6 Measurement and correlates of in situ growth

View the most recent version.

In the preceding analysis, in situ growth emerges as an important driver of degree-holder growth, particularly among the young. The intent of this chapter is to evaluate the likelihood of degree attainment relative to the place of residence prior to university. That is, we are interested in whether the relatively low degree shares in small urban and rural areas are due to lower rates of degree attainment among those that spent their formative years in these areas.

To address this issue, we focus on new degree holders aged from 20 to 24 in 2001. We make the implicit assumption that it is the ability of urban and rural areas to create and/or retain these young degree holders that plays a major role in determining their overall degree holder share in the long-run.

To locate degree holders in their place of residence prior to university, we 'relocate' them back to their stated place of residence in 1996. In effect, this means that we are imposing the counterfactual that no persons that obtained a degree left the place where they spent their formative years.

Table 5 illustrates degree attainment across the urban-rural hierarchy, with various age restrictions corresponding to the average age of entry into university by region. We introduce age restrictions in order to better identify individuals that spent their formative years in the location they identified as their residence five years previous. The 'no restriction' case includes all individuals aged from 15 to 19 in 1996 (from 20 to 24 at the time of the census), with their age in 1996 corresponding to a time when they would most likely still be residing in a parental/guardian home prior to university. The group, however, includes persons aged 18 and 19 in 1996. Many of these would be in university, particularly in Atlantic and Western Canada, and may have cited the town where they went to university as where they resided five years previous. The 'low restriction' scenario includes those aged from 16 to 18 in 1996 living in Western Canada and the Atlantic provinces, and aged from 17 to 19 (in 1996) in Ontario and Quebec. Further constraints on age are imposed in the 'high restriction' case, corresponding to entry ages from 16 to 17 in Western and Atlantic Canada and from 17 to 18 in Ontario and Quebec. In short, these age restrictions account for differences in the average starting age for university across provinces. As such, they increase the likelihood that we are capturing the location where degree holders spent their formative years, rather than where they attended university.

Regardless of age restrictions, a clear gradient in degree attainment is observed across the urban- rural hierarchy, with the greatest rate of degree attainment in large urban centres. Nearly similar proportions of individuals resident in medium urban centres attain a degree. Residents of large and medium urban centres are, on average, at least two times more likely to obtain a degree than residents of rural areas.

The results presented in Table 5 illustrate the effect oflocation on the rate of degree attainment. However, they are limited in the sense that other factors that influence degree attainment are not controlled for. In particular, individual characteristics like age, gender and immigrant status may influence the choice of where and when to obtain a degree. Province of residence may also influence this choice, for example, because of differences in education policies and opportunities. These are, of course, not an exhaustive set of correlates, but they do allow us to 'narrow the field' of factors that might explain why rates of degree attainment vary by 'location' in the urban-rural hierarchy.

Table 5
Percentage share of the late-teen population resident in 1996 that obtained degrees by 2001, by urban-rural class

The effect of these correlates on the choice to obtain a degree is estimated through a binary choice (logistic) model (Table 6). Paralleling the tabulations presented in Table 5, three separate models are presented, with Model 1 being the least restrictive in terms of age effects, and Model 3 the most restrictive. Reported are the marginal effects on the probability of obtaining a degree. The reference person is a non-immigrant, non-Aboriginal female living in rural Ontario at about the age of 22. In Model 1 the predicted probability that this person has obtained a degree is 0.072 or 7.2%. The marginal effects of all other variables are made with respect to this predicted probability, with binary variables treated as either 0 or 1.

Turning first to differences across the urban-rural hierarchy, a strong gradient appears that echoes the results obtained earlier. For Model 1, the additional probability of obtaining a degree for residents of large urban centres is, on average, 6.7 percentages points higher than for their rural counterparts. Moving down the hierarchy, the propensity to obtain a degree decreases monotonically, with residents of medium and small urban centres approximately 4.3 and 1.6 percentage points, respectively, more likely to obtain a degree than rural residents. The effect of city size remains qualitatively the same across all three models. Hence, regardless of how we restrict the ages of degree holders, our results remain the same.

Of particular interest is the difference in the probability of obtaining a degree between large urban and medium urban areas. ⁶ While residents of large and medium urban centres essentially have equal access to universities, the probability of obtaining a degree is higher in large centres than in medium centres. This suggests that the relatively high rates of degree attainment in large urban areas is not just due to ready access to postsecondary education, but is reflective of other characteristics of large urban centres.

Table 6
Logistic model of the propensity to earn a degree

Supporting the broader literature, other effects that modify degree attainment are noted as well. For instance, males and Aboriginals are less likely to obtain a degree. The coefficient associated with males is somewhat surprising, but reflects increased university participation among females over time. Conversely, and not surprisingly, age has a positive effect. There is no significant difference between immigrants and the Canadian born. We do not, however, control for the effect ofimmigrant parents on the probability of obtaining a degree.

Overall, there exists a clear urban-rural gradient for degree attainment. Location along this gradient can have a significant effect on the probability of obtaining a degree, one that rivals that of Aboriginal status or gender. Hence, part of the reason why we observe a significant difference in the share of degree holders across the urban-rural hierarchy is due to this effect. To the extent that migratory flows of degree holders flow toward larger urban areas, migration will tend to exacerbate these urban-rural differences.

6 . The null hypothesis that the marginal probabilities of obtaining a degree are the same for large urban and medium urban centres was rejected for all three models at a significance level of 0.0001, or less.