Multivariate analyses – factors related to dropping out

Warning View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Thus far, this report has focused on bivariate results. These, however, do not account for other differences that might be present. For example, dropout rates of males were presented and shown to be higher than females. However, this result might have been driven by the fact that boys tended to be less engaged in learning. It is therefore necessary to compare dropout rates of boys and girls with similar levels of engagement to derive the true effect of gender. This section uses multivariate analysis on postsecondary education dropouts only, to obtain profiles of postsecondary education dropouts in light of multiple individual, family, school and regional characteristics.

Four identical models were estimated to profile four different scenarios: dropouts from universities; dropouts from colleges; dropouts from other type of postsecondary education, and an overall postsecondary education dropout measure. Each of the four models was estimated using a logistic regression. The estimate for each analyzed variable represents odds ratios of falling into a dropout category. The estimates for the four logistic models are presented in Table 12.

The overall dropout category was derived using the conventional method, which overlooks dropout incidences if an individual graduated or was still in education. It was included in the analysis because it represents the combined outcomes of all attempts in all types of institutions.

Dropping out of university

When compared to youth in Ontario, young adults in several provinces had significantly different odds of dropping out of university. The probabilities were significantly lower in Quebec, with Nova Scotia, Saskatchewan, Alberta and British Columbia having significantly higher probabilities of young adults dropping out (Table 12, column 1).

Furthermore, women were 28% less likely to drop out of university than men1 . In Table 12, an estimate below 1 means that the individual with this characteristic had a lower chance of being a postsecondary education dropout compared to an individual in the base group. Among other characteristics associated with lower chances of being a university dropout were: being raised in a two parent family or single parent family as compared to other family type; and obtaining high school marks in the ranges of 90% and over or 80% to 89% as compared to those in the 60% to 69% grade category.

Table 12 Likelihood of dropping out by type of institution – results of four logistic regression analyses. Opens a new browser window.

Table 12
Likelihood of dropping out by type of institution – results of four logistic regression analyses

Among significant variables found to increase the chance of dropping out of university were: spending less than three hours per week on homework while in high school and having parents who started some postsecondary education but did not complete as compared to those with only high school education.

Dropping out of college / CEGEP

Among college dropouts, as compared to Ontario, significant differences were found only in two provinces (Table 12, column 2). Increased probabilities of dropping out from college / CEGEP were found in the provinces of Quebec and Alberta. Some possible explanations could be that for Quebec, CEGEP are prerequisites for university. As Quebec 's average university dropout rates were significantly lower, this may imply that the dropout process takes place in CEGEP with graduating students more likely to persist in universities.

The probabilities of dropping out of college / CEGEP were lower for young adults with the following sets of characteristics: women; those with student loans; who reported their high school marks in the categories of 90% and over and 80% to 89% as compared to those in the 60% to 69% category; who had parents with postsecondary education credentials compared to those with high school education only; and having experienced a dropout episode in high school.

On the other hand, variables found to be associated with higher odds of dropping out of college / CEGEP were: being raised in a single parent family as compared to those from other types of families; not being Canadian by birth; and reporting to have spent less than three hour per week on homework while in high school.

Dropping out of other postsecondary education

Compared to Ontario, none of the provinces obtained significantly lower odds ratios of dropping out of other types of postsecondary education (Table 12, column 3). As in the case of university students, residents of British Columbia and Nova Scotia had significantly higher probabilities of dropping out of other types of postsecondary education.

Although they had lower odds of dropping out of both, university and college, women had higher probabilities of dropping out of other types of institutions. Among variables associated with lower odds of dropping out of other type of postsecondary education were: having student loans; being 25 or 26 years old compared to 24 years old; and reporting to have been very engaged while in high school.

Dropping out of all types of postsecondary education

Overall postsecondary education dropouts face more risk of not obtaining postsecondary credentials at all. Many university, college or other type of institution dropouts might have pursued other types of postsecondary education and they might have been able to get a different status by doing so, as illustrated in Table 11. Dropouts in this group however, had not made a successful transition into another form of postsecondary education in the six years covered by the survey, meaning that if they tried another program in another type of institution, they dropped out again.

As shown in Table 12, column 4, compared to Ontario, only Quebec students had significant higher odds of being an overall type of a dropout.

Rural students were also more likely to have dropped out of all forms of postsecondary education. Increased odds of being an overall dropout were also found among: students from single parent families; students reporting spending less than three hour per week on homework while in high school; and students who also had a dropout experience while in high school.

Women, those with student loan, those reporting high school grades in the range of 90% and over and 80% to 89% and those having parents who completed postsecondary education have lower odds of being an overall dropout.

Differences between dropouts from the different types of postsecondary education

The four separate models analyzed dropouts in the three types of institutions as well as the overall postsecondary education dropout measure. Although, it is difficult to compare the differences in magnitudes between the four models, the cross relevance of these estimates can be discussed in terms of significance across models. Table 13 provides a quick summary of variables found significant in each of the models.

Table 13 Variables associated with higher or lower odds of dropping out by type of institution. Opens a new browser window.

Table 13
Variables associated with higher or lower odds of dropping out by type of institution

Clearly, certain variables were significant in more than one model. This means that these characteristics were consistently associated with higher odds of dropping out from any type of postsecondary education. These were: being a resident of Alberta, British Columbia or Quebec; having spent less than three hours on homework per week while in high school; and finally, experiencing a dropout spell while in high school.

Living in Alberta and British Columbia, as alluded to earlier in the report, may be a reflection of very favourable labour market conditions in the those provinces.

Residents of Quebec also experienced higher odds of being dropouts, but these were concentrated among college students. With Quebec 's unique CEGEP system, with mandatory college participation prior to entering university, young adults appear to drop out of college / CEGEP before entering university.

Finally, low homework intensities while in high school, as well as dropout episodes in secondary education were also found to be repeatedly related to higher dropout rates. Not only were these variables significant in several models but their magnitude was often higher than other variables. These might be an indication of a learning culture that was developed prior to entering postsecondary education and which persists during their learning careers.

On the other hand, certain variables were repeatedly associated with lower odds of dropping out. They were: having a student loan and reporting good grades when they were in high school.

Few variables were also found to have very different odds of dropping out when compared across the models. For instance, students from Quebec, and students coming from a single parent family were less likely to drop out of university and much more likely to do so from college / CEGEP.


Note :

  1. The ratios are obtained by subtracting 1 from the estimate. For exact ratios, for example, an estimate of 0.669 for Quebec in the university model, (Table 12) means that students from Quebec are 33% (0.0.669-1) less likely to drop out than Ontario youth, and an estimate of 1.654 for Newfoundland and Labrador means that students from that province are 65% (1.654-1) more likely to drop out than students from Ontario which is the province of reference in the model.