The influence of neighbourhoods on wellbeing and mental health : ethnic diversity, neighbourhood deprivation and neighbourhood perceptions in Rotterdam

(1)

MASTER’S THESIS

The Influence of Neighbourhoods on Wellbeing and Mental Health

Ethnic diversity, neighbourhood deprivation and neighbourhood perceptions in Rotterdam

University of Amsterdam Master’s Thesis Sociology ‘Migration and Ethnic Studies’ Evelien Damhuis (11265949) Supervisor: dr. Sonja Fransen / Second reader: dr. Bram Lancee Amsterdam: July 10, 2017

(2)

2

Acknowledgements

First of all, I would particularly like to thank my supervisor, Sonja Fransen, for her interesting ideas and challenging feedback during my master’s thesis process. Due to her keen supervis io n, I was able to develop my thesis in the best way possible. I am very grateful for her support and the learning opportunity provided to me. Moreover, I would like to thank my contacts at the municipality of Rotterdam for helping me with my master’s thesis by providing me the opportunity to use their data of the Health monitor 2012. I would like to thank Özcan in particular for his continuing interest in my project. I would also like to take this opportunity to thank Koen Damhuis and Rik Damhuis for critically reading my draft version. Finally, I would like to thank my parents. Without their loving support none of this would have been possible. I am very proud to present to you my master’s thesis.

(3)

3

The Influence of Neighbourhoods on Wellbeing and Mental

Health

Ethnic diversity, neighbourhood deprivation, and neighbourhood perceptions in Rotterdam

Abstract. The aim of this study was to gain insights in social contexts influenc i ng wellbeing and mental health, specifically the impact of the residential area. This study added to previous literature by combining three elements, namely the impact of neighbourhood ethnic diversity, the focus on both wellbeing and mental health as outcome measures, and by including both neighbourhood effects and neighbourhood perceptions. Following the contact theory, this study predicted ethnic diversity to positively influence wellbeing and mental health, and moreover that perceived neighbourhood social cohesion and social capital positive ly mediated this effect. It was also assumed that living in less deprived neighbourhoods and having better neighbourhood perceptions would be beneficia l for both wellbeing and mental health. Finally, it was assumed that the impact of neighbourhood characteristics and neighbourhood perceptions on wellbeing and mental health would be stronger for ethnic minorities than for native Dutch residents. Using both individual- and neighbourhood- level data from the municipality of Rotterdam, a multilevel design was created. Findings suggested no impact of neighbourhood ethnic diversity on wellbeing and mental health and negative results were found for the mediation mechanisms. Some evidence was found that suggested that living in less deprived neighbourhoods is beneficial for wellbeing and mental health for both native Dutch and ethnic minority residents. This also holds for neighbourhood perceptions. Finally, evidence was found that both supports and contradicts the assumption that neighbourhood effects and neighbourhood perceptions on wellbeing and mental health is stronger for ethnic minorities than native Dutch. Overall, on the basis of the findings in this study, it could be suggested that it is rather neighbourhood deprivation than neighbour hood ethnic diversity influencing wellbeing and mental health of residents. Finally, this study suggests that individual perceptions of the neighbourhood and individ ua l characteristics better explain wellbeing and mental health of both native Dutch and ethnic minority residents than neighbourhood-level characteristics.

KEYWORDS: wellbeing, mental health, ethnic diversity, neighbourhood deprivation, neighbourhood perceptions, perceived social cohesion, social capital

(4)

4

1. Introduction

A person’s wellbeing is of central importance to achievements and capabilities in a wide range of life domains. Substantial research has been undertaken examining the links between wellbeing and work, social relations and longevity amongst others (Diener & Ryan, 2009). Individual wellbeing is often understood as a social phenomenon. It is argued that individ ua ls are social beings whose lives are embedded in and shaped by social structures (e.g. Larson, 1996; Keyes & Shapiro, 2004). In recent years, the urban and more specifically the resident ia l environment as possible explanation of one’s wellbeing has received increased attention (e.g. Ettema & Schekkerman, 2016). Still, this residential relevance was acknowledged somewhat earlier. Leyden, Goldberg & Michelbach (2011), for example, described neighbourhoods as ‘the stage’ where interaction between individuals happens and where individuals participate in social activities that can contribute to one’s wellbeing.

The existing literature on mental health underscores the argument that individuals are embedded in social structures and that the latter also determines mental health conditio ns (Stockdale, Wells, Tang, Belin, Zhang & Sherbourne, 2007). For instance, studies found that the residential environment influences depressive disorders (Mair, Diez-Roux, Osypuk, Rapp, Seeman & Watson, 2010; Termorshuizen, Braam & Van Ameijden, 2015). Similar investigations are of great importance, since depressive symptoms can cause mult ip le unfavourable health outcomes and could have a major impact on daily functionin g, comparable to the effects of other major chronic physical diseases (Licht, De Geus, Zitma n, Hoogendijk, Van Dyck & Penninx, 2008; Buist-Bouwman, De Graaf, Vollebergh, Alonso, Bruffaerts & Ormel, 2006). Providing insights in (social) factors explaining wellbeing and mental health is therefore relevant. This study is interested in a particular social context influencing wellbeing and mental health, namely that of neighbourhoods.

In the literature on neighbourhood effects, various neighbourhood characteristics are argued to be influential on individual wellbeing and mental health. First, the neighbourhood’s ethnic composition is found to be important in explaining wellbeing and mental health. In their article, Veldhuizen, Musterd, Dijkshoorn & Kunst (2015) argue that urban ‘societies’ in Western Europe, including the Netherlands, have undergone some demographic changes. One of these changes include that cities have become more ethnically diverse. In 2016, more than 20% of the Dutch population had a non-Dutch background. In the four largest cities in the Netherlands (Amsterdam, Rotterdam, The Hague & Utrecht) this percentage is even higher

(6)

6

(Statistics Bureau, 2017). On the neighbourhood level, this has led to various neighbourhoods becoming more ethnically heterogeneous. This has made neighbourhood ethnic composition an interesting focus area in the literature on neighbourhood effects.

Scholars studying the association between the ethnic composition of the neighbour hood and individual health found somewhat contradicting outcomes (e.g. Veldhuizen et al., 2015; Schrier et al., 2014; Termorshuizen et al., 2015; Bécares, Stafford, Laurence & Nazroo, 2011). Focusing on suicide risks, Termorshuizen et al. (2015), for example, argued in their article that having a higher share of individuals from the same ethnic group in the neighbourhood (also referred to as own-ethnic density), has beneficial impacts on suicide risk among non-Western minorities. Furthermore, Knies, Nandi & Platt (2016) for example found that a higher concentration of the own-ethnic group in the neighbourhood is associated with higher life satisfaction. On the contrary, Schrier and colleagues (2014) found no association between neighbourhood ethnic density and psychological distress. More recently, Erdem, Burdorf & Van Lenthe (2017) found that the mental health of individuals residing in high ethnic diverse neighbourhoods tends to be worse than those of residents in low ethnic diverse neighbourhoods. When it comes to the interpretation of studies on neighbourhood’s ethnic compositio n and wellbeing and mental health, it is important to distinguish between the different contexts in which the studies were conducted. Whereas some research is conducted in the United States (e.g. Mair et al., 2010), other research is conducted in countries in Western Europe (e.g. Erdem et al., 2017). However, urban areas in European welfare states, like the Netherlands, are quite different than in the United States (Bolt & Van Kempen, 2012). Bolt & Van Kempen argue that in the US, for example, there are high concentrations of ethnic minorities in one ’s residential area. Such ethnic concentrations are quite rare in the Netherlands, where urban areas are more ethnically diverse. It is therefore that this study will focus on neighbour hood ethnic diversity instead of ethnic density.

Besides ethnic composition, a second characteristic that is frequently studied in literature on neighbourhood effects is the association between neighbourhood deprivation and individual health (e.g. Letki, 2008). Letki, for example, found that it is not ethnic divers it y that influences health, but rather neighbourhood deprivation. There is, however, no consensus on how neighbourhood deprivation should be defined and operationalized in existing literature (Van Vuuren, Reijneveld, Van der Wal & Verhoeff (2014). Van Vuuren and colleagues did an attempt and defined neighbourhood deprivation as the relatively low physical, social, and economic position of a neighbourhood. Various studies found that neighbourhood deprivation is associated with multiple health indicators, such as poor

(7)

self-7

rated health and poor mental health (e.g. Poortinga, Dunstan & Fone, 2008). Whereas some studies have shown that the impact of neighbourhood deprivation on individual health is mainly the result of a concentration of individuals with a somewhat low socioecono m ic position in these urban areas, others suggest that there truly is an effect of neighbour hood deprivation on individual health by controlling for individual differences in socioecono m ic position (ibid.). This study is also interested in the role of neighbourhood deprivation on wellbeing and mental health.

Next to neighbourhood- level characteristics, individual perceptions of the neighbourhood are found to be influential on individual wellbeing and mental health too (Leslie & Cerin, 2008; Poortinga et al., 2008; Stafford & Marmot, 2003). Leslie and Cerin, for example, argue that area-level effects on health have been explored extensively, but that the relation between perceptions of the physical and social neighbourhood environment and health is less clear. Some studies found that ethnic composition of the neighbourhood influences (the perception of ) social cohesion and social capital in the neighbourhood (Cramm, Van Dijk & Nieboer, 2013; Putnam, 2007). Ettema & Schekkerman (2016) emphasize that neighbour hood characteristics and individual neighbourhood perceptions have to be taken into account separately, since these concepts differ both conceptually and in their effects on wellbeing and mental health.

This study aims to build on previous literature on neighbourhood effects and individ ua l health in several ways. First, instead of solely looking at risk of mental health problems or individual wellbeing, this study will focus on both. It is relevant to look at as well wellbe i ng as mental health, since evidence is found that wellbeing and mental health are affected differently by neighbourhood characteristics (e.g. Ettema & Schekkerman, 2016). Furthermore, Cramm and her colleagues (2013) argue that the association between neighbourhood effects and individual health has been researched extensively, but that this effect on wellbeing is investigated to a lesser extent. Second, studies that did look at the effects of neighbourhood characteristics on both mental health and wellbeing (e.g. Ettema & Schekkerman, 2016; Leslie & Cerin, 2008), did not look, to my knowledge, at neighbour hood ethnic diversity as a predictor, which is a focus area in this study. Third, whereas the studies that did look at the ethnic composition of the neighbourhood and its effect on wellbeing and mental health, these studies often only focus on the neighbourhood- level effects on health (e.g. Schrier et al., 2014). This study argues that both individual neighbourhood perceptions and neighbourhood- level characteristics are important to consider.

(8)

8

Additionally, previous studies found evidence that in Western societies, there are health inequalities between different ethnic minority groups (e.g. Nielsen & Krasnik, 2010; Devillé, Uiters, Westert & Groenewegen, 2006). Devillé and colleagues stated that the prevalence of poor self-reported health of Dutch natives is around 15%, this prevalence is considerably higher among people with a Turkish background (45%), Moroccan background (39%), and Surinamese background (29%). Moreover, according their results, Bécares, Nazroo, Albor, Chandola & Stafford (2012) indicated that the association between neighbourhood effects and individual health varies among various ethnic groups. Important is therefore to consid er differences in effects between different ethnic groups1_.

According to the Statistics Bureau, around 20% of the Dutch population in 2016 had a ‘non-Dutch’ background2_{(2017). A frequent distinction in the Netherlands is between}

immigrants from ‘western’ and ‘non-western’ descent. Among the Dutch population with a migrant background, 55,3% has a ‘non-western’ background (ibid.). Comparing Dutch cities, Rotterdam has the largest share of ethnic minorities in comparison with other Dutch cities, with 38% of its residents having a non-western migrant background (Statistics Bureau, 2016). With Rotterdam being an interesting ethnic melting-pot, and for reasons on data availability, I opted for the case of Rotterdam.

Considering the above, the following research question is sought to be answered:

Focusing on Rotterdam: to what extent do neighbourhood effects and individual perceptions of the neighbourhood influence individual wellbeing and mental health of native Dutch and ethnic minority residents?

This study will be subdivided into five sections. The following section will elaborate on the different concepts and theories employed in this study concerning individual wellbeing and health, neighbourhood effects, and perceptions of neighbourhoods. Subsequently, the data used in this study will be described as well as the operationalization of the concepts. Subsequently, the analytical strategy will be discussed, before presenting and discussing the results of the analyses.

1_{Due to data limitations, this study is restricted to distinguish between native Dutch and ethnic minorities.} 2_{This percentage refers to both persons who live in the Netherlands, but who were born abroad (so-called}

first-generation immigrants), and persons who live in the Netherlands from whom one or both of the parents were born abroad (so-called second generation immigrants).

(9)

9

2. Theoretical framework

This study is interested in explaining individual wellbeing and mental health with neighbourhood effects and neighbourhood perceptions as possible influencers. The neighbourhood effects that this study is particularly interested in are neighbourhood ethnic diversity and neighbourhood deprivation. These concepts will be elaborated on in this chapter after discussing both concepts that are related to the dependent variables in the present study: wellbeing and mental health. Strikingly, in the existing literature there is no consensus on the definition and operationalization of both concepts. This also holds for the concept of neighbourhood deprivation. This study aims to provide a more comprehens ive conceptualization and operationalization of these concepts. After discussing wellbeing, mental health, and neighbourhood effects, individual- level neighbourhood perceptions will be discussed. The neighbourhood perceptions that are considered in this study are perceived social cohesion, social capital, and the perceived physical environment in the neighbourhood. Finally, differences in neighbourhood effects and neighbourhood perceptions on wellbeing and mental health between native Dutch and ethnic minorities are argued.

2.1 Individual wellbeing

According to Steptoe, Deaton & Stone (2015), people’s wellbeing has become a promine nt focus of debates in economics and public policy. Improving the wellbeing of individuals has therefore emerged to one of the main societal aspirations in various European countries. In academia, different disciplines focused on explaining individual wellbeing, includ ing philosophy, psychology, economics, and sociology (Ettema, Gärling, Olsson & Friman, 2010). Psychological wellbeing is generally defined as ‘the degree to which an individual positive ly evaluates the overall quality of their lives’ (ibid., p. 725). According to Ettema & Schekkerman (2016) the concept of mental health is often related, and sometimes seen as an equivalent of individual wellbeing. However, they argue that effects of neighbourhood characteristics can differ between mental health and wellbeing. Their results suggested that wellbeing was more affected by perceptions of the residential environment, while mental health was more affected by neighbourhood- level effects. Therefore, both concepts will be elaborated and researched separately.

In the literature on wellbeing, various conceptualizations exist. Already in 1985, Diener, Emmons, Larsen & Griffin mentioned an increase in the academic attention on wellbeing.

(10)

10

Specifically, they referred to subjective wellbeing [SWB]. In their article, Diener and colleagues identified three components of SWB: positive affect, negative affect, and life satisfaction. Where the first two referred to a more emotional aspect of wellbeing, the latter referred to a more cognitive aspect. They argued that previous research at that time focused more on the assessment of (positive/negative) affect, and less on the measurement of general life satisfaction. Diener and colleagues developed a multi- item scale to measure life satisfact io n, known as the Satisfaction With Life Scale (SWLS).

According to Keyes (2006), the focus on individual wellbeing rose after World War II, when more tolerance for the diversity of people and viewpoints developed along with the appreciation of the individual. Keyes (2006) argued that since then, also in the academic literature, scholars increasingly focused on individual’s viewpoints and perceptions, and the individual meaning of life. He underscores different types of wellbeing, such as subjective, hedonic and eudemonic wellbeing. He explained subjective wellbeing as a multifactor ia l concept that is concerned with various dimensions of individual wellbeing. Hedonic wellbeing is explained as the perceptions on interest in life, happiness, life satisfaction, and positive and negative affect. Keyes argued eudemonic wellbeing to be the evaluation of one’s psychologica l wellbeing.

Diener & Ryan (2009) follow the idea that subjective wellbeing is a rather multidimensional concept which is used to describe overall wellbeing according to individua l’s subjective evaluations of their lives. Subjective wellbeing can be measured many ways, for example on a more global scale, such as life satisfaction or how much certain feelings are experienced. Although such measures differ in the academic literature, Diener & Ryan (2009) emphasize that these measures all concern individual wellbeing ‘from the subjective standpoint of the respondent’ (p. 391).

More recently, Steptoe and his colleagues (2015) referred to subjective wellbeing as psychological wellbeing. They elaborated on three different aspects of psychological wellbeing: life evaluation (or life satisfaction), hedonic wellbeing, and eudemonic wellbeing. Life evaluation refers to individuals’ thoughts about the quality of their lives. In other terms, how satisfied or happy they are with their lives. Hedonic wellbeing implies the everyday feelings or moods of individuals (e.g. anger, sadness, stress, and happiness, but then the mood not the evaluation of life). Finally, they mention eudemonic wellbeing that focuses on individ ua l judgment about the purpose and meaning of their life.

Clearly, in the academic literature, there are many components and dimens io ns concerning wellbeing, along with various definitions and terms used to describe this concept.

(11)

11

Bobowik, Basabe & Páez (2015) discuss this complexity of the multi-dimensionality of wellbeing and state that, concerning the structure of wellbeing, ambiguities arose because each conceptualization consists of both strengths and weaknesses. However, the above descriptions of wellbeing do have some components in common. First, wellbeing concerns the individ ua l evaluation of their life and/or emotions. Second, life satisfaction is one recurring feature of wellbeing. This study will therefore refer to individual wellbeing as the individual assessment of their overall life satisfaction.

2.2. Mental health

In addition to individual wellbeing, mental health is important to take into consideration in studying neighbourhood effects and neighbourhood perceptions. Even though the concepts of mental health and wellbeing are often related (and mental health is sometimes seen as an equivalent of individual wellbeing), evidence suggests differences in neighbourhood effects on both. Strikingly, many studies on neighbourhood characteristics and mental health neglect elaborating on the concept of mental health (e.g. Leslie & Cerin, 2008; Wandersman & Nation, 1998). This subsection attempts to provide a clearer explanation of the concept of mental health, or at least, how this study will conceptualize mental health.

Keyes (2006) argues that mental health combines both the (complete) state of subjective wellbeing (both hedonic and eudemonic wellbeing), and the absence of mental disorders. In their article, Wandersman & Nation (1998) provide an overview of previous studies that looked at the effects of urban neighbourhoods on mental health. Although not going in-depth in explaining the concept of mental health, some features related to mental health are described in their article. These are psychiatric disorders, depression, anxiety, and somatic symptoms. One feature of mental health returns frequently in the three models Wandersman & Nation describe in their article, namely depression.

Other scholars also mention various features of mental health. First, Echeverría, Diez-Roux, Shea, Borrell & Jackson (2008), who researched the association between the social and physical living environment on mental health, operationalized mental health using depression measures. Leslie & Cerin (2008), who specifically focus on individual perceptions of the neighbourhood on mental health, mention some aspects of mental health, namely (psychological di)stress, depression, and anxiety. More recently, Erdem and colleagues (2017), used psychological distress as indicator for depression.

Looking at racial differences in physical and mental health, Williams, Yu, Jackson & Anderson (1997) described the concept of mental health to be a more overarching concept.

(12)

12

They use two measures for mental health, namely psychological wellbeing and psychologica l distress. They referred to psychological wellbeing as the overall satisfaction with life. Willia ms and colleagues use psychological distress as the sum of the frequency in which respondents felt nervous, sad, hopeless, restless, worthless, and like that everything was an effort. This operationalization is in line with the argumentation of Ettema and Schekkerman (2016), who argued that mental health is sometimes seen as equivalent of wellbeing. However, as argued before, this study will distinguish both concepts. Following the operationalization of many studies described above, this study will refer to mental health as lacking (or holding) depressive symptoms.

2.3. Neighbourhood effects on wellbeing and mental health

This study suggests that it is important to distinguish between neighbourhood- le ve l characteristics and individual perceptions of the neighbourhood, since it is argued that there are conceptual differences as well as differences in effects (Ettema & Schekkerman, 2016). In this subsection, the neighbourhood- level effects will be discussed. These are often referred to as the ‘factual’ characteristics of a neighbourhood and are usually based on land use data and official statistics (ibid.). According to Bécares and colleagues (2012), two main domains of neighbourhood effects are important in terms of its impact on individual wellbeing and mental health, namely the neighbourhood’s social as well as its physical environment. First, ethnic diversity will be discussed and subsequently neighbourhood deprivation.

2.3.1. Ethnic diversity

The first factor that is assumed to affect wellbeing and mental health is neighbourhood ethnic diversity. Thereby, as explained in the introduction, urban spaces in the United States and Western Europe cannot simply be compared (Bolt & Van Kempen, 2012). Bolt & Van Kempen, for example, state that urban areas with higher shares of ethnic minorities in the US (e.g. poor and ‘black’ ghettos) are very different than ‘ethnic concentra te d’ neighbourhoods in, for example, the Netherlands. Such high concentrations of ethnic minorities in one residential area are quite rare in the Netherlands, where the cities are rather ethnically diverse than concentrated. In that sense, Dutch residential areas are not simp ly comparable to the situation in the US. In line with this argumentation -i.e. that there is less ethnic concentration in cities in Western-Europe, but more ethnic heterogeneity- this study will focus on neighbourhood ethnic diversity.

Veldhuizen and her colleagues define ethnic diversity as the degree of ethnic heterogeneity within a neighbourhood (2015). Studies on the association between

(13)

13

neighbourhood ethnic diversity and individual health found somewhat contradic t i ng outcomes. A recent study on the relation between ethnic diversity and psychological distress in the four largest cities in the Netherlands (Amsterdam, Rotterdam, The Hague, and Utrecht) found that the mental health of residents in highly ethnic diverse neighbourhoods to be worse than the mental health of residents in low ethnic diverse neighbourhoods (Erdem et al., 2017). On the contrary, focusing on children, Flink and colleagues found that ethnic inequalities in behavioural and emotional problems may be smallest in ethnic a ll y heterogeneous neighbourhoods (2013). A study on adolescent’s health that was conducted in Canada, found no significant evidence that living in ethnically diverse neighbourhoo ds affects health in general (Abada, Hou & Ram, 2007).

Although there is no general theory in previous research explaining the relatio n between ethnic diversity and individual wellbeing and mental health, scholars often follo w one of two contradicting theories: the contact theory or the conflict theory (Veldhuizen et al., 2015). These theories are often used in combination with three frequently hypothes ized mechanisms of neighbourhood ethnic diversity: social cohesion, social capital, and perceived racial discrimination (e.g. Letki, 2008; Lancee & Dronkers, 2008; Gesthuizen, Van der Meer & Scheepers, 2009; Das-Munshi, Bécares, Stansfeld & Prince, 2010; Bécares et al., 2011; Veldhuizen et al., 2015). For example, Putnam (2007), who conducted research in the United States, found evidence supporting the conflict theory. He found that in ethnically diverse neighbourhoods, people tend to ‘hunker down’, with lower social cohesion and social capital in the neighbourhood. While Putnam not studied these impacts on individual health, other scholars found evidence of these mechanisms affecting individual wellbeing and menta l health (e.g. Cramm et al., 2013). On the other hand, researchers in the Netherlands found that having ethnically diverse neighbours increased the inter-ethnic trust in the neighbourhood (Lancee & Dronkers, 2008).

The previously mentioned mechanisms and theories will be elaborated more in-depth in the subsection 4.2.1. on perceptions of the neighbourhood . For now, this study chooses to follow the contact theory, assuming positive effects of neighbourhood ethnic diversity. This study opts to follow the contact theory, since Putnam’s study is conducted in the US where neighbourhoods are more ethnically concentrated than neighbourhoods in the Netherlands (Bolt & Van Kempen, 2012), assuming that lower concentrations of ethnic minorities results in contact rather than conflict. This leads to the following hypothesis:

(14)

14

H1: More ethnic diversity in the neighbourhood positively influences resident s’ wellbeing and mental health.

2.3.2. Neighbourhood deprivation

Bolt & Van Kempen (2012) emphasized the different contexts between, for example, the United States and western European countries. They argued that ethnic concentrations in urban areas in the U.S. are rather rare in western European countries. Additionally, they stated that relatively deprived areas in the Netherlands are much cleaner and have fewer residents than such areas in the United States. Also, being a welfare state, being poor in the Netherlands is probably less ‘harsh’ than being poor in a more market-driven US (ibid.).

According to Knies, Nandi & Platt (2016), neighbourhood deprivation has not previously, at least before their study, been incorporated into models focusing on ethnic differences in life satisfaction. They argue it is a quite new focus field and therefore important here to take into consideration. Not taking ethnic differences into account, simp ly the association between neighbourhood deprivation and individual health is researched extensively (e.g. Letki, 2008).

It is important to study the association between neighbourhood deprivation and wellbeing and mental health, since a previous finding (although focusing health in genera l) was that it is not neighbourhood ethnic diversity influencing health, but rather neighbourhood deprivation (ibid.). Strikingly, many scholars neglect to elaborate on the concept of neighbourhood deprivation. Verhaeghe and Tampubolon (2012), for examp le, researched the link between neighbourhood deprivation and self-rated health in Engla nd. They did not go in-depth on the clarification of neighbourhood deprivation, but in their operationalization they explained that they use an index for neighbourhood deprivation that was gathered by the Office of the Deputy Prime Minister in 2004. This index focuses on seven domains: income, health, employment, housing, education, crime, and living environment. Various scholars used this Index of Multiple Deprivation [IMD] (e.g. Lang, Llewellyn, Langa, Wallace, Huppert & Melzer, 2008). However, like Verhaeghe and Tampubolon, most scholars neglect to further elaborate on neighbourhood deprivation and the reasons why they have chosen the IMD.

Focusing on youth (0-18 years), Van Vuuren, Reijneveld, Van der Wal & Verhoeff (2014) did go somewhat more in-depth on the effects of neighbourhood deprivation. They explained neighbourhood deprivation as the relatively low physical (e.g. graffiti), social (e.g. unemployment), and economical position (e.g. income) of a neighbourhood. They

(15)

15

furthermore argue that among scholars, there is no consensus on how to define and operationalize neighbourhood deprivation. Therefore, they analysed 19 studies and their operationalization of neighbourhood deprivation. Their conclusion is that most studies used measures of income/wealth, education, and employment in operationalizing neighbour hood deprivation. A few scholars (also) used measures such as housing and living environment.

Although not specifically talking about neighbourhood deprivation, Wandersman & Nation (1998) discuss in their literature overview various neighbourhood characteristics that partially touch upon the above description, and their impact on mental health. They differentiate three conceptual models. First, they discuss the impact of structura l neighbourhood characteristics, including socioeconomic status, ethnic composit io n, residential patterns, and family disruption. Second, they explain the relationship between neighbourhood disorder and mental health. Finally, they talk about the association between environmental stressors and mental health.

Wandersman & Nations explain structural characteristics such as the demograp hic characteristics of a population, which in this case are neighbourhood residents. These include, for example, the percentage of residents living in poverty, ethnic composition, and the percentage of families with high-risk characteristics (e.g. single-parent households ). They state that previous research follows the basic assumption that distressed neighbourhoods, that lack economic and social resources, are associated with more social problems that in turn influences mental health. Furthermore, the neighbourhood disorder model refers to social and physical incivilities. Under social incivilities, public drunkennes s, corner gangs, and street harassment are explained as possible factors influencing menta l health. Abandoned buildings, vandalism, and litter are examples of the physical incivilit ie s. Finally, environmental stressors are discussed as neighbourhood characteristics influenc i ng individual health. They argue noise pollution, crowding, and general pollution to be predictors of mental health.

Keeping the idea in mind that there is no clear uniform definition and operationalization of neighbourhood deprivation, previous studies did find effects of living in a deprived neighbourhood on health in general. Pickett & Pearl (2001), for example, reviewed 25 studies that researched the association between the status of the neighbour hood and health. Of these 25 studies, 23 reported a statistically significant association between (at least) one neighbourhood measure and health, controlling for individual characteristics. Some studies, however, suggest that the effect of neighbourhood deprivation on individual health is mainly the result of the concentration of people holding a low socioeconomic position in these

(16)

16

neighbourhoods (e.g. Browning and Cagney, 2003). Other scholars did find harmful effects of neighbourhood deprivation on individual health, taking individual socioeconomic characteristics into account (e.g. Jones & Duncan, 1995; Malmström, Johansson & Sundquis t, 2001; Fone & Dunstan, 2006). More recently, Poortinga and colleagues (2008) researched the link between neighbourhood deprivation and self-rated health. Their results suggest that there is a statistically significant, negative effect of neighbourhood deprivation on individual health3_.

However, this effect substantially reduced when controlling for individual socioeconomic status. They found individuals’ perceptions of the neighbourhood to also being an important indicator of individual health.

Poortinga, Dunstan and Fone (2008) furthermore argue that there is a well-know n problem regarding neighbourhood effect studies, namely that results are open to interpretat io n. There is a possibility that individual socioeconomic status drives the relation between neighbourhoods and individual health. However, they do suggest that there are also mult ip le examples of genuine neighbourhood effects, where living in a deprived urban area does har m one’s health. However, since individual characteristics sometimes do change the neighbourhood effects, this study will additionally control for some individual characterist ics (see subsection 3.5). For now, following (most) previous studies on neighbourhood deprivatio n and individual health, the following hypothesis will be tested:

H24_{: Living in less deprived neighbourhoods has a positive impact on residents’} wellbeing and mental health.

2.4. Neighbourhood perceptions on wellbeing and mental health

2.4.1. Perceived neighbourhood social cohesion & social capital

Besides neighbourhood effects, this study argues that also perceptions of the resident ia l environment influence individual wellbeing and mental health. In their research, Ettema & Schekkerman (2016) found that perceptions of the neighbourhood had a higher explanato r y power than the ‘factual’ neighbourhood- level characteristics. Leslie & Cerin (2008) also studied the relationship between perceptions of the local environment and mental health.

3_{In the study of Poortinga et al. (2008), self-rated health is indicated as individual health in general.}

Respondents were asked to rate their own health on a 5-point scale. Poortinga et al. made it a dichotomous variable with 1 representing fair/poor health and 0 good/very good/ excellent health.

4_{In order to more easily interpret the outcomes later, the hypothesis talks about less neighbourhood}

(17)

17

They focus on both the physical and social environment of the neighbourhood. They argue that neighbourhood perceptions affect the level of satisfaction with one’s residential area, which, in turn, may affect aspects of mental health (e.g. stress, depression, or anxiety). The perceived neighbourhood characteristics that will be discussed here are perceived social cohesion5_{, social capital, and the perceived physical environment. As explained earlie r,}

perceptions of social cohesion and social capital could act as mechanisms of ethnic divers it y. Here, both the direct and mediating effect of both concepts will be discussed.

Before turning to the earlier- mentioned theories (conflict theory vs. contact theory), a brief explanation of the concepts of social cohesion and social capital is necessary. Bolt & Van Kempen (2012), for example, explain social cohesion, in its most general meaning, as a kind of ‘glue’ which holds a ‘society’ together. Social cohesion includes various aspects, such as social solidarity, social control, social networks, a strong bond with the place one lives, and a feeling of belonging to each other through a common identity. Forrest & Kearns (2001), for example, define social cohesion as the ‘need for a shared sense of morality and common purpose, aspects of social control and order, the threat to social solidarity of income and wealth inequalities between people, groups and places, the level of social interac t io n with communities or families, and a sense of belonging to place’ (p. 2128). Lacking social cohesion could lead to extreme social inequality, low levels of place attachment, and low levels of social interaction within communities (ibid.). Additionally, Echeverría and colleagues (2008) described social cohesion as ‘the degree of connectedness and solidarity that exists among people living in defined geographic boundaries’ (p.854). In turn, individ ua l health could be influenced by neighbourhood social cohesion through its promoting role in adopting health- related behaviours (ibid.). Perceptions of neighbourhood social cohesion are thus individual evaluations of the extend of connectedness and solidarity they experience in the neighbourhood.

Social capital, on the other hand, is referred to as not solely to a set of social contacts, but also as the means individuals (or households) get out of these contacts (Bolt & Van Kempen, 2012). Putnam (2007) sees social capital as ‘social networks and the associated norms of reciprocity and trustworthiness’ (p. 137). Lancee & Dronkers (2011) argue that it is useful to distinguish between structural and cognitive social capital. Structural social

5_{Whereas some studies see neighbourhood social cohesion as a neighbourhood-level characteristic, this study}

chooses to refer to individual perceptions of neighbourhood social cohesion, following some other scholars on neighbourhood effects and health (e.g. Abada, Hou & Ram, 2007). The discussion of the concept of social cohesion will mostly be equal to earlier research on neighbourhood social cohesion, but the operationalization will be on individual-level perception of neighbourhood social cohesion.

(18)

18

capital implies the ‘wires’ in the network, that is the ‘frequency and intensity of the links that contribute to the exchange of resources’ (p. 599). According to Lancee & Dronkers, this also involves a behavioural component. Cognitive social capital, on the other hand, is referred to as the ‘nodes’ in one’s social network, including the attitudes and values (e.g. perceptions of support, reciprocity, and trust) that contribute to resource exchange. Here, however, social capital will be referred to as individual’s social contacts and the means they get out of these contacts.

As mentioned earlier, two often used and contradicting theories to explain the link between ethnic diversity and social capital/social cohesion (and in turn mental health), are the contact theory and conflict theory (Veldhuizen et al., 2015). Lancee & Dronkers (2011) discuss both theories. According to the contact theory, more (ethnic) divers ity will lead to more inter-ethnic tolerance and more social solidarity. The idea is that individuals of different ethnic groups tend to trust ‘others’ more if they have more contact with people ‘unlike themselves’. The idea is that then initial barriers between the ‘in-group’ and ‘out-group’ will be overcome. Although focusing on prejudice reduction, the first influentia l intergroup contact hypothesis was introduced by Allport in 1954 (Pettigrew, 1998). Allpor t discussed four conditions for optimal intergroup contact, namely equal status within the situation, intergroup cooperation, common goals, and support of authorities. Some critic is m exists on the first condition, equal status, since this is a rather vague definition and used in different ways. Still, most research support this idea. Furthermore, Allport argued that in order to have good intergroup contact, an active-goal oriented effort is needed. In this way, in order to achieve this goal, people have to cooperate. Therefore, intergroup cooperation is linked with the condition of common goals. Finally, it is argued that explicit social sanctio ns are necessary for intergroup contact to be more accepted.

On the other hand, the conflict theory implies that ethnic diversity strengthens distrus t between various ethnic groups and creates more ‘in-group’ solidarity. Meuleman, Davidov & Billiet (2009) talk about increasing negative ‘outgroup’ attitudes, that is, negative attitud es toward persons who are not part of one’s ‘in-group’. These group distinctions can be based on multiple factors, such as ethnicity, but also gender or age. The idea of the conflict theory is that privileges of one’s group are threatened by other groups (ibid.). In other words, negative attitudes toward other groups could be a result of competition of (scarce) goods between groups. These goods can refer to as well material goods (e.g. jobs or welfare-s ta te resources), as immaterial goods (e.g. status or power). This threat is influenced by a context of perceived intergroup competition. According to Meuleman and his colleagues (2009), in

(19)

19

previous literature on group conflict, minority group size and economic conditions are used to operationalize competitive conditions. The idea is that the bigger the minority group, the greater the perceived threat, since there would be more ethnic competitors for (scarce) goods then and more potential for mobilization. Also, the idea exists that goods are scarcer when economic conditions are worse. Putnam (2007) also argues that ethnic diversity leads to lower levels of trust between groups. Basically, his argumentation is that when a context is more ethnically diverse, there are more people ‘unlike you’ and less ‘like you’, resulting in fewer people with whom one can identify. Consequently, this results in less trust and fewer social contacts in this context.

The above theories clearly show opposite effects of ethnic diversity on perceived social cohesion and social capital. How then do both social cohesion and social capital in turn influence individual health? Although focusing on wellbeing only, Cramm et al. (2013) explain the effects of social capital and social cohesion on individual wellbeing. They found evidence that suggests that neighbourhood social capital and social cohesion are as well significantly as independently associated with wellbeing of elderly residents. They argue that both social capital and social cohesion in the neighbourhood result in higher provision of support to neighbours. This idea that neighbours take care of each other and watch over each other, which is visible in small favours (e.g. support in times of sickness, support in groceries, throwing away garbage), translate into better wellbeing outcomes for older adults. Focusing on mental health, Echeverría et al. (2008) hypothesized that social cohesion influences individual health through, for example, psychosocial processes. They argue that individuals could be provided with meaningful connections, mutual respect, and an increasing sense of purpose and meaning in life among residents, which all positively affects mental health. Their findings support their hypothesis, indicating that (self-reported) social cohesive neighbourhoods positively influences mental health.

Abada and colleagues explained how perceived neighbourhood social cohesion can affect (general) health outcomes (2007). They argued that individuals living in specific neighbourhoods are influenced by who they socialize with in terms of values, attitudes, and aspirations. They moreover argued that a lack of social integration in an/(a) (resident ia l) environment may contribute to feelings of hopelessness, which increases the risk of depressive symptoms.

In short, perceived social cohesion and social capital in the neighbourhood are argued to have direct implications for wellbeing and mental health. Moreover, both concepts can serve as mediating mechanism for the effect of neighbourhood ethnic diversity on wellbe i ng

(20)

20

and mental health. While in the latter case both conflict and contact theory could be follow ed, I again choose to go with the contact theory, assuming that ethnic diversity in a neighbourhood could lead to better perceptions of social cohesion and more social capital which in turn affects wellbeing and mental health positively. Again, I assume that ethnic concentrations in Dutch neighbourhoods are not at such high levels that other ethnic groups feel their privileges will be threatened, which could be more likely in the US where ethnic concentrations in neighbourhoods are (much) higher possibly leading to a stronger sense of threat. Considering this, the following hypotheses will be tested (also: see figure 1):

H3: More perceived neighbourhood social cohesion and social capital leads to higher wellbeing and better mental health of residents.

H4: The positive effect of neighbourhood ethnic diversity on wellbeing and mental health of residents is mediated by perceived neighbourhood social cohesion and social capital.

Figure 1. Predicted mediation mechanism of perceived social cohesion and social capital for the effect of ethnic diversity on wellbeing and mental health.

(21)

21

2.4.2. Perceived neighbourhood physical environment

According to Leslie & Cerin (2008), besides the neighbourhood’s social environment, the physical environment is also important in explaining individual health outcomes6_{. Other}

scholars also emphasized and researched this association (e.g. Ettema & Schekkerman, 2016; Gidlow, Cochrane, Davey, Smith & Fairburn, 2010). Galster (2010), for example, provides an overview of various mechanisms of neighbourhood characteristics, including three environmental mechanisms: exposure to violence, physical surroundings, and toxic exposure.

First, the relation between exposure to violence and mental health was explained. The idea is that if individuals feel that their property or person is in danger this may have an impact on their wellbeing. Second, physical surroundings refer to the idea that deteriorated conditio ns in the residential area can have negative psychological effects for its residents. These physical surroundings include many conditions, such as area structures, public infrastructure, litter, and graffiti. Bell, Greene, Fisher & Baum (1996), for example, argue that noise (created by public infrastructure) can create psychological distress due to ‘environmental overload’. Finally, toxic exposure implies the exposure to high levels of soil-, air-, and/or water-borne pollution that it is unhealthy for residents.

Leslie & Cerin (2008) state in their article that perceptions of the neighbourhood can influence the level of satisfaction with living in a neighbourhood, which in turn influe nces multiple aspects of mental health. Leslie & Cerin argue the following perceived neighbour hood physical characteristics to be related to self-reported health: crime, green space, access to amenities, traffic load and safety, and recreation areas (2008). Referring to previous studies, they argue that perceived green spaces in the neighbourhood to be beneficial to mental health as well as access to amenities. Perceived characteristics that negatively influence self-reported health are argued to be perceptions of crime and traffic load. A person who perceives his or her neighbourhood as dangerous could create emotional distress. Perceptions of neighbour hood physical disorder, for example, could lead to ideas that neighbourhoods are dangerous. Examples of visible physical disorder are graffiti, litter, and vandalism. In the long run, they argue, signs of crime in the neighbourhood could lead to anxiety and depression.

According to Stansfeld, Brown & Haines (2000), traffic load is associated with noise pollution. Stansfeld and colleagues studied the effects of noise on individual health. They argued that the impact of noise is the strongest for outcomes such as ‘quality of life’ rather than

6_{Again, some scholars would argue the neighbourhood physical environment to be neighbourhood -level}

characteristic. However, this study chooses to follow the idea of Leslie & Cerin (2008) who argue the importance of perceived neighbourhood social and physical characteristics.

(22)

22

‘illness’. Furthermore, they provide evidence that psychological wellbeing of individuals is reduced in areas exposed to high traffic noise. They state noise pollution may lead to depressiveness and nervousness.

Following the idea of Leslie & Cerin, this study argues that positive perceptions of the neighbourhood physical environment (i.e. green spaces, access to amenities, lack of crime, and lack of traffic load) are beneficial for wellbeing and mental health. This leads to the follow ing hypothesis:

H5: Residents who perceive their neighbourhood’s physical environmental more positively will have a higher wellbeing and better mental health.

2.5. Differences between native Dutch and ethnic minorities

As mentioned earlier, it is argued that life satisfaction is an important outcome measure, since it is a probable source for inequalities between ethnic groups (Knies, Nandi & Platt, 2016). Moreover, many scholars found that life satisfaction is lower among ethnic minorities than the among the ‘native’ population (e.g. Bobowik et al., 2015; de Vroome & Hooghe, 2014; Koczan, 2013). Also within ethnic groups, differences in life satisfaction are visible (e.g. Amit, 2010). Focusing on Great-Britain, Bécares et al. (2012) also found that the association between individual health (poor self-rated health) and neighbourhood effects varied between ethnic groups and that this association was the weakest for white British people. This subsection aims to find explanations for differences in neighbourhood effects on wellbeing and health.

Karlsen, Nazroo & Stephenson (2002) already looked at ethnic inequalities in health, focusing on the social and environmental factors that could influence this. They argued that previous research has focused mainly on biological, genetic, cultural, and behavioural factors and not that much on social/contextual factors. Their findings suggest that ethnic minorit ies, despite them residing in more deprived areas, perceive the amenities in their residential area more positively than the ‘native’ British people. Bécares et al. (2012) refer to this finding and suggest that this could, for example, reflect ethnic differences in perceptions of objective contexts, different expectations of the context and its amenities, or investme nt by ethnic minorities in their neighbourhood facilities instead of actual differences in neighbour hood facilities and environment. Their findings confirm this idea, indicating that the association between neighbourhood effects and individual health varies between ethnic groups, with the weakest association for ‘native’ British people.

(23)

23

Although focusing on physical health among residents in Amsterdam, Agyemang, Van Hooijdonk, Wendel-Vos, Lindeman, Stronks, & Droomers (2007), also discuss ethnic differences in regard to perceptions of the (residential) environment. They state that these perceptions can differ due to, for example, differences in culture, socioeconomic position, migration history, and/or language. They furthermore argue that many of the ethnic minor it y populations live in disadvantaged neighbourhoods, which is a disadvantage for their health. This study, focusing on Rotterdam, will include both ‘disadvantaged’ and ‘advantaged’ neighbourhoods. Even so, following Agyemang et al., (2007), assuming that ethnic minorit ies are overrepresented in more ‘disadvantaged’ neighbourhoods, it could be assumed that neighbourhood effects on wellbeing and mental health are stronger for ethnic minorities than for native Dutch. Moreover, this study follows the findings of Karlsen and colleagues (2002), assuming that the effect of neighbourhood perceptions on wellbeing and mental health will be stronger for ethnic minorities than for native Dutch. Considering this, the following hypothesis will be tested:

H6: Neighbourhood effects and the effect of neighbourhood perceptions on wellbeing and mental health are stronger for ethnic minorities than for native Dutch.

(24)

24

3. Data & Methods

3.1. Data

This study will combine individual- and neighbourhood- level data, creating a multilevel design for data analysis. Regarding individual- level data, this study will make use of the ‘Gezondheidsmonitor 2012’ (i.e. Health Monitor 2012) from the municipality of Rotterdam. The Health monitor includes questions on individual characteristics, individual health and wellbeing, as well as individual perceptions of the neighbourhood. Initially, 36,730 persons residing in Rotterdam were asked to participate in the survey. Only 38,4% responded, which is the lowest response rate in comparison with the other 18 municipalities where the mean response rate was 49%7_{. This resulted in a sample size of 14,113 respondents for the Health}

monitor 2012 in Rotterdam, of whom 6,352 are men and 7,761 are women. The age of respondents ranges from 17 up to and including 100 years. The survey data is corrected by using big weighting factors from the Dutch Statistics Bureau (Schouten, 2013). This way, the sample reflects the Dutch population.

The following subsections will discuss the operationalization of the concepts used in this study. First, the outcome measures (individual wellbeing and mental health) will be elaborated, then the predictors on both the neighbourhood- level and the individual-le ve l. Subsequently, the control variables that will be used in the analyses are discussed. Finally, the analytical strategy of this study will be explained.

3.2. Wellbeing & mental health

First, the dependent variables ‘wellbeing’ and ‘mental health’ are both individual-le ve l measures. As explained earlier, wellbeing is conceptualized in this study as the individ ua l evaluation of the overall quality of their lives. In other words, it is concerned with the overall satisfaction of life. Following other scholars, this study uses life satisfaction as proxy for wellbeing (e.g. Bobowik et al., 2015; Ettema & Schekkerman, 2016). The 2012 Health monitor includes the following question: ‘How (dis)satisfied are you, in general, with your current life?’ Respondents answered this question by providing a mark ranging from 1 ‘dissatisfied’ to 10 ‘satisfied’. Psychological distress will be used here as proxy for mental health, again following

7_{Unfortunately, it is not clear why the response rate in the city of Rotterdam was this low. In 2008, the}

response rate was 47,2%. It appears the response rates were higher when age increases. Also, in every age category, women responded more than men. Response rates among native Dutch were higher than each ethnic minority group. The survey was distributed in other languages too (Turkish, English, Moroccan Arabic, and Cape Verdean). Therefore, language barriers cannot serve as explanation for a lower response rate.

(25)

25

other scholars (e.g. Williams et al., 1997). The Health monitor included ten questions: ‘How often did you feel 1) very tired without a reason? 2) nervous? 3) so nervous you could not get calm again? 4) helpless? 5) restless or fidgety? 6) so restless you could not sit still anymore? 7) sad or depressed? 8) like everything takes much effort? 9) so sad that nothing could cheer you up? 10) inferior? These questions refer specifically to the previous four weeks and the response categories range from ‘always’ to ‘never’. Before creating a scale, a principal component analysis (pca) was executed. Results show one component with an eigenvalue >1. Following Samuels (2016), the factor loadings here should be at least above 0.3. In this case, all factor loadings are bigger than 0.3. A Cronbach’s alpha was measured to check the reliability of the scale. An often-used rule of thumb in regard to the value of the Cronbach’s alpha is: “_ > .9 – Excellent, _ > .8 – Good, _ > .7 – Acceptable, _ > .6 – Questionable, _ > .5 – Poor, and _ < .5 – Unacceptable (e.g. George & Mallery, 2003). Here, the Cronbach’s alpha is .92, which is, according to the above categorization defined as ‘excellent’. Finally, to check the valid it y, correlations between the items were measured. According to Hinkle, Wiersma & Jurs (2003), the classification of Pearson correlation coefficients goes as follows: .00 to .03 implies little if any correlation, .30 to .50 implies low correlation, .50 to .70 entails moderate correlation, .70 to .90 implies a high correlation, and .90 to 1.00 implies very high correlation. See appendix 1 for the outputs of the pca, Cronbach’s alpha, and the Pearson correlations. A scale is constructed by taking the means of the items8_{. Cases were coded as missing values if respondents had}

missing values on 4 or more items.

3.3. Neighbourhood effects

Several neighbourhood- level variables will be merged to the individual- level data from the Health monitor 2012. The Health monitor already listed neighbourhoods in Rotterdam, differentiating 59 neighbourhoods. This neighbourhood distribution (see appendix 2) is the foundation for the neighbourhood characteristics.

3.3.1 Ethnic diversity

An ethnic diversity index is used as predictor for individual wellbeing and mental health. Data on residents’ ethnicity in Rotterdam neighbourhoods is collected from the ‘Rotterdam Buurtmonitor’ website (n.d.). From this ethnicity data of 2012, the Herfindahl index of ethnic diversity was calculated for each neighbourhood. Following the formula of Lancee & Dronkers (2008), the subsequent formula is used:

8_{This study chose to take the mean of the items instead of letting PCA create a scale, since in this way cases}

(26)

26

Dc = -((fraction ethnic group 1)2 + (fraction of ethnic group 2)2 +….+ (fraction of

ethnic group n)2 )

Here, Dc is the level of neighbourhood ethnic diversity. The ethnic groups that are included are

people from Turkish, Moroccan, Antillean, Surinamese, and Cape Verdean descent, along with other non-western immigrants, Western immigrants, other Europeans, and native Dutch. Theoretically, the index ranges from -1 to 0, where ‘-1’ indicates no diversity at all (i.e. there is only one ethnicity in the neighbourhood), and ‘0’ implies total ethnic diversity (i.e. every resident in the neighbourhood has a different ethnicity). See table 1 for further information.

3.3.2. Neighbourhood deprivation

Neighbourhood deprivation was explained by Verhaeghe & Tampubolon (2012) as the relatively low physical (e.g. graffiti), social (e.g. unemployment), and economical positio n (e.g. income) of a neighbourhood. Some scholars used the Index of Multiple Depriva t io n [IMD], which covers neighbourhood deprivation in various domains, namely: income, health, employment, housing, education, crime, and living environment (e.g. Verhaeghe & Tampubolon, 2012; Lang et al., 2008). Due to limits in data gathering, this study includes : neighbourhood income, physical index, and pollution & hassle.

Data on average disposable household income in 2012 was provided to me by the municipality of Rotterdam (Gemeente Rotterdam, 2015). Most data on neighbourhood income was transferred easily. However, some neighbourhoods in the health monitor are combined9_.

Therefore, the average income was calculated for these neighbourhoods by first, multipl ying average neighbourhood income with the number of households in that neighbourhood, then by summing up these outcomes, and dividing them by the total number of households of these neighbourhoods. For some neighbourhoods (Nieuw-Mathenesse, Blijdorpse Polder, Noord Kethel, Kralingse-bos, and Rijnpoort) there was no information on average disposable household income. In this study, the variable ranges from the least average disposable household income to the most.

9_{The combined neigbhourhoods in the Health monitor 2012 are: Stadsdriekhoek + C.S. Kwartier; Cool + Nieuwe}

Werk + Dijkzigt; Nieuw Mathenesse + Oud Mathenesse + Witte Dorp; Schieven + Zestienhoven + Overschie + Landzicht + Spaanse Polder; Blijdorp + Blijdorpse Polder; Terbregge + Molenlaankwartier; Nieuw Crooswijk + Oud Crooswijk; Kralingen-Oost + Kralingse-bos; Kop van Zuid + Kop van Zuid-Entrepot; ’s-Gravenland + Kralingseveer; Charlois Zuidrand + Zuidplein + Zuiderpark; Oud-Charlois + Heijplaat; Strand en Duin + Dorp + Rijnpoort (see appendix 2).

(27)

27

The physical Index is one out of three indexes of ‘Wijkprofiel Rotterdam’. Besides the physical index, there are also the safety index and social index. Unfortunately, it was not possible to get data on the safety index. The scores of these indexes are based on feasible facts and numbers, and on the experiences of Rotterdam’s residents. The physical index entails scores on real estate, public spaces, facilities, and environment in neighbourhoods. The average score ranges from 0 to 200, where 0 indicates a ‘bad’ physical neighbourhood environment and 200 a good one. Again, for some neighbourhoods, a mean had to be calculated. There were missing scores for Charlois Zuidrand (therefore, the mean was calculated of Wielewaal, Zuidplein & Zuiderpark), Nieuw-Mathenesse (therefore, the mean was calculated of Oud Mathenesse & Witte dorp), and Landzicht and Spaanse polder (therefore, the mean was calculated for Noord-Kethel, Overschie, Schieveen & Zestienhoven). Moreover, there was a missing score for the neighbourhood Rozenburg. This neighbourhood, therefore, has a missing value on the physical index. See table 1 for the descriptive statistics of this index for this study sample.

The last variable that will be used as measurement for neighbourhood deprivation is the score on pollution and hassle. This is an independent score within the social index. I chose to take this score and not the overall score of the social index (as I have done with the physical index), since otherwise there are some overlapping themes with the physical index. I chose pollution and hassle, since this touches upon Index of Multiple Deprivation that focuses on the domains: income, health, employment, housing, education, crime, and living environment, as elaborated on earlier. The score on pollution and hassle indicates a mark that ranges from 0 to 10, with ‘0’ implying a lot of pollution and hassle and ‘10’ (almost) no pollution and hassle. Scores are missing for the neighbourhoods: Blijdorpse Polder (therefore, the mark for Blijdorp is used for both neighbourhoods), and Charlois Zuidrand (therefore, the mean is calculated for Zuiderpark, Zuidplein and Wielewaal).

3.4. Neighbourhood perceptions

The predictors on individual- level entail individual perceptions of one’s neighbourhood. As explained earlier, these are considered to be individual evaluations of the residential area, and are based on personal experiences and preferences of one’s neighbourhood (Ettema & Schekkerman, 2016). Such questions are available in the Health monitor (2012). As explained in the previous section, several mechanisms are sought to be studied here. Social capital and perceived social cohesion in the neighbourhood are studied here as possible mediation effects for ethnic neighbourhood composition on individual wellbeing and mental health. Furthermore,