Walk and Be Moved: How Walking Builds Social Movements

Brian B. KnudsenUrban Innovation Analysis(Corresponding author)

1920 S Street NW, Apt. 105Washington, DC 20009Phone: 202-641-8151

Email: knudsenbb@gmail.com

Terry N. ClarkUniversity of Chicago

Email: tnclark@uchicago.edu

1

Biographical information: Brian Knudsen is Research Associate at Urban

Innovation Analysis. He received his PhD from the Heinz College at Carnegie

Mellon University. His research focuses on walking, experience of cities, and

new social movements. Terry Nichols Clark is Professor of Sociology at the

University of Chicago. He holds MA and PhD degrees from Columbia University.

He has worked on how cities use culture to transform themselves, esp. in books

on The City as an Entertainment Machine, and Building Post-Industrial Chicago. His most recent

work focuses on urban scenes.

Abstract: Recent scholarship recognizes the city’s role as “civitas” – a

“space of active democratic citizenship” and “full human realization” based on

open and free encounter and exchange with difference. The current research

emerges from and fills a need within this perspective by examining how local

urban contexts undergird and bolster social movement organizations (SMOs).

Our theory elaborates and linear regressions assess the relationships between

four urban form variables and SMOs. In addition, our theory also examines how

urban walking mediates the relationships between these local contextual traits

and SMOs. Drawing primarily from the ZIP Code Business Patterns and U.S.

Census, we generate a dataset of approximately 30,000 cases, permitting

regression analyses that distinguish strong direct effects of density,

2

connectivity, housing age diversity and walking on the incidence of SMOs.

Sobel tests indicate that for both density and connectivity, walking mediates

the relationships with SMOs in a way consistent with the mechanisms of the

hypotheses.

Keywords: Walking, Social Movements, Organizations, Cities

1. Introduction

This paper investigates whether city contexts activate and undergird current-

day social movement organizations (SMOs).1 We elucidate how one’s unique

experiences in cities – the ways in which an individual interacts with and

makes use of urban environments, neighborhoods and spaces – relates to this

particular type of political phenomena. Moreover, we identify the precise

urban contextual traits that pertain to these effects, postulate how these

traits influence the ways in which individuals experience cities, and show the

subsequent implications for political outcomes. In short, this research

explores the intricate inter-relations between the distinct traits and

qualities of cities, our individual experiences of them, and SMOs.

We contend that the physical accessibility characteristic of dense, diverse,

walkable cities enables a social accessibility to a variety of ideas, actions,

3

and happenings. Encounter with difference is fundamental to the orientation

of many current-day SMOs. As such, through their capacity to generate and

facilitate encounter with difference, cities serve as a primary social setting

for these organizations. Furthermore, we propose that pedestrian activity –

i.e. walking – bridges between the physical accessibility of urban context and

the social accessibility central to SMOs. Through walking, people experience

both the physical and social diversity of their city in a direct and engaged

manner. Therefore, we contend that walking mediates the relationships between

urban context and SMOs. This paper develops these ideas, formulates them into

testable hypotheses, and submits them to empirical test.

There are several primary motivations for this research. First, examining the

links between urban context, walkability, and contemporary SMOs contributes to

and extends the prominent work on Political Opportunity Structures (POS).

Describing the POS as the “structure and dynamics of the political

environment”, McCright and Clark (2006) note that the POS conceptualization

helps to probe the factors that cause “variation in the mobilization of social

movements across U.S. communities.” They contend that existing accounts of

political opportunity structures too rigidly define the concept as equivalent

to institutional and state-based factors. By looking at the relationships

between urban contexts, walkability, and SMOs, this work extends the concept

of POS by examining a number of other contextual variables that help explain

the emergence and mobilization of movement organizations.

4

Second, by postulating the city as a context for current-day SMOs, we add to

the growing scholarship revealing the ways in which modern cities enable

important social processes. Jennings (2001) suggests that in the Western

tradition, cities have been conceptualized as venues for two different forms

of social activity, “urbs” and “civitas”. The “urbs” perspective portrays

cities primarily as economic markets, and recent social scientific research

(e.g., Acs and Armington 2004; Florida 2002; Glaeser 1994) explores the city

as the setting for economic accumulation, market exchange, and the force

behind growth. This study is instead situated within a different tradition,

described by Jennings as “the city as ‘civitas’”: the city (88) as a “space of

active democratic citizenship, equality under the law, and civic virtue”, as

well as “full human self-realization”. Instead of a city based upon (91)

individualistic “instrumental relationships of economic transaction”, the city

as civitas is organized around (89) the “active pursuit of shared purpose”,

and (91) the “sharing of a common moral space, common commitment to each

other, and a common political identity”.

By hypothesizing cities as spaces for contemporary SMOs, this paper presents a

specific characterization of the city as civitas. There are a number of

reasons for focusing on this particular account. One primary reason is that

although current research has looked at other important overlaps between

politics and the city, there is a distinct lack of scholarship concerning this

intersection of contemporary movement organizations and cities. For instance,

a number of studies (e.g., Gainsborough 2001; Oliver 2001; Walks 2004) have

5

pointed to differences in voting behavior, party preferences, and political

attitudes between people living in and out of cities and urban contexts.

Other important work (e.g., Ferman 1996; Logan and Molotch 1987; Peterson

1981; Stone 1993) has focused on urban governance and the characteristics of

the coalitions that influence municipal government and policy formation.

Castells’ (1983) study on historical urban movements remains an important

contribution. By contrast, little analysis probes the ways in which modern-

day movements relate to cities. For instance, Nicholls (2008, 841) points out

that “…few have actually opened up the urban ‘black box’ to identify the

processes and mechanisms that allow cities to play specific roles in broad

social movements.” We contend that the “processes and mechanisms” linking

present-day SMOs and cities stem from the centrality of diversity to these

SMOs and the role that cities play as spaces of difference. Therefore, this

paper makes an important contribution by hypothesizing and testing specific

mechanisms and processes by which cities enable these organizations.

2. Concepts

2.1 Contemporary Social Movement Organizations

This paper seeks to explain the incidence of present-day SMOs. Movement

organizations have diversified since the 1970s, and grown more complex to

capture conceptually. Noting such diversity, social movement theorists tried

to explain variations by introducing concepts like political opportunity

structure, framing (adapted from Goffman and Snow) and the internal

organizational structure (Della Porta, Kriesi, and Rucht l999; McAdam,

6

McCarthy, and Zald 1996). Some groups have adopted feminist issues; others

have embraced government and foundation policies; still others have linked

with political parties—all of which can be cited as evidence of the success of

movement agendas. From the 1970s to early 1990s many of the New Social

Movements (Offe 1985) became “institutionalized” and “professionalized” (Clark

and Kallman 2011) as they mimicked political parties and retained many

hierarchical elements, official membership, voting for officers, and

bureaucratic domination. Some were even dominated by unions or political

parties or other actors, such as the Tuscany environmental movements that were

supported by mayors and their parties (Della Porta and Andretta 2002). By

contrast, some later groups seek to recover the non-institutional politics

foregone by many New Social Movements in recent decades by deemphasizing

hierarchical structure, by more explicitly emphasizing organizational

horizontality, spontaneity, and leaderless decentralization, and by allowing

individuals to act autonomously (Graeber 2003; Clark and Kallman 2011;

McDonald 2006). Examples of this later movement activity include alter-

globalization protests at IMF and World Bank meetings in the late 1990s as

well as the Occupy concerns of 2011. We have termed them New New Social

Movements, but there has been minimal reconceptualization. This may be due in

part to the joint participation of these multiple types of organizations at a

single protest site, the refusal of overall coordination among activists, and

mutual ignorance of many. Deeper ethnography is one approach to seek to

disentangle such subtleties.

7

Our approach recognizes such distinctions, but moves in a broader comparative

direction. We think we are the first to discover and analyze electronic

unpublished data for environmental, human rights, and other types of social

movement organizations, from the US Census of Business. These data are

advantageous in that they provide SMO totals for each of the approximately

32,000 Zip Code Tabulation Areas, the Census version of US Postal Service Zip

Codes. This broad coverage permits analysis of how the density of such

organizations compares across the entire U.S., as well as investigations into

why it varies. We have only the self-reported categories of organizations in

terms of North American Industrial Classification System (NAICS) codes, and

thus cannot assess how much each may be strongly egalitarian, idealistic, or

narrowly NIMBY in style. Clearly there is variation among “environmental

groups,” but this is equally true of many social movement studies, and our

Business Census items can also be compared to “religious groups,” “business

groups” and other categories. The variations are so large that it is

appealing to investigate their sources. We thus refer to SMOs in a broad

manner, operationalized with these Census categories of social, environmental,

and human rights groups (detail below).

2.2 Cities and social movement organizations

The idea that democracy depends on diverse interactions and should thrive in

cites has a longstanding conceptual basis. For instance, Sennett (1999, 276)

writes that a democracy “supposes that people can consider views other than

their own. This was Aristotle’s notion in the Politics. He thought that the

8

awareness of difference occurs only in cities because every city is formed by

synoikismos, a drawing together of different families and tribes, of competing

economic interests, of natives with foreigners.” More specifically, Nicholls

(2008, 841) writes that cities contribute to the formation of movement groups

by acting as “relational incubators” that generate diverse social forms

possessing high-grade specialized resources available for use by

organizations.

Substantial earlier literature from sociology, economics, urban planning, and

political theory depicts cities as environments that generate diversity and

that enhance accessibility to it (e.g. Park 1915; Wirth 1938; Jacobs 1961;

Fischer 1975; Jacobs and Appleyard 1987; Glaeser 2000; Knudsen et al 2008).

This work suggests (Talen 2006, 234) a “complex encounter between the physical

world and the social world”, between urban context and enhanced access to and

interaction with diverse social influences, insights, and activities. These

relationships between urban context and enhanced accessibility to social

diversity have clear implications for current-day SMOs.2 Specifically, we

suggest that contexts with greater density, mixed urban uses, connectivity,

and walkability generate and offer the possibility of interaction with a

diversity of physical destinations, and in so doing permit and encourage

encounters with a wide variety of social influences, ideas, and people. These

encounters undergird the formation of SMOs.3 Below we develop a number of

more specific hypotheses relating these components of neighborhood context and

SMOs.

9

2.2.1 Density, mixed urban uses, connectivity, and social movement

organizations

Density – the concentration of people, dwelling units, and activity in a

location – both generates diversity and facilitates encounter with it.

Fischer (1975) notes that by concentrating many people together in close

range, dense cities construct the infrastructure and obtain thresholds needed

to support diverse social forms such as unique commercial enterprises,

minority ethnic communities, divergent cultural movements, and dissenting

ideas. By contrast, places with thinly spread populations are homogenizing

and can typically support more mainstream enterprises and views. Density also

facilitates encounter with diversity by creating low transport costs for both

people and ideas. Glaeser (2000, 484) writes that dense “cities excel in

permitting combinations [of diverse ideas] because of the absence of physical

distance.” New ideas and enterprises emerge from the combination of diverse

perspectives, making dense cities conducive to innovativeness. We assert that

dense urban contexts generate and facilitate interaction with diverse social

forms, and therefore support and undergird the formation of SMOs. We offer

the following hypothesis: H1: In locales with higher density, there is a higher incidence4 of

SMOs.

Mixed urban use refers to (Saelens, Sallis, and Frank 2003, 81) the “level of

integration within a given area of different types of uses for physical space,

including residential, office, retail/commercial, and public space.” Urban

10

contexts characterized by mixed urban use generate diversity and encounter

with it. As Jacobs (1961) influentially wrote, mixed use generates diversity

by drawing a variety of people to a district for different reasons at

different times, thus supporting diverse establishments, services, cultural

enterprises, and ideas. Intermixing of such uses also brings diverse

individuals to them who intermix as they pass between the various

destinations. Thus, mixed use urbanism contributes to diverse interactions

encouraging SMOs. H2: In locales with more mixed land uses, there is a higher incidence of SMOs.

Mixed building age is a variant on mixed urban uses that also (Jacobs 1961, 212)

“has a direct, explicit connection with diversity of population, diversity of

enterprises, and diversity of scenes.” The relationship between diversity and

mixed building age assumes an association between the age of a building and

the nature of the activities held in it. Well-established, standardized

establishments inhabit more new buildings, since they can support the higher

costs. By contrast, less established, experimental, and higher risk

enterprises can seek out older and less expensive buildings. Writing about

the transformation of a deserted factory site in the Canton neighborhood into

a retail shopping area, Merrifield (2002, 45) reports that “provision for

small-scale businesses incur high-risks: those catering exclusively to a

Canton catchment area would be unlikely to have extensive monetary turnover.

Here small businesses would be hard-pressed to pay any market rent, especially

one that would give the developer an adequate return for their initial

investment.” For these reasons mixed building ages contributes to a diverse

11

urban social setting which undergirds SMOs. H3: In locales with more mixed building

ages, there is a higher incidence of SMOs.

Connectivity refers to (Saelens, Sallis, and Frank 2003, 82) the “ease of moving

between origins (e.g., households) and destinations (e.g., stores and

employment) within the existing street and sidewalk-pathway structure”, and is

higher when streets are arranged in a grid pattern with short blocks. Urban

areas with high connectivity provide direct as well as alternative routes to a

destination. By contrast “low connectivity is found in the layout of modern

suburbs and is characterized by a low density of intersections (e.g., long

block sizes), barriers to direct travel (e.g. cul de sacs), and few route

choices.” Connectivity decreases travel costs, increases alternatives and

choices, and creates opportunities for interactions with diversity. Whereas

streets with long blocks are (Jacobs 1961, 179) “self-isolating”, short blocks

enlarge the situation of one’s everyday life by removing impediments to

movement and interaction. Districts with short blocks do not have (Jacobs

1961, 180) “mutual isolation of paths”, but instead streets that are “mixed

and mingled with each other.” The connectivity of short city blocks bolsters

the diversity behind SMOs. Therefore: H4: In locales with higher connectivity (in the form of

short city blocks), there is a higher incidence of SMOs.

2.2.2 Walking and social movement organizations

We can add further specificity to the links between urban context, encounter

with diversity, and contemporary SMOs by exploring how pedestrian activity –

12

walking – mediates between these components that undergird SMOs. Many

sociologists perceive a barrier between the physical and social; walking is a

specific mechanism, analogous to social interaction, which can transform its

participants. In their classic accounts, Balzac, Baudelaire, and Benjamin

depict flâneurs playfully traversing cities on foot to (Kramer and Short 2011,

323) absorb “[their] affective intensities for aesthetic translation”,

therefore linking the physical and social in cities. Walking is an engaged

way of experiencing and interacting with the physical and social forms of a

neighborhood, through which individuals can react to density, connectivity,

and mixed-uses; it is a key process through which the values inhering in these

urbane qualities can be enacted, as we elaborate below.

A bridge between cities, walking, diversity, and SMOs depends first on the

existence of a path between urban context and walkability. Substantial

conceptual and empirical scholarship shows that density, mixed urban use, and

connectivity enables pedestrian activity. Saelens, Sallis, and Frank (2003)

argue that these urban traits enable walking because they promote proximity

and directness of travel. Density – of people, housing, retail, etc. – leads

to walking by increasing the number and variety of destinations in an area,

and thus increasing the proximity to any one destination. Mixed urban use

also enables walking: when land uses are sufficiently co-located, distance

between uses decreases and walking becomes feasible and attractive (Smith et

al 2008; Boarnet and Sarmiento 1998). Directness of travel – i.e.

connectivity – is the second factor enabling walking. As noted, connectivity

13

reflects the ease of moving between origins and destinations. Connectivity

and directness is achieved when (Saelens, Sallis, and Frank 2003, 82) “route

distance is similar to straight-line distance.” This is higher when streets

are laid out in grid patterns with shorter blocks. As Smith et al note (2008,

238) directness is “expected to enhance walkability by making walking trips

relatively short, direct, and convenient”, by “slowing car traffic via

multiple stopping points”, and by providing alternative routes to one’s

destination. Demerath and Levinger (2003, 218) suggest that connectivity

enhances walkability by placing few “constraints on a person’s chosen route

between two destinations”, thus enabling freedom of movement and freeing

people to take part in as full a range of encounters as possible. In

addition, a growing body of statistical work (Boer et al 2007; Craig et al

2002; Berrigan and Troiano 2002; Greenwald and Boarnet 2002; Frank and Pivo

1994; Handy, Cao, and Mokhtarian 2006) reveals empirical relationships between

measures of these urban environmental traits and walking. For example,

Berrigan and Troiano (2002, 75) find that even after controlling for gender,

race, education, income, and health level, walking is significantly more

prevalent among U.S. adults who live in older homes and traditional urban

neighborhoods with “sidewalks”, “denser interconnected networks of streets”,

and “a mix of business and residential uses”.

Therefore, since walking is a prime means by which individuals directly

experience their city, we stress how it links urban physical and social

activities to create what we could term “socio-physical capital”. We thus

14

extend the social capital concept (e.g. Coleman 1988) to explicitly join with

physical surroundings. This undergirds SMOs. Walking permits individuals to

have (du Toit et al 2007, 1679) “frequent casual [face-to-face] contact,

whether intentional or spontaneous”, which is important for several reasons.

First, traits like trust, solidarity, friendship, and predictability enhance

encounters between diverse individuals and encourage collective political

action like SMOs. These traits may be facilitated by the face-to-face contact

of walking. Routledge (2003, 339) writes: there “are features of face-to-face

interaction [such as gestures, tone, and pitch]…that are highly informative;

these features are concealed in computer-based interactions.” These features

combine with (Storper and Venables 2004, 355) “visual contact” and “emotional

closeness” to build trust, solidarity, and predictability. Second, face-to-

face contact improves communication between diverse social elements,

especially when uncodified knowledge is involved. Uncodified information,

Storper and Venables (2004, 354) posit, is “only loosely related to the symbol

system in which it is expressed.” For instance, one can learn a formal symbol

system (i.e. the “syntax” and “grammar” of a language), but fail to decipher

certain information, subtleties, and idiosyncrasies in the system (i.e.

“metaphors”). We contend that much information inhering in the diverse social

forms of cities and neighborhoods is of this “metaphorical” variety, in that

(Storper and Venables (2004, 354) it “includes much linguistic, words-based

expression”, as well as reflecting the different experiences, dispositions,

and backgrounds of urban inhabitants. To successfully transmit the metaphors

15

of diverse groups to each other, face-to-face contact permits “a depth and

speed of feedback that is impossible in other forms of communication.”

In summary, by enabling face-to-face encounters with diverse social

influences, walking raises socio-physical capital and thus empowers

individuals to utilize diverse information for political ends. Solnit (2006)

writes that the “exercise of democracy begins as exercise, as walking around,

becoming familiar with the streets, comfortable with strangers, able to

imagine your own body as powerful and expressive rather than a pawn. People

who are at home in their civic space preserve the power to protest and revolt,

whereas those who have been sequestered into private space do not.”

Experiencing urban contexts through walking produces emboldened, empowered

individuals, who can then be forces for change through innovative political

forms like current-day SMOs.5 Thus, codifying in more specific hypotheses:

H5: In locales with more walking, there is a higher incidence of SMOs.

H6: Walking mediates the relationships between elements of urban context and SMOs.

H6a: Walking mediates the relationship between density and SMOs.

H6b: Walking mediates the relationship between mixed land uses and SMOs.

H6c: Walking mediates the relationship between mixed building ages and SMOs.

H6d: Walking mediates the relationship between connectivity and SMOs.

3. Data and Results

3.1 Data

16

Empirical analyses6 of the above hypotheses are pursued by combining old and

new data sources, described below. The Zip Code Tabulation Area (ZCTA) is the

unit of analysis. Developed as a “new statistical entity” for the 2000

Census, ZCTAs “are generalized area representations of U.S. Postal Service

(USPS) ZIP Code service areas” created to “overcome the difficulties in

precisely defining the land area covered by each ZIP Code”.7 There is a

nationwide total of 33,322 ZCTAs. We use ZCTAs primarily because they are the

smallest geography for which the dependent variables are available.8 ZCTAs

allow for more fine-grained measurement of urban contextual elements than

larger units like entire cities or counties used in much urban research.9

Descriptive statistics for all variables are provided in Table 1.

<Table 1 about here>

3.1.1 Dependent Variables – Social movement organizations

The data used to construct measures of SMOs are drawn from the 2007 ZIP Code

Business Patterns. This is a rich, new data source for social movements and

local political research; we are the first to use it to our knowledge. The

ZIP Code version of County Business Patterns, ZBP is “an annual series that

provides subnational economic data by industry.”10 ZBP data include the

number of establishments and employees in ZIP Codes, as well as payroll data.

Establishments are classified according to the 2002 North American Industrial

Classification System (NAICS), defined as “a single physical location at which

business is conducted or services or industrial operations are

performed….Establishment counts represent the number of locations with paid

17

employees any time during the year.” We use ZBP to attain the number of SMO

establishments in all ZIP Codes for a given year. ZBP includes three NAICS

categories we utilize as measures of SMOs: (1) Environment, Conservation, and

Wildlife Organizations (NAICS 813312); (2) Human Rights Organizations (NAICS

813311); and (3) Other Social Advocacy Organizations (NAICS 813319)11. We use

factor analysis to combine these three measures into a single SMO index, and

then use this index as the dependent variable in OLS regressions. These three

measures load highly on a single factor12, indicating an underlying SMO

dimension. We compute the factor scores and use them as values for the

dependent variable in subsequent regressions.

These ZBP data reflect some ideal-typical traits described above, but clearly

individual organizations vary substantially. One issue is how much groups

with paid employees may differ: do they deviate from the idealized conception

of SMOs as decentralized, anti-hierarchical, and informal organizations? To

assess this, we used Guidestar nonprofit financial data in a validation

exercise. Given the relationship between Guidestar and ZBP13, if the majority

of sampled Guidestar organizations possess modest financial resources, we can

conclude that the ZBP groups are more informal and decentralized SMOs. From a

sample of 50 Zip Codes with 1431 downloaded Guidestar organizations14, we find

that a majority are small to moderately sized: 10% have Total Assets of

$25,811 or less, and at least 25% of the organizations have less than $100,000

in Total Assets. 50% of the organizations have no more than $414,363 in

assets, and 40% have less than $250,000.15 These results show a substantial

18

overlap with the construct of small, volatile, and non-establishment SMOs.

Still, ZBP also contains older, larger, and more established organizations as

well.

3.1.2 Independent Variables

3.1.2.1 Urban contextual variables

Using data from the 2000 U.S. Census, we measure walking as the percent of

ZCTA workers 16 years and older who walk to work. Of course, individuals walk

for other additional purposes, including for leisure and exercise. We would

prefer to have measures of walking for additional destinations and purposes,

but walking to work is the only walking measure in the US Census.16 Freeman

(2001, 72) concurs. Despite this clear measurement error, we find important

results for walking thus measured. Second, we use data from the 2000 U.S.

Census and the 2000 ZBP ZIP Code Industry Detail File to calculate population

density, housing density, retail density, and employment density. The

denominator in these density quotients is ZCTA land area in square feet.

Third, our measure of connectivity is an estimate of city block density. We

divide the number of Census blocks in a ZCTA by land area in square miles.

Connectivity is enhanced with short city blocks, implying high block density.

We approximate city blocks with Census blocks, and measure connectivity as the

above quotient. Fourth, we employ an entropy variable to measure mixed urban

uses17. Ranging from 0 to 1, this variable takes higher values – indicating

mix – for more even distributions of ZCTA square footage across land uses.

Finally, we use 2000 Census data and Blau’s (1977) index of heterogeneity to

19

measure housing age diversity. This variable measures the probability that

two randomly chosen housing units in a ZCTA were built in different year

ranges.

3.1.2.2 Demographic and socioeconomic control variables

In addition to urban context variables, a number of other factors potentially

predict the dependent variables, and are therefore included. These are from

the 2000 U.S. Census for ZCTAs: population, median age, percent of population

15 years and older that is married, percent of households with children under

18, mean travel time to work in minutes for workers 16 years and older,

percent of population 25 years and older with a bachelor’s degree or above,

percent of population 5 years and older living in the same house for five or

more years, percent of housing units that are renter occupied, median

household income in 1999, median gross rent for specified renter-occupied

housing units, racial diversity, and foreign born diversity18. These are

classic in many urban studies. [TERRY: A reviewer asked for references to

studies that employ these “classic” measures. Can we leave the text as it

currently reads, or do we need to insert references here?] Yes lets ignore.

3.2 Results

3.2.1 Regression results

We estimate several regressions to examine the direct effects of density,

connectivity, mixed urban use, and walking on SMO activity, controlling for

relevant demographic and socio-economic factors. We report standardized slope

20

coefficients, computed after mean centering each independent variable and then

dividing by two standard deviations. Gelman (2007) argues for using two

standard deviations as it places all variables – continuous and binary – on a

common scale.

Table 2 reports Gelman standardized coefficients for regressions relating the

SMO index to independent variables described above: (a) social and economic

controls19, (b) urban contextual factors, and (c) walking. The sixteen

estimations in Table 2 are combinations of our density, land use mix, and

walking variables.

<Table 2 about here>

Coefficients on population density, housing density, retail density, and

employment density variables test H1. All four density variables indicate

significant increases in the incidence of SMOs with increases in density. For

instance, in estimations 5 – 8 of Table 2 we observe significant standardized

coefficients on housing density ranging from 0.25025 to 0.31932, close to the

largest magnitudes – and possible importance – of the variables in these

columns. As such, the consistently strong positive and significant

coefficients on these four density measures provide substantial evidence in

support of H1.20

The coefficients on the land use entropy variable provide tests of H2. We

expect positive and significant coefficients, but observe in Table 2

21

consistently negative and significant coefficients. H2 is therefore not

supported. However, these results may reflect entropy’s inadequacy as an

indicator of mixed urban use. For instance, Brown et al (2009) describe six

ways in which entropy measures may depict situations other than mixed use. It

remains for future research to identify, construct, and utilize better

measures of mixed land uses, since data limitations do not permit this here.21

Coefficients of housing age diversity test H3. Table 2 shows positive and

significant standardized coefficients: diversity of housing covaries with

SMOs, although the magnitude is smaller than other variables.

Coefficients on “Census blocks / sq. mi” test H4, which suggests that urban

connectivity facilitates SMOs. In all estimations in Table 2, we observe

positive relationships between Census blocks / sq. mi and the SMO index,

consistent with the hypothesis. The magnitudes of the standardized

coefficients on Census blocks / sq. mi are large compared to the other

variables.

Finally, the coefficients on “walked to work” in Table 2 test H5, which

suggests walking heightens SMOs. All models show significant increases in

SMOs with walking. The magnitudes of the standardized coefficients of walking

are larger than most other variables, suggesting the relative importance of

walking in predicting SMOs. An “alternative” walking measure – the percentage

of all workers who walk, bike, ride the bus, or ride the train to work – is

22

similarly positive on the incidence of SMOs. Underlying this alternative

measure is the idea that walking, biking, and public transportation are

complementary to and reinforce one another, since places conducive to biking

and public transport are also likely to be conducive to walking. By contrast,

we find a clear and strong negative relationship between SMOs and the

percentage of adults who drive to work. These results provide substantial

evidence in support of H5.22

The regression coefficients of interest are robust to tests for spatial

autocorrelation, multicollinearity, heteroskedasticity, and outlier deletion.

Results are also robust to the inclusion of a dummy variable which takes the

value of 1 if a ZCTA is in a Census Defined Place23 with at least 250,000

population. Results are also similar to Table 2 when regressions are

estimated on the smaller subset of 2,300 ZCTAs from Census Defined Places with

populations of at least 250,000. We also consider the possibility that

liberal ideology (in the American sense) may confound these relationships. In

county-level regressions with the percentage voting Democratic as a proxy for

liberalism, we find that liberalism neither predicts SMOs nor alters the

relationships between urban context and SMOs. Finally, these results still

pertain when the three SMO measures (Human rights organizations, Environmental

and Wildlife Organizations, and Other Social Advocacy Organizations) are

analyzed separately as dependent variables, instead of employing the SMO

index.

23

In summary, we find evidence in support of H1, (density), H3, (mixed building

ages), H4, (connectivity), and H5, (walking), but contrary to H2. (mixed land

use). Furthermore, the large magnitudes on the housing, retail, and

employment density variables, as well as on the connectivity and walking

variables suggest the relative importance of these factors in predicting SMOs

when compared to the other variables included in the regressions.

3.2.2 Mediation

H6 suggests a specific conceptual relationship between a third variable

(walking), the independent variable (urban context), and the dependent

variable (SMOs), specifically that walking mediates between urban context and

SMOs. Mediation implies a causal path (Judd and Kenny 1981; Baron and Kenny

1986) running from an independent variable, through a mediating third

variable, to the dependent variable. The most common test for mediation is

the Sobel test, of the following form: z = αβ αβ /

√α2σβ2+β2σα

2

√α2σβ2+β2σα

2, with the mediated effect calculated as a product of the

coefficients, αβ αβ , where α α is the slope coefficient on density,

connectivity, housing age diversity, or mixed urban use in a regression with

walking as the dependent variable, and β is the slope coefficient on the

walking variable in Table 2. For consistency with our hypotheses, α α and β

must take on positive values. The denominator is the standard error, where

σα2 σα

2 is the variance of α α and σβ

2 σβ2 is the variance of β .

24

Sobel tests indicate that for both density (H6a) and connectivity (H6d),

walking mediates the relationships with SMOs consistently with the mechanisms

of the hypotheses. There are two possible Sobel tests for each of the four

density measures, all eight of which show statistically significant mediation

effects. This is substantial evidence that the effects of density on SMOs are

transmitted through walking. Four out of eight Sobel tests reveal mediation

by walking of the relationship between connectivity and SMOs, thus some

evidence that effects of connectivity on SMOs are likewise transmitted through

walking. Sobel tests do not support H6b or H6c.

3.2.3 Alternative Dependent Variables

Walkable, dense, high connectivity, mixed use urban locales may be conducive

to other kinds of organizations or even for-profits. To explore local

conduciveness to SMOs relative to other types of organizations, we estimate

regressions with different organizational entities as the dependent variable

and compare the effects to regressions with SMOs as the dependent variable.

We use the same ZBP source to construct thirteen alternative dependent

variables. These variables are the factor scores from factor analyses of ZCTA

counts of entities in several NAICS categories. We create an artistic

industry composite index, ten separate composite indices from the financial,

insurance, and real estate (FIRE) industries, and two composite indices of

“Religious, Grantmaking, Civic, Professional, and Similar Organizations”

(NAICS 813).24

25

Using these thirteen measures as dependent variables, we estimate regressions

using the same independent variables in Table 2. In general, after estimating

these additional regressions we conclude that the SMO variable – as compared

to most of the thirteen other dependent variables – is most consistently and

significantly related as predicted by the hypotheses. Specifically, only two

of the thirteen alternative dependent variables – a Securities Dealing and

Brokerage composite index and a Business, Professional, and Political Party

Organizations composite index – have regression coefficients comparable in

direction and significance to those of the SMO index as the dependent

variable. However, these results shed light on the main findings by showing

that our urban diversity measures are distinctly important for phenomena –

like Securities Trading and Business and Professional Organizations – that are

information intensive, rely on knowledge transfer and spillovers, and depend

on face-to-face contact and deal-making, as stressed by Jacobs (1961) and

Sassen (2001). The proximity afforded by dense, connected, and walkable

cities enables this kind of efficient information transfer and face-to-face

contact. Cities are locations for both economic and political outcomes.

3.2.4 Limitations

Our current data and analyses have characteristic limitations of correlational

studies as compared to controlled experiments.

But experiments are unrealistic for most urban research of this sort. theThe

power of our ZCTA-level25 results frankly surprised us by their robustness.

26

Still the walking effects are so dramatic that we are the first to recommend

that others help us explore where and how they may hold or not.

Another data limitation is that the Census ZBP provides no information

on where a group’s membership comes from or where its activism occurs, instead

only the location of its “registered office”. To try to incorporate more

adequately the regions of activity and membership, we repeated our zctaZCTA

regressions for US counties. Using identical variables, we find minor

differences. At the county-level, coefficients on walking are positive and

significant half of the time; density is always positive and significant;

connectivity is positive and significant in all estimations except one; and

housing age diversity is always insignificant. 

4. Conclusions

Cities possess size, density, connectedness, and walkability, which combine to

enable learning, speed the creation and transfer of ideas, and generate and

enable bridging across diversity. Cities therefore become locales for social

change and hubs of innovativeness of many kinds – economic, political,

cultural, and even ethical. However, neighborhoods even within large cities

are not homogeneously urbane; drilling down to zip codes provides sharper

detail, an advantage of the zip code data that we suggest others consider

using. Encounters with diverse influences, perspectives, ideas, and issues

are fundamental to the orientation of SMOs. Although permitted by their

27

decentralized organizational form, we find that neighborhood specific social

diversity facilitates SMOs. Given their capacity to generate and facilitate

encounter with diversity, we report that walkable, dense, mixed-use, high-

connectivity urban contexts serve as a social setting to undergird SMOs. We

find that walking mediates between these urban contextual traits and the

encounters with social diversity central to SMOs.

The results in Part 3 support most of the hypotheses.26 The regression

analyses offer strong and consistent empirical evidence linking population

density, housing density, retail density, employment density, connectivity,

and mixed housing ages to higher incidence of SMOs in some 30,000 U.S. ZCTAs.

Potentially most interesting are the findings for walking, which we determine

raises the incidence of SMOs. “Walkability” may therefore be considered a

separate dimension of urban life that should be explored further by social

scientists more generally. Sobel tests provide considerable evidence that the

relationships of both density and connectivity with SMOs are enhanced by

walking. It is primarily through walking that an individual interacts with

and makes use of urban spaces, and is a key means by which the social

diversity inhering in these spaces is realized and utilized. To find that

walking effects are larger than income, local rent, or racial diversity and

that walking builds socio-physical capital is perhaps our major contribution.

These are simple and logical ideas, but this is the first systematic national

study to document such relations.

28

References

Acs, Zoltan, and Catherine Armington. 2004. “Employment Growth and

Entrepreneurial Activity in Cities.” Regional Studies 38:911-927.

Baron, Reuben M., and David A. Kenny. 1986. “The Moderator-Mediator Variable

Distinction in Social Psychological Research: Conceptual, Strategic, and

Statistical Considerations.” Journal of Personality and Social Psychology 51:1173-

1182.

Berrigan, David, and Richard P. Troiano. 2002. “The Association Between Urban

Form and Physical Activity in U.S. Adults.” American Journal of Preventive

Medicine 23 (Supplement 1):74-79.

Blau, Peter M. 1977. Inequality and Heterogeneity. New York: Free Press.

Boarnet, Marlon, and Sharon Sarmiento. 1998. “Can Land-use Policy Really

Affect Travel Behaviour? A Study of the Link between Non-work Travel and

Land-use Characteristics.” Urban Studies 35:1155-1169.

Boer, Rob, Yuhui Zheng, Adrian Overton, Gregory K. Ridgeway, and Deborah A.

Cohen. 2007. “Neighborhood Design and Walking Trips in Ten U.S.

Metropolitan Areas.” American Journal of Preventive Medicine 32:298-304.

Brown, Barbara, Ikuho Yamada, Ken Smith, Cathleen Zick, Lori Kowaleski-Jones,

and Jessie Fan. 2009. “Mixed land use and walkability: Variations in

land use measures and relationships with BMI, overweight, and obesity.”

Health & Place 15:1130-1141.

Carr, Lucas J., Shira I. Dunsiger, and Bess H.Marcus. 2011. “Validation of

Walk Score for estimating access to walkable amenities.” British Journal of

Sports Medicine 45: 1144-1148.

29

Castells, Manuel. 1983. The City and the Grassroots. Berkeley and Los Angeles, CA:

University of California Press.

Clark, Terry N., and Meghan Kallman. 2011. The Global Transformation of Organized

Groups: New Roles for Community Service Organizations, Non-Profits, and the New Political

Culture. Beijing, China: Ministry of Civil Affairs.

Coleman, James S. 1988. “Social Capital in the Creation of Human Capital.”

American Journal of Sociology 94 (Supplement: Sociological and Economic

Approaches to the Analysis of Social Structure): S95-S120.

Craig, Cora, Ross C. Brownson, Sue N. Cragg, and Andrea L. Dunn. 2002.

“Exploring the Effect of the Environment on Physical Activity: A Study

Examining Walking to Work.” American Journal of Preventive Medicine 23

(Supplement 1):36-43.

Della Porta, Donatella, Hanspeter Kriesi, and Dieter Rucht. 1999. Social

Movements in a Globalizing World. Houndmills England and New York: Macmillan

Press and St. Martin's Press.

Della Porta, Donatella and Massimiliano Andretta. 2002. “Changing Forms of

Environmentalism in Italy: The Protest Campaign on the High Speed

Railway System.” Mobilization 7:59-77.

Demerath, Loren, and David Levinger. 2003. “The Social Qualities of Being on

Foot: A Theoretical Analysis of Pedestrian Activity, Community, and

Culture.” City and Community 2:217-237.

du Toit, Lorinne, Ester Cerin, Evie Leslie, and Neville Owen. 2007. “Does

Walking in the Neighborhood Enhance Local Sociability?” Urban Studies

44:1677-1695.

30

Duncan, Dustin T., Jared Aldstadt, John Whalen, Steven J. Melly, and Steven L.

Gortmaker. 2011. “Validation of Walkscore for Estimating Neighborhood

Walkability: An Analysis of Four Metropolitan Areas.” International Journal of

Environmental Research and Public Health 8: 4160-4179.

Ferman, Barbara. 1996. Challenging the Growth Machine. Lawrence, KS: University

Press of Kansas.

Fischer, Claude S. 1975. “Toward a Subcultural Theory of Urbanism.” American

Journal of Sociology 80:1319-1341.

Florida, Richard. 2002. The Rise of the Creative Class. New York: Basic Books.

Frank, Lawrence D., and Gary Pivo. 1994. “Impacts of Mixed Use and Density on

Utilization of Three Modes of Travel: Single-Occupant Vehicle, Transit,

and Walking.” Transportation Research Record 1466:44-52.

Freeman, Lance. 2001. “The Effects of Sprawl on Neighborhood Social Ties: An

Explanatory Analysis.” Journal of the American Planning Association 67:69-77.

Gainsborough, Juliet. 2001. Fenced Off: The Suburbanization of American Politics.

Washington, DC: Georgetown University Press.

Gelman, Andrew. 2007. “Scaling regression inputs by two standard deviations.”

Statistics in Medicine 27:2865 – 2873.

Glaeser, Edward L. 1994. “Cities, Information, and Economic Growth.” Cityscape

1:9-47.

Glaeser, Edward L.------. 2000. “Cities and Ethics: An Essay for Jane Jacobs.”

Journal of Urban Affairs 22:473-493.

31

Graeber, David. 2003. “The Globalization Movement and the New New Left.” In

Implicating Empire: Globalization and Resistance in the 21st Century World Order, edited by

Stanley Aronowitz and Heather Gautney, 325-338. New York: Basic Books.

Greenwald, Michael, and Marlon Boarnet. 2002. “The built environment as a

determinant of walking behavior: analyzing non-work pedestrian travel in

Portland, Oregon.” Transportation Research Record 1780:33-42.

Grubesic, Tony H., and Timothy C. Matisziw. 2006. “On the use of ZIP codes and

ZIP code tabulation areas (ZCTAs) for the spatial analysis of

epidemiological data.” International Journal of Health Geographics 5:58. doi:

10.1186/1476-072X-5-58.

Handy, Susan, Xinyua Cao, and Patricia L. Mokhtarian. 2006. “Self-Selection in

the Relationship between the Built Environment and Walking.” Journal of the

American Planning Association 72:55-74.

Jacobs, Jane. 1961. The Death and Life of Great American Cities. New York: Vintage

Books.

Jacobs, Allen, and Donald Appleyard. 1996. “Toward an Urban Design Manifesto.”

In The City Reader, edited by Richard T. Legates and Frederic Stout, 164-

175. London: Routledge.

Jennings, Bruce. 2001. “From the Urban to the Civic: The Moral Possibilities

of the City.” Journal of Urban Health: Bulletin of the New York Academy of Medicine

78:88-103.

Judd, Charles M., and David A. Kenny. 1981. “Process Analysis: Estimating

Mediation in Treatment Evaluations.” Evaluation Review 5:602-619.

32

Knudsen, Brian. 2011. “The Local Ecology of New Movement Organizations.” PhD

diss., Carnegie Mellon University.

Knudsen Brian, Richard Florida, Kevin Stolarick, and Gary Gates. 2008.

“Density and Creativity in US Regions.” Annals of the Association of American

Geographers 98:461-478.

Kramer, Kathryn, and John Rennie Short. 2011. “Flânerie and the Globalizing

City.” CITY: analysis of urban trends, culture, theory, policy, action 15:323-342.

Krizek, Kevin J. 2003. “Residential Relocation and Changes in Urban Travel:

Does Neighborhood-Scale Urban Form Matter?” Journal of the American Planning

Association 69:265-281.

Logan, John, and Harvey Molotch. 1987. Urban Fortunes: The Political Economy of Place.

Berkeley and Los Angeles, CA: University of California Press.

McAdam, Doug, John D. McCarthy, and Mayer N. Zald, eds. 1996. Comparative

Perspectives on Social Movements. Cambridge: Cambridge University Press.

McCright, Aaron M., and Terry N. Clark. 2006. “The Political Opportunity

Structure of the Environmental Movement in U.S. Communities.” In

Community and Ecology: Dynamics of Place, Sustainability, and Politics, edited by Aaron

M. McCright and Terry N. Clark, 199-240. Amsterdam: Elsevier/JAI Press.

McDonald, Kevin. 2006. Global Movements: Action and Culture. Oxford: Blackwell.

Merrifield, Andy. 2002. Dialectical Urbanism. New York: Monthly Review Press.

Nicholls, Walter. 2008. “The Urban Question Revisited: The Importance of

Cities for Social Movements.” International Journal of Urban and Regional Research

32:841-859.

33

Offe, Claus. 1985. “New Social Movements: Challenging the Boundaries of

Institutional Politics.” Social Research 52:817-868.

Oliver, J. Eric. 2001. Democracy in Suburbia. Princeton, NJ: Princeton University

Press.

Park, Robert E. 1915. “The City: Suggestions for the Investigation of Human

Behavior in the City Environment.” The American Journal of Sociology 20:577-

612.

Peterson, Paul E. 1981. City Limits. Chicago: University of Chicago Press.

Routledge, Paul. 2003. “Convergence space: process geographies of grassroots

globalization networks.” Transactions of the Institute of British Geographers 28:333-

349.

Sassen, Saskia. 2001. The Global City: New York, London, Tokyo. Princeton: Princeton

University Press.

Saelens, Brian E., James F. Sallis, and Lawrence D. Frank. 2003.

“Environmental Correlates of Walking and Cycling: Findings From the

Transportation, Urban Design, and Planning Literatures.” Annals of

Behavioral Medicine 25:80-91.

Sennett, Richard. 1999. “The Spaces of Democracy.” In The Urban Moment:

Cosmopolitan Essays on the Late-20th-Century City, edited by Robert Beauregard and

Sophie Body-Gendrot, 273-285. Thousand Oaks: Sage Publications Inc.

Smith, Ken R., Barbara B. Brown, Ikuho Yamada, Lori Kowaleski-Brown, Cathleen

D. Zick, and Jessie X. Fan. 2008. “Walkability and Body Mass Index:

Density, Design, and New Diversity Measures.” American Journal of Preventive

Medicine 35:237-244.

34

Solnit, Rebecca. 2006. “Democracy should be practiced regularly, on foot.” The

Guardian,

June 6.

Stone, Clarence N. 1993. “Urban Regimes and the Capacity to Govern: A

Political Economy Approach.” Journal of Urban Affairs 15:1-28.

Storper, Michael, and Anthony Venables. 2004. “Buzz: face-to-face contact and

the urban economy.” Journal of Economic Geography 4:351-370.

Talen, Emily. 2006. “Design That Enables Diversity: The Complications of a

Planning Ideal.” Journal of Planning Literature 20:233-249.

Walks, R. Alan. 2004. “Place of Residence, Party Preferences, and Political

Attitudes in Canadian Cities and Suburbs.” Journal of Urban Affairs 26:269-

295.

Wirth, Louis. 1938. “Urbanism as a Way of Life” American Journal of Sociology 44:1-

24.

35

Footnotes

36

Variable N Mean StandardDeviation

2007 Human rights orgs. 35546 0.079808 0.5734572007 Environmental, wildlife, and conservation orgs.

35546 0.173217 0.690069

2007 Other social advocacy orgs. 35546 0.181118 1.0151022007 Social Movement Organization Index 35546 0.000000 0.9429942007 Securities Dealing and Brokerage Index

39646 0.000000 0.938594

2007 Artistic Index 39646 0.000000 0.8738872007 Depository Credit Intermediation Index

39646 0.000000 0.892089

2007 Real Estate Index 39646 0.000000 0.99801.002007 “813” Index #1 39646 0.000000 0.9337932007 “813” Index #2 39646 0.000000 0.8729872000 Population (in 1000s) 33167 8.599860 12.9790982000 Median Age 33178 36.745875 8.4464452000 Pct. Bachelor’s Degree and Above 32153 18.055106 13.7868792000 Racial Diversity Index 32165 0.219422 0.2112212000 Median Gross Rent (in 100s) 33178 4.746175 2.4141411999 Median Household Income (in 1000s) 32096 39.537554 16.2628262000 Travel Time to Work (in minutes) 32064 26.321232 8.3308332000 Pct. living in same house in 1995 32105 60.896390 12.7621762000 Pct. of population 15 years and older, Married

32165 60.522052 10.367637

2000 Pct. of households with children 0-17 years old

32096 33.055306 10.335134

2000 Foreign Born Diversity Index 27164 0.403940 0.2452252000 Pct. of occupied housing units, renter occupied

32097 24.832883 15.933493

2000 Population Density (in 1000s) 32406 1.132913 4.3362342000 Housing Density (in 1000s) 32406 0.478348 2.0117012000 Retail Density 31821 6.127513 126.0909092000 Employment Density (in 1000s) 34072 1.550155 30.7211722000 Census blocks per square mile 32406 161.803980 14588.1392142000 Housing Age Diversity 32140 0.788679 0.09740972001 Land Use Mix Entropy 32698 0.315632 0.2746282000 Pct. walked to work 32072 3.807581 7.079608

37

Table 1 – Descriptive Statistics of Key Variables Analyzed, ZCTA level

38

(1) (2) (3) (4) (5) (6) (7) (8)Social and EconomicPopulation (1000s)

0.16817** 0.08108** 0.24525** 0.19319** 0.15015** 0.07007** 0.21421** 0.16517**

Median Age -0.19219**

-0.20220**

-0.20020**

-0.20320**

-0.18719**

-0.19520**

-0.19720**

-0.19920**

Education 0.34434** 0.40941** 0.32533** 0.39640** 0.32733** 0.38739** 0.30931** 0.37838**Racial Diversity

-0.00400 -0.02402 0.01101 -0.01502 0.00100 -0.01802 0.01602 -0.01001

Median Rent (100s)

-0.06206**

-0.08308**

-0.02803 -0.04304* -0.07007**

-0.09009**

-0.04104 -0.05606*

Median Income (1000s)

-0.08709**

-0.10711**

-0.04104 -0.04404 -0.07908**

-0.09810**

-0.04204 -0.04505

Travel Time to Work

0.03303* -0.01702 0.01502 -0.03403* 0.01602 -0.03103* 0.00000 -0.04705**

Same House, 5 Years

-0.07608**

-0.09109**

-0.07307**

-0.09810**

-0.09409**

-0.10711**

-0.09109**

-0.11512**

Percent Married

-0.05906**

-0.10210**

-0.09309**

-0.14214**

-0.07007**

-0.11111**

-0.10010**

-0.14615**

Percent withChildren

-0.30931**

-0.31231**

-0.31431**

-0.31932**

-0.29229**

-0.29329**

-0.29930**

-0.30330**

Foreign BornDiversity

0.03003* 0.00901 0.04905** 0.03003* 0.03103* 0.01001 0.04905** 0.03003*

Percent Renters

0.04805* 0.12813** 0.05205* 0.14314** 0.01702 0.09610** 0.01401 0.10410**

DensityPopulation Density (1000s)

0.13413** 0.17718** 0.09409** 0.12012**

Housing Density (1000s)

0.28128** 0.31932** 0.25025** 0.27327**

Retail

39

DensityEmployment Density (1000s)ConnectivityCensus blocks / sq.mi

0.31431** 0.31431** 0.36036** 0.41542** 0.25526** 0.25626** 0.28529** 0.34034**

Land-use MixHousing Age Diversity

0.10210** 0.07307** 0.10210** 0.07407**

Land Use Entropy

-0.21421**

-0.31231**

-0.17918**

-0.27427**

WalkingWalked to Work

0.28929** 0.29730** 0.27628** 0.28829**

n 25789 25789 25578 25578 25789 25789 25578 25578R-squared (adj)

0.19520 0.18018 0.19920 0.18619 0.20220 0.18919 0.20521 0.19319

Dependent Variable: Social Movement Organization Index, 2007, **p<0.01, *p<0.05; Table reports Gelman standardized regression coefficients

Table 2 – Regression Results, ZCTA level: Many Significant Variables, but Walking is Robust(9) (10) (11) (12) (13) (14) (15) (16)

Social and EconomicPopulation 0.19720** 0.15215** 0.23824** 0.20921** 0.21922** 0.15516** 0.28028** 0.24324**

40

(1000s)Median Age -

0.19019**-

0.19720**-

0.20220**-

0.20420**-

0.19119**-

0.20020**-

0.19920**-

0.20220**Education 0.32833** 0.36537** 0.31732** 0.35836** 0.35636** 0.41742** 0.33834** 0.40240**Racial Diversity

-0.01502 -0.02903 -0.00300 -0.02102 -0.01001 -0.02903 0.00501 -0.01902

Median Rent (100s)

-0.08308**

-0.09409**

-0.06507**

-0.07407**

-0.06907**

-0.08809**

-0.04004 -0.05405*

Median Income (1000s)

-0.08308**

-0.09510**

-0.06106* -0.06406* -0.10210**

-0.12312**

-0.05706* -0.06306*

Travel Time toWork

0.01802 -0.01101 0.00801 -0.02002 0.05205** 0.01401 0.03103* -0.00801

Same House, 5 Years

-0.07508**

-0.08108**

-0.07508**

-0.08909**

-0.04204* -0.04805**

-0.04805**

-0.06406**

Percent Married

-0.11111**

-0.14014**

-0.13013**

-0.16216**

-0.05706**

-0.09409**

-0.09209**

-0.13313**

Percent with Children

-0.27628**

-0.27528**

-0.28529**

-0.28629**

-0.29630**

-0.29630**

-0.30330**

-0.30531**

Foreign Born Diversity

0.02803* 0.01401 0.04304** 0.03103* 0.02903* 0.00901 0.04805** 0.03103*

Percent Renters

0.03604 0.08509** 0.02703 0.08008** 0.09209** 0.17317** 0.08609** 0.17117**

DensityPopulation Density (1000s)Housing

Density (1000s)Retail Density 0.56857** 0.61061** 0.55355** 0.58959**Employment Density (1000s)

0.24525** 0.29129** 0.22623** 0.26527**

ConnectivityCensus 0.15616** 0.15015** 0.16517** 0.19019** 0.30631** 0.31031** 0.33834** 0.38639**

41

blocks / sq. miLand-use MixHousing Age Diversity

0.09610** 0.07708** 0.09910** 0.07307**

Land Use Entropy

-0.12312**

-0.17618**

-0.20220**

-0.28529**

WalkingWalked to Work 0.17718** 0.18118** 0.25826** 0.26226**n 25368 25368 25162 25162 25763 25763 25555 25555R-squared (adj)

0.24525 0.24024 0.24625 0.24124 0.20420 0.19319 0.20721 0.19820

Dependent Variable: Social Movement Organization Index, 2007, **p<0.01, *p<0.05; Table reports Gelman standardized regression coefficients

Table 2, contcontinued. – Regression Results, ZCTA level: Many Significant Variables, but Walking is Robust

BRIAN: How about trying to CUT the regressions by one digit, from 3 to 2; see if it causes problems. Then report that we changed to make more readable?

42

1 This research draws from Knudsen’s (2011) doctoral dissertation, The Local Ecology of

New Movement Organizations. We thank the UAR editors and reviewers, Daniel Silver,

Sam Braxton, and Scenes Project participants for careful suggestions. Most raw

data are available to interested researchers.

2 We seek to go beyond the urban label to specify distinct neighborhood

characteristics, or at least Zip Codes, the lowest level for which our social

movement organization data are available.

3 Of course, scholarship has not reached consensus regarding the relationships

between urban context, diversity, and social processes. Other theories (e.g. Wirth

1938) contend that large and dense cities subject individuals to friction,

disorientation and stress, which lead to atomization and undermines interaction,

trust, and social engagement. This perspective would be more consistent with the

emergence of NIMBY-type organizations out of homogeneous suburbs. By contrast,

regular exposure to social injustice that comes from frequent exposure to conflict

in dense and diverse cities may increase engagement with social movement

organizations. More subtle data are needed to disentangle such specifics.

4 Incidence is here understood to mean the fact of occurrence, and while we aim for

causality, most data only permit documenting associations.

5

Our analysis probably applies more to modern Western societies since

comparatively, these cities may have more open civic activities and active

citizens. We have work in progress to replicate these results in other societies.

6 Any calculations or results not in Tables 1 or 2 are available upon request.

7 For additional information on ZCTAs, see the ZCTA Technical Documentation at

www.census.gov.

8 The dependent variables use data from the Zip Code Business Patterns (ZBP),

described below. Whereas ZCTAs are the units for most of the independent

variables, ZBP uses USPS ZIP Codes. The ZIP Codes used by ZBP are those reported

by the businesses or establishments, or on administrative address lists. The

Census Bureau built ZCTAs in 2000 based on both residential and commercial

addresses, but prioritized residential addresses because they were verified during

the decennial Census operations. The Census Bureau did not have the same level of

verification for commercial address locations, and as such the resulting ZCTAs may

not match commercial addresses quite as well as residential address locations.

This, along with imperfect correspondence in areal representation between ZCTAs and

ZIP Codes (Grubesic and Matisziw 2006), may be a small source of measurement error

in our analyses. Most of the 30,000+ ZIP Codes from ZBP match the Census ZCTAs.

9 Also, past studies (Krizek 2003) show that neighborhood-scale urban form

variables affect travel behavior, which is an important consideration for our

analysis of walking.

10 See description of County Business Patterns at www.census.gov.

11 Other Social Advocacy Organizations are establishments engaged in social

advocacy except human rights and environment, conservation, and wildlife

preservation. For more information on these three NAICS categories, see

description of NAICS at www.census.gov.

12 Human rights organizations (NAICS 813311) have a factor loading of 0.850;

Environmental, wildlife, and conservation organizations (NAICS 813312) have a

factor loading of 0.586; Other social advocacy organizations (NAICS 813319) have a

factor loading of 0.911.

13 ZBP establishments are categorized by NAICS, while Guidestar organizations are

categorized by National Taxonomy of Exempt Entities (NTEE) codes. Correspondences

between NAICS and NTEE are available from the National Center for Charitable

Statistics. For instance, the NAICS code (813311) for Human Rights organizations

corresponds to the following 21 NTEE codes: Q30 - International Development; Q31 -

International Agricultural Development; Q32 - International Economic Development;

Q33 - International Relief; Q70 - International Human Rights; Q71 - International

Migration and Refugee Issues; Q99 - International, Foreign Affairs, and National

Security, Not Elsewhere Classified; R20 - Civil Rights; R22 - Minority Rights; R23

- Disabled Persons' Rights; R24 - Women's Rights; R25 - Seniors' Rights; R26 -

Lesbian and Gay Rights; R30 - Intergroup and Race Relations; R40 - Voter Education

and Registration; R60 - Civil Liberties; R61 - Reproductive Rights; R62 - Right to

Life; R63 - Censorship, Freedom of Speech, and Press; R67 - Right to Die and

Euthanasia; R99 - Civil Rights, Social Action, and Advocacy, Not Elsewhere

Classified.

14 Specifically, we download organizations from 71 NTEE codes that are linked with

the three NAICS codes that we use for the dependent variable.

15 In the upper half of the distribution, 60% of the organizations have less than

$750,000 in assets and 65.33% less than $1,000,000. Total Assets become large in

the very upper deciles, but 80% of the organizations still have assets of less than

$3 million. The skew of the distribution comes into view as Total Assets rapidly

enlarge to a maximum of $644 million from about $9 million at the 90th percentile.

Overall, even though there are large, wealthy organizations at the upper end of the

distribution, a majority of the groups are small to moderately sized, at least in

assets.

16 “Walkscores”, the walkability data from the website walkscore.com, are

calculated for an address based on the number and diversity of destinations nearby

and the distance to them. As such, this measure may better reflect the multiple

destinations to which persons may walk. One could calculate walkscores for random

samples of addresses from all U.S. ZCTAs, and then calculate estimated average ZCTA

walkscores. These average walkscores could be used as a walking variable in

regressions similar to those performed in this paper. However, one shortcoming is

that the destination-based walkscore data only reflects opportunities to walk

(Duncan et al 2011; Carr, Dunsiger, and Marcus 2011) whereas our Census measure

captures actual walking behavior.

17 Downloaded from GeoDa Center for Geospatial Analysis and Computation, calculated

by Lee Mobley as RTI Spatial Impact Factor Data – beta version 2.

18 Racial and foreign born diversity are calculated as Blau’s index of

heterogeneity.

19 Our regressions include as control variables those measures that are relevant to

the analysis and are most readily available given the data we employ and our chosen

unit of observation. However, given data limitations, we are currently unable to

analyze several other potentially relevant controls. These include accessibility

to public transit, availability to office space, and median commercial rent.

20 Coefficients on log-transformed density measures are generally similar to the

above, with some minor differences.

21 Land use entropy is calculated as a combination of the percentages of ZCTA

square feet in different land use categories. As an alternative, one could utilize

these separate percentages as their own measures of mixed land uses. The analysts

who calculated the entropy measure used in this paper were unable to provide us

with separated percentages.

22 Is the effect of walking on social movement organizations confounded by omission

of accessibility to the central business district (CBD)? New, small firms or

organizations may find accessibility to the CBD desirable. CBD districts may also

be more walkable. We thus added a distance to CBD mreasuremeasure in our

regressions. The Census 2010 report “Patterns of Metropolitan and Micropolitan

Population Change: 2000 to 2010” geocodes the addresses of the city hall in the

largest principal city of each metropolitan area. Since city halls are usually

located near their city’s original CBD, they are a reasonable proxy measure of the

CBD. Using ARCGIS, we calculated distances from each surrounding ZCTA centroid to

the CBD within the 363 metro areas. Social movements are fewer in ZCTAs farther

from the CBD. However, coefficients on walking to work, density, connectivity, and

housing age diversity all remain significant. Richard Greene, Northern Illinois

University and University of Chicago, assisted with the CBD computations.

23 Census Defined Places (CDP) can be thought of as urban municipalities. For

example, the New York city CDP has a population of 8 million. By contrast, the New

York City – Northern New Jersey – Long Island Metropolitan Statistical Area (of

which the New York city CDP is a part) has a population of over 21 million.

24 To create the artistic industry composite index, we save the factor scores from

a factor analysis of the ZCTA counts in the following NAICS categories: Theatre

companies and dinner theatres (NAICS 711110), Dance companies (NAICS 711120),

Musical groups and artists (NAICS 711130), and Independent artists, writers, and

performers (NAICS 711510). We similarly use factor analysis on the ZCTA counts in

a variety of other NAICS categories to calculate the following ten FIRE industry

composite indices: (i) Depository Credit Intermediation composite index; (ii) Non-

depository Credit Intermediation composite index; (iii) Activities related to

Credit Intermediation composite index; (iv) Commodity Contracts Dealing and

Brokerage composite index; (v) Securities Dealing and Brokerage composite index;

(vi) Other Financial Investment Activities composite index; (vii) Insurance

Carriers composite index; (viii) Insurance Agencies, Brokerages, and Other

Insurance Related Activities composite index; (ix) Real Estate composite index; and

(x) Holding Company composite index. Information on the specific NAICS categories

that comprise these ten indices can be provided upon request. Finally, we perform

a factor analysis on the ZCTA counts for the following: Religious Organizations

(NAICS 813110), Civic and Social Organizations (NAICS 813410), Business

Organizations (NAICS 813910), Professional Organizations (NAICS 813920), Labor

Unions and Similar Organizations (NAICS 813930), and Political Organizations (NAICS

813940). The factor analysis extracts two factors, with business, professional, and

political organizations loading on the first factor, and religious, civic/social,

and labor organizations loading on the second factor. The composite indices are the

factor scores of the two factors.

25 DELETE FOOTNOTE. TOO ELEMENTARY. Sorry idont know what #26 is; deleted with the

text that it was part of.

26 With one-time data we cannot measure causality but only association in all

relations in Part 3. In further work using data over time, we find some evidence

of reciprocal effects. However, to suggest that people who like to walk move to

neighborhoods with more SMOs, and vice versa, is not inconsistent with the main

hypotheses discussed.