen
whereas other factors, such as street connectivity, were
excluded because this dimension best captured our defini-
tion of walkability (ability to walk to get things done in
everyday life). Participants reported whether they could
walk to each of nine places—to a job, school, supermarket,
restaurant, gym or fitness center, library, post office, park,
coffee or tea place, and bank or ATM machine— on a
5-point scale (1 ⫽less than 5 min walk, 2⫽6 –10 min walk,
3⫽11–20 min walk, 4⫽21–30 min walk, 5⫽over 30 min
walk or cannot walk). Results were reverse-scored so that
higher scores indicated greater walkability (M⫽2.33,
SD ⫽.99; ␣⫽.89). We also asked participants to indicate
how often they walked to each of these places on a 5-point
scale (1 ⫽almost always, 2⫽usually, 3⫽occasionally,
4⫽rarely, 5⫽not at all; M ⫽1.81, SD ⫽1.04; ␣⫽.94).
We reverse-scored the results so that higher scores indicated
more walking.
We measured participants’ sense of belonging by asking
them on a 5-point scale ranging from 1 (strongly disagree)
to5(strongly agree) the degree to which they agreed with
the following item: “I feel a sense of belonging in my
community” (Su, Tay, & Diener, 2014; M⫽3.31, SD ⫽
1.18).
Upward social mobility was obtained by comparing the
current SES reported by participants with that reported by their
parents when participants were growing up, measured on a
5-point scale (1 ⫽lower/working, 2⫽lower middle, 3⫽
middle, 4⫽upper middle, 5⫽upper). Parents’ social class
rating was subtracted from participants’ current social class,
and this score was used as an index for upward social mobility,
with higher scores indicating greater actual mobility
(M⫽⫺.08, SD ⫽1.05). Demographic questions including
gender, age, race, highest level of education, employment
status, and annual household income were asked at the end of
Study 3. All items can be found at https://osf.io/4mv3t/register/
565fb3678c5e4a66b5582f67?view_onlyd4bd7df3065b42679a
57db30b57f18ea.
Results and Discussion
As expected, perceived walkability and actual frequency of
walking were positively correlated, r(1825) ⫽.54, p⬍.001,
indicating that people who live in walkable neighborhoods do
indeed walk more. Because the size of the correlation is far
from perfect, we conducted the same set of analyses for walk-
ability and frequency of walking separately.
Walkability, sense of belonging, and upward social
mobility. Unlike in Study 1, in this study there was no
direct relationship between walkability and upward mobil-
ity, r(1811) ⫽⫺.013, p⫽.581. Despite the lack of a direct
association, we went on to explore the mediation analyses
presented later because some prominent methodologists and
researchers have pointed out that a direct association is not
a necessary condition for mediation and indeed have rec-
ommended dispensing with the direct effect as a require-
ment for mediation (e.g., MacKinnon, 2008; Rucker,
Preacher, Tormala, & Petty, 2011; Shrout & Bolger, 2002).
Conceptually, the lack of a direct association between walk-
ability and upward social mobility at the level of individuals
could be driven by unmeasured suppressor variables. For
example, income inequality and air pollution are both likely
to be positively associated with walkability (Marshall,
Brauer, & Frank, 2009) and are also likely to be negatively
associated with upward social mobility (Chetty et al., 2014).
As predicted, the more walkable one’s residential area
was, the greater was the sense of belonging the person was
likely to experience, r(1819) ⫽.127, p⬍.001, and the
greater the sense of belonging one experienced, the more
likely the person was to achieve higher social status,
r(1817) ⫽.160, p⬍.001 (see Table S2 in the online
supplemental materials for correlations between key vari-
ables including control variables). To test for the indirect
effect, we used a bootstrapping procedure (5,000 iterations)
with 95% bias-corrected confidence estimates (PROCESS
Model 4; Hayes, 2013). This analysis suggested that even
3
We originally preregistered the use of an attentional filter, which 1,680
participants passed. However, because the results of our analyses do not
differ between a filtered and an unfiltered sample, we report here the more
complete data.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
7
WALKABILITY AND UPWARD MOBILITY
after controlling for age, gender (1 ⫽male, 0 ⫽female),
race (1 ⫽White, 0 ⫽others), education, employment
status, and income, participants who resided in a more
walkable place experienced a greater sense of belonging
(b⫽.1640, SE ⫽.0281), t(1713) ⫽5.8341, p⬍.001, R
2
⫽
.07; that the greater one’s sense of belonging, the more
likely a person achieved upward social mobility (b⫽.0857,
SE ⫽.0207), t(1712) ⫽4.1509, p⬍.001, R
2
⫽.16; and
that a sense of belonging mediated the association between
walkability and upward social mobility (indirect effect ⫽
.0141, 95% CI [.0064, .0237]; see Table 3).
We also tested an alternate hypothesis, that walkability,
above and beyond our control variables, would lead to
upward social mobility, which would then lead to a sense of
belongingness, but found no evidence for that pathway
(indirect effect ⫽⫺.0014, 95% CI [⫺.0071, .0042]).
Frequency of walking, sense of belonging, and upward
social mobility. We found similar results when we sub-
stituted frequency of walking for walkability. Like per-
ceived walkability, frequency of walking was not correlated
with upward social mobility, r(1815) ⫽.002, p⫽.947.
However, we found that the more frequently one tended to
walk, the greater the sense of belonging the person was
likely to experience, r(1824) ⫽.165, p⬍.001. We then
tested for mediation with possible confounding variables
using a bootstrapping procedure (5,000 iterations) with 95%
bias-corrected confidence estimates (PROCESS Model 4;
Hayes, 2013) and found evidence that frequency of walking
enhanced upward social mobility mediated through a sense
of belonging above and beyond other control variables
(indirect effect ⫽.0160, 95% CI [.0077, .0263]; see Table
4). That is, participants who tended to walk more (vs. less)
often to nearby places tended to experience a greater sense
of belonging (b⫽.1825, SE ⫽.0276), t(1718) ⫽6.6112,
p⬍.001, R
2
⫽.07. The greater the sense of belonging one
experienced, the more likely one was to achieve higher
social status than did one’s parents, controlling for the effect
of frequency of walking (b⫽.0879, SE ⫽.0207),
t(1717) ⫽4.2503, p⬍.001, R
2
⫽.16.
We also tested the alternate hypothesis that frequency of
walking enhances a sense of belonging mediated through
upward social mobility above and beyond the control vari-
ables but found no evidence for this pathway (indirect
effect ⫽⫺.0016, 95% CI [⫺.0077, .0039]).
In a preregistered, nationally representative sample of
Americans, we found that living in a more walkable neigh-
borhood was associated with stronger feelings of belonging
in that neighborhood and that those feelings themselves
were associated with upward social mobility. In short, Study
3 revealed one potential psychological mechanism underly-
ing the association between walkability and upward social
mobility.
Study 4
Cross-cultural research has shown that some of the find-
ings from North America and other Western, educated,
industrialized, rich, and democratic (WEIRD) societies are
not replicated in non-WEIRD samples (Henrich, Heine, &
Norenzayan, 2010). Thus, it is important to examine
whether the findings of Study 3 in the United States would
extend beyond the American context. We chose to test the
generalizability of our findings in Korea, because personal
achievement is tightly linked to success and power in ver-
tical individualistic cultures (such as the United States),
whereas they are less tightly linked in more vertical collec-
tivistic cultures (such as Korea; Torelli & Shavitt, 2010).
For instance, family members of a zaibatsu in Japan (e.g.,
Sumitomo, Mitsui) or a chaebol in Korea (e.g., Samsung)
could have power and success, even if their personal
achievement is limited. Therefore, it is possible that our key
finding that walkability is associated with upward social
mobility is applicable to only the United States or other
individualistic, meritocratic societies and not generalizable
to other collectivistic, nepotistic societies.
Method
Participants. Participants were members of Micromill
Embrain, a nationwide online research panel in South Ko-
Table 3
Mediation Analyses in Studies 3 and 4: Testing the Links
Between Walkability and Belonging and Between Belonging and
Upward Mobility
Path
Unstandardized path
coefficients (SE)p
Walkability
Walkability ¡belonging
Study 3 .1640 (.0281) ⬍.001
Study 4 .0527 (.0405) .1931
Belonging ¡upward mobility
Study 3 .0857 (.0207) ⬍.001
Study 4 .0804 (.0233) ⬍.001
Walking
Walking ¡belonging
Study 3 .1825 (.0276) ⬍.001
Study 4 .2022 (.0361) ⬍.001
Belonging ¡upward mobility
Study 3 .0879 (.0207) ⬍.001
Study 4 .0837 (.0235) ⬍.001
Table 4
Indirect Effect (95% Confidence Intervals) of Walkability and
Frequency of Walking on Upward Social Mobility in Studies 3
and 4
Variable
Study 3
(United States)
Study 4
(Korea)
Walkability [.0064, 0237] [⫺.0021, .0134]
Frequency of walking [.0077, .0263] [.0071, .0316]
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
8OISHI, KOO, AND BUTTRICK
rea. Micromill Embrain’s panel has more than one million
individuals, ages 18 and above, who voluntarily opt in to be
a panel member and receive monetary incentives in return
for the completion of surveys. Among the panel members,
participants in this study were invited to take part in the
survey online in 2016. Data for this study were obtained as
part of a larger longitudinal study on social judgment, and
this study was presented with other measures that are unre-
lated to this research. The study involved two waves of
online surveys among Korean adults ages 18 years and
older. The items related to social mobility were not
included in the first round of the survey and were added
in the second round. Therefore, we analyzed the data from
the second round of the survey. In the first round, among
4,350 participants who received the invitation, 1,880 par-
ticipated in the survey. Of those participants, 1,466 (49.2%
female; M
age
⫽41.09 years) completed the survey in the
second round (attrition rate 23%).
Materials and procedure. Participants’ perceived
walkability of their residential area and actual walking
activity was measured as in Study 3 using the land-use mix
diversity dimension of the NEWS⫺A (Cerin et al., 2006;
␣⫽.77; M⫽3.27, SD ⫽.69) and self-reported walking
(␣⫽.81; M⫽3.30, SD ⫽.76). Then participants com-
pleted a two-item scale adapted from past research to indi-
cate their sense of belonging to their neighborhood (i.e., “I
feel a strong sense of belonging to my neighborhood” and “I
feel a strong sense of belonging to the city/town I live in
now”; r⫽.607; Keyes, 1998) on a 6-point scale ranging
from 1 (strongly disagree)to6(strongly agree; M ⫽2.75,
SD ⫽.99). As in Study 3, we operationalized participants’
upward social mobility by subtracting their current social
class from their parents’ social class rating, with higher
scores indicating greater actual mobility (M⫽.06, SD ⫽
.95). Demographic questions including gender, age, educa-
tion, employment status, and income were asked at the end
of Study 4.
Results and Discussion
As in the United States, the perceived walkability of the
neighborhood was correlated with actual frequency of walk-
ing, albeit more weakly,
4
r(1464) ⫽.222, p⬍.001. Thus,
we conducted the same set of analyses for walkability and
frequency of walking separately.
Walkability, sense of belonging, and upward social
mobility. As in Study 3, there was no direct relationship
between walkability and upward social mobility, r(1464) ⫽
⫺.004, p⫽.883. Also, there was no correlation between
walkability and a sense of belonging, r(1464) ⫽.016, p⫽
.540. However, the greater the sense of belonging one
experienced, the more likely the person was to achieve
higher social status, r(1464) ⫽.113, p⬍.001 (see Table S2
in the online supplemental materials for correlations be-
tween key variables including control variables). A boot-
strapping procedure (5,000 iterations) with 95% bias-
corrected confidence estimates (PROCESS Model 4; Hayes,
2013) suggested that controlling for age, gender (1 ⫽male,
0⫽female), education, employment status, and income,
participants who resided in a more walkable place did not
hold a greater sense of belonging (b⫽.0527, SE ⫽.0405),
t(1459) ⫽1.3022, p⫽.1931, R
2
⫽.05. The greater was
one’s sense of belonging, however, the more likely a person
achieved upward social mobility (b⫽.0804, SE ⫽.0233),
t(1458) ⫽3.4531, p⬍.001, R
2
⫽.05. In addition, a sense
of belonging did not mediate the association between walk-
ability and upward social mobility (indirect effect ⫽.0042,
95% CI [⫺.0021, .0134]).
Frequency of walking, sense of belonging, and upward
social mobility. There was no direct relationship between
frequency of walking and upward social mobility,
r(1464) ⫽⫺.011, p⫽.669. However, as in Study 3, the
more frequently one walked, the greater the sense of be-
longing the person was likely to experience, r(1464) ⫽.124,
p⬍.001, and the greater the sense of belonging one
experienced, the more likely the person was to achieve
higher social status, r(1464) ⫽.113, p⬍.001. A bootstrap-
ping procedure (5,000 iterations) with 95% bias-corrected
confidence estimates (PROCESS Model 4; Hayes, 2013)
suggested that even after controlling for age, gender, edu-
cation, employment status, and income, participants who
frequently walked were more likely to feel a greater sense of
belonging in their neighborhoods (b⫽.2022, SE ⫽.0361),
t(1459) ⫽5.5970, p⬍.001, R
2
⫽.06. Also, the greater
one’s sense of belonging, the more likely that one had
actually been economically mobile, achieving higher social
status than the status their parents had while they them-
selves were growing up, controlling for the effect of the
aforementioned demographics and the frequency of walking
(b⫽.0837, SE ⫽.0235), t(1458) ⫽3.5603, p⬍.001, R
2
⫽
.05. We then found that a sense of belonging significantly
mediated the relationship between frequency of walking and
achieving upward social mobility (indirect effect ⫽.0169,
95% CI [0071, .0316]).
We also tested an alternate hypothesis and found no
evidence that frequency of walking enhances a sense of
belonging mediated through upward social mobility above
and beyond the control variables (indirect effect ⫽⫺.0019,
95% CI [⫺.0106, .0046]).
In Study 4, we extend the findings of Study 3 to a new
country (Korea). Study 4 shows, in a nationwide sample,
that people who frequently walked in neighborhoods felt a
greater sense of belonging in those neighborhoods, which is
4
The weaker correlation may be due to restricted range, because Korea’s
population is highly concentrated in Seoul and the surrounding areas,
which are generally walkable, with an extensive public transportation
system.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
9
WALKABILITY AND UPWARD MOBILITY
associated with upward socioeconomic mobility. However,
unlike in Study 3, perceived walkability was not associated
with a sense of belonging. The null findings for perceived
walkability may be explained by the fact that unlike the
United States, Korea is a small and densely populated
country, significantly more walkable than is the United
States (M
Korea
⫽3.24, SD ⫽.68; M
United States
⫽2.33, SD ⫽
.99), t(3225.388) ⫽⫺30.873, p⬍.001, d⫽1.060. Most
relevant to our null findings, variance in walkability scores
was substantially smaller in Korea than in the United States,
Levene’s test F(1, 3298) ⫽288.352, p⬍.001, d⫽.59.
General Discussion
From a socioecological perspective (Oishi, 2014; Stokols,
1992; Yamagishi, 2011), the current research explored the
link between walkability and upward social mobility and
tested whether walkability is associated with a greater sense
of belonging, which in turn would be associated with up-
ward social mobility. Whereas much of psychological re-
search on upward social mobility has focused on internal
factors such as intelligence and motivation (Deary et al.,
2005; Snarey & Vaillant, 1985), we have instead examined
the impact of a concrete built environmental factor: walk-
ability.
Using tax records from approximately nine million Amer-
icans, in Study 1 we first established that commuting zones
with higher walkability also have higher intergenerational
upward mobility and that this relationship is robust to con-
trol variables such as political climate and physical health.
In Study 2, using data from over 3.66 million Americans,
we explored one possible mechanism, finding that employ-
ment and wages in walkable cities are less dependent on car
ownership. That is, economic success had less to do with car
ownership in walkable cities than in less walkable cities. In
Studies 3 and 4, we tested a psychological mechanism,
whether living in a walkable neighborhood is associated
with a greater sense of belonging, which in turn would be
associated with upward social mobility. Study 3 found that
although the direct association between walkability and
upward social mobility was not significant, those living in a
walkable neighborhood and those who walked more in their
everyday lives felt a greater sense of belonging, which was
in turn associated with upward social mobility. Study 4 was
a direct replication of Study 3 in South Korea. We did not
find the predicted indirect effect from walkability to sense
of belonging to upward social mobility, perhaps due to a
smaller variance in walkability. It should be noted that
frequency of walking was indeed associated with a greater
sense of belonging, which was in turn associated with
upward social mobility. In both Studies 3 and 4, the fre-
quency of walking was associated with a greater sense of
belonging, which in turn was related to more upward social
mobility. Thus, it appears that walking was a more proximal
predictor of upward social mobility than was walkability of
the neighborhood per se. In other words, walkability matters
(at least in the United States) to the extent that it encourages
residents to walk more. The more an individual walks, the
greater the sense of belonging one feels toward the city.
Additional Mechanisms
In the current work, we focused on one intrapsychic
mechanism: one’s sense of belonging. However, walkability
and walking may also impact upward mobility through
other, intrapsychic as well as interpersonal means. Imagine
people living in a walkable city. In their daily commute, on
foot or via public transportation, they will, almost by ne-
cessity, run into lots of other people directly engaged in the
same commute. By contrast, people in a more unwalkable
city, commuting by car, are practically hermetically sealed
off from social interaction for the duration of their com-
mute. The social intermingling of a more walkable city
brings people together from all socioeconomic strata; the
lack of social contact in a more unwalkable city precludes
that possibility. In a walkable city, people from lower so-
cioeconomic strata are more likely to see “successful” peo-
ple in a daily basis, and these repeated interactions may
make success feel more attainable, because it is something
that they see every day. It is well recognized that having
positive role models is an important aspect of future suc-
cess; for instance, when exposed to a successful role model,
students estimated that future career goals are more attain-
able (Lockwood & Kunda, 1997; Lockwood, Shaughnessy,
Fortune, & Tong, 2012). In a walkable city, success may
seem more attainable and there are many more accessible
role models whom struggling people can look up to and
emulate. The firsthand contact with successful people might
inspire people with limited means to work harder and be
successful.
Similarly, recent work has suggested that walking encour-
ages locomotive motivation, creativity, and forward prog-
ress (Webb, Rossignac-Milon, & Higgins, 2017). It might
be that living in a walkable neighborhood encourages more
walking, which in turn increases locomotion motivation
(action orientation) and creative problem solving. Future
work exploring these interpersonal, cognitive, and motiva-
tional aspects of highly walkable cities will enhance the
understanding of the relation between walkable environ-
ments and upward social mobility.
Limitations and Future Directions
We would like to point out several limitations of the
current research. The primary limitation is a function of our
analytical approach: Because the current analyses are based
on correlational data, there is the possibility that unmea-
sured third variables account for the link between walkabil-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
10 OISHI, KOO, AND BUTTRICK
ity and upward mobility, and thus we cannot make causal
claims. In a similar vein, we cannot conclusively demon-
strate chains of causality. In Study 3, for example, though
we did not find any evidence for the mediational role of
upward social mobility in explaining links between walk-
ability and a sense of belonging, it is possible that both
walkability and upward social mobility induce a sense of
belonging, and our findings could be driven by this alter-
native specification. Because we could not manipulate ei-
ther walkability or a sense of belonging, we could not fully
disambiguate these two accounts. In Studies 2⫺4, there is a
possibility of selection bias in that people who like to walk
chose to live in walkable cities or neighborhoods and vice
versa. Study 1 does not have this problem, because children
are unlikely to be able to choose where to live during their
childhood. In Study 1, the city in which the participants
grew up, although not randomly assigned, is not an endog-
enous variable, and the selection bias in Study 1 is not a
major concern. Nevertheless, it is important to explore
whether there is a causal effect of living in a walkable city
in the future.
Second, our analyses treat walkability and access to pub-
lic transportation as interchangeable. Although they are
strongly correlated, it is likely that each has its own separate
impact on upward mobility. For instance, Lachapelle,
Frank, Saelens, Sallis, and Conway (2011) found that even
within equally walkable neighborhoods, individuals who
used public transportation to commute had more moderate-
intensity physical activity than did those who drove (see
also Saelens, Vernez Moudon, Kang, Hurvitz, & Zhou,
2014). The availability and use of public transportation thus
could have an independent effect on an individual’s upward
social mobility. As measures of public transit availability
become more comprehensive (as of now, there are far more
cities with Walk Scores than Public Transit scores available
at www.walkscore.com), future work disentangling the two
factors will be important for guiding policy recommenda-
tions.
Third, Studies 3 and 4 relied on self-reported walkability,
whereas other researchers interested in walkability have
used more objective measures based on geographic infor-
mation systems (Todd et al., 2016). As Studies 3 and 4
found, it is related to self-reported frequency of walking.
However, because walkability is a multidimensional con-
struct—ranging from purely physical aspects such as inter-
section density and connectivity to the presence of side-
walks; to proximity to restaurants, banks, and other
amenities; to recreational opportunities (Forsyth, 2015)—
different aspects of walkability may affect different paths to
upward social mobility. In addition to the hardscape (built
environment) and accessibility of amenities that is captured
in the Walk Score measure used in Studies 1 and 2 and that
may have a more direct effect on the link between car
ownership and employment, the softscape of a neighbor-
hood, such as its green spaces and lighting, also affect
perceptions of an area’s walkability (Hajna, Dasgupta, Hal-
parin, & Ross, 2013) and may have a more direct effect on
a sense of belonging. Our current analyses, based as they are
on either the hardscape-limited Walk Score or global indi-
vidual perceptions of walkability, cannot disentangle all
these distinct aspects of walkability, and future studies with
more focused definitions of walkability, or which manipu-
late perceptions of walkability even without changing the
hardscape, will be useful, especially when it comes to policy
recommendations.
Fourth, whereas we found a robust association between
walkability and upward social mobility in the city-level
analyses of Study 1, we did not observe the direct associa-
tion in the individual-level analyses of Studies 3 and 4. It
may be that walkability is an emergent property most
clearly visible at the level of aggregate, not at the level of
each individual. This seeming paradox can be illustrated
with a reference to epidemiology. Studies clearly indicate,
for example, a strong association between air quality and
prevalence of lung cancer when examined at the level of a
city or county (e.g., Hemminki & Pershagen, 1994), yet
when examined at the level of individuals, the association is
null or nonsignificant (e.g., Beelen et al., 2008). This is in
part because lung cancer is rare. When examined at the level
of a city, cancer prevalence could range from 0% to even
more than 10%, with gradation in each level. When ob-
served at the level of individuals, however, most of them do
not have cancer, and therefore the effect of air quality is
hard to discern. Similarly, the rate of lower SES children’s
moving up the economic ladder ranges from ⬃5%–13%
between cities, and so when examined at the level of indi-
viduals, those who moved up the ladder in adulthood are
small in number.
Finally, in an effort to test generalizability, we conducted
Study 4 in South Korea. Although we assumed that a sense
of belonging would be measured more or less equivalently
between the United States and Korea, this assumption must
be tested rigorously in the future using sophisticated tech-
niques such as item response theory (e.g., Reise, Widaman,
& Pugh, 1993).
Societal progress is often measured by whether the life of
the current generation has gotten better than that of the
previous generations, with intergenerational upward eco-
nomic mobility as a critical indicator of the fairness of a
society. We found that the walkability of a city is an
important predictor of upward social mobility and that this
might be due in part to the fact that in walkable cities
residents can get access to employment without owning a
car and due in part to more walking and a greater sense of
belonging, which we showed have real-world relationships
with individual upward mobility. We found walking effects
(but not perceived walkability effects) cross-nationally and
cross-culturally, both in the individualistic United States
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
11
WALKABILITY AND UPWARD MOBILITY
and in the more collectivist Korea, implying that the link
between walking, sense of belonging, and upward social
mobility may be widespread and robust. It is not easy to add
sidewalks or make public roads more walkable by adding
more intersections and crossings. Adding additional bus
lines or putting in new train lines is also not cheap (Duany,
Plater-Zyberk, & Speck, 2000; Speck, 2012). However,
these might be wise societal investments if, as our results
suggest, they may help rebuild the fading American dream.
References
Aaronson, D., & Mazumder, B. (2008). Intergenerational economic mo-
bility in the United States, 1940 to 2000. The Journal of Human
Resources, 43, 139 –172.
Allen, K., Kern, M. L., Vella-Brodrick, D., Hattie, J., & Waters, L. (2018).
\What schools need to know about fostering school belonging: A meta-
analysis. Educational Psychology Review, 30, 1–34. http://dx.doi.org/10
.1007/s10648-016-9389-8
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for
interpersonal attachments as a fundamental human motivation. Psycho-
logical Bulletin, 117, 497–529. http://dx.doi.org/10.1037/0033-2909.117
.3.497
Beelen, R., Hoek, G., van den Brandt, P. A., Goldbohm, R. A., Fischer, P.,
Schouten, L. J.,...Brunekreef, B. (2008). Long-term exposure to
traffic-related air pollution and lung cancer risk. Epidemiology, 19,
702–710. http://dx.doi.org/10.1097/EDE.0b013e318181b3ca
Carr, L. J., Dunsiger, S. I., & Marcus, B. H. (2011). Validation of Walk
Score for estimating access to walkable amenities. British Journal of
Sports Medicine, 45, 1144 –1148. http://dx.doi.org/10.1136/bjsm.2009
.069609
Cerin, E., Saelens, B. E., Sallis, J. F., & Frank, L. D. (2006). Neighborhood
Environment Walkability Scale: Validity and development of a short
form. Medicine & Science in Sports and Exercise, 38, 1682–1691.
http://dx.doi.org/10.1249/01.mss.0000227639.83607.4d
Chetty, R., Grusky, D., Hell, M., Hendren, N., Manduca, R., & Narang, J.
(2017). The fading American dream: Trends in absolute income mobility
since 1940 (NBER Working Paper No. 22910). Cambridge, MA: Na-
tional Bureau of Economic Research.
ⴱ
Chetty, R., Hendren, N., Kline, P., & Saez, E. (2014). Where is the land
of opportunity? The geography of intergenerational mobility in the
United States. Quarterly Journal of Economics, 129, 1553–1623. http://
dx.doi.org/10.1093/qje/qju022
Davidai, S., & Gilovich, T. (2015). Building a more mobile America—One
income quintile at a time. Perspectives on Psychological Science, 10,
60 –71. http://dx.doi.org/10.1177/1745691614562005
Deary, I. J., Taylor, M. D., Hart, C. L., Wilson, V., Smith, G. D., Blane, D.,
& Starr, J. M. (2005). Intergenerational social mobility and mid-life
status attainment: Influences of childhood intelligence, childhood social
factors and education. Intelligence, 33, 455– 472. http://dx.doi.org/10
.1016/j.intell.2005.06.003
Duany, A., Plater-Zyberk, E., & Speck, J. (2000). Suburban nation: The
rise of sprawl and the decline of the American dream. New York, NY:
North Point Press.
Duncan, D. T., Aldstadt, J., Whalen, J., Melly, S. J., & Gortmaker, S. L.
(2011). Validation of Walk Score for estimating neighborhood walk-
ability: An analysis of four U.S. metropolitan areas. International Jour-
nal of Environmental Research and Public Health, 8, 4160 – 4179.
http://dx.doi.org/10.3390/ijerph8114160
Ewing, R., Hamidi, S., Grace, J. B., & Wei, Y. D. (2016). Does urban
sprawl hold down upward mobility? Landscape and Urban Planning,
148, 80 – 88. http://dx.doi.org/10.1016/j.landurbplan.2015.11.012
Forsyth, A. (2015). What is a walkable place? The walkability debate in
urban design. Urban Design International, 20, 274 –292. http://dx.doi
.org/10.1057/udi.2015.22
Frank, L. D., Sallis, J. F., Conway, T. L., Chapman, J. E., Saelens, B. E.,
& Bachman, W. (2006). Many pathways from land use to health:
Associations between neighborhood walkability and active transporta-
tion, body mass index, and air quality. Journal of the American Planning
Association, 72, 75– 87. http://dx.doi.org/10.1080/01944360608976725
Hajna, S., Dasgupta, K., Halparin, M., & Ross, N. A. (2013). Neighbor-
hood walkability: Field validation of geographic information system
measures. American Journal of Preventive Medicine, 44(6), e55– e59.
http://dx.doi.org/10.1016/j.amepre.2013.01.033
Hayes, A. F. (2013). Introduction to mediation, moderation, and condi-
tional process analysis: A regression-based approach. New York, NY:
Guilford Press.
Hemminki, K., & Pershagen, G. (1994). Cancer risk of air pollution:
Epidemiological evidence. Environmental Health Perspectives,
102(Suppl. 4), 187–192.
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in
the world? Behavioral and Brain Sciences, 33, 61– 83. http://dx.doi.org/
10.1017/S0140525X0999152X
Hillman, C. H., Pontifex, M. B., Raine, L. B., Castelli, D. M., Hall,
E. E., & Kramer, A. F. (2009). The effect of acute treadmill walking
on cognitive control and academic achievement in preadolescent
children. Neuroscience, 159, 1044 –1054. http://dx.doi.org/10.1016/j
.neuroscience.2009.01.057
Hochschild, J. L. (1996). Facing up to the American dream: Race, class
and the soul of the nation. New Jersey: Princeton University Press.
http://dx.doi.org/10.1515/9781400821730
Keyes, C. L. M. (1998). Social well-being. Social Psychology Quarterly,
61, 121–140. http://dx.doi.org/10.2307/2787065
Kraus, M. W., & Tan, J. J. X. (2015). Americans overestimate social class
mobility. Journal of Experimental Social Psychology, 58, 101–111.
http://dx.doi.org/10.1016/j.jesp.2015.01.005
Lachapelle, U., Frank, L., Saelens, B. E., Sallis, J. F., & Conway, T. L.
(2011). Commuting by public transit and physical activity: Where you
live, where you work, and how you get there. Journal of Physical
Activity & Health, 8(Suppl. 1), S72–S82. http://dx.doi.org/10.1123/jpah
.8.s1.s72
Leip, D. (2012). Dave Leip’s atlas of U.S. presidential elections. Retrieved
from http://uselectionatlas.org
Lockwood, P., & Kunda, Z. (1997). Superstars and me: Predicting the
effect of role models on the self. Journal of Personality and Social
Psychology, 73, 91–103. http://dx.doi.org/10.1037/0022-3514.73.1.91
Lockwood, P., Shaughnessy, S. C., Fortune, J. L., & Tong, M. O. (2012).
Social comparisons in novel situations: Finding inspiration during life
transitions. Personality and Social Psychology Bulletin, 38, 985–996.
http://dx.doi.org/10.1177/0146167212447234
MacKinnon, D. P. (2008). Introduction to statistical mediation analysis.
Mahwah, NJ: Erlbaum.
Marshall, J. D., Brauer, M., & Frank, L. D. (2009). Healthy neighborhoods:
Walkability and air pollution. Environmental Health Perspectives, 117,
1752–1759. http://dx.doi.org/10.1289/ehp.0900595
National Park Services. (n.d.). National register database and research.
Retrieved from https://www.nps.gov/subjects/nationalregister/database-
research.htm
Oishi, S. (2014). Socioecological psychology. Annual Review of Psychol-
ogy, 65, 581– 609. http://dx.doi.org/10.1146/annurev-psych-030413-
152156
Ong, P., & Blumenberg, E. (1998). Job access, commute and travel burden
among welfare recipients. Urban Studies, 35, 77–93. http://dx.doi.org/
10.1080/0042098985087
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
12 OISHI, KOO, AND BUTTRICK
Power, C., Matthews, S., & Manor, O. (1996). Inequalities in self rated
health in the 1958 birth cohort: Lifetime social circumstances or social
mobility. British Medical Journal, 313, 449 – 453.
Raphael, S., & Rice, L. (2002). Car ownership, employment, and earnings.
Journal of Urban Economics, 52, 109 –130.
Raudenbush, S. W., Bryk, A. S., & Congdon, R. (2004). HLM 6 for
Windows [Computer software]. Lincolnwood, IL: Scientific Software
International, Inc.
Reise, S. P., Widaman, K. F., & Pugh, R. H. (1993). Confirmatory factor
analysis and item response theory: Two approaches for exploring mea-
surement invariance. Psychological Bulletin, 114, 552–566. http://dx.doi
.org/10.1037/0033-2909.114.3.552
Rubin, D. B., & Thomas, N. (1996). Matching using estimated propensity
scores: Relating theory to practice. Biometrics, 52, 249 –264. http://dx
.doi.org/10.2307/2533160
Rucker, D. D., Preacher, K. J., Tormala, Z. L., & Petty, R. E. (2011).
Mediation analysis in social psychology: Current practices and new
recommendations. Social and Personality Psychology Compass, 5, 359 –
371. http://dx.doi.org/10.1111/j.1751-9004.2011.00355.x
Saelens, B. E., Vernez Moudon, A., Kang, B., Hurvitz, P. M., & Zhou, C.
(2014). Relation between higher physical activity and public transit use.
American Journal of Public Health, 104, 854 – 859. http://dx.doi.org/10
.2105/AJPH.2013.301696
Sallis, J. F., Saelens, B. E., Frank, L. D., Conway, T. L., Slymen, D. J.,
Cain, K. L.,...Kerr, J. (2009). Neighborhood built environment and
income: Examining multiple health outcomes. Social Science & Medi-
cine, 68, 1285–1293. http://dx.doi.org/10.1016/j.socscimed.2009.01.017
Sargan, J. (1958). The estimation of economic relationships using instru-
mental variables. Econometrica, 26, 393– 415.
Shnabel, N., Purdie-Vaughns, V., Cook, J. E., Garcia, J., & Cohen, G. L.
(2013). Demystifying values-affirmation interventions: Writing about
social belonging is a key to buffering against identity threat. Personality
and Social Psychology Bulletin, 39, 663– 676. http://dx.doi.org/10.1177/
0146167213480816
Shrout, P. E., & Bolger, N. (2002). Mediation in experimental and non-
experimental studies: New procedures and recommendations. Psycho-
logical Methods, 7, 422– 445.
Solon, G., Haider, S. J., & Wooldridge, J. M. (2015). What are e weighting
for? Journal of Human Resources, 50, 301–316.
Sorkin, A. R., & Thee-Brenan, M. (2014, December 10). Many feel the
American Dream is out of research, Poll shows. The New York Times.
Retrieved from http://nyti.ms/1zxDn8i
Snarey, J. R., & Vaillant, G. E. (1985). How lower- and working-class
youth become middle-class adults: The association between ego defense
mechanisms and upward social mobility. Child Development, 56, 899 –
910. http://dx.doi.org/10.2307/1130102
Speck, J. (2012). Walkable city: How downtown can save America one step
at a time. New York, NY: North Point Press.
Stephens, N. M., Markus, H. R., & Phillips, L. T. (2014). Social class
culture cycles: How three gateway contexts shape selves and fuel in-
equality. Annual Review of Psychology, 65, 611– 634. http://dx.doi.org/
10.1146/annurev-psych-010213-115143
Stokols, D. (1992). Establishing and maintaining healthy environments:
Toward a social ecology of health promotion. American Psychologist,
47, 6 –22. http://dx.doi.org/10.1037/0003-066X.47.1.6
Su, R., Tay, L., & Diener, E. (2014). The development and validation of the
Comprehensive Inventory of Thriving (CIT) and the Brief Inventory of
Thriving (BIT). Applied Psychology: Health and Well-Being, 6, 251–
279. http://dx.doi.org/10.1111/aphw.12027
Todd, M., Adams, M. A., Kurka, J., Conway, T. L., Cain, K. L., Buman,
M.P.,...King, A. C. (2016). GIS-measured walkability, transit, and
recreation environments in relation to older adults’ physical activity: A
latent profile analysis. Preventive Medicine, 93, 57– 63. http://dx.doi
.org/10.1016/j.ypmed.2016.09.019
Torelli, C. J., & Shavitt, S. (2010). Culture and concepts of power. Journal
of Personality and Social Psychology, 99, 703–723. http://dx.doi.org/10
.1037/a0019973
U.S. Department of Labor, Bureau of Labor Statistics. (2009-2013). Oc-
cupational employment statistics. Retrieved from https://www.bls.gov/
oes/tables.htm
Van Cauwenberg, J., Van Holle, V., De Bourdeaudhuij, I., Van Dyck, D.,
& Deforche, B. (2016). Neighborhood walkability and health outcomes
among older adults: The mediating role of physical activity. Health &
Place, 37, 16 –25. http://dx.doi.org/10.1016/j.healthplace.2015.11.003
Walton, G. M., & Cohen, G. L. (2011, March 18). A brief social-belonging
intervention improves academic and health outcomes of minority stu-
dents. Science, 331, 1447–1451. http://dx.doi.org/10.1126/science
.1198364
Webb, C. E., Rossignac-Milon, M., & Higgins, E. T. (2017). Stepping
forward together: Could walking facilitate interpersonal conflict resolu-
tion? American Psychologist, 72, 374 –385. http://dx.doi.org/10.1037/
a0040431
Yamagishi, T. (2011). Micro-macro dynamics of the cultural construction
of reality. Advances in Culture and Psychology, 1, 251–308.
Yeager, D. S., Walton, G. M., Brady, S. T., Akcinar, E. N., Paunesku, D.,
Keane, L.,...Dweck, C. S. (2016). Teaching a lay theory before college
narrows achievement gaps at scale. Proceedings of the National Acad-
emy of Sciences of the United States of America, 113, E3341–E3348.
http://dx.doi.org/10.1073/pnas.1524360113
Received March 16, 2018
Revision received September 27, 2018
Accepted October 7, 2018 䡲
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
13
WALKABILITY AND UPWARD MOBILITY
A preview of this full-text is provided by American Psychological Association.
Content available from American Psychologist
This content is subject to copyright. Terms and conditions apply.