SOCIAL NETWORKS, GEOGRAPHY, AND

NEIGHBOURHOOD EFFECTS

Ron Johnston and Charles Pattie

THIS ARTICLE IS TO APPEAR IN THE

ENCYCLOPEDIA OF SOCIAL NETWORKS

Not to be cited without

the authors’ consent

It is through personal networks that society is structured and the individuals

integrated into society. … daily life proceeds through personal ties: workers

recruit in-laws and cousins for jobs on a new construction site; parents choose

their children‟s pediatricians on the basis of personal recommendation; and

investors get tips from their tennis partners. … The interactions among the

abstract parts of society – “the family”, “the economy”, and so on – usually

turn out to be personal dealings between real individuals who know one

another, turn out to be operations of personal networks. … All through life, the

facts, fictions, and arguments we hear from kin and friends are the ones that

influence our actions most. Reciprocally, most people affect their society only

through personal influences on those around them. Those personal ties are also

our greatest motives for action: to protect relatives, impress friends, gain the

respect of colleagues, and simply enjoy companionship.

(Tilly, 1982: 3)

In one of social science‟s classics, Festinger et al. (1950) showed a strong relationship

between distance and social interaction. Social networks are spatially structured, as

both cause and effect. In towns and cities, people congregate into residential areas to

live among those they want to have social contact with, and to avoid others – as

demonstrated by the increasing popularity of gated communities around the world

(Atkinson and Blandy, 2006). Their social interactions are then structured within this

patterning of residential choices, creating a geography of social networks. The

existence of such a geography underpins much work across the social sciences: people

create urban environments which structure their social lives (Michelson, 1970). But

that geography is more often assumed than demonstrated, at least explicitly. Few

geographers have explored the network geographies (early exemplars rarely followed

up include Stutz, 1971; Wheeler and Stutz, 1971; Johnston, 1974), even though one of

their discipline‟s most innovative concepts – time geography (Hägerstrand, 1976,

1982, 1984) – provided an analytical framework. Instead they, like many other social

scientists, have built models and analytical schemas on the assumption that social

interaction is spatially structured, and interpreted observed behavioural patterns as

evidence that this is so. Similarly, many studies that focus on social networks focus on

2

the interactions among members and their impacts, without paying great attention to

their geography.

One area of research founded on the key assumptions that social interaction has a

spatial architecture and that information flows through those geographically-

structured spaces strongly influence attitudes and behaviour concerns neighbourhood

effects. People talk to their neighbours and the outcome of their conversations may be

changes in what they know and think about a subject – such as the candidates

standing at a forthcoming election. The result may be a behavioural change – they

vote for another candidate than the person initially preferred. A single conversation

may be enough to change one‟s mind in some cases; many may be needed in others.

But the result is likely to be the same – a spatial polarisation of opinions as more

people respond to the information reaching them through their neighbours and thus, in

this case, a polarisation of voting patterns.

Knowing the geography of social networks – who talks to whom, and where – is thus

key to appreciating how opinions are shaped and behavioural patterns formed.

Building that knowledge base involves drawing together information from a range of

studies – of network structures and geographies, information flows, attitude change

and behavioural consequences. The following sections follow that sequence, focusing

on neighbourhood effects in the geography of voting.

RESIDENTIAL CHOICE AND NEIGHBOURHOOD EFFECTS

The United States, according to Bishop (2008: 5), has recently developed a „stark

geographical pattern of political belief, one that has grown more distinct in

presidential elections since 1976. … [It] has been sorting itself, sifting at the most

microscopic levels of society, as people have packed children, CDs, and the family

hound and moved‟. The criteria people deploy in selecting new homes at the

neighbourhood scale include making „choices about who their neighbors will be and

who will share their new lives‟ which, he contends, have political impacts. Thus at the

county scale, whereas in 1976 less than one-quarter of all Americans lived in places

where one candidate for the presidency won by a landslide, by 2004 the proportion

was more than one-half. The country became spatially more polarized politically

because in their migration patterns „people were creating new, more homogeneous

relations‟ (p. 6: though see Cain, 2009, on the scales of sorting and representation).

Those homogeneous neighbourhoods become self-perpetuating societal divisions:

„The like-minded neighborhood supported the like-minded church, and both

confirmed the image and beliefs of the tribe that lived and worshipped there‟ (Bishop,

2008, 6). Other local institutions – schools, formal clubs and associations etc. –

sustain and enhance these processes, as do informal interactions with neighbours. A

greater homogeneity of ways of living shapes greater spatial polarization of political

beliefs and voting patterns (Gimpel and Lay, 2005).

In the United Kingdom, too, Curtice and Steed (1982) noted growing spatial

polarization of the electorate – later confirmed at a variety of scales (Johnston and

Pattie, 2006). This was also – at least in part – associated with selective migration

(although see Denver and Halfacree, 1992; McMahon et al., 1992). To others, a much

more important influence has been the operation of contextual or neighbourhood

3

effects. The seminal work of Butler and Stokes (1969: 182) concluded that „once a

partisan tendency becomes dominant in a local area processes of opinion formation

will draw additional support to the party that is dominant‟, and Miller (1977: 48)

argued that:

… contact is a condition for consensus … social contacts are structured by

family, choice of friends, social characteristics and locality. If party appeals to

group interest or group attitudes evoke any differential political responses, the

patterns of contact between individuals will tend to increase the political

consensus within high-contact groups.

Indeed, locality proved a better predictor of how people voted than their social

characteristics because „people who talk together vote together‟ (Miller, 1977: 65).

Others argued similarly, Andersen and Heath (2002: 126) contending that:

… we would expect to find tendencies towards class voting to be reinforced

among voters who regularly associate with others from the same social class.

On the other hand, we would expect to find the tendency towards class voting

to be undermined among voters who frequently interact with people from

other social classes since the interaction will tend to move them toward

agreement with members of other social classes. Simply put, the more that

people interact with members of other social classes, the weaker we expect

class voting to be.

Many studies using aggregate (ecological) data have generated findings consistent

with this hypothesis – that there is a greater spatial polarization of voting for a given

party than there is of the social groups who tend to support that party. But almost all

provide circumstantial evidence only; the patterns are consistent with the

neighbourhood effect, but the processes are unobserved (Doreian, 2001).

Underpinning the argument for the neighbourhood effect is a series of propositions.

1. Locational decisions involve a considerable degree of social selection; people

choose to live in residential neighbourhoods where people like themselves

dominate.

2. The neighbourhood social networks that people join are thus dominated by

people like themselves, not only in their individual characteristics but also

their ideologies, attitudes and behaviour. Interaction with them sustains and

may even strengthen their own positions; living among people who think and

act like you can make it even more likely to that you will think and act

accordingly.

3. Nevertheless, for a variety of reasons few local areas are entirely

homogeneous. Some social contacts within the neighbourhood are thus likely

to expose people to attitudes and behaviours different from their own. The

majority locally will have less exposure to such „deviant‟ tendencies than

those in the minority will have to the majority‟s norms, so it is more likely that

the minority will be converted to the majority view – „conversion by

conversation‟ – than vice versa. The result will be the observed polarization.

These propositions in turn rest on a series of assumptions, the most important of

which is that much social interaction takes place in localized social networks. Such

networks are extremely unlikely to be isolated – many members will have links to

either or both of other, non-local networks (based on workplaces or family/kin, for

example) and separate networks in adjacent neighbourhoods. Such „external‟ links

(which may be weaker in intensity than the strong links within their core, localized

4

networks: Granovetter, 1973) are continual sources of new information, providing

stimuli that in some cases generate altered attitudes and behaviour. The local social

network is thus a structure not only within but also to and from which information

(almost invariably interpreted information) flows. But relatively little is known about

such flows: they are assumed to exist because observed patterns of behaviour are

consistent with models and hypotheses predicated on their existence.

Such arguments apply to a much wider set of attitudes and behaviours than those

associated with electoral decision-making. Just as political information flows through

such networks so too does material linked to other types of behaviour – such as the

adoption of innovations (see Mark, 1998, on musical preferences). Furthermore, such

networks are also major conduits for other flows – infectious diseases, for example.

Again, aggregate patterns are consistent with such models but the underlying

processes – the actual flows along the network links – are often not revealed, with

some exceptions (such as Rothenberg et al., 2005; Rothenberg, 2007).

In reviewing literature on the links between local social networks, information spread

and attitudes/behaviour, therefore, we have to combine works directly addressing the

hypotheses that social networks are geographically concentrated and that flows

through those networks influence attitudes and behaviour with a wider set which

identifies patterns consistent with those hypotheses but does not reveal the ongoing

processes. Much of our attention focuses on political attitudes and voting behaviour as

one example of such processes and patterns.

ON THE SPATIALITY OF SOCIAL NETWORKS

A core social science argument has identified a major difference between rural and

urban areas in the nature of social interaction there; the former were characterized by

gemeinschaft – or community – whereas the latter were characterized by gesellschaft

– or association (Tönnies, 1887). Rural areas and relatively small settlements were

assumed to display intense patterns of social contact in small, tightly-knit social

networks, whereas urban areas were characterized by more diverse and transient

contact patterns, with relatively few intense relationships, reflecting the fragmentation

of such places (as with the spatial separation of home and workplace). Empirical

research challenged this, however, identifying both communities exemplifying

gemeinschaft within urban areas (especially working-class residential areas and

minority immigrant enclaves) and also many „urban‟ patterns of living spreading into

rural areas, leading to the breaking down of well-established local communities. (The

classic studies of the former include Gans, 1962, and of the latter Pahl, 1965; see also

Dunbar, 2008.)

If we want, in Tilly‟s words, to „protect relatives, impress friends, gain the respect of

colleagues, and simply enjoy companionship‟ then we must interact with them –

usually, though not necessarily, through face-to-face contact. As geographers have

stressed since a pioneer established that theirs is a „discipline in distance‟ (Watson,

1955; Johnston, 2003), most encounters, especially frequent encounters with kin,

colleagues and friends, are spatially constrained, not least because of the time, cost

and effort involved in overcoming the friction of distance: social worlds are

geographically structured.

5

These contentions were substantially exemplified by research in a variety of

Californian places (Tilly, 1982). Most respondents had social networks comprising

15-19 identified individuals; the largest group within those networks comprised kin

(over 40 per cent of all those named), with work colleagues and neighbours each

comprising a further 10 per cent. Non-kin, non-work associated neighbours did not

dominate, therefore, but were a significant component of people‟s contact circles.

Nevertheless, local people – who could be in two or even all three of those categories

– were a major component of the average social network: this comprised some 16

persons, of whom 5 lived within five minutes drive of the respondent‟s home and a

further 6 between 5-60 minutes drive away. And while there were differences

between type of settlement (semi-rural, town, metropolitan, regional core) in the

number of local (within a five-minute drive) kin named, there were few differences in

the number of non-kin; respondents in each type of place had the same number

(averaging c.3.6) living within 5 minutes drive of their home (i.e. neighbours, though

not necessarily individuals commonly involved in chance meetings), although those

living in the metropolitan area and its core named more people living further away

(i.e. they had both larger and spatially more dispersed social networks than their rural

counterparts). Overall, people from small towns were more involved with their fellow

residents than those living in larger settlements and „urbanites [especially high income

urbanites] substitute more distant relations for the foregone local ones‟ (Tilly, 1982:

167): Tonnies‟ binary split had not entirely dissipated.

Tilly (1982: 174-175) also found variations in the spatial boundedness of contacts

according to their nature: the percentage that were with near-neighbours varied

considerably depending on whether they were sociable (e.g. visiting and having

dinner together) through discussing a hobby or personal issues, obtaining advice on

important matters, and lending money and:

As one moves from exchanges for which distance is crucial to ones for which

it is a marginal cost, from contacts requiring frequent physical presence to

ones calling for occasional interactions possibly by telephone or mail, and

from casual matters to critical matters, the advantage of close associates

declines. For sociable interactions, distant associates were much less often

cited than nearby ones. For discussion of hobbies, which often involves

engaging in the hobby together, nearby associates were again more commonly

cited, though not as much more. Physical presence promotes discussing

personal matters, but it is not essential and the advantage of local associates is

marginal. Giving advice on important decisions and lending money in an

emergency can easily be done occasionally and at a distance, and there is no

advantage to proximity.

If, therefore, much of the politically-relevant information flowing through social

networks – much of which may be unstructured and unplanned – involves face-to-face

interaction, neighbourhood circles are likely to be important.

This conclusion is sustained by Huckfeldt‟s (1983) Detriot study. His respondents‟

networks were very much structured by social class, and across all classes around 40

per cent had a majority of their friends drawn from within a 10-minute drive, with less

than one-third having no friends within that radius. Local social context was an

important influence on friendship choice, however; people living in areas where class

x dominated were more likely to have one or more friends drawn from that class,

whatever their own class. Thus people in a minority in an area were likely to be

6

exposed to the majority view, which was likely to have the hypothesized effects on

political attitudes. Huckfeldt (1986: 50) found in Buffalo that working class

respondents were much more likely to identify with the Democratic party if they lived

in strong rather than weak working class neighbourhoods (0.60 as against 0.48: for

non-working-class individuals the proportions were 0.49 and 0.37); non-working class

individuals were less likely to identify as Democrats but both classes were more likely

to do so if they lived in working class neighbourhoods. Furthermore, members of the

middle class (according to their occupation etc.) who identified with the working class

were much more likely to identify as Democrats the more working-class friends they

had and the more working-class the neighbourhood in which they lived. These

differences according to context extended well beyond political affiliation: friendship

selection, ethnic loyalties, and residential satisfaction were also linked to

neighbourhood social characteristics and were influenced by both structured, primary-

group, and unstructured (casual) interactions within the local milieu.

A study of contextual effects on voting in South Bend, Indiana at the 1984

Presidential election found that most discussion partners lived in the same

neighbourhoods (Eagles et al., 2004: the original study by Huckfeldt and Sprague,

1995, is discussed in more detail below). The survey interviewed not just the sampled

respondents (clustered within neighbourhoods) but also their main discussion

partners. Most conversation partnerships were quite local: on average, discussants

lived 2.76 miles apart. Conversations with neighbours tended to be over very short

distances – on average they lived half a mile apart. Conversations between family

members, work colleagues or fellow church members spanned larger distances – 3.8,

3.6 and 3.1 miles respectively on average. That said, how far apart discussion partners

lived had no independent impact on the transfer of political influence within the

conversation partnership (Eagles et al., 2004: 215). Other factors were much more

important in this regard: the more alike discussion partners were – the closer their

friendship, sharing the same religious affiliation, sharing the same partisanship, and so

on – the more likely they were to agree.

These survey data pre-date widespread use of the internet and mobile phones, which

many argue have changed the nature of conversation networks; proximity is no longer

as important for sustaining contacts and delivering information, it is claimed.

Wellman and Potter (1999) identified three types of community – lost, saved and

liberated – which differ, among other characteristics, on the importance of face-to-

face and phone contact. Surveys of Toronto residents in the 1960s showed that

although „people who live near each other continue [as in traditional gemeinschaft

communities] to have more frequent contact‟ (p. 64) nevertheless proximity in this

case is a relative concept. Most of the networks involving non-kin friends were

metropolitan-wide in their spatial range rather just the respondents‟ immediate

neighbourhoods; only 13 per cent of their most intimate relationships were with

people living in the same neighbourhood (Wellman, 1979, 1996).

Wellman‟s data allow exploration of whether distance was an important constraint on

social interaction prior to the internet‟s creation, and provided a baseline against

which later studies could be assessed. A 1978 survey showed not only the expected

distance-decay pattern in the intensity of social interaction with both kin and non-kin

intimates, but also a marked decrease in the frequency of face-to-face contacts if the

distance between their addresses exceeded five miles (Mok et al., 2007) – although no

7

respondents had most of their active social ties with individuals living within one

mile‟s walking distance of their homes (Wellman et al., 1988); telephone contact only

starts to decline beyond a distance of 100 miles. A 2005 survey of residents in the

same area showed that, as in 1978, face-to-face contact declined above an inter-home

distance of five miles, and phone contact at about 100 miles, but email contact was

only slightly sensitive to distance. There is thus a continuing pattern of face-to-face

contact being locally-, and phone contact regionally-structured, but internet-enabled

communication is largely unconstrained by distance (Mok et al, 2009). Social

networks have not been transformed but rather extended, therefore, and the relative

importance of local as against regional and distant contacts depends on whether they

are kin or non-kin, intimates or not, since the intensity of contact varies by group, as

well as by income (Carrasco et al., 2008).

A parallel study of movers to a „wired‟ suburb (Hampton and Wellman, 2001) found

that those who did not realize the internet‟s potential experienced a decline in their

social contacts after moving, whereas those who did experienced no change – but this

included contacts with their neighbours as well as with more distant others. Hampton

and Wellman thus concluded – p. 491 – that although for some the move reduced both

contact with and support from friends and relatives, for internet users being wired

fostered contact and support both near and far, in what Wellman (1999) describes as a

more „loosely-coupled world‟.

As cyberspace becomes more important and localized, intense communities

apparently decline in their significance we move towards what Wellman (2001, 3; see

also Zelinsky and Lee, 1998) terms glocalization:

Except in situations of ethnic or racial segregation, contemporary Western

communities are usually loosely-bound, sparsely-knit, ramifying networks of

specialized ties. Rather than being full members of one solidary

neighbourhood or kinship group, community has become “glocalized”.

Contemporary urbanites juggle limited memberships in multiple, specialized,

far-flung, interest-based network communities as they deal with shifting

amorphous networks of kin, neighbours, friends, workmates, and

organizational ties. Only a minority of network members are directly

connected with each other. Most friends and relatives live in different

neighbourhoods; many live in different metropolitan areas. At work, people

often work with distant others and not those sitting near them. People usually

obtain support, sociability, information and a sense of belonging from those

who do not live in the same neighbourhood.

Piselli (2007: 872), on the other hand, asks „have places – or local areas comprising

the values, knowledge, institutions, productive skills, and feelings of belonging on

which the recognition and self-recognition of local identity are grounded – lost their

importance?‟ and answers „It appears that they have not‟. New types of community

may be emerging „based largely on interactions devoid of physical contact and

reciprocal recognition of identities‟ (p. 875) but this is just one of the many ways in

which people interact: it enriches and expands, rather than replaces the „social

networks that define and redefine places, which change their functions, features, and

symbolic meanings‟ (p. 875).

INFORMATION FLOW THROUGH NETWORKS AND NEIGHBOURHOOD

EFFECTS

8

Whatever their spatial dimensions, social networks are communication conduits

through which people exchange information (which may be either factual only or

involve value judgements). Those flows can influence beliefs, attitudes and

behaviour, so knowing who speaks to whom, about what, can be crucial in exploring

who thinks and does what. In political contexts, for example, this can include

discussions about a variety of matters – parties‟ policies, governments‟ performance

and individual politicians‟ leadership credentials. In turn, such material may influence

how people vote. Study of social networks in operation can thus enable investigations

of political action and advance understanding of election outcomes. Further if many

social networks, especially those based on face-to-face interaction, are spatially

configured, then information flowing through locally-focused networks should

generate clear patterns of political behaviour.

Following this argument, research should focus on the flow of electorally-relevant

information through local networks but few have adopted this format, for a variety of

reasons – many associated with the cost of such intensive research strategies. Thus, as

exemplified here, most research outcomes have either: identified voting patterns

consistent with the neighbourhood effect concept, inferring that these have been

generated by local residents‟ conversations; or studied decision-making in the context

of people‟s conversations, which if they show that these lead to some changing their

minds can imply an outcome consistent with neighbourhood effects. The two should

be integrated, but few studies have done so, hence the organization of the following

sections.

Local social context and neighbourhood effects

During conversations within a social network a majority of whose members support

one view/candidate/party, the weight of opinion encountered is more likely to lead to

adherents of a minority view switching to the majority than vice versa. The outcome

would be the majority view dominating the network to a greater degree than could be

predicted from knowledge of individual members‟ personal characteristics alone. And

if those conversation networks are spatially constrained, the political complexion of

areas should be more polarized than their social composition implies – as suggested

by Cox (1969) in a seminal paper.

This argument has underpinned a large number of ecological studies using aggregate

areal data at a variety of scales: the independent variables, representing the local

context, are usually census socio-economic and demographic data and the dependent

variables are election results. In many the areal units deployed are at much larger

spatial scales than the local neighbourhoods within which much social interaction is

assumed to occur. Nevertheless, if they identify polarized patterns associated with the

neighbourhood effect this offers circumstantial evidence sustaining that argument. If,

say, the Labour party in a dominantly working-class British Parliamentary

constituency (with c.70,000 voters) has a greater share of the votes cast than predicted

from its class composition (Johnston et al., 1988), this could indicate that: (a) more

neighbourhoods in the constituency have predominantly working-class than non-

working-class populations; and (b) that therefore a pro-Labour neighbourhood effect

dominates across the constituency. There may also be differences between types of

neighbourhood within urban areas: Cox (1968), for example, found strong evidence of

9

neighbourhood effects in London‟s suburbia – where local social networks might be

more developed than in inner city areas characterized by much greater population

mobility (suburban gemeinschaft but inner-city gesellschaft) – and Walks (2004,

2005) has identified growing inner city-suburban polarization in both British and

Canadian cities more recently, with surveys suggesting that much of this in the

Canadian context is a consequence of households selecting to live in areas among

others with whom they (assume that they) share basic, including political, values

(Walks, 2006).

Such circumstantial evidence (reviewed in Johnston and Pattie, 2006; see also Cho

and Rudolph, 2008, and, more generally, Blasius et al., 2007) has increasingly been

sustained by studies which merged survey data on individuals‟ voting behaviour with

census data on their local contexts. (Early examples were Wright, 1977, and Harrop et

al., 1992.) Such work has been extended recently with studies using contextual data

for what are termed „bespoke neighbourhoods‟ in which very small area census data

are used to identify the characteristics of the immediate area around each separate

respondent‟s home. These too have found patterns entirely consistent with the

neighbourhood effect; voters from any class background were more likely to vote for

a party the larger the proportion of the local population drawn from that party‟s

„natural‟ class supporters (McAllister et al., 2001). Furthermore, this relationship was

found for bespoke neighbourhoods defined at a variety of spatial scales: the greater

the intensity of local support for a particular party, the greater the polarization of

voting towards it (Johnston et al., 2001, 2005a, 2007); such polarization was also

much stronger among respondents, the higher their levels of local social capital and

interaction with their neighbours (Johnston et al., 2005b; see also Fone et al., 2006:

Walks, forthcoming, looks at political behaviour within Canadian gated communities,

locales where households have clearly selected to distance themselves from other

groups within society).

Social interaction is only one of the processes that can generate neighbourhood effect-

consistent patterns, however. Many voting decisions are based on people‟s

evaluations of government policy, especially economic policy. They tend to reward

governments that have delivered prosperity by voting for their return to power, but

punish them by voting for an opposition party (especially one that seems likely to

govern well) if they have not. Such calculi operate at a variety of scales: the

individual („Have I prospered over the last year?‟; „Do I think my income/quality of

living will improve over the next year?‟); the national („Has the national economy

improved recently?‟; „Will it during the immediate future?‟); and the regional/local

(„Have things improved locally recently?‟; „Will they continue to do so?‟: Pattie and

Johnston, 1995). And they refer to a range of government policies (Johnston and

Pattie, 2001a, 2001b). Context is important in these calculi too: studies using the

bespoke neighbourhood approach found that people economically optimistic about

their own financial situations were less likely to vote for the government if they lived

in relatively deprived areas than if they lived in places where their neighbours were

prospering too (Johnston et al., 2000). Context, it seems, stimulated altruistic

behaviour – implying that people were in contact with, and were concerned for, their

neighbours.

Another geographically-variable influence on voting decisions is party campaigning,

much of it spatially focused to ensure that a party‟s supporters turn out in those

10

constituencies where their participation is most needed – marginal seats that could be

won or lost depending on who abstains. Such campaigning has become increasingly

targeted through a range of advertising and other strategies aimed at contacting and

mobilising individual voters (and hoping that they will mobilize others through their

social networks) – with clear impacts, especially as a consequence of the intensity of

campaigns mounted by opposition candidates/parties (Pattie and Johnston, 2009). The

outcome is also likely to be a pattern of voting consistent with that generated by the

classic neighbourhood effect: it could mean that social interaction is an irrelevant

influence, although those contacted by a party may well transmit the message to their

neighbours. However, analyses incorporating both economic voting and party

campaign intensity into the bespoke neighbourhood approach have shown that all

three are complementary: parties perform better in areas where they have stimulated

prosperity, where they have campaigned most intensively, and where the local social

networks are favourably inclined towards them (Johnston et al., 2007). In sum,

therefore, a substantial body of research findings is entirely consistent with the

neighbourhood effect hypothesis, strongly implying that information flowing through

local social networks influences voting decisions – but that evidence is

overwhelmingly circumstantial only.

Conversion through conversation

Discussion between citizens lies at the heart of most theories of democracy. For

democracy to function there has to be scope for diversity of opinion, free expression

of those opinions, and resolution of differences and conflicts. Political conversations

provide one means for spreading salient information, opinion and argument, and can

enable individuals to determine their positions on the relevant issues and/or

personalities by testing their views against others‟. A celebrated two-step flow model

of political communication, for instance, argues that local „opinion leaders‟ pick up

political information from the media and in turn pass this on, often in an evaluated

form, to others in their communities with whom they are in contact (Katz and

Lazarsfeld, 1955).

Individuals‟ social networks might be dominated by people who largely share their

own views, prejudices and values, but they may encounter others with very different

attitudes. Conversations are likely to reinforce one‟s own beliefs in the former case,

whereas in the latter they may well cause individuals to question their opinions –

especially if they are not strongly committed to any position, candidate or party and

hold minority views within the conversation network. Other things being equal,

therefore, holders of minority views may change their minds and agree with the

network‟s majority. Evidence supports this argument; the more supporters of a

particular party individuals talk to, the more likely they are to switch their vote to that

party if they previously either voted for an alternative or abstained (Huckfeldt and

Sprague, 1995; Pattie and Johnston, 2000, 2001; Levine, 2005).

Involvement in relevant conversations could also encourage political participation

(Putnam, 2000), providing information on how to take part and indirect confirmation

that one‟s associates are likely to participate, hence enhancing the impact of social

norms – although those already likely to participate may, as a result of their

commitment, be more likely to discuss politics with others than those not intending to

participate. Studies confirm that those with extensive conversation networks

11

participate more than those with limited networks, particularly if conversations are

politically focused (Leighley, 1990; McClurg, 2003, 2006a, 2006b; Kotler-Berkowitz,

2005).

The content of conversations is likely to be at least as important as their prevalence. It

is one thing to be surrounded by individuals who all confirm the correctness of one‟s

own opinions (which some may seek to achieve by their choice of networks: Finifter‟s

(1974) study of American car plant workers demonstrated that individuals who held a

minority view – supporting the Republicans in a predominantly Democrat-voting

environment – not only were more likely to form friendships with like-minded people

at work, but were also less likely than their Democrat-supporting workmates to

discuss politics outside the workplace). It is potentially quite another to be faced with

widespread disagreement. Most encounter at least some disagreement within their

discussion networks, however, and few can entirely insulate themselves from

heterogeneous opinions; pressures towards homogeneity within networks

notwithstanding, disagreement is an endemic feature of conversation (Huckfeldt et al.,

2004). And, of course, some disagreement is essential for influence to occur (McPhee,

1963): where people agree entirely, they cannot persuade.

The extent of the impact of intra-network disagreement on participation has proved

controversial. Classic pluralist accounts of democracy suggest that where differences

of opinion exist, people and/or groups will be mobilized to represent the various

views expressed, thereby acting as mobilising forces themselves (Dahl, 1989). But

psychological models suggest that as most individuals are conflict-averse they will try

to avoid it, by either acquiescence or silence (Festinger, 1957; Ulbig and Funk, 1999).

Some have shown that countervailing opinions in discussion networks can discourage

participation, in part by increasing uncertainty (Mutz, 2002a, 2002b, 2006; Mutz and

Mondak, 2006; Pattie and Johnston, 2008; Campus et al., 2008). McClurg (2006a), on

the other hand, reports that disagreement‟s impact is modified by context, in particular

by whether an individual is part of a local political majority or minority. His results

suggest that political participation by individuals sharing their local context‟s majority

view is unaffected by exposure to disagreement but those in the local minority

become less likely to participate as their exposure to disagreement increases:

disagreement disincentivizes participation by the latter.

Individuals are more likely to participate if they feel their discussants are politically

sophisticated and less likely if they feel discussants lack expertise (McClurg, 2006b).

Furthermore, they are more likely to discuss politics with those they perceive to have

some political expertise and knowledge than with those whose knowledge they doubt

– irrespective of whether they share the partisan leanings of the person they see as

politically expert (Huckfeldt, 2001). Judgements regarding discussants‟ political

expertise are solidly based, drawing on knowledge of their education, political

understanding and partisanship: people who are likely to know more about politics are

perceived as indeed knowing more, and are hence more likely to be chosen as sources

of politically-relevant information (Huckfeldt, 2001; Huckfeldt et al., 2000).

Work on UK political conversation networks sheds further light on this, much of it

based on data from the 1992 and 2005 British Election Studies (BES), which asked

survey respondents for information on those they discussed politics with. In 1992,

respondents were asked to give information on three such individuals (including their

12

relationship to the respondent, their employment, the extent to which the respondent

generally agreed with them, and the respondent‟s view of their partisanship). In 2005,

respondents were asked the extent to which they discussed politics separately with

their partners, family members, friends, neighbours, and workmates: they were also

asked to assess – very roughly – the proportion of their political discussants in each

group with whom they shared the same party affiliation. While the earlier survey

gives insights into particular discussion partnerships, therefore, the latter gives a less

detailed but wider idea of the extent of individuals‟ discussion networks, and the

extent of political agreement within these networks.

These respondents were influenced by the views of those they discussed politics with.

Among those who did not vote for the Conservatives at the 1987 general election, for

example, 54 per cent of those who changed their mind and voted for the party in 1992

named a Conservative supporter as their most important discussant, compared to only

12 per cent of those who did not vote Conservative at either election (Pattie and

Johnston, 1999: 882). Similar effects hold for conversations with supporters of the

other major parties – individuals who changed their vote in 1992 were much more

likely to have talked to a supporter of that party than were individuals whose vote did

not change. The corollary also held: talking to a party‟s supporter decreased the

likelihood of switching one‟s vote to that party‟s rivals. Thus of those who did not

vote Conservative in 1987, just under 9 per cent of those who did vote for the party in

1992 named a Labour supporter as their main discussant, and 6 per cent named a

Liberal Democrat: the comparative figures for those who did not vote Conservative in

either election were 32 and 12 per cent respectively. (The lower figure for discussions

with Liberal Democrat supporters reflects their third party status – there were fewer

Liberal Democrat partisans to encounter than either Conservative or Labour partisans,

and hence fewer opportunities for conversation with them.) Not surprisingly, the more

supporters of a particular party people talked to, the more likely they were to vote for

that party in 1992 if they had not done so at the previous election (Pattie and

Johnston, 1999: 885). Discussions with spouses and family members were particularly

influential (see also Stoker and Jennings, 2005; Zuckerman et al., 2005, 2007; Verba

et al., 2005), but conversations with non-relatives had an impact too, and respondents

were more influenced by the partisan leanings of discussants they generally agreed

with on most matters than by discussions with those they felt they generally disagreed

with: people were more likely to listen to those whose views they felt generally made

sense than with those whose opinions they routinely discounted (Pattie and Johnston,

2002). The influence of conversation remained even after controls were introduced for

factors such as class, age, gender, and economic evaluations (Pattie and Johnston,

2000). And analyses of inter-election change between 1992 and 1997 using panel data

(hence tracking the same individuals over time) yielded similar results (Pattie and

Johnston, 2001).

The impact of political conversations among British voters is not limited to vote

choice but also influences underlying political attitudes. Using BES panel data for the

1992-1997 inter-election period, Pattie and Johnston (2001: 35) show that the more

Conservatives respondents reported talking to, the greater the probability that their

attitudes would become more right-wing over time. Conversely, the more Labour

supporters they discussed politics with, the more left-wing (and the more libertarian

rather than authoritarian) their views became.

13

Recent interest in theories of deliberative democracy is driven in part by notions of

the wider beneficial effects of political discussion (e.g. Fearon, 1998; Fishkin, 1995).

Interest here focuses not only on „conversion by conversation‟ effects but also on the

putative impact of discussion on individuals‟ abilities to accept opponents‟ views as

legitimate and reasoned – even if not ultimately convincing. To the extent that

discussion exposes individuals to opposing views, and forces both discussants to

explain the reasons for their opinions, then it should (according to the theory) make

them more understanding of views they do not hold. Deliberative democracy theorists

therefore see political discussion as central to the development of habits of tolerance.

While the theory itself is notoriously difficult to pin down empirically, some progress

can be made via middle-level theories and analyses (Mutz, 2006, 2008).

Much of the empirical research on the topic reinforces this claim (Conover et al.,

2002; Mutz, 2006; Mutz & Mondak, 2006; Pattie and Johnston, 2008; Searing et al.,

2007). The more people discuss politics, and the more political disagreement

individuals encounter in their discussion networks, the more readily are they able to

understand why others hold views which diverge from their own (Mutz, 2006; Mutz

and Mondak, 2006). The more they discuss, and the more they encounter divergent

views in those discussions, the more tolerant they become, both of others‟ political

opinions (and their right to hold those views) and of different lifestyle choices (Mutz,

2006; Pattie and Johnston, 2008). Discussion in general and encountering opposing

views in particular also encourages individuals to clarify their own views –

presumably the better to defend them in discussion. Responses to attitude questions in

the 1992 and 2005 BES show that individuals who report relatively high levels of

political discussion and of disagreement in those networks give fewer „don‟t know‟

responses to opinion questions, and more clear opinions, than those who report little

or no disagreement or little political discussion of any sort (Pattie and Johnston,

2008). Furthermore, those who talk politics extensively and those who encounter

much disagreement also report higher levels of perceived political efficacy than those

whose discussion networks are limited, consensual or apolitical: those encountering

disagreement are more confident of their political abilities than are those who do not.

One possible objection to such an argument might be that causation runs from

attitudes of tolerance and feelings of efficacy to reports of discussion and

disagreement, not vice-versa. Those who are most confident of their own views and

political efficacy, and most tolerant of others, therefore, would be most likely to seek

out political conversations and least concerned by the risks of encountering

countervailing views – and hence most likely to report disagreement. To some extent,

this is undoubtedly the case. However, recourse once again to 1992-97 BES panel

data shows that, even when we take pre-existing levels of tolerance and personal

efficacy into account, the greater the degree of political discussion and general

disagreement in particular individuals report, then over time the more tolerant they

become, and the more politically efficacious they feel (Pattie and Johnston, 2008:

695).

Active political discussion in social networks not only influences vote choice,

therefore, but also has wider beneficial effects on civic attitudes, especially where this

exposes individuals to views counter to their own. But political disagreement in social

networks is not necessarily an unalloyed good; encountering disagreement may

demotivate people from taking part in politics, even as it also increases their levels of

14

tolerance (see especially Mutz, 2002a, 2002b, 2006). Being better able to understand

others‟ views may lead to confusion over, or falling confidence in, one‟s own

opinions, for instance. Or individuals encountering active disagreement may come to

feel theirs is a minority view, and hence see little incentive for political participation:

if I am consigned to a permanent minority, what point is there in my taking part? To

the extent that this is the case, therefore, political disagreement within social networks

might be counterproductive for democracy: a citizen body of tolerant, understanding

people may be pleasant to live in, but of limited political use if no-one participates as

a result.

Much of the empirical research demonstrating the demotivating effects of

disagreement in discussion networks has focused on the USA, however, and on one

form of political participation there, voting. Looking at different political contexts and

different forms of political activism suggests a more complex, and less bleak, picture.

McLurg‟s (2006a) work on the importance of political context found that people who

hold the local majority view are just as likely to participate in politics whether or not

they encounter disagreement: it is those who hold a locally minority opinion whose

chances of participation fall if they encounter divergent views. But recent work using

British data suggests that the sort of political activism envisaged may also be

important. Voting is not the only way of getting involved politically and the British

Citizen Audit survey of 2000 reported three quite distinct styles of political

engagement (Pattie et al., 2004): individual activity, such as voting, petition-signing,

and ethical consumption (actions which a citizen can engage in alone and which do

not require the co-operation of others); contact activism (contacting those in power –

politicians, the media, etc. – to achieve some end); and collective activism (forms of

political engagement which require active co-operation between like-minded people –

strikes, various forms of public protest, and so on). The 2005 BES data suggest that it

is important to take into account what form political activism will take (Pattie and

Johnston, forthcoming). Other things being equal, British voters with large political

discussion networks were more likely to turn out at the 2005 general election than

were those with small networks. But the extent of disagreement encountered within

those networks had a negative – albeit only weak– effect on turnout: once network

size and a wide range of other factors were taken into account, those who reported a

relatively high density of disagreement within their discussion networks were less

likely to vote than those whose networks were more politically homogeneous. That

said, the negative effect was very small (although statistically significant), and was

comfortably overshadowed by the positive impact of larger discussion networks. But

for other forms of political activism, the picture was somewhat different. Individuals‟

probabilities of volunteering to take part in political or community campaigns actually

increased the greater the level of disagreement they encountered in their discussion

networks (although the effect was again small and swamped by the much larger and

positive impact of larger discussion networks per se). Similar positive effects emerged

for various measures of potential future activism – including voting, community

involvement and participation in party politics. It is premature to argue that

encountering conflicting views discourages political involvement, therefore: it may

actually encourage it.

Putting it all together

15

What much of the literature reviewed above lacks is an integration of the two main

approaches to studying neighbourhood effects – information flow through social

networks, and spatially-polarized aggregate voting and other behavioural patterns

(reflecting similarly polarized norms). Its conclusions provide strong circumstantial

support for the argument that locally-focused social interaction influences people‟s

voting decisions in ways that are very likely to generate polarized patterns, but the

evidence is not conclusive.

A few have sought to remedy this – largely through very small-scale studies of

selected locales (e.g. Fitton, 1973) or deploying sample data in which contact with

neighbours was surveyed (Curtice, 1995). By far the most important studies of social

networks in their spatial settings, however, have been the large sample surveys

conducted by Huckfeldt and his collaborators. The original, seminal work (Huckfeldt

and Sprague, 1995) used a survey of the residents of South Bend, IN, who were

interviewed on three occasions before and after the 1984 US presidential election and

provided information about those with whom they discussed political issues

(subdivided into spouses, other kin, and non-kin: these discussion partners were also

interviewed). These and other data on the individual were integrated with data on the

neighbourhoods in which they lived; one-third of the nominated discussants lived in

the respondents‟ home neighbourhoods and 40 per cent worked at the same location.

(Only 6 per cent were both workmates and neighbourhood co-residents.) The impact

of discussants‟ political choices was greatest when the respondent‟s nominated main

discussant was her/his spouse; outwith the immediate household, the main

discussant‟s influence was greatest when the respondent correctly identified that

person‟s own political preferences (in 1984, whether or not he/she supported Reagan

or Mondale).

Huckfeldt and Sprague‟s (1995: 189) conclusion that „vote preferences are socially

structured, not only by the characteristics of the voter, but also by the characteristics

and preferences of others with whom the voter discusses politics‟ was extended by

incorporating further variables to represent not only the political characteristics of

their respondents‟ neighbourhood socio-political milieux but also the apparent

intensity of the election campaign there. They found that partisan contact with voters

in a neighbourhood influenced not only those contacted but also others in the locality,

with the initial contact thus acting as „a catalyst that sets into motion a series of

events‟ (p. 255) because „people know their neighbour‟s politics, and one reason they

know is party organization aimed at informing them‟ (p. 254). Thus the proportion of

a neighbourhood‟s respondents who supported a particular party (the sample was

spatially clustered to allow this to be estimated) was strongly related to whether an

individual living there also identified with that party, whatever her/his individual

characteristics. Not everybody is converted to the local majority view, of course,

especially if disagreement is inconsistent with the general tenor of opinion within an

individual‟s network (Huckfeldt et al., 2002). And reaction to the local milieu

depends upon the nature of that awareness: Baybeck and McClurg (2005) found that a

substantial majority of the South Bend respondents could accurately represent the

characteristics of their home neighbourhood – and then, as they put it, „When a

neighborhood‟s majority becomes obvious, even opposing voters seem capable of

figuring that out‟ (p. 509).

16

Although the early Huckfeldt studies integrated social networks and neighbourhood

contexts much more firmly than almost all other analysts had, nevertheless the

network geographies were to a considerable extent inferred. Later work in two cities

took the work further, by using post-coded information on the respondents‟ and their

main discussants‟ homes to establish the degree of network spatial dispersion. As

anticipated from other work on the geography of social interaction reviewed earlier,

the networks were not intensely localized: for kin (excluding spouses) only 23 per

cent lived within 1km. of the respondent‟s home and the average distance between the

two locations was 6.4km.; for non-kin, the percentage was only 15 and the mean

distance 8.4km. – nevertheless over half of this group lived within 15 minutes‟

driving time of the respondent‟s home (Baybeck and Huckfeldt, 2002a). The more

dispersed the network, however, the less intense the discussions taking place within it.

Information is spread more widely through the more dispersed networks, across a

wider range of neighbourhoods contexts – so that even though they do not necessarily

connect individuals who are socially and politically more diverse than is the case with

the spatially more clustered networks, they act as the „bridges between socially and

politically diverse locales‟ (p.273). This is what Granovetter (1973) terms the „weak

ties‟ that introduce (perhaps dissonant) information to otherwise separate networks

and locales, even though two individuals so connected are less likely to converge in

their opinions over an election campaign than are two similar individuals who are

members of spatially higher density networks, among whom contact is also more

frequent (Baybeck and Huckfeldt, 2002b). Spatially dispersed networks, it seems,

create a politically homogeneous overlay on a politically diverse urban area – a

conclusion that is now being explored with experimental data (e.g. Ahn, Huckfeldt

and Ryan, 2007).

IN CONCLUSION

Social networks are key to many of the myriad flows within society that diffuse

information and knowledge, processes that in turn influence people‟s behaviour and

actions. A great deal of evidence – much of it reviewed elsewhere in this volume –

has been generated to sustain this argument and show that who you talk to can be very

influential on what you learn and what you do.

One aspect of this argument that has received somewhat less attention than others is

its spatiality, or the geography of social networks. It is generally assumed that

proximity/propinquity are fundamental to network structures and operations – we are

more likely to know near than distant neighbours, more likely to interact with those

who live close to our homes and whom we may encounter in a range of structured as

well as unstructured arenas. If this is so, then the spatial concentration of social

networks can have substantial implications for the geography of attitudes and

behaviour; those who live in relative close proximity are more likely to think and act

in similar ways because of the spatial selectivity of information flows – a spatial

polarization often referred to as a neighbourhood effect.

This argument – as illustrated here – is frequently tested with circumstantial evidence

only; if attitude and behavioural patterns are spatially concentrated in the predicted

fashion this can be taken as support for the neighbourhood effect hypothesis. And

many social networks are spatially concentrated, as we have demonstrated using

examples taken from electoral studies. Further circumstantial evidence is provided by

17

studies of the processes rather than the patterns, of the flows of information through

networks and their impacts on behaviour – but often without any close attention to the

networks‟ geography: conversation leads to conversion (in the electoral context,

„those who talk together vote together‟, whether within the household or beyond:

Johnston et al., 2005; Pattie and Johnston, 2000) irrespective of the relative locations

of those involved. But few studies link all the parts together – the geography of social

networks, the information flowing through them, and the impacts this has on people‟s

attitudes and behaviour. Those that do – such as those by Huckfeldt and his

collaborators – sustain the entire thesis; networks are spatially biased, information

flowing through them influences behaviour – especially information favouring the

views of the network majority – and as a consequence neighbourhood effects are

generated.

We have largely illustrated this argument using examples taken from one area of

social science inquiry only – voting studies. But similar cases have been presented,

evaluated empirically and found valid in a range of other subject areas. In education,

for example, there is substantial evidence to show that parental and student attitudes

to participation and attainment are in part linked to predominant attitudes in their local

communities – with consequences for the geography of educational performance (see,

for example, Kohen et al., 2008; McCulloch, 2006; Sampson et al., 2002). The same

applies to a wide range of health-related behaviours – smoking, for example, and

eating unhealthy foods – with consequences for morbidity and life chances (e.g.

McCulloch, 2003). There are geographies of criminal behaviour too which are

strongly supportive of the neighbourhood effect hypothesis (e.g. Oberwittler, 2007),

as also with geographies of unemployment and job-seeking. Where you are influences

who you interact with; who you interact with influences what you learn and how you

interpret the information and knowledge gained; and such local sources of „valued‟

information influence how you behave – all of which takes place in spatially-defined

contexts, in linked geographies of inputs, processes and outputs.

REFERENCES

Ahn, T. K., Huckfeldt, R. and Ryan, J. B. (2007) „Networks, groups, and contextual

constraints on political communication‟. Available at

http://faculty.psdomain.ucdavis. edu/rhuckfeldt/files/apsa_paper_final.pdf

Andersen, R. and Heath, A. F. (2002) „Class matters: the persisting effects of

contextual social class on individual voting in Britain, 1964-97‟, European

Sociological Review, 18: 125-38.

Atkinson, R. and Blandy, S., editors (2006) Gated Communities. London: Routledge.

Baybeck, B. and Huckfeldt, R. (2002a) „Spatially dispersed ties among interdependent

citizens; connecting individuals and aggregates‟, Political Analysis, 10: 261-

75.

Baybeck, B. and Huckfeldt, R. (2002b) „Urban contexts, spatially dispersed networks

and the diffusion of political information‟, Political Geography, 21: 195-210.

18

Baybeck, B. and McClurg, S. D. (2005) „What do they know and how do they know

it? An examination of citizen awareness of context‟, American Politics

Research, 33: 492-520.

Bishop, B. (2008) The Big Sort: Why the Clustering of Like-Minded Americans is

Tearing us Apart. Boston: Houghton Mifflin.

Blasius, J., Friedrichs, J., and Galster, G. (2007) „Frontiers of quantifying

neighbourhood effects‟, Housing Studies, 22: 627-36.

Cain, B. E. (2009) Sorting it out. The California Journal of Politics and Policy, 1,

Article 8.

Campus, D., Pasquino,G. and Vaccari, C. (2008) „Social networks, political

discussion, and voting in Italy: a study of the 2006 election‟, Political

Communication, 25, 423-44.

Carrasco, J.-A., Miller, E. J. and Wellman, B. (2008a) „How far and with whom do

people socialize? Empirical evidence about the distance between social

network members‟, Transportation Research Record: Journal of the

Transportation Research Board, 2076: 114-22.

Carrasco, J.-A., Hogan, B., Wellman, B. and Miller, E. J. (2008b) „Agency in social

activity interactions: the role of social networks in time and space‟, Tijdschrift

voor Economische en Sociale Geografie, 99: 562-83.

Cho, W. K. T. and Rudolph, T. J. (2008) „Emanating political participation:

untangling the spatial structure behind participation‟, British Journal of

Political Science, 38: 273-89.

Conover, P. J., Searing, D. and Crewe, I. (2002). „The deliberative potential of

political discussion‟, British Journal of Political Science. 32: 21-62.

Cox, K. R. (1968) „Suburbia and voting behaviour in the London metropolitan area‟,

Annals of the Association of American Geographers, 58: 111-127.

Cox, K. R. (1969) „The voting decision in spatial context‟, Progress in Geography, 1:

83-117.

Curtice, J. (1995) „Is talking over the garden fence of political import?‟, in M. Eagles

(ed.), Spatial and Contextual Models in Political Research. London: Taylor

and Francis, pp. 195-209.

Curtice, J. and Steed, M. (1982) „Electoral choice and the production of government:

the changing operation of the electoral system in the United Kingdom since

1955‟, British Journal of Political Science, 12: 249-98.

Dahl, R. (1989) Democracy and its Critics. New Haven: Yale University Press.

19

Denver, D. T. and Halfacree, K. (1992) „Inter-constituency migration and turnout at

the British general election of 1983‟, British Journal of Political Science, 22:

248-54.

Doreian, P. (2001) „Causality in social network analysis‟, Sociological Methods and

Research, 30, 81-114.

Dunbar, R. I. M. (2008) „Mind the gap, or why humans aren‟t just great apes‟,

Proceedings of the British Academy, 154: 403-23.

Eagles, M., Bélanger, P and Calkins, H W. (2004) „The spatial structure of urban

political discussion networks‟, in M. F. Goodchild and D. G. Janelle (eds.),

Spatially Integrated Social Science. Oxford: Oxford University Press, pp. 205-

18.

Fearon, J. (1998). „Deliberation as discussion‟, in J. Elster (ed.) Deliberative

Democracy. Cambridge: Cambridge University Press, pp. 44-68.

Festinger, L. (1957). A Theory of Cognitive Dissonance. Palo Alto: Stanford

University Press.

Festinger, L., Schachter, S. and Back, K. (1950) Social Pressure in Informal Groups.

New York: Harper & Row.

Finifter, A. (1974). „The friendship group as a protective environment for political

deviants‟, American Political Science Review, 68: 607-625.

Fishkin, J. (1995). The Voice of the People: Public Opinion and Democracy. New

Haven: Yale University Press.

Fitton, M. (1973) „Neighbourhood and voting: a sociometric explanation‟, British

Journal of Political Science, 3: 445-472.

Fone, D. L., Farewell, D. M. and Dunstand F. D. (2006) „An ecometric analysis of

neighbourhood cohesion‟, Population Health Metrics, 4: 1-17.

Gans, H. J. (1962) The Urban Villagers. New York: The Free Press.

Granovetter, M. (1973) „The strength of weak ties‟, American Journal of Sociology,

78: 1360-1380.

Gimpel, J.G. and Lay, J.C. (2005) „Party identification, local partisan contexts, and

the acquisition of participatory attitudes‟, in Zuckerman, A.S. (ed.), The Social

Logic of Politics: Personal Networks as Contexts for Political Behavior,

Philadelphia: Temple University Press, 209-227.

Hägerstrand, T. (1976) „Geography and the study of interaction between society and

nature‟, Geoforum, 7: 329-34.

20

Hägerstrand, T. (1982) „Diorama, path and project‟, Tijdschrift voor Economische en

Sociale Geografie, 73: 323-39.

Hägerstrand, T. (1984) „Presence and absence: a look at conceptual choices and

bodily necessities‟, Regional Studies, 18: 373-80.

Hampton, K. and Wellman, B. (2001) „Long distance community in the network

society: contact and support beyond Netville‟, American Behavioral Scientist,

45: 476-95.

Harrop, M., Heath, A. and Openshaw, S. (1992) „Does neighbourhood influence

voting behaviour – and why?‟, in I. Crewe, P. Norris, D. Denver and D.

Broughton (eds.) British Elections and Parties Yearbook 1. Hemel

Hempstead: Harvester Wheatsheaf, pp. 103-20.

Huckfeldt, R. R. (1988) „Social contexts, social networks, and urban neighborhoods;

environmental constraints on friendship choice‟, American Journal of

Sociology, 89: 651-69.

Huckfeldt, R. R. (1996) Politics in Context: Assimilation and Conflict in Urban

Neighborhoods. New York: Agathon Press.

Huckfeldt, R. (2001) „The social communication of political expertise‟, American

Journal of Political Science, 45: 425-38.

Huckfledt, R., Ikeda, K. and Pappi, F. U. (2000). „Political expertise, interdependent

citizens and the value added problem in democratic politics‟, Japanese

Journal of Political Science, 1: 171-95.

Huckfeldt, R., Johnson, P. E. and Sprague, J. (2002) „Political environments, political

dynamics, and the survival of disagreement‟, Journal of Politics, 64: 1-21.

Huckfeldt, R., Johnson, P. E. and Sprague, J. (2004) Political Disagreement: the

Survival of Diverse Opinions within Communication Networks. Cambridge:

Cambridge University Press.

Huckfeldt, R. and Sprague, J. (1995) Citizens, Politics and Social Communication:

Information and Influence in an Election Campaign. Cambridge: Cambridge

University Press.

Johnston, R. J. (1974) Social distance, proximity and social contact: eleven culs-de-

sac in Christchurch, New Zealand. Geografiska Annaler, 56B, 57-67.

Johnston, R. J. (2003) „Order in space: geography as a discipline in distance‟, in R. J.

Johnston and M. Williams (eds.) A Century of British Geography. Oxford:

Oxford University Press for the British Academy, pp. 303-46.

Johnston, R. J., Dorling, D. F. L., Tunstall, H., Rossiter, D. J., McAllister, I. and

Pattie, C. J. (2000) „Locating the altruistic voter: context, egocentric voting

21

and support for the Conservative party at the 1997 general election in England

and Wales‟, Environment and Planning A, 32: 673-94.

Johnston, R. J., Jones, K., Propper, C., Sarker, R., Burgess, S. and Bolster, A. (2005)

„A missing level in the analysis of British voting behaviour: the household as context as shown by analyses of a 1992-1997 longitudinal survey‟, Electoral Studies, 24: 201-25.

Johnston, R. J., Jones, K., Propper, C. and Burgess, S. (2007) „Region, local context,

and voting at the 1997 general election in England‟, American Journal of Political Science, 51: 640-54.

Johnston, R. J. and Pattie, C. J. (2001a) „Dimensions of retrospective voting:

economic performance, public service standards and Conservative party

support at the 1997 British general election‟, Party Politics, 5: 39-54.

Johnston, R. J. and Pattie, C. J. (2001b) „„It‟s the economy, stupid‟ – but which

economy? Geographical scales, retrospective economic evaluations and voting

at the British 1997 general election‟, Regional Studies, 35: 309-19.

Johnston, R. J. and Pattie, C. J. (2006) Putting Voters in their Place: Geography and

Elections in Great Britain. Oxford: Oxford University Press.

Johnston, R. J., Pattie, C. J., Dorling, D. F. L., McAllister, I., Tunstall, H. and

Rossiter, D. J. (2001) „Housing tenure, local context, scale and voting in

England and Wales, 1997‟, Electoral Studies, 20: 195-216.

Johnston, R. J., Pattie, C. J. and Allsopp, J. G. (1988) A Nation Dividing? The

Electoral Map of Great Britain 1979-1987. London: Longman.

Johnston, R. J., Propper, C., Burgess, S., Sarker, R., Bolster, A. and Jones, K. (2005a)

„Spatial scale and the neighbourhood effect: multinomial models of voting at

two recent British general elections‟, British Journal of Political Science, 35:

487-514.

Johnston, R. J., Propper, C., Jones,K., Sarker, R., Bolster, A. and Burgess, S. (2005b)

„Neighbourhood social capital and neighbourhood effects‟, Environment and

Planning A, 37: 1443-57.

Katz, E. and Lazarsfeld, P. F. (1955). Personal Influence: the Part Played by People

in the Flow of Mass Communications. Glencoe, IL: Free Press.

Kohen, D. E., Leventhal, T., Dahinten, V, S. and McIntosh, C. N. (2008)

„Neighborhood disadvantage: pathways for young children‟, Child

Development, 79: 156-69.

Kotler-Berkowitz, L. (2005) „ Friends and politics: linking diverse friendship

networks to political participation‟, in Zuckerman, A.S. (ed.), The Social Logic

of Politics: Personal Networks as Contexts for Political Behavior,

Philadelphia: Temple University Press, pp. 152-70.

22

Leighley, J. E. (1990) „Social interaction and contextual influences on political

participation‟, American Politics Quarterly, 18: 459-75.

Levine, J. (2005) „Choosing alone? The social network basis of modern political

choice‟, in Zuckerman, A.S. (ed.), The Social Logic of Politics: Personal

Networks as Contexts for Political Behavior, Philadelphia: Temple University

Press, pp. 132-52.

McAllister, I., Johnston, R. J., Pattie, C. J., Tunstall, H., Dorling, D. F. L. and D. J.

Rossiter (2001) „Class dealignment and the neighbourhood effect: Miller

revisited‟, British Journal of Political Science, 31: 41-59.

McClurg, S. D. (2003). „Social networks and political participation: the role of social

interaction in explaining political participation‟, Political Research Quarterly

56: 449-64.

McClurg, S. D. (2006a) „Political disagreement in context: the conditional effect of

neighborhood context, disagreement and political talk on electoral

participation‟, Political Behavior, 28: 349-66.

McClurg, S. D. (2006b) „The electoral relevance of political talk: examining

disagreement and expertise effects in social networks on political

participation‟, American Journal of Political Science, 50: 737-54.

McCulloch, A. (2003) „An examination of social capital and social disorganisation in

neighbourhoods in the British Household Panel Study‟, Social Science and

Medicine, 56: 1425-38.

McCulloch, A. (2006) „Variation in children‟s cognitive and behavioural adjustment

between different types of place in the British National Child Development

Study‟, Social Science and Medicine, 62: 1865-79.

McMahon, D., Heath, A. F., Harrop, M. and Curtice, J. (1992) „The electoral

consequences of north-south migration‟, British Journal of Political Science,

22: 419-443.

McPhee, W. N. (1963) Formal Theories of Mass Behavior. New York: The Free Press

of Glencoe.

Mark, N. (1998) „Birds of a feather sing together‟, Social Forces, 77: 453-485.

Michelson, W. (1970) Man and his Urban Environment. Reading PA: Addison-

Wesley.

Miller, W. L. (1977) Electoral Dynamics in Britain since 1918. London: Macmillan.

Miller, W. L. (1978) „Social class and party choice in England: a new analysis‟,

British Journal of Political Science, 8: 259-284.

23

Mok, D., Wellman, B. and Basu, R. (2007) „Did distance matter before the internet?

Interpersonal contact and support in the 1970s‟, Social Networks, 29: 430-61.

Mok, D., Wellman, B. and Carrasco, J.-A. (2009) „Does distance matter in the age of

the internet?‟, Urban Studies, 46:

Mutz, D. (2002a) „The consequences of cross-cutting networks for political

participation‟, American Journal of Political Science, 46: 838-55.

Mutz, D. (2002b) „Cross-cutting social networks: testing democratic theory in

practice‟, American Political Science Review, 96: 111-26.

Mutz, D. (2006). Hearing the Other Side: Deliberative versus Participatory

Democracy, Cambridge: Cambridge University Press.

Mutz. D. (2008) „Is deliberative democracy a falsifiable theory?‟, Annual Review of

Political Science, 11: 521-38.

Mutz, D. and Mondak, J. (2006) „The workplace as a context for cross-cutting

political discourse‟, The Journal of Politics, 68: 140-55.

Oberwittler, D. (2007) „The effects of neighbourhood poverty on adolescent problem

behaviours: a multi-level analysis differentiated by gender and ethnicity‟,

Housing Studies, 22: 781-803.

Pahl, R. E. (1965) Urbs in Rure. London: London School of Economics, Department

of Geography, Occasional Paper 2.

Pattie, C. J. and Johnston, R. J. (1995) „It‟s not like that round here: region, economic

evaluations and voting at the 1992 British General Election‟, European

Journal of Political Research, 28: 1-32.

Pattie, C. J. and Johnston, R. J. (1999) „Context, conversation and conviction: social

networks and voting at the 1992 British General Election‟, Political Studies.

47: 877-889.

Pattie, C. J. and Johnston, R. J. (2000) „People who talk together vote together: an

exploration of contextual effects in Great Britain‟, Annals of the Association of

American Geographers, 90: 41-66.

Pattie, C. J. and Johnston, R. J. (2001) „Talk as a political context: conversation and

electoral change in British elections‟, Electoral Studies, 20: 17-40.

Pattie, C. J. and Johnston, R. J. (2002) „Political talk and voting: does it matter to

whom one talks?‟, Environment and Planning A, 34: 1113-35.

Pattie, C. J. and Johnston, R. J. (2008) „It‟s good to talk: talk, disagreement and

tolerance‟, British Journal of Political Science, 38: 677-98.

24

Pattie, C. J. and Johnston, R. J. (2009) „„Still talking but is anybody listening?‟ The

changing face of constituency campaigning in Britain, 1997-2005‟, Party

Politics, 14:

Pattie, C. J. and Johnston, R. J. (forthcoming) „Conversation, disagreement and

political participation‟, Political Behavior,

Pattie, C. J., Seyd, P. and Whiteley, P (2004). Citizenship in Britain: Values,

Participation and Democracy. Cambridge: Cambridge University Press.

Piselli, F. (2007) „Communities, places, and social networks‟, American Behavioral

Scientist, 50: 867-78.

Potterat, J. J. et al. (2002) „Sexual network structure as an indicator of epidemic

phase‟, Sexually Transmitted Infections, 78: 152-58.

Rothenberg, R. (2007) „Maintenance of endemicity in urban environments: a

hypothesis linking risk, network structure and geography‟, Sexually

Transmitted Infections, 83: 10-15.

Rothenberg, R. et al. (2005) „Social and geographic distance in HIV risk‟, Sexually

Transmitted Disease, 32: 506-12.

Sampson, R. J., Morenoff, J. D. and Gannon-Rowley, D. (2002) „Assessing

“neighbourhood effects”: social processes and new directions in research‟,

Annual Review of Sociology, 28: 443-78.

Searing, D. D., Solt, F. Conover, P. J. and Crewe, I. (2007) „Public discussion in the

deliberative system: does it make better citizens?‟, British Journal of Political

Science, 37: 587-618.

Stoker, L. and Jennings, M. K. (2005) „Political similarity and influence between

husbands and wives‟, in Zuckerman, A.S. (ed.), The Social Logic of Politics:

Personal Networks as Contexts for Political Behavior, Philadelphia: Temple

University Press, pp. 51-74.

Stutz, F. P. (1973) „Distance and network effects on urban social travel fields‟,

Economic Geography, 49: 134-44.

Tilly, C. (1982) To Dwell among Friends: Personal Networks in Town and City.

Chicago: University of Chicago Press.

Tönnies, F. (1887) Gemeinschaft und Gesellschaft. Leipzig: Fue‟s Verlag. (Translated

by C. P. Loomis as Community and Society. New York: Harper and Row,

1940.)

Ulbig, S. G. and Funk, C. L. (1999) „Conflict avoidance and political participation‟,

Political Behavior, 21: 265-82.

25

Verba, S., Schlozman, K.L. and Burns, N. (2005) „Family ties: understanding the

intergenerational transmission of political participation‟, in Zuckerman, A.S.

(ed.), The Social Logic of Politics: Personal Networks as Contexts for

Political Behavior, Philadelphia: Temple University Press, pp. 95-114.

Walks, R. A. (2004) „Place of residence, party preferences, and political attitudes in

Canadian cities and suburbs‟, Journal of Urban Affairs, 26: 269-95.

Walks, R. A. (2005) „City-suburban electoral polarization in Great Britain, 1950-

2001‟, Transactions of the Institute of British Geographers, NS30: 500-17.

Walks, R. A. (2006) „The causes of city-suburban political polarization? A Canadian

case study‟, Annals of the Association of American Geographers, 96: 390-414.

Walks, R. A. (forthcoming) Electoral behaviour behind the gates: partisanship and

political participation among Canadian gated community residents.

Watson, J. W. (1955) „Geography: a discipline in distance‟, Scottish Geographical

Magazine, 71: 1-13.

Wellman, B. (1979) „The community question: the intimate networks of East

Yorkers‟, American Journal of Sociology, 84: 1201-1231.

Wellman, B. (1996) „Are personal communities local? A Dumptarian

reconsideration?‟, Social Networks, 18: 347-354.

Wellman, B. (1999) „From little boxes to loosely-bound networks: the privatization

and domestication of community‟, in J. Abu-Lughod (ed.) Sociology for the

Twenty-First Century: Continuities and Cutting-Edges. Chicago: University of

Chicago Press, pp. 94-114.

Wellman, B. (2001) The Persistence and Transformation of Community: from

Neighbourhood Groups to Social Networks. Report to the Law Commission of

Canada. Toronto: Wellman Associates.

Wellman, B., Carrington, P. and Hall, A. (1988) „Networks as personal communities,‟

in B. Wellman and S. D. Berkowitz (eds.) Social Structures: a Network

Approach. Cambridge: Cambridge University Press, pp. 130-184.

Wellman, B. and Potter, S. (1999) „The elements of personal communities‟, in B.

Wellman (ed.) Networks in the Global Village. Boulder CO: Westview Press,

pp. 49-82.

Wheeler, J. O. and Stutz, F. P. (1971) „Spatial dimensions of urban social travel‟,

Annals of the Association of American Geographers, 61: 371-86.

Wright, G. C. (1977) „Contextual models of electoral behaviour: the southern Wallace

vote‟, American Political Science Review, 71: 497-508.

26

Zelinsky, W. and Lee, B. A. (1998) „Heterolocalism: an alternative model of the

sociospatial behaviour of immigrant ethnic communities‟, International

Journal of Population Geography, 4: 281-98.

Zuckerman, A. S., Dasović, J. and Fitzgerald, J. (2007) Partisan Families: the Social

Logic of Bounded Partisanship in Germany and Britain. Cambridge:

Cambridge University Press.

Zuckerman, A. S., Fitzgerald, J. and Dasović, J. (2005) „Do couples support the same

political parties? Sometimes: evidence from British and German household

panel surveys‟, in Zuckerman, A.S. (ed.), The Social Logic of Politics:

Personal Networks as Contexts for Political Behavior, Philadelphia: Temple

University Press, pp. 75-94.