Master thesis Sociology
Urban track
Feiko Michiel de Grip
Student-ID: 6065120
July 2016
Supervisor: Yannis Tzaninis
Second reader: Linda op den Kamp
2
Table of contents
Page
Summary 4
Part 1.
Introduction
6
1.1 Studying gentrification using social trust 4
1.2 The case: The Indische Buurt 5
Part 2.
Theoretical framework
9
2.1 Social trust 9
2.1.1 Group formation due to emphasized differences 9
2.1.2 The relevance of social trust 9
2.1.3 Diversity and social trust 10
2.1.4 Applying social trust to gentrification research 11
2.2 Gentrification 12
2.2.1 The Dutch context and social mix strategies 13
2.2.2 Consequences for original residents 13
2.3 Another perspective on trust 14
Part 3.
Research question and hypotheses
18
3.1 Research question 18
3.2 Hypotheses for the quantitative study 19
3.3 Dimensions of trust for the qualitative study 20
Part 4. Research methods 22
4.1 Quantitative research method 22
3
4.1.2 Variables in the analysis 23
4.2 Qualitative research method 26
Part 5.
Results
28
5.1 General trends in the Indische Buurt 29
5.1.1 Growing share of the socioeconomic higher class 29
5.1.2 The dimension of origin 30
5.1.3 Components of social trust 32
5.2 Regression analyses 34
5.3 Comparison with other neighborhoods 38
5.4 Exploring the influence of other factors 40
5.4.1 Origin 40
5.4.2 Length of residence 43
5.5 Findings from the in-depth interviews with residents 45
5.5.1 General impressions 45 5.5.2 Physical developments 46 5.5.3 Class differences 48 5.5.4 Ethnic differences 50 5.5.5 Feeling at home 52 5.5.6 Active citizenship 54
Part 6.
Conclusions
57
References 61 Appendices 65Appendix 1: SPSS output Cronbach’s Alpha analysis for the Social Trust scale.
65
Appendix 2: Inflation correction 66
Appendix 3: Interview transcription with color coded dimensions of social trust
4
Summary
Social trust in a changing neighborhood
This thesis studies the development of social trust in the context of a gentrifying neighborhood. The focus is on the experiences of residents from the lower social class, who are defined by a low level of education. Gentrification is defined by an increasing share of higher educated residents and residents with high incomes. This is an operationalization of Hackworth’s (2002) definition of the concept: “The production of urban space for progressively more affluent users”. Theories on the social effects of gentrification led to the expectation that people in the lower social class would either suffer from social isolation or alienation due to the income of middle- and higher class residents (Marcuse, 1985; Blokland, 2003; Uitermark, Duyvendak and Kleinhans, 2007; Reijndorp, 2007; Galster et al, 2008; Lees et al, 2008), or would experience emancipatory effects (Atkinson, 2004; Metaal en Teijmant, 2008). Also, theories on social trust show a problematic relationship between both socio-economic- and ethnic diversity and social trust in a society (Putnam, 2000; 2007; Alesina and la Ferrara, 2002; Lancee and Dronkers, 2008).
In this study, the relationship between an increasing share of higher class residents and the social trust of lower class residents has been studied in a gentrifying neighborhood in Amsterdam: The Indische Buurt. Apart from being a gentrifying neighborhood, this is also an ethnically diverse neighborhood. The role of this second factor of diversity has been considered as well. Two methods have been used: A longitudinal quantitative analysis of self-reported social trust data on the neighborhood scale has been combined with the analysis of in-depth interviews with residents.
Despite the expectation that gentrification would influence the social trust of lower class residents, such a relationship has not been found in the quantitative analysis. In the neighborhood as a whole, social trust has grown faster than the Amsterdam average and it grew among all socioeconomic classes. Talking to residents led to more understanding of the mechanisms that underlie this growth. The physical renewal that neighborhood dwellings and public space have undergone is highly appreciated by the respondents and make them feel more comfortable in the neighborhood. Furthermore, the serious improvement of the social infrastructure that has predominantly been initiated by residents themselves is experienced as a
5 factor that connects residents. This familiarity with others might contribute to people’s social trust. The expectation that a confrontation between social classes influences people’s social trust, has not been verified. Some respondents say to experience a certain level of alienation from the neighborhood due to the adaptation of its facilities to a wealthier population, but this development does not worry people very much. The ‘newcomers’ are seen as decent people, a factor that might cause people’s social trust not to be affected.
A possible explanation for the seemingly positive configuration of different groups in the Indische Buurt is offered by Oosterlynck’s (2016) perspective on solidarity. He argues that solidarity between people cannot only be traced along historical ethnic- and cultural lines, but can also be found in the here and now. When people live together in a neighborhood, the multiplicity of interactions (be they superficial or not) in the streets can lead to a sense of belonging and a comforting feeling of solidarity. People in the Indische Buurt are used to diversity and have seen many newcomers in their neighborhood’s streets. Since they do not feel threatened by the current newcomers, they seem to be barely bothered by their arrival.
6
Part 1. Introduction
1.1 Studying gentrification using social trust
This thesis studies the development of social trust in the context of a gentrifying neighborhood. It defines gentrification as the process in which the share of people with a high income and high education in a neighborhood grows. This definition aims to operationalize Hackworth’s (2002) notion of the concept -“The production of urban space for progressively more affluent users”- into measurable dimensions of income level and educational level. Social trust is defined as a person’s belief that others will not, at worst, knowingly or willingly harm him, and will, at best, act in his interests (ESS, 2013). To find out how this social trust develops among the generalized population of a neighborhood, a quantifiable measure is used. The social effects of gentrification have not often been studied in such a way, i.e. by looking at the development of the generalized attitude of residents towards each other. This thesis combines such a method with in depth interviews with residents, in order to enlarge the understanding of how the mechanisms that are studied really work for residents. Does an increased presence of the higher socioeconomic class in a neighborhood lead to the experience of class polarization among other groups, or is there peaceful co-existence?
Attempts to answer such questions have predominantly concentrated on qualitative accounts of interaction between social classes in gentrifying neighborhoods (Uitermark et al, 2007; Lees et al, 2008). Elementary research on the co-existence of different groups (Elias and Scotson, 1994) and the sociology of group formation (Brubaker, 2004) leads to a concern for social struggle as a result of group diversity. Meanwhile, social research into the relationship between increasing ethnic diversity and the trust people have in their neighborhood and their fellow residents predominantly shows a negative relationship between the two (Putnam, 2007; Lancee and Dronkers, 2008) Also, studies have found that an increasing socioeconomic diversity (meaning the income of a certain socioeconomic class in an area where another class has been the dominant group) is harmful for the social trust of residents in this neighborhood (Alesina and La Ferrara, 2002). However, this has been studied in specific circumstances, which makes it still important to transfer the study to different contexts. While using a different approach, there are studies examining cases of gentrification in the Dutch context (Metaal and Teijmant, 2008; Veldboer and Bergstra, 2011) that show a trajectory in which residents are not so much threatened by increasing socioeconomic diversity. These studies point at the mitigating role of regulation related to the Dutch welfare state, which makes gentrification less of an
7 infringement on people’s lives. The contrary is shown for instance by Marcuse (1985) in the New York context, where tenants’ rights have been far more contested and Desmond (2012) who studied the threat of eviction for the urban poor, also in the American context.
Summarizing, the social effects of gentrification are still a topic of debate. Different approaches are used to study these effects. Also, the specific context of a neighborhood matters. This thesis is concerned with the Indische Buurt, a neighborhood in eastern Amsterdam. It aims to examine the effect of gentrification on the social trust of residents on the longer term. Residents of the Indische Buurt have reported their attitudes on questions about social trust every two years since 2001. Applying this generalized measure of social trust to the study of gentrification gives the opportunity to study a long term development. This study may enlarge our understanding of how a changing neighborhood population affects the way in which the people living there feel about each other. The study looks specifically into the experiences of the lower socioeconomic class in order to find out whether they feel threatened by an income of the higher socioeconomic class in the neighborhood. Socioeconomic classes are defined here along lines of educational level and income. The neighborhood and its characteristics will now be described to complete this introduction. In the following chapter the theoretical framework of the thesis will be discussed.
1.2 The case: The Indische Buurt
The Indische Buurt is an interesting case of gentrification. At this moment, the neighborhood is increasingly popular as a place to live for the socio-economic middle class (Gemeente Amsterdam, 2016). The strategy of the municipality and housing corporations is to provide for these groups, transforming the housing stock of the neighborhood from predominantly social housing to a more heterogeneous housing composition (Sakizlioglu, 2014). This means that a part of social housing is sold on the private market and another part is redeveloped and transferred to the private rent sector. The share of social housing in the total housing stock was very large in 2005 with 85%. In 2015 it had decreased to 63%, which is still above the Amsterdam average. As a result, the population has changed. Wealthier people have moved into the neighborhood, and the population’s ethnic composition has changed as well. The Indische Buurt has for a long time been home to many migrants from non-western countries, predominantly from Turkey and Morocco. Because of the income of wealthier native Dutch people and the leaving of a part of the relatively poor migrant population, the share of non-western migrants has decreased from 59% in 2005 to 47% in 2015, whereas their share had only been growing in the three decades before. Still, the neighborhood is ethnically diverse (see
8 table 1), which possibly adds a dimension to gentrification compared to neighborhoods that start the process with a more mono-ethnic population. More on this is discussed in the theoretical framework.
Tabel 1: Amount of residents by origin in the Indische Buurt (januari 1, 2016) Origin Native Dutch Moroccan Western migrants Non-Western Turkish Surinames e Antillean Total Amount 8379 4434 3155 2662 2114 1854 271 2269 6 Source: Gemeente Amsterdam, Dienst Onderzoek en Statistiek.
The growing share of wealthier people is clearly illustrated by one group, the so called ‘new urbanites’. The Amsterdam municipality uses this term to refer to people of western origin between 18 and 54 years old that have settled in the city after the age of 17 (Gemeente Amsterdam, 2016). According to the municipality, this group is often highly educated and serves as an indicator of a neighborhood’s attractiveness to the socioeconomic middle class. As such, ‘new urbanites’ are considered a possible catalyst of gentrification. (Gemeente Amsterdam, 2016) The share of this group in the total neighborhood population has risen from 22% in 2005 to 33% in 2015.
9 Map 1 and 2: Location of the Indische Buurt
Source: CBS
Source: Google Maps
The Indische Buurt is located in the Zeeburg borough in Northeastern Amsterdam. The Zeeburgerdijk separates the neighborhood in the north from the eastern port area of the city that is now a residential area. In the West and Southwest the neighborhood is edged by the railway connecting Amsterdam to Utrecht. The Muiderpoort railway station lies in the neighborhood along this railway. In the south the neighborhood borders the Ringvaart canal, while in the East the Flevopark separates the neighborhood from the water of the river IJ. The Indische Buurt was built in the beginning of 20th century, to house the workers of the city’s port that expanded fast towards the east by that time. But in the 1960’s, the port moved all its functions towards the west of the city, and the Indische Buurt lost its primary reason for existence. As a result of the moving out of port workers and the more prosperous residents, the neighborhood started to suffer from housing vacancy and physical dilapidation. A combination of residents, activists and entrepreneurs from the neighborhood started protesting against the downfall of the neighborhood and they were successful to the extent that the municipality decided to renovate old houses and built new blocks in the 1970’s and 1980’s. The very high share of social housing remained intact. But the outflow of people that could afford to move did not stop. Migrants from Morocco, Turkey and Surinam moved into their houses, which made the Indische Buurt
10 into the multicultural neighborhood it still is, but it also became a relatively poor neighborhood with all kinds of related problems (Heijdra, 2000).
In the early 1990’s, Anderiesen and Reijndorp (1990) wrote about the perspectives of people on the Indische Buurt. Both policy makers and others called it a ‘failed neighborhood’, that had only become less attractive and more problematic because of the character of the newly build social housing blocks. It increasingly suffered from poverty, high crime rates and public space was deteriorating. This developments contributed to the awareness that the neighborhood had to be drastically restructured. In the earlier policy interventions the idea was that improving- and increasing the share of social housing was the solution for urban areas in decline, but this had now reversed. In the 1990’s policy makers started to see a large share of social housing as the very reason for decline (Van Gent, 2012). For the Indische Buurt and similar urban problem areas new policies were developed, aimed at the removal of the problems of a high share of low-income groups and ethnic minorities. In order achieve this, the housing market had to be restructured. Social housing was to be privatized and redeveloped in order to attract middle class people. The first interventions were done in 2004 by housing corporation Ymere in the Berlageblokken at the Javasquare, after quite some opposition of the former inhabitants1. But the responsible actor continued quickly with the neighborhood renewal, using the promising perspective of a ‘liveable’ neighborhood to put residents at ease (Sakizlioglu, 2014).
1 It even came to lawsuits when inhabitants refused to leave their houses (Amsterdam Court of
Justice, 2004). Link to the verdict of the court of justice:
11
Part 2. Theoretical framework
2.1 Social trust
2.1.1 Group formation due to emphasized differences
When the demographics of a neighborhood change, this means in concrete that new people come to live there and/or that others leave. Consequently, it may be the case that social connections in the neighborhood change as well. People get new neighbors, or their children get new classmates. Be they intense or not, be they satisfactory to old and new residents or not, new connections will likely arise. According to several studies (Elias and Scotson, 1994; Putnam, 2000; Uitermark et al, 2007), the new connections made are likely to be superficial at best and conflicting at worst when people identify less with newcomers than they did with those that lived in their neighborhood before. And differences do not have to be abundant to lead to disidentification, like Elias and Scotson (1994) showed by studying the dynamics of group formation in a working class neighborhood of Winston Parva, UK. The two groups were only separated by the fact that one group lived there already and the other group moved in at a certain moment. The researchers show that exclusion by means of stigmatization and gossip reinforce the group separation. In-group solidarity grows while people get more anxious and distrustful towards the other group. Thinking from Elias and Scotson’s observations, established and newcomers will have a hard time trusting each other, and thus living together in harmony. Although their observations are case specific, the insights in the more general human tendency to form groups are widely acclaimed in psychology (for example Turner, 1987) and sociology (Brubaker, 2004). Because of entirely different circumstances it is likely that the Indische Buurt shows different social dynamics then Winston Parva does, but we must keep in mind that people are inclined to form a group and possibly distrust the other. Brubaker (2004) explains why forming groups is a cognitive necessity for human beings. Dividing people and things into clear, understandable categories is a way to make people able to deal with the complexity and diversity of the world. Just like Elias and Scotson, Brubaker (2004) sees that forming groups and reinforcing group separation is partly done by stigmatization that encourages the distrust in the other group.
2.1.2 The relevance of social trust
A lack of trust in another group will not directly affect people’s everyday life when they do not interact with the other group on a frequent basis. But this changes when people and groups
12 anxious or reserved towards each other meet time after time in their daily life, as is shown by studies on the social effects of group diversity in societies (Alesina and La Ferrara, 2002; Costa and Kahn, 2003; Putnam, 2007). So when people with whom the original residents can hardly identify move into a neighborhood, distrust caused by disidentification can make life less enjoyable for both old and new residents.
It is important to note that if we add social to trust we deal with the form of trust that relates to other people and groups of people. Academic literature thoroughly deals with the relevance of social trust. In doing so, social trust is often lifted up from the micro-perspective of an individual’s trust in others to the combined social trust of people that form a society, in order to assess the larger implications of the presence or absence of trust in a society. The assumption underlying such generalized research into social trust is that an individual’s trust can be objectively assessed and taken together for a group. Looking for a working definition of social trust, the European Social Survey (ESS, 2013) offers one that covers most of the academic approaches examining it: “Social trust is the belief that others will not, at worst, knowingly or willingly do you harm, and will, at best, act in your interests.” This definition combines the notions of three scholars: Hardin (1998, p.12) sees social trust as an "encapsulated self-interest”, Warren (1999, p. 311) argues that trust in others emerges out of shared interests and the absence of threat, and Gambetta (1988, p. 217) adds to this that trust involves the belief that others will perform in a way that is beneficial to us. Approaching social trust in ways more or less similar to the definitions above, researchers aim to find how social trust influences people and societies and how trust is influenced by other factors. In doing so researchers have shown the importance of social trust for both people’s physical wellbeing2 and the relative
success of societies as a whole. For instance, La Porta et al. (1997) as well as Knack and Keefer (1997) showed in comparative- (between societies) and longitudinal (at multiple moments in one society) studies that social trust stimulates economic growth. The general explanation for this dynamic is that there is more (and more productive) cooperation in societies where people have more social trust.
2.1.3 Diversity and social trust
When social trust is important for societies and the wellbeing of the people that form it, it is important to see which factors threaten it. Putnam (2000) studied the relation between ethnic
2 The focus here is not on the relation between social trust and physical wellbeing, but a lot of
trust-research deals with it and has shown for example that someone’s health benefits from a high level of trust in others (Subramania, Kim and Kawachi, 2002)
13 diversity and social trust in the United States. He sees that people in ethnically diverse neighborhoods have a lower level of social trust. This connects with Elias and Scotson (1994) and Brubaker’s (2004) analyses: (Ethnic) group differences seem to go hand in hand with a lack of trust. Putnam (2007) also finds that neighborhoods that are socio-economically mixed have relatively low trust rates. This is in line with the earlier findings of Alesina and La Ferrara (2002), who found a negative relation between class differences and social trust of neighborhood residents as well. Costa and Kahn (2003) study the role of volunteering work in societies in relation to social trust. They argue that volunteering in societal activities has a two-sided relationship with social trust as both a contributor to- and an outcome of it: People who have a higher level of social trust are more likely to volunteer, and a high level of volunteering contributes positively to the experience of a committed society that is beneficial to someone’s wellbeing. They found that increasing income inequality has enforced a decline in volunteering rates in the United States between 1950 and 1990. Letki (2008) elaborated on the previous studies on predictors of social trust in the United Kingdom, and argues that the role of ethnicity should be nuanced. She says that ethnic diversity only has an indirect effect on social trust: The factor that really accounts for lower trust in neighborhoods is the socio-economic status. People in economically disadvantaged neighborhoods have lower levels of social trust. However, Fieldhouse and Cutts (2010) find that diversity does account for lower trust by itself. They additionally find that only autochthonous residents experience lower levels of trust as a result of diversity. Ethnic newcomers are not negatively affected by diversity. In this light it is interesting to examine how different groups in the ethnically diverse Indische Buurt react to diversity.
The discussed findings are not univocal, and the same goes for research into diversity in the Dutch context. Lancee and Dronkers (2008), in a Dutch replication of Putnam’s (2007) study find that ethnic diversity negatively influences social trust. But they add that it is mainly the effect of having neighbors of a different ethnicity that negatively influences social trust in the neighborhood. However, Tolsma et al. (2009) find no relation between ethnic diversity and social trust at the neighborhood level, but only at the municipal level. Just like Letki (2008) they find that inside neighborhoods, lower social trust is caused by income and educational differences.
2.1.4 Applying social trust to gentrification research
According to several studies, people’s social trust is related to the diversity they are confronted with. It is interesting to apply this perspective to the study of gentrification, a process in which
14 the population of a neighborhood changes. If increasing diversity threatens people’s social trust (Putnam, 2007; Lancee and Dronkers, 2008) and if people see newcomers as a distinct group with whom they cannot identify (see Elias and Scotson, 1994), gentrification might be harmful for social trust. As will become clear in the next paragraph, gentrification literature is very critical of the effects of an insertion of people from the middle- and higher social classes in a neighborhood with primarily lower class residents. Looking from this class perspective, several studies see in gentrification a threat for the lower class (Blokland, 2003; Uitermark, Duyvendak and Kleinhans, 2007; Reijndorp, 2007; Galster et al, 2008; Lees, 2010). But how is their social trust threatened? Applying social trust to the analysis of residents confronted with socioeconomic diversity might give insights in the attitude of people in a gentrifying neighborhood towards each other during the process of gentrification. The next paragraph elaborates on the topic of gentrification.
2.2 Gentrification
A large body of literature covers the gentrification process. Also, many definitions have been used. A broad definition comes from Hackworth (2002): “The production of urban space for progressively more affluent users”. This definition seems to be applicable to all forms of what is seen as gentrification, since it describes the active process of making an urban area more attractive for middle-class people to live in. However, it does not touch on how this production of space happens and what can be the results of it. Van Gent (2012) proposes to handle the concept by discussing the actors of transformation and the social consequences. In this part, the role of governmental institutions as actors of transformation will be discussed first. Afterwards, the social consequences of gentrification that are mediated by specific circumstances of a case are looked into.
Over time, the process of gentrification has been pushed by different actors and it has occurred in different ways. Most recent processes of gentrification in The Netherlands, the Indische Buurt included, show the characteristics of what Hackworth and Smith (2001) call the third wave of gentrification. They see a first wave starting in the late 1960’s, when gentrification was mainly a state-led effort to prevent inner city neighborhoods to deteriorate. The second wave, that started when the economic depression of the nineteen-seventies came to an end, was fueled more by the private market than the state, and fitted in a time of neoliberal politics that pursued the increasing power of the free market (Peck and Tickle, 2002). The third wave of gentrification described by Hackworth and Smith starts from the mid-1990s, and the most
15 important difference with wave two is that the state takes back the leading role. How this can be observed in The Netherlands at this moment will now be discussed.
2.2.1 The Dutch context and social mix strategies
Although Hackworth and Smith (2001) admit that most of their empirical evidence is from the US, they argue that gentrification is a general process that shows similarities in the places where it occurs. Van Gent (2012) subscribes that there are general characteristics of gentrification, but also sees the importance of looking at local contexts. He studies the process of gentrification in Amsterdam in relation to the change of relevant institutions. In doing so Van Gent sees that the role of the state in the gentrification process is very strong in The Netherlands, a parallel to what Hackworth and Smith (2001) observe in the third phase of gentrification.
Van Gent (2012) argues that in Western-European cities, municipal strategies of gentrification are often related to the upgrading of deprived neighborhoods by pursuing a more mixed population when it comes to financial power. Lees (2000) points out that such strategies that mix social classes are used to create what policy makers call ‘liveable neighborhoods’. However, the livability concept is subjective, and policy makers mainly use it as a cover for increasing the value of housing by attracting wealthier residents, Lees says. Van Gent (2010) describes that physical and social interventions through housing are done to achieve this social mix. The Dutch government’s policy to reduce the social housing stock is an example. Social housing is sold to private developers who sell or rent houses on the market for much more money than the former residents can afford. In this way, policy makers actively insert middle-income households into low-middle-income neighborhoods. (Van Gent, 2012)
2.2.2 Consequences for original residents
One of the arguments policy makers use in favor of social mixing is the proclaimed emancipatory effect of the presence of middle-class people for lower income groups (Uitermark, Duyvendak and Kleinhans, 2007). According to Atkinson (2004) social mixing could indeed be positive for original residents when it leads to a better physical state of the neighborhood dwellings and an improved neighborhood management. Uitermark, Duyvendak and Kleinhans (2007) show that original residents often believe in this promising future as well, but see in fact social divisions between middle-class and lower income groups as a result. Others (Blokland, 2003; Galster et al., 2008) also see a minimal amount of interaction between socioeconomic groups and observe few other emancipatory effects for original residents in socially mixed neighborhoods. Reijndorp (2007, p. 149) points out that neighborhood
16 regeneration using social mix strategies causes a “dichotomy between the disadvantaged and the educated”. Lees et al. (2010) in their overview of gentrification theories show that many authors see a clash of lifestyles in disadvantaged neighborhoods that are being rediscovered by the middle class. A sense of being taken over prevails among original residents when neighborhood facilities (shops, bars) adapt to the new, more affluent residents. This results in a decreasing level of trust in the neighborhood. Marcuse (1986) provides a conceptualization for the feeling of being taken over. He speaks of displacement pressure. The concept refers to a sense of alienation with the neighborhood that original residents may experience when their neighborhood changes both culturally (by the replacement of old facilities with new, different ones) and demographically. As a result, original residents can feel a pressure to leave the neighborhood: displacement pressure. Marcuse (1986) also describes the direct form of displacement of original residents. This relates to the eviction of people that cannot afford higher rents anymore.
In the Dutch contemporary context, the process of changing neighborhoods populations is mediated by the welfare state and housing market regulation (Veldboer and Bergstra, 2011, p.6). The term “mild gentrification” was used by Van Weesep and Wiegersma (1990) to describe the process in Amsterdam’s Jordaan neighborhood, where they found little social friction. A similar process was found in Amsterdam Westerpark by Metaal and Teijmant (2008). Veldboer and Bergstra (2011) see a very gradual change of the housing stock (from social rent to owner occupation) as the main advantage of Dutch cases of mild gentrification. They argue that when a population changes gradually and without forced eviction, it feels natural and more diversity can be sustainable.
2.3 Another perspective on trust
The emergence of a gap between social classes in gentrifying neighborhoods is observed by several studies (Blokland, 2003; Uitermark, Duyvendak and Kleinhans, 2007; Reijndorp, 2007; Galster et al, 2008). However, in none of these cases real tensions occur, the problem lies more in social separation of social classes. It is interesting to examine if lower class people really feel threatened by wealthy newcomers, or if they just interact with them superficially and feel allright about this. In this perspective, Oosterlynck’s (2016) ideas are interesting. He argues that superficial interactions in the everyday life are a good base for solidarity with people with other backgrounds. According to him, solidarity has until now been understood as being grounded in the spatial boundedness of territorial states and the intergenerational continuity of supposedly culturally homogeneous nations (Oosterlynck, 2016, p. 2). When we assume that
17 solidarity exists only inside groups that are defined by a common nationality or culture that traces back for centuries, we may be blind for solidarity that emerges when people that live together at a certain place at the current moment. This solidarity can be a result of meeting in the grocery store or in the streets of a neighborhood. In the case of the Indische Buurt, Oosterlynck’s (2016) perspective can be useful to understand how people with so many different cultural and ethnic backgrounds live together here. When this ethnic diversity is followed by a process of gentrification that brings socioeconomic diversity, it may again be possible that solidarity develops between people from different social classes just because people live together.
18
Part 3. Research question and hypotheses
3.1 Research question
The academic literature on gentrification provides useful insights on the trajectory of the gentrification process and the consequences it can have for the people living in gentrifying neighborhoods. However, the social consequences of gentrification have mainly been studied by talking to a select sample of residents (Doucet, 2011). Not often have they been studied on the longer term in a quantified way, trying to find generalizable relationships between changes and their consequences. This is understandable to an extent, because it seems hard to grasp complex feelings of belonging and connecting to others by using a survey. But using reports of social trust as a quantified measure to grasp these feelings, does offer the opportunity to look into the factors influencing this feelings longitudinally on the neighborhood level. Still, we should pay attention to the micro-differences and complex qualitative mechanisms that underlie these processes. Only such an effort may enable us to study the relationship between gentrification and social trust in a more comprehensive way. Therefore, this research combines a quantitative analysis on indicators of social trust with the analysis of in-depth interviews with residents.
The focus in this study is on the experiences of residents from the lower socioeconomic class. In gentrification, the lower socioeconomic class is confronted with the income of people from higher classes, and this study aims to find out how this influences them. Socioeconomic classes are defined here by differences in educational level and income, which is a common approach in (primarily statistical) analysis of social class differences, and is also used by the Dutch Central Bureau of Statistics (Verweij and Mulder, 2014). The theoretical visions and empirical work that have been discussed see different consequences of a growing class diversity, some in which social trust of the lower class is threatened and some showing more positive effects. Cases of gentrification differ largely, so this scattered image is understandable. However, this makes it difficult to predict how the process works out in the Indische Buurt. Neighborhood characteristics like the ethnic diversity that is present, as well as the phase in which the process is now, might give gentrification here a specific face. Before the research, no expectations could be formulated about the relationship between demographic changes and the social trust of the lower socioeconomic class. Therefore the research design is exploratory. This results in a research question that aims to find out how demographic changes influence the
19 social trust of the socioeconomic lower class (indicated by a low level of education) in the Indische Buurt.
Research question: How does the social trust of lower educated residents in the Indische Buurt evolve in the context of a gentrifying neighborhood?
In the attempt to answer the research question in a comprehensive and detailed way, two research methods have been used: A quantitative analysis of social trust reported by residents between 2007 and 2015 is combined with a qualitative analysis of in-depth interviews with long-term, mainly lower educated residents. The quantitative approach tested two hypotheses, they are presented in paragraph 2.4.2. For the qualitative approach several dimensions of social trust have been determined, they are discussed in paragraph 2.4.3.
3.2 Hypotheses for the quantitative study
In the statistical analysis that was conducted, the most important variables are social trust (dependent) and the share of people with a high educational level and income level (independent). An increasing share of the higher socioeconomic class (people with a high education and -income) in the total population is used as an indicator for gentrification. This is similar to other studies that approach gentrification in a statistical way (Hochstenbach et al, 2014). The effect of a growing share of the higher socioeconomic class has been examined. In the variable paragraph of the methods section it will be explained in more detail what is seen as a high- and a low level of education and income.
For the statistical analyses, the exploratory design means that two hypotheses are tested, both expecting an inversely proportional relationship between gentrification and social trust. On the one hand, hypothesis 1 formulates the expectation that the lower socioeconomic class experiences a decreasing level of trust in fellow residents when the socioeconomic higher class moves in. This is based on studies on social trust explaining that diversity relates to social trust negatively. Both ethnic diversity (Putnam, 2007) and socio-economic diversity (income inequality) (Alesina and La Ferrara, 2002; Costa and Kahn, 2003) in a neighborhood have been ascribed a negative influence on the social trust of people living there. As we can trace from elementary observations about group formation (Elias and Scotson, 1994), the slightest differences between people results in group separation, leading to and reinforced by a lack of trust in the other group. This, combined with more applied gentrification theories that see the process as a threat for lower socioeconomic classes (Lees, 2010; Reijndorp, 2007; Marcuse,
20 1986), leads to the expectation that trust will decrease among the lower educated. On the other hand, hypothesis 2 expects the contrary. The studies more applied to Dutch cases of gentrification picture more nuanced consequences of the process. Mild gentrification, as Veldboer and Bergstra (2011) call gentrification that is mitigated by tenants’ rights and welfare state institutions, is a form that is more often seen in The Netherlands. Furthermore, there are voices saying that gentrification can have positive effects for lower socioeconomic classes (Atkinson, 2004; Doucet, 2011). Insights from these studies led to a reversed expectation from the first hypothesis in hypothesis 2. The two hypotheses areas follows:
H1: An increasing share of the high educated and high income groups positively influences the social trust reported by the lower educated.
H2: An increasing share of the high educated and high income groups negatively influences the social trust reported by the lower educated.
3.3 Dimensions of trust for the qualitative study
Social trust is a multidimensional concept that is not easy to grasp. If we would ask someone “how high is your level of social trust?”, that person would have a hard time answering something that makes sense. This complexity of the concept is the reason that a scale of five questions is used (more about the scale in the methods section, § 3.1.2) in an attempt to measure it in the quantitative part of this research. For a qualitative approach of social trust another mode of operation is needed. Here we want to look deeper into the mechanisms that cause possible changes in the attitudes people have towards their fellow residents. Therefore, the interviews that have been conducted break down the concept into several dimensions that can be related to the indicators that form the scale of social trust. These dimensions serve as an attempt to capture the wide range of factors influencing social trust. It is not said that they all together represent the social trust of residents, but they inform us on how people experience the neighborhood and their fellow residents and as such can approach their social trust. A dynamic in each of the dimensions might have a positive or negative influence on people’s social trust. The dimensions are the red line in the qualitative study. The interview has been structured via these dimensions and the results will also be discussed using these dimensions. The interviews have also been used to find out whether there are other dimensions that play a role in the social trust of residents. The first dimension is the general impression of living in the neighborhood. Presumably, people are not breaking down their thoughts on their neighborhood into several elements of social trust. It is more likely that they have general feeling that is the unconscious
21 result of several aspects of living in a neighborhood. Therefore, the first dimension to look into is this general feeling. Afterwards this feeling will be broken down into more dimensions.
Table 2: Operationalization scheme Social trust and gentrification
Concept Quantitative indicators Qualitative dimensions
Social trust
Overall satisfaction with the neighborhood. - General impression: combination of all other dimensions.
- Physical development of buildings and public spaces.
Group interaction. - Experience of social class differences and separation.
- Experience of ethnic differences and separation.
Feeling at home. - Social connections
- Neighborhood facilities Commitment of residents with the
neighborhood.
- Active citizenship.
Future perspective. - Changing experiences over time.
22
Part 4. Research methods
It has become clear by now that this research uses a two-way strategy. Quantitative data analysis is conducted to identify long term demographic changes and to explore if there is statistical evidence that these changes have affected the social trust of the lower socioeconomic class in the Indische Buurt. In order to get a better understanding of the mechanisms that determine the relationship between a changing neighborhood and social trust, in-depth interviews have been conducted with long term residents. The aim is to combine these two methods and get a comprehensive understanding of the process. This mixed methods strategy serves a pragmatic approach of social reality. While qualitative and quantitative researchers often reject each other’s methods, they can also be cumulative. Feilzer (2010) presents a pragmatic paradigm that sees social reality as so complex and multidimensional that researchers should make use of the added value of both qualitative and quantitative methods. For example, using only a qualitative approach, it would be difficult to find out whether social change has occurred in time. Meanwhile, a quantitative longitudinal study by itself gives limited understanding of what people really experience. This research aims at combining insights from both, while acknowledging possible limitations, to get a broad as well as a specific insight in the effect of demographic change on social trust.
4.1 Quantitative research method
4.1.1 The WiA data
The research design of this part is longitudinal. The long-term evolution in social trust will be studied. This is mainly possible because of the data that are used. Access was obtained to data of the Wonen in Amsterdam (WiA) study, conducted by the Amsterdam municipality. This exploring study into all kinds of social aspects of living in Amsterdam is conducted every two years since 2001. Self-completion questionnaires are filled in by a sample of the city’s population. This provides the opportunity to study long-term processes by using these data. All respondents are 18 years or older, which means that information always concerns the grown up population of the neighborhood.
23 4.1.2 Variables in the analysis
In this paragraph all variables in the analyses that are conducted are described. Their construction and other important things will be discussed.
Social trust (dependent variable)
The dependent variable in all the analyses that are conducted is social trust. Just like in the work of Putnam (2000; 2007) the trust of people in their fellow residents is measured using survey questions. However, Putnam uses only one question to measure social trust, namely: “To what extent do you trust your neighbors?” This is a relatively superficial measure, as Boutellier et al (2007) show. Social trust can depend on more than only trust in neighbors. Therefore, the measure of trust in the WiA data is based on the scale of social trust of Boutellier et al. (2007). This scale is based on five questions. The answer categories to each question range from 1 to 10, 1 representing a very negative answer and 10 representing a very positive answer. The scale of social trust is the mean of these five answers. The five questions are:
How would you describe your satisfaction with the neighborhood? How do different groups of people interact with each other?
How much are residents involved with what happens in the neighborhood? Do you feel at home in the neighborhood?
How do you see the evolvement of the neighborhood in the coming years? Translated from Oirschot et al. (2011, p.76).
The 5-item scale has only been used in the WiA survey since 2007. In previous years, only three items were used. Therefore the decision has been made to concentrate the analysis on the time period 2007-2015. The internal reliability of the scale of social trust has been tested using the Cronbach’s Alpha test. This test controls if the different items correlate with an extent so large that they measure a similar component all together. In this case, Cronbach’s Alpha α =0.855 is generously above the critical limit of 0.6, which means it is reasonable to use the five indicators all together as a scale. The output of the Cronbach’s Alpha can be found in appendix 1.
Level of education
As one of the indicators of social class, the level of education is an important variable. This variable is an ordinal variable that shows the level of education for each respondent. People were asked to report the level of their highest education certificate. They could report a wide range of educations. These have been divided into three levels, according to the division used
24 by the CBS (Verweij, 2014). The table underneath shows how different educations have been divided into a lower- middle- and high level of education.
Table 3: Levels of education
Level of education Highest certificate
Low No certificate or only a few years of
elementary school, elementary school, VSO (special education),
VBO/LBO/ULO.
Middle MAVO/MULO/VMBO, MBO,
HAVO/VWO/Gymnasium, HBS.
High HBO, University
Lower educated
Because the research examines the development of social trust reported by lower educated residents, a variable is needed that separates lower educated respondents from all others. Therefore this dummy variable, in which all lower educated respondents have the value 1 and all others have the value 0, was created.
Income
This is the variable that shows the income of each respondent’s household. Each respondent was asked to report the net income of his or her household. This means that in households with more than one working member, these incomes are added up.
Percentage high incomes
This variable indicates the share of the higher social class in the Indische Buurt. The variable has been constructed using the ordinary income variable. This variable shows for each year the percentage of respondents that has an income that falls within the highest quartile of all incomes of Amsterdam. This makes it possible to see whether the Indische Buurt becomes increasingly wealthy in relation to Amsterdam as a whole. Because the research deals with a development in time, inflation leads to a natural increase in incomes. But this does not mean that the financial capacity of respondents increases. In order to be really able to observe a change in the share of high incomes, there has been corrected for inflation. An income that was in the highest quartile in 2007, could not be in there in 2015. Therefore it would be impossible to see whether the percentage of high incomes has increased or decreased in this period. So again, the data have
25 been corrected for inflation. The border of the highest income quartile of 2007 on Amsterdam scale has been indexed using the CPI for each of the following years. The variable shows the share of people in each year that have an income that would have belonged to the highest 25% in 2007. These are incomes that have the same financial capacity as those above €2800 in 2007. See appendix 2 for the indexed borders of the income quartiles.
Time
Because a development in time is examined here, it is necessary that each respondent can be linked to the year in which this person took part in the survey. Initially, the data of each year were in distinct datasets. These sets were merged into one big dataset. In the process of merging the datasets, all cases in a certain year got a value ascending from 1 (2007) to 5 (2015) on the newly constructed variable called Time.
Origin
The dimension of origin of respondents might play a role in their social trust. Therefore this variable is used. It shows the origin of respondents in six categories. The categories are the four groups of origin that are the most represented in the Indische Buurt (Dutch, Moroccan, Turkish, Surinamese/Antillean) complemented with the categories of Western immigrants and Non-western immigrants. The latter two are not very informative because they contain numerous nationalities, but the WiA data provided no other options. For each of the categories dummy variables have been constructed, in order to be able to examine the developments of each distinct group.
Length of residence
This variable is used to examine the role of someone’s length of residence in the neighborhood. It simply shows the number of years someone lives in the Indische Buurt for each respondent.
Age (control variable)
This variable displays the age as it was reported by each respondent. It is used as a control variable.
26 4.2 Qualitative research method
In order to get a better understanding of what underlies the relationship between a changing neighborhood and the social trust of residents, in-depth interviews have been conducted with people that already lived in the Indische Buurt before 2007. This is the first year that is included in the longitudinal data analysis and also a good criterion to separate people that lived in the neighborhood before the large scale interventions from those that moved in during the process of change. In this way, the experiences of people that have lived in the Indische Buurt throughout the changes, have been scrutinized.
Seven semi-structured interviews were conducted. Al interviews have been transcribed and subsequently analyzed using Atlas.TI software. Passages in the interviews that said something about a certain dimension of social trust, have been coded with a color attributed to that dimension. This helped in analyzing all information per dimension. An example of how the interviews have been transcribed and analyzed can be found in appendix 3.
In finding respondents, the focus has been on lower educated people. This might give insights in the mechanisms influencing the social trust of the lower socioeconomic class. However, also some higher educated people have been interviewed (a student and a resident active in a neighborhood community), trying to provide insights from more perspectives. Table 4 informs us on their age, their origin, the length of their residence in the Indische Buurt, their level of education and whether they are active citizens or not.
Table 4: Respondents Responden
t
Age Years living here
Origin Level of education
Active citizen
1. Mandy 29 29 Dutch Middle Yes
2. Jeffrey 40 35 Surinamese Low No
3. Lenie 74 33 Dutch Low No
4. Isha 29 29 Surinamese
2nd gen.
High Yes
5. Herman 47 14 Dutch High Yes
6. Niek 25 25 Dutch High No
27 The level of education contains three categories; lower- middle- and higher educated. Knowing the educational level of respondents may be useful in analyzing their perspective on a changing level of education around them. People’s attitude towards educational groups may be influenced by their own level of education. The final category, about active citizenship is based on the information obtained in the interviews. Active citizens are people that are involved in one of the neighborhood communities that organize all kinds of events, which according to the respondents aim primarily at having fun and connecting residents. It is important to be aware of the activeness of respondents, because their very activeness might influence the social configuration of the neighborhood as well as their own perception of how different groups go along. The names of respondents are fictive because of the respondents’ privacy.
28
Part 5. Results
In this chapter the results of the research are presented. First, the findings based on the quantitative data will be discussed. This part begins with a brief discussion of important general developments on education, income and social trust in the Indische Buurt. Thereafter, the results of the regression analyses are presented, showing if and how people’s social trust has been influenced by the demographic changes from 2007 to 2015. When these main results have been discussed, we continue by exploring two other factors that might influence the relationships between social trust and a changing neighborhood here, namely ethnic differences and length of residence. Subsequently, the elaboration of the analysis will be continued by involving the interviews with residents of the neighborhood. These conversations serve to further interpret the quantitative findings and also add some insights to areas of the analysis that remained unexplained until then.
29 5.1 General trends in the Indische Buurt
5.1.1 Growing share of the socioeconomic higher class
Gentrification, if defined by the settlement of the socioeconomic middle- and higher classes in neighborhoods previously inhabited by mainly the lower class, can be traced by a rise in the general educational level of residents (Hochstenbach et al. 2015). An assumption underlying this research was that this development could be seen in the Indische Buurt, and this appears to be the case. Graph 1, that is based on the WiA data, shows such a trend. It depicts three lines, representing the percentages of the population above 17 years old with lower-, middle- and higher education. It shows the development between 2007 and 2015. There has been a considerable decrease of the share of lower educated people (from 51% in 2007 to 34% in 2015) and a growing share of highly educated people from 14% in 2007 to 33% in 2015. It is interesting that the shares of the three educational levels have become much more equal over this period. Higher educated people were a minority and are now almost equally represented in the Indische Buurt as lower educated people.
Graph 1: Share (%) of the three education levels - development 2007-2015
Source: Own analysis on WiA-data
Related to the rising share of higher educated residents in the Indische Buurt, is a rise in income. This is confirmed by a regression model with the net household income as the outcome variable and level of education as the predictor: B=512,93, p<.001, meaning that people that are highly educated are likely to have an income of €512,93 more than those who are mid-level educated. Consequently, graph 2 shows that the average level of income of people above 17 in the
0 10 20 30 40 50 60 2007 2009 2011 2013 2015
30 Indische Buurt rised from €1.816 in 2007 to €2.524 in 2015. However, these are the actual income data. In order to be able to compare the financial capacity of residents in time in a fair way, the incomes have been corrected for inflation, more about this procedure can be read in the methods section (§3.1.2). In that section can also be found the procedure for the creation of the variable percentage high incomes that uses the lower limit of the highest quartile of 2007’s incomes in the whole of Amsterdam as the demarcation line to be counted among the high incomes. This limit in 2007 (which was €2800,-) has been indexed for each of the following years using the price index of the Dutch central statistics bureau. Graph 3 shows that by 2007 9,8% of the population of the Indische Buurt had a high income. In 2015 the share of the high income group rised to 18,8%.
Graph 2: Mean household income over time Graph 3: Percentage high incomes over time (indexed)
Source: Own analysis on WiA-data
5.1.2 The dimension of origin
The picture of socioeconomic change in the Indische Buurt is clear: The share of higher educated residents and that of residents with a high income has increased between 2007 and 2015. Before testing if and how these developments influence the social trust of residents, we will shortly look into the dimension of diversity of origin. Before the changes in socio-economic diversity started to show, the Indische Buurt already had an ethnically diverse population. Table
31 1 (page 4) shows the amount of residents per country or region of origin according to the most recent data of the Amsterdam municipality.
Alltough it is the socioeconomic diversity that is scrutinized here, we should consider the opportunity that the diversity of origin influences the relationships we are looking for. An argument for this call is shown by graph 4, that shows for each of the six most represented origins in the Indische Buurt the share of group members that has a certain level of education. The groups of Dutch residents and migrants from Western countries have by far the highest shares of people with high education. All other origin groups show a reversed representation of the education levels. Graph 5 shows the average income of each group of origin, and here we see a trend related to the educational differences. Dutch residents and western migrants have higher incomes then those of other origin groups. The facts that ethnic minorities are generally lower educated and most autochtonous people are high educated might give differences between educational levels an ethnic dimension. This will be considered in the regression analyses testing the relation between the growing share of the higher socioeconomic class and the social trust of the lower socioeconomic class.
32 Graph 4: Share of educational levels per origin
Source: Own analysis on WiA-data
Graph 5: Average householdincome per origin
33 5.1.3 Components of social trust
It has been discussed that social trust is quite an abstract term when it is not broken down into the indicators that define it. To improve the understanding of residents’ trust attitude towards fellow residents, we will briefly consider the attitudes that have been reported on each of the five indicators of social trust. Graph 6 and 7 show the development of these attitudes between 2007 and 2015. Graph 6 does so for only lower educated residents and graph 7 displays the attitudes of the entire sample. The mean of the five indicators has been added as well.
Graph 6/7. Development of the components of social trust reported by lower educated residents (left) and all residents (right).
Source: Own analysis on WiA-data
The most notable thing that becomes clear here is the positive trend in all attitudes, for both lower educated and the population as a whole. The reported social trust has improved between 2007 and 2015. The lower educated are a little less positive on all indicators compared to the population as a whole. The attitude with the most spectacular rise for both lower educated residents and the entire sample is the one towards the future development of the neighborhood (see the actual means per year at the yellow lines). Furthermore it is interesting that, despite more critical attitudes towards other social circumstances, people are relatively positive about feeling at home in the Indische Buurt.
5,94 6,45 6,35 6,61 6,91 5 5,5 6 6,5 7 7,5 8 2007 2009 2011 2013 2015 How do different
groups get along? Do you feel at home here?
How much are people here involved with the neighborhood? How positive are you about the future development? How is your overall appreciation of the neighborhood? Mean Social trust
6,47 7,24 7,14 7,37 7,72 6,25 6,8 6,73 6,94 7,14 5 5,5 6 6,5 7 7,5 8 2007 2009 2011 2013 2015
34 5.2 Regression analyses
Now that we have seen the development of educational level, income and social trust, it is time to conduct statistical analyses to see whether changes in these variables influence each other. First, the development of social trust in the period scrutinized here will be discussed, as well as the development of the social trust reported by lower educated residents. The next step is to test the hypotheses that have been formulated, by examining the influence of a growing share of higher educated people and high incomes on the social trust of the lower educated.
First, table 5 shows the results of two regression models with social trust as outcome variable and level of education and lower educated in time3 as the predictor variables. Both models are significant predictors of social trust (model 1: F=22.59, p <.001; model 2: F=39,41, p <.001), which means that they predict the development of social trust better than if simply the development in the past was projected at the future development. Model 1 shows that the level of education is a significant predictor of social trust (B=2.04, p <.001). The positive coefficient means that higher educated residents report a higher level of social trust. This model also shows a significant relationship between social trust and the interaction variable with the dummy for lower educated and development in time (B=.75, p <.001). The positive coefficient means that the social trust reported by lower educated residents increased.
Table 5. Social trust & level of education, change for lower educated
Model 1 Model 2 Level of education 2.04*** (0.24) 0.84** (0.26) Predictors Lower educated x Time 0.75*** (0.15) 0.03 (0.16) Age 0.03 (0.05) 0.00 (0.05) Control variables Age square 0.00 (0.00) 0.00 (0.00)
Time variable Development over time 1.00*** (0.10) Constant 28.22*** (1.43) 29.41*** (1.40) Adjusted R2 0.04 0.08 Observations Dependent variable: Social trust. * p < 0.05, ** p < 0.01, *** p < 0.001
3 This is the interaction of the dummy separating the lower educated from other respondents. The
35 However, when we look at model 2 (F=39,41, p <.001) we see that the development of social trust among lower educated becomes insignificant. This is the effect of adding the time variable itself to the model. This improves the variance in social trust explained by the model, as we can see from the growing R2. The time variable appears to be a predictor of social trust (B=1.00, p <.001). This shows us that social trust has increased significantly in the Indische Buurt between 2007 and 2015. The fact that adding the time variable to the model makes the lower educated/time interaction variable insignificant, means that the increase in social trust among lower educated is not explained by their position as lower educated, but by the general trend that social trust rises in the Indische Buurt. The graph underneath shows the development of the social trust reported by the three education groups. It confirms the image that higher educated residents report higher levels of social trust. This is a distinction that remains unchanged between 2007 and 2015, bearing in mind that trust has increased among all three education groups
Graph 8. Mean social trust per education group, development 2007-2015
Source: Own analysis on WiA-data
We continue with the next regression models. Models 3 and 4 (see table 6) test the relationship between the social trust reported by lower educated residents and the share of higher educated residents in the Indische Buurt. This is the first part of testing the hypotheses H1 and H2 that were formulated in the research question section (§2.4.2). The hypotheses formulate expectations about the relationship between social trust of lower educated residents on the one hand and the share of high educated people and high incomes on the other hand. As model 3 and 4 look into the share of higher educated people, they bring us halfway in testing the
6 6,2 6,4 6,6 6,8 7 7,2 7,4 2007 2009 2011 2013 2015
36 hypotheses. And as we have seen in paragraph 4.1 (graph 1) the share of high educated people has increased between 2007 and 2015.
Both models can predict the social trust in the Indische Buurt better than if no model was used (model 3: F=21.32, p <.001; model 4: F=39.55, p <.001). We will look directly at the relationship between the social trust of lower educated and the share of higher educated people present in the neighborhood. This relationship is tested by the interaction predictor lower educated x percentage higher educated. Model 3 shows that lower educated residents report higher levels of social trust when more high educated people move into the neighborhood (B=.11, p <.001). But model 4 adds the time variable to the analysis, and the significant effect of the interaction variable disappears. The time variable (B=.99, p <.001) takes away the effect of the interaction variable, similar to what happened in model 2. This means that the growing social trust among lower educated resident is not explained by the effect we tested, namely the growing share of higher educated residents. This gives a statistical argument to state that the social trust of lower educated residents is not affected by the share of higher educated people present in the neighborhood.
Table 6. Social trust & share of the higher educated
Model 3 Model 4 Level of education 2.29*** (0.30) 1.01** (0.31) Predictors Lower educated x Percentage higher educated 0.11*** (0.24) 0.02 (0.25) Age 0.03 (0.05) 0.00 (0.05) Control variables Age square 0.00 (0.00) 0.00 (0.00)
Time variable Development over time 0.99*** (0.10) Constant 27.54*** (1.51) 28.95*** (1.48) Adjusted R2 0.03 0.08 Observations Dependent variable: Social trust. Standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001
