Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Health, hidden homelessness, and gender : a multivariate analysis Ferguson, Max 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2018_november_ferguson_max.pdf [ 738.24kB ]
Metadata
JSON: 24-1.0372353.json
JSON-LD: 24-1.0372353-ld.json
RDF/XML (Pretty): 24-1.0372353-rdf.xml
RDF/JSON: 24-1.0372353-rdf.json
Turtle: 24-1.0372353-turtle.txt
N-Triples: 24-1.0372353-rdf-ntriples.txt
Original Record: 24-1.0372353-source.json
Full Text
24-1.0372353-fulltext.txt
Citation
24-1.0372353.ris

Full Text

  HEALTH, HIDDEN HOMELESSNESS, AND GENDER: A MULTIVARIATE ANALYSIS   by Max Ferguson  B.Sc.N., R.N. The University of Ottawa, 2011  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF PUBLIC HEALTH/MASTER OF SCIENCE IN NURSING in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  October 2018  © Max Ferguson, 2018  ii  The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis/dissertation entitled:  Health, Hidden Homelessness, And Gender: A Multivariate Analysis  submitted by Max Ferguson in partial fulfillment of the requirements for the degree of Master of Public Health/Master of Science in Nursing in The School of Population and Public Health  Examining Committee: Dr. Elizabeth Saewyc Supervisor  Dr. Victoria Bungay  Supervisory Committee Member  Dr. Sabrina Wong Supervisory Committee Member  Additional Examiner   Additional Supervisory Committee Members:  Supervisory Committee Member  Supervisory Committee Member iii  Abstract Research topic: My analyses were designed to evaluate the burden of physical and mental illness associated with different shelter strategies among homeless youth. I investigated the health outcomes of hidden homelessness versus other shelter strategies in youth across different genders – including male youth, female youth, and trans & gender diverse youth. Research questions: What is the association between using hidden homelessness shelter strategies and health outcomes among homeless and street involved youth living in British Columbia? Is there a difference in the relationship between hidden homelessness and health outcomes for people of different genders, and, if so, what is the nature of this relation? Participants: This cross-sectional descriptive study uses the British Columbia Homeless and Street Involved Youth Survey (HSIY). Data were collected by the McCreary Centre Society. The total sample is comprised of 681 homeless or street involved youth from 13 communities across British Columbia. Participants ranged in age from 12 to 19. Methods: I conducted a series of logistic regression models to determine the statistical effect size of the relation of hidden homelessness and health on youth. The outcome health variables included poor or fair self-rated general health, poor or fair self-rated mental health, having at least one mental health condition, having foregone mental health care, having foregone medical health care, having experienced an injury serious enough to require medical attention in the past year, experiencing extreme levels of stress, experiencing extreme levels of despair, self-harming in the last year, having considered suicide in the last year, and having attempted suicide in the last year. I investigated whether the effect size differed for youth of different genders.   Results: Youth who have used hidden homelessness shelter strategies reported significantly worse health outcomes except for experiences of extreme despair. Compared to male youth, iv  female youth were more likely to experience foregone medical care and have experienced an injury serious enough to require medical attention. Trans & gender diverse youth had worse health outcomes associated with hidden homelessness in descriptive analyses, although the sample was insufficient to include in the models with interaction terms.   v  Lay Summary This work was designed to test the commonly held assumption that youth who stay in shelters or on the street have worse health than hidden homeless youth who are homeless in other ways including couch surfing; staying in hotels, motels, single room occupancy buildings, or hostels; staying in a tent or car; living in a transition house; squatting in an empty or abandoned building; staying in an extreme weather shelter; or being homeless in another, invisible way.  My work showed that this assumption was, in fact, incorrect and that hidden homeless youth had poorer access to health services, as well as poorer physical and mental health outcomes. Furthermore, the association between hidden homelessness and poorer health outcomes was more exaggerated in female youth in comparison to male youth. An advisory committee of youth with lived experience emphasized the importance of large scale societal change by strengthening programs for homeless youth, including creating services for hidden homeless youth within existing services for homeless people, bolstering affordable housing, and giving young people the tools to find the services they might need before they experience homelessness.  vi  Preface This work was a secondary data analysis of the Homeless and Street Involved Youth Survey conducted by the McCreary Centre Society.  I designed the plan of analysis with help from Drs. Elizabeth Saewyc, Vicky Bungay, and Sabrina Wong, my supervisor and thesis committee members. I conducted all descriptive analyses, logistic regression models, and multiple imputation. Dr. Gu Li confirmed my multiple imputation code in R. I consulted with Mike Marin on the statistical methods that I used. My thesis work did not require ethics approval as I did not collect any primary data. My work has not been published.  vii  Table of Contents  Abstract ......................................................................................................................................... iii!Lay Summary .................................................................................................................................v!Preface ........................................................................................................................................... vi!Table of Contents ........................................................................................................................ vii!List of Tables ............................................................................................................................... xii!List of Figures ............................................................................................................................. xiv!List of Symbols .............................................................................................................................xv!List of Abbreviations ................................................................................................................. xvi!Glossary ..................................................................................................................................... xvii!Acknowledgements .................................................................................................................. xviii!Dedication ................................................................................................................................... xix!Chapter 1: Introduction ................................................................................................................1!1.1! Statement of Purpose ...................................................................................................... 3!1.2! Research Questions ......................................................................................................... 4!1.3! Gender ............................................................................................................................. 4!1.3.1! Women and Homelessness ..................................................................................... 5!1.3.2! Transgender People and Homelessness .................................................................. 7!1.4! Hidden Homeless Youth ................................................................................................. 9!1.5! My Connection to Hidden Homelessness ....................................................................... 9!Chapter 2: Literature Review .....................................................................................................10!2.1! Methods......................................................................................................................... 10!viii  2.2! Prevalence of youth hidden homelessness and shelter strategy trends ......................... 11!2.3! Pathways to homelessness ............................................................................................ 12!2.4! What is hidden homelessness? ...................................................................................... 12!2.5! Which groups are made ‘hidden’ by current enumeration policies? ............................ 13!2.6! What ages were conceptualized as youth or adolescents? ............................................ 14!2.7! Subgroups ..................................................................................................................... 14!2.8! Context of youth hidden homelessness ......................................................................... 15!2.8.1! Shelter strategies ................................................................................................... 17!2.9! Gender and hidden homelessness ................................................................................. 17!2.10! How does hidden homelessness affect health? ............................................................. 18!Chapter 3: Methodology ..............................................................................................................20!3.1! Research Questions ....................................................................................................... 20!3.2! Data Source ................................................................................................................... 20!3.3! Explanatory variables .................................................................................................... 22!3.3.1! Hidden homelessness shelter strategies ................................................................ 22!3.4! Potential Confounders or Effect Modifiers ................................................................... 23!3.4.1! Gender ................................................................................................................... 23!3.4.2! Age ........................................................................................................................ 24!3.4.3! Location ................................................................................................................ 24!3.5! Outcome Health Variables ............................................................................................ 25!3.6! Model Building ............................................................................................................. 27!3.6.1! Hypotheses ............................................................................................................ 27!3.7! Missing Data ................................................................................................................. 29!ix  3.8! Multiple Imputation in Homelessness Research ........................................................... 30!3.9! Limitations .................................................................................................................... 31!3.10! Gender in Research ....................................................................................................... 31!3.11! Advisory Committee ..................................................................................................... 33!Chapter 4: Results ........................................................................................................................34!4.1! Descriptive Analyses .................................................................................................... 34!4.2! Adjusted Odds Ratios of Hidden Homelessness on Various Health Outcomes ........... 38!4.3! Multiple Imputation ...................................................................................................... 39!4.3.1! Descriptive Analyses ............................................................................................ 39!4.3.2! Adjusted Odds Ratios of Hidden Homelessness on Various Health Outcomes ... 43!4.4! Multiple Imputation ...................................................................................................... 44!4.5! Advisory Committee Observations & Recommendations ............................................ 45!Chapter 5: Discussion: .................................................................................................................47!5.1! Summary of Findings .................................................................................................... 48!5.2! Contributions to the Literature ...................................................................................... 49!5.3! Limitations .................................................................................................................... 51!5.4! Implications ................................................................................................................... 51!5.5! Conclusions ................................................................................................................... 55!References .....................................................................................................................................57!Appendices ....................................................................................................................................61!Appendix A -  Literature Review Inclusion & Exclusion Criteria ........................................... 61!Appendix B -  Model Differences ............................................................................................. 69!B.1! Original Dataset ........................................................................................................ 69!x  B.2! Imputed Dataset ........................................................................................................ 70!Appendix C Model Building: Original Dataset ........................................................................ 71!C.1! Self-Reported Poor or Fair General Health .............................................................. 71!C.2! Self-Reported Poor or Fair Mental Health ................................................................ 71!C.3! Mental Health Condition ........................................................................................... 72!C.4! Foregone Medical Care ............................................................................................. 72!C.5! Foregone Mental Health Care ................................................................................... 73!C.6! Injury ......................................................................................................................... 74!C.7! Stress ......................................................................................................................... 74!C.8! Despair ...................................................................................................................... 75!C.9! Self-Harm .................................................................................................................. 75!C.10! Consider Suicide ....................................................................................................... 76!C.11! Attempt Suicide ........................................................................................................ 77!Appendix D Model Building: Imputed Dataset ........................................................................ 77!D.1! Self-Reported Poor or Fair General Health .............................................................. 77!D.2! Self-Reported Poor or Fair Mental Health ................................................................ 78!D.3! Mental Health Condition ........................................................................................... 78!D.4! Foregone Medical Care ............................................................................................. 79!D.5! Foregone Mental Health Care ................................................................................... 80!D.6! Injury ......................................................................................................................... 80!D.7! Stress ......................................................................................................................... 81!D.8! Despair ...................................................................................................................... 81!D.9! Self-Harm .................................................................................................................. 82!xi  D.10! Consider Suicide ....................................................................................................... 83!D.11! Attempt Suicide ........................................................................................................ 83!Appendix E Unadjusted Odds Ratios ....................................................................................... 85!E.1! Unadjusted Odds Ratios of Hidden Homelessness on Various Health Outcomes ... 85!E.2! Unadjusted Odds Ratios of Hidden Homelessness on Various Health Outcomes ... 85! xii  List of Tables Table 3.1 Variables and variable dichotomization ....................................................................... 27!Table 3.2 Number and percentage missing for each variable of interest ...................................... 29!Table 4.1 Demographic breakdown of sample. ............................................................................ 34!Table 4.2 Percentage of variables of interest in hidden homeless sample versus the remainder of the sample and chi square analyses p value. ................................................................................. 35!Table 4.3 Percentage of variables of interest among different genders and chi square analyses p value. ............................................................................................................................................. 36!Table 4.5 Adjusted odds ratios of hidden homelessness on various health outcomes .................. 38!Table 4.6 Demographic breakdown of imputed sample ............................................................... 39!Table 4.7 Percentage of variables of interest in hidden homeless sample versus the remainder of the sample in imputed sample and chi square analyses p value ................................................... 40!Table 4.8 Percentage of variables of interest among different genders in imputed sample and chi square analyses p value. ................................................................................................................ 41!Table 4.10 Adjusted odds ratios of hidden homelessness on various health outcomes in imputed sample ........................................................................................................................................... 43!Table A.1 Literature review inclusion & exclusion criteria ......................................................... 68!Table C.10 Model building: Consider suicide .............................................................................. 76!Table C.11 Model building: Attempt suicide ............................................................................... 77!Table D.1 Model building: Self-reported poor or fair general health ........................................... 78!Table D.2 Model building: Self-reported poor or fair mental health ............................................ 78!Table E.1 Unadjusted odds ratios of hidden homelessness on various health outcomes………. 85 xiii  Table E.2 Unadjusted odds ratios of hidden homelessness on various health outcomes in imputed sample…………………………………………………………………………………………... 86  xiv  List of Figures Figure 3.1 Theorized relationship of variables ............................................................................. 25! xv  List of Symbols Δ: Change  xvi  List of Abbreviations HH: Hidden Homelessness HSIY: Homeless and Street Involved Youth Survey xvii  Glossary Cis: A cisgender person identifies with the gender they were assigned at birth. Two-Spirit: Two-spirit people are those who have both a masculine and a feminine spirit. The term is used by some Indigenous people to describe their sexual, gender, or spiritual identity (Ross, 2017).   xviii  Acknowledgements I would like to sincerely thank Dr. Elizabeth Saewyc, my supervisor, and Dr. Victoria Bungay and Dr. Sabrina Wong, my committee members. Additionally, I want to thank Dr. Gu Li for his help checking my multiple imputation code in R and Mike Marin for discussing statistical methods with me. The support of the Lyle Creelman Endowment Fund awarded by the School of Nursing, University of British Columbia, was invaluable in allowing me to explore this topic in the depth that it deserves.  This work would not have been possible without my partner, Frederick, who listened to endless discussion of my thesis topic, fed me, and was the best editor a queer could hope for.  I would like to especially thank the youth advisory committee. I cannot say how much I appreciate the critical thought and insight that this group of young people brought. They have agreed to have their names listed in thanks. By alphabetical order: Skyla Barahona, Ashley Crossan, Breezy Arielle Hartley, Fialka Jack-Flesh, Barbara-Jean Johnson, and Joshua Wasilenkoff – Thank you so much for your time and your expertise! Without you this work would just be numbers.  xix  Dedication  I would like to dedicate this thesis to all the people with whom I have worked. It has been a privilege to be let into the lives of so many of my community members. I have learned more from these people than I ever thought possible.  1  Chapter 1:!Introduction         Homelessness can devastate young peoples’ health. Homelessness exposes young people to a variety of harms and traumas that affect their health and passage into adulthood (Fast, Shoveller, Shannon, & Kerr, 2009; May, 2015; O'Grady & Gaetz, 2004; Watson, 2011). These harms often differ by gender due to the sexism and transphobia that young people experience (Johnson et al., 2017; May, 2015; North & Smith, 1993; Richter & Chaw-Kant, 2010; Tomas & Dittmar, 1995; Watson, 2016 Watson, 2011). Most population studies of youth homelessness narrowly focus on people either living in shelters or on the street, yet there are many different ways that people navigate homelessness. Federally, there is a push to both better understand peoples’ experiences of homelessness and to create accurate and standardized population estimates; however, people with atypical experiences of homelessness – that is experiences of homelessness where people do not stay exclusively on the streets or in shelters such those using hidden homelessness shelter strategies – are still largely overlooked (Gaetz, Dej, Richter, & Redman, 2016; Homelessness Partnering Strategy, 2017; Johnson, Ribar, & Zhu, 2017; Watson & Cuervo, 2017). Many Canadians continue to struggle as a result of a deficit of affordable housing and many of these Canadians navigate homelessness in a way hidden to health policy makers (Laird, 2007). My thesis work investigated the relationships between hidden homelessness, health outcomes, and gender among youth.   Hidden homeless people are those who use specific shelter strategies (Eberle, Kraus, Serge, Mustel Research Group, & Inc, 2009). These can include sleeping in homes without security of tenure – that is without the legal protection of tenancy afforded to someone leasing or subletting a place – in either overcrowded conditions, or staying on a tenant’s or homeowner’s 2  couch, or else living in inadequate shelters such as in cars or squats (Eberle, Kraus, Serge, Mustel Research Group, & Inc, 2009). Using these strategies influences peoples’ health and what services they access (Harter et al., 2005; May, 2015; McLoughlin, 2013). Visible homelessness shelter strategies include staying at shelters or on the street. Throughout this work I will refer to the hidden homeless and those using hidden homelessness shelter strategies interchangeably. These strategies are especially used by women and trans people of all genders for multiple different reasons (Abramovich, 2012; Johnson et al., 2017; Lyons, Krüsi, Pierre, Smith, Small & Shannon, 2016; Maccio & Ferguson, 2016; May, 2015; North & Smith, 1993; Richter & Chaw-Kant, 2010; Sakamoto, Chin, Chapra, & Ricciardi, 2009; Tomas & Dittmar, 1995). There is no strict agreement in the literature on how to define hidden homelessness. While some researchers describe the population as couch-surfers, most sources describe hidden homelessness more broadly. A Vancouver-based study conducted a scan of how community agencies were using the term “hidden homeless” across Canada, and while some agencies exclusively cited couch-surfing, hidden homelessness was widely conceptualized as applying to those who do not have “security of tenure” (Eberle, Kraus, Serge, Mustel Research Group, & Inc, 2009). Documented ways of being hidden homeless include being temporarily housed in motels, drug houses, correctional facilities, sheds/garages, empty UHAULs, emergency departments, shower and laundry services, backyards and campgrounds, cars, abandoned buildings, church basements, as well as accepting housing from romantic partners (potentially in dangerous domestic situations), and living in the care of the Children’s Aid Society (in the case of children) (Eberle et al., 2009; Social Planning and Research Council of BC, 2011). The hidden homeless are often portrayed as being less at risk than visible homeless people, yet a Canadian 3  study showed that the mortality rates of the marginally housed were not significantly different than of those who live in shelters (Hwang, Wilkins, Tjepkema, O'Campo, & Dunn, 2009). Whereas the hidden homeless population is extremely heterogeneous, there are some common health deficits presented in the literature. One BC-based study reported that 65% of this group of people have mental health concerns compared to 20% lifetime prevalence among the general Canadian population; 52% have used substances in a way deemed problematic; 48% have physical health challenges; and most have difficulties securing employment (Social Planning and Research Council of BC, 2011).  1.1! Statement of Purpose My work on hidden homelessness focuses on the experiences of adolescents – a group defined by the World Health Organization as people between 10 and 19 years of age (World Health Organization, 2014) yet which will be operationalized in this work as 12-19 due to the data collection protocol of the Homeless and Street Involved Youth Survey that is the source of data. While a limited amount of research has been conducted on hidden homeless adults, even less has been conducted on adolescents. The purpose of this work is to examine the burden of physical and mental illness among male youth, female youth, and trans & gender diverse youth, aged 12-19, and whether this burden is associated with hidden homelessness shelter strategies. Prior research suggests that the health discrepancies between securely sheltered and homeless youth are either caused by increased exposure to risk (such as being in proximity to more people who may harm them and directly cause poor health outcomes) or are place- and process-based stress responses (related to uncertainty and fear surrounding occupying insecure housing or being homeless) (Shankardass, 2012). This suggests that harms related to homelessness are not exclusive to people experiencing visible homelessness. Women more frequently use hidden 4  homelessness shelter strategies, which may be associated with exposure to violence (including intimate partner violence) due to lack of choice of other shelter strategies.  1.2! Research Questions  What is the association between using hidden homelessness shelter strategies, and physical and mental health outcomes among homeless and street involved youth living in British Columbia?  H0: The null hypothesis would suggest that there is no difference in health outcomes between those who use hidden homelessness shelter strategies relative to those who use visible homelessness shelter strategies. Are there gender differences in the association between hidden homelessness shelter strategies and health outcomes, and, if so, what is the nature of this relation? H0: The null hypothesis would venture that the relation between hidden homelessness and health outcomes is the same for those of different genders.  1.3! Gender I have included gender-based analyses in my study due to the breadth of work connecting oppressed genders (including women and trans people of all genders) to poor health outcomes as well as being more likely to experience hidden homelessness. There is conflicting evidence around gender and homelessness. While women and trans people are systematically paid less in comparison to cis men, they are underrepresented in current figures of homelessness (Johnson et al., 2017; Richter & Chaw-Kant, 2010). One explanation might be that women and trans people are simply not homeless at the same rates as cis men, despite financial and structural barriers to housing (Gaetz et al., 2016). Another, potentially more sound explanation is that women and trans people are unsafe in many public places, and are forced to use hidden homelessness shelter 5  strategies. It is established that women are more likely to be part of the hidden homeless population, and are therefore less likely to be enumerated in point in time homelessness counts (Johnson et al., 2017; May, 2015; North & Smith, 1993; Richter & Chaw-Kant, 2010; Watson, 2011).  1.3.1! Women and Homelessness Health disparities exist between the genders in Canada – while Canadian women experience greater longevity they also report a greater burden of illness in psychiatric illness, chronic illnesses, and general distress (Denton, Prus, & Walters, 2004). As Canadian women are more likely to live in poverty and have a higher burden of mental illness, it is curious that they comprise such a small proportion of people enumerated in homeless counts (Denton et al., 2004; Johnson et al., 2017). Canadian women are more likely to report stressful life events and childhood trauma, and have significantly lower levels of self-esteem (Denton et al., 2004). The associations between homelessness and physical illness including HIV/AIDS are stronger for women compared to men (Johnson et al., 2017). The primary health consequences of homelessness on women include poorer physical and mental health, higher reported rates of physical and sexual abuse, more problems associated with drugs and alcohol, hunger, and lower rates of and less stable employment (Johnson et al., 2017).  Women, both trans and cis (people who are not trans), experience homelessness differently and for different reasons than men do. Women become homeless for different reasons than men – more women than men become homeless for reasons such as family break-up, marital and parental disputes, or through domestic violence (North & Smith, 1993; Tomas & Dittmar, 1995). Canadian homeless women are less likely to be employed, more likely to have 6  lower incomes, and more likely to be a single parent compared to homeless Canadian men (Denton et al., 2004). Violence against women is understood to play an important role in women’s homelessness (Gaetz et al., 2016; Johnson et al., 2017). Researchers found that many people who were discharged from violence-against-women-shelters report leaving to insecure housing or shelters (Tutty, Koshan, Jesso, Ogden, & Warrell, 2011). Women with children may not have the option of sleeping rough (sleeping outside) and family shelters are often at capacity, contributing to family hidden homelessness (Gaetz et al., 2016; Vacha & Marin, 1993). A quarter of participants in one study reported that violence in their last residence was a cause for their homelessness (Johnson et al., 2017). Due to lack of affordable housing and social supports for women who have experienced domestic violence, many must decide between homelessness and returning to their abuser (Thurston et al., 2013). This is especially the case for immigrant women who may not have access to appropriate social services, may have limited work options related to immigration policies, may experience housing discrimination, or may not have social support networks to support them due to recent relocation (Thurston et al., 2013).  Young women experiencing homelessness encounter gender-based violence and are often held solely responsible for navigating their personal safety (Biederman & Nichols, 2014; Watson, 2016). A strategy that some young women use to minimize the risk of violence is to engage in intimate relationships with men as a form of protection, yet protection-based relationships can be unstable and potentially a source of violence while undermining access to other resources such as stable accommodation and support from family and friends (Johnson et al., 2017; Watson, 2016). For some women, intimate partner violence was sometimes perceived as a viable alternative used to manage the external violence of homelessness (Johnson et al., 7  2017; Watson, 2016). Finding temporary residence through short term sexual relationships is an overlooked element of women’s homelessness (Johnson et al., 2017; Tomas & Dittmar, 1995).   Young homeless women described the places they went when homeless (including the street and hidden shelters such as squats) as dangerous environments in which they were in constant fear for their safety and lives and faced frequent sexual violence (Watson, 2016). Many young homeless women internalize the sense that they alone are responsible for managing their safety from gender-based violence (Watson, 2016). Homeless women face greater barriers accessing care because of stigma around gendered homelessness; their most frequent point of access is the Emergency Department (Biederman & Nichols, 2014; Shiller & Bierman, 2010). This trend may be because there are fewer services available to women, and because of the general distrust that many women have due to more extreme enacted stigma they face from the combination of classism and misogyny (Richter & Chaw-Kant, 2010). This has serious implications on women’s health and the sustainability of health care provision to the homeless population. There is no research comparing youth and adults’ experiences of hidden homelessness, yet looking at the relationship between gender and homelessness for adults may shed light on the associations among youth. Women’s experiences of homelessness have been well researched and serve as a good jumping-off point in discussions of gender and homelessness.   1.3.2! Transgender People and Homelessness Even less understood is how transgender people experience hidden homelessness. While there is no current population-level statistical data, trans people, and specifically trans women, are believed to be disproportionately represented amongst the homeless in Canada (Abramovich, 2012; Lyons et al., 2016; Maccio & Ferguson, 2016; Sakamoto, Chin, Chapra, & Ricciardi, 8  2009). Trans people are financially disadvantaged because of systemic transphobia and transmisogyny often resulting in homelessness. However, shelters, drop-in centers, and other services are often inaccessible to transgender people due to the stigma and violence they face there at the hands of both other service users and employees (Abramovich, 2012; Lyons et al., 2016; Maccio & Ferguson, 2016; Sakamoto et al., 2009). Gender non-conforming people and trans women especially face violence and discrimination in the shelter systems (Abramovich, 2016; Lyons et al., 2016). Trans youth often use hidden homelessness shelter strategies to avoid discrimination (Maccio & Ferguson, 2016).  Trans people experience health inequities such as experiencing higher rates of negative health outcomes, including depression and violence linked to transphobia, stigma, and social exclusion, yet many report having limited access to care because of discrimination in health systems (Lyons et al., 2016). Trans people are overrepresented among Canadian homeless communities due to complex, interrelating reasons such as discrimination in the labour force and the housing market (Lyons et al., 2016).  Gender presentation is difficult for homeless trans people to navigate – trans women and some two-spirit people reported that while presenting as stereotypically feminine allowed them to access the social and health services that more closely matched their identity and where they were more safe, presenting as more stereotypically masculine helped them avoid violence and harassment on the streets (Lyons et al., 2016). The risks associated with being street involved differ between cis and trans people due to stigma and social exclusion (Watson, 2011). Similarly to women versus adolescent girls, the differences and similarities between trans & gender diverse adults and youth have not been researched. One might presume that systems of 9  oppression operate on youth just as they do on adults, but the nature of those differences is not described in the literature.   1.4! Hidden Homeless Youth Hidden homeless youth often remain highly mobile to remain invisible to other street involved people or the police, which many perceive as dangerous (Fast, Shoveller, Shannon, & Kerr, 2009; May, 2015). Family conflict and violence are some of the complex and multilayered reasons that youth become homeless (O'Grady & Gaetz, 2004; Watson, 2011). Pregnancy was found to be a contributing reason for homelessness in young women, such as when there is insufficient room for a baby or when a youth’s pregnancy acts as a precipitator of family conflict around sexual behaviour (Watson, 2011).  1.5! My Connection to Hidden Homelessness Hidden homelessness has played a role in my life both personally and professionally. As a bi genderqueer teen, I experienced rural homelessness. I had no access to (or knowledge of) social services, so, despite living in a wealthy country, I starved for periods during my youth and did not expect to finish high school.   As an adult and a registered nurse, I wanted to give back to the community that I had come from and pursued a career in street nursing. The people with whom I worked came from many different backgrounds, and had many different experiences, but I noticed how gender influenced strategies of homelessness, and how these strategies shaped the health and the health care access of the people who employ them. These personal and professional experiences sparked my passion for health equity for hidden homeless people.  10  Chapter 2:!Literature Review Experiences of homelessness are not homogenous. In this chapter I outline how the literature review was conducted, how different authors describe hidden homelessness, what groups of youth are invisible to the Canadian government with current enumeration strategies, which subgroups of people tend to be hidden homeless, how youth come to be hidden homeless, and the social and health effects of hidden homelessness. I knew, from familiarity with the literature, that there is little work on hidden homeless youth, so I tried to synthesize all the pertinent available articles, as a review of hidden homeless youth health or a discussion of hidden homeless youth and gender would result in a quite limited review. Similarly, I included articles with a wide range of definitions of youth to be able to have enough literature to present a broad picture of what is known about hidden homeless youth.   2.1! Methods  While it is not valid to simply generalize what we know of hidden homeless adults to youth, I start with an exploration of the general literature on hidden homelessness. To get a comprehensive overview of the youth-specific literature, I searched four health databases – Medline, CINAHL, Sociological Abstracts, and Psycinfo – for articles related to hidden homelessness. Mesh headings, subject headings, and keywords used to search for young people included ‘youth’, ‘teen’, ‘adolescent’, ‘adolescence’, and to identify articles on hidden homelessness the search terms ‘hidden homeless*’, ‘couch surf*’, ‘invisible homeless*’, and ‘homeless youth’.  I used Google Scholar, the Homeless Hub, and the McCreary Centre Society report database to expand the scope of my literature review to capture grey literature in addition to peer-reviewed articles. The Homeless Hub is a Canadian site dedicated to collecting literature on 11  homelessness. The Homeless Hub was searched with the keywords ‘youth hidden homelessness’ and ‘hidden homeless’. Google scholar was searched with ‘youth AND hidden homelessness’ and ‘adolescent AND hidden homelessness’. I searched the McCreary Centre reports website, as I had prior knowledge that they had released reports on homeless youth with disaggregated information on hidden homelessness.   The titles and abstracts of 1,170 articles and reports were reviewed for relevance. Theses, newspaper articles, and book reviews were excluded. Only English-language articles were included for review. Databases were searched from December 2016 – April 2017; all relevant literature was searched up until the date that I reviewed the databases. There was no limit set on how far back literature was published. I reviewed the title and abstract of each article for potential inclusion – most non-applicable studies were obvious at this point. Considering the small size of pertinent literature, duplicates were manually eliminated. Articles were excluded if they did not specifically have disaggregated information on hidden homelessness versus other shelter strategies. Articles were excluded if they did not have any disaggregated information on youth (however the authors chose to identify this group – the diverse definitions are discussed below). Most articles were eliminated through a review of their abstracts. Twenty articles were read in their entirety and 10 were identified as being significant for this literature review. Decisions on inclusion and exclusion are described in appendix B.  2.2! Prevalence of youth hidden homelessness and shelter strategy trends The Canadian demographics of homelessness are changing – the country has seen a growing proportion of women, children, youth, and families navigating homelessness systems  (May, 2015). Indeed, young people under 24 comprise the fastest growing segment of the Canadian homeless population (Centre for Addiction and Mental Health Hidden in Our Midst: 12  Homeless Newcomer Youth in Toronto – Uncovering the Supports to Prevent and Reduce Homelessness, 2014). Homeless youth have become more likely to report using hidden homelessness shelter strategies. The numbers of British Columbia youth who reported couch surfing rose from 37% in 2006 to 50% in 2014 (Smith et al., 2015). This shift has been attributed to the effect of neoliberalism on the housing and employment markets starting in the 1980s, as well as increasing racial segregation in the same timeframe (May, 2015). Hidden homelessness as a shelter strategy is becoming increasingly widespread as public spaces are increasingly policed, causing individuals to remain invisible (and, as a result, excluded from policy statistics) (May, 2015). Eighty percent of all homeless people in Canada are theorized to be using hidden homelessness shelter strategies (Crawley et al., 2013). Understanding youth hidden homelessness is becoming increasingly important for Canadian health policy makers.   2.3! Pathways to homelessness Common pathways to youth homelessness include being evicted from their parents’ home, voluntarily choosing to leave, being driven out by conflict, or being ‘crowded out’ due to limited living space at home for too many people (Fitzpatrick, 1998). Homeless youth are less likely to receive social welfare or be in the legal system compared to adults and are thus less likely to be identified as being in need (Anthony, 2014). Lack of confidence or initiative and refusal to use services directed at homeless people due to the perception of danger were reported reasons why youth did not approach social service agencies, as was a reluctance to leave an individual’s local area (Fitzpatrick, 1998). 2.4! What is hidden homelessness? Hidden homelessness has been defined as using provisional accommodation without security of tenure (Crawley et al., 2013; Smith, 2013). There have been many ways that people 13  navigate hidden homelessness, but couch surfing is the most studied strategy (Centre for Addiction and Mental Health Hidden in Our Midst: Homeless Newcomer Youth in Toronto – Uncovering the Supports to Prevent and Reduce Homelessness, 2014; Fitzpatrick, 1998; McCoy & Hug, 2016; Perez & Romo, 2011; Robert, Pauzé, & Fournier, 2005). Couch surfing – or temporarily staying with a family member, friend, or stranger – is a frequently used strategy (McLoughlin, 2013). Young people rely on their own social resources to navigate lack of housing, yet couch surfing can entrench people in social and economic marginalization and remains an important subject of health research (McLoughlin, 2013). Many couch-surfing youth must continuously renegotiate temporary living conditions – for many this is not a sustainable solution (McLoughlin, 2013). Out of a sample of 218 runaway adolescents, 59% reported staying with friends, acquaintances or relatives; 18% simply did not sleep; 14% slept in a park, abandoned house, or public place; and only 9% used a homeless shelter (Robert et al., 2005). Out of the 10 adolescents who had used shelters, 7 had also stayed with friends, acquaintances or relatives at a different time (Robert et al., 2005). This Canadian study found that less than 1% of homeless youth never used hidden homelessness shelter strategies (Robert et al., 2005). 2.5! Which groups are made ‘hidden’ by current enumeration policies? The Government of Canada states that point-in-time counts are not designed nor best suited to enumerate hidden homeless people, but emphasizes the importance of furthering understanding of who the population is comprised of and what their service needs are (Homelessness Partnering Strategy, 2017). The upcoming coordinated Canadian counts will not intentionally look for hidden homeless people, but will ask individuals “Can you stay [at your housing situation] as long as you want or is this a temporary situation?” and “Do you have a house or apartment that you can safely return to?” in order to screen individuals into the count 14  who may otherwise be excluded (Homelessness Partnering Strategy, 2017). The youth count toolkit available at the Homeless Hub states that youth use hidden homelessness shelter strategies more frequently than adults, and so hidden homelessness is more intentionally sampled in this population (The Homeless Hub, 2017). They will attempt to “screen in” youth by asking respondents if they are couch surfing (The Homeless Hub, 2017). 2.6! What ages were conceptualized as youth or adolescents?   Authors’ delineations of who would be included in their samples of “youth” were heterogeneous. While one author simply sampled those under 18 (Robert et al., 2005), most included older youth, ages 20-25, as well. The articles reviewed described the age range of their participants (where included) as: 16-24 (Centre for Addiction and Mental Health Hidden in Our Midst: Homeless Newcomer Youth in Toronto – Uncovering the Supports to Prevent and Reduce Homelessness, 2014), 16-25 (Fitzpatrick, 1998), 17-21 (Harter, Berquist, Scott Titsworth, Novak, & Brokaw, 2005), 17-26 (May, 2015), 16-23 (McCoy & Hug, 2016), 15-25 (McLoughlin, 2013), 18-22 (Perez & Romo, 2011), and 12-19 (Smith et al., 2015). The Canadian Observatory for Homelessness has identified homeless youth as 13-24 year-olds who do not have means or ability to access a stable, safe, consistent residence (Gaetz et al., 2016). Clearly the descriptor ‘youth’ is being used in disparate ways.  2.7! Subgroups Racialized, immigrant, newcomer youth are disproportionately represented in Canada’s hidden homeless population (Centre for Addiction and Mental Health Hidden in Our Midst: Homeless Newcomer Youth in Toronto – Uncovering the Supports to Prevent and Reduce Homelessness, 2014). In British Columbia, a starkly disproportionate number of homeless youth are Indigenous; around half of youth in a study of homeless youth identified as Aboriginal 15  (53%), compared to 7% in a provincially representative school-based survey conducted a year before (Smith et al., 2015). One Toronto-based study on homeless newcomer youth reported that almost two thirds of the study’s participants came from Africa or the Caribbean, 10% came from the Middle East, 9% came from South America, and 27% were refugees (Centre for Addiction and Mental Health Hidden in Our Midst: Homeless Newcomer Youth in Toronto – Uncovering the Supports to Prevent and Reduce Homelessness, 2014). Queer and trans youth are disproportionately represented in the homeless population (Abramovich, 2013). In British Columbia, 3% of homeless youth identified as transgender or gender diverse, and 62% identified as straight, compared to 1% of youth identifying as transgender or gender diverse and 81% identifying as straight in a provincially representative school-based study conducted one year prior (Smith et al., 2015; Smith, Stewart, Poon, Peled, McCreary Centre Society, 2014). In a study conducted in Toronto, Ontario, over one third of homeless newcomer youth identified as LGBTTIQ (Centre for Addiction and Mental Health Hidden in Our Midst: Homeless Newcomer Youth in Toronto – Uncovering the Supports to Prevent and Reduce Homelessness, 2014).  Having a history of government care, such as being in foster care or staying in a group home, is associated with experiencing homelessness. In BC, over half (51%) of homeless youth had lived in a foster home or group home at some point in their lives (Smith et al., 2015). 2.8! Context of youth hidden homelessness  Hidden homelessness is a key shelter strategy used by youth, yet many do not self-identify as being homeless, which presents difficulties to providers and renders youth invisible in surveys (Harter et al., 2005). Labeling a person as ‘homeless’ has detrimental effects on peoples’ self-image and may cause youth to feel socially devalued due to enacted and internalized stigma 16  (Harter et al., 2005; May, 2015).  Many youth wish to remain invisible, which can prevent them from accessing formalized help due to many services’ mandates (Harter et al., 2005; May, 2015; McLoughlin, 2013). Youth are often placed in the difficult position of needing to choose between remaining invisible to avoid stigma or being open about their homelessness to access health or social services (Harter et al., 2005).  Young people report the desire to not feel or look desperate, worthless, pathetic, or needy, or be a burden on others (McCoy & Hug, 2016). Some youth reported staying at strangers’ places rather than with family or friends to avoid being a burden in relationships where they felt they had more at stake (McCoy & Hug, 2016). Staying with social networks was perceived as more immediately available to youth than other support services, making them more likely to stay with friends, family, or strangers in times of urgent need for housing (McCoy & Hug, 2016).  Legal issues complicate how younger people navigate housing. Underage youth may avoid official social and health services because they do not wish to be reported to the police or child protective services (Robert et al., 2005). Racialized young Canadian men reported the importance of remaining hidden from the police who often profile them, and who they describe as predatory (May, 2015). In urban areas, there have been increasing numbers of police projects designed to clear the streets of homeless people, pushing them into hidden homelessness (May, 2015). As gentrification changes the urban landscapes within Canada, the numbers of visible homeless people will decline and hidden homelessness will become increasingly common (May, 2015).  Hidden homeless youth are sometimes not guided to appropriate housing services by adults in their lives who do not understand the severity of their situations (McLoughlin, 2013). 17  As many youth do not independently know of these resources without being introduced, they, in effect, do not have access (McLoughlin, 2013). 2.8.1! Shelter strategies  Couch surfing is undoubtedly an important shelter strategy, yet there is ambiguity in what couch surfing functionally means – youth report that it can range from relatively safe and supportive situations with family and friends, to quite dangerous arrangements with near strangers (McCoy & Hug, 2016). Even family members are not always able to offer safe or stable places to live (Perez & Romo, 2011). Couch surfing is an immediate, accessible strategy through which to mitigate risks associated with visible homelessness (McLoughlin, 2013).  One under-reported strategy is for youth to simply avoid sleeping (McCoy & Hug, 2016; Robert et al., 2005). Attending all-night raves or parties is a reported way that youth use to address lack of shelter, yet this strategy reportedly put youth at high risk of harm (McCoy & Hug, 2016).  2.9! Gender and hidden homelessness In the social and economic space of the streets, male youth typically have considerably more power and control than female youth – young women face increased risk of sexual and physical assault (O'Grady & Gaetz, 2004). Young women are more likely to identify parental conflict, physical abuse, sexual abuse, and mental health issues as significant reasons for leaving home (O'Grady & Gaetz, 2004). There are fewer young women on the street, but they are more likely to report violence forcing them away from home, leading authors to theorize that the street is often seen as a ‘male space’ where females are reluctant to go unless forced by extreme violence (O'Grady & Gaetz, 2004). Young women experience personal safety differently than males, affecting their mobility, comfort in certain environments, and income generation 18  (O'Grady & Gaetz, 2004). Female youth are less likely to see shelters as ‘safe’ (O'Grady & Gaetz, 2004). Young homeless men in BC were more than twice as likely to report staying in a safe house or shelter compared to young homeless women (15% vs 7%) (Smith et al., 2015). 2.10! How does hidden homelessness affect health?  Experiences of hidden homelessness range widely in terms of potential for danger, levels of support, and practical circumstances (McCoy & Hug, 2016). Shelters for the homeless and longer-term temporary accommodation are often unsuitable and unsafe for youth, yet some of the ways youth avoid shelters can also be harmful (McCoy & Hug, 2016).  Some potential harms of couch surfing include strains on relationships, damage to health, exposure to dangerous environments, and sexual exploitation (McCoy & Hug, 2016). While couch surfing with family members is widely perceived as benign, it is very much dependent on an individual’s circumstance – youth describe very positive and supportive temporary living situations as well as harmful or traumatizing ones (McCoy & Hug, 2016). Shelter is sometimes exchanged for domestic services or sex – both of which can come with associated risks of abuse in the context of extreme power differentials (McCoy & Hug, 2016). Youth reported physical and mental health consequences of being forced into temporary living conditions – couch surfing often resulted in disrupted sleeping due to lack of space, privacy, or control, decreased access to food, increase in substance use, poor mental health related to stress, violence and uncertainty, and increased risk of physical harm (McCoy & Hug, 2016). Rape, physical assault, intense bullying, and in one case, being set on fire featured in the stories told about temporary housing by an English cohort of couch surfing youth (McCoy & Hug, 2016). School and work were impacted by couch surfing – in one study, fatigue due to 19  unstable housing situations led to low performance, frequent absenteeism, and loss of employment (McCoy & Hug, 2016).  20  Chapter 3:!Methodology This chapter describes the research questions guiding my work, how the Homeless and Street Involved Youth Survey was conducted, what variables I analyzed and what hypotheses I tested. I describe what statistical methods, and community engagement methods I used, and discuss the underpinnings of my statistical and theoretical work. 3.1! Research Questions What is the association between having used hidden homelessness shelter strategies and physical and mental health outcomes among homeless and street involved youth living in British Columbia?  Are there gender differences in the association between hidden homelessness and health outcomes, and, if so, what is the nature of this relation? 3.2! Data Source This cross-sectional descriptive study uses the British Columbia Homeless and Street Involved Youth Survey (HSIY). The HSIY is a survey from 13 communities across British Columbia; data are collected by the McCreary Centre Society. The HSIY was steered by an advisory committee of youth with lived experience of homelessness and adult social service professionals who work with homeless youth; these advisory committee members assisted in developing the survey, guided interpretations of findings, and took responsibility for recruitment and support of community co-researchers (a youth worker and at least one youth with experience of homelessness in each sampled community) who helped collect the data. Secondary data analysis – or the use of existing research data to answer research questions which differ from the original work –was chosen to balance the generation of high-quality public health research specific to homeless youth with the competing ethical issue of ensuring that homeless youth are 21  not over researched nor over exposed to sensitive questions (Tripathy, 2013). The Homeless and Street Involved Youth Survey aimed to present a broad overview of the health experiences and needs of homeless and street involved youth in different communities across BC and therefore had survey questions that answered my research questions concerning a variety of health measures, gender, and where youth had stayed (Smith et al., 2015).   The total sample is comprised of 689 youth (with 681 usable surveys) from 13 communities across British Columbia. The 2006 HSIY selected communities with the highest prevalence of homelessness (including hidden homelessness) and sexual exploitation and included more communities in 2014 based on geographic diversity and community consultation (Smith, personal communication, July 16, 2018). Participants ranged in age from 12 to 19. Rolling data collection and convenience sampling were used – in some communities, community researchers were confident that they had achieved a census of the homeless youth population. Sample size was predetermined based on the expert estimates – made by experiential youth and health care providers – of the numbers of homeless youth in each community. The survey was pen and paper, and the data were collected between October 2014 and January 2015. Participation was voluntary and anonymous. Prior to survey administration, the implications of the survey were explained to participants, and consent was obtained. Participants received a small honorarium. Variables of interest were selected from the literature review and then I identified which survey questions captured those variables from!the BC Adolescent Health Survey. The research aim of the HSIY was to provide information about risk and protective factors and the broader health picture of BC youth who were homeless, precariously housed, or involved in a street lifestyle. Youth with lived experiences of homelessness were involved from survey development and delivery to interpretation of findings. 22  3.3! Explanatory variables  Key variables of interest were identified based on my literature review and my experience in clinical practice. If the HSIY had a pertinent survey question, I included it as an outcome variable for analysis.  The independent variable of interest was youth’s hidden homelessness shelter strategies. I drew on responses from the question, “People can live a lot of different places for short or long periods of time. Please tell us where you live now (yesterday) and any place you’ve ever lived”: where youth report staying at their parents’ home, at a house or apartment, at another relative’s home, nowhere/couch-surfing, at a safe house or shelter, on the street, at a hotel/motel/ single room occupancy buildings (SROs)/hostel, in a tent or car, in a transition house, in a squat or abandoned building, at an extreme weather shelter, in a college dorm, or another place that they specified, I included them in the hidden homelessness sample.  3.3.1! Hidden homelessness shelter strategies In the data collected through the Homeless and Street Involved Youth Survey, youth reported staying at their parents’ home, at a house or apartment, at another relative’s home, nowhere/couch-surfing, at a safe house or shelter, on the street, at a hotel/motel/SRO/hostel, in a tent or car, in a transition house, in a squat or abandoned building, at an extreme weather shelter, in a college dorm, or other.  A subset of this population was extracted for analysis and comparison based on whether youth have experiences of hidden homelessness. This was operationalized by looking at how homelessness is currently measured and analyzing the subpopulation of individuals who may not be counted by more traditional homelessness research.  23  Despite a common preconception that staying with family is a typically stable and low risk strategy, people staying with a family member may not have security of tenure and may not necessarily be safe (McCoy & Hug, 2016; Perez & Romo, 2011). However, as it is not possible to differentiate those with security of tenure in this sample, these individuals were excluded from the hidden homeless subset. Similarly, many people that are sleeping outside intentionally remain hidden, yet the data collected as part of the HSIY survey did not allow for that distinction, therefore unsheltered youth were not included in the hidden homelessness subset.  To align with the government of Canada’s descriptions of homelessness and hidden homelessness and to meaningfully distinguish from other ways that homeless youth are counted, youth who report the following shelter strategies were considered as having experienced hidden homelessness: nowhere/couch surfing, hotel/motel/SRO/hostel, tent or car, transition house, squat or abandoned building, extreme weather shelter, or other. This study looked at whether youth had ever had experience of hidden homelessness. 3.4! Potential Confounders or Effect Modifiers Confounding variables were retained in the model if the statistical effect changed more than 10% and if the standard error of the effect did not increase.  3.4.1! Gender Reported gender identities include female, male, transgender, gender fluid, genderless, and pan-gendered. Gender was split into three variables: female youth, male youth, and trans & gender diverse youth. Non-binary transgender people, binary transgender people, and other gender diverse people were grouped in the analysis to maintain adequate sample size. It is not my intention to suggest that this is a homogenous group, but to look at the relation of trans and gender-diverse gender oppression and health outcomes. As some trans person do not identify 24  with “trans” as their gender but their true gender (“male” or “female”), and some gender diverse youth do not identify as “trans” I cannot conclude if participants are cis men or women. Therefore, the gender categories I used were “female”, “male”, “trans and gender diverse”. This is not to imply that the male or female gender of people who selected “trans” is any less valid.  Trans and gender diverse youth comprised 3% of the sample of 681. The limited number of trans and gender diverse youth meant that there was limited power in the analyses. Analyses were completed with the awareness of this limitation except for logistic regression models with disaggregated gender where trans & gender diverse youth did not meet the 30-person threshold for inclusion.  Conceptually, gender could act as an effect modifier in the relationship between health and hidden homeless status. By checking for statistical interaction, I looked for evidence of gendered social processes that affect the health of people using hidden homelessness shelter strategies (Bauer, 2014). By analyzing the effect modification of gender, I aimed to separate the effect of hidden homelessness between genders and determine the gendered relationship between hidden homelessness and health outcomes.    3.4.2! Age The numerical ages of participants at the time of the survey were analyzed to see if they had a confounding effect on the relationship between health and hidden homelessness. As participants range in age from 12 to 19, I reasoned that there would be heterogeneous effects depending on age.  3.4.3! Location  Location was split into Northern, Interior, Fraser, Vancouver Coastal, and Vancouver Island areas to separate the effect of location from the relationship of interest. Due to the small 25  number of categories, and the clustering of homeless youth in larger city centers, location was only included if the statistical effect changed significantly (more than 10%).  3.5! Outcome Health Variables    The following health outcome variables were descriptively analyzed and included in the logistic regression models: Having experienced an injury serious enough to require medical attention, self-rated poor or fair general health, self-rated poor or fair mental health, having at least one mental health condition, having foregone medical care, having foregone mental health care, having experienced a serious injury, level of extreme stress, level of extreme despair, self-harm, having considered suicide in the last year, and having attempted suicide in the last year.    Hidden  Homelessness Gender Health  Outcomes Age Location Figure 3.1 Theorized relationship of variables 26  Variable Name Dichotomization Explanatory Variable of Interest Hidden Homelessness Yes: Ever having stayed in the following housing situations: nowhere/couch surfing, hotel/motel/SRO/hostel, tent or car, transition house, squat or abandoned building, extreme weather shelter, or other No: Never having stayed in the preceding housing situations Moderator of Interest Gender Trans & gender diverse youth Female youth  Male youth Confounders of Interest Age 12 13 14 15 16 17 18 19 Location Northern Interior  Fraser Vancouver Coastal  Vancouver Island Outcome Variables of Interest Having experienced an injury serious enough to require medical attention in the past year Yes No Self-rated general health Poor/Fair Good/Excellent Self-rated mental health Poor/Fair Good/Excellent Having at least one mental health condition Yes  No Foregone mental health care Yes  No Foregone medical health care Yes  No Level of stress Not at all/A little/Some, enough to bother me/Quite a bit Extremely so, to the point I couldn’t work or deal with things 27  Level of despair  Not at all/A little/Some, enough to bother me/Quite a bit Extremely so, to the point I couldn’t work or deal with things Self-harm in the last year I have never deliberately cut or injured myself 0 times/1 time/2 or 3 times/4 or 5 times/6 or more times Having considered suicide in the last year Yes  No Having attempted suicide in the last year 0 times 1 time/2 or 3 times/4 or 5 times/6 or more times Table 3.1 Variables and variable dichotomization 3.6! Model Building I built the regression models starting with a review of important factors impacting both hidden homelessness and health outcomes in the literature. I subsequently scanned the HSIY database for questions that measured these factors. Following the selection of potential confounders, I tested for confounder significance by seeing if the addition of the variable to the model significantly altered the change in log odds of the health outcome given the hidden homeless condition (±10%) without inflating the standard error (+10%). 3.6.1! Hypotheses 1)! Youth who have used hidden homelessness shelter strategies will have poorer health outcomes in comparison to those who have not.  H0: The null hypothesis would suggest that there is no difference in health outcomes between those who use hidden homelessness shelter strategies relative to those who use visible homelessness shelter strategies. 2)! Female youth and trans & gender diverse youth will have poorer health outcomes in comparison to male youth. See below for the pertinent null hypothesis.  3)! Multiplicative interaction will be present in the relationship between health outcomes and hidden homelessness by gender. 28  H0: The null hypothesis would be that the relation between hidden homelessness and health outcomes is the same for those of different genders. This will be evident in both a lack of difference in descriptive analyses and insignificant interaction terms.  Descriptive statistics were drawn on the demographic make-up of the youth who have used hidden homelessness shelter strategies and those who have not. The average age and gender breakdown for these two groups were calculated. Additionally, age and gender distributions for the outcome measures were calculated. The percentages of missing data for each variable of interest were calculated to prepare for multiple imputation. If any variables were missing over 5%, multiple imputation was implemented (Enders, 2010).  While multiple imputation cannot duplicate a randomly sampled dataset, it can allow for the approximation of a convenience sampled dataset with complete answers assuming missing values are missing completely at random or missing at random (Enders, 2010). Multiple imputation was used as some variables in the dataset were missing over 5% of responses. The MICE package in R was used to multiply impute the dataset (Enders, 2010). Data were imputed 100 times. Multiple imputation created results that may better reflect the true standard error, versus the falsely small standard error observed in single imputation (Enders, 2010).  Logistic regression effect size models were fit to the different health outcome variables and used to determine whether there were differences in health outcomes by hidden homelessness. I used logistic regression on all health outcomes as they were binary variables. From data collection we know that observations were independent and the outcome variables are dichotomous – the assumptions of logistic regression were checked to ensure that the linearity assumption is met, and the log odds of the outcome variables were a linear function of the explanatory variables (Vittinghoff, Glidden, Shiboski, & McCulloch, 2007). With these models, 29  I aimed to paint an overall picture of health by housing status. All calculations were conducted in the statistical program, R.  Interactions between hidden homelessness and gender were evaluated to see how the intersection of gender and shelter strategies impacted the relationship with health. It was my intention to use this aspect of regression analyses to expand our intersectional knowledge of the gendered social processes of homelessness (Bauer, 2014). Following my statistical analyses, I consulted with an advisory committee of young people who have experienced homelessness as to the meaning behind the findings, and to create recommendations for how to address youth hidden homelessness. This research was informed by critical theory (Bauer, 2014; Watson & Cuervo, 2017). 3.7! Missing Data Since some variables had more than 5% missing, multiple imputation was used.  Variable Name Item Missing n (%) Hidden Homelessness 40 (5.9%) Location 0 (0%)  Gender 2 (0.3%) Age 5 (0.7%) Having experienced an injury serious enough to require medical attention in the past year 71 (10.4%) Poor/fair self-rated general health 37 (5.4%) Poor/fair self-rated mental health 79 (11.6%) Level of stress 90 (13.2%) Level of despair 87 (12.8%) Having considered suicide in the last year 90 (13.2%) Having attempted suicide in the last year 95 (14.0%) Self-harm in the last year 86 (12.6%) Having at least one mental health condition 89 (13.1%) Foregone Medical Care 69 (10.1%) Foregone Mental Health Care 63 (9.3%) Table 3.2 Number and percentage missing for each variable of interest 30  3.8! Multiple Imputation in Homelessness Research Due to societal discrimination, homeless youth may feel distrust for researchers and have high levels of unit nonresponse (Hudson et al., 2010). As such, multiple imputation may be a valuable method for meaningfully addressing missing at random variables (Enders, 2010). If a variable is missing at random, the probability of missing data is related to some other measured variable (or variables) (Enders, 2010). Missing HSIY data follow a general pattern as missing values were dispersed throughout the data matrix at random, yet are – conceptually – likely to have a systematic cause behind them such as certain subgroups finding certain questions more sensitive (Enders, 2010). Intersectional identities may result in more stigmatization in some groups compared to others – women who experience mental illness encounter more stigma than their male counterparts and may be more likely to skip questions related to mental illness (Mizock & Russinova, 2015).  Multiple imputation is a technique which accounts for missing responses in a way that does not violate the assumption that the data is missing completely at random – that is not related to any other variables (Little, Jorgensen, Lang, & Moore, 2014). Other imputation techniques like mean substitution, and other methods for dealing with missing data, like list wise or case wise deletion, violate the assumption that the data is missing completely at random and do not account for missingness related to demographic data, for example (Little et al., 2014). While consensus has not been reached on widespread use of multiple imputation, and results should be considered with the theoretical implications of multiple imputation in mind, it may be a key technique in homelessness research. 31  3.9! Limitations  A limitation inherent to the study design is that the data were collected through non-probability sampling due to lack of a sampling frame because of the hard-to-sample nature of homeless communities.    Since there were fewer than 30 trans & gender diverse youth in the sample, they were not included in the models separated by gender, as the results would have an inflated risk of type 2 errors – or false negatives. The trans & gender diverse youth sample was only excluded in the logistic regression models with significant interaction terms where ORs for the different genders were calculated separately.   Multiple imputation will continue trends observed in the sample, and if the sample does not reflect the true population of homeless and street involved youth, these trends may be amplified.   Many different aspects of peoples’ identities change how they access health care and housing – for the sake of this analysis I will be exclusively focusing on the intersectionality of housing and gender. However, while analyses by race, sexual orientation, and disability are outside the scope of this project, I by no means wish to undercut their importance (Bauer, 2014). 3.10! Gender in Research The Canadian government has explicitly called for the integration of both sex and gender in health research (“Canadian Institutes of Health Research: How to integrate sex and gender into research” 2018). Funded population health research in 2010-2011 had an especially low proportion of studies that included sex or gender components compared to funded clinical studies, health systems studies, or biomedical studies (with 21%, 44%, 54%, and 81% of funded 32  research integrating sex and gender respectively) (“Canadian Institutes of Health Research: Sex, gender and health research in Canada” 2014). In the fields of homelessness, housing, and poverty research, it is especially important to include gender considerations, since the experiences of trans people of all genders and cis women are qualitatively different than cis men; the risks that people of different genders are exposed to and the barriers to care that they face are fundamentally changed by the intersection of poverty and gender (Abramovich, 2012; Johnson et al., 2017; Lyons et al., 2016; Maccio & Ferguson, 2016; North & Smith, 1993; Richter & Chaw-Kant, 2010; Watson, 2011; 2016; Whitzman, 2006).  Especially important to the consideration of gendered homelessness is the inclusion of trans peoples’ experiences. Trans people’s experiences are an important piece of the Canadian homeless population because of the employment and housing discrimination they face, as well as young trans people being kicked out of their homes because of their gender identity (Abramovich, 2012; Maccio & Ferguson, 2016).  While quantitative analysis has no way of discerning what the nature of these differences is, it can shed light on the health consequences of gender differences. Existing literature can provide explanations of potential gender differences as was detailed in the introduction of this thesis.  Some studies exclude trans people since there are fewer trans people relative to cis people in most samples and resulting in power constraints. There are problems with this approach, both philosophical and practical. There is a philosophical problem with the erasure of trans people aligning with societal norms that reject gender deviance (Bauer et al., 2009). Public health decision makers need research on trans people. In the absence of even attempting to look at trans 33  population health, or generation of descriptive statistics, there is no evidence on which to base health services and policies (Bauer et al., 2009). This work includes descriptive statistics of trans & gender diverse youth, yet the sample size is inadequate for gender-specific odds ratios to be generated.  In this study, gender is not examined with the intention of describing peoples’ identities, but to describe health inequities based on gender oppression. While grouping trans and gender diverse youth together as a monolith is not reflective of their identities, it is designed to pragmatically capture common experiences of transphobia.  3.11! Advisory Committee An advisory committee of youth with lived homelessness experience between the ages of 16 and 24 was formed to give context to the findings and generate policy recommendations. The advisory committee meetings lasted about an hour. The committee was compensated for their time and insight with a meal, transit, and a $20 gift certificate.  Guiding questions:  •! What barriers to health might hidden homeless youth face? •! Which programs can hidden homeless youth not access? •! What parts of these programs are especially beneficial to youth? •! What might help hidden homeless youth health in the future? •! What might help hidden homeless youth health for people of different genders? Diversity of opinion and perspective was valued over consensus building. I wrote out the group’s points and they were encouraged to interject if their point had not been adequately captured.     34  Chapter 4:!Results This chapter discusses the findings of the descriptive analyses, the logistic regression models, impacts on the sample from multiple imputation, and recommendations from the advisory committee.  4.1! Descriptive Analyses The sample had approximately even numbers of male and female youth. There were a disproportionately large number of trans and gender diverse youth – 3.4% compared to less than 1% reported in the Adolescent Health Survey which was administered to youth in public schools around the same period (Smith et al., 2014). The average age was approximately 17 years old. Most participants were from the Vancouver Coastal Region (which includes Vancouver) although many participants were from the Fraser Region. Most youth had experienced hidden homelessness.   Percentage Gender (%) Male  Female Trans & Gender Diverse  47.0% 49.6% 3.4% Hidden Homelessness 63.8% Age Mean (Standard Deviation) 16.8 (1.6) Region  Northern Interior  Fraser Vancouver Coastal  Vancouver Island  17.0% 13.8% 27.0% 29.1% 13.1% Table 4.1 Demographic breakdown of sample.  Note: Hidden homelessness was operationalized as having experienced homelessness while not having stayed with family other than parents, stayed in a shelter, or slept outside.   35  Variable Percentage among hidden homeless sample (n=409) Percentage among the remainder of the sample (n=232) Chi Square P-Value Region  Northern Interior  Fraser Vancouver Coastal  Vancouver Island  15.2% 16.4% 26.2% 26.9% 15.4%  22.4% 10.8% 25.9% 32.8% 8.2% <0.01 Gender Trans & Gender Diverse Female Male  3.9% 50.4% 45.7%  2.2% 47.8% 50.0% 0.34 Poor/Fair self-rated general health 48.4% 37.0% <0.01 Poor/Fair self-rated mental health 55.3% 43.1% <0.01 Having at least one mental health condition 73.2% 49.7% <0.01 Having foregone physical health care 32.7% 13.4% <0.01 Having foregone mental health care 40.3% 19.3% <0.01 Having experienced an injury serious enough to require medical attention in the past year 38.3% 17.5% <0.01 Extreme stress 25.8% 12.3% <0.01 Extreme despair 12.5% 7.5% 0.07 Self-harm 48.1% 39.9% 0.06 Having considered suicide in the last year 48.4% 29.9% <0.01 Having attempted suicide in the last year 35.3% 21.2% <0.01 Table 4.2 Percentage of variables of interest in hidden homeless sample versus the remainder of the sample and chi square analyses p value. Note: Hidden homelessness was operationalized as having experienced homelessness while not having stayed with family other than parents, stayed in a shelter, or slept outside.  Unfavourable health outcomes were more commonly seen among the hidden homeless sample. Using a 5% significance level, having self-reported poor or fair general and mental health, having at least one mental health condition, having foregone physical or mental health care, having experienced a serious injury, having experienced extreme stress, or having 36  considered or attempted suicide in the last year were more common in the hidden homeless sample than in the other homeless youth sample. There is no evidence to conclude that having experienced extreme despair or having self-harmed were different between the two subsamples. These relationships were explored further using logistic regression models.  Variable Percentage among trans and gender diverse youth (n=23) Percentage among female youth (n=337) Percentage among male youth (n=319) Chi Square P-Value Region  Northern Interior  Fraser Vancouver Coastal  Vancouver Island  0.0% 21.7%  30.4%  34.8%  NR  15.7% 12.8% 26.7% 32.1% 12.8%  19.8% 14.4% 27.3% 25.7% 12.9% 0.29 Poor/Fair self-rated general health 57.1%  48.6% 41.4% 0.03 Poor/Fair self-rated mental health 76.2% 57.6% 41.4% <0.01 Having at least one mental health condition 95.7%  72.8% 53.9% <0.01 Having foregone physical health care 27.3% 26.7% 24.9% 0.87 Having foregone mental health care 50.0% 37.5% 26.0% <0.01 Having experienced an injury serious enough to require medical attention in the past year 36.4% 32.7% 28.2% 0.43 Extreme stress 33.3% 24.4% 15.2% <0.01 Extreme despair NR 13.3% 6.7% 0.02 Self-harm 77.3% 61.2% 23.1% <0.01 Having considered suicide in the last year 68.2% 51.4% 29.7% <0.01 Having attempted suicide in the last year 33.3% 40.5% 19.9% <0.01 Table 4.3 Percentage of variables of interest among different genders and chi square analyses p value.  Note: NR: Not reportable (n<5) 37  Unfavourable health outcomes were more commonly seen among the female and trans & gender diverse youth samples. Using a 5% significance level, having self-reported poor or fair general or mental health, having at least one mental health condition, having foregone mental health care, having experienced extreme stress or despair, having self-harmed, or having considered or attempted suicide in the last year were more common in the female and trans & gender diverse youth sample than in the male youth sample. There was no evidence to conclude that having foregone physical health care or having experienced a serious injury were different between the three subsamples. These relationships were further explored using logistic regression models.   38  4.2! Adjusted Odds Ratios of Hidden Homelessness on Various Health Outcomes Model Odds Ratio of Hidden Homelessness (95% Confidence Interval) Self-Reported Poor or Fair General Health 1.59 (1.14-2.24) Self-Reported Poor or Fair Mental Health* 1.73 (1.21-2.48) Mental Health Condition 2.77 (1.92-4.00) Foregone Medical Care 3.15 (2.01-5.09) Foregone Mental Health Care 2.82 (1.90-4.26) Injury 2.93 (1.94-4.52) Stress 2.47 (1.54-4.10) Despair* 1.64 (0.90-1.28) Self-Harm* 1.80 (1.24-2.63) Consider Suicide 2.20 (1.53-3.19) Attempt Suicide* 2.33 (1.54-3.60) * Adjusted for age Table 4.4 Adjusted odds ratios of hidden homelessness on various health outcomes Adjusted logistic regression showed that the odds of self-reported poor or fair general health, self-reported poor or fair mental health, having a mental health condition, having foregone medical care, having foregone mental health care, having experienced a serious injury, extreme stress, self-harm, considering suicide, or attempting suicide were larger among hidden homeless youth compared to other youth. As the confidence intervals for the odds ratio for the model on extreme despair crosses 1, there was insufficient evidence to state that the odds of hidden homeless youth having more extreme levels of despair are greater in comparison to other 39  homeless youth. Age acted as a confounder for self-reported poor or fair mental health, extreme despair, self-harm, and attempting suicide.  4.3! Multiple Imputation 4.3.1! Descriptive Analyses  Percentage Gender (%) Male  Female Trans and gender diverse  47.0% 49.6% 3.4% Hidden Homelessness 63.9% Age Mean (Standard Deviation) 16.8 (1.7) Region  Northern Interior  Fraser Vancouver Coastal  Vancouver Island  17.0% 13.8% 27.0% 29.1% 13.1% Table 4.5 Demographic breakdown of imputed sample   40  Variable Percentage among hidden homeless sample (n=435) Percentage among the remainder of the sample (n=246) Chi Square P-Value Region  Northern Interior  Fraser Vancouver Coastal  Vancouver Island  14.7% 15.6% 27.1% 27.1% 15.4%  21.1% 10.6% 26.8% 32.5% 8.9% <0.01 Gender Trans & Gender Diverse Female Male  3.9% 50.4% 45.8%  2.4% 48.4% 49.2% 0.47 Poor/Fair self-rated general health 47.6% 37.4% 0.01 Poor/Fair self-rated mental health 54.5% 44.7% 0.01 Having at least one mental health condition 72.9% 48.0% <0.01 Having foregone physical health care 31.3% 13.4% <0.01 Having foregone mental health care 38.9% 19.1% <0.01 Having experienced an injury serious enough to require medical attention in the past year 37.2% 17.1% <0.01 Extreme stress 25.3% 14.2% <0.01 Extreme despair 12.4% 8.5% 0.12 Self-harm 45.8% 37.8% 0.04 Having considered suicide in the last year 47.1% 30.1% <0.01 Having attempted suicide in the last year 34.9% 21.6% <0.01 Table 4.6 Percentage of variables of interest in hidden homeless sample versus the remainder of the sample in imputed sample and chi square analyses p value  Similar to the original dataset, unfavourable health outcomes are more commonly seen among the hidden homeless sample. Using a 5% significance level, having self-reported poor or fair general and mental health, having at least one mental health condition, having foregone physical or mental health care, having experienced a serious injury, having experienced extreme stress, having self-harmed, or having considered or attempted suicide in the last year are more common in the hidden homeless sample versus the other homeless youth. There is no evidence to 41  conclude that having experienced extreme despair was different between the two subsamples. Following imputation, having self-harmed became significant. These relationships will be explored further using logistic regression models.   Variable Percentage among trans and gender diverse youth (n=23) Percentage among female youth (n=338) Percentage among male youth (n=320) Chi Square P-Value Region  Northern Interior  Fraser Vancouver Coastal  Vancouver Island  0.0% 21.7%  30.4%  34.8%  NR  15.7% 12.7% 26.6% 32.0% 13.0%  19.7% 14.4% 27.2% 25.6% 13.1% 0.29 Poor/Fair self-rated general health 56.52% 48.5% 41.9% 0.01 Poor/Fair self-rated mental health 73.91% 58.0% 41.9% <0.01 Having at least one mental health condition 95.65% 71.9% 53.1% <0.01 Having foregone physical health care 30.43%  26.3% 22.8% 0.47 Having foregone mental health care 52.17% 35.5% 26.3% <0.01 Having experienced an injury serious enough to require medical attention in the past year 39.13%  31.4% 27.8% 0.38 Extreme stress 34.78%  25.2% 16.3% <0.01 Extreme despair NR 15.2% 7.2% <0.01 Self-harm 73.91% 59.8% 22.8% <0.01 Having considered suicide in the last year 65.22% 50.3% 29.4% <0.01 Having attempted suicide in the last year 34.78%  38.8% 20.6% <0.01 Table 4.7 Percentage of variables of interest among different genders in imputed sample and chi square analyses p value.  Note: NR: Not reportable (n<5) 42  Imputation did not change the results of the chi square tests for gender. Unfavourable health outcomes are more commonly seen among the female and trans & gender diverse youth samples. Using a 5% significance level, having self-reported poor or fair general or mental health, having at least one mental health condition, having foregone mental health care, having experienced extreme stress or despair, having self-harmed, or having considered or attempted suicide in the last year are more common in the female and trans & gender diverse youth sample versus the male youth. There is no evidence to conclude that having foregone physical health care or having experienced a serious injury were different between the three subsamples. These relationships will be further explored using logistic regression models.   43  4.3.2! Adjusted Odds Ratios of Hidden Homelessness on Various Health Outcomes Model Odds Ratio of Hidden Homelessness (95% Confidence Interval) Self-Reported Poor or Fair General Health 1.52 (1.11-2.10) Self-Reported Poor or Fair Mental Health** 1.48 (1.07-2.06) Mental Health Condition 2.91 (2.10-4.05) Foregone Medical Care Female Youth:  3.22 (0.90-12.57) Male Youth: 3.22 (1.19-3.59) Foregone Mental Health Care 2.69 (1.87-3.93) Injury Female Youth:  4.49 (1.21-17.36) Male Youth: 1.82 (1.08-3.14) Stress 2.04 (1.36-3.13) Despair* 1.64 (0.90-1.28) Self-Harm** 1.72 (1.22-2.45) Consider Suicide 2.20 (1.53-3.19) Attempt Suicide 2.30 (1.58-3.39) * Adjusted for age ** Adjusted for age and location Table 4.8 Adjusted odds ratios of hidden homelessness on various health outcomes in imputed sample Adjusted logistic regression with an imputed dataset showed that the odds of having self-reported poor or fair general health, having self-reported poor or fair mental health, having a mental health condition, having foregone mental health care, extreme stress, self-harm, considering suicide, or attempting suicide are greater among hidden homeless youth compared to 44  other homeless youth. As the confidence intervals for the model on extreme despair crosses 1, there is insufficient evidence to state that the odds of hidden homeless youth having more extreme levels of despair in comparison to other homeless youth. Age alone acted as a confounder in the model of extreme despair, and age and location acted as a confounder in the models of self-reported poor or fair mental health and self-harm. Interaction with gender was present in the foregone medical care and serious injury models. The odds of male youth with experience of hidden homelessness having foregone medical care are greater than the odds of their male peers to have foregone medical health care. The odds ratios for female youth crossed 1; therefore, there was insufficient evidence for a difference in odds. The odds of female and male youth with experience of hidden homelessness having experienced a serious injury are greater than their same-gender peers.  4.4! Multiple Imputation  While the multiple imputation did not greatly change the frequency values in the different variables, it changed some of the conclusions to be drawn from the analyses. In the multiply imputed data, there were some potentially important nuances in the findings. In the original dataset, location was not a confounder for any of the models. Hidden homelessness acted as the independent variable, the health outcomes acted as the dependent variable, gender was assessed for significance as a mediator, and location and age were tested as confounders. In the imputed dataset, location was a confounder for self-reported mental health and self-harm. This potentially sheds light on location-based mental health outcomes. In the original dataset, the interaction term between gender and hidden homelessness was non-significant in all models. In the imputed dataset, significantly different statistical effects of hidden homelessness were found by gender for having foregone medical care and having had an injury in the last year. 45  4.5! Advisory Committee Observations & Recommendations  The fact that hidden homeless youth had worse health outcomes was taken to be common knowledge by the advisory committee. The barriers to health that they believed hidden homeless youth experienced included not having access to a supportive adult and not wanting to access services because of anxiety or issues with certain services. The committee echoed the literature in describing youth not accessing services to avoid the label of being ‘homeless’ due to the associated stigma (Harter et al., 2005; May, 2015).  The committee cited several anecdotal experiences that demonstrated how hidden homeless youth were disadvantaged, including youth not being able to register for school as they were couch surfing and did not have the support of a parent or guardian, and youth getting fired because they were homeless.   According to the committee, hidden homeless youth have difficulty accessing housing programs. This is especially difficult for queer or Indigenous youth. Youth “in the middle” do not get support because they are not considered to be in an acute enough situation. The committee described “completely” homeless youth being eligible for services hidden homeless youth could not access.  In terms of service provision, the committee felt that staff had a lot to contribute to hidden homeless youth health. They stated that staff with lived experience, who treat them like equals, and know the youth well are especially helpful. Staff knowledge including knowledge of mental health, family mediation, and the government care system was especially important. Flexible hours helped youth access service.   Going forward, the advisory committee cited several systems level changes as being most important for hidden homeless youth health. Rent freezes, rent reduction, and rent subsidies were 46  important policy recommendations which the advisory committee believe would help stabilize youth. The advisory committee felt that it was important to have protection against landlords declining to rent to youth based on whether they were people of colour, queer, a different religion, on social assistance, or looked different (looked homeless).  The committee observed that youth lacking information was detrimental to homeless youth health. The committee stated that youth were often unaware of shelters and social services. They suggested that youth should be taught about where they can access services and resources while they are young and in school, along with life skills.   The advisory committee recommended investment in bigger shelters and more accessible BC housing. Youth with pets were noted to be excluded from many of these programs, leading to inequitable health despite the positive mental health impact of having pets. The committee identified that it is also crucial to have youth housing for people over 19.  To help youth of different genders, the advisory committee discussed inclusive housing as well as queer-specific housing. One individual mentioned that the gendered nature of shelters can make youth uncomfortable.   The advisory committee observed that young men were often less desirable as tenants and are sometimes excluded from family shelters. Young men also need access to housing and shelters.    The advisory committee mentioned the importance of more research in hidden homelessness.  47  Chapter 5:!Discussion: Entwined in human rights is the concept that people have the right to the highest standard of attainable health (Braveman, 2014). This study addresses a gap in the literature on hidden homelessness among youth and how hidden homelessness as a housing strategy affects health in British Columbia. Health equity research aims to help people access the highest standard of attainable health by investigating and intervening in the social determinants of health negatively influencing these individuals’ lives on a systemic level. This study demonstrates that youth hidden homelessness is associated with poor health outcomes suggesting that hidden homelessness is a key factor when we talk about youth homelessness.   Social exclusion and stigma occur differentially within homeless youth communities – I decided to focus on gender within this analysis, while recognizing that factors outside of this study’s purview – including ethnicity, racialization, disability, and sexual orientation – also play into homeless youths’ experiences of health (Abramovich, 2012; 2013; Conroy & Heer, 2003; Maccio & Ferguson, 2016; May, 2015; Perez & Romo, 2011; Watson, 2016; Watson & Cuervo, 2017). My intention is to combat the erasure of homeless young women and young trans people by discussing hidden homelessness and testing the commonly held assumption that the hidden homeless have better health than other homeless people. One of the reasons that the needs of hidden homeless people must be centered is that they are excluded from some of the most widely funded homelessness programs such as Housing First, and thus are unable to access financial benefits or social services (Katz, Zerger, & Hwang, 2016; Watson & Cuervo, 2017). This is especially important for youth, as young people are especially vulnerable and will bring their health behaviours into adulthood. Additionally, I hope to disrupt discourses which primarily 48  frame homelessness as not having a place of residence by discussing the factors of stigma and social exclusion that lead to this disenfranchisement (Katz et al., 2016; Watson & Cuervo, 2017).     5.1! Summary of Findings I put forth the following hypotheses:  1.! Youth who have used hidden homelessness shelter strategies will have poorer health outcomes in comparison to those who have not.  Adjusted logistic regression provides evidence that youth using hidden homelessness shelter strategies have significantly worse odds of the following outcomes: self-rated general health, self-rated mental health, having at least one mental health condition, having forgone medical health care, having foregone mental health care, having experienced an injury serious enough to require medical attention, harmful levels of stress, self-harm, having considered suicide in the last year, and having attempted suicide in the last year. Adjusted logistic regression did not provide supporting evidence that youth using hidden homelessness shelter strategies have significantly worse odds of harmful levels of despair.  2.! Young women and trans & gender diverse youth will have poorer health outcomes in comparison to young men.  Chi square analyses largely supported the hypothesis that young women and trans & gender diverse youth have poorer health outcomes in comparison to young men. Having self-rated poor or fair general health, having self-rated poor or fair mental health, having at least one mental health condition, having foregone mental health care, harmful levels of stress, harmful levels of despair, self-harm, having considered suicide in the last year, and having attempted suicide in the last year were significantly different between genders. Having foregone medical health care and having experienced a serious injury were not significantly different between genders. 49  3.! Multiplicative interaction will be present in the relationship between health outcomes and hidden homelessness by gender. The interaction terms between hidden homelessness and gender were significant for having foregone medical health care, and having experienced an injury serious enough to require medical attention, but only in the multiply imputed dataset. These analyses provided evidence to support rejecting the null hypothesis that the statistical effect of hidden homelessness on health is the same among genders for these health outcomes. The interaction terms between hidden homelessness and gender were not significant for the following outcomes: self-rated general health, self-rated mental health, having at least one mental health condition, having foregone mental health care, harmful levels of stress, harmful levels of despair, self-harm, having considered suicide in the last year, and having attempted suicide. Disaggregated odds ratios for trans & gender diverse youth could not be calculated within the logistic regression models due to small sample size.  These findings support the hypothesis that youth using hidden homelessness shelter strategies have poorer health outcomes than other homeless youth in every domain studied except for extreme despair. Moderation analysis supported difference of the statistical effect of hidden homelessness on health by gender for foregone medical care (for male youth) and a history of injuries (for both female and male youth), but with larger effects for female youth. 5.2! Contributions to the Literature Canadian research shows that the mortality rates of marginally housed adults are not different from those who live in shelters (Hwang, Wilkins, Tjepkema, O'Campo, & Dunn, 2009). While my work did not look at mortality measures, it similarly challenges the assumption that people on the streets or in shelters more acutely need services (Hwang et al., 2009). The finding 50  that hidden homeless youth largely have worse health outcomes relative to other homeless youth challenges common assumptions around the mechanisms behind homelessness-related ill health. One potential implication is that poor health is not solely due to housing as much as the situations that surround poverty, such as being in proximity to more people who may harm them and directly cause poor health outcomes or uncertainty and fear surrounding occupying insecure housing or being homeless (Shankardass, 2012). Hidden homeless youth my get inadequate support for finding resources such as appropriate housing or health services by the adults in their lives who do not understand the severity of their situations (McLoughlin, 2013). Some of the physical and mental health consequences of being forced into temporary living conditions may account for the health disparities described by this work – disrupted sleeping due to lack of space, privacy, or control, decreased access to food, increase in substance use, poor mental health related to stress, violence and uncertainty, and increased risk of physical harm – both physical and sexual – may lead to the ill health of the youth in this sample (McCoy & Hug, 2016).  The literature states that women are more likely to be part of the hidden homeless population (Johnson et al., 2017; May, 2015; North & Smith, 1993; Richter & Chaw-Kant, 2010; Watson, 2011). Female and trans & gender diverse youth took up a slightly larger proportion of the hidden homeless sample compared to the visible homeless sample. However, a chi squared test showed that there were no significant differences by gender between the hidden homeless sample and the rest of the overall sample. This may be due to a difference between youth and adult populations, since hidden homelessness is a key shelter strategy used by youth (Harter et al., 2005). 51  5.3! Limitations There are limitations to this work. The sampling was undertaken using convenience methods rather than systematic ones. As such, the original sample may not be a true reflection of the populations, and multiple imputation would amplify the patterns of the original sample (Enders, 2010). The most ideal design for this work would be a systematically sampled longitudinal study on homeless youth examining hidden homelessness. This design would be able to provide evidence on causation, and would be much more likely to be representative of the true population of homeless youth in British Columbia. The population of homeless youth is not static, and its members often intentionally try to remain hidden, therefore, there is not aknown sampling frame from which systematic sampling can be based. The cost and attrition of a longitudinal study has not proven feasible either. The HSIY provided a wealth of data given the difficulty of sampling and studying the population.  While measures were taken to make respondents feel comfortable, such as the surveys being conducted by a youth worker and a community co-researcher (a youth with experience of homelessness in a specific community), some of the questions asked were on stigmatized topics such as sexual activity, drug use, and mental health. Some of the youth may not have felt safe answering these sensitive questions. These limitations are somewhat offset by strengths of study design. The HSIY was conducted in multiple cities across British Columbia which helps with generalizability. The sample size of the HSIY was large for this sort of study.  5.4! Implications  Health inequities are a social injustice. In order to pursue health equity, public health must improve the health of the economically or socially disadvantaged – to improve them we 52  must determine if they exist and to what magnitude (Braveman, 2014). It behooves public health nurses and public health researchers to investigate the nature of gender-based health inequities within homeless populations to combat them. Homeless women and trans people of all genders suffer from poor health not only from homelessness but also from transphobia and misogyny. The healthcare system must examine if structures put in place to help homeless people may widen the health deficit gap between genders. The assumption that more men are homeless than other genders may result in an underestimation of the needs of homeless women, trans people, and gender diverse people (Gaetz et al., 2016). Indeed, these people receive the harms of not only being homeless, but being a homeless trans or gender diverse person or a homeless woman (Abramovich, 2013). This intersection of identities put these people in further risk of harm – my study suggests that this risk of harm is increased among the youth hidden homeless population.    Gender as a social determinant of health has been well established (Bungay, Johnson, Varcoe, & Boyd, 2010; Lyons et al., 2016; Poteat, Reisner, & Radix, 2014). Women who use drugs face stigma based on these two identities which can decrease access to health and overall health (Bungay et al., 2010). Trans women face barriers to accessing HIV prevention, testing and treatment, and HIV positive transwomen can experience extreme stigma due both their gender identity and serostatus (Poteat, Reisner, & Radix, 2014). Trans women and two-spirit people experience discrimination in housing programs leading to further health inequities (Lyons, Krüsi, Pierre, Smith, Small & Shannon, 2016). Intersections of identities, housing situation, health status, or substance use create structural barriers to those with oppressed gender identities – these barriers are detrimental to health (Bungay et al., 2010).    The constructs of gender and homelessness are notoriously difficulty to study; homeless people can be hard to find and recruit for research, and gendered experiences can be 53  misinterpreted (Golden, 2013; Richter & Chaw-Kant, 2010). As a result, there is a lot of pure conjecture circulating on how the two interrelate. As an example, an article from “A Voice for Men” suggested that the root cause of men’s homelessness is misandry, and cited men having been falsely accused (of sexual assault, presumably) and divorce as the reason they clearly bear the burden of homelessness (Golden, 2013). While a certain agenda is clearly at play in this example, it demonstrates a gap in our knowledge that is being filled without consulting those communities affected. From my professional work experience as a street nurse I know that gender complicates how many homeless youth navigate housing – there is more to the story than that women and trans people seldom become homeless relative to cis men. This work strengthens our understanding of how youth of different genders navigate homelessness, and what health associations they report.    While qualitative research is well positioned to explore intersectional population health research, quantitative methods (such as testing for interaction terms) have the potential to help us to explore intersectionality from a different angle (Bauer, 2014). This approach can contribute to our understanding of homelessness, and why we, as a society, need to address it. By teasing apart the relationship between health, shelter strategies, and gender, we can help build knowledge on the burden of ill-health and injury associated with gendered hidden homelessness. Combining quantitative and qualitative may provide the most robust and well developed evidence base to support policy makers.    An important element of gender-based health equity is studying existing gendered health dynamics to design health policies and systems to combat health disparities (Panisello & Pastor, 2015). Since interactions were present in two of the models studied, I have some evidence to 54  support that the relationship between hidden homelessness and health outcomes is different for different genders for foregoing medical care and having experienced a serious injury.  My findings suggest that hidden homelessness puts youth at much higher risk of experiencing worse health outcomes – which could have lasting impacts for the rest of their lives. There are gendered implications of this work – through these analyses I have demonstrated that the odds of female youth experiencing a serious injury associated with hidden homelessness was greater than the odds of the same for young men, yet the odds of foregoing medical care associated with hidden homelessness was not significant for female youth, yet were significant for male youth. Although I was not able to include trans & gender diverse youth in the gender-specific logistic regression models, it is clear from the descriptive analyses that they also experience very detrimental health outcomes in association with hidden homelessness.  With this research, I hope to emphasize the heterogeneity of homeless youth, and to point to the needs of some of those groups. I have applied a social justice lens to this work, looking not exclusively at the experiences of homeless people in general, but also at the intersections of identity within homelessness (Watson & Cuervo, 2017). Two of the objectives of the World Health Organizations’ Gender Policy include the following: to “provide qualitative and quantitative information on the influence of gender on health and health care” (World Health Organization, 2002) and to support the “integration of gender considerations…in all policies and programmes [through policy development and research]” (World Health Organization, 2002). I have addressed these objectives by conducting a gender-based analysis. 55  5.5! Conclusions  Public health should be wary of what goes uncounted. The hidden homeless as well as trans people are surely difficult to account for in research, yet neglecting to do so erases them for policy makers and health care services (Bernard et al., 2010).   Rooted in critical theory, I hope this thesis challenges current societal assumptions about homelessness, poverty, and gender (Wellmer, 2014). These results should caution policy makers, and service providers against overlooking the needs of hidden homeless youth.   Knowledge on hidden homeless youth will only become more crucial to public health decision making. The numbers of British Columbia youth who reported couch surfing rose from 37% in 2006 to 50% in 2014 (Smith et al., 2015). Hidden homelessness as a shelter strategy is becoming increasingly widespread as public spaces are increasingly policed, causing individuals to remain invisible (and, as a result, excluded from policy statistics) (May, 2015). Eighty percent of all homeless people in Canada are theorized to be using hidden homelessness shelter strategies (Crawley et al., 2013).   How can public health improve health outcomes moving forward? The youth advisory committee largely pointed at large structural barriers to health, including insufficient protective legislation for landlords and lack of education on how to access health and social services. The youth advisory committee discussed how to help hidden homeless youth with the understanding that some youth will inevitably experience homelessness.   As gentrification changes the urban landscapes within Canada, the numbers of visible homeless people will decline and hidden homelessness will become increasingly common, and we need to have public policy and health research to support the challenges that these people face (May, 2015). We, as a society, need to be ready to accommodate the changing 56  demographics of homelessness – the growing proportion of women, children, youth, and families who do not have security of tenure – and look at the reasons why this has come about (May, 2015).  I strongly believe in the importance of this work. I hope to leverage quantitative methods to support the work being done worldwide on how hidden homelessness affects people’s health. Inquiry into hidden homelessness from a trans-inclusive position will shed light on a group of people who are generally ignored in the context of homelessness research (Abramovich, 2012). I think it is important to end this thesis with the advisory committee’s conclusions, as young people with lived experience can provide us with the most meaningful recommendations. The advisory committee emphasized the importance of large scale societal change by strengthening programs for homeless youth, include hidden homeless youth in existing services for homeless people, bolstering affordable housing, and giving young people the tools to find the services they might need before they experience homelessness. The changes that homeless youth need must happen at a systemic level.  57  References Abramovich, A. (2012). No Safe Place to Go. Canadian Journal of Family and Youth, 4(1), 29–51. Abramovich, A. (2013). No Fixed Address: Young, Queer, and Restless. In S. Gaetz, B. O'Grady, K. Buccieri, J. Karabanow, & A. Marsolais (Eds.), Youth Homelessness in Canada: Implications for Policy and Practice (pp. 387–403). Abramovich, A. (2016). Understanding How Policy and Culture Create Oppressive Conditions for LGBTQ2S Youth in the Shelter System. Journal of Homosexuality, 00(00), 1–18. http://doi.org/10.1080/00918369.2016.1244449 Anthony, I. (2014). Meeting the Challenge. Clinical Pediatrics, 53(1), 38–40. http://doi.org/10.1177/0009922813499067 Bauer, G. R. (2014). Incorporating intersectionality theory into population health research methodology: Challenges and the potential to advance health equity. Social Science & Medicine, 110(C), 10–17. http://doi.org/10.1016/j.socscimed.2014.03.022 Bauer, G. R., Hammond, R., Travers, R., Kaay, M., Hohenadel, K., & Boyce, M. (2009). “I Don't Think This Is Theoretical; This Is Our Lives”: How Erasure Impacts Health Care for Transgender People. Journal of the Association of Nurses in AIDS Care, 20(5), 348–361. http://doi.org/10.1016/j.jana.2009.07.004 Bernard, H. R., Hallett, T., Iovita, A., Johnsen, E. C., Lyerla, R., McCarty, C., et al. (2010). Counting hard-to-count populations: the network scale-up method for public health. Sexually Transmitted Infections, 86(Suppl 2), ii11–ii15. http://doi.org/10.1136/sti.2010.044446 Biederman, D. J., & Nichols, T. R. (2014). Homeless Women's Experiences of Service Provider Encounters. Journal of Community Health Nursing, 31(1), 34–48. http://doi.org/10.1080/07370016.2014.868733 Braveman, P. (2014). What is Health Equity: And How Does a Life-Course Approach Take Us Further Toward It? Matern Child Health J, 18, 366–372. http://doi.org/10.1007/s10995-013-1226-9 Bungay, V., Johnson, J. L., Varcoe, C., & Boyd, S. (2010). Women's health and use of crack cocaine in context: Structural and 'everyday' violence. International Journal of Drug Policy, 21(4), 321–329. http://doi.org/10.1016/j.drugpo.2009.12.008 Canadian Institutes of Health Research: How to integrate sex and gender into research. (2018). Canadian Institutes of Health Research: How to integrate sex and gender into research. Retrieved March 2, 2018, from Canadian Institutes of Health Research: Sex, gender and health research in Canada. (2014). Canadian Institutes of Health Research: Sex, gender and health research in Canada. Retrieved March 2, 2018, from Centre for Addiction and Mental Health Hidden in Our Midst: Homeless Newcomer Youth in Toronto – Uncovering the Supports to Prevent and Reduce Homelessness. (2014). Hidden in Our Midst: Homeless Newcomer Youth in Toronto – Uncovering the Supports to Prevent and Reduce Homelessness (pp. 1–36). Centre for Addiction and Mental Health. Conroy, S. J., & Heer, D. M. (2003). Hidden Hispanic Homelessness in Los Angeles: The “Latino Paradox” Revisited. Hispanic Journal of Behavioral Sciences, 25(4), 530–538. http://doi.org/10.1177/0739986303258126 Crawley, J., Kane, D., Atkinson-Plato, L., Hamilton, M., Dobson, K., & Watson, J. (2013). 58  Needs of the hidden homeless - no longer hidden: a pilot study. Public Health, 127(7), 674–680. http://doi.org/10.1016/j.puhe.2013.04.006 Denton, M., Prus, S., & Walters, V. (2004). Gender differences in health: a Canadian study of the psychosocial, structural and behavioural determinants of health. Social Science & Medicine, 58(12), 2585–2600. http://doi.org/10.1016/j.socscimed.2003.09.008 Eberle, M., Kraus, D., Serge, L., Mustel Research Group, & Inc, M. R. (2009). Results of the pilot study to estimate the size of the hidden homeless population in Metro Vancouver (pp. 1–59). Vancouver. Enders, C. K. (2010). Applied Missing Data Analysis (pp. 1–401). New York & London: The Guilford Press. Fast, D., Shoveller, J., Shannon, K., & Kerr, T. (2009). Safety and danger in downtown Vancouver Understandings of place among young people entrenched in an urban drug scene. Health & Place, 16(1), 51–60. http://doi.org/10.1016/j.healthplace.2009.07.004 Fitzpatrick, S. (1998). Hidden Homelessness Amongst Young People. Youth and Policy, 8–22. Gaetz, S., Dej, E., Richter, T., & Redman, M. (2016). The State of Homelessness in Canada 2016 (pp. 1–85). Canadian Observatory on Homelessness. Golden, T. (2013). Is homelessness a men’s issue? Retrieved from https://www.avoiceformen.com/men/mens-issues/is-homelessness-a-mens-issue/ Harter, L. M., Berquist, C., Scott Titsworth, B., Novak, D., & Brokaw, T. (2005). The Structuring of Invisibility Among the Hidden Homeless: The Politics of Space, Stigma, and Identity Construction. Journal of Applied Communication Research, 33(4), 305–327. http://doi.org/10.1080/00909880500278079 Homelessness Partnering Strategy. (2017). Everyone Counts: A guide to point-in-time counts in Canada. Employment and Social Development Canada. Hudson, A. L., Nyamathi, A., Greengold, B., Slagle, A., Schneider, A., Koniak-Griffin, D., et al. (2010). Health-Seeking Challenges Among Homeless Youth. Nursing Research, 59(3), 212–218. http://doi.org/10.1097/NNR.0b013e3181d1a8a9 Hwang, S. W., Wilkins, R., Tjepkema, M., O'Campo, P. J., & Dunn, J. R. (2009). Mortality among residents of shelters, rooming houses, and hotels in Canada: 11 year follow-up study. British Medical Journal, 339(3), b4036–b4036. http://doi.org/10.1136/bmj.b4036 Johnson, G., Ribar, D. C., & Zhu, A. (2017). Women’s Homelessness: International Evidence on Causes, Consequences, Coping and Policies (pp. 1–42). IZA Institute of Labor Economics. Katz, A. S., Zerger, S., & Hwang, S. W. (2016). Housing First the conversation: discourse, policy and the limits of the possible. Critical Public Health, 1–9. http://doi.org/10.1080/09581596.2016.1167838 Laird, G. (2007). Homelessness in a growth economy: Canada’s 21st century paradox, 1–98. Little, T. D., Jorgensen, T. D., Lang, K. M., & Moore, E. W. G. (2014). On the Joys of Missing Data. Journal of Pediatric Psychology, 39(2), 151–162. http://doi.org/10.1093/jpepsy/jst048 Lyons, T., Krüsi, A., Pierre, L., Smith, A., Small, W., & Shannon, K. (2016). Experiences of Trans Women and Two-Spirit Persons Accessing Women-Specific Health and Housing Services in a Downtown Neighborhood of Vancouver, Canada. LGBT Health, 3(5), 373–378. http://doi.org/10.1089/lgbt.2016.0060 Maccio, E. M., & Ferguson, K. M. (2016). Services to LGBTQ runaway and homeless youth: Gaps and recommendations. Children and Youth Services Review, 63(C), 47–57. http://doi.org/10.1016/j.childyouth.2016.02.008 59  May, J. (2015). “Gone, leave, go, move, vanish”: race, public space and (in)visibilities. Social Identities: Journal for the Study of Race, Nation and Culture, 489–505. http://doi.org/10.1080/13504630.2015.1093468 McCoy, S., & Hug, B. (2016). Danger Zones and Stepping Stones: Young people's experiences of hidden homelessness (pp. 1–33). McLoughlin, P. J. (2013). Couch surfing on the margins: the reliance on temporary living arrangements as a form of homelessness amongst school-aged home leavers. Journal of Youth Studies, 16(4), 521–545. http://doi.org/10.1080/13676261.2012.725839 Mizock, L., & Russinova, Z. (2015). Intersectional Stigma and the Acceptance Process of Women with Mental Illness. Women & Therapy, 38(1-2), 14–30. http://doi.org/10.1080/02703149.2014.978211 North, C. S., & Smith, E. M. (1993). A Comparison of Homeless Men and Women: Different Populations, Different Needs. Community Mental Health, 423–431. O'Grady, B., & Gaetz, S. (2004). Homelessness, Gender and Subsistence: The Case of Toronto Street Youth. Journal of Youth Studies, 7(4), 397–416. http://doi.org/10.1080/1367626042000315194 Panisello, M. L., & Pastor, I. (2015). Health with equality: a proposal for the incorporation of the gender perspective in health care systems. Ciência & Saúde Coletiva, 20(5), 1555–1563. http://doi.org/10.1590/1413-81232015205.10942014 Perez, B. F., & Romo, H. D. (2011). ``Couch surfing“” of Latino foster care alumni: Reliance on peers as social capital. Journal of Adolescence, 34(2), 239–248. http://doi.org/10.1016/j.adolescence.2010.05.007 Poteat, T., Reisner, S. L., & Radix, A. (2014). HIV epidemics among transgender women. Current Opinion in HIV and AIDS, 9(2), 168–173. http://doi.org/10.1097/COH.0000000000000030 Richter, S., & Chaw-Kant, J. (2010). Canadian homeless women: Gaps in the research agenda. International Journal of Child Health and Human Development, 3(1), 7–16. Robert, M., Pauzé, R., & Fournier, L. (2005). Factors associated with homelessness of adolescents under supervision of the youth protection system. Journal of Adolescence, 28(2), 215–230. http://doi.org/10.1016/j.adolescence.2005.02.004 Sakamoto, I., Chin, M., Chapra, A., & Ricciardi, J. (2009). A “normative” homeless woman?: Marginalisation, emotional injury and social support of transwomen experiencing homelessness. Gay Lesbian Issues and Psychology Review, 5(1), 2–19. Shankardass, K. (2012). Place-Based Stress and Chronic Disease: A Systems View of Environmental Determinants. In P. O Campo & J. R. Dunn (Eds.), Rethinking Social Edidemiology (pp. 113–136). Dordrecht: Springer. http://doi.org/10.1007/978-94-007-2138-8 Smith, A., Stewart, D., Poon, C., Peled, M., McCreary Centre Society. (2014). From Hastings Street to Haida Gwaii (pp. 1–92). Vancouver, BC. Smith, A., Stewart, D., Poon, C., Peled, M., Saewyc, E., McCreary Centre Society. (2015). Our communities, our youth: The health of homeless and street-involved youth in BC, 1–72. Smith, J. (2013). Methodological problems of sampling young homeless people in four European societies with different levels of service provision and definitions of homelessness. Housing, Care and Support, 16(2), 64–75. http://doi.org/10.1108/HCS-03-2013-0005 Social Planning and Research Council of BC. (2011). Knowledge for Action: Hidden 60  Homelessness in Prince George, Kamloops, Kelowna, Nelson and Nanaimo (pp. 1–150). Burnaby, BC. The Homeless Hub. (2017). Youth Count Toolkit. Retrieved April 4, 2017, from Thurston, W. E., Roy, A., Clow, B., Este, D., Gordey, T., Haworth-Brockman, M., et al. (2013). Pathways Into and Out of Homelessness: Domestic Violence and Housing Security for Immigrant Women. Journal of Immigrant & Refugee Studies, 11(3), 278–298. http://doi.org/10.1080/15562948.2013.801734 Tomas, A., & Dittmar, H. (1995). The Experience of Homeless Women: An Exploration of Housing Histories and the Meaning of Home. Housing Studies, 10(4), 493–516. Tripathy, J. P. (2013). Secondary Data Analysis: Ethical Issues and Challenges, 42(12), 1478–1479. Tutty, L., Koshan, J., Jesso, D., Ogden, C., & Warrell, J. G. (2011). Evaluation of the Calgary Specialized Domestic Violence Trial Court & Monitoring the First Appearance Court: Final Report (pp. 1–209). Vacha, E. F., & Marin, M. V. (1993). Doubling Up: Low income Households Sheltering the Hidden Homeless. Journal of Sociology Social Welfare, 20(3), 25–41. Vittinghoff, E., Glidden, D. V., Shiboski, S. C., & McCulloch, C. E. (2007). Statistics for Biology and Health: Linear, Logistic, Survival, and Repeated Measures Models (pp. 1–346). Watson, J. (2011). Understanding survival sex: young women, homelessness and intimate relationships. Journal of Youth Studies, 14(6), 639–655. http://doi.org/10.1080/13676261.2011.588945 Watson, J. (2016). Gender-Based Violence and Young Homeless Women: Femininity, Embodiment and Vicarious Physical Capital. The Sociological Review, 64(2), 256–273. http://doi.org/10.1111/1467-954X.12365 Watson, J., & Cuervo, H. (2017). Youth homelessness: A social justice approach. Journal of Sociology, 461–475. Wellmer, A. (2014). On Critical Theory, 1–31. Whitzman, C. (2006). At the Intersection of Invisibilities: Canadian women, homelessness and health outside the “big city.” Gender, Place & Culture, 13(4), 383–399. http://doi.org/10.1080/09663690600808502 World Health Organization. (2002). Integrating Gender Perspectives in the work of WHO. (G. H. Brundtland, Ed.) (pp. 1–6). World Health Organization. (2014). Health for the world's adolescents: A second chance in the second decade. Retrieved from http://apps.who.int/adolescent/second-decade/section2/page1/recognizing-adolescence.html  61  Appendices Appendix A  -  Literature Review Inclusion & Exclusion Criteria  Author (Year) Article Title Eligibility Location Methodology Population Definition of Hidden Homelessness Definition of Youth Anthony, I. (2014) Meeting the Challenge: Responding to the Health Care Needs of Homeless Youth Excluded  Article does not discuss HH specifically.       Auerswald, C.L. (2013)   Hidden in Plain Sight: An Assessment of Youth Inclusion in Point-in-Time Counts of California’s Unsheltered Homeless Population Excluded  Article does not discuss HH specifically.      CAMH (2014)   Hidden in Our Midst: Homeless Newcomer Youth in Toronto – Included Toronto, Ontario Mixed methods – Multi-method qualitative and quantitative approach Homeless youth, service providers in the City of Toronto, those serving homeless Staying with family or friends. Having no permanent, safe, affordable, and adequate home.  16-24 62  Author (Year) Article Title Eligibility Location Methodology Population Definition of Hidden Homelessness Definition of Youth Uncovering the Supports to Prevent and Reduce Homelessness including use of participatory action research youth, & those serving newcomer communities Crawley J. (2013) Needs of the hidden homeless e no longer hidden: a pilot study Excluded  Article does not disaggregate youth.       Fitzpatrick, S. (1998) Hidden Homelessness Amongst Young People Included Drumchapel, Glasgow, UK Qualitative (unstated) Youth 16-25 People who move between friends’ and relatives’ houses and sleep rough in unknown sites, staying in an institution because there is nowhere else to go, enduring relationships which or violent or highly unsatisfactory, or living in intolerable or insecure 16-25 63  Author (Year) Article Title Eligibility Location Methodology Population Definition of Hidden Homelessness Definition of Youth accommodation (vs visible homelessness who are recorded in official statistics, are in contact with homeless agencies, and/ or staying in the official network of homeless accommodation or sleeping rough in visible or known areas) Gausvik, C. (2015) “I Spent Most of My Life Just Walking Around, Trying to Find a Place to Sleep”: Describing the Experience of Youth Hidden Homelessness Excluded  Thesis       Harter, L. (2005) The Structuring of Invisibility Included 2 Midwestern Qualitative - Ethnographic Educators, case managers, and  1st Site: School 64  Author (Year) Article Title Eligibility Location Methodology Population Definition of Hidden Homelessness Definition of Youth Among the Hidden Homeless: The Politics of Space, Stigma, and Identity Construction communities of 200,000 & 250,000, USA youth at a support program for youth without homes aged children 2nd Site: 17-21 Johnson, E. (2014) Ronald E. Hallett: Educational Experiences of Hidden Homeless Teenagers: Living Doubled-Up Excluded  Book Review      May, J. (2013) Race, Gender, Youth, and Urban Space: Young Men of Colour and Homelessness in the Greater Toronto Area Excluded  Thesis      May, J. (2015)   ‘Gone, leave, go, move, vanish’: race, public space Included Greater Toronto Area, Canada Qualitative (unstated) Young men of colour Invisible homelessness refers to those obscured from 17-26 65  Author (Year) Article Title Eligibility Location Methodology Population Definition of Hidden Homelessness Definition of Youth and (in)visibilities public view through house hopping or couch surfing, living in unaffordable or inadequate housing or living in temporary shelters.   Visible homeless people might sleep rough, or panhandle. McCoy, S. (2016) Danger Zones and Stepping Stones: Young people’s experiences of hidden homelessness Included Oldham, Sheffield, Whitley Bay, London, and Gravesend.  Qualitative (unstated)  18 young people  Exclusively looked at sofa (couch) surfing. Sofa surfing is an umbrella term that encompasses different living arrangements, and is ambiguous and complex (who people are staying with and how safe that place is).  16-23 66  Author (Year) Article Title Eligibility Location Methodology Population Definition of Hidden Homelessness Definition of Youth McCloughlin, P. J. (2011) Couch surfing through homelessness in the lives of Australian youth Excluded  Thesis       McCloughlin, P. J. (2013) Couch surfing on the margins: the reliance on temporary living arrangements as a form of homelessness amongst school-aged home leavers Included South Australia  Qualitative – Grounded Theory Young men and women who have couch surfed Couchsurfing for a period lasting for at least 2 months.  15-25 O’Grady, B., & Gaetz, S. (2004) Homelessness, Gender and Subsistence: The Case of Toronto Street Youth   Included Toronto, Ontario Mixed Methods (unstated) – Participatory Action Research 360 youth – questionnaire, 20 youth - open-ended/semi-structured, tape recorded interviews Does not specifically describe HH.   O’Grady, B., & Gaetz, S. (2009) Street Survival: A Gendered Analysis Included Toronto, Ontario Mixed Methods (unstated) – Participatory 360 youth – questionnaire, 20 youth - open-Does not specifically describe HH.  67  Author (Year) Article Title Eligibility Location Methodology Population Definition of Hidden Homelessness Definition of Youth of Youth Homelessness in Toronto Action Research ended/semi-structured, tape recorded interviews Perez, B.F. & Romo, H.D (2011) “Couch surfing” of Latino foster care alumni: Reliance on peers as social capital Included Major urban area in Texas Qualitative- emerging research design Latino youth that had been in foster care Couch surfing Post foster care experiences 18-22 Robert, M., Fournier, L., & Pauzé, R. (2003) Examen des caractéristiques cliniques familiales et personnelles associées a I'itinérance cachée d'adolescent(e)s en difficulté Excluded   French language      Robert, M., Pauzé, R., & Fournier, L. (2005) Factors associated with homelessness of adolescents under supervision of the youth Included Quebec Quantitative (unstated) – cross sectional Children under 18 under the supervision of four Youth Protection Centres in Quebec Using social network rather than public spaces. Those not living on the street or in shelters. Under 18 68  Author (Year) Article Title Eligibility Location Methodology Population Definition of Hidden Homelessness Definition of Youth protection system Smith, A., Stewart, D., Poon, C., Peled, M., Saewyc, E., and McCreary Centre Society (2015) Our communities, our youth:  The health of homeless and street-involved youth in BC  Included British Columbia Quantitative – cross sectional (unstated) 681 youth aged 12-19 across BC  Does not specifically describe HH. 12-19 Smith, J. (2013) Methodological problems of sampling young homeless people in four European societies with different levels of service provision and definitions of homelessness Excluded  Article does not discuss HH specifically.      Table A.1 Literature review inclusion & exclusion criteria 69  Appendix B  -  Model Differences B.1! Original Dataset  B1 SEB1 Change % B1 Change % SEB1 ModelSRGH 0.47 0.17   ModelSRGH1 0.44 0.18 -6.38 5.88 ModelSRGH2 0.46 0.17 -2.13 0.00 ModelSRMH 0.49 0.18   ModelSRMH1 0.55 0.18 12.24 0.00 ModelSRMH2 0.50 0.19 -9.09 5.56 ModelMHC 1.02 0.19   ModelMHC1 0.97 0.19 -4.90 0.00 ModelMHC2 0.96 0.19 -5.88 0.00 ModelFMC 1.15 0.24   ModelFMC1 1.13 0.24 -1.74 0.00 ModelFMC2 1.12 0.24 -2.61 0.00 ModelFMHC 1.04 0.21   ModelFMHC1 1.05 0.21 0.96 0.00 ModelFMHC2 1.04 0.21 0.00 0.00 ModelI 1.07 0.22   ModelI1 1.09 0.22 1.87 0.00 ModelI2 1.03 0.22 -3.74 0.00 ModelS 0.90 0.25   ModelS1 0.95 0.25 5.56 0.00 ModelS2 0.89 0.27 -1.11 8.00 ModelD 0.56 0.31   ModelD1 0.49 0.31 -12.50 0.00 ModelD2 0.52 0.32 6.12 3.23 ModelSH 0.33 0.18   ModelSH1 0.59 0.19 78.79 5.56 ModelSH2 0.56 0.19 -5.08 0.00 Model CS 0.79 0.19   Model CS1 0.85 0.19 7.59 0.00 Model CS2 0.75 0.19 -5.06 0.00 Model AS 0.71 0.21   Model AS1 0.85 0.22 13.33 4.76 Model AS2 0.86 0.22 1.18 0.00 Table B.1 Model differences: Original dataset 70  B.2! Imputed Dataset  B1 SEB1 Change % B1 Change % SEB1 ModelSRGH 0.42 0.16   ModelSRGH1 0.39 0.17 -7.14 6.25 ModelSRGH2 0.41 0.16 -2.38 0.00 ModelSRMH 0.39 0.16   ModelSRMH1 0.44 0.17 12.82 6.25 ModelSRMH2 0.39 0.17 -11.36 0.00 ModelMHC 1.07 0.17   ModelMHC1 1.02 0.17 -4.67 0.00 ModelMHC2 1.01 0.17 -5.61 0.00 ModelFMC 1.08 0.21   ModelFMC1 1.08 0.22 0.00 4.76 ModelFMC2 1.06 0.22 -1.85 4.76 ModelFMHC 0.99 0.19   ModelFMHC1 1.01 0.20 2.02 5.26 ModelFMHC2 0.98 0.19 -1.01 0.00 ModelI 1.06 0.20   ModelI1 1.05 0.20 -0.94 0.00 ModelI2 1.03 0.20 -2.83 0.00 ModelS 0.71 0.21   ModelS1 0.71 0.22 0.00 4.76 ModelS2 0.71 0.22 0.00 4.76 ModelD 0.42 0.27   ModelD1 0.35 0.28 -16.67 3.70 ModelD2 0.38 0.28 8.57 0.00 ModelSH 0.33 0.16   ModelSH1 0.60 0.18 81.82 9.38 ModelSH2 0.54 0.18 -10.00 2.86 Model CS 0.73 0.17   Model CS1 0.79 0.18 8.22 5.88 Model CS2 0.69 0.17 -5.48 -5.56 Model AS 0.67 0.19   Model AS1 0.83 0.19 20.29 0.00 Model AS2 0.82 0.20 -1.20 5.26 Table B.2 Model differences: Imputed dataset  71  Appendix C  Model Building: Original Dataset C.1! Self-Reported Poor or Fair General Health Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.46 0.17 1.59 (1.14-2.24)  HH + Age 0.44 0.18  Δ in B1: -6.38% Δ in SEB1: 5.88%  Age did not significantly change the B1, therefore it will not be included in the model.  HH + Location 0.46 0.17  Δ in B1: -2.13% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model. HH + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.1 Model building: Self-reported poor or fair general health  C.2! Self-Reported Poor or Fair Mental Health Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.49 0.18 1.63 (1.15-2.31)  HH + Age 0.55 0.18 1.73 (1.21-2.48) Δ in B1: 12.24% Δ in SEB1: 0%  Age significantly increased the B1 without increasing the SEB1, therefore it will be included in the model. HH + Age + Location 0.50 0.19  Δ in B1: -9.09% Δ in SEB1: 5.56%  72  Location did not significantly change the B1, therefore it will not be included in the model. HH + Age + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.2 Model building: Self-reported poor or fair mental health C.3! Mental Health Condition Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 1.02 0.19 2.77 (1.92-4.00)  HH + Age 0.97 0.19  Δ in B1: -4.90% Δ in SEB1: 0%  Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 0.96 0.19  Δ in B1: -5.88% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model. HH + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.3 Model building: Mental health condition C.4! Foregone Medical Care Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 1.15 0.24 3.15 (2.01-5.09)  HH + Age 1.13 0.24  Δ in B1: -2.61% Δ in SEB1: 0%  73  Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 1.12 0.24  Δ in B1: -.88% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model. HH + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.4 Model building: Foregone medical health care C.5! Foregone Mental Health Care Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 1.04 0.21 2.82 (1.90-4.26)  HH + Age 1.05 0.21  Δ in B1: 0.96% Δ in SEB1: 0%  Age did not significantly change the B1, therefore it will not be included in the model. HH + Age + Location 1.04 0.21  Δ in B1: 0% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model HH + Age + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.5 Model building: Foregone mental health care 74  C.6! Injury Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 1.07 0.22 2.93 (1.94-4.52)  HH + Age 1.09 0.22  Δ in B1: 1.87% Δ in SEB1: 0%  Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 1.03 0.22  Δ in B1: -3.74% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model HH + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.6 Model building: Injury C.7! Stress Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.90 0.25 2.47 (1.54-4.10)  HH + Age 0.95 0.25  Δ in B1: 5.56% Δ in SEB1: 0%  Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 0.89 0.27  Δ in B1: -1.11% Δ in SEB1: 8.00%  Location did not significantly change the 75  B1, therefore it will not be included in the model HH + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.7 Model building: Stress C.8! Despair Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.56 0.31 1.75 (0.97-3.33)  HH + Age 0.49 0.31 1.64 (0.90-1.28) Δ in B1: -12.50% Δ in SEB1: 0%  Age significantly decreased the B1 without increasing the SEB1, therefore it will be included in the model. HH + Age + Location 0.52 0.32  Δ in B1: 6.12% Δ in SEB1: 3.23%  Location did not significantly change the B1, therefore it will not be included in the model HH + Age + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.8 Model building: Despair C.9! Self-Harm Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.33 0.18 1.40 (0.99-1.99)  HH + Age 0.59 0.19 1.80 (1.24-2.63) Δ in B1: 78.79% Δ in SEB1: 5.56%  76  Age significantly increased the B1 without significantly increasing the SEB1, therefore it will be included in the model. HH + Age + Location 0.56 0.19  Δ in B1: -5.08% Δ in SEB1: 3.23%  Location did not significantly change the B1, therefore it will not be included in the model HH + Age + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.9 Model building: Self-harm C.10! Consider Suicide Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.79 0.19 2.20 (1.53-3.19)  HH + Age 0.85 0.19  Δ in B1: 7.59% Δ in SEB1: 0%  Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 0.75 0.19  Δ in B1: -5.06% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model. HH + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.2 Model building: Consider suicide 77  C.11! Attempt Suicide Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.71 0.21 2.03 (1.36-3.09)  HH + Age 0.85 0.22 2.33 (1.54-3.60) Δ in B1: 13.33% Δ in SEB1: 4.76%  Age significantly increased the B1 without significantly increasing the SEB1, therefore it will be included in the model. HH + Age + Location 0.86 0.22  Δ in B1: 1.18% Δ in SEB1: 0%  Location did not  significantly change the B1, therefore it will not be included in the model. HH + Age + Gender + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table C.3 Model building: Attempt suicide Appendix D  Model Building: Imputed Dataset D.1! Self-Reported Poor or Fair General Health Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.42 0.16 1.52 (1.11-2.10)  HH + Age 0.39 0.18  Δ in B1: -7.14% Δ in SEB1: 6.25%  Age did not significantly change the B1, therefore it will not be included in the model.  HH + Location 0.41 0.16  Δ in B1: -2.38% Δ in SEB1: 0%  78  Location did not significantly change the B1, therefore it will not be included in the model. HH + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table D.4 Model building: Self-reported poor or fair general health D.2! Self-Reported Poor or Fair Mental Health Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.39 0.16 1.48 (1.08-2.03)  HH + Age 0.44 0.17 1.55 (1.12-2.15) Δ in B1: 12.24% Δ in SEB1: 6.25%  Age significantly increased the B1 without increasing the SEB1, therefore it will be included in the model. HH + Age + Location 0.39 0.17 1.48 (1.07-2.06) Δ in B1: -11.36% Δ in SEB1: 0%  Location significantly changed the B1, therefore it will be included in the model. HH + Age + Location + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table D.5 Model building: Self-reported poor or fair mental health D.3! Mental Health Condition Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 1.07 0.17 2.91 (2.10-4.05)  HH + Age 1.02 0.17  Δ in B1: -4.67% Δ in SEB1: 0% 79   Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 1.01 0.17  Δ in B1: -5.60% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model. HH + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table D.3 Model building: Mental health condition D.4! Foregone Medical Care Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 1.08 0.21 2.94 (1.95-4.52)  HH + Age 1.08 0.22  Δ in B1: 0% Δ in SEB1: 4.76%  Age did not change the B1, therefore it will not be included in the model. HH + Location 1.06 0.22  Δ in B1: -1.85% Δ in SEB1: 4.76%  Location did not significantly change the B1, therefore it will not be included in the model. HH + Gender + HH + HH*Gender   Trans & Gender Diverse Youth: 0.31 (0.23-46.96) Female Youth:  3.22 (0.90-12.57) Male Youth: 3.22 (1.19-3.59) The interaction term for trans & gender diverse*male youth was significant at a p-value of 0.02 therefore, it will be included in the model. Table D.4 Model building: Foregone Medical Care 80  D.5! Foregone Mental Health Care Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.99      0.19 2.69 (1.87-3.93)  HH + Age 1.01 0.20  Δ in B1: 2.02% Δ in SEB1: 5.26%  Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 1.98 0.19  Δ in B1: -1.01% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model HH + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table D.5 Model building: Foregone mental health care D.6! Injury Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 1.06 0.20 2.88 (1.98-4.28)  HH + Age 1.05 0.20  Δ in B1: -.94% Δ in SEB1: 0%  Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 1.03 0.22  Δ in B1: -2.83% Δ in SEB1: 0%  Location did not significantly change the 81  B1, therefore it will not be included in the model HH + Gender + HH + HH*Gender   Trans & Gender Diverse Youth: 4.44 (0.31-169.53) Female Youth:  4.49 (1.21-17.36) Male Youth: 1.82 (1.08-3.14) The interaction term for female*male was significant at a p-value of 0.03 therefore, it will be included in the model. Table D.6 Model building: Injury D.7! Stress Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.71 0.21 2.04 (1.36-3.13)  HH + Age 0.71 0.22  Δ in B1: 0% Δ in SEB1: 4.76%  Age did not significantly change the B1, therefore it will not be included in the model. HH + Location 0.71 0.22  Δ in B1: 0% Δ in SEB1: 4.76%  Location did not significantly change the B1, therefore it will not be included in the model HH + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table D.7 Model building: Stress D.8! Despair Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.42 0.27 1.52 (0.91-2.63)  HH + Age 0.35 0.28 1.41 (0.83-2.48) Δ in B1: -16.67% Δ in SEB1: 3.70% 82   Age significantly decreased the B1 without increasing the SEB1 significantly, therefore it will be included in the model. HH + Age + Location 0.38 0.28  Δ in B1: 8.57% Δ in SEB1: 0%  Location did not significantly change the B1, therefore it will not be included in the model HH + Age + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table D.8 Model building: Despair D.9! Self-Harm Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.33 0.16 1.39 (1.01-1.91)  HH + Age 0.60 0.18 1.81 (1.29-2.57) Δ in B1: 81.82% Δ in SEB1: 9.38%  Age significantly increased the B1 without significantly increasing the SEB1, therefore it will be included in the model. HH + Age + Location 0.54 0.18 1.72 (1.22-2.45) Δ in B1: -10.00% Δ in SEB1: 2.86%  Location significantly decreased the B1, therefore it will be included in the model. HH + Age + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. 83  Table D.9 Model building: Self-harm D.10! Consider Suicide Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.73 0.17 2.07 (1.49-2.90)  HH + Age 0.79 0.18  Δ in B1: 8.22% Δ in SEB1: 5.88%  Age did not significantly change the B1, therefore it will not be included in the model.  HH + Location 0.69 0.17  Δ in B1: -5.48% Δ in SEB1: -5.56%  Location did not significantly change the B1, therefore it will not be included in the model. HH + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model.  Table D.10 Model building: Consider suicide D.11! Attempt Suicide Model B1 SEBI Estimated Measure of Association (95% Confidence Interval) Decision on Variable Inclusion in Model HH 0.67 0.19 1.96 (1.37-2.83)  HH + Age 0.83 0.19 2.30 (1.58-3.39) Δ in B1: 20.29% Δ in SEB1: 4.76%  Age significantly increased the B1 without significantly increasing the SEB1, therefore it will be included in the model. HH + Age + Location 0.82 0.20  Δ in B1: -1.21% Δ in SEB1: 5.26% 84   Location did not  significantly change the B1, therefore it will not be included in the model. HH + Age + Gender + HH + HH*Gender    Interaction term was not significant; therefore, it will not be included in the model. Table D.11 Model building: Attempt suicide   85  Appendix E  Unadjusted Odds Ratios E.1! Unadjusted Odds Ratios of Hidden Homelessness on Various Health Outcomes Model B1 SEB1 Unadjusted Odds Ratio of Hidden Homelessness (95% Confidence Interval) Self-Reported Poor or Fair General Health 0.46 0.17 1.59 (1.14-2.24) Self-Reported Poor or Fair Mental Health 0.49 0.18 1.63 (1.15-2.31) Mental Health Condition 1.02 0.19 2.77 (1.92-4.00) Foregone Medical Care 1.15 0.24 3.15 (2.01-5.09) Foregone Mental Health Care 1.04      0.21 2.82 (1.90-4.26) Injury 1.07 0.22 2.93 (1.94-4.52) Stress 0.90 0.25 2.47 (1.54-4.10) Despair 0.56 0.31 1.75 (0.97-3.33) Self-Harm 0.33 0.18 1.40 (0.99-1.99) Consider Suicide 0.79 0.19 2.20 (1.53-3.19) Attempt Suicide 0.71 0.21 2.03 (1.36-3.09) Table E.3 Unadjusted odds ratios of hidden homelessness on various health outcomes  E.2! Unadjusted Odds Ratios of Hidden Homelessness on Various Health Outcomes Model B1 SEB1 Unadjusted Odds Ratio of Hidden Homelessness (95% Confidence Interval) Self-Reported Poor or Fair General Health 0.42 0.16 1.52 (1.11-2.10) Self-Reported Poor or Fair Mental Health 0.39 0.16 1.48 (1.08-2.03) 86  Mental Health Condition 1.07 0.17 2.91 (2.10-4.05) Foregone Medical Care 1.08 0.21 2.94 (1.95-4.52) Foregone Mental Health Care 0.99      0.19 2.69 (1.87-3.93) Injury 1.06 0.19 2.88 (1.98-4.28) Stress 0.71 0.21 2.04 (1.36-3.13) Despair 0.42 0.27 1.52 (0.91-2.63) Self-Harm 0.33 0.16 1.39 (1.01-1.91) Consider Suicide 0.73 0.17 2.07 (1.49-2.90) Attempt Suicide 0.67 0.18 1.96 (1.37-2.83) Table E.4 Unadjusted odds ratios of hidden homelessness on various health outcomes in imputed sample  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0372353/manifest

Comment

Related Items