“A Bad Neighbour Is as Great a Plague as a Good One Is a Great Blessing”: On Negative Relationships between Neighbours

Abstract

With research on social relations hitherto, it is not clear how and why negative relationships between neighbours emerge. In this study, arguments are developed on the conditions within neighbourhoods and on individual characteristics that facilitate negative relations amongst neighbours. The arguments are divided according to three perspectives: diversity, uncertainty and social influence. In the Dutch context, most support is found for the social influence perspective, and both the neighbourhood and the individual level seem important in explaining negative relationships. Important factors that explain the likelihood for negative relationships are the willingness of residents to intervene on behalf of the neighbourhood, religious diversity and individually perceived conflicts in the neighbourhood. However, people who have more relationships outside the neighbourhood, undergo less influence of perceived conflict.

A bad neighbour is as great a plague as a good one is a great blessing; he who enjoys a good neighbour has a precious possession. … Take fair measure from your neighbour and pay him back fairly with the same measure, or better, if you can; so that if you are in need afterwards, you may find him sure (Hesiodos (1970), from Works and Days, around 700 BC).

Introduction

The neighbourhood is a place where people usually interact with each other on a daily basis, ranging from greeting each other in the street to organising block parties together. Many people and many sociologists as well, however, believe that there has been a decline of neighbourhood interaction, which has been under investigation as an important part of the ‘loss of community’ question (for example, Campbell and Lee, 1992; Guest and Wierzbicki, 1999; Putnam, 2000; Wellman, 1979, 1996). Nevertheless, the literature on neighbourhood relations—and relations in general—is positively biased and focuses on how relations can help to achieve important goals in life. Everybody knows, however, that relationships between neighbours are not always positive. Conflict relationships can be a source of severe problems and might have an even greater impact on people’s life than positive interactions (Labianca et al., 1998; Labianca and Brass, 2006). Negative relationships can be linked to the broader field of social cohesion research, where neighbourhood interactions are used as a measure for social cohesion (Forrest and Kearns, 2001). Often, however, lack of social cohesion is measured as lack of interactions, while negative relationships are largely ignored in theoretical arguments and empirical research as well (for exceptions, see: Moerbeek and Need, 2003; Völker and Flap, 2007a).

The neighbourhood is an interesting place to look at conflict relationships, because it is, to some extent, difficult to avoid interactions completely—for example, when one goes to work in the morning, when doing groceries or bringing the children to school; in all situations when one leaves the home, there is the chance of meeting a neighbour. With such repeated interactions and difficulty in avoiding each other, the accumulation of small nuisances and retaliations between certain neighbours can potentially result in conflict between these neighbours (Peper and Spierings, 1999; Peper et al., 2002). To gain better insight in conflict relations between neighbours, this study will first describe at what scale negative neighbourhood relationships occur and, secondly, explain why negative relationships occur between neighbours.

Theoretical arguments and empirical research on the emergence of negative relationships between individuals are relatively scarce. The current study draws its theoretical arguments from several research fields that investigate for example conflicts between groups, but also friendship formation and deviant behaviour. From these different fields, a number of explanatory arguments are formed on the individual and contextual levels. In our analysis, we want to determine whether individuals or contexts are more important in explaining the variation in neighbourhood contacts.

Theories and Hypotheses

In the following, we discuss three general perspectives which might explain the emergence of negative relationships—i.e. diversity, uncertainty and social influence. Because of the lack of existing theoretical perspectives about negative relationships, the proposed perspectives are reformulations of theories from different strands of research. Besides negative relationships, two other types of outcome of contact or meeting are possible: positive relationships and no relationships between neighbours. This threefold nature of relationships must be taken into account, because negative relationships can be opposed to positive relationships, or to no relationships, or both.

Diversity

Between neighbourhoods, heterogeneity in the social composition can differ substantially. And within neighbourhoods, neighbours can be in a number of dimensions similar or dissimilar to each other. In the literature on the formation of friendships and other positive relationships, similarity between people is among the most important factors in explaining the emergence of relationships. People who are more similar to each other are also more likely to interact, because interactions with similar others are expected to be more rewarding due to shared knowledge and therefore better mutual understanding (Homans, 1961/1974; Kalmijn, 1998; McPherson et al., 2001). In the case of neighbour relations, it is argued that, when neighbours are more similar to each other regarding important lifestyle features or other social background characteristics, they are more likely to interact (Völker et al., 2007). Conversely, interactions with dissimilar people are more demanding, since mutual understanding is more difficult due to the differences. So if the argument about friendship formation is turned around, it can be argued that one’s foes are probably more dissimilar to a person than one’s friends (Völker and Flap, 2007a).

In the neighbourhood, residents try to achieve a pleasant living environment for themselves. Problems arise when what is considered a pleasant living environment differs between neighbours with different lifestyles and beliefs (Peper and Spierings, 1999), which results in a situation where for example only some residents can realise their idea of a pleasant living environment. Clashing opinions and lifestyles can for instance be that some want to play loud music all day, while others need a quieter environment. Or that some like cooking with fragrant spices, while others cannot stand oriental smells. Different opinions can also exist about the use of public spaces like parking lots or playgrounds. These annoyances seem trivial, but when they accumulate they can result in serious conflicts between neighbours (Peper et al., 2002). Competing ideas about how the neighbourhood should function can exist between residents from different ethnicities, socioeconomic status, religion or age. People from these different categories have different lifestyles that cannot always coincide, possibly leading to conflict.

However, while there are many arguments justifying that dissimilarity enhances conflicts, it can also be argued that competition is more likely between people who are more similar to each other, because they strive for the same resources (Engels, 1845/1976; Völker and Flap, 2007a). For example, individuals of similar socioeconomic background compete with each other for the same jobs (Jencks and Mayer, 1990). Since competition over scarce resources is a zero-sum game, there are both winners and losers. When people are poor sports, losing in competition enhances the chance of negative relationships (Dietz, 2002; Galster, 2008). The latter are found to exist in all social layers of society for competition-based labour markets like those found in the West (Engels, 1845/1976). In the case of competition over scarce resources like jobs, similarities between people have the possibility to produce negative relationships. This association is supported by findings from the workplace, where people who have negative relationships with each other, usually have a similar education and the same gender and where a higher difference in status is related to a lower likelihood of having a negative relationship (Völker and Flap, 2007a).

The previous arguments show that two opposing predictions can be made about how similarity between two neighbours relates to conflict relations. So, no directed hypothesis can be derived. However, to explore further the question of when similarity leads to negative relations, conflict theory is applied to include a macro-level explanation. Conflict theory assumes that people attain their identity by categorising themselves and others in different groups, based on for example, class, ethnicity or age (Tajfel, 1982). Conflict theory argues that more heterogeneity enhances out-group distrust and in-group solidarity. In the case of competition over scarce resources, people tend to trust others who are similar to themselves (in-group members) and other groups are perceived as a threat, therewith creating negative attitudes towards out-group members (LeVine and Campbell, 1972). Putnam (2007) expands this argument by stating that higher heterogeneity leads to worsened relationships in general, regardless of the group of the other person. From this follows the prediction that

With increased heterogeneity in income, ethnicity, religion or age in a neighbourhood, residents have a higher likelihood to have a negative relationship with their neighbours (Hypothesis 1).

Uncertainty

The second perspective from which arguments are derived to predict negative relationships is uncertainty. This perspective deals with situations where neighbours have a short ‘shadow of the future’—i.e. where residents are uncertain about whether they will have a common future with their neighbours (for example, in the case of high turnover expectations). In the case of a short shadow of the future, people are less likely to co-operate to produce a common good, since they are less likely to interact with the same person again, resulting in low expected returns of co-operative behaviour (Axelrod, 1984; Flap, 2005). Few co-operative activities between neighbours have been shown to result in few neighbourhood relationships (Völker et al., 2007); however, a short common future can also lead to conflict relationships, because people have less incentive to behave in a friendly manner when expected returns of co-operative behaviour are low (Völker and Flap, 2007a). Hence, a short shadow of the future is expected to raise the chance of hostile behaviour towards neighbours in situations that would otherwise be solved in a more friendly way, because the likelihood of interacting with a specific person again is much smaller than in case of a long shadow of the future.

At the neighbourhood level, uncertainty about a common future results from neighbourhood turnover expectations: if the turnover rate is expected to be high, residents are uncertain about whether specific neighbours will still live next to them next month. The length of the shadow of the future can also be derived from individual characteristics. First, if residents have an intention to leave the neighbourhood, their common future with neighbours is short. And, secondly, homeownership can be considered a larger barrier to mobility compared with renting, increasing the shadow of the future. Additionally, homeowners will have a bigger incentive to invest in the local community, since a better-organised community will increase the value of their property (DiPasquale and Glaeser, 1999). Homeowners will therefore be more likely to solve disagreements in a friendly way than renters. The neighbourhood-level prediction for uncertainty is

With higher turnover expectations in a neighbourhood, residents have a higher likelihood of having a negative relationship with their neighbours (Hypothesis 2a).

The individual-level hypothesis is

With higher intentions to leave the neighbourhood, or when residents are renters, they have a higher likelihood of having a negative relationship with their neighbours (Hypothesis 2b).

An objection to these hypotheses is the problem of causality: do residents have negative relationships because they have an intention to leave the neighbourhood (and thus a short shadow of the future), or do they have an intention to leave the neighbourhood because they have negative relationships with their neighbours? To account for this problem, additional analyses are employed using data with two time points. With these data, the proposed direction of the effect can be examined.

Social Influence

Social influence is the third perspective related to conflict relationships between neighbours and encompasses the extent to which residents are influenced by their neighbours when considering conflict behaviour. Social influence can take place in several ways (for example, imitation or norm enforcement), which are described in what follows.

The contagion model is based on the assumption that perceived behavioural patterns in the neighbourhood spread amongst residents through imitation and interaction with residents who hold deviant norms (Akers et al., 1979; Friedrichs and Blasius, 2005). Conflict behaviour is incorporated in the norms of residents by perceiving the conflict behaviour of other neighbourhood residents. The people showing this behaviour are less likely to criticise the same behaviour in other residents, since behaving in such a way indicates that they find conflict behaviour acceptable. So when the group showing conflict behaviour is larger, the conflict behaviour of individuals is less likely to be criticised (Akers et al., 1979). In a given neighbourhood, the more conflict behaviour is perceived by its residents, the more likely an individual is to think that this behaviour is acceptable; the less likely an individual is to be criticised for this behaviour and thus the more likely this individual is to adopt this behaviour in situations where it would be applicable. The hypothesis is that

The more conflict a resident perceives in his/her neighbourhood, the higher the likelihood this resident has negative relationships with his/her neighbours (Hypothesis 3).

However, investigating in which situations residents perceive disorder, Sampson and Raudenbush (2004) found that only a minor part of perceived disorder is explained by observed disorder. The larger part is explained by the stratification of the neighbourhood: residents in concentrated poverty neighbourhoods perceive more disorder independently of the observed disorder. This process might work similarly for perceived conflict, leading to the expectation that, in concentrated poverty neighbourhoods, more conflict is perceived and residents therefore are more likely to adopt conflict behaviour.

An additional argument as to why concentrated poverty would lead to more conflict behaviour is that residents of poor neighbourhoods are socially isolated (i.e. they lack sustainable contact with representatives of mainstream society) (Wilson, 1987) and are often stigmatised by society (Sampson and Raudenbush, 2004; Wacquant, 2008). Social isolation and stigmatisation negatively affect the chances people have to overcome their disadvantaged situation, leading to frustration and aggression amongst residents (Peper et al., 2002), making a positive stance on conflict behaviour more likely (Wilson, 1987). For both arguments, the prediction is

The higher the concentration of poverty in a neighbourhood, the higher the likelihood a resident has negative relationships with his/her neighbours (Hypothesis 4).

The contagion model implies interaction amongst neighbourhood residents; however, residents can also have a social network outside the neighbourhood, which also influences them in their behaviour. Besides, it has been found that relationships outside the neighbourhood decrease the likelihood of having relationships inside the neighbourhood (Völker and Flap, 2007b). It can be argued that, when residents have more alternative relationships, they are less likely to embrace deviant norms (about conflict behaviour) that exist in the neighbourhood. The expectation therefore is an interaction with hypotheses 3 and 4

The more alternative relationships a resident has outside the neighbourhood, the smaller the effect of perceived conflict and concentrated poverty on the likelihood that that resident has negative relationships with his/her neighbours (Hypothesis 5).

A possible solution for conflicts between neighbours is that the residents of a neighbourhood solve the problems themselves. However, residents should also be able and willing to do this. If the residents of a neighbourhood have a tight network and mutual trust is high—i.e. social cohesion is high—they are better able to exercise social control (Galster, 2012; Portes, 1998; Portes and Landolt, 1996). First, residents of cohesive neighbourhoods have more opportunity to form common norms than residents of neighbourhoods who hardly meet each other. Secondly, individuals in cohesive neighbourhoods have more incentive to criticise neighbours who defy the norm, because they know this is supported by other neighbours who share that norm (Coleman, 1990). And thirdly, if individuals do not conform to the norm in a cohesive neighbourhood, this is noticed by other residents and consequently these individuals’ reputations are affected, which can be detrimental to future interactions (Raub and Weesie, 1990). Thus, an individual in a cohesive neighbourhood with norms that disfavour conflict is less likely to solve his/her problem with conflict, because he/she wants to avoid punishment by other residents (examples of punishment are that neighbours end their relationship with the norm-defying person or expel him/her from neighbourhood activities). In a neighbourhood without strong ties between neighbours, residents have less incentive to behave in conformity with the dominant norm, since social control is less strong. However, besides being able to enforce norms, neighbours should also be willing to enforce norms. The willingness of residents to intervene on behalf of the neighbourhood depends on the social cohesion in the neighbourhood (Coleman, 1990; Putnam et al., 1993; Sampson et al., 1997), making it a good indicator to predict how well the neighbourhood residents can enforce norms. The expectation is that

The lower the willingness of residents to intervene on behalf of the neighbourhood, the higher the likelihood a resident has negative relationships with his/her neighbours (Hypothesis 6).

Similar to the contagion model, the effect of the willingness to intervene may be influenced by social networks outside the neighbourhood. Individuals who have more relationships outside the neighbourhood are less likely to comply with norms that are enforced by the other residents, because sanctions can more easily be avoided by turning to their alternative relations and reducing interest in the neighbourhood (Coleman, 1990). This interaction is hypothesised as follows

The more alternative relationships a resident has outside the neighbourhood, the smaller the effect of the willingness of residents to intervene on behalf of the neighbourhood on the likelihood that that resident has negative relationships with his/her neighbours (Hypothesis 7).

Data and Measurements

Data

This study utilises the Survey on the Social Networks of the Dutch (SSND), first conducted in 1999/2000, and with a second wave conducted in 2007. To collect the sample from the first wave (SSND1), 40 of the approximately 500 Dutch municipalities were selected. These are representative of the different provinces and regions, taking into account the rate of urbanisation and the number of inhabitants. In every selected municipality, four neighbourhoods based on five position zip code areas were randomly selected. Per neighbourhood, between six and twelve persons between the ages of 18 and 65 were interviewed. The overall response rate is 42 per cent, which is common for Dutch surveys nowadays (Stoop, 2005). The first wave contains 1007 respondents, who are nested within 161 neighbourhoods. There is a slight overrepresentation of men, married people, higher educated people and people with a paid job (Mollenhorst, 2009). In the analysis, these characteristics are controlled for.

For the data collection from the second wave (SSND2), 863 of the addresses were traced back and, in total, 604 respondents were re-interviewed. The sample comprises people between 26 and 72 years of age. Married people, older people and higher educated people were more likely to participate again in the second wave, resulting in a slight overrepresentation of these groups (Mollenhorst, 2009). These characteristics are controlled for in the analysis. An additional 394 residents within the same neighbourhoods were newly selected to enlarge the sample from the second wave. From this additional sample, there is no information at the time of the first wave. This study uses neighbourhood-level variables that are constructed from the individual data of SSND respondents who reside in the same neighbourhood. However, it is possible that respondents with negative relationships have a more negative perception of the neighbourhood. Therefore, to increase the reliability of neighbourhood-level variables, respondents of neighbourhoods with less than five residents in the dataset are omitted from the analysis. This way, the impact of individual biases on neighbourhood-level variables is reduced. As a result, in this study, a total sample of the SSND2 is used of 519 respondents who are nested within 98 neighbourhoods. The data from the first wave is only used to do additional tests with the variables ‘perceived conflict’ and ‘negative relationships’ (see the results section for more information).

In addition, contextual information about neighbourhoods has been added from Statistics Netherlands.¹

Dependent Variable

Because of the novelty of empirical research on negative relationships, directly usable measurements for negative relationships are not readily available in the literature. To stay on the path taken by the few previous research articles on negative ties (for example, Völker and Flap, 2007a), negative relationships are defined as whether the respondent indicates having been annoyed by the behaviour of the alter. Negative relationships are thus not equated with actual encounters, but comprise a negative judgement about the alter’s behaviour. Respondents could mention two direct neighbours for the name-generating question: “Who are your direct neighbours?”. The majority mentioned two direct neighbours (n = 405) and a smaller group mentioned one neighbour (n = 114). The following question from the SSND2 is used to measure negative relationships between neighbours: “Do your direct neighbours sometimes annoy you?”. This is asked for seven categories of annoyances: ‘noise’, ‘(cooking) smells’, ‘pets’, ‘garbage/trash on the street’, ‘inconveniently parked car’, ‘children’ and ‘other’. For each respondent, a dichotomous variable is created that measures whether he/she feels annoyed by his/her direct neighbours, if the respondent answered having been annoyed by his/her neighbours on at least one of the seven categories. Descriptive information on this and other variables is found in Table 1.

Table 1.

Descriptive statistics of the variables

Variable name	N	Mean	S.D.	Minimum	Maximum
Neighbourhood level	98
Religious heterogeneity	98	0.30	0.04	0.19	0.42
Age heterogeneity	98	0.66	0.16	0	0.92
Ethnic heterogeneity	98	0.34	0.19	0.04	0.91
Income heterogeneity	98	0.24	0.03	0.15	0.34
Turnover expectations	98	0.76	0.11	0.50	0.96
Mean income	98	0.41	0.12	0.18	0.74
Property value	98	0.29	0.16	0	0.73
Willingness to intervene	98	0.77	0.07	0.58	0.94
Individual-level	519
Negative relations (dependent variable)	519	0.18	0.39	0	1
Intention to leave	519	0.76	0.24	0	1
Homeownership	519	0.72	0.45	0	1
Alternative relationships	519	90.78	50.11	0	26
Perceived conflict	519	0.25	0.16	0	1
Male	519	0.52	0.50	0	1
Married	519	0.74	0.44	0	1
Employed	519	0.37	0.48	0	1
Years of education	519	0.53	0.28	0	1
Immigration status
Dutch	519	0.90	0.30	0	1
1st-generation immigrant	519	0.05	0.22	0	1
2nd-generation immigrant	519	0.05	0.22	0	1
Age	519	54.72	12.55	21	89

Source: SSND2.

Independent Variables

Neighbourhood-level variables

At the neighbourhood level, four diversity measurements are created for income, age, ethnicity and religion. Following Marsden (1987), to measure the heterogeneity in age, the standard deviation of the age of all residents in a neighbourhood is used. Data on the distribution of age are only available divided into groups of five years of age (0–4, 5–9, etc.), of which the centre is taken for the analysis. The data are available for four position zip code areas from StatLine for 2007 (Statistics Netherlands, 2010). Ethnic and religious heterogeneity are measured using a standardised version of the diversity index (see Agresti and Agresti, 1978).² For every zip code area, the ethnic heterogeneity index is created with the categories ‘Dutch’, ‘Western non-Dutch’ and ‘non-Western non-Dutch’, for which the data are also available from StatLine (Statistics Netherlands, 2010). Religious heterogeneity is constructed using the religion of all members of a certain neighbourhood in the SSND2, with the categories ‘Roman Catholic’, ‘Protestant’, ‘Muslim’, ‘other’ and ‘no religion’. Income heterogeneity is constructed with the diversity index, using SSND2’s income measure (see later) of all respondents in a certain neighbourhood.

Expected turnover at the neighbourhood level for both waves is measured by aggregating the answers of all respondents in a certain neighbourhood on the statement: “I’m not planning on living here for more than a couple of years”. The answer to this statement in the SSND2 is measured on a five-point scale (‘totally agree’, ‘agree’, ‘not agree, but not totally disagree’, ‘disagree’, ‘totally disagree’) and is standardised on a scale from zero to one.

The concentration of poverty is measured with two continuous variables. First, the mean property value of the dwellings in a neighbourhood is used, available for five position zip code areas for the year 2004 from StatLine (Statistics Netherlands, 2010). This measure excludes the value of property that is used primarily for non-residential purposes, even though people may reside there as well—like, for example, farms. Also this variable is standardised.

The second value used to measure concentration of poverty is the mean of the income of all respondents in a certain neighbourhood. In the second wave of the SSND, people’s net income in euros is asked with answering categories ‘less than 250’, ‘between 251 and 500’, ‘between 501 and 750’, …, ‘more than 4000’. For every category, the centre is taken as the corresponding value. Thus category ‘less than 250’ gets ‘125’ and ‘more than 4000’ gets ‘4125’. Consequently, this variable is standardised, to prevent distortions when it is used in an interaction effect.

The willingness of neighbourhood members to intervene is measured by aggregating the answers of respondents in a neighbourhood on the question of whether they would expect neighbourhood members to take action if they see something unwelcome, asked about the following subjects: “children hanging around”, “youngsters writing graffiti on the wall”, “a rough argument on the street”, “houses are broken into”, “someone is meddling with the parked car of one of the neighbours”, “fighting children”, “the local authorities have plans to build a centre for addicts in the neighbourhood” and “a club/discotheque will be built in this neighbourhood”. The SSND2 contained five answering categories (‘yes, sure’, ‘probably yes’, ‘not improbable, not probable’, ‘probably not’, ‘surely not’). The willingness to intervene is made into a standardised scale with an alpha of 0.76.

Individual-level variables

The individual intention to move out of the neighbourhood as indicator of a short shadow of the future is measured in the SSND2 with the statement “I’m not planning on living here for more than a couple of years”, to which the respondents could answer on a five-point scale how much they agreed with this. This measure is also standardised on a scale from zero to one.

A second individual measure for the shadow of the future is whether residents own their house or rent, which is used as a dichotomous variable in the analysis. Owning indicates a longer shadow of the future, renting a shorter one.

Perceived conflict in a neighbourhood at the time of the SSND2 is measured with the five-point scale answer to the statement: “There is little conflict between people in this neighbourhood”, which is reverse coded and standardised for the analysis. For an additional analysis, this variable is also measured for the first wave, to verify the causal order.

The number of alternative relationships an individual has outside the neighbourhood is measured as the total number of network members mentioned by a respondent as a result of the 12 name-generating questions in the SSND2 (see Fischer, 1982, for more information about name-generating questions), excluding people from the respondent’s own neighbourhood. The sizes of the obtained networks range from zero to 30 alters.

Control variables

Control variables are added for the gender, marital status, age, employment status and years of education of the respondent, because on these factors, the data are not fully representative of the Dutch population. Marital status is a dichotomous variable that measures whether people are married or cohabiting. Age is measured in years of age. In the data, education is provided as the highest finished education; this is recoded into years of education using a 1996 recoding key from the International Stratification and Mobility File (Ganzeboom, 2010). Besides, a control variable is added for immigration status, which is dummified into three variables: ‘Dutch’, ‘first-generation immigrant’ (i.e. born outside the Netherlands) and ‘second-generation immigrant’ (i.e. born in the Netherlands and at least one parent born outside the Netherlands).

Whether or not a person had negative relationships in the first wave is also controlled for in an additional analysis. This variable is measured in the same way as the dependent variable, but with information from SSND1. The variable controls for whether respondents have had negative relationships for a longer time, which indicates that negative relationships can be long lasting, and in that case are not explained by the independent variables used in the analysis.

Method

Variables in the dataset are available on two levels: the neighbourhood level and the individual level. The likelihood of having negative relationships—i.e. the dependent variable—is considered an individual characteristic. Because individual respondents are nested within neighbourhoods, the most appropriate method to analyse the data is with multilevel techniques (Hox, 2002). And, because of the dichotomous nature of the dependent variable, a logistic variant of multilevel analysis is used. The main analyses are conducted using the SSND2. In the text, additional analyses are mentioned that make use of SSND1 variables; these are not presented in the table.

Results

When looking at how many people actually have negative relationships with their direct neighbours, more than 18 per cent of the respondents state that they have been annoyed by at least one of their direct neighbours. By far the most mentioned annoyance is noise, other often-mentioned annoyances are children, pets, inconveniently parked cars, verbal violence and conflicts about garden borders. The most common way for people to deal with annoyances is actively to find a solution by talking about it with the neighbour who annoys them (80 per cent), or by talking with other neighbours about the problem (5 per cent). Avoiding the neighbour and doing nothing about the annoyance is a much less favoured behaviour (10 per cent).

The data are structured in two levels, with the neighbourhood as the highest level and individuals at the lowest. The variance at the individual level is fixed (see Table 2), as is standard in logistic multilevel models. Comparing individual- and neighbourhood-level variance, it shows that most variance is at the individual level, while 15 per cent³ of the variance is at the neighbourhood level, indicating that there is clustering of negative relationships within neighbourhoods.

Table 2.

Logistic two-level models on negative relationships (N = 519)

	M0: Control variables		M1: Neighbourhood-level variables		M2: M1 + individual-level variables		M3a: M2 + interaction		M3b: M2 + interaction		M3c: M2 + interaction		M3d: M2 + interaction
	Coefficient	S.E.	Coefficient	S.E.	Coefficient	S.E.	Coefficient	S.E.	Coefficient	S.E.	Coefficient	S.E.	Coefficient	S.E.
Neighbourhood level
Religious heterogeneity^a		2.93	2.72	3.64	2.70†	2.88	2.77	3.54	2.71†	3.53	2.71†	3.89	2.71†
Age heterogeneity^a		−0.66	0.85	−0.88	0.85	−0.95	0.86	−0.93	0.85	−0.85	0.85	−0.89	0.84
Ethnic heterogeneity^a		0.19	0.75	−0.04	0.75	−0.04	0.76	−0.03	0.75	−0.07	0.75	−0.08	0.74
Income heterogeneity^a		−0.41	2.41	−0.76	2.41	−0.54	2.45	−0.96	2.42	−1.08	2.40	−0.85	2.44
Turnover expectations^a		−0.82	1.18	−0.43	1.27	−0.14	1.29	−0.35	1.27	−0.34	1.26	−0.38	1.25
Mean income^a		−0.75	1.26	−0.63	1.25	−3.79	2.44†	−0.62	1.24	−0.75	1.25	−0.61	1.24
Property value^a		1.73	0.99	1.86	0.98	1.78	1.00	1.58	1.80	1.90	0.98	1.84	0.97
Willingness to intervene^a		−5.38	2.04**	−4.75	2.04**	−4.71	2.05*	−4.71	2.04*	−8.45	3.55**	−5.28	2.03**
Individual level
Intention to leave ^a			−0.08	0.54	−0.17	0.55	−0.13	0.55	−0.15	0.55	−0.17	0.56
Homeownership^a			−0.30	0.30	−0.28	0.31	−0.27	0.31	−0.26	0.31	−0.28	0.31
Perceived conflict^a			1.53	0.72*	10.47	0.72*	10.53	0.73*	1.57	0.73*	4.64	1.42**
Alternative relationships				−0.11	0.09	0.01	0.05	0.10	0.07	0.12	0.05**
Interaction variables
Alt. relations*mean income				0.30	0.20
Alt. relations*property value					0.02	0.14
Alt. relations*willingn. to interv.						−0.34	0.27
Alt. relations*perceived conflict							−0.35	0.14*
Constant	0.13	0.66	4.03	2.00*	2.90	2.06	4.08	2.25†	2.82	2.07	5.60	3.01†	2.13	2.06
Variance neighbourhood-level	0.58	0.25	0.33	0.36	0.24	0.48	0.24	0.48	0.21	0.53	0.21	0.53	0.14	0.80
Variance individual-level	3.29		3.29		3.29		3.29		3.29		3.29		3.29
Log likelihood	−242.55		−235.89		−232.89		−231.39		−232.56		−231.73		−229.34
LR chi² (df)^b			13.31(8)		6.00(3)		3.00(2)^c		0.65(2)^c		2.32(2)^c		7.09(2)* ^c

based on one-sided test.

Likelihood-ratio test compares a model with the previous model.

M3a–d are compared with M2.

Notes: All models contain the following control variables: male, married, employed, years of education, immigration status, and age. ** p <0.01; * p <0.05; † p <0.10.Source: SSND2.

In model 1, the neighbourhood-level variables are added. Compared with the control variables model, this model explains a large part of the variance at the neighbourhood level. The first hypothesis about neighbourhood-level effects states that, with increased heterogeneity in income, ethnicity, religion and age, the likelihood of people having a negative relationship increases (H1). Inspecting model 1, no evidence is found that heterogeneity in income, ethnicity and age lead to more negative relationships. However, the effect of religious diversity is marginally significant (see model 2). Thus, support is found for the hypothesis that, with increased religious diversity, people have a higher likelihood of negative relationships with their neighbours. Furthermore, additional analyses tested whether, when different cleavages coincide, the effect of diversity would be stronger. However, no such result was established.

Turnover expectations at the neighbourhood level as an indicator for a short shadow of the future are found not to support the second hypothesis, which predicts that with higher turnover expectations in the neighbourhood, residents have a higher likelihood for negative relationships (H2a).

The hypothesis that predicts a higher likelihood of having negative relationships with a higher concentration of poverty (H4) can be tested using two indicator variables: mean income and mean property value of the dwellings in a neighbourhood. For both indicators, no significant effect is found and thus no support is found for the fourth hypothesis. However, the coefficient of mean property value seems strongly to indicate that the effect is in the opposite direction, suggesting that, with higher mean property values, the likelihood of negative relationships increases. This finding rejects the hypothesised relation between concentrated poverty and negative relationships.

Furthermore, model 1 shows support for the hypothesis that, the lower the willingness to intervene on behalf of the neighbourhood, the higher the likelihood that a resident has negative relationships with his/her neighbours (H6). The effect of willingness to intervene is significant and of considerable size compared with other neighbourhood-level variables.

In model 2, the individual-level hypotheses are tested. Intention to leave the neighbourhood and homeownership are used as individual indicators of the shadow of the future; having a higher intention to leave means a shorter shadow of the future, and being a homeowner meaning a longer shadow of the future. The model shows significant effects for neither of these two variables, therewith finding no support for the hypothesis that an intention to leave or being a renter leads to a higher likelihood of negative relationships (H2b).

Hypothesis 3 states that the more conflict residents perceive in their neighbourhood, the higher the likelihood that they have a negative relationship with their direct neighbours (H3). The result shows that indeed when residents perceive more conflict, this leads to having a higher likelihood of negative relationships. However, because having negative relationships can also enhance the likelihood that residents perceive conflict in their neighbourhood, perceived conflict is also measured at the first SSND wave to explain negative relationships in the second wave, to give a better indication about the causal order. When estimating the model with perceived conflict at the first wave, the result remains, which supports the hypothesised causal order.

The last models (M3a–d) include the interaction effects that are intended to show that the effects of perceived conflict, concentrated poverty (H5) and the willingness to intervene (H7) are smaller for persons who have more alternative relationships outside the neighbourhood. Three of the interaction effects are insignificant, so the hypotheses are not supported for concentrated poverty and willingness to intervene. As model 3d shows, the interaction effect between alternative relationships and perceived conflict is significant, therewith supporting the hypothesis. Thus, the effect of the conflict residents perceive in their neighbourhood on the likelihood that they have negative relationships, is smaller when they have more alternative relationships outside the neighbourhood. Also, adding the interaction effect reveals the effect of alternative relationships on negative relationships—i.e. the more alternative relationships a person has outside the neighbourhood, the higher the likelihood of having negative relationships. This effect, however, is very small.

Besides the models shown in Table 2, an additional model was estimated using information from the first wave of the SSND about whether respondents had negative relationships at that time. The variable showed a significant positive effect, while at same time the other effects remained as they are presented in Table 2. This indicates that negative relationships tend to be long-lasting or that there are still unobserved variables that explain negative relationships.

Conclusion

In this study, negative relationships between neighbours were investigated according to three perspectives: diversity, uncertainty and social influence. The explanatory arguments based on these perspectives covered both the neighbourhood and the individual level. With the current results, most support is found for the social influence perspective, and both the neighbourhood and the individual level seem important in explaining negative relationships. This perspective supports the idea that norms and shared expectations on behaviour in neighbourhoods explain the emergence of animosity. In neighbourhoods where people do not control each other to intervene on behalf of the collective good, more negative relationships exist. It is unlikely that there exists only a one-way causal relationship, since individual negative relationships may well lead to less social cohesion in a neighbourhood. However, by additional analyses using independent variables measured at an earlier point in time, the objection of reversed causality was met to some extent, and the current interpretations remain valid.

Considering the diversity perspective, the argument was that higher income, religious, age and ethnic diversity in the neighbourhood lead to a higher likelihood for negative relationships between neighbours. This argument is supported only for religious diversity, in this regard corroborating our usage of Putnam’s (2007) theory that higher diversity leads to people behaving more negatively towards others. It is possible that the neighbours with whom people have negative relationships are from a different religion, but for now, we only know that religious diversity at the neighbourhood level increases negative relationships among neighbours. A deeper inquiry into that issue would be interesting.

Note, however, that Putnam formulated his claim with regard to ethnic diversity, which we could not prove. A possible explanation for this different finding is that Putnam looks at trust, while we look at negative relationships. It is possible that more diversity does indeed lead to more distrust between neighbours, but not to more negative relationships. Perhaps in neighbourhoods with higher levels of distrust, residents make more effort to avoid each other. When neighbours avoid each other, they may have a lower likelihood of clashing. In this scenario, an increase in distrust does not necessarily have to lead to an increase in negative relationships and this might explain why we did not find any evidence that ethnic diversity increases problems and negative relationships among neighbours.

The uncertainty perspective argument is that a shorter shadow of the future leads to a higher likelihood of negative relationships. At the neighbourhood level this means that higher turnover expectations leads to a higher likelihood of negative relationships; however, this was not found. At the individual level, the argument was that having the intention to leave the neighbourhood or being a renter leads to a higher likelihood of negative relationships. Similar to the neighbourhood-level argument, the individual-level argument was also not supported by the results. An explanation for the lack of support could be that a short shadow of the future does not lead to negative relationships, but only to fewer relationships with neighbours. When people know a relationship will probably not last long, they do not invest in this relationship. Similarly, previous research showed that an individual intention to leave the neighbourhood leads to less investment in the neighbourhood (Völker et al., 2007).

With the current study, most support is found for the social influence perspective. First, support was found for the hypothesis that when residents perceive more conflict in their neighbourhood, they have a higher likelihood of having negative relationships with their neighbours. The argument is that residents can adopt the behaviour they perceive in the neighbourhood when the frequency of that behaviour is higher and the likelihood of sanctions is lower. Moreover, perceived conflict has a lasting effect, since the perceived conflict at an earlier point in time still leads to a higher likelihood of negative relationships. An interesting finding also is that alternative relationships outside the neighbourhood have a decreasing effect on the strength of the effect of perceived conflict on negative relationships. This means that residents with more alternative relationships, undergo less influence of perceived conflict on the likelihood of having negative relationships.

A higher concentration of poverty is expected to lead to a higher likelihood of negative relationships. However, this expectation was not fulfilled. The first indicator was the mean income in a neighbourhood, and the second the mean property value of the dwellings in a neighbourhood. For mean income, the expected effect was not found, and for property value the results show a strong indication that the direction of the effect is actually the other way around. It seems to be the case that, in neighbourhoods with more expensive dwellings, residents are more likely to have negative relationships. A possible explanation for this can be that more affluent people have less need for the support and help of their neighbours and are therefore less restrained in their behaviour towards them, leading to a higher likelihood for negative relationships.

Another argument from the social influence perspective is that residents have a higher likelihood of negative relationships with their neighbours in neighbourhoods where residents are less willing to intervene on behalf of the neighbourhood. Strong support for this argument is shown in the results. It was expected that the influence of the willingness to intervene would differ for residents with more alternative relationships outside the neighbourhood, but this is not found. Willingness to intervene seems to have an effect on all neighbourhood members, regardless of their network outside the neighbourhood.

Considering the three perspectives, first it can be concluded that social influence is the most suitable perspective to predict negative relationships between neighbours. It appears that sanctions are an important mechanism in explaining why people are influenced by others: people can alter their behaviour because of possible sanctions by others (willingness to intervene), and people can adopt the behaviour of others when the likelihood of sanctions is lower (perceived conflict). Secondly, diversity as a perspective to predict negative relationships seems promising, but the notion has to be investigated further. Characteristics of the neighbour could be explored, since now it is not clear whether diversity leads to more negative relationships in general, or that it leads to negative relationships with specific (out-group) neighbours. And finally, the uncertainty perspective was found not to explain negative relationships. This suggests that people are not as calculative in their behaviour towards their neighbours as the shadow of the future argument suggests.

Due to limitations of the data, we cannot test whether a selection effect exists for the perceived conflict argument. It is possible that an underlying factor causes both higher perceived neighbourhood conflict and more negative relationships between neighbours. One possible underlying factor could be that economic or lifestyle differences between residents cause them to compete over resources and values, leading to a higher likelihood of conflict at both the neighbourhood and the individual levels. Competition over values has been captured in the three heterogeneity variables (religion, age and ethnicity). The income heterogeneity variable accounts for competition over resources. Characteristics of the alters could give a relational-level explanation, to get a more precise idea about competition between neighbours.

Pertaining to the different levels, negative relationships are mainly explained by individual characteristics, although neighbourhood characteristics also play a meaningful role. With the neighbourhood characteristics used in the analysis, we get a good idea about what goes on at the neighbourhood level. At the individual level, we have now some idea about how to predict negative relationships, but more work still lies here for future research.

These outcomes imply for policy-makers that they should reconsider their focus. Interventions like community mediation or stimulation of neighbourhood participation are usually set up in problematic neighbourhoods, which, in the Netherlands, are defined by aggregate levels of unemployment, poverty and crime rates (Peper and Spierings, 1999; VROM, 2007, 2009). These problems though, do not necessarily indicate anything conclusive about negative relationships between neighbours, which is shown by the finding that more-affluent neighbourhoods contain more negative relationships. When looking at neighbourhoods, policy-makers would do better to look into group dynamics—as for instance, the collective willingness of residents to intervene in case of problems. This approach would demand the collection of different data, however; when such processes are adequately mapped, policy-makers would be better able to identify neighbourhoods that need attention. Furthermore, the causes of conflict between two neighbours can possibly also be found at the individual level or at the relational level, such as lifestyle differences (Peper and Spierings, 1999)—thus two neighbours could have a negative relationship, while their neighbourhood hardly has any problems. Policy-makers should therefore look beyond the scope of so-called problematic neighbourhoods, and assess conflict relationships between neighbours both on the basis of the neighbours’ interrelational characteristics and the extent to which neighbourhood characteristics influence the residents’ behaviour.

Furthermore, our results provide implications for social-mix policies, which are being endorsed by many European governments (Galster, 2007; Kleinhans, 2004). Social mix is often seen as a cure-all policy: mixing up residents of different ethnicity or socioeconomic background should improve the well-being of individual residents. Social-mix research looks at, amongst other things, (the quality of) social interactions (Kleinhans, 2004). This is the focus of our analyses, although we looked at it at a reversed angle. We tested whether religious, age, ethnic and income diversity lead to a worsened quality of neighbourhood-based relationships. We did not find support for these hypotheses, nor for diversity leading to better-quality relationships, suggesting that policy-makers should be wary of applying social mix to quality of neighbourhood-based relationships. Probably, it would be most fruitful for policy-makers to look at how to increase the collective willingness of residents to intervene in neighbourhood-based problems. In such situations, residents would be able to solve problems even with negative relationships amongst themselves.

Footnotes

Acknowledgements

The authors would like to thank George Galster and Jürgen Friedrichs, who commented on an earlier draft of this article. They would also like to thank the two anonymous referees of Urban Studies for their helpful comments.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Agresti

B. F.

(1978) Statistical analysis of qualitative variation, Sociological Methodology, 9, pp. 204–237.

Akers

R. L.

Krohn

M. D.

Lanza-Kaduce

Radosevich

(1979) Social learning and deviant behavior, American Sociological Review, 44(4), pp. 636–655.

Axelrod

(1984) The Evolution of Cooperation. New York: Basic Books.

Campbell

K. E.

Lee

B. A.

(1992) Sources of personal neighbor networks: social integration, need, or time?, Social Forces, 70(4), pp. 1077–1100.

Coleman

J. S.

(1990) Foundations of Social Theory. Cambridge, MA: Belknap Press.

Dietz

R. D.

(2002) The estimation of neighbourhood effects in the social sciences, Social Science Research, 31, pp. 539–575.

DiPasquale

Glaeser

E. L.

(1999) Incentives and social capital: are homeowners better citizens?, Journal of Urban Economics, 45, pp. 354–384.

Engels

(1845/1976) Die Lage der arbeitenden Klasse in Engeland, in: K. Marx and F. Engels, Werke, Band 2, pp. 225–506. Berlin: Dietz.

Fischer

C. S.

(1982) To Dwell among Friends. Berkeley, CA: University of California Press.

10.

Flap

(2005) Creation and returns of social capital, in: Flap

Völker

(Eds) Creation and Returns of Social Capital, pp. 3–23. Oxford: Routledge.

11.

Forrest

Kearns

(2001) Social cohesion, social capital and the neighbourhood, Urban Studies, 38(12), pp. 2125–2143.

12.

Friedrichs

Blasius

(2005) Social norms in distressed neighbourhoods, in: Friedrichs

Galster

Musterd

(Eds) Life in Poverty Neighbourhoods, pp. 11–30. London: Routledge.

13.

Galster

(2007) Should policy makers strive for neighbourhood social mix? An analysis of the Western European evidence base, Housing Studies, 22(4), pp. 523–545.

14.

Galster

G. C.

(2008) Quantifying the effect of neighbourhood on individuals, Schmollers Jahrbuch, 128, pp. 7–48.

15.

Galster

G. C.

(2012) The mechanism(s) of neighbourhood effects, in: van Ham

Manley

Bailey

Simpson

Maclennan

(Eds) Neighbourhood Effects Research, pp. 23–56. Dordrecht: Springer.

16.

Ganzeboom

(2010) International stratification and mobility file. ISMF website (http://home.fsw.vu.nl/~ganzeboom/ismf/ismf.htm; accessed 4 April 2010).

17.

Guest

A. M.

Wierzbicki

S. K.

(1999) Social ties at the neighborhood level, Urban Affairs Review, 35(1), pp. 92–111.

18.

Hesiodos (1970) The Homeric Hymns and Homerica, transl. by Evelyn-White

H. G.

Cambridge, MA: Harvard University Press.

19.

Homans

G. C.

(1961/1974) Social Behavior. New York: Harcourt Brace Jovanovich.

20.

Hox

(2002) Multilevel Analysis. Mahwah, NJ: Lawrence Erlbaum Associates.

21.

Jencks

Mayer

S. E.

(1990) The social consequences of growing up in poor neighbourhood, in: Lynn

L. E.

McGeary

M. G. H.

(Eds) Inner-city Poverty in the United States, pp. 111–186. Washington, DC: National Academy Press.

22.

Kalmijn

(1998) Intermarriage and homogamy, Annual Review of Sociology, 24, pp. 395–421.

23.

Kleinhans

(2004) Social implications of housing diversification in urban renewal: a review of recent literature, Journal of Housing and the Built Environment, 19, pp. 367–390.

24.

Labianca

Brass

D. J.

(2006) Exploring the social ledger: negative relationships and negative asymmetry in social networks in organizations, Academy of Management Review, 31(3), pp. 596–614.

25.

Labianca

Brass

D. J.

Gray

(1998) Social networks and perceptions of intergroup conflict: the role of negative relationships and third parties, Academy of Management Journal, 41(1), pp. 55–67.

26.

LeVine

R. A.

Campbell

D. T.

(1972) Ethnocentrism. New York: John Wiley.

27.

Marsden

P. V.

(1987) Core discussion networks of Americans, American Sociological Review, 52(1), pp. 122–131.

28.

McPherson

Smith-Lovin

Cook

J. M.

(2001) Birds of a feather: homophily in social networks, Annual Review of Sociology, 27, pp. 415–444.

29.

Moerbeek

H. H. S.

Need

(2003) Enemies at work: can they hinder your career?, Social Networks, 25, pp. 67–82.

30.

Mollenhorst

(2009) Networks in Context. Utrecht: ICS-dissertation.

31.

Peper

Spierings

(1999) Settling disputes between neighbours in the lifeworld: an evaluation of experiments with community mediation in the Netherlands, European Journal on Criminal Policy and Research, 7, pp. 483–507.

32.

Peper

Spierings

Meere

F. de

Jong

W. de

(2002) Moderne buren en hun ruzies, Tijdschrift voor herstelrecht, 2(2), pp. 32–43.

33.

Portes

(1998) Social capital: its origins and applications in modern sociology, Annual Review of Sociology, 24, pp. 1–24.

34.

Portes

Landolt

(1996) Social capital: promise and pitfalls of its role in development, Journal of Latin American Studies, 32(2), pp. 529–547.

35.

Putnam

R. D.

(2000) Bowling Alone. New York: Simon & Schuster.

36.

Putnam

R. D.

(2007) E pluribus unum: diversity and community in the twenty-first century, Scandinavian Political Studies, 30(2), pp. 137–174.

37.

Putnam

R. D.

Leonardi

Nanetti

R. Y.

(1993) Making Democracy Work. Princeton, NJ: Princeton University Press.

38.

Raub

Weesie

(1990) Reputation and efficiency in social interactions. American Journal of Sociology, 96(3), pp. 626–654.

39.

Sampson

R. J.

Raudenbush

S. W.

(2004) Seeing disorder: neighbourhood stigma and the social construction of ‘broken windows’, Social Psychology Quarterly, 67(4), pp. 319–342.

40.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997) Neighbourhoods and violent crime, Science, 277(5328), pp. 918–924.

41.

Statistics Netherlands (2010) StatLine (http://statline.cbs.nl/statweb/; accessed 28 February 2010).

42.

Stoop

I. A. L.

(2005) The hunt for the last respondent. Social and Cultural Planning Office, The Hague.

43.

Tajfel

(1982) Social psychology of intergroup relations, Annual Review of Psychology, 33, pp. 1–39.

44.

Völker

Flap

(2007a) Feinde am Arbeitsplatz: Zur Entstehung von Feindschaften in Organisationen, in: Hummell

H. J.

(Ed.) Die Analyse von Gesellschaften, Organisationen und Individuen in ihrem Zusammenhang, pp. 133–156. Bonn: GESIS-IZ.

45.

Völker

Flap

(2007b) Sixteen million neighbors: a multilevel study of the role of neighbors in the personal networks of the Dutch, Urban Affairs Review, 43(2), pp. 256–284.

46.

Völker

Flap

Lindenberg

(2007) When are neighbourhoods communities?, European Sociological Review, 23(1), pp. 99–114.

47.

VROM (2007) Wijkenselectie voor de wijkenaanpak. Bijlage 1. De indicatoren van de aandachtswijken. Ministry of Housing, Spatial Planning and the Environment, The Hague (www.vrom.nl/get.asp?file=docs/kamerstukken/Tue24Apr20071715180200/DGW2007033161A.doc; accessed 23 March 2009).

48.

VROM (2009) Buurtbewoners in balans. Ministry of Housing, Spatial Planning and the Environment, The Hague.

49.

Wacquant

(2008) Urban Outcasts. Cambridge: Polity Press.

50.

Wellman

(1979) The community question: the intimate networks of East Yorkers, American Journal of Sociology, 84, pp. 1201–1231.

51.

Wellman

(1996) Are personal communities local?, Social Networks, 18, pp. 347–354.

52.

Wilson

W. J.

(1987) The Truly Disadvantaged. Chicago, IL: University of Chicago Press.