Roots and STEMS? Examining field of study choices among northern and rural youth in Canada

ABSTRACT

Despite several decades of postsecondary expansion, new research finds youth from northern and rural areas in Canada still experience difficulties making the transition to postsecondary education, and those who do attend take longer to do so. Proximity, we argue, may also have a considerable impact on one’s field selection, as many of Canada’s larger universities and colleges, who offer considerably more program and degree options, tend to be concentrated in large, urban centers, and in the southern regions of Canada’s provinces. This study draws on Cycles 1–4 of Statistics Canada’s Youth in Transition Survey – Cohort A to examine regional inequalities in accessing Science, Technology, Engineering and Mathematics (STEM)-related fields at both the university and non-university levels. Indeed, our findings suggest that location of residence does impact field choices, as students from northern and rural areas were less likely to enter STEM as well as non-STEM, university programs.

KEYWORDS:

• STEM (Science, Technology, Engineering and Mathematics)
• field of study
• higher education
• proximity
• rural
• parental socio-economic status (SES)

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Canada’s PSE system is predominately publicly-governed and comprised of trades schools, colleges, and universities. The college system in Canada is analogous to the community college system in the United States, where programs are generally one to two years in length, and diplomas are conferred upon completion. Canada’s university system lacks an internationally-recognized set of elite institutions akin to Harvard or Oxford, as most Canadian universities are comprehensive, and provide a range of programs to diverse populations. Canada’s PSE market is also locally-driven in many respects especially in comparison to the US system, as few students travel great distances for their initial PSE programs (Davies and Hammack 2005) and departments of higher education are housed at the provincial level and not governed nationally.

2 The YITS data unfortunately did not survey youth residing in Canada’s three territories. As a result, our northern analyses include only those who reside in Canada’s provincial North, and any northern inequalities accessing certain fields may be underestimated.

3 Youth who were homeschooled, special needs schools, and schools on First Nation reserves were also excluded, which represents approximately 4 percent of 15-year-olds in Canada.

4 Given wave attrition and our focus on a small sub-population of northern and rural youth, we necessarily draw on students’ field of study measured up to age 21 (Cycle 4). In supplemental analyses, we ran similar models using field of study of highest level of education taken across all programs and institutions as of December 2007 (Cycle 5), or when the youth were age 23. These models revealed very similar findings to those currently presented with the exception that fewer significant differences across population subgroups were observed – a direct result of the lower sample size available for estimation by Cycle 5. This additional analysis also revealed that while there may be some switching of degree majors over the course of college and university careers, there is a high correlation (0.89) between fields of study at age 21 and field of study of the highest level of education taken across all programs and institutions at age 23. Moreover, in the case of respondents who started PSE in university, well over half remained in the same type of institution and program by age 23 in Cycle 5: 66% in the case of STEM programs in university, and 90% for non-STEM programs in university. This discrepancy between STEM and non-STEM is likely due to the much larger number of CIP codes belonging to non-STEM programs than STEM programs. The story at the non-university level was more mixed, while the greatest proportion remained in their particular program type, the numbers are not as large as observed for those in university programs.

5 Due to small sample sizes, we combine colleges with trade and vocational schools captured by the non-university category.

6 In some analyses, we included ‘agriculture, natural resources, and conservation’ into STEM; however, the results did not change, so we opted to keep them in the non-STEM group.

7 Recently, Statistics Canada developed the term BHASE, which can be used in place of non-STEM. (see Statistics Canada 2017). This variant of the definition encompasses business, humanities, health, arts, social science, education, legal studies, trades, services, natural resources and conservation. This definition and terminology was not used in this current paper as the groupings do not perfectly align with those of the 2000 CIP codes which were used for defining program of study in YITS (see Statistics Canada 2005).

8 McNiven and Puderer (2000) describe in detail 16 different indicators that were employed to define north and south boundaries including the OECD rural north definition, Revenue Canada’s northern and intermediate income tax zones, resourced areas and native north, living cost differential, boundaries of the boreal forest, growing degree-days, heating degree-days, all-season road and railway transportation networks, accessibility index, discontinuous permafrost, summer concentration of thermal efficiency, agroclimatic resource index, agriculture ecumene, and population ecumene.

9 Unfortunately, the YITS does not survey individuals who are living in the territories, and as such, we believe that any northern inequalities uncovered in this paper may represent an underestimate of the overall level of inequalities faced by residents of northern Canada.

10 To examine the effects of particular explanatory variables x_k, all variables except for x_k were held constant at their sample means or proportions.

11 While the coefficients produced in the multinomial logistic regressions take on values of zero for the base outcome, the marginal effects are based on the change in the predicted probabilities across values of covariates and are calculated post-estimation for each of the five outcomes.

12 All estimates in Table 1 are weighted by the YITS sample weights.

13 It is important to note the provincial differences in educational systems, which have implications for our findings here. In terms of the non-university STEM programs as well as non-university non-STEM programs, Quebec shows the highest proportions at 0.179 and 0.520 respectively. Compared to other provinces, youth in Quebec typically finish high school after grade 11 and begin their postsecondary studies one year earlier than the other provinces in Canada, and they enter university one or two years later than those in the other provinces. In Quebec, youth typically enter the CEGEP (College d’enseignment general et professional) system after high school. For those heading towards university, they normally complete a prerequisite program which takes two to three years. Alternatively, youth could choose a college program, which typically takes three years. As such, many youth from Quebec may still be completing their CEGEP programs (classified as non-university in our analyses), and contribute to higher relative proportions selecting non-university programs compared to the other provinces.

14 An alternative version of Table 2 with the odds ratios presented is available upon request.

15 The estimates shown in Figure 1 are the predicted probabilities and 95 percent confidence intervals from Model 1, holding all other variables constant at their sample means and/or proportions.

16 The estimates shown in Figure 2 are the predicted probabilities and 95 percent confidence intervals from Model 7, holding all other variables constant at their sample means and/or proportions.

17 We also explored the possibility that the relationships among youth from various regions and their educational outcomes might vary by gender. The results from the multiple parameter Wald tests revealed that the interaction was not statistically significant, so this additional model is not presented here.

Additional information

Funding

This research was undertaken, in part, thanks to funding from the Canada Research Chairs program and the Social Sciences and Humanities Research Council of Canada [grant number # 430-2017-0661].