Social mobility and mental health in Canada

Abstract

Objectives

To investigate whether intergenerational mobility in education and income are associated with levels of psychological distress in Canada, a context in which rates of intergenerational mobility are higher than those of the United States but lower than those of Nordic countries.

Methods

The data came from the Longitudinal and International Study of Adults (LISA) linked to tax records from the Canada Revenue Agency (N = 4100). Diagonal reference models were used to investigate whether educational mobility and income mobility were associated with levels of psychological distress in adulthood as assessed by the Kessler (K-10) scale. The models controlled for sociodemographic characteristics and were stratified by gender.

Results

Although we did not find that mobility in general was associated with greater levels of psychological distress, we found that downward educational mobility in particular corresponded to higher levels of psychological distress (b = 0.15 with 95% CI = 0.00, 0.31) among men.

Conclusion

Overall, we found no strong evidence that social mobility in general is impactful for levels of psychological distress, but downward educational mobility in particular may have negative consequences for the mental health of men. In addition, a notable gradient between income and psychological distress in adulthood was observed for both women and men.

Introduction

Since Sorokin (1927) first proposed that mobility can be a psychologically taxing and socially disruptive experience, numerous studies have investigated how changes in social status across and within generations can have consequences for mental health, physical health, and life satisfaction. In recent years, methodological advances have enabled researchers to establish the effects of mobility over and above the effects of individuals’ positions of origin and destination, thereby disentangling the effects of movement between different social positions from the effects of being in these positions (Bulczak et al., 2021; Gugushvili et al., 2019; Houle & Martin, 2011; Jonsson et al., 2017; Schuck & Steiber, 2018; Steiber, 2019; van der Waal et al., 2017). However, previous studies have typically relied on few indicators of mobility (i.e., educational or occupational mobility) and few mental health outcomes (i.e., depressive symptoms) and have focused mostly on Europe or the United States. In this article, we investigate whether intergenerational mobility in education and income are associated with mental health, as assessed by the Kessler (K-10) scale of psychological distress, in the understudied Canadian context. In Canada, education and income mobility rates have typically been higher than those of the USA but lower than those of Nordic countries (Corak, 2013; Narayan et al., 2018). However, recent research suggests trends toward increasing income inequality and decreasing income mobility (Connolly et al., 2021) which may arguably make mobility experiences more impactful for the mental health of Canadians.

The mental health consequences of social mobility have been theorized in multiple ways. Inspired by Sorokin’s account, the dissociative thesis posits that any type of social mobility—but especially upward mobility—leads to poor mental health because the experience of mobility leaves individuals torn between social worlds with different norms and lifestyles and disconnected from previously strong ties in their community of origin. They are left with feelings of guilt due to a perceived betrayal of their roots and face difficulties forging a coherent sense of self (Ellis & Lane, 1967; Friedman, 2016; Lee & Kramer, 2012). Others have argued that the experience of upward mobility can potentially generate a greater sense of control and confidence in overcoming difficulties as well as opportunities for creative adaptations and feelings of gratitude for being welcomed into a higher status social environment (Gugushvili et al., 2019; Reay et al., 2009). According to the latter perspective, upward mobility tends to lead to better mental health. Downward mobility, by contrast, can produce a sense of powerlessness, anger, and frustration from having to abandon a previously comfortable and secure way of life (Newman, 1999).

Conventional regression methods can provide mobility estimates by (a) controlling for position of origin but not position of destination, (b) controlling for position of destination but not position of origin, or (c) controlling for position of origin and destination. All of these are now deemed suboptimal strategies (Schuck & Steiber, 2018). In the first strategy, mobility effects are confounded by the effect of position of destination; in the second strategy, mobility effects are confounded by the effect of position of origin; and the third strategy is unsuitable because mobility is linearly dependent on both origin and destination (Schuck & Steiber, 2018). An approach that faces the same problem is the comparison of different mobile groups to an immobile reference category (van der Waal et al., 2017). In light of these limitations, diagonal reference models have become widely accepted as the gold standard for mobility research (Zang et al., 2023).

In a sample of young Dutch adults, van der Waal and colleagues (2017) showed that mobility variables were significant predictors of being overweight or obese in logistic regression models but not in diagonal reference models. Similarly, Schuck and Steiber (2018) applied different methods to compare how educational mobility is associated with life satisfaction in European countries. Using linear regression models, they found mobility effects on life satisfaction in all groups of countries; using diagonal reference models, they found mobility effects only in continental countries where upward mobility corresponded to higher life satisfaction and downward mobility corresponded to lower life satisfaction. Consistent with the latter findings, a study using diagonal reference models applied to pooled data from 27 European countries found that upward educational mobility was associated with fewer depressive symptoms while downward educational mobility was associated with more depressive symptoms, as measured by the CES-D scale, with stronger effects for long-range mobility (Gugushvili et al., 2019). Interestingly, no mobility effects were found among women in gender-stratified analyses. In a recent study from the USA, no effects were found for education, income, and occupation mobility on depressive symptoms (Bulczak et al., 2021). Using data from the Wisconsin Longitudinal Study, Houle and Martin (2011) found that sons of farmers who were mobile into nonmanual occupations reported lower levels of depressive symptoms at age 52 compared to non-mobile men or other upwardly mobile men. In two Chinese studies, neither upward nor downward intergenerational occupational mobility variables were associated with levels of happiness when using diagonal reference models, although effects on well-being were found for intragenerational mobility (Zang & De Graaf, 2016; Zhao et al., 2017). In Canada, one study using diagonal reference models found links between intergenerational mobility and self-rated health (Veenstra & Vanzella-Yang, 2021), with downward educational and income mobility corresponding to poorer health among women and upward income mobility corresponding to poor health among men, but the effects of mobility on mental health have yet to be investigated.

Though it is now clear that mobility effects are not as strong as previously estimated using conventional regression methods, the links between social mobility and subjective well-being outcomes (particularly those related to mental health) are still not well understood. We contribute to the growing body of research on the consequences of social mobility for mental health using diagonal reference models. We investigate associations between intergenerational education and income mobility and psychological distress in Canada, testing the following mobility hypotheses:

• Hypothesis 1: Mobility of any kind—but especially upward mobility—is associated with higher levels of psychological distress (dissociative thesis)

• Hypothesis 2: Upward mobility is associated with lower levels of psychological distress (rising from rags thesis)

• Hypothesis 3: Downward mobility is associated with higher levels of psychological distress (falling from grace thesis)

Methods

Data

We draw from the second wave of the Longitudinal and International Study of Adults (LISA) collected by Statistics Canada in 2014 and linked to T1 Family Files from Canada Revenue Agency. The LISA provides information on labour market, education and training, skills, health, and family experiences (Statistics Canada, 2016). In the second wave, the LISA also provides information on mental health. The target population of the LISA was all residents of Canada’s ten provinces aged 15 and older excluding individuals living on reserves and other Aboriginal settlements, official representatives of foreign countries living in Canada and their families, members of religious and other communal colonies, members of the Canadian Armed Forces stationed outside of Canada, persons living full-time in institutions, and persons living in other collective dwellings. In total, 11,458 of 15,907 (72.0%) randomly selected households participated in the first wave of the study. Attempts were made to survey all members of each participating household who were aged 15 and older, with a person-level response rate of 89.0%. This led to a final survey sample of 23,926 respondents in 2012 (wave 1). In 2014 (wave 2), there were 16,895 respondents who remained in the study. We restricted our analyses to the approximately 15,400 respondents who were aged 25 or older in 2014 to ensure that most study participants had completed their educational training. At the time we began this study, only waves 1 to 3 were available, and mental health items were assessed from wave 2 onward. Due to attrition in the later waves of the LISA, we chose to work with the earliest wave containing mental health items (i.e., wave 2).

Statistics Canada used Social Insurance Numbers to link the survey respondents to their T1 Family File (T1FF) income tax data. This databank of Canadian tax filers grouped into families includes income data, both personal and family, before and after taxation. A T1FF family is composed of a married couple with or without children of either or both spouses, a common-law couple with or without children of either or both partners, a lone parent living with at least one child, or a person living alone. We further restricted the analyses to survey respondents who were not designated as a “filing child” in the 2013 tax year, i.e., they were not an adult child in a parent’s family. We determined the year in which each respondent first filed a T1 income tax form with the CRA and further restricted the sample to respondents who were designated as a “filing child” at the time of first filing, i.e., they were a member of a T1FF family containing at least one of their parents. This produced a sample of n = 4800. Most of these respondents (86.28%) were between the ages of 15 and 21 at the time of first filing; we further restricted the sample to this group, producing a final sample composed of approximately 2100 women and 2000 men with ages ranging from 25 to 53. We then calculated variables representing personal family income in 2013 and family income at the time of first filing on the part of the LISA survey respondent.

Variables

The dependent variable in the analyses is the Kessler scale of psychological distress (Kessler et al., 2003). The Kessler scale was created from respondents’ responses to questions asking how often during the last 30 days they felt (i) tired out for no good reason, (ii) nervous, (iii) so nervous that nothing could calm them down, (iv) hopeless, (v) restless or fidgety, (vi) so restless they could not sit still, (vii) depressed, (viii) that everything was an effort, (ix) that nothing could cheer them up, and (x) worthless. Response categories for each item were 1 = none of the time, 2 = a little of the time, 3 = some of the time, 4 = most of the time, and 5 = all of the time. The Kessler scale variable ranged from 10 to 50. We logged this variable to address right skewness using the Stata command lnskew0. Several studies have assessed the psychometric properties of the K-10 scale, finding high levels of validity and reliability (Bougie et al., 2016; Fassaert et al., 2009; Sampasa-Kanyinga et al., 2018).

The predictors of interest are educational mobility and income mobility. Respondents provided information on the educational attainment of their parents as well as their own. Education was coded in three categories: high school diploma or less, college certificate or diploma, and bachelor’s degree from university or higher. Educational mobility was assessed by comparing the highest credential obtained by the respondent (the column variable in the mobility table, i.e., the position of destination) to the highest credential obtained by a parent (the row variable in the mobility table, i.e., the position of origin). Income mobility was assessed by comparing the respondent’s family income tertile in adulthood (position of destination) to parental family income tertile upon first filing between the ages of 15 and 21 (position of origin). One-step and two-step mobility dummy variables were created to assess short- and long-range mobility in education and income. We controlled for age, square of age, marital status (distinguishing between married or common-law and others), and immigrant status (distinguishing between individuals born in Canada and individuals who immigrated to Canada) across all models.

Analytic strategy

We used diagonal reference models (DRMs) to investigate the mental health effects of mobility per se as well as the relative importance of positions of origin and destination. The DRMs compared the mental health of mobile individuals to the mean levels of mental health of non-mobile members in the positions of origin and destination. The outcome variable is specified as “the weighted sum of the estimated mean scores in the non-mobile origin group and the non-mobile destination group” (Schuck & Steiber, 2018, p. 1246). The baseline DRM with the inclusion of covariates can be represented as:

$$Y_{\mathit{ijk}}=w\times {\mu }_{\mathit{ii}}+(1-w)\times {\mu }_{\mathit{jj}}+\sum {\beta x}_{\mathit{ijkl}}+e_{\mathit{ijk}}$$

Subscript i and j respectively refer to positions of origin and destination. Y_ijk represents the value of the dependent variable in cell ij of the mobility table which has k observations. µ_ii is the estimate of Y in the diagonal cell in the row denoting the position of origin while µ_jj is the estimate of Y for the diagonal cell in the column denoting the position of destination. The w parameter, also referred to as “weight,” estimates the strength of the effect of the position of origin relative to the position of destination and usually falls between 0 and 1. $\beta x$_ijkl refers to the l covariates in the model. We executed the analyses using the Stata command drm with a linear specification (Kaiser, 2018). We applied the responding person weights provided by Statistics Canada to all models. This study was approved by the University of British Columbia’s Behavioural Ethics Board.

Results

Among women, we find no evidence that intergenerational mobility in education or income is associated with psychological distress (Table 1). Consistent with the well-known SES-health gradient, the diagonal intercepts show that women who were stably low in educational attainment (i.e., low parental education and low personal education) and stably low in family income (i.e., their parents were in the bottom family income tertile and they themselves are in the bottom family income tertile) had greater levels of psychological distress relative to the overall average. In parallel, women who were stably high in family income had lower levels of distress relative to the overall average. The w parameter falls closer to 0 than to 1 in all models for women, indicating that social position of origin matters less than social position of destination for psychological distress.

Table 1.

Diagonal reference models with linear link on logged Kessler score (women)

Educational mobility	Model 1	Model 2	Model 3
Diagonal intercepts
µ11: low	0.11 (0.00, 0.21)*	0.10 (0.00, 0.21)*	0.10 (− 0.00, 0.21)†
µ22: medium	− 0.06 (− 0.15, 0.04)	− 0.07 (− 0.16, 0.03)	− 0.07 (− 0.18, 0.04)
µ33: high	− 0.05 (− 0.13, 0.03)	− 0.04 (− 0.12, 0.05)	− 0.03 (− 0.13, 0.06)
w: weight of origin	0.20	0.31	0.34
Upward educational mobility		0.02 (− 0.10, 0.14)
Downward educational mobility		0.10 (− 0.07, 0.27)
Upward educational mobility—one step			0.03 (− 0.10, 0.15)
Upward educational mobility—two steps			0.01 (− 0.19, 0.21)
Downward educational mobility—one step			0.10 (− 0.08, 0.28)
Downward educational mobility—two steps			0.13 (− 0.35, 0.61)
AIC	8,145,496	8,141,271	8,141,149
BIC	8,145,547	8,141,333	8,141,223
Income mobility	Model 1	Model 2	Model 3
Diagonal intercepts
µ11: low	0.15 (0.06, 0.25)**	0.15 (0.06, 0.25)**	0.15 (0.04, 0.26)**
µ22: medium	− 0.06 (− 0.13, 0.01)†	− 0.06 (− 0.14, 0.01)	− 0.05 (− 0.16, 0.06)
µ33: high	− 0.09 (− 0.17, − 0.01)*	− 0.09 (− 0.18, − 0.01)*	− 0.10 (− 0.19, − 0.01)*
w: weight of origin	− 0.05	− 0.02	− 0.22
Upward income mobility		− 0.04 (− 0.21, 0.12)
Downward income mobility		− 0.03 (− 0.18, 0.13)
Upward income mobility—one step			− 0.03 (− 0.27, 0.21)
Upward income mobility—two steps			0.04 (− 0.49, 0.56)
Downward income mobility—one step			− 0.06 (− 0.33, 0.22)
Downward income mobility—two steps			− 0.07 (− 0.51, 0.38)
AIC	8,105,372	8,104,152	8,103,751
BIC	8,105,372	8,104,215	8,103,824

Each model adjusts for age in years, age squared, marital status, and immigrant status. The reference group for mobility variables is the immobile group. Brackets contain 95% confidence intervals for the estimates

†p < 0.10; *p < 0.05; **p < 0.01

Among men, we find evidence that downward education mobility is associated with greater levels of psychological distress, consistent with the “falling from grace” thesis, as shown in Table 2. In model 2, downward educational mobility corresponded to higher psychological distress scores (b = 0.15 with 95% CI = 0.00, 0.31). In model 3, one-step downward educational mobility also corresponded to greater levels of psychological distress (b = 0.13 with 95% CI = − 0.02, 0.28). However, these coefficients were only significant at the 10% level. Two-step downward mobility in education was more strongly, though not significantly, associated with greater levels of psychological distress (b = 0.24 with 95% CI = − 0.16, 0.63). We find no compelling evidence that income mobility impacts psychological distress among men (like the findings for income mobility among women). The diagonal intercepts show, as expected, that men who were stably low in family income had greater levels of psychological distress and men who were stably high in family income had lower levels of psychological distress in comparison to the overall average. Interestingly, men who were stably low or stably high in educational attainment did not have significantly different levels of distress relative to the overall average. In the models for men, the w parameter suggests that parental education matters more than personal education for mental health. The AIC and BIC statistics are lower in models 2 and 3 than in model 1 across all analyses, indicating that the inclusion of the mobility variables in the model improved the model fit.

Table 2.

Diagonal reference models with linear link on logged Kessler score (men)

Educational mobility	Model 1	Model 2	Model 3
Diagonal intercepts
µ11: low	− 0.02 (− 0.05, 0.02)	0.06 (− 0.05, 0.17)	0.06 (− 0.05, 0.17)
µ22: medium	− 0.03 (− 0.22, 0.16)	− 0.10 (− 0.21, 0.01)†	− 0.09 (− 0.21, 0.24)
µ33: high	− 0.05 (− 0.14, 0.24)	0.04 (− 0.06, 0.14)	− 0.03 (− 0.07, 0.13)
w: weight of origin	2.11	0.87	0.91
Upward educational mobility		− 0.05 (− 0.17, 0.09)
Downward educational mobility		0.15 (0.00, 0.31)†
Upward educational mobility—one step			− 0.04 (− 0.19, 0.10)
Upward educational mobility—two steps			− 0.07 (− 0.29, 0.16)
Downward educational mobility—one step			0.13 (− 0.02, 0.28)†
Downward educational mobility—two steps			0.24 (− 0.16, 0.63)
AIC	9,764,900	9,752,878	9,751,061
BIC	9,764,950	9,752,939	9,751,134
Income mobility	Model 1	Model 2	Model 3
Diagonal intercepts
µ11: low	0.14 (0.04, 0.23)**	0.14 (0.05, 0.23)**	0.11 (0.0, 0.22)**
µ22: medium	0.02 (− 0.05, 0.09)	0.02(− 0.06, 0.09)	0.07 (− 0.05, 0.18)
µ33: high	− 0.16 (− 0.25, − 0.06)**	− 0.15 (− 0.25, − 0.05)**	− 0.18 (− 0.28, − 0.08)**
w: weight of origin	− 0.12	− 0.20	0.44
Upward income mobility		0.05 (− 0.21, 0.30)
Downward income mobility		− 0.01 (− 0.27, 0.30)
Upward income mobility—one step			− 0.07 (− 0.26, 0.12)
Upward income mobility—two steps			− 0.07 (− 0.41, 0.26)
Downward income mobility—one step			0.07 (− 0.16, 0.31)
Downward income mobility—two steps			0.25 (− 0.08, 0.59)
AIC	9,724,148	9,723,095	9,719,563
BIC	9,724,198	8,723,156	9,719,636

Each model adjusts for age in years, age squared, marital status, and immigrant status. The reference group for mobility variables is the immobile group. Brackets contain 95% confidence intervals for the estimates

†p < 0.10; *p < 0.05; **p < 0.01

Discussion

Despite the popularity of Sorokin’s dissociative thesis which posits that social mobility—and especially upward mobility—leads to poor mental health, no large-scale studies have offered compelling empirical support for the thesis. Indeed, we did not identify significant associations between upward mobility and psychological distress, consistent with previous studies using similar methodological strategies. This body of literature therefore continues to offer little evidence of a “cleft habitus,” the sociologist Pierre Bourdieu’s term to designate a fragmented and incoherent sense of self which results from the experience of navigating social environments with different norms, values, and lifestyles. However, this does not mean that upward mobility is never detrimental for mental health and well-being. It is possible that extreme cases of upward mobility, which tend to be quite rare and therefore harder to capture in statistical models, may be particularly impactful. Individuals who experience long-range mobility within a short period of time may be particularly vulnerable to experiencing the psychological consequences of clashing social worlds (Friedman, 2016). Furthermore, given the structural expansion of the Canadian educational system in previous decades (Chow & Guppy, 2021), with ever more people attaining higher educational credentials, it is possible that the models used here do not meaningfully capture upward educational mobility, even when considering two-step mobility. Given the historical trend toward greater participation in higher education, one could make a case that upward educational mobility should be considered the norm rather than the exception, especially for women who had a higher rate of increase in university attendance in the second half of the past century (Chow & Guppy, 2021). Additionally, individuals who undergo very extreme forms of mobility may be more compelled to share their stories, which may be why the dissociative thesis remains popular in qualitative studies but mostly unsupported in quantitative studies that use diagonal reference models. Future studies should aim to investigate upward mobility with a focus on the more extreme cases of upward mobility and, wherever possible, distinguish between absolute and relative educational mobility (Gugushvili et al., 2019). Alternatively, it may be that the detrimental effects of upward mobility are offset by the benefits that upward mobility brings, such as improved material conditions and greater sense of control over one’s life.

We found weak evidence that downward educational mobility corresponded to greater levels of psychological distress as the “falling from grace” thesis would predict, but only among men. This finding is consistent with the study from Gugushvili and colleagues (2019) who found that downward educational mobility in Europe was associated with more depressive symptoms among men but not among women. However, our findings differ from that study in that we do not find significant effects for two-step (i.e., long-range) mobility. These higher levels of psychological distress may result from the frustration associated with not being able to take advantage of an expanding educational system or from failing to meet parental and societal expectations (Culatta & Clay-Warner, 2021). It is unclear, however, why this was observed only for short-range and not long-range movements between social positions.

Strengths and limitations

The income data used in this paper are uniquely valid and precise given the linkage of survey data with T1FF Family Files from the Canada Revenue Agency. However, the linkage of parental income data to respondents required respondents to have been living with at least one of their parents upon first filing. In addition, it was not possible to determine the age of the parents when respondents first filed. In an analysis of possible sample selection issues with linked LISA data, other researchers have found that respondents with a parental link were relatively likely to be Canadian-born, male, employed, living in a single detached home, and living in a rural area (Simard-Duplain & St-Denis, 2020). Linked respondents were also more likely to be living with two parents at age 15 and to have parents with higher levels of education and were less likely to be a visible minority (Simard-Duplain & St-Denis, 2020). Our sample is, therefore, not fully representative of the Canadian population. Another important limitation is that the LISA survey design excludes individuals living on reserves and other Indigenous settlements. First Nations communities have endured hundreds of years of colonial oppression and continue to experience chronic disadvantage and marginalization in Canadian society (Palmater, 2012). The income and earnings gap between Indigenous people and majority-group workers is significant and is largest for Indigenous people living on reserves (Pendakur & Pendakur, 2011). Thus, the absence of these individuals in the LISA survey means that our analysis may fail to capture impactful mobility experiences.

Regarding methodological limitations, DRMs cannot offer insights into how more detailed trajectories of socioeconomic hardship (reflective of the dynamic nature of resources) are associated with mental health and mental health trajectories over time. We discussed the limitation of the educational mobility variable above and, though the income tertile approach arguably does better in assessing relative mobility (rather than absolute or structural mobility), it still oversimplifies the potential trajectories of hardship and financial well-being that individuals may experience throughout the life course. In an alternative approach, Willson and Shuey (2016) use Repeated Measure Latent Class Analysis to identify trajectories in economic hardship from early life to early adulthood and their associations with health risk trajectories from early adulthood into middle age. This approach illuminates a heterogeneity of pathways, including mobility out of economic hardship in childhood and mobility out of economic hardship in adulthood, something that DRMs cannot do. They found that moving out of hardship early in life corresponded to a lower likelihood of being in a trajectory of high health risk, while moving out of hardship later in life did not diminish these high health risks (Willson & Shuey, 2016). When possible, approaches to studying the health consequences of mobility should, they argue, “include the stability of resources, the timing of mobility, and the degree and direction of change” (p. 419).

Conclusion

In summary, we found no support for the dissociative thesis (hypothesis 1) or the rising from rags thesis (hypothesis 2). We found some support for the falling from grace thesis (hypothesis 3) in our analysis of educational mobility among men. Overall, however, we found little of note regarding intergenerational social mobility and psychological distress. Our analysis makes an important contribution for three reasons. First, while most previous work has only investigated one form of mobility (e.g., educational or occupational mobility), we examined how two different kinds of mobility (educational and income mobility) may impact the mental health of adults. Income mobility has rarely been assessed in previous research, and we were able to execute this analysis using exceptionally valid and precise income data. Second, while most previous studies use depressive symptoms and general health/well-being as the dependent variable, ours is the first study to assess the effects of mobility using the Kessler scale of psychological distress, a validated instrument for mental health assessment. Finally, most existing research on social mobility and mental health uses data from Europe or the USA, and we offer the first analysis of this topic in the understudied Canadian context.

Contributions to knowledge

What does this study add to existing knowledge?

• This study is the first to explore whether intergenerational mobility is associated with mental health in the Canadian context.

• This study uses uncommonly valid and precise income data for respondents and their parents.

What are the key implications for public health interventions, practice, or policy?

• Though social mobility appears not to be impactful for mental health, individuals who are stably low in income experience greater levels of psychological distress.

• It is imperative to provide mental health resources and support for individuals who are socioeconomically disadvantaged. Given the links between psychological distress and other health outcomes, such support could additionally prevent the onset of other health problems and reduce costs to individuals and society.

Author contributions

Both authors contributed to the study conception and design. Vanzella-Yang conducted statistical analyses and wrote the original draft. Both authors read and approved the final manuscript.

Funding

This research was supported by a Joseph-Armand Bombardier Canada Graduate Scholarship awarded to AVY and an Insight Grant awarded to GV, both from the Social Sciences and Humanities Research Council of Canada (SSHRC). Currently, AVY is supported by a SSHRC Postdoctoral Fellowship (#756–2022-0048).

Availability of data and material

The data came from the Longitudinal and International Study of Adults (LISA) linked to T1 Family Files from the Canada Revenue Agency. Data are available in the Research Data Centres (RDC) in 33 universities across Canada for researchers who meet the criteria for access to confidential data. Statistical analyses were conducted at the UBC-RDC.

Code availability

Stata syntax files require vetting from the RDC staff.

Declarations

Ethics approval

The project was approved by The University of British Columbia Behavioural Ethics Board (certificate number H18-02461).

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Conflict of interest

The authors declare no competing interests.