SOCIAL CAPITAL AND THE BUILT ENVIRONMENT : A CASE STUDY OF PUBLIC HOUSING IN SAGINAW, MI CHIGAN By Zachary Vega A THESIS Submitted to Michigan State University i n partial fulfillment of the requirements for the degree of Urban and Regional Planning - Master in Urban and Regional Planning 2019 ABSTRACT SOCIAL CAPITAL AND THE BUILT ENVIRONMENT : A CASE STUDY OF PUBLIC HOUSING IN SAGINAW, MICHIGAN By Zachary Vega This study builds off of the literature regarding the relativ ely new and popular theory of social capital to understand its relationship with the built environment. Social capital is best described as the accumulation of the perceived benefits that develop through interpersonal relationships and social networks. In measures comprise the dependent variables. The independent variables include landscape factors such as housi ng type, median home values, Walk Score, vacancy, housing density, road conditions, blight/abandonment, junk piles/illegal dumping, street segment connectivity, street segment integration. The relationships between these landscape factors and social capita l variables are discussed anecdotally and later tested using a linear regression model. This study finds no significant relationships between landscape factors and measures of social capital, although demographic controls such as access to a personal vehic le, education, employment status and income are found to h ave a significant relationship. However, regression analyses in this study are significant and indicate that landscape variables do account for around 20 percent of the variation in levels of social capital, though the effect of individual factors remains unclear. Therefore, the decisions of urban planners and landscape architects appear to contribute to the social connections of a place and future research should continue to estimate which factors p lay the greatest roles. iii Copyright by ZACHARY VEGA 2019 iv ACKNOWLEDGEMENTS I would like to thank all of th e people who have helped me complete this degree program. First, thank you to my adviser, Dr. Noah Durst , for being the trues t definition of a teacher through his unwavering patience, willingness to challenge me and dedication to my success . To Dr. Linda Nubani and Dr. Trish Machemer for their contributions to my studies and especially to this research. To my mentors Holly Madil l and Wayne Beyea for leading by example and for dedicating so much time and effort towards my development . To my cohort for their consistent grit and for making the most difficult challenges enjoyable. Finally, thank you to my mother for supporting me in all of my endeavors and for teaching me the importance of working hard and being kind. v TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ...................... vii LIST O F FIGURES ................................ ................................ ................................ ................... viii Chapter 1. Introduction ................................ ................................ ................................ ................ 1 Chapter 2. Literature Review ................................ ................................ ................................ ...... 4 2.1 Pop ularity of the Term ................................ ................................ ................................ ....... 4 2.2 Main Divide: Defining and Measuring Social Capital ................................ ..................... 5 2.3 Cause, Effect, or Response ................................ ................................ ................................ . 7 2.4 Focus on Positive Outcomes ................................ ................................ ............................... 8 2.5 Other Contributing Factors Considered ................................ ................................ ........... 9 2.6 Using a Combination of M easures ................................ ................................ ................... 10 2.7 How Social Capital is Formed ................................ ................................ .......................... 11 2.8 The Effects of Social Capital ................................ ................................ ............................ 15 2.9 Geography and the Built Environment ................................ ................................ ........... 15 2.10 Conclusion ................................ ................................ ................................ ........................ 19 Chapter 3. Case Study of Saginaw, MI ................................ ................................ ..................... 21 3.1 Introduction ................................ ................................ ................................ ....................... 21 3.2 Socio - economic, Demographic, and Housing Profile and Population Projection ....... 21 3.3 Temp oral Analysis ................................ ................................ ................................ ............. 23 ................................ ................................ ....................... 26 Chapter 4. Methodology ................................ ................................ ................................ ............. 28 4.1 Research Framework ................................ ................................ ................................ ........ 28 4.2 Survey Design ................................ ................................ ................................ .................... 31 4.3 Sample Selection and Survey Implementation ................................ ............................... 34 4.4 Analytical Method ................................ ................................ ................................ ............. 37 4.5 Measurement of Trust and Neighborhood Cohesion ................................ ..................... 39 4.6 Measurement of Landscape Factors ................................ ................................ ................ 41 4.7 Control Variables ................................ ................................ ................................ .............. 46 vi 4.8 Study Limitations ................................ ................................ ................................ .............. 49 Chapter 5. Analysis ................................ ................................ ................................ ..................... 51 5.1 Survey Results ................................ ................................ ................................ ................... 51 5.2 Trust and Landscape Factors ................................ ................................ ........................... 53 5.3 Neighborhood Cohesion a nd Landscape Factors ................................ ........................... 58 5.4 Linear Regression Models ................................ ................................ ................................ 65 Chapter 6. Discussion and Conclusion ................................ ................................ ...................... 71 APPENDIX ................................ ................................ ................................ ................................ .. 76 B IBLIOGRAPHY ................................ ................................ ................................ ....................... 84 vii LIST OF TABLES Table 1: Demographic Comparisons for Sag inaw, MI ................................ ............................ 22 Table 2: Housing Characteristics for Saginaw, MI ................................ ................................ . 23 Table 3: Survey Layout ................................ ................................ ................................ .............. 33 Table 4: Survey Mailings by Household Type ................................ ................................ ......... 37 Table 5: General Trust and Landscape Factors ................................ ................................ ...... 54 Table 6: Neighbor Trust and Lands cape Factors ................................ ................................ .... 56 Table 7: Housing Type and Neighborhood Cohesion ................................ .............................. 59 Table 8: Connectivity, Integration and Neighborhood Cohesion Mean Value s ................... 61 Table 9: Road Conditions, Blight, Junk and Neighborhood Cohesion ................................ .. 63 Table 10: Neighborhood Spatial Factors and Cohesion ................................ .......................... 64 Table 11: Linear Regression Results ................................ ................................ ......................... 69 viii LIST OF FIGURES ial Capital ................................ ......... 8 Figure 2: Population Change with 2 - year Moving Average ................................ ................... 24 Figure 3: Age Group Distribution ................................ ................................ ............................. 25 Figure 4: Theoretical Framework ................................ ................................ ............................. 29 Figure 5:General Locations of Public Housing Unit Respondents ................................ ........ 35 Figure 6 : Connectivity Output Generated by Depthmap ................................ ....................... 46 1 Ch apter 1. Introduction The social sciences field has paid particularly close attention to the concept of social capital as a measure of community well - being after s famous book Bowling Alone : The Collapse and Revival of American Community ( Putnam, 2000) popularized the term and introduced its suggested influence into the American psyche. Analyses of social capital primarily examine the ways in which social connecti ons act as a source of capital to leverage other resources, and that in many instances, these positive social connections can actually be a resource in themselves (Bourdieu, 1986; Portes & Landolt, 2000) . Past studies , discussed in Ch apter 2, provide evide nce that locations with higher levels of social capital thrive in comparison to those places with lower levels: they tend to be healthier (Cattell, 2001) , more economically viable (Bourdieu, 1986; Wacquant & Wilson, 1989) and hold a greater quality of life overall (Putnam, 2000) . However, the research on the link between social capital and th e built environment is limited and t absent from the literature since the 2000s . The field of Urban and Regional Planning is dedicate d to creating urban landscapes that promote the health, safety and quality of life for people. Therefore, it seems imperative that planning academics should pay attention to how the built environment affects social ties, the quality of which has implicatio ns for everything from mental health to the ability to attract a talented workforce. This research project examined the relationship between the built environment and measures of social capital in Saginaw, MI. Saginaw is a city suffering from high crime r ates, high unemployment and decaying structures (these data are discussed in chapter 3) . To fewest social connections, or those with the lowest social capital, this st udy focuses on residents 2 in public housing in Saginaw. The purpose was to investigate whether the siting of public housing sites may contribute to access to higher or lower levels of social capital for the residents. Therefore, t h e central research questio n in this study is the following : Are levels of trust for public housing residents in the city related to landscape factors such as building density, walkability , the presence of blighted properties , and road conditions ? The broader goal of this research is to better inform where municipalities across the country site their public housing units. As previously mentioned and detailed further in the literature review chapter, s ocial networks are vital for an individual to gain access to job prospects, feel p art of their community and become civically engaged. T his study attempts to determine if geography factors play role in the success or failure of public housing siting as it relates to measures of social capital . The following analysis of social capital an d the built environment proceeds in four chapters . First, there is a review of the literature on social capital as a theory and empirical construct , including contradictions and critiques of the concept . This review then frames the methodology used in this research study, which includes the survey instrument, data collection and statistical methods chosen for analysis. Data comes from a mail - in survey instrument sent to housing units managed by the Saginaw Housing Commission. Data that could not be measured - year estimates. All of the landscape factor data, as well as demographic constants, made up the independent variables for this study, with trust in neighbors and neighborhoo d cohesion acting as the dependent variables. Using a linear regression , the relationships of these landscape factors and the two measures of social capital are measured . This research framework is detailed in the methodology chapter and findings are discu ssed in the analysis chapter. The null hypothesis that differences in social capital levels for public housing residents in Saginaw were not related to 3 landscape factors could not be rejected by this study, though these results may have been the result of research design errors discussed in the conclusion. This report contributes to socia l sciences literature in two ways: First, the research here indicates that landscape likely does not relate to neighborhood trust levels and neighborhood cohesion as much a s other factors not measured in this study. Second, this study could be replicated using a revised design framework with the possibility of finding relationships between landscape and social capital as other studies have done ( Rahimi, S., Martin, M. J. R., Obeysekere, E., Hellmann, D., Liu, X., & Andris, C. ( 2017) . Though the findings herein may promote more scrutiny of the research designs that have found a connection between these factors. Ultimately, these relationships remain unclear. 4 Ch apter 2. Liter ature Review Social capital, when applied to research studies, is used for a wide range of purposes and is measured very differently depending on circumstances . This makes it imperative for researchers to consider how social capital is conceptualized and m easured in order to discern whether or not it can be effectively and appropriately utilized in social science research . This chapter main divide in the literature. The r eview then addresses three main contentions with social literature pertaining to how social capital is formed, what its effects are, and how social capital studies have become more prevalent in analyses of the built environment are included. It is important to understand past w ork involving social capital to, first, be able to develop a study of social capital that can be replicated, and, second, to avoid many of th e shortcomings of completed studies attributable to misconceptions of what the term entails. 2.1 Popularity of the Term Social capital has its origins in the field of sociology but has become popular in other social sciences fields as well as health studie s (Portes, 2000). Jane Jacobs (1961), Pierre Bourdieu and Jean - Claude Passeron (1970), and Glenn Loury (1977) are credited with coining the term, while James Coleman (1988), Robert Putnam (2000), and Alejandro Portes (Portes & Sensenbrenner, 1993) have wor ked to cultivate its theoretical usage in more recent years Bowling Alone: The Collapse and Revival of American Community (2000), has been the most influential text in popularizing social capi tal in a diverse range of academic fields (Macinko & Starfield, 2001). Capital refers to something that is created through labor, which is then used by groups to achieve more of itself or 5 other types of capital. Social capital, in its simplest form, refers to group membership that is disinterested in that its function is not directly or explicitly for economic gain (Bourdieu, 1986). In other words, people form social connections because they perceive group membership to be beneficial in some way; however, u benefits are not necessarily monetary. 2.2 Main Divide: Defining and Measuring Social Capital Social capital is a theory whose theoretical framework has not been widely agreed upon (Muntaner, Lynch, & Smith, 2000). The key problem that plagues the social capital literature is that there is a strong divide in whether social capital is a good held by communities or if it is one held by individuals (Macinko & Starfield, 2001). Individual level assessmen ts are typically utilized when considering economic, social, and health benefits for someone in a given community. The concept as a collective resource is used by theorists like Robert Putnam (2000) to discuss larger societal issues like civic participatio n, democracy, and the effectiveness of political institutions. Currently, there are four distinct measure s of social capital: macro, neighborhood, individual - level behaviors, and individual - level attitudes (Macinko & Starfield, 2001). The individual measur es preceded the community level ones chronologically (Portes, 2000). Scholars are more critical of the use of communi ty level measures (Portes, 2000; Portes & Landolt, 2000; Portes & Sensenbrenner, 1993), which were popularized more recently by Robert Put Bowling Alone (2000). Putnam views social capital as a community asset defined by trust, reciprocity, and civic participation. Proponents of the community level measure argue that social capital is ecological, meaning that there is an environmen tal element that must be considered (Bothwell, Gindroz, & Lang, 1998; Lochner, Kawachi, & Kennedy, 1999; Leyden, 6 2003; Lindstrom, Merlo, & Ostergren, 2002; Rahimi, Martin, Obeysekere, Hellmann, Liu, & Andris, 2017). However, different levels of aggregated data will result in differing contributors to the presence or absence of social capital (Macinko & Starfield, 2001). For instance, a local measure, such as a neighborhood, would be influenced by the everyday contact between residents, while a country - wide approach would likely be more dependent on culture and policy (Lochner et al., 1999). The problem, however, is that researchers often do not explain why they use certain measures and ignore others when determining levels of social capital (Macinko & Starfi eld, 2001). Additionally, researchers favoring the collective community interpretation have never formally defined the idea, meaning that its usage in recent years has made it hard to discern what is meant when referring to social capital (Portes, 2000). L ochner et al. (1999) attempt to bridge this gap in the literature by making social capital at the community level easier to measure; they do so by defining the term as the existence of collective efficacy, psychological sense of community, neighborhood coh esion, and community competence. Though these four characteristics of social capital still do not create a shared definition amongst scholars in all fields, they do include overlapping themes such as civic engagement, mores defined by support and exchange, and trust between group members, which could inform uniform measures (Lochner et al., 1999). But even these terms are ambiguous when applied to a study, as researchers typically do not specify what they are attempting to measure. For example, trust, whose existence is a key indicator of social capital, can take on many different meanings such as trust in family, trust in institutions, or trust in neighbors (Macinko & Starfield, 2001). On the other side of the literature divide, Portes and Landolt (2000) po int out that the most accepted and used definition of social capital is the capacity of an individual to attain 7 resources due to their connection to a particular social network. One benefit of viewing social capital from an individual perspective is that s for associating with a network, or what resources that person perceives as desirable (Portes & networks, as oppos as the number of people in the network and the amount of resource s they control (Bourdieu, 1986), a s opposed to viewing social capital as a material held by the community as well as an ecological trait (Putnam, 2000). O thers like Bourdieu (1986) argue that social capital is a means to an end, not an end in itself. This leads to another argument against community level assessments, namely that these researchers fail to define socia l capital as a tool, a result, or an individual response (Macinko & Starfield, 2001). Macinko and Starfield (2001) point out that Portes (2000) , who studies social capital at the individual or family level, is one of the few researchers who is able to clea rly describe and justify his definition of social capital, and thus, his measurements . 2. 3 Cause, Effect, or Response It is unclear whether social capital is a means or an end; in other words, whether it is a resource that can be used to obtain something o r if it itself is the product of forming a social network (Woolcock, 1998). In sociology, especially, social capital can hold three separate - mediated benefits, and a source of resources mediated by non - social capital as a means to an end, in which the outcomes of its existence can be positive or negative depending on how society perceives that end (de Souza Briggs, 1998). This view woul d suggest that the outcomes of social capital as a resource are most important. Others argue that 8 researchers using social capital theory should pay more attention to what causes social capital as opposed to what actually results from high levels of social capital (Woolcock, 1998). In both instances, however, social capital is seen as a middle actor. Portes and Sensenbrenner (1993) understand social capital similarly: social processes lead to built or collected social capital; social capital is therefore th being connected, or having high social capital, can be positive or negative. discussed in chapter 4. Figure 1 2.4 Focus on Positive Outcomes Some argue that studies of social capital, especially at the community level, tend to only concentrate on its positive outcomes (McMillan & Chavis, 1986; Muntaner et al., 2000; Portes, 2000; Portes & Sensenbrenner, 1993). There are various drawbacks to social capital (Portes, 2000; Portes & Landolt, 2000; Woolcock, 1998). For example, while a strong sense of c ommunity is often perceived favorably, this outcome could also create forms of isolation for those viewed as outsiders (McMillan & Chavis, 1986). Additionally, as previously mentioned, 9 high social capital can prevent economic development in a community by fostering an attitude of intragroup trust and a mistrust of outsiders; it can also create high demands on group members, limit their individual mobility, and instigate free - riding (Portes & Landolt, 2000). Social capital can also promote, not reduce, level s of discrimination as membership and access to resources deter outside interaction; in other words, a group can have too little and too much social capital (Woolcock, 1998). 2.5 Other C ontributing F actors C onsidered ing a community may be overstated, as other factors such as race, class or gender more than likely play a role in determining a person or (Muntaner et al., 2000). For instance, high levels of civic participation are th ought to be a component of high social capital (Putnam, 2000). By stating that high levels of civic participation lead to better government there tends to be little consideration for alternative explanations for effective political institutions. By not con sidering external to differentiate correlation from causation (Portes & Landolt, 2000). For example, the success of an immigrant student from culture A as o pposed to a student from culture B may have more to she belongs (Portes, 2000). For this reason, some argue that researchers need to consider the role that th e direct family has on the individual youth as it pertains to social capital (de Souza Briggs, 1998). Additionally, in order to measure social capital, one may need to consider how political and economic conditions in the geographic area in question influe nce social capital (Goode & 10 governments are also believed t o have strong impacts on social capital (Woolcock, 1998). Besides institutions, local economic and soc ial conditions such as rates of vacancy and homeownership, housing value s, and diversity in social amenity types all have a correlation with levels of trust in a community (Rahimi et al., 2017). Aside from the built environment, Cattell (2001) argues t hat community involvement is not enough for strong social connections to form in the modern neighborhood; work opportunities are the missing link, where bonds and networks can form to promote more social capital building and better community health. Cattel l (2001) also finds that it is necessary to evaluate amenities, housing, opportunities for social interaction and overall reputation when considering its ability to foster higher social capital . Income inequality h as also been shown to have a strong inverse relationship with civic participation (Kawachi, Kennedy, Lochner, & Prothrow - Smith, 1997). Because there is little agreement on what social capital truly is, researchers typically use differing empirical construc ts depending on what question they are trying to answer . 2.6 Using a Combination of Measures Although social capital can either be considered an individual or community asset, and thus there are important distinctions between the two constructs, they are n ot necessarily mutually exclusive (Portes, 2000). For example, Lindstrom et al. (2002) find that, when controlling for individual characteristics, differences in levels of social participation still vary in different neighborhoods . Therefore research desig ns that measure social capital at the community or group level are justified because the result in that study is that individual level measures do not completely account for the variance in social participation . H owever, the study also finds that individua l measures such as e ducation level and occupation status reduce this variance somewhat and are therefore, necessary for analysis. Thus, Lindstrom et al. (2002) indicate that social capital 11 studies should not choose either individual or community measures, but should include both for a clearer understanding of why social capital levels vary. As research using social capital has developed, studies have f ou nd that social capital and its benefits reinforce one another. For example, Brehm and Rahn (1997) find th at civic participation is more likely to increase trust than trust is to increase participation; however, trust and participation form a cyclical sequence: participation fosters more trust and more trust then encourages more participation. Ultimately, Breh m and Rahn (1997) demonstrate that social capital can be measured at both community and individual levels, since these forms of social capital, often perceived as separate, actually function together. 2.7 How Social Capital is Formed Much debate surrounds not only how social capital is measured, but as a consequence, how it is developed. Bourdieu (1986) identifies three types of capital: economic, cultural, and social. According to Bourdieu, c ultural and social capital are derived from economic capital, an d are merely a conversion of that original form. They can later be converted back into economic of constant exchanges, where the acquisition of one type mea ns the surrender of some amount of another. In other words, people form groups to exchange their social connectedness for material benefits. According to Putnam (2000), t here are two main types of connections: bonding social capi tal and bridging social ca pital . These can also be considered what Xavier de Souza Briggs (1998) terms social support and social leverage. The first, bonding or social support, refers to relationships that foster personal well - being and can directly contribute to better health (Kaw achi et al., 1997; Sabin, 1993). The latter, bridging or social leverage, has more to do with social 12 expose him or her to better job opportunities (Boxman, De Gra af, & Flap, 1991). for the individual to have in order to deal with not entirely separate from one another in how they function (Putnam, 2000). It is also important to recognize that the value of these bonds ultimately depends on the resources that can be derived from the m , not just the existence of connections; it must be considered that communities with strong social ties may not necessarily result in individuals with adequate resources (Portes & of social capital, Wacquant and Wilson (1989) find that blacks in extreme poverty not only lack in number of social connections, but those social connections they do have are deficient in resources. Therefore, they have neither the emotional nor economic s upport of network connectivity, or social capital; though even if they were connected with bonding social capital, they would likely still not garner as many advantageous resources. In order for a person to reap the benefits of bonding or bridging capital , he or she must become a member of a group. Even when he or she becomes part of this social network, it is important to recognize that capital takes time to develop, the benefits of which are not realized immediately (Bourdieu, 1986). As previously mentio ned, Portes and Sensenbrenner (1993) identify four criteria that make up social capital for discernable groups. These include value introjection, reciprocity transactions, bounded solidarity, and enforceable trust. Value introjection consists of the learne d behaviors that members of a group a dopt by socializing with others in the group. Reciprocity transactions are derived from the assumption that good deeds 13 will likely be paid back in the future. Bounded solidarity are the sentiments or actions shared by m embers of a group as a result of a shared situational circumstance. Enforceable trust involves the obligations each member has to the group as a result of the perceived benefits of being a member. The four criteria can be labeled as either principled or in strumental to explain the forces that drive the individual to comply. Value introjection and bounded solidarity are principled in that they are followed out of learned moral obligation; reciprocity exchanges and enforceable trust are instrumental, meaning reward or punishment depending on if they choose to adhere (Portes and Sensenbrenner, 1993) . of a social network , the y do not fully explain what leads an individual to join a particular network in the first place. The existence of social capital can depend on: real or symbolic material exchanges that work to bond individuals, the presence of a shared name, and the sharin g of space by those in the group (Bourdieu, 1986). In order for the individual to garner social capital, or, in other words, become a member of a particular social network, he or she must devote some significant resources to the group, and 8 6) conceptualization theory of McMillan and Chavis (1986). They argue that a sense of community involves four criteria: membership, influence, integ ration and fulfillment of needs and shared emotional conn ection. Membership consists of boundaries, emotional safety, sense of belonging and iden tification, personal investment and a common symbol system. Influence relates to how the individual forms the group and how the group forms the individual. Integration and fulfillment of needs concerns the notion that people join groups because of shared values and because they believe they can derive some benefit by participating as a member, which is another way of 14 saying they seek to exchange one form of capital for a nother (Bourdieu, 1986). Finally, a shared emotional connection involves members who recognize themselves and the group as having a common history; this history contributes to their interactions and norms. While Portes and (1986) four criteria discuss why and how individuals form communities. Furthermore, there are (C attell, 2001). Cattell (2001) describes five different types of networks that individuals in a community can be a part of, each part of a scale that describes the level of connection a person experiences. These networks include the following: Socially Exc luded, Homogeneous, Traditional, Heterogeneous, and Network of Solidarity. The networks are separated by the number of connections and the type of connections; the quantitative and qualitative distinctions are important to recognize as the sheer number of connections may be less significant when considering the characteristics of said connections (Sabin, 1993). Those at the lower level, otherwise known as Socially Excluded individuals, would have few people in their network and those people would likely inc lude family members or close friends. In contrast, a person with a Network of Solidarity would have connections in a multitude of organizations and geographic spaces (Cattell, 2001). These different levels of social capital can have an influence on what be nefits or restraints a person receives by being more or less connected, or by being connected to a network with greater or fewer resources. One example that illustrates the importance of distinguishing between the quantity and quality of relationships is t hat elderly people with socioexpressive, meaning voluntary, connections have been shown to have a lower mortality rate than those with connections consisting of family members or caretakers (Sabin, 1993). 15 2.8 The Effects of Social Capital There are a varie ty of studied benefits to high social capital levels at both the individual and community levels. In terms of income attainment, social capital can be a determinant of career mobility (Boxman et al., 1991) . In addition, Boxman et al. ( 1991 ) find that human capital, which includes education and work experience, matters less to job attainment and increased incomes when the individual has a greater amount of social capital. Also, in regard to wealth attainment, social capital can greatly aid minority groups, s uch as immigrants, in advancing their economic well - being (Port e s and Sensenbrenner, 1993). Greater links to social networks can help youth in low - income neighborhoods access better resources and gain more social support (de Souza Briggs, 1998). Bonds of s olidarity, one component of how social capital is built, within a community can be determinants of whether or not an economic or political initiative will be successful (Portes & Landolt, 2000). Additionally, increased levels of trust have been shown to en courage more civic participation (Brehm & Rahn, 1997), which is a key facet to a functioning democracy (Putnam, 2000). Finally, communities with greater levels of social capital are often perceived as safer (de Souza Briggs, 1998). 2.9 Geography and the Bu ilt Environment Although individual level assessments of social capital are the most widely accepted and utilized measures (Portes & Landolt, 2000), the importance of the built environment and on. The reason for this change is that the measureable attributes of individuals may be less important when location is considered. As m entioned earlier, w hen controlling for individual characteristics such as age, education, immigration status, employment status, whether th e respondent lives alone or not and sickness, variations in levels of social capital still exist when studying different geographic locations 16 (Lindstrom et al., 2002). Scholars typically focus on institutions and culture as agents affect ing what would be regarded as social capital (Rahimi et al., 2017). However, while educational level and employment status are more important determinants of social capital than other individual level factors, there is also evidence that landscape factors influence levels of social capital, contributing to the notion that the ecology of an area must also be considered (Rahimi et al., 2017). While it may seem that the built environment is less important nowadays due to the advent of online social networking , the internet can act to both increase or decrease social capital (Blanchard & Horan, 1998). Blanchard and Horan (1998) find that, through the internet, people can improve bridging social capital through increased exposure to others outside their physical proximity, but at the cost of lower bonding social capital within their own community. Therefore, social capital, as it pertains to civic participation, is most likely to improve communities when virtual and physical communities are built around one anoth er (Blanchard & supports both bridging and bonding connections. For these reasons, some research suggests that concepts like walkable neighborhoods (Leyden, 2 003) and Traditional Neighborhood Design (Bothwell et al., 1998) can be emphasized emerging, in part to act as a response to evidence that Americans are becoming less t rusting of others decade after decade (Rahn & Transue, 1998). Walkable neighborhoods and Traditional Neighborhood design both stress mixed - use planning and a wide diversity of public amenities accessible via non - motorized transportation (Bothwell et al., 1 998; Leyden, 2003). Similar to 17 neighbors, civic participation, trust in others, and the num ber of social organizations to which they belong. This study finds th at all of these elements of social capital were greater in walkable neighborhoods because these environments set a stage for face - to - face interactions. Therefore, the argument is that neighborhoods should be built to the human scale, not to the automobile scale. Design of the built environment influences levels of social networking; this increased sociality promotes civic participation, which improves both the social and economic vitality of a particular area (Bothwell et al., 1998). While some argue that homeownership and tenure greatly 2017). Other landscape factor example, vacancy rates have a negative correlation with trust, whereas the diversity of amenities drives up social connections (Rahimi et al., 2017). To the latter, diversity of amenitie s gives people more options for places to reach by foot; Leyden (2003) finds that there is a positive correlation with social capital measures and the number of places people claim they can walk. However, one drawback to solutions that emphasize landscape factors is that variables relating to spatial context, or the built environment, did not have a significant impact on improving trust between different races, meaning that physical space may not necessarily improve bridging social capital (Rahimi et al., 2 017). Nevertheless, research related to urban planning has begun to stress that government policies should attempt to discourage sprawling development tendencies in the same way that Leyden, 2003). The alternative to sprawl, from one perspective, is that American development should go back to 18 emphasizing Traditional Neighborhood Design, as it promotes social interactions that improve neighborhood cohesion (Bothwell et al., 1998). Thoug h Rahimi et al. (2017) only considers the presence or absence of amenity types in their study, the planned landscape in a given area can be important to social connections as well (Bothwell et al., 1998) . While social capital is often associated with posi tive outcomes for conceivably all people, this study focuses exclusively on public housing residents to determine if their built environment was promoting or deterring social networking. T his decision was made under the hypothesis that economic and social disadvantaged residents (Wacquant & Wilson, 1989). This study also recognized that reliance on the social capital for those residents who could seemingly benefit greatly from increased social networking. Often, public housing sites do not cultivate these desir able outcomes. For example, Bothwell et al. (1998) states that , the stigma created by the visual differences between public housing and its surrounding environment can instill both a neighborhood fear towards the low - income residents living there , as well as low self - esteem for those same residents (Bothwell et al., 1998). Low - income youth that move to a housing project in a more affluent neighborhood have lower levels of social support, but higher levels of social leverage (de Souza Briggs, 1998), possibly because where they live visibly categorizes them as separate from others in the community, making it difficult to form connections with their neighbors (Bothwell et al., 1998). This means that when a family moves from one neighborhood to another, they may gain in connections that provide mobility, but lose the emotional personal support that they had in their previous neighborhood (de Souza Briggs, 1998). 19 Bonds of solidarity are very difficult for policymakers to create, meaning that social capital buildi ng itself requires a strategic approach (Portes & Landolt, 2000). Those in opposition believe the use of social capital and its promotion by state governments is simply a way of ignoring other policies and institutions that may be more important to improvi ng public health, such as resource distribution and political representation (Muntaner et al., 2000). However, local governments have the potential to promote positive connections in a community, as evidence suggests trust can be created through policy ini tiatives that nurture participation (Brehm & Rahn, 1997). The question now is whether policy initiatives influencing the built environment can have similar must consider the id entity and politics of the area by including community members in the planning process . The purpose of doing so is to avoid disrupting what social capital already exists in that area 2.10 Conclusion theoretical term used in research studies appears to be both an advantage and disadvantage. The term can be modified to fit into a variety of fields and levels of study, making it an adaptable unit of measure to explain phenomena related to group membersh ip. However, defining social capital in such a wide scope makes it difficult to replicate studies as there is little agreement on what social capital truly is and how it should be measured (Macinko & Starfield, 2001) . Therefore, its usage in academic studi es must be joined with detailed explanations, informing others why a particular measure and certain variables were chosen and others were not. As a newer research term, there is very little pertaining to how social capital is influenced by the built enviro nment (Rahimi et al., 2017) ; urban planners and community stakeholders may benefit significantly by understanding how land use decisions can 20 aid or detract trust and civic participation, especially since the literature indicates that social t reaches to many important aspects of well - being and good government. T he following chapters describe the research framework and justifications for the chosen measurement s of social capital and the results of this undertaking. 21 Chapter 3. Case Study of Saginaw, MI 3 .1 Introduction To examine whether or not relationships exist between landscape factors and social capital, public housing in Saginaw, Michigan was targeted . T his municipality was selected with the hope to better inform how the c ity sites public housing to facilitate higher levels of social capital. Saginaw is the county seat of Saginaw County, located in the Great Lakes Bay region of Michigan with a n estimated population of 48,753 . To contextually frame the study area in regard to s ocial capital, t his chapter focuses on demograph ic and housing trends in the city . S aginaw is a legacy c ity, and like o ther legacy c ities it is generally defined by rapid population declin e ity has become a majority minority p opulation with very high levels of poverty compared to nearby Bay City and Midland, the metro politan area and the State of Michigan. This poverty appears to contribute to a poor housing market and damaged family structures. These data are used to describe the soc ial makeup of a post - industrial city during a period of disinvestment and rapid population loss; specifically, this chapter describes the c ity of Saginaw its population, race, family structure and housing instability to provide context on the br oader city . 3 .2 Socio - economic, Demographic, and Housing Profile and Population Projection T he c ity of Saginaw is similar to other legacy c ities throughout the Midwest United States, defined largely by a loss of industry that has resulted in h ite f light , high poverty and a minority community unable to attract investment. Unlike nearby Bay City and Midland, Saginaw is a majority minority community. With a population that is 45 .5 percent black and 15.6 percent Latino, the c ity has a far greater minority r ace make - up compared to nearby Midland and Bay City, which are both a r o und 90 percent white. This and other demographic differences are shown 22 in Table 1. This minority population is also much higher than those of Saginaw County (26.4 %) and the State of Mic (22.4 %) racial makeup (Community Profile, 2019). Table 1 : Demographic Comparisons for Saginaw, MI Total population White alone percent Percent of households with an individual younger than 18 with a single mother Percent of those age 25+ with college degree Percent of renters who spend more than 50% of gross income on rent Median household income Percent of people older than 15 who never been married Saginaw 48,753 42.5% 18.7% 8.3% 37.1% $29,542 46.3% Saginaw County 1 93,725 73.6% 9.5% 14.3% 26.7% $47,282 34.0% Midland 42,499 89.7% 6.6% 25.5% 22.1% $60,456 29.0% Bay City 34,099 88.0% 12.0% 11.1% 23.3% $37,230 37.3% Michigan 10,057,191 77.6% 8.0% 17.4% 25.6% $53,680 33.5% Besides racial demographics, Saginaw also d iffers from the other nearby municipalities in terms of family status. Another likely result of high poverty , the c ity has nearly double the percent of single mothers (18.7 %) than Saginaw County as a whole (ACS Population Summary, 2019). This, however, is not the result of divorce, as divorce rates in Saginaw are similar to other geographies; instead, Saginaw has a much higher population of people over the age of 15 who have never been married (Community Profile, 2019) . Low marriage rates and non - married co uple families could be the result of instability associated with poverty. The median household income in the c ity is around $24,000 less than the Michigan average and $18 ,000 a year less than Saginaw County as a whole (Community Profile, 2019) . The fact th at nearby localities enjoy middle - class incomes while the c ity has very low median incomes adds to the notion that Saginaw is suffering from disinvestment (see population trends in next section). As for housing in Saginaw, the market is grim compared to Mi dland, Bay City, Saginaw County and the State of Michigan (see Table 2) . 20.4 percent of residential structures are vacant 23 in the City, roughly double the percent vacant in the metro politan area (Community Profile, 2019) . This is clearly the result of the City losing roughly half of its peak population in a short time span . Additionally, despite having the lowest contract rent costs of the geographies compared here , a higher percentage (37.1%) of Saginaw renters have to use 50 percent or more of their gros s monthly income on rent costs compared to nearby cities, the metro area and the state (ACS Population Summary, 2019) averag e compared to the state, with a median value of $ 5 9,083 , far lower than the ot her (ACS Housing Summary, 2019; Community Profile, 2019) . This suggests a housing stock in poor condition that is still largely unaffordable to the 32.5 percent of households under the pover ty line of the population in poverty, another contributor to social instability (ACS Population Summary, 2019) . Table 2 : Housing Characteristics for Saginaw, MI Residential v acancy rate Median year structure built Median home value Median contract rent Saginaw 20.4 % 1949 $59,083 $456 Saginaw County 11.3% 1965 $120,154 $549 Midland 6.8% 1970 $158,873 $658 Bay City 11.5% 1944 $77,762 $463 Michigan 14.9% 1970 $156,034 $649 3 .3 Temporal Analysis This section will analyze trends and p rojections for population, race and h ousing for the City of Saginaw to examine whether the data discussed in section 3.2 are expected to change in the near future . the built environment. Figure 2 s population decline starting from 1970 and continuing to the present. The 2 - year moving average shows that the City will likely continue to lose population during the next Census count in 2020. B usiness Analyst Online also projects a 24 decline of around 1,5 00 people from 2018 to 2023 (Community Profile) . This is largely rcent of youth, seen in Figure 3 , indicating that the City is having a hard time attracting young professiona ls to its neighborhoods. Figure 2 : Population Change with 2 - year Moving Average Data Source: Manson, S., Schroeder, J., Van Riper, D., & Ruggles, S. (2018). Analysis by the author. 25 Figure 3 : Age Group Distribution Data Source: Manson, S. et al. (2018); Analysis by the author. As for race , minority population has grown quickly in recent decades . The areas east of the Saginaw River have seen the greatest decline in white residents, espec ially in the neighborhoods on the easternmost side of the city. It is clear that a rising minority population coupled with an overall decline in population is not necessarily the result of high migration of minorities to Saginaw, but rather white people ch oosing to move elsewhere. These poverty levels, according to Business Analyst Online, are not projected to c hange by 2023 ; median household income in Saginaw is only expected to rise by around $5 ,000 from 2018 to 2023 , which is a relatively small increase considering the median income is still very low, currently just $29,542 (Community Profile, 2019) . also not expected to improve much in t he next five years. Esri forecasts on Business Analyst Online show that vacancy rates will continue on their current 26 trend, rising from 16 percent in 2010 to 22.9 percent in 2023 (Community Profile, 2019) . While housing markets across the country have largely stabilized after the Great Recession in 2008, it appears Saginaw will continue to exper ience growing housing abandonment. Low property values suggest low demand, while high poverty rates still mean that the most affordable housing is not affordable to many residents. This also sugges ts that property owners in the c ity have little money avail able for property maintenance, meaning the housing quality is probably diminishing over time. 3.4 Conclusions on This chapter understands Saginaw a s a post - industrial city that has been in decline for decades. As conditions worsened, thos e who were able to leave did so, leaving behind an impoverished community with few resources for resurgence. The conditions for the people living in Saginaw, as well as its built environment have continued to decline for decades. The city, when compared to nearby municipalities, the state of Michigan and the US as a whole, is more impoverished, has a high er number of single mothers, a housing stock suffering from prolonged vacancy and lower homeownership rates. These data lead to two suppositions that help ed with the overall creation of the research framework. First, the built environment in the city is largely in a state of neglect, but likely more so in some areas compared to others. These differences in landscape factors may have a relationship with the neighborhood as cohesive. These findings, if substantiated, could help the city site public housing in areas that allow these residents to form networks to leverage more resource s. Second, many residents in Saginaw are experiencing difficult conditions that may not be remedied entirely, but possibly softened by a built environment that facilitates the opportunity for more 27 social interactions and greater levels of social capital. T he limits to choice for public housing study show that social capital and the built environment are related. The following chapter describes the methodology f implication for future research. 28 Chapter 4. Methodology In this study, a survey of public housing residents and geographic analysis is used to examine the potential relationsh ip between measures of trust and landscape factors in the c ity of Saginaw . As one can see from the literature review, studies on these topics allow for a wide range of choices pertaining to research area, social capital measures and the theoretical framewo rk that forms the research design. Therefore, this section will describe the context in which social capital is understood in this research , the concepts from the literature that were selected for this study and the reasoning behind these selections and su bsequent omissions, as well as a description of the survey instrument used for data collection. The context of the study discussed in t he Research Framework section (4 hypothesis statement, which is later ev aluated by the survey instrument discusse d in the Survey Design section (4.2 ). 4.1 Research Framework A s previously mentioned, social c measured differently across studies. Therefore, it was import ant to provide a reasoning for this measured and the data collection instrument chosen . This study considers key aspects that frequently make social capital res definition. The first problem was deciding whether social capital is a tool, a result or an individual response. In other words, whether it is a means to an end, the end itself, or somethi ng in between. en the means and end. Figure 4 illustrates definition of social capital, which was 29 adapted from the Portes and Landolt (2000) conceptua lization discussed previously . This choice was predicated on the notion that while social networking can help to garner more resources (means to an end), typical measures of social capital such as trust and civic participation are in themselves an acquired good (end). Therefore, it is reasonable to assert that social capital is a type of individual response. Because of this theoretical decision, this study does not attempt to understand how a social network acts to acquire more resources nor does it disrega rd the idea that individuals can increase their resources by raising their social capital. Instead, I merely wanted to measure the social capital held by the study population. Figure 4 : Theoretical Framework The second concept t hat needed reconciliation was the question of whether social capital should be measured at the individual level or at the community level. T hree reflective questions were considered when deciding on a quasi - individual - community scale for the methodological could indicate a scale as large as an entire country, as Robert Putnam uses, or as small as a group of people who regularly attend religious sermons together. The am biguity of the term community made this level of measurement on its own more difficult to defend because it often fails to capital. Second, from a data analysis perspective, it seemed that a research study can more easily measure a phenomenon at the individual level and then aggregate it to a community level than in 30 the reverse order. Third, as was mentioned in the literature review, community - level measurements typically attempt to understand higher societal concepts such as civic participation, democracy and political institutions, whereas individual - level assessments often measure this idea as it relates to the economic, social and health benefits for someone i n a given community (Macinko & Starfield, 2001). However, as Lindstrom et al. (2002) argue, social participation could act to bridge the gap between these measures. Thus, one can attempt to measure civic participation, for example, at the individual level as well as the community level. Therefore, decision was made to measure social capital as something held by individuals, but hypothetically shared within a particular group or community. This study employed a quasi - community - individual level of measuremen t based on the following assumptions drawn from parts of the literature review: Individuals, more than groups, vary in their behaviors and attitudes on a day - to - day basis. This idea is largely self - evident. However, it is also reasonable to assume that ind ividuals with similar traits who live in a relatively uniform geographic location, built environment or who share certain demographic traits will show similar characteristics. T his study sought to measure levels of social capital held by public housing re sidents in Saginaw, Michigan to evaluate if this poorer, more vulnerable group with less control over their place of residence varies in their social capital because of landscape factors near their housing unit . The shared housing status, income and reside nce within the same city could be considered a community measure. However, the variation in public housing type, location, exposure to certain landscape factors and differing demographic features allowed for more individualistic measures as well. T his meth odological approach was instituted with the understanding that its replication may or may not be suitable for another research population and that the results may only provide 31 a narrative for public housing residents in Saginaw, Michigan and their levels o f social capital. Thus, this study should be considered a case study that examines the relationship between trust and the built environment among a specific population within a single city . 4.2 Survey Design The survey ins trument, presented as a n Appendix , was informed by the concepts discussed in the literature review, borrowing questions from existing national level surveys in the U.S. for the purpose of generating quantitative data to explore the relationship between social capital and the built environm ent for public housing residents in Saginaw, Michigan. The survey borrows questions from the General Social Survey (2016 ballot 1), the Saguaro Seminar Social Capital Benchmark Test and the American Housing Survey (AHS). Factor analysis was used to deve lop preliminary indicators of social capital among public housing residents in the country using data from the 2013 wave of the AHS. The AHS survey was used because it asks questions regarding landscape factors as well as factors that could be considered f orms of social capital. This analysis suggest ed that there are five distinct measures of social capital: Neighborhood Cohesion/Norms, Hypothetical Civic Engagement, Actual Civic Engagement, Informal Social Networks and Formal Social Networks. These factors represent distinct types of social capital that could be measured in other studies. Preliminary regression analysis examined whether selected landscape factors measured by the AHS were associated with the five indicators of social capital identified above . Only Neighborhood Cohesion/Norms was found to correlate with the selected landscape factors. This factor is primarily composed of - 32 were used to inform the questions included on the survey instrument, while other survey questions were derived from the General Social Survey and the Saguaro Survey, both of which are used frequently in Social Capital literature. Although my preliminary analysis of AHS data showed that only neighborhood cohesion related questions correlated with landscape f actors, this study also include d measurements of trust, civic participation and social participation. Measures of trust were included for multiple participati on and vice versa. Thus, if urban planners and policymakers intend to alleviate the social detriments typically associated with low levels of trust, as well as promote civic participation from a traditionally marginalized population, it was determined wort hwhile to measure trust and civic participation levels for these public housing residents. Additionally, when measured at the local municipal level, as oppos ed to the national level, which includes the AHS dataset. This study, therefore, incorporated survey questions from the Saguaro Seminar study by measuring trust a s general trust in people, trust in neighbors and trust in people from 02) finding that this factor can bridge the gap between individual and community level measures of social capital. Because this study, again, is a quasi - community - individual level measure, it was important that social participation be included. 33 Table 3 : Survey Layout A: Location B: Social Trust C: Community D: Activities E: Neighborhood Conditions F: Household Section A: Location asked questions to confirm that the resident lived at that address. e included in the analysis to avoid collecting sentiments from someone who may not have lived in the neighborhood. This section also asked about his or her length of tenure at that residence and if he or she chose to live at that specific location or if it was assigned housing unit. Section B: Social Trust used questions from the General Social Survey and the Saguaro Seminar Capital Benchmark Test. These questions helped to form each of the three types of trust originally intended to act as dependent varia trust in people generally, followed by his or her trust towards people in the neighborhood and concluded with trust in white people, African American or black people, Asian people and Hispanic or Latino people . Section C : Community of the survey helped to inform the neighborhood cohesion index described in section 4.5. The first two questions sought to understand how many friends the respondents had in their neighborhood and how often they spend social ev enings with their neighbors. These two questions came directly from the General Social Survey. The remaining four questions in this section were derived from the American Housing Survey. These were selected specifically because of their demonstrated relati onships were certain landscape variables analyzed in the initial factor analysis described earlier in this section. 34 Section D: Activities sought to understand levels of civic participation for the survey population. This section was made up of the civic participation questions from the 2013 AHS. Section E : Neighborhood Conditions contained the built environment questions that could not be measured using US Census data or GIS. These questions asked about the condition of roads and the presence of blight a nd junk piles/illegal dumping within a half block of the and my dependent variables are discussed in s ection 4.6 . Section F: Household was the final group of ques tions in the survey and all asked demographic questions that were used as control variables in the analysis. These questions included whether or not the participant had access to a personal vehicle, what race(s) were represented in his or her household, hi s or her yearly household income, his or her education role in the analysis are both described in section 4.7. 4.3 Sample Selection and Survey Implementation The Saginaw Housing Commission (SHC) provided all survey population information including addresses and housing type for all of the properties that the organization manages. This included 81 single - family scattered sites, 92 townhomes located in the Northeast area of the cit y and 458 elderly - disabled high - rise units spread across five locations. F igure 5 shows the actual location for the five high rises, and general locations for the other two housing types in order to preserve respondent anonymity. 35 Figure 5 : General Locations of Public Housing Unit Respondents High Rise Scattered Site cluster Townhouses garner as much geographic variation as possible, the following proportion s of each housing type resident were include d in the first survey mailing: All 81 scattered sites (100% of housing type) 36 27 townhome sites (29.3% of housing type) 64 high rise units (13.9% of housing type) Total of 172 mailings (27.2% of population) Surveys were sent to the study population by mail after the study received exempt status from the Institutional Review Board . The first survey mailing earned 28 responses and was followed by a second mailing of 144 surveys t o those who did not respond to the first. This follow - up mailing yielded another 25 responses and led to a third wave of 119 surveys. In summary, after three attempts, the original sample population of 172 residents had a response rate of 29.6% (51 respons es). After the third mailing, the sample population was expanded to try to get more responses from the high rise and townhome residents. This fourth mailing included an additional 188 high rise units and 34 more townhome units totaling a final sample popul ation of 394 public housing units, or 62.4% of the properties manages by SHC. Table 4 displays the number of surveys that went to each housing type for the four waves. At the end of data collection process, 83 valid surveys (respondents indicated that they currently lived at the mailing address) were collected for a 21% response rate. All respondents were paid a $25 VISA gift card as an incentive to complete and return the survey. To help the person return their questionnaire, each mailing contained a pre - s tamped envelope with the return address already placed on the front. In addition, the final page of the survey instrument provided instructions for mailing the survey back to Dr. Noah Durst at the School of Planning, Design and Construction at Michigan Sta te University . If additional information was needed, the first page of the survey instrument also contained a brief description of the research project, contact information for myself and Dr. Noah Durst, as well as contact information to the Michigan State University Research Protection Program. A Thompson Research Endowment Award, the College of Social 37 Sciences and the School of Planning, Design & Construction funded all of the costs associated with the survey. Table 4 : Survey Mail ings by Household Type High Rises Townhomes Scattered Sites Total First Mailing 64 27 81 172 Second Mailing 48 24 72 144 Third Mailing 37 20 62 119 Fourth Mailing (sample population expanded) 188 34 0 222 4.4 Analytical Method The central research q uestion for this study was whether there is a relationship between measures of social capital and certain landscape factors. The results of which were intended to better inform urban planners, policymakers and landscape architects on the potential effect t hat certain factors may have on the trust and cohesion in a neighborhood. A ll data were aggregated to ensure respondent confidentiality and to consider the conditions of public housing residents in Saginaw as a study population. All of the 83 responses wer e imported into SPSS and Stata software for data analysis, coded and modified to ensure their information was measured in the most logical way. These modifications are discus sed in sections 4.5 and 4.6 of this chapter. I first analyzed the data using cross tabs between categorical variables and by comparing means for scale variables. This was completed for the three types of trust and all independent variables, as well as for the six measures of neighborhood cohesion in sect ion C of the survey instrument an d all independent variables. R ow and column percentages were analyzed for each of these tables . The purpose of this initial descriptive analysis was to observe trends and to ensure that each variable was measured in a manner that made logical sense before moving 38 forward with the linear regression analysis. Linear regression 1 was used to analyze these variables, later including controls to evaluate changes when certain factors such as race, income, education and others were held constant . In order to account for a heteroscedasticity (non - constant variance of the error term) the model used robust standard errors . Later, standard errors were clustered by block group to account for spatial dependen ce due to the fact that many of the observations were drawn from the same neighborhood . The following model was estimated: where Y, the dependent variable, is a series of measures of either neighbor tr ust or neighborhood cohesion for respondent i, L is a vector of the landscape factors described above, and X is a series of socio - economic and demographic control variables. is a vector of coefficients which represent the association between landsca pe factors and the selected measures of neighborhood trust or cohesion, while is a vector of coefficients representing the relationship between the dependent variable and selected control variables. is the error term. The hypothesis for this model was that, when controlling for non - landscape factors such as demographic variables, show variation explained in part by differences in the built environment . The hypotheses for each independent variable are discussed in sections 4.6 and 4.7. 1 O rdered logit regression was more appropriate for analyses of ordinal variables, in this case, general trust and trust in neighbors. An ordered l ogit regression model was performed and the results (the sign of the coefficient and measures of statistical significance) were nearly identical to the linear regression. Therefore, due to the ease of interpretation for linear regression, the ordered logit analysis is not presented here. 39 4.5 Measurement of Trust and Neighborhood Cohesion Trust was measured in three different ways in order to see if one or more types of trust are related to landscape fact ors. These questions drew from the Saguaro Seminar Social Capital Benchmark Test . The purpose of having multiple measures of trust was to ascertain whether general trust, trust in neighbors and trust in people of other races varied for each respondent and if this variation was related to differences in location and landscape factors. General trust : This question asked residents Generally speaking, would you say that Poss would not show a relationship with landscape factors. Trust in neighbors : This question people in your neighborhood. Generally speaking, would you say that you can trust them T his neighborhood variable seemed most interesting because it appear ed to logically relate to neighborhood landscape factors. For this reason, and because the other trust variables did not show noticeable results in the anecdotal analysis, it is the only variable that appeared as a dependent variable in the regression model. My hypotheses for how each independent variable would relate to trust in neighbors are discussed in section 4.6. Trust in other races: This variable was calculated by asking survey participants the same question as the trust in neighbors variable, but inst ead asked about trust in particular rac ial groups. These responses were used to create a trust in other races index. This index was calculated by summing the trust levels in all races from 1 (trust them not at all) to 3 40 (trust them a lot) and calculating a n average. Trust in a particular race was excluded if that was t he same race as those races represented in the respondent or if the t in the races represented in his or her household. This comparison prompted a new final category where trust in other races was either lower, the same or higher than trust in the races represented in the household. Neighborhood cohesion index: To measure neighborhood cohesion with one combined variable , the six questions in Section C of the survey were averaged to create an index. As mentioned previously, t hese questions asked residents the number of friends they had in their neighborhood; the number of s ocial evenings they spend with others in their neighborhood; and asked the degree to which the participant agreed or disagreed with the following statements: o People in my neighborhood are willing to help their neighbors o People in my neighborhood get along with each other o I live in a close - knit neighborhood o People in my neighborhood share the same values Each response was averaged from zero to five for number of friends, zero to six for how often they spend a social evening with a neighbor and zero to three for each of the four AHS agree or disagree statements. These averages were then added together and divided by six (the number of questions included in the index) to create a fin al neighborhood cohesion score. Only those respondents that answered all six of these quest ions were included in the index , with the lowest possible score as zero and the highest possible 41 score as six. My hypotheses for how each independent variable would relate to trust in neighbors are discussed in section 4.6. This process resulte d in the creation of two dependent variables for the regression models. The first was trust in neighbors and the second was neighborhood cohesion. These variables were selected as the dependent variables because they seemed to be the ones that would most l ikely show a relationship with landscape factors. General trust and trust in other races were deemed more to individual demographic factors such as age, race, in come and education. This hypothesis was confirmed by the anecdotal analysis discussed in section 5.2. 4.6 Measurement of Landscape Factors The following landscape factors were measured because of their supposed relationship with measures of trust. The dat a for first three variables mentioned in this section, presence of blight, junk pile or illegal dumping and road conditions made up Section E: Neighborhood Conditions of the survey instrument. The remaining variables were analyzed using American Community Survey (ACS) 2013 - 2017 Five - Year Estimates from the United States Census Bureau , GIS software or Google Maps and information provided by the Saginaw Housing Commission . Presence of blight: abando ned buildings within half a block of this building? Is there more than one recoded to only signify yes, there is blight within a half block, or no, there is no blight. I hypothesized that the presence of blight would have a negative relationship with trust and 42 neighborhood cohesion because its existence may signify unsafe conditions or n eighborhood neglect. Junk or illegal dumping same as the presence of blight, that junk piles or illegal dumping would signify either criminal activity or neighborhood decay, thereby reducing levels of trust and cohesion. Road conditions : This question sought to understand if road conditions reduced trust and cohesion with the notion that poor conditions, si milar to the previous two variables discussed, represent a neighborhood in distress. This variable also related to the initial factor analysis for the neighborhood cohesion variables used by AHS, which is primarily the reason it was included in this analys is. However, I hypothesized that this variable would not show a relationship to either of the dependent variables. Housing type: This information was denoted in the address list provided by the Saginaw Housing Commission. Public housing residents in Sagin aw live in one of five high rise units 2 , townhouses or single - family detached scattered sites. The high rises were located in varying locations throughout the city. The townhouses were situated in the northeastern part of the city, while the scattered site s were clustered in primarily five neighborhoods. This variable was included in the analysis to test whether housing units with more people living in closer quarters fostered more opportunities for social interaction. Therefore, I predicted that high rise unit residents would show higher levels 2 The high - Commission. While it does not appear that all residents who answered the survey held these demographic characteris tics, the many who did resulted in a mostly homogenous respondent pool from this housing type. 43 of trust and cohesion, followed by townhouse residents, and scattered site residents showing the lowest levels. Housing density: Th ese 2013 - 2017 F ive - Y ear E stimate s. The number of housing units within the block group was number of housing units per square mile. This hypothesis for this variable was similar to housing unit in tha t higher densities would relate to higher trust and cohesion because of an increased opportunity for social interaction. Walk Score: Walk Score measures walkability from 0 - 100 by mapping routes to typical amenities including supermarkets, schools, parks, r estaurants and retail businesses . s 40 out of 100. Although the high rise units in this study had an average score of 53.93, the scattered sites and townhouses were much lower at 24.68 and 17.30, respectively. T he score provi ded by the company from 0 - 100 was identified for each addre ss that responded to the survey. I surmised that greater walkability and access to amenities without the need of a car would help to promote social interactions and would, therefore, show a positiv e relationship with trust and cohesion. Property vacancy: These data came from the A - Y ear E stimates. Property vacancy was measured at the census block group level as the number of properties vacant for a reason other than for rent, for sale, seasonal usage or for migrant workers divided by the total number of housing units within the block group. In other words, this number measures housing units vacant due to abandonment. Similar to junk piles and blight, this variable was supposed to show a negative relationship with neighbor trust and 44 neighborhood cohesion because it creates a visual eye sore and can be a health and safety risk to nearby residents. Median housing values: This variable was also measured at the census block group level from t he A CS 2017 Five - Y ear E stimates. Twenty of the 83 respondents lived in census block groups that did not have median home value data available for this estimate. For these responses, census tract median relations hip with the dependent variables was more difficult to hypothesize because of the study population comprising public housing residents. One perspective was that public housing units in block groups with higher median housing values would promote higher tru st and cohesion because of a likely reduction in crime and visual blight. who moved to a wealthier area did not necessarily gain new social networks. In other words, hig her median housing values could just as likely hamper trust and cohesion for this study population as promote these measures of social capital. Connectivity: This variable refers to the number of streets that intersect the street on which the respondent l ived. Following space syntax theory, c urved streets were broken into series of straight lines intersecting one another, se en in Figure 6 , which shows highly connected streets in red and orange, less connected streets in yellow and green and highly disconne cted streets in blue. This map data was completed in Depthmap program and was that highly disconnected housing units would show le ss trust and cohesion because they would lite rally have less connection to their neighborhood and the city as a whole. 45 Global integration: This variable measures how many turns are required to reach a housing unit from anywhere in Saginaw. Also, following the space syntax theory, t his data was collec ted in the same way as the connectivity factor, wherein the integration value of all street segments in Saginaw were calculated using Depthmap program , where each line intersection is considered a turn. The higher the value for this value, the more integra ted the street segment is considered. Global integration measures how easy any street segment is accessible from any street segment in the city. I hypothesized that higher integration would relate to greater trust and cohesion values. Local integration: L ocal integration is the same as global integration but only measures the number of connected line segment turns up to three segments. Local integration measures how accessible street segments are within a certain neighborhood. My hypothesis for this varia ble was the same as global integration. Sidewalks: Each housing unit was observed to determine whether or not there was a sidewalk adjacent to the property. Because all of the units had a sidewalk present, this variable was dropped from the analysis. 46 Figure 6 : Connectivity Output Generated by Depthmap 4.7 Control Variables The following control variables were added to the analysis and eventual linear regression models to account for individual characteristics that would possibly relate to variations in trust finding that landscape factors and individual respondent characteristics can both explain variations in social capital. This section addresses each control variable used in the model and provides an initial hypothesis that was later tested and discussed in Chapter 5: Analysis. Chose housing unit : This question was intended to determine whether not the resident was assigned to t he housing unit without a choice, or if they requested that location and were granted their choice by the housing commission. 47 Length of t enure in unit : The second question of the survey asked residents how long they had lived at their current SHC managed residence. The responses were broken down However, due to a lack of variation and low overall response rate, this variable was recoded to a 1 or 0 value to define tho se residents who had lived at their current address for at least 2 years. The hypothesis for this variable was that longer tenure would result in higher social capital, as more time spent in the neighborhood would likely indicate more familiarity with neig hbors and more social interactions. Race : The survey asked the survey participants to identify what race(s) were represented by the people who lived in their household. He or she was able to check all that applied The purpose of this variation in levels of trust and cohesion . I did not necessarily have a preconceived idea of how this variation would present itself in the results. Because of low variation in the races represented in this study, with the majority of respondents indicating either black or white, this variable wa s recoded to indicate 1 for black residents and 0 for all other races. Education : Respondents were asked to provide their highest grade of school or year of My supposition was that respondents with more education would have higher levels of trust and cohesion, perhaps because they may have been more open to social interactions and new experiences. For the analysis, 48 this var iable was recoded as a binary 1 for yes, 0 for no on whether or not the respondent had completed more than a high school diploma or GED. Income : S urvey participants were also asked to indicate their yearly household income with responses ranging from less than $10,000 to $50,000 or more, and five choices total. I hypothesized a negative relationship between household income and trust and cohesion. This was based on the notion that greater financial stability within the household would allow for more leisur e time and chances to build social connections with others. Employment status: Respondents were asked to indicate the scenario that best described their employment status. This question was originally intended to be a single response variable, but some of the respondents checked more than one choice so it was later coded as a multiple response variable. For the regression model, this variable was recoded as a My hypot hesis was that people who were working would have higher levels of trust and social capital, as their work environment and relatively greater financial stability would provide more opportunities for networking. Access to personal vehicle : R esidents were as ked if they had access to a personal vehicle with multiple ideas on how this factor could relate to trust and cohesion. First, residents with access to a personal vehicle may have been more likely to spend time outside the neighborhood and therefore, be le ss likely to indicate trust in their neighbors or perceptions of neighborhood cohesion. Along that line of thinking, it was also possible that people with a vehicle may have spent more time walking to destinations or using public transportation, thereby ex posing the respondent to more of his or her neighbors. 49 For these reasons, I predicted that respondents without access to a personal vehicle would show higher levels of trust and cohesion. 4.8 Study Limitations and design presented some limitations to the analysis and subsequent results. First, by design, the studied population comprised only residents in Saginaw, Michigan and, further, only those living in public housing managed by the Saginaw Housing Commission. The sample size of respondents was further limited by the difficulty in gathering responses. Even with a $25 gift, just 83 out of the 394 surveyed units returned the survey, leading to a 21 percent response rate. This low response rate and the fact that many of the scattere d sites were clustered in close proximity to one another led to limited spatial variation in landscape factors . This clustering was not observed until after data had been collected and mapped. I t was also recognize d that the different types of trust should have all been measured in the same way rather than taken directly from the Saguaro survey without modification. This made it difficult to discern differences between each type of trust. In addition , although the representatives of the Saginaw Housing Comm ission reported that a relatively small number of public housing residents were able to choose their place of residence (i.e., their unit), I cannot confirm that this was in fact the case. R espondents were asked whether they had chosen the location of thei r housing unit or if they were randomly assigned. A substantial share of respondents ( 69 %) reported that they had chosen the unit. This question was likely worded inaccurately and thus resulted in a variable that could not be used for analysis. This means that m y analysis cannot provide insight into the causal effect of landscape factors on trust because I cannot reliably distinguish between two potential explanations for lower trust among residents with certain landscape characteristics: for example, it is possible that an individual residing areas 50 with particular landscape factors may have lower trust as a result of said factors, or the y may have chosen to live in such an area as a result of their pre - existing lower levels of trust. These limitations shoul d considered when interpreting the results of the analysis, discussed in the following chapter. 51 Chapter 5. Analysis This chapter addresses the initial research question as to whether or not there is a relationship between landscape factors and measures of social capital. The ultimate goal, as previously stated, was to help municipal planners, residents and other public officials consider most at risk of m arginalization. This chapter begins with a summary of the frequencies for each section of the survey instrument. I then discuss notable relationships between trust and landscape factors and neighborhood cohesion and landscape factors using cross tabs and b y comparing means. Finally, these noted relationships are tested for their significance with linear regression models. It is important to note that this section and the study as a whole do not attempt to claim causality between variables and is intended to act merely as an analysis of trends and relationships. 5.1 Survey Results Of the 83 respondents, 19 ( 23 %) were from scattered sites, 10 ( 12 %) lived in townhomes and 54 ( 65 %) lived in high rises, totaling 25 unique locations. Most respondents earned less than $10,000 a year in household income (71 %) and had not com pleted any college courses (52 %). The length of t enure varied but most respondents had lived in their SHC managed u nit for at least two years (78 %). This indicated that most of the respondents p resumably had a deeper understanding of their neighborhood, its built environment and their neighbors. When asked represented 63% of households, white people rep resented 23%, Latinos represented 1 1 % and 9% selected ther. Respondents could report more than one race in the household, which is why these percentages do not sum to 100. Personal vehicle ownership was practically split down the 52 middle with 51% indicat ing they had access to a personal vehicle. Finally, just 16% of the respondents said that they were currently working. 40% stated that they were perma nently disabled and another 19 % were retired. This is largely attributable to the fact that 65% of the res ponses came from residents in the elderly - disabled high rises. When asked ab out people their trust in people generally, most (52% People can be % stated that they trust thei r neighbors a lot. However, 19 % did not trust their neighbors at all, meaning that the majority (77%) we re somewhere in the middle. For the questions in Section C: Community, 23 % indicated that they do not have any friends who lived in t he neighborhood, while another 40 % said they have only 1 or 2 neighborhood friends. A s discussed in section 4.7 , the number of friends likely depends length of tenure in the housing unit. Because 78% of the survey po pulation indicated that they had lived at their current address for at least two years, it was somewhat surprising to find that 63% had just two neighborhood friends or fewer. This showed that factors other than tenure likely contributed to this variation in local friends. Adding to this trend of low sociality , 36% said that they never spend a social evening with someone who lived in their neighborhood , 27 % do so once or twice a week , and 13 % socialize with a neighbor every day. It is worth noting again tha - - rises and that this was a possible explanation for these few community relationships/ Section C, as described in section 4.8 of this report, also asked respondents to indicate the degree to w hich they agree or disagree with four statements regarding their perception of their The frequencies of each of these questions was fairly uniform, with between 45% and 55% somewhat agreeing with the statement. The strongly dis agree option 53 was the least selected for all neighborhood cohesion questions meaning that this population of public housing residents feel s that their neighborhood is at least somewhat cohesive. 48% of respondents said that major repair wor k is needed on t heir block. 64 % said there is a blighted structu re within a half block with 49 % indicating that t here is more than one. Only 18 % said that there is a junk pile or illegal dumping on their blo ck. The high level of variation in perceptions of road conditions and the presence of blight/abandonment helped to provide some capability to compare how these differing landscape experiences may have related to trust in neighbors and neighborhood cohesion. The following sections summarize the notable anecdotal findings on these relationships, followed by a more rigorous statistical analysis using all independent variables to test if these relationships were significant. 5.2 Trust and Landscape Factors This section examines preliminary evidence regarding the link betwee n landscape factors and trust. To do so, each landscape factors is examined for respondents who reported differing levels of trust (high, medium, low) . Trends emerged for each of the three types of trust and their relationship to the measured landscape fa ctors , though neighbor trust, as hypothesized in section 4.5, showed the most associations with landscape variables . General trust in people showed relationships with block group home value, Walk Score and block group housing density. Table 5 , which shows general trust in relation to all landscape variables reveals than d in block groups averaging a home at $40,630. This trend suggests that areas with a struggling housing market may act as a 54 in others. However, it should also be noted that the presence of blight within a half block, another indication of a struggling housing market, did not share this trend. This pattern for general trust in others also appeared when considering the Walk Sco re of each unit. People who declared a trust in others were more likely to live in a more walkable area than people who stated a cautionary approach to others. While both levels of trust had a neighborhood may have play ed a role in their perceived trust in people generally. Although I hypothesized th at housing density would only show an association with neighbor trust under the assumption that more social interactions often lead to higher trust in people living nearby , the analysis showed that there is also a notable trend with general trust. Responde just 1,186 housing units per square mile. As stated previously, one shou ld not assume causality fr om these relationships, especially for this example analyzing general trust and housing density. There are other factors that likely contribute as well. Table 5 : General Trust and Landscape Factors Genera l Trust Row percentages Column percentages Housing Type High Rise 12.0 56.0 32.0 66.7 68.3 55.2 Townhouse 10.0 70.0 20.0 11.1 17.1 6.9 Scattered Site 10.5 31.6 57.9 22.2 14.6 37.9 55 General Trust Row percentages Column perce ntages Block group median home value (In thousands of dollars) 38.95 17.743 40.63 38.95 17.74 40.63 Walk Score 38.22 23.42 44.59 38.22 23.42 44.59 Block group vacancy (avg %) 10.91 6.17 9.78 10.91 6.17 9.78 Block group housing density (In hundreds of housing units/sq mile) 18.99 11.86 19.42 18.99 11.86 19.42 Road conditions Major repair work 14.3 42.9 42.9 55.6 38.5 57.7 Minor repair work 10.0 60.0 30.0 33.3 46.2 34.6 No repair work 11.1 66.7 22.2 11.1 15.4 7.7 Blight within half block 11.1 55.6 33.3 66.7 73.2 62.1 Junk pile within half block 8.3 50.0 41.7 11.1 18.2 20.8 Connectivity 12.00 15.20 11.79 12.00 15.20 11.79 Global i ntegration 1.35208 1.35895 1.34089 1.35208 1.35895 1.34089 Local i ntegration 3.04800 3.16067 2.98110 3.04800 3.1 6067 2.98110 Unemployed 33.3 66.7 0 11.1 12.2 30.8 Retired 11.1 24.4 11.5 Permanently disabled 66.7 36.6 46.2 Homemaker 0 0 7.7 Student 11.1 4.9 0 d relationships to housing unit type, the presence of bligh t and junk , road conditions, home values and housing density , shown in Table 6 . The notion that more landscape factors relate to neighbor trust makes sense, as the built environment constitutes local conditions and therefore presumably has an association w ith trust in people within that local environment. 56 Looking at row percentages in Table 6 , one can observe the following p attern: high rise residents varied widely i n their trust in neighbors and we re the only group to say that they trust the people in thei r neighborhood a lot. None of the scattered site residents stated that they trust their neighbors a lot, nor did they indicate that they do not trust them at all. Instead, they are somewhere in the middle. Townhouse residents were clearly the least trustin g of their neighbors. Half said that they do not trust their neighbors at all, 30% trust them only a little and no one said they trust their neighbors a lot. It is worth noting that the townhouse housing type is confined to one area of the city so location , amongst other factors, may have played a key role in these trends. Table 6 : Neighbor Trust and Landscape Factors Neighbor Trust Row percentages Column percentages Housing Type High Rise 6.0 32.0 42.0 20.0 100.0 57.1 65.6 66.7 Townhouse 0 20.0 30.0 50.0 0 7.1 9.4 33.3 Scattered Site 0 55.6 44.5 0 0 35.7 25.0 0 Block group median home value (In thousands of dollars) 54.97 39.94 39.99 41.23 54.97 39.94 39.99 41.23 Walk Score 50.67 38.93 43.84 43.27 50.67 38.93 43.84 43.27 Block group vacancy (avg %) 6.99 10.5 9.74 6.79 6.99 10.5 9.74 6.79 Block group housing density (In hundreds of housing units/sq mile) 27.01 19.21 20.79 23.23 27.01 19.21 20.79 23.23 Road conditions Major repair work 2.9 32.4 38.2 26.5 50.0 42.3 41.9 69.2 Minor repair work 3.4 44.8 41.4 10.3 50.0 50.0 38.7 23.1 No repair work 0 22.2 66.7 11.1 0 7.7 19.4 7.7 Blight within half block 3.7 29.6 42.6 24.1 66.7 57.1 71.9 86.7 Junk pile within half block 9.1 18.2 45.5 27.3 33.3 8.7 17.2 37. 5 57 Connectivity 16.00 12.32 14.00 15.80 16.00 12.32 14.00 15.80 Global i ntegration 1.43260 1.30239 1.38332 1.36606 1.43260 1.30239 1.38332 1.36606 Local i ntegration 3.46353 2.96734 3.11252 3.25479 3.46353 2.96734 3.11252 3.25479 In te rms of observable blight, 24 % of people who said there is a vandalized or abandoned building within a half block of their residence stated they do not trust their neighb ors at all. This compares to 8 % of people without blight on their block. Roughly the sa me trend appeared for junk piles. Of the respondents who said there is a junk pile or ille gal dumping on their block, 27 % do not trust their neighbors at all, while just 10 % of people without junk on their block indicated the same low level of neighbor tru st. In regard to road conditions, residents who stated roads within a half block of their housing unit were in need of major repairs were those most likely to indicate that they do not trust the people in their neighborhood. Response rates for minor or rep air work and no repair work were relatively similar. O nly three respondents said that they trust their neighbors a lot, meaning that this does not give a clear indication of a relationship between high neighbor trust and landscape factors . However, t here are multiple similarities amongst these three residents. All three resided in high rise s , but in three different units and locations . Besides the housing style, the location of these units had a higher median block group home value and a higher housing den sity compared to those for people with less trust in their neighbors. Trust in other races seemed to associate with housing unit type, housing density and vacancy rates. People in scattered sites were slightly more likely to trust other races more, while people in high rises were slightly more likely to trust other races less. Townhouse residents 58 trusted other races the same. Both housing density and vacancy rates showed a small negative relationship wi th trust in other races. 5.3 Neighborhood Cohesion an d Landscape Factors Similar to section 5.2, which analyzed possible associations between trust and landscape factors, t his section discusses the relationships between measures of neighborhood cohesion and those same landscape variables. The purpose of doi ng this was to evaluate, before completing the linear regression models, whether or not certain facets of the built environment showed goal, mentioned throughou t this report, was to understand these relationships to help urban planners and local officials to effectively influence the creation of landscapes conducive to high social capital levels. In this part of the anecdotal analysis, h ousing type showed the mos t are described here for the six questions that sought to und erstand these dynamics. High rise residents were the least likely to indicate that they have zero neighborhood friends (18 %) and the most likely to state that th ey have 10 or more friends (20 %) when measuring percentages within each of the three housing t ypes, though the homogeneity in age for these residents may have driven this finding. Table 7 illustrates the trends described for housing type. Scattered site residents appeared to have the fewest friends in close geographic proximity , while townhouse res idents primarily stated having one or two neighbor friends. High rise residents were also those most likely to spend a social evening with their neighbors almost every day (17 %) and least likely to never do so (29 %) . Townhouse respondents varied more in th eir social freq uency, showing no discernable trend. More than half (53 %) of scattered site residents 59 never spend a social evening with neighbors. These trends lead to the notion that multi - family housing may facilitate mo re social interactions. In this sen se, public housing units with higher density may actually provide social benefits , assuming they are aesthetically pleasing and designed in a manner that deters criminal behavior. This would also lead to the notion that single - family detached units, becaus e they are by definition spatially isolated compared to other housing types, prevent the acquisition of social capital. However, it was also probable that this variation was driven by the fact that high rise residents tended to share more demographic chara cteristics such as age and employment status. Table 7 : Housing Type and Neighborhood Cohesion Independent variable Housing type High rise Townhouse Scattered site Friends in Neighborhood None 50.0 (18.4) 11.1 (20.0) 38.9 (36. 8) 1 or 2 51.6 (32.7) 22.6 (70.0) 25.8 (42.1) 3 to 5 78.6 (22.4) 0 (0) 21.4 (15.8) 6 to 9 75.0 (6.1) 25.0 (10.0) 0 (0) 10 or more 90.9 (20.4) 0 (0) 9.1 (5.3) Frequency of social evenings in neighborhood Never 51.9 (29.2) 14.8 (40.0) 33.3 (52.9) O nce/year to once/month 61.5 (16.7) 7.7 (10.0) 30.8 (23.5) Several times/month to once or twice/week 72.0 (37.5) 16.0 (40.0) 12.0 (17.6) Almost every day 80.0 (16.7) 10.0 (10.0) 10.0 (5.9) People in neighborhood are willing to help their neighbors St rongly disagree 50.0 (7.5) 33.3 (25.0) 16.7 (6.3) Somewhat disagree 63.6 (17.5) 9.1 (12.5) 27.3 (18.8) Somewhat agree 60.0 (52.5) 11.4 (50.0) 28.6 (62.5) Strongly agree 75.0 (22.5) 8.3 (12.5) 16.7 (12.5) People in neighborhood get along with each other Strongly disagree 66.7 (5.1) 33.3 (12.5) 0 (0) Somewhat disagree 72.7 (20.5) 18.2 (25.0) 9.1 (6.7) Somewhat agree 60.6 (51.3) 12.1 (50.0) 27.3 (60.0) Strongly agree 60.0 (23.1) 6.7 (12.5) 33.3 (33.3) I live in a close - knit neighborhood Strongl y disagree 81.8 (20.9) 9.1 (14.3) 9.1 (6.7) Somewhat disagree 69.2 (20.9) 7.7 (14.3) 23.1 (20.0) Somewhat agree 56.7 (39.5) 16.7 (71.4) 26.7 (53.3) Strongly agree 72.7 (18.6) 0 (0) 27.3 (20.0) 60 Independent variable Housing type High r ise Townhouse Scattered site People in neighborhood share same values Strongly disagree 80.0 (21.6) 20.0 (33.3) 0 (0) Somewhat disagree 60.0 (24.3) 20.0 (50.0) 20.0 (30.0) Somewhat agree 70.8 (45.9) 4.2 (16.7) 25.0 (60.0) Strongly agree 75.0 (8.1) 0 (0) 25.0 (10.0) The trends shown for connectivity and loca l integration, s hown in Table 8 , are somewhat explained intuitively: housing units located on streets that had a higher number of intersections with other streets (connections) s howed more socia l connections . In addition, units on a street with a higher local integration score, meaning fewer turns are required to reach those streets, also showed signs of a possible relationship with number of neighborhood friends. Another way of understanding the se variables, described in section 4.6, is that they are measures of spatial isolation . In other words, this initial analysis indicated that spatially isolated housing units related to lower opportunity for neighborhood connections. Although, because of th e small amount of locational variation offered by this study population, it is probable that this variation in number of friends is prompted by other factors. This idea is also likely true for the observation that housing density had a positive relationshi p with both the number of friends a respondent had in their neighborhood as well as the frequency of their social evenings. Though this would confirm my hypothesis for these variables, mentioned in section 4.6, low variation made this claim difficult to su pport without a regression model. The remaining paragraphs in this section describe the perceived associations between landscape factors and each of the four neighborhood cohesion statements outlined in section 4.5. I n regard to the first statement , that 83 % of people who strongly disagreed stated their road needed major repairs, 61 whereas no one who strongly disagreed indicated that repairs were unnecessary. In addition, those who strongly agr eed with the statement were more likely to indicate good road conditions near their residence. This finding supported my hypothesis that better road conditions would show an association with higher opinions of neighborhood cohesion. The second of these st atements, which dealt with how well neighbors get along, only showed a relationship with housing type. Over 93% of scattered site residents somewhat agreed or strongly agreed with the statement compared to just 62% of townhouse residents. 74 % of high rise residents agreed to some extent. Townhouse residents were most l ikely to strongly disagree (12 %). apparent associations, was likely created by demographic similarities between these respondents rather than their housing type. In other words, high rise residents may agree that the people in their neighborhood get along well because the people in their neighborhood share characteristics such as age and employment status t hat would likely facilitate more positive interactions than a neighborhood where residents differ more. Table 8 : Connectivity, Integration and Neighborhood Cohesion Mean Values Independent variable Connectivity Global integration L ocal integration Friends in Neighborhood None 11.78 1.33857 2.99907 1 or 2 12.58 1.26260 2.90405 3 to 5 14.36 1.41601 3.18176 6 to 9 14.50 1.37067 3.14338 10 or more 15.82 1.50332 3.45395 Frequency of social evenings in neighborhood Never 12. 30 1.35023 2.98908 Once/year to once/month 11.69 1.27537 2.85914 Several times/month to once or twice/week 14.48 1.33062 3.15127 Almost every day 17.60 1.52982 3.51336 People in neighborhood are willing to help their neighbors Strongly disagree 13. 17 1.29343 3.02027 Somewhat disagree 19.09 1.46106 3.39342 Somewhat agree 11.97 1.31724 2.99535 62 Strongly agree 15.75 1.40253 3.24450 People in neighborhood get along with each other Strongly disagree 19.00 1.43371 3.45385 Somewhat disagree 19.27 1.44224 3.48135 Somewhat agree 11.27 1.30422 2.94154 Strongly agree 13.27 1.37778 3.09815 I live in a close - knit neighborhood Strongly disagree 17.82 1.44385 3.35683 Somewhat disagree 15.23 1.38924 3.23895 Somewhat agree 11.70 1.29 912 2.95902 Strongly agree 13.36 1.40515 3.12693 People in neighborhood share same values Strongly disagree 19.40 1.46335 3.44325 Somewhat disagree 14.27 1.28689 3.08922 Somewhat agree 13.21 1.33641 3.05783 Strongly agree 15.25 1.55026 3.32645 Further expanding on this idea, respondents from high rise units were most likely to strongly disagree with the statement that their neighborhood is close - knit (21%). This group was followed by those from the townhouses and scattered sites (14% and 7 %, res pectively). Scattered site residents were most likely to agree that their neighborhood is close - knit (20%) . The presence of junk or illegal dumping within a half block also showed associations with this statement. Of those who strongly disagreed with the n otion that their neighborhood is close - knit , 40% stated that there was junk near their housing unit. This frequency was much smaller for those who somewhat disagreed or agreed. These findings can be seen in Table 9 , which shows column percentages in parent heses. showed relationships with housing type, vacancy, road conditions, junk and blight. None of the scattered site residents strongly disagreed with the statement while 33% of the townhouse residents did. However, very few of the respondents from all three of the housing types strongly agreed that their neighbors share the same values. Most answered somewhat agree or somewhat 63 disagree. Block group vacancy, interestingly, showed a positive association: as vacancy rates went up so did the level to which respondents agreed with the statement, display ed in Table 10 . Of those who strongly agreed, 50% had nearby roads needing major repairs. In comparison, 80% of those who strong ly disagreed stated very poor road conditions. No one who strongly disagreed lived on a block needing no repairs. This trend was similar for the presence of blight within a half block. Those who agreed were less likely to observe dilapidated or vacant stru ctures, whereas those who disagreed were very likely to notice blight. As for junk, 44% of those who strongly disagreed had illegal dumping or junk accumulation within a half block. Just 25% of those who strongly agreed and 9% of those who agreed noticed t he same landscape factor. Table 9 : Road Conditions, Blight, Junk and Neighborhood Cohesion Road conditions within half block Major repair work needed Minor repair work needed No repair work needed Blight present within half bloc k Junk pile or illegal dumping present within half block Friends in Neighborhood None 52.9 (25.0) 41.2 (23.3) 5.9 (12.5) 72.2 (24.5) 15.4 (16.7) 1 or 2 41.4 (33.3) 48.3 (46.7) 10.3 (37.5) 61.3 (35.8) 25.9 (58.3) 3 to 5 42.9 (16.7) 42.9 (20.0) 14. 3 (25.0) 71.4 (18.9) 9.1 (8.3) 6 to 9 50.0 (5.6) 25.0 (3.3) 25.0 (12.5) 75.0 (5.7) 25.0 (8.3) 10 or more 70.0 (19.4) 20.0 (6.7) 10.0 (12.5) 72.7 (15.1) 10.0 (8.3) Frequency of social evenings in neighborhood Never 46.2 (34.3) 38.5 (37.0) 15.4 (44. 4) 63.0 (11.4) 10.0 (18.2) Once/year to once/month 58.3 (20.0) 41.7 (18.5) 0 (0) 69.2 (15.9) 20.0 (18.2) Several times/month to once or twice/week 41.7 (28.6) 45.8 (40.7) 12.5 (33.3) 68.0 (59.1) 12.5 (27.3) Almost every day 66.7 (17.1) 11.1 (3.7) 22.2 ( 22.2) 90.0 (13.6) 50.0 (36.4) People in neighborhood are willing to help their neighbors Strongly disagree 83.3 (15.2) 16.7 (5.0) 0 (0) 83.3 (11.4) 60.0 (30.0) Somewhat disagree 50.0 (15.2) 40.0 (20.0) 10.0 (12.5) 63.6 (15.9) 22.2 (20.0) Somewhat agree 50.0 (51.5) 38.2 (65.0) 11.8 (50.0) 74.3 (59.1) 14.3 (40.0) Strongly agree 54.5 (18.2) 18.2 (10.0) 27.3 (37.5) 50.0 (13.6) 9.1 (10.0) 64 Road conditions within half block Major repair work needed Minor repair work needed No repa ir work needed Blight present within half block Junk pile or illegal dumping present within half block People in neighborhood get along with each other Strongly disagree 100.0 (9.7) 0 (0) 0 (0) 100.0 (7.3) 100.0 (27.3) Somewhat disagree 80.0 (25.8) 10.0 (5.0) 10.0 (12.5) 90.9 (24.4) 37.5 (27.3) Somewhat agree 43.8 (45.2) 43.8 (70.0) 12.5 (50.0) 60.6 (48.8) 14.3 (36.4) Strongly agree 42.9 (19.4) 35.7 (25.0) 21.4 (37.5) 53.3 (19.5) 7.1 (9.1) I live in a close - knit neighborhood Strongly disag ree 81.8 (28.1) 18.2 (8.7) 0 (0) 72.7 (19.5) 40.0 (44.4) Somewhat disagree 30.8 (12.5) 53.8 (30.4) 15.4 (33.3) 61.5 (19.5) 9.1 (11.1) Somewhat agree 48.3 (43.8) 37.9 (47.8) 13.8 (66.7) 60.0 (43.9) 12.0 (33.3) Strongly agree 62.5 (15.6) 37.5 (13.0) 0 (0) 63.6 (17.1) 12.5 (11.1) People in neighborhood share same values Strongly disagree 80.0 (32.0) 20.0 (10.5) 0 (0) 80.0 (22.9) 44.4 (40.0) Somewhat disagree 40.0 (24.0) 40.0 (31.6) 20.0 (42.9) 66.7 (28.6) 25.0 (30.0) Somewhat agree 40.9 (36.0) 45.5 (52.6) 13.6 (42.9) 62.5 (42.9) 8.7 (20.0) Strongly agree 50.0 (8.0) 25.0 (5.3) 25.0 (14.3) 50.0 (5.7) 25.0 (10.0) Table 10 : Neighborhood Spatial Factors and Cohesion Block group median home value (In thousands of dollars) Avera ge Walk Score Block group average vacancy (%) Block group average housing density (In hundreds of housing units/sq mile) Friends in Neighborhood None 39.09 40.11 10.46 17.63 1 or 2 42.01 38.35 8.88 18.98 3 to 5 42.21 49.50 9.63 23.80 6 to 9 41.35 45.25 8.32 22.27 10 or more 34.38 37.64 11.18 23.94 Frequency of social evenings in neighborhood Never 37.91 41.74 11.29 18.34 Once/year to once/month 42.94 37.62 8.33 19.96 Several times/month to once or twice/week 41.0 36.76 8.99 21.82 Almost e very day 38.14 53.60 7.94 24.85 People in neighborhood are willing to help their neighbors Strongly disagree 44.82 43.33 6.48 20.62 Somewhat disagree 36.45 49.09 8.66 13.97 65 Block group median home value (In thousands of dollars) Average Walk Score Block group average vacancy (%) Block group average housing density (In hundreds of housing units/sq mile) Somewhat agree 39.20 35.89 10.41 11.17 Strongly agree 44.89 46.92 10.16 24.12 People in neighborhood get along with each other Strongly disagree 44.07 51.00 4.52 25.90 Somewhat disagree 44.28 53.73 5.97 25.43 Somewhat agree 39.65 33.91 10.47 18.19 Strongly agree 39.92 45.07 11.37 20.29 I live in a close - knit neighborhood Strongly disagree 48.21 54.09 6.53 25.62 Some what disagree 45.02 46.15 9.45 22.07 Somewhat agree 36.67 33.40 9.90 18.68 Strongly agree 43.38 49.18 11.95 20.69 People in neighborhood share same values Strongly disagree 41.74 49.50 6.70 25.90 Somewhat disagree 42.97 41.87 7.96 21.74 Somewhat agree 43.29 38.50 10.02 20.82 Strongly agree 24.98 42.50 16.99 22.81 5.4 Linear Regression Models These anecdotal observations helped to delineate certain noticeable patterns from the survey collection and landscape factor measuring process. While these patterns matched many of my preconceptions , specifically that levels of trust and cohesion varied by landscape factor , it was important to run statistical models in order to clarify whether these trends might be attributable to sampling variation and whet her they persisted after controlling for other landscape factors or socio - economic and demographic characteristics of respondents. As previously mentioned, the following linear regression model was used : 66 where Y , the dependent variable, is a series of measures of either neighbor trust or neighborhood cohesion for respondent i, L is a vector of the landscape factors described above, and X is a series of s ocio - economic and demographic control variables. is a vector of coefficients which represent the association between landscape factors and the selected measures of neighborhood trust or cohesion , while is a vector of coefficients representing the relationship between the dependent variable and selected control variables . is the error term. I ran a variety of analyses to determine whether or not a relationship exists between landscape factors, neighborhood trust and neighborhood cohes ion. The results of the initial robust regression model , shown in the second column of Table 11 , indicated that living in a townhouse had a significant negative relationship with neighborhood trust. However, after including control variables, none of the l andscape factors were significantly related this suggested that individual - level factors may play a more important role in determining levels of trust than landscape factors, or perhaps that in the present study public housing residents with particular d emographic or socio - economic traits may be more likely to reside in neighborhoods with varying levels of landscape amenities or disamenities . Furthermore , the R - squared value without controls was .198, but with the addition of controls, this value increase d to .441 . This means that individual demographic characteristics account ed for a greater percentage of the variation in trust than landscape factors. This trend was similar for neighborhood cohesion. Because the initial robust regression analysis did not show many significant findings, the model was altered. Rather than using robust regression, the standard errors were clustered by block group to account for the fact that respondents in the same neighborhood are exposed to similar landscape conditions. Thi s changed the p - value of some of the variables, but did not result in any conclusive findings. To confirm that these findings were accurate and that there is 67 little evidence to support the notion that landscape and social capital are related, I tested for multicollinearity between the independent variables. This indicated that housing density and local integration had standard errors inflated by up to 25 times their actual value. These variables were removed to reduce the likelihood of multicollinearity fro m the model; however, there were no substantive Global integration was the only significant landscape factor related to the neighborhood cohesion index ( - 4.760 coefficient ), which was scaled from zero to six . The findings , i n other words, revealed that as global integration decreases, neighborhood cohesion increases. - statistic of 0.003 (p<.01 ). This result is contrary to my hypothesis that greater locational isolation would als o show lower neighborhood cohesion . However, because global integration refers to connections from the entire city, it is possible that this negative relationship indicates that untrustworthy people from outside a neighborhood are more likely to travel int o that neighborhood. This spatial integration may promote criminal activity, which would explain why its value relates to lower neighborhood cohesion. C ertain control variables were significant in the robust model, the clustered model or in both . In the fi rst linear regression model that used robust standard errors, access to a personal vehicle was significant at the p<.05 level with a coefficient indicating that having a vehicle related to a .697 increase in neighbor trust, measured from zero to three, wit h three indicating high trust. This was contrary to my hypothesis that, without a vehicle, residents would be more likely to walk their neighborhoods and thereby have increased opportunities for social interactions. Residents who completed some education b eyond a high school degree showed a significant negative relationship in both models for trust (p<.05 in the robust model and p<.01 in the clustered model). Having more education demonstrated 68 ignificant for its relationship to trust in both models (p<.05) with a coefficient of - .654. These results for education and employment may be explained by the fact that this study population was made up of low - income public housing residents. While I pres umed that having more education and a job would relate to increases in social capital, I did and perception of his or her neighbors/neighborhood. In other wo rds, individuals higher in these traits are possibly isolated because of these qualities. Finally, income $20,000 - $29,999 was significant in both models (p<.05) and had a coefficient of 2.847 for the neighborhood cohesion index scaled zero to six. Just six respondents were in this income category, earning more than 72 other participants. This significant positive relationship may, somewhat obviously, conclude that those earning the most within an overall low - income group have more positive perceptions of th eir neighborhood. T he null hypothesis that there is no significant relationship between landscape factors, neighborhood trust and ne ighborhood cohesion was rejected in this study. The F - statistic for trust and landscape factors, including controls, was si gnificant at the p<.01 level. For neighborhood cohesion and landscape factors, with controls, the F - statistic was also significant (0.001). This means that landscape factors account for around 20% of the variation in neighbor trust and neighborhood cohesi on. However, this research model could not effectively conclude which landscape factors drive this variation. Future studies on this topic should ensure a larger sample size and greater variation in geographic location, possibly across multiple cities, to more robustly evaluate the relationship between landscape factors and social capital . 69 Table 11 : Linear Regression Results No Controls Controls, Robust Controls, Clustered No Controls Controls, Robust Controls, Clustered Dependent Variable Trust in Neighbors Trust in Neighbors Trust in Neighbors Neighborhood Cohesion Index Neighborhood Cohesion Index Neighborhood Cohesion Index Townhouse - 1.172* - 0.771 - 0.771** 0.124 0.619 0.619 (0.46) (0.49) (0.21) (1.01) (1.08) (1.06) High Ris e - 0.495 0.135 0.135 0.775 2.879 2.879 (0.43) (0.44) (0.48) (1.73) (1.79) (2.19) Block group median housing value - 0.0681 - 0.115 - 0.115 - 0.0719 - 1.318 - 1.318 (0.39) (0.43) (0.39) (0.90) (1.05) (0.96) Block group avg Walk Score - 0.000692 - 0.000715 - 0. 00072 - 0.0214 - 0.0127 - 0.0127 (0.01) (0.01) (0.01) (0.02) (0.02) (0.03) Block group avg percent vacant 0.00172 - 0.0206 - 0.0206 0.0991 0.00935 0.00935 (0.02) (0.02) (0.02) (0.05) (0.07) (0.08) Block group avg housing density (1,000 houses/sq mile) 0.3 32 0.0264 0.0264 0.512 - 0.569 - 0.569 (0.24) (0.30) (0.22) (0.94) (1.01) (1.13) Connectivity - 0.0364 - 0.0137 - 0.0137 0.0101 0.0409 0.0409 (0.04) (0.04) (0.02) (0.06) (0.07) (0.06) Global integration - 1.264 - 0.568 - 0.568 - 2.809 - 4.760* - 4.760** (0.84 ) (0.98) (1.11) (1.74) (2.01) (1.22) Local integration 0.469 0.0949 0.0949 0.831 1.363 1.363 (0.43) (0.46) (0.33) (0.86) (1.00) (1.02) No road repairs needed - 0.284 - 0.26 - 0.26 0.439 0.468 0.468 (0.22) (0.25) (0.30) (0.42) (0.44) (0.43) Blig ht within half block - 0.112 - 0.451 - 0.451* - 0.261 - 0.534 - 0.534 (0.23) (0.23) (0.17) (0.43) (0.51) (0.56) 70 No Controls Controls, Robust Controls, Clustered No Controls Controls, Robust Controls, Clustered Dependent Variable Trust in Neighbors Trust in Neighbors Trust in Neighbors Neighborhood Cohesion Index Neighborhood Cohesion Index Neighborhood Cohesion Index Junk pile within half block - 0.225 0.0721 0.0721 - 0.221 - 0.288 - 0.288 (0.32) (0.36) (0.26) (0.49) (0.57) (0.50) Lived at address 2 years or longer 0.399 0.399 1.177 1.177 (0.36) (0.37) (0.59) (0.60) Has Access to Personal Vehicle 0.697* 0.697 0.487 0.487 (0.30) (0.38) (0.53) (0.44) Black - 0.18 - 0.18 0.202 0.202 (0.32) (0.33) (0.80) (0.43) Income $10,000 - $19,999 0.427 0.427 - 0.0269 - 0.0269 (0.30) (0.37) (0.63) (0.90) Income $20,000 - $29,999 0.334 0.334 2.847* 2.847* (0.56) (0.56) (1.10) (1.24) Income $30,000 - $39,999 0.168 0.168 - 0.764 - 0.764 (0.79) (0.14) (0.69) (0.50) Attended some co llege - 0.508* - 0.508** - 0.148 - 0.148 (0.20) (0.15) (0.44) (0.67) Working - 0.654* - 0.654* - 1.037 - 1.037 (0.27) (0.28) (0.65) (0.81) _cons 2.259 2.596 2.596 2.504 7.754 7.754 (1.57) (1.73) (1.65) (4.22) (4.96) (4.19) N 63 63 63 68 66 66 F - s tatistic 0.0298* 0.001** .1563 .003** R - sq 0.198 0.441 0.441 0.188 0.336 0.336 Notes: Standard errors in parentheses * p<0.05, ** p<0.01, *** p<0.001 71 Chapter 6. Discussion and Conclusion in measures of social capital could be explained by certain landscape factors. P ossess ing high amounts of social capital is certainly beneficial to the individual as well as whole communities , prompting a better quality of life through the perceived abilit y to access more resources. It seems that few else stand to gain more from increased social capital than public housing residents, especially those in cities experiencing prolonged decline and poverty such as Saginaw. For this reason, I hoped to better und erstand how the built environment and social capital are associated in order to provide evidence that certain urban planning practices and policy initiatives can either deter or advance the social networks within a local context. In order to measure these phenomena, I developed a research framework based on the literature and This framework understood social capital as both a tool and a product; a mean to acquire more of another good a nd a good in itself . Therefore, my methodology for this study was created to reflect the idea that landscape factors may contribute to higher levels of social capital, and that this in itself is good, but that it may also lead to higher civic participation . To test this fram ework, I surveyed public housing residents in Saginaw, Michigan to measure their trust in people generally, trust in their neighbors and trust in people of other races. I also collected data on their civic participation and demographics. These data were combined with landscape data that I gathered using the survey instrument, the US Census Bureau and a street segment mapping platform generated by space syntax based program, Depthmap . All of the relationships between my dependent variables (neighbor trust and neighborhood cohesion) and landscape variables were anecdotally reviewed 72 for notable trends. This helped me to identify mistakes in coding and to adjust my variables before using them in a linear regression analysis. I tested these relationship s with a linear regression model using robust standard errors to account for heteroscedasticity. When this model did not show many significant findings, the model was adjusted and standard errors were clustered by block group instead. When this model also did not indicate many significant relationships, I tested for multicollinearity between independent variables and removed those with inflated standard errors. None of the landscape factors displayed significant relationships with either trust or cohesion, though s ome of the controls such as access to a vehicle, education, employment and income were significant. This can likely be explained by the low spatial had been surveyed, it is possible that individual landscape factors would have shown significance, similar to the findings for the control variables. Despite finding no significant relationships between specific landscape factors and measures of social capital, th is thesis report did find that statistically significant regression models had an increased R - squared value when landscape factors were included. In other words, it seems that the built environment is related to different levels of social capital, but this study cannot discern which variables determine this. The reason, as previously mentioned, is almost Overall, this research should do well to inform future studies through its results, d esign and limitations. Perhaps the most important limitation of this study was the limited response collection. O nce the location s for respondents were collected and mapped after Institutional Review Board approval and completion of the research design I noticed that the housing sites were highly clustered, thus reducing the ability to measure variation in trust and cohesion effectively. This 73 was unanticipated. I was aware that residents in the townhouses would be exposed to identical landscape factors ; however, after preliminary conversations with the Saginaw Housing Commission, I was led to believe that the scattered sites would provide sufficient geographic variation to capture a variety of landscape factors. As it turned out, these scattered sites w ere located in a limited number of neighborhoods, thus reducing the variation in environmental conditions . Therefore, future studies should ensure that their study population is varied in individual locations to more accurately discern if landscape factors are related to measures of social capital. An increase in spatial variation would help to better discern whether differences in social capital are most influenced by environment or by individually held characteristics such as race and education. If studyi ng one community rather than a national dataset, future researchers should attempt to garner responses from various neighborhoods, street segments and locations that offer clear differences in the built environment. In this study in particular, this would have meant expanding the study population to all Saginaw residents rather than limiting the population to public housing residents. The second encountered problem was that each of the three types of trust were measured differently. This was due to the ch oice to take questions directly from surveys used in past social capital academic studies. The decision to not alter the questions was likely too restrictive and these measures should have been adapted to more closely match the overall research question. F races was a scale index. The categorical dependent variables also m ade the regression analysis more difficult to interpret since the possible responses (such as ) have a natural order but not a natural scale. In other words, it may be more problematic 74 to have a community where neighbors do not trust each other at all than it is favorable to have a community with high trust levels. For instance, it could be possible that very low levels of social capital contribute to undesirable conditions such as crime, poverty and the mental health prob lems that presumably arise due to prolonged social isolation. In contrast, the benefits to very high levels of trust may improve neighborhood conditions only slightly. Therefore, t his study and those that follow should use a scale to measure each type of t rust from 0 - 100. This would help to more accurately measure the variation in levels of trust from one person or group to the next. Finally, this study was limited by the time period allotted for completion and most certainly excluded landscape factors th at may have a relationship with social capital. For example, the presence of landscaping, bars on windows, parking lots and various other factors likely would have been included in the study if the study period had been longer in duration. It should be not ed that an increase in variables would have to be accompani ed by a larger population size to account for the loss of degrees of freedom due to the inclusion of additional control variables. This could have been accomplished in this study by sampling all Sa ginaw residents rather than only those in public housing. In this research, I was challenged by the difficulty in recruiting responses despite offering a cash incentive. Many of the units I surveyed were vacant during one mailing wave but were occupied for another. Even when units were occupied, very few residents chose to fill out and return the questionnaire. I utilized a mail survey to account for the fact that this study population was low - income and unlikely to complete an online survey. However, if fu ture studies examine a more diverse population, they could more easily incorporate both an online and mail survey to increase response rates. 75 Although this study lacks definite findings, social capital as a theory is still an important topic for research ers and the wider public to understand. This is especially true as the concept relationship with measures of social capital can be effectively measured and replicated, this could help to promote comm unities with greater cohesion, trust, economic stability and health . These findings would help urban planners, landscape architects, local politicians and developers have a clear indication of the effects that their land use decisions have on the well - bein g of the people who live within their scope of work. It is almost obvious community, neighborhood or block. The goal then is to design a research method that effectively measures and d escribes these factors so that certain planning practices can either be emphasized or reduced. In this way, those involved in the formation of the built environment could advocate for practices that yield the types of communities that are highly desirable to a wider range of people, thu s becoming attractive places for people to live, work and play. 76 APPENDIX 77 APPENDIX Survey: Built Environment & Trust Researchers from Michigan State University would like to learn about Sag inaw public housing of trust and their neighborhood. We are asking for 5 - 10 minutes of your time to participate in a research survey to understand your thoughts and feelings about trust and your neighborhood. You may not benefit personal ly from being in this study. However, we hope that, in the future, other people might benefit from this study as it will provide data on social trust and neighborhoods, which could be used by local leaders and officials responsible for housing . Rest assur ed, all of the information that you provide will be kept confidential and will only be used to help us understand the issues mentioned directly above. Your name, address, and any other identifying information will NOT be associated with any specific survey responses in any publications, presentations, or other materials. to this information. Thank you for taking the time to consider this request for your participation in our survey. Participation is voluntary. You may choose not to answer certain questions, to discontinue your participation at any time or not to participate at all, without consequence. If you have any concerns or questions about this study, please contact Zach Vega (517 - 432 - 8800; vegazach@msu.edu ) or Dr. Noah Durst (517 - 353 - 3184; durstnoa@msu.edu). If you have any questions or concerns about your role and rights as a research participant in this study, you may contact, anonymously if you wish, the MSU Human Research Protection Program at 517 - 355 - 2180, or by email: irb@msu.edu. If you respond to this survey and send it back in the return envelope provided, y ou will receive a $25 gift card mailed t o this same address. By submitting this survey, you are consenting to participate. SECTION A. LOCATION: The following questions ask about where you live. 1. Do you currently live at [mail merge address]? Yes No 2. How long have you lived at [mail merge addr ess]? Less than 1 year 1 year 2 to 5 years More than 5 years 3. Did you choose to live at [mail merge address] or were you assigned to live there? I chose to live at [mail merge address] 78 I was assigned to live at [mail merge address] without a ch oice SECTION B. SOCIAL TRUST: people. 4. Generally speaking, too careful with people? People can be trusted be too careful It depends For each of the following groups , would you say that you trust them a lot, some, only a little, or not at all? 5. People in your neighborhood Trust them a lot Trust them some Trust them only a little Trust them not at all 6. White people Trust them a lot Trust them some Trust them only a little Trust them not at all 7. African American or Black people Trust them a lot Trust them some Trust them only a little Trust them not at all 8. Asian people 79 Trust them a lot Trust them some Trust them only a little Trust them not at all 9. Hispanic or Latino people Trust them a lot Trust them some Trust them only a little Trust them not at all SECTION C. COMMUNITY : The following questions address community relationships. 10. How many friends do you have who live in this neighborhood? None 1 or 2 3 to 5 6 to 9 10 or more 11. How often do you spend a social evening with someone lives in your neighborhood? Almost every day Once or twice a week Several times a month About once a month Several times a year About once a year Never Please read the following statements about your neighborhood and say whether you strongly agree, somewhat agree, somewhat disagree, or strongly disagree . 12. People in my neighborhood are willing to help their neighbors. 80 Strongly agree Somewhat agree Somewhat disagree Strongly disagree 13. People in my neighborhood get along with each other. Strongly agree Somewhat agree Somewhat disagree Strongly di sagree 14. I live in a close - knit neighborhood. Strongly agree Somewhat agree Somewhat disagree Strongly disagree 15. People in my neighborhood share the same values. Strongly agree Somewhat agree Somewhat disagree Strongly disagree now SECTION D. ACTIVITIES : The following questions address community activities. 16. Have you or a neighborhood member attended a meeting of a block or neighborhood group about a neighborhood problem or neighborhood improvement? Yes No 17. Have you o r anyone in your household gotten together with neighbors to do something about a neighborhood problem or to organize neighborhood improvement? 81 Yes No 18. Have you or anyone in your household spoken with a local politician like you Ward committeepe rson or an elected official like your alderperson about a local problem? Yes No 19. Have you or anyone in your household talked to a person or group causing a problem in the neighborhood? Yes No 20. Over the past 12 months, have you or a member of your household volunteered or helped out with activities in your community? Yes No SECTION E. NEIGHBORHOOD CONDITIONS : The following questions address physical conditions in your neighborhood. 21. What is the condition of the streets w ithin a half a block of this building? Do these streets need major repairs, minor repairs or no repair work? Major repair work Minor repair work No repair work 22. Are there any vandalized or abandoned buildings within half a block of this building ? Is there more than one vandalized or abandoned building? Yes, one Yes, more than one No 23. Are there any junk piles/illegal dumping within half a block of this building ? 82 Yes No SECTION F. HOUSEHOLD : The following questions ask about t he characteristics of your household or family. 24. Do you have access to a personal vehicle? Yes No 25. Are you or is any member of your household of Hispanic, Latino, or Spanish descent? Yes No 26. What race(s) are represented by the people who live in your household (check all that apply) Asian American Indian/Alaska Native Black/African American Native Hawaiian/Other Pacific Islander White/Caucasian Other 27. What is your yearly household income? Less than $10,000 $10,000 - $19,999 $20,000 - $29,999 $30,000 - $39,999 $40,000 - $49,999 $50,000 or more 28. What is the highest grade of school or year of college you have completed? Less than high school (Grade 11 or less) High school diploma (including GED) Some college 83 or specialized technical training Some graduate training Graduate or professional degree 29. Which of the following best describes your current employment status? Working Temporarily laid off Unemployed Retired Permanently disabl ed Homemaker Student INSTRUCTIONS FOR RETURNING SURVEY Thank you for completing this survey and helping with this study. Please place this completed survey inside the return envelope provided. You do not need to include postage on the return en velope. Place the return envelope containing the completed survey in an outgoing mailbox. The MSU rese arch team will send you your $25 gift card to the same address the survey was mailed to as a thank you for helping with this study. 84 B IBLIOGRAPHY 85 BIBLIOGRAPHY Blanchard, A., & Horan, T. (1998). Virtual Communities and Social Capital. Social Science Computer Review , 16 (3), 293 307. http://doi.org/10.1177/08944393980160030 6 Bothwell, S. E., Gindroz, R., & Lang, R. E. (1998). Restoring Community through Traditional Neighborhood Design: A Case Study of Diggs Town Public Housing. Housing Policy Debate , 9 (1), 89 114. http://doi.org/10.1080/10511482.1998.9521287 Bourdieu, P. (1986). The Forms of Capital. In J. Richardson (Ed.), Handbook of Theory and Research for the Sociology of Education (pp. 241 258). Westport, CT: Greenwood. Bourdieu, P., & Passeron, J. (1970) . Reproduction in Education, Society and Culture. London: Sage. Boxman, E. A. W., De Graaf, P. M., & Flap, H. D. (1991). The Impact of Social and Human Capital on the Income Attainment of Dutch Managers. Social Networks , 13 (1), 51 73. http://doi.org/10.1016/0378 - 8733(91)90013 - J Brehm, J., & Rahn, W. (1997). Individual - Level Evidence for the Causes and Consequences of Social Capital. American Journal of Political Science , 41 (3), 999 1023. http://doi.org/10.2307/2111684 Cattell, V. (2001). Poor People, Poor Places, and Poor Health: The Mediating Role of Social Networks and Social Capital, 52 (10), 1501 1516. http://doi.org/10.1016/S0277 - 9536(00)00259 - 8 Coleman, J.S. (1988). Social Capital in the Creation of Human Capital. American Journal of Sociology, 94 , S95 - S120 de Souza Briggs, X. (1998). Brown Kids in White Suburbs: Housing Mobility and the Many Faces of Social Capital. Housing Policy Debate , 9 (1), 177 221. http://doi.org/10.1080/ 10511482.1998.9521290 DiPasquale, D., & Glaeser, E. L. (1999). Incentives and Social Capital: Are Homeowners Better Citizens? Journal of Urban Economics , 45 (2), 354 384. http://doi.org/10.1006/juec. 1998.2098 Esri & United States Census Bureau, Census 2010 Summary File 1. (2019). Community Profile. Retrieved April 1, 2019 from the Business Analyst Online database. Esri & United States Census Bureau, Census 2010 Summary File 1. (2019). Demographic and Income Profile. Retrieved April 1, 2019 from the Business Analyst Online database. 86 Esri & United States Census Bureau, 2012 - 2016 American Community Survey. (2019). ACS Housing Summary. Retrieved April 1, 2019 from the Business Analyst Online database. Esri & United States Census Bureau, 2012 - 2016 American Community Survey. (2019). ACS Population Summary. Retrieved April 1, 2019 from the Business A nalyst Online database. Projects Disrupt Local Social Relations. In R. Dilworth (Ed.), Social Capital in the City: Community and Civic Life in Philadelphia (pp. 159 176). Philadelphia: Temple University Press. Retrieved from http://www.jstor.org/stable/j.ctt14bt0jz.14 Jacobs, J. (1961). The Death and Life of Great American Cities . New York: Random House Kawachi, I., Kennedy, B. P., Lochner, K., & Prothrow - Smith, D. (1997). Social Capital, Income Inequality, and Mortality. American Journal of Public Health , 87 (9), 1491 1498. http://doi.org/10.2105/AJPH.87.9.1491 Leyden, K. M. (2003). Social Capital and the Built Environment: The Importance of Walkable Neighborhoods. American Journal of Public Health , 93 (9), 1546 1551. http://doi.org/ 10.2105/AJPH.93.9.1546 Lindstrom, M., Merlo, J., & Ostergren, P. - O. (2002). Individual and Neighbourhood Determinants of Social Participation and Social Capital: A Multilevel Analysis of the City of Malmo, Sweden. Social Science & Medicine , 54 (2), 1779 1791. http://doi.org/10. 1016/S0277 - 9536(01)00147 - 2 Lochner, K., Kawachi, I., & Kennedy, B. P. (1999). Social Capital: A Guide to its Measurement. Health and Place , 5 (4), 259 270. http://doi.org/10.1016/S1353 - 8292(99)00016 - 7 Loury, G. (1977). A Dynamic Theory of Racial Income Differences. In P. A. Wallace and A. LeMund (Eds.), Women, Minorities, and Employment Discrimination (pp. 153 - 188). Lexington, MA: Lexington Books Macinko, J., & Starfield, B. (2001). The Utility of Social Capital in Research on Health Determinan ts. The Milbank Quarterly , 79 (3), 387 427. http://doi.org/10.1111/1468 - 0009.00213 Manson, S., Schroeder, J., Van Riper, D., & Ruggles, S. (2018). IPUMS National Historical Geographic Information Syst em: Version 13.0 [Database]. Minneapolis: University of Minnesota. http://doi.org/10.18128/D050.V13.0 McMillan, D. W., & Chavis, D. M. (1986). Sense of Community: A Definition and Theory. Journal of Comm unity Psychology , 14 (1), 6 23. http://doi.org/10.1002/1520 - 6629(198601)14:13.0.CO;2 - I 87 Muntaner, C., Lynch, J., & Smith, G. D. (2000). Social Capital and the Third Way in Public Health . Critical Public Health , 10 (2), 107 124. http://doi.org/10.1080/713658240 Portes, A. (2000). The Two Meanings of Social Capital. Sociological Forum , 15 (1), 1 12. http://doi.org/10.1023/A:1007537902813 Portes, A., & Landolt, P. (2000). Social Capital: Promises and Pitfalls of its Role in Development. Journal of Latin American Studies , 32 (2), 529 547. http://doi.org/10.1017 /S0022216X00005836 Portes, A., & Sensenbrenner, J. (1993). Embeddedness and Immigration: Notes on the Social Determinants of Economic Action. The American Journal of Sociology , 98 (6), 1320 1350. http://doi.org/10.1086/230191 Putnam, R. D. (2000). Bowling Alone: The Collapse and Revival of American Community . New York, NY: Simon & Schuster Paperbacks. Rahimi, S., Martin, M. J. R., Obeysekere, E., Hellmann, D., Liu, X., & Andri s, C. (2017). A Geographic Information System (GIS) - Based Analysis of Social Capital: Landscape Factors that Correlate with Trust. Sustainability , 9 (3), 365. http://doi.org/10.3390/su 9030365 Rahn, W., & Transue, J. E. (1998). Social Trust and Value Change: The Decline of Social Capital in American Youth, 1976 - 1995. Political Psychology , 19 (3), 545 565. http://doi.org/10.1111/0162 - 895X.00117 Sabin, E. P. (1993). Social Relationships and Mortality Among the Elderly. The Journal of Applied Gerontology , 12 (1), 44 60. http://doi.org/10.1177/073346489301200105 United States Census Bureau, 2013 - 2 017 American Community Survey. (2019). Retrieved April 1, 2019 from the American FactFinder database. Wacquant, L. J. D., & Wilson, W. J. (1989). The Cost of Racial and Class Exclusion in the Inner City. The Annals of the American Academy of Political and Social Science , 501 (1), 8 25. http://doi.org/10.1177/0002716289501001001 Woolcock, M. (1998). Social Capital and Economic Development: Toward a Theoretical S ynthesis and Policy Framework. Theory and Society , 27 (2), 151 208. http://doi.org/10.1023/A:1006884930135