Hands Community Health Centre, Toronto, Ontario, Canada, 3Department of ... 5Faculty of Information and Media Studies Doctoral Program FIMS & Nursing ...
Health and Social Care in the Community (2017) 25(3), 1139–1150
doi: 10.1111/hsc.12414
Depression and discrimination in the lives of women, transgender and gender liminal people in Ontario, Canada Charmaine C. Williams PhD1 , Deone Curling PhD2, Leah S. Steele PhD3, Margaret F. Gibson 4 Andrea Daley PhD , Datejie Cheko Green MES5 and Lori E. Ross PhD6
4
PhD
,
1
Factor-Inwentash Faculty of Social Work, University of Toronto, Toronto, Ontario, Canada, 2Women’s Health in Women’s Hands Community Health Centre, Toronto, Ontario, Canada, 3Department of Family and Community Medicine, St. Michael’s Hospital, University of Toronto, Toronto, Ontario, Canada, 4School of Social Work, York University, Toronto, Ontario, Canada, 5Faculty of Information and Media Studies Doctoral Program FIMS & Nursing Building, Room 2050 The University of Western Ontario London, Ontario, Canada and 6Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, Canada
Accepted for publication 27 November 2016
Correspondence Charmaine C. Williams Factor-Inwentash Faculty of Social Work, University of Toronto 246 Bloor Street West Toronto, Ontario M5S 1V4, Canada E-mail: charmaine.williams@ utoronto.ca
What is known about this topic
• •
•
Research suggests that women are at higher risk for depression and are treated for it more often than men. Intersectionality theory has emerged as a framework for understanding how women with intersecting marginalised identities (i.e. racial minority, gender minority, sexual minority, living in poverty) may face even higher risks for depression. It is unclear how quantitative empirical analyses can best account for intersectionality and associated converging vectors of oppression.
What this paper adds
• •
•
Analyses revealed that race, gender, class and sexuality all correspond to significant differences in exposure to discrimination, experiences of depression and unmet needs for mental healthcare. Everyday discrimination is the strongest predictor of both depression and unmet need for mental healthcare, explaining more variance than any single identity variable or interaction term representing intersecting marginalised identities. There are significant limitations in using quantitative data to explain links between identities and health disparities, but intersectionality theory points us towards what may be a more appropriate focus on marginalisation and discrimination.
© 2017 John Wiley & Sons Ltd
Abstract This article uses an intersectionality lens to explore how experiences of race, gender, sexuality, class and their intersections are associated with depression and unmet need for mental healthcare in a population of 704 women and transgender/gender liminal people from Ontario, Canada. A survey collecting demographic information, information about mental health and use of mental healthcare services, and data for the Everyday Discrimination Scale and the PHQ-9 Questionnaire for Depression was completed by 704 people via Internet or pen-and-paper between June 2011 and June 2012. Bivariate and regression analyses were conducted to assess group differences in depression and discrimination experiences, and predictors of depression and unmet need for mental healthcare services. Analyses revealed that race, gender, class and sexuality all corresponded to significant differences in exposure to discrimination, experiences of depression and unmet needs for mental healthcare. Use of interaction terms to model intersecting identities and exclusion contributed to explained variance in both outcome variables. Everyday discrimination was the strongest predictor of both depression and unmet need for mental healthcare. The results suggest lower income and intersections of race with other marginalised identities are associated with more depression and unmet need for mental healthcare; however, discrimination is the factor that contributes the most to those vulnerabilities. Future research can build on intersectionality theory by foregrounding the role of structural inequities and discrimination in promoting poor mental health and barriers to healthcare. Keywords: Canada, depression, discrimination, gender liminality, healthcare access, intersectionality, mental health, transgender, women
Introduction Decades of research suggest that women are at a higher risk for depression, and are treated for it more often than men (summarised in WHO, 2000, Kessler 2003). Despite testing of biological and psychological hypotheses about why women experience more depression, consistent explanations are elusive (Kessler 2003, Slaunwhite 2015). We are even less able to explain how elevated risk for depression may be further 1139
C. C. Williams et al.
understood to be a factor in the lives of women who are racial minorities, sexual/gender minorities or living in poverty. The question of why there are differences within gender is just as compelling as the question of why there are differences between genders (Sheilds 2008). In women’s health research, there is growing understanding that treating women as a homogenous population is insufficient to find explanations for the distribution of health risks and disparities within that group. In response, intersectionality has emerged as an explanatory framework for the multiple ways in which individuals are identified, and those identifications correspond to converging axes of discrimination and exclusion that can affect health (Hankivsky & Christofferson 2008, Seng et al. 2012). Intersectionality is a theoretical framework that describes how intersecting effects of privilege and marginalisation can contribute to differences we observe in social circumstances and health outcomes. Its associations with the work of black feminists like Lorde (1984), Crenshaw (1993) and Collins (2009) align it with a politicised reading of how these differences correspond to health disparities for marginalised groups. An intersectionality analysis requires moving beyond additive models that treat marginalisation as the sum of unfortunate social designations, to consideration of the multiplicative effects of simultaneous and converging vectors of exclusion that have consequences for the material, social and physical circumstances of marginalised groups (Hankivsky & Christofferson 2008, Veenstra 2011). As an analytic tool in health research, intersectionality compels us to interpret the mutually reinforcing risk factors that undermine physical and mental health. Simultaneously, conceptualisations of double and triple jeopardy (King 1988) do not fully represent intersectional impact, as we must also find ways to account for how identities associated with privilege may insulate individuals from experiences of marginalisation. This has been explored with particular attention to the juxtaposition of whiteness with sexual and gender minority status (Riggs 2012, Parent et al. 2013). Health researchers have struggled to find effective ways to account for intersectionality empirically; at this point, demographic variables are typically used to stand in as proxies for the structural inequalities associated with them (Veenstra 2011). We believe that this may be a step towards integration of an intersectional lens, but an intersectionality analysis is defined by interpretation that calls attention to the structural inequities associated with those identities. Without this explicit critical analysis, including multiple identities in health research is little more than comparison 1140
of groups and subgroups. Structural inequities, and their manifestation in institutional processes, undermine opportunities for health by exposing marginalised people to stress, discrimination, risk and danger (Kurtz et al. 2008, Montesanti & Thurston 2015). Health research needs to shift from attributing health disparities to status as a woman, sexual minority, gender minority, racial minority or person living in poverty. The appropriate focus is how sexism, homophobia/heterosexism, cisnormativity, racism and classism converge and work interdependently to promote harm and risk for certain individuals and communities. Evaluating impacts of discrimination is vital to understanding the factors that contribute to mental health disparities for marginalised groups (Sheppard 2002, Schwartz & Meyer 2010). In research on women’s depression, intersectionality gives us the context to interpret findings pointing to similarities and differences in the experience of depression for various groups of women as indicative of the role that structural inequities play in their lives. In addition to the elevated risk that women face for depression, research demonstrates even higher risks for women with intersecting marginalised identities. For example, research shows that depression is prevalent among women living in poverty (DeCarlo Santiago et al. 2013), lesbian and bisexual women (Needham 2012, Mustanski & Liu 2013, Przedworski et al. 2015), transgender women (Nemoto et al. 2011, Rotondi Khobzi et al. 2011, Bariola et al. 2015) and racial minority women (Nemoto et al. 2011, Curry Owens & Mask Jackson 2015, Hardeman et al. 2015). In many of these studies, lives are recognised as gendered (albeit, often assumed to be cisgendered) and otherwise affected by social stratification, but the tendency to look at these experiences as doubly disadvantaged, rather than multiplicatively marginalised, points to the necessity of integrating intersectionality in research on women’s mental health. In our analysis, we represent the effects of intersecting marginalisations by developing two- and three-way interaction terms for use as predictors in regression analyses. This approach builds on the examples provided by other researchers modelling intersectionality, for example, Veenstra (2011, 2013), Hinze et al. (2012) and Seng et al. (2012). As Veenstra (2011) described, adding interaction terms as predictors in regression equations makes it possible to evaluate the amount of added explained variance that the intersections contribute. This additional explained variance can be understood to indicate the extent to which adding terms to address intersectionality explains variance in outcomes above and beyond what is explained by using single variables. Yet, we © 2017 John Wiley & Sons Ltd
Depression and discrimination in the lives of women
recognise a contradiction in this method. As Sheilds (2008) pointed out, these statistical methods assume that the constituents of interaction terms are independent. This assumed independence of factors is inconsistent with intersectionality theory’s foundational assertion that an intersectional lens recognises the interdependence of identity experiences that are multiply configured. Therefore, we moved forward with this analysis noting its limitations, but open to how statistical tools could inform more complex understandings of connections between race, gender, sexuality, class and depression. Our approach to gender in this research requires some explanation, as mental health research tends to define women as a homogenous group distinct from men. In our study, we needed to raise the question of whether treating women as a single gender, or even as two genders (cisgender, transgender) was adequate. We worked with a community advisory board comprised of service providers and service users working in agencies serving women and transgender people and they advised us to construct 10 gender categories for use in our survey, allowing respondents to check all that applied. When we looked at our data, we had a multitude of self-identifications based on varying configurations and combinations of woman, man, gender fluidity, genderqueerness, twospiritedness, trans-experienced, etc. It was necessary to move beyond a focus on the mental health of cisgender women and transgender people to acknowledge other gender identifications that are conventionally unnamed and unrecognised, but have meaning and consequence in people’s lives. The expansion of the gender binary to include trans-identification still represented only a fraction of the gender possibilities participants wished to claim. Failure to recognise those other possibilities render lives invisible and ignored (Turner 1969). Hence, in this study, we expanded the focus on women’s and transgender people’s health to encompass gender liminality. Gender liminality is often named and sanctioned outside of western cultural contexts and the difficulty encountered when describing and naming it in our research is an indication of how such experiences are voided through conventions of English language and Euro-western culture (Besmer 1996). By identifying gender liminality, and choosing liminality as a term to describe lives at, or on both sides of a socially constructed boundary between genders, we challenge the erasure of a spectrum of lived experiences that are either disregarded in the gender binary, or confined in fixed notions of trans or third genders (Besmer 1996, Besnier 1997, Booth 2011, Dutta 2015). Consequently, this paper’s starting point was the © 2017 John Wiley & Sons Ltd
experiences of women, but it was expanded to a more inclusive representation of experiences of cisgender women, transgender women, and gender liminal people facing various experiences of exclusion. We identify marginalised identities as our reference categories in the analyses. For gender, our population of interest is those gender liminal people who do not access the privileges associated with membership in a category called cisgender. For race, our population of interest is racial minority women who do not access the privileges associated with membership in the category identified as white women. For sexuality, our population of interest is sexual minority women who do not access the privileges associated with being identified as heterosexual. Finally, for class, our population of interest is those women who live in conditions of lower income and do not have access to the privileges associated with higher income. Furthermore, we are focused on exploring the implications of experiences of marginalisation and privilege that exist at the intersection of all of these identities. This analysis addresses the following questions: 1 What differences do we see among women, transgender and gender liminal people defined by axes of race, gender, sexuality and class with respect to reported depression and access to mental healthcare? 2 How do separate and intersecting identities of race, gender, sexuality and class explain variance in reported depression and access to mental healthcare?
Methods Sample This paper is based on secondary analysis of crosssectional survey data collected during a study of access to depression treatment for Lesbian, Bisexual and Transgender Women in Ontario, Canada (Ross et al. 2011), which, on the basis of advice from our community advisory board, was ultimately expanded in scope to include transgender people of any gender identity. A primary analysis focused on sexual and gender minority mental health is reported elsewhere (Steele et al. 2017). The study was reviewed and approved by the Research Ethics Board of the Centre for Addiction and Mental Health and the University of Toronto. We used targeted convenience sampling to recruit adult women, transgender and gender liminal people from across Ontario, with selection to ensure that we had representation of different intersections of socio-demographic factors including lower 1141
C. C. Williams et al.
socioeconomic status, racial minority identities, and minority sexual and gender identities. Further details of this recruitment strategy are explained in our previous paper (Steele et al. 2017). Racial classifications were based on those used in the Canadian census. Socioeconomic data were collected in the areas of household income, employment and education but this analysis uses income as a class indicator to emphasise its correspondence to experiences of material and social deprivation (MacKinnon 2013). Inclusion criteria for participation included Ontario residency, age 18 years or older and sufficiently fluent in English to understand the consent form and questionnaires. All self-identified women or transgender people were eligible to participate. Recruitment Our primary method of recruitment was through electronic posting of flyers advertising a study of ‘experiences with depression treatment in Ontario.’ Notices were circulated through women’s and LGBT online networks, health and social service agencies. To supplement electronic recruitment, study flyers were posted in agencies serving women and LGBT people, and a research co-ordinator visited community settings to promote the project. All participants had the option of participating anonymously. Individuals provided contact information received a $10 gift card via mail. Those who participated anonymously had the option of donating $10 honoraria to a registered charity. Data collection Data were collected via an Internet-based survey posted on a dedicated website. The data collection period was from June 2011 to July 2012. Questions about eligibility criteria were programmed into the survey so that only eligible participants could provide data. Participants without access to a computer could request a hard copy version of the survey, and collaborating community agencies provided space for our research co-ordinator to bring laptops for participants to use. Participants had the option of completing the survey during multiple online sessions and could save and resume the survey by entering a username and password. Data from completed surveys were automatically entered to an SPSS database. Instruments Usability and functionality of the online survey were tested extensively prior to recruitment. The survey collected socio-demographic information and asked questions about mental health and experiences 1142
accessing mental healthcare. The survey included the following standardised measures: • We selected questions from the Canadian Community Health Survey asking about mental health service use and barriers (Statistics Canada 2003). These items have been used previously to study mental health service utilisation in Canada (e.g. Vasihadis et al. 2005, Slaunwhite 2015). • The PHQ-9 Questionnaire for Depression (Spitzer et al. 1999), a brief, self-report instrument that scores each of the DSM-IV criteria for depression from 0 (not at all) to 3 (nearly every day). Total scores range from 0 to 27, with higher scores indicating more severe depression. The PHQ-9 is used widely to screen for depression in clinical and research settings and has been assessed positively for construct and criterion validity (Kroenke et al. 2001). • A modified version of the Everyday Discrimination Scale (EDS) (Williams et al. 1997), using eight of nine items asking about experiencing various types of discrimination on a scale from 1 (never) to 5 (very often). Follow-up questions were asked to identify if the respondent believed the discrimination was based on specific identities (e.g. race, sexuality, age, etc.). Total scores range from 8 to 40, with higher scores indicating more experiences of discrimination. The EDS has been used extensively for health research and demonstrated to have high levels of reliability and validity (Taylor et al. 2004, Krieger et al. 2005). For this analysis, we only used the questions assessing frequency of discrimination experiences. Although other researchers have tried to sum discrimination across identity categories as a way of quantifying the burden of oppression (e.g. Jefferson et al. 2013), we chose to treat discrimination as an integrated experience. As women embodying intersectional existences across race, gender, sexuality and class, we understood them to have unique experiences of discrimination as it is directed at those intersections. As Bowleg (2008) asserted, it may not be possible, or even useful, to untangle what specific aspects of identities are targeted by discrimination. Data analysis We recoded several variables to create mutually exclusive categories for statistical analysis. Bowleg (2008) described a conundrum created when we “transform observations into data for analysis (p. 317)”. Does what we gain in statistical capacity compensate for what we lose in specificity and © 2017 John Wiley & Sons Ltd
Depression and discrimination in the lives of women
authenticity in representing experience? We recoded our data deciding that a strategic essentialism (referencing Spivak) was necessary for carrying out our analyses, but mindful of the consequence that such action obscured the diversity and specificity that we were trying to foreground. The 14 racial categories from the Canadian census were recoded into racial minority versus white people. Ten sexual identity categories were recoded into sexual minority versus heterosexual. Eight income range categories were recoded into a binary variable of combined household income less than $39,999 CDN (lower income) versus $40,000 CDN or higher (higher income), based on review of low-income thresholds in Canada (Grant 2014). For gender, we created categories defining cisgender women and gender liminal people (which included people identifying as transgender). The sample includes no cisgender men, as they were excluded from participation in the study. All analyses were conducted using IBM SPSS 22. The group comparisons suggested by the first research question were evaluated using chi-squared tests for the following categorical outcome variables: ever diagnosed with depression; depression experienced in the last 12 months; depression diagnosed in the last 12 months; treatment for diagnosis in the last 12 months; and unmet need for emotional or mental health treatment in the last 12 months. As part of analyses, SPSS calculates a phi coefficient (a correlation between categorical values). We used ANOVA analyses to evaluate group differences in PHQ-9 scores and EDS scores. The predictive research questions were evaluated by developing stepwise linear and logistic regression models. The regression analyses were conducted in four steps with a first model to test the main effects of each identity category, a second step adding two-way interactions, a third step adding three-way interactions and a final step adding Everyday Discrimination scores. Discrimination was added as a predictor because EDS scores emerged as significant in bivariate analyses, and the inclusion of discrimination as a predictor was consistent with an intersectionality analysis. Based on guidelines presented in Green (1991), our sample size was adequate to conduct these analyses. We explored the data to ensure it met assumptions for each statistical test conducted and analyses were based on cases with no missing values for variables included. The threshold significance level for all analyses was a = 0.05.
Results We collected complete data sets from 704 participants; socio-demographic information for the study © 2017 John Wiley & Sons Ltd
participants is presented in Table 1. The information provided about education and employment reveals that the participants were well educated, but had a wide range of employment situations, requiring multiple categories to describe them. The high education level may be an indicator that most of the study population had access to computers/the Internet and skills to complete the survey. Summary statistics for key variables in the study are presented in Table 2 with significant differences between subgroups in bold type. ANOVA tests revealed that depression scores were significantly higher for lower income than higher income women (F(1, 607) = 25.945, P < 0.0001) and gender liminal people than cisgender women (F (1, 617) = 10.788, P = 0.001). We also found significant differences in the experiences of everyday discrimination between racial minority and white women (F(1, 594) = 26.635, P < 0.0001), sexual minority women and heterosexual women (F(1, 594) = 16.911, P < 0.0001), lower income women and higher income women (F(1, 586) = 26.956, P < 0.0001), and gender liminal people and cisgender women (F(1, 594) = 26.395, P < 0.0001). Chi-squared Table 1 Summary of demographic variables in the sample (n = 704) Variable
Categories
n (% of sample)
Race
Racial minority White Gender liminal person Cisgender Woman Sexual minority Heterosexual
