Persistent inequality or liberation from social origins? Determining who attends graduate and professional schools in Canada's expanded postsecondary system.

Zarifa, David

INTRODUCTION

AS UNDERGRADUATE ENROLLMENTS at universities continue to rise, many Canadian students are confronted with the decision to attend even "higher" education. While graduate degrees have long been an option for those who wished to further their expertise in their field or continue their professional development on a part-time basis, an increasing proportion of the population may be considering graduate-level credentials as the next step after the bachelor's in securing more favorable labor market outcomes. Many students holding a bachelor's degree are finding themselves underemployed or working outside their field of study (Boudarbat and Chernoff 2009; Frenette 2000; Yuen 2010). While further research on the motivations behind graduate school enrollment is warranted, it is possible that these unfulfilled promises of a university degree have prompted a new motivating logic for pursuing advanced degrees in their field. Greater proportions of students may be seeking further credentials that will yield a "good" job, while smaller proportions of students may be attending graduate programs in efforts of making a contribution to their field. (1)

At the same time, governments have been encouraging universities to increase access to graduate studies, providing greater allotments of provincial (e.g., OGS, OGSST) and federal funding (e.g., SSHRC, NSERC, CIHR, CGS). Governments, seeking to increase Canada's research and development profile worldwide as well as a concern for projected professor shortages, have proposed a variety of innovative strategies in recent years (CAGS 2004). These actions have prompted universities to expand their existing graduate programs and develop new programs. The end result has been a dramatic increase in graduate enrollments nationwide. A recent report released by The Higher Education Quality Council of Ontario (HEQCO) found that graduate enrollments increased approximately 44 percent from about 75,000 students in 1999 to 2000 to just over 100,000 in 2008 to 2009 (Wiggers, Lennon, and Frank 2011).

While policymakers have recognized the value of advanced degrees in the labor market and have increased graduate funding over the last few decades, it is not clear whether some groups are more likely to attend than others. Despite increased scholarship opportunities, Canadian graduate students and those in advanced professional degrees face greater tuition increases compared to their undergraduate counterparts. In 2010/11, the average tuition fees for full-time undergraduate students was in the order of $4,000 to $5,000, while tuition fees in many graduate and professional programs exceed $10,000, $15,000, and even $20,000. (2) These trends may create difficulties for individuals from low socioeconomic backgrounds.

Drawing on the 2000 cohort of the National Graduates Survey (NGS), this paper examines who attends graduate and first-professional degrees five years after university graduates have completed a bachelor degree. (3) Specifically, two formal research questions guide the analyses. First, to what extent does socioeconomic status (SES; parent's education and income), gender, race, or age impact one's likelihood of obtaining a graduate degree in Canada? Are there educational factors (academic ability, aspirations, academic confidence, field of study, type of undergraduate degree) that minimize the effects of social origins on completing a professional or graduate program? (4)

This study considers two possible hypotheses on the relationship between social origins and post-undergraduate decisions. On the one hand, theories of Maximally Maintained Inequality (MMI) predict that inequality will persist as individuals from advantaged backgrounds will continually migrate toward higher levels of credentials, once access opens up and becomes universal at lower levels (Raftery and Hout 1993). On the other hand, an alternative position suggests that social origins may not yield any direct effects on graduate school decisions. Graduates may be "liberated" from their social origins by virtue of their undergraduate experiences, and thus the effects of family background on graduate school should be substantially weaker than those at lower credential levels (Mare 1980, 1981; Stolzenberg 1994).

A growing body of research has become increasingly concerned with new modes of differentiation in expanded systems of higher education. As access to undergraduate education continues to increase for traditionally disadvantaged groups, sociologists are paying closer attention to who goes into particular fields of study, selective institutions, or continues into graduate school (for a review, see Gerber and Cheung 2008; Ayalon and Yogev 2005; Davies and Guppy 1997; Ethington and Smart 1986; Goyette and Mullen 2006; Hearn 1991; Karen 2002; Mullen, Goyette, and Soares 2003; Stolzenberg 1994; van de Werfhorst, Graaf, and Kraaykamp 2001; Zarifa 2008a). These studies, however, have not compared the post-undergraduate educational experiences of Canadian university graduates. In Canada, existing studies typically focus on the labor market outcomes of undergraduates and graduates (e.g., Betts, Ferrall, and Finnie 2000; Boudarbat and Chernoff 2009; Desjardins and King 2011; Finnie 1999; Finnie and Frenette 2003; Frenette 2000; Walters and Zarifa 2008; Yuen 2010; Zarifa and Walters 2008), while other studies describe graduate enrollment trends and the characteristics of degree-holders (e.g., CAGS 2006; Gluszynski and Peters 2005; Wiggers et al. 2011; Williams 2005). As increasing proportions of baccalaureate degree-holders aspire to attend graduate school, the results of this study will also be informative to policymakers and education officials who may seek to continue to improve access to post-undergraduate programs.

REVIEW OF THE LITERATURE

Much of the existing literature in the United States has found mixed results on the relationship between social background characteristics and continuing on to higher levels of education. The debate centers around two major questions: (1) does inequality persist beyond the well-documented trends at postsecondary entry? (2) Or, are individuals "liberated" from their social origins during their undergraduate careers?

Some early studies in the status attainment tradition suggested that the role of SES declines substantially over the course of one's undergraduate years and may have little (if any) influence on graduate school decisions. Originating in Mare's (1980, 1981) work and gaining momentum in recent replications of this research (see Stolzenberg 1994), researchers found graduate school admissions to be "uncontaminated" by social origin effects. Mare's (1980) data from the late 1960s and early 1970s revealed little influence of father's education or socioeconomic index scores on decisions to attend graduate programs. According to this perspective, most of the selection effects (e.g., by SES) occur when students enter into the postsecondary system. That is, social background effects may be largely eliminated by the time one graduates with their undergraduate degree. Put differently, as one climbs the postsecondary ladder, one becomes further removed from one's social origins and engages in a more meritocratic process. Contrary to the well-established selection processes that occur at lower credential levels, the effects of social background among those individuals who have graduated high school, entered university, and completed their degrees may be comparatively smaller.

Similarly, Stolzenberg's (1994) cohort analyses of the National Longitudinal Study (1972, 1976, and 1986) uncovered minimal effects of family background on students' likelihood of taking a graduate entrance examination (LSAT, GMAT, or GRE) or entering into an MBA program. For Stolzenberg (1994), SES effects may decline over the course of the college career, though not necessarily linearly. As students gain more and more experiences during college, the effects of status origins decline. Parental background effects on graduate school admissions (if they exert any influence at all) may operate through students' educational aspirations. For college graduates, the link between aspirations and educational continuation is quite strong. However, Stolzenberg (1994) argues that the formative years for student aspirations to attend graduate school occur during the undergraduate degree. That is, students develop, revise, and reshape their educational aspirations and goals in an atmosphere and at a time in their life when parental influences have declined or even subsided.

In the current study, support for the "liberation" argument may take the form of small or nonsignificant social background influences on attending graduate or professional school. At the same time, if strong, marginal SES effects weaken or subside when factors such as academic achievement, educational aspirations, and undergraduate experiences are included in the models, SES may be exerting its influence indirectly. Such a situation would also point to declining social origins effects.

An alternative explanation focuses on the ways in which inequality persists, and suggests that educational expansion may not necessarily reduce class inequalities (for a review, see Arum, Gamoran, and Shavit 2007). Emerging in Raftery and Hout's (1993) work at the secondary school level and more recently applied at the tertiary level (see Shavit and Blossfeld 1993; see also Shavit, Arum, and Gamoran 2007), theories of "MMI" argue that inequality at a given transition point will persist until the point at which nearly all members of privileged origins attain that level of education. According to this perspective, the recent structural expansion and proliferation of graduate programs and enrollments in Canada may not translate into improved access and a reduction in inequality. In fact, for MMI, demand from advantaged groups may be a key driving force leading to expansion of graduate and professional programs in order to perpetuate their advantages. As one level of education becomes more accessible and becomes universal, socially privileged groups will migrate toward higher levels of education. That is, social background influences should continue to be strong, as one examines higher level transitions beyond high school graduation. Given the relatively recent expansion of graduate level education, social origins should continue to play an important role in determining entry patterns.

In the analyses, support for the MMI perspective would take the form of strong, marginal, and partial SES effects on graduate school decisions. In other words, SES should show strong direct effects on post-undergraduate decisions. Additionally, social origins should play a more prominent role in predicting graduate school entry in comparison to aspirations and prior academic achievements.

While theories of MMI have yet to be systematically explored in studies aimed at predicting graduate and professional school behaviors, previous research has found parental levels of education to have strong and lasting influences on student's decisions to attend graduate or professional school (Ethington and Smart 1986; Mullen et al. 2003). For example, Ethington and Smart's (1986) study in the United States examined decisions to enroll in graduate school for just over 6,000 college freshmen in the fall of 1971 and again in a follow-up survey in 1980. The authors found that socioeconomic characteristics strongly influenced the initial choice of undergraduate institution. However, SES was also shown to indirectly affect the decision to enroll in graduate school through intervening variables of their undergraduate experience.

Similarly, in their analyses of the 1992 to 1993 cohort of the Baccalaureate and Beyond Study in the United States, Mullen et al. (2003) found that parents' education did matter for predicting graduate school attendance, and entry into professional programs. In addition to the status attainment studies above, the authors draw on Pierre Bourdieu's social reproduction theory (Bourdieu 1984; Bourdieu and Passerson [1977] 1990) to explain the parental background effects they uncovered at the graduate level. The authors argue that parent education does not simply determine one's choice of graduate program; the impact of such advantages is largely indirect. The type and prestige of parental education does not merely provide their children with a passport to a Master's or Doctorate degree. Rather, family background effects operate through the student's academic achievements, career, and educational expectations (i.e., habitus) and the institutional characteristics of their undergraduate university. Moreover, the authors also found that the strength of this relationship varies by post-graduate program. That is, parents' education has stronger effects on entering first-professional programs and doctoral programs and weaker (albeit significant) effects on entering Master's programs. According to these findings, the role of postsecondary institutions is somewhat contradictory. On the one hand, academic institutions provide avenues for conferring the educational advantages of parents. On the other hand, students from disadvantaged backgrounds can overcome some of their disparities by performing well in their undergraduate careers.

The relationship between undergraduate school characteristics and graduate school attendance has been explored in various ways. U.S. research suggests that graduating from a high-quality undergraduate institution increased the likelihood of enrolling in a high-quality graduate school (Ethington and Smart 1986; Lang 1987; Mullen et al. 2003). Others have reached similar conclusions, as individuals who graduated from elite private colleges also had higher probabilities of attending graduate school, and graduate schools at more prominent research universities (see Eide, Brewer, and Ehrenberg 1998). Zhang (2005) also found graduates from high-quality colleges were not only more likely to enroll in graduate programs, graduate programs at high-quality colleges, but also doctoral programs. The relationship between the quality of the undergraduate institution and graduate school enrollment may in part reflects indirect family background influences, as a long list of studies in the United States have uncovered strong family background effects on entering high-quality/selective undergraduate institutions (Davies and Guppy 1997; Hearn 1991; Karen 2002; Zarifa 2011).

At the bachelor's level, recent research has shown that family background impacts one's choice of undergraduate major or field (e.g., Davies and Guppy 1997; Goyette and Mullen 2006; Zarifa 2008b). In addition to the selectivity or quality of the undergraduate institution, field of study or college major may serve as an additional intervening mechanism for indirect family background effects on attending graduate school (Bedard and Herman 2008; Mullen et al. 2003; Perna 2004; Sax 2001; Zhang 2005). A recent critique of MMI (see Lucas 2001), Effectively Maintained Inequality suggests that once a given point of selection becomes saturated, social background may continue to exert its influence on the track, type, or selectivity of the program. Several scholars have taken a similar line of inquiry to show that when education expands at the undergraduate level, social origins may become "effectively maintained" through the qualitative forms of education pursued (Ayalon and Shavit 2004; Breen and Jonsson 2000; Tsai and Shavit 2007; Zarifa 2008b). Therefore, it is important to examine the field of study effects on graduate school choices, as fields represent an increasingly important point of differentiation in expanded systems of higher education.

A final concern throughout much of the literature is the impact of gender on graduate and professional degree enrollment. That is, are women underrepresented in graduate and advanced professional programs? While enrollments and completion rates for women at the undergraduate levels are significantly higher compared to men (see Buchmann, DiPrete, and McDaniel 2008), enrollments at the graduate and professional school level have traditionally been higher for men. Mullen et al. (2003) found the odds of entering a Master's program were greater for women, and the odds of entering all other programs were greater for men. Perna (2004) reached similar conclusions, as men were more likely to enroll in both doctoral and professional programs compared to women, who were overrepresented in enrollments at the Master's level. Recent Canadian data shows that about 56 percent of students enrolled in Master's programs are female, compared to only 47 percent of doctoral students (Wiggers et al. 2011:11). Still, it is unclear what role (if any) gender may continue to play on graduate school admissions, or even how the relationships may differ in Canada.

METHODS

Data

The data for this study are drawn from the 2000 cohort of Statistics Canada's National Graduate Survey (NGS) and follow-up survey. The NGS was conducted via computer-assisted telephone interviews in 2002, two years after respondents had graduated from a postsecondary institution. The NGS surveys provide extensive information on the educational experiences, program history, and early labor market experiences of recent graduates. The NGS is derived from stratified, systematic random sample designs without the replacement of graduates within strata. The respondents are initially surveyed two years following graduation, and then a follow-up survey is conducted five years after graduation (in 2005). The survey population is composed of all graduates of Canadian postsecondary educational institutions who had completed the requirements for degrees, diplomas, or certificates during the 2000 calendar year.

The 2000 NGS and follow-up of graduates contain information on approximately 35,000 postsecondary graduates of various programs across all provinces and territories. Given conceptual and sample requirement issues, trades/vocational graduates in the 2000 NGS were not followed-up in 2005. This sampling design produced a final sample size of 34,304 (68.5 percent) responding graduates in the five-year follow-up of graduates. The analyses for this paper, however, were restricted to only university graduates who completed their first bachelor's degree (e.g., B.A., B.Sc., B.Ed., B.A.Sc., B.Eng.) in 2000 and were Canadian citizens. The final sample includes a total 7,325 respondents.

Variables and Procedures

The analyses include descriptives and multinomial logistic regression models, appropriate for a categorical response variable with multiple unordered categories (Fox 2008; Long 1997; Long and Freese 2000). Essentially, the effects of the independent variables are allowed to differ for each outcome of the response variable. The multinomial logit model can be written as follows

ln [[OMEGA].sub.m|b] (x) = in Pr(y = m|x)/Pr(y = b|x) = x [[beta].sub.m|b] for m = 1 to J

where b is the base category and the log odds of an outcome compared with itself must equal zero (i.e., ln [[OMEGA].sub.b|b] (x)= ln 1 = 0), and the effects of the independent variables are also zero (Long and Freese 2000).

To further grasp the results of the statistical analyses, predicted probabilities and 95 percent confidence intervals around the estimates are calculated and presented in graphical displays. The predicted probability that y = m given x is calculated as follows

Pr(y = m|[x.sub.i]) = exp ([x.sub.i][[beta].sub.m])/[summation.sup.J.sub.j=1] exp([x.sub.i][[beta].sub.j])

where [[beta].sub.1] : 0 and, [[beta].sub.m] is a vector of coefficients for the effect of [x.sub.k] on outcome m and J is the number of outcomes of the dependent variable (Long 1997). To examine the effects of particular explanatory variables [x.sub.k], all variables except for [x.sub.k] were held constant at their means or proportions.

For the multinomial logistic regressions, a number of key explanatory variables are entered in several stages. In order to explore the direct and indirect effects of social background on post-undergraduate outcomes, base models include only family background measures, and subsequent models include a number of sociodemographic control variables, measures of ability and aspirations, and variables related to the respondents' undergraduate education.

The sociodemographic variables include respondent's age (in years), marital status, presence of dependent children, gender, and visible minority status. To measure the effects of family background on attending graduate school, the models include sets of dummy regressors for the highest level of education of either parent. (5) Unfortunately, the NGSs do not contain conventional measures of family income. The total dollar amount of government-sponsored student loans obtained during respondents' undergraduate degrees is employed to provide an estimate of respondents' level of income. Student loans obtained after the undergraduate degree (for those who were enrolled in further education) are not included in this measure. Based on the highly skewed distribution of the loans variable, which included a large number of meaningful zero values, a set of dummy variables were created with the following categories: "no loans," "loans of $15,000 or less," and "loans above $15,000." Government-sponsored student loans provide the best alternative, since family income is directly involved in students' eligibility and the amounts of loans offered. Moreover, since undergraduate tuitions are fairly similar across disciplines in Canadian institutions, the amount of loans may be only slightly affected by students' entering a more expensive program.

Measures of student ability and aspirations have also been identified in the literature as important predictors of graduate school attendance. Since the NGS does not obtain information on students' grades or GPA in their undergraduate program, the total dollar amount of the respondents' achievement-based scholarships, awards, and fellowships obtained during their undergraduate degree is used as a proxy measure for ability. (6) Scholarships awarded to students who may have enrolled in further education beyond the undergraduate level are not included in this measure. Since the distribution of scholarships was very highly skewed to the right and included a large proportion of students with meaningful zero values, the proxy for ability was treated as a categorical variable with three categories: "no scholarships," "scholarships of $5,000 or less," and "scholarships above $5,000." In addition, the NGS also includes a self-assessed measure of students' perceived academic standing during their undergraduate degree. That is, students are asked to indicate where they ranked academically in comparison to their graduating class (e.g., top 10 percent, below top 10 percent but in the top 25, below the top 25 percent). While the distribution of this variable casts serious doubt on its ability to measure academic performance, this variable is included as an indicator of academic confidence--one that previous work shows influences graduate school attendance. (7) In terms of aspirations or educational expectations (see Mullen et al. 2003), a set of dummy variables indicating whether or not students were planning on pursuing a Master's degree are included in the models.

Finally, several variables that relate to the respondents' undergraduate educational program and experience are also included in the models. (8) Previous studies have pointed to a number of factors that may influence graduate enrollment decisions. Field of study consistently shows significant effects (Bedard and Herman 2008; Goyette and Mullen 2006; Mullen et al. 2003; Sax 2001; Zhang 2005). The analyses include students' undergraduate field of study based on the Classification of Instructional Programs (CIP), developed by the National Center for Education Statistics in the United States. The analyses contain an aggregated version of the field of study classification containing six categories. The final categories used in the analyses are displayed below.

1. Arts and Humanities,

2. Social and Behavioral Sciences,

3. Business and Management,

4. Physical and Life Sciences,

5. Engineering, Mathematics, and Computer Science, and

6. Other.

In addition to field of study, we also consider the impact of type of undergraduate education. The analyses include a dichotomous variable indicating whether or not the student completed a co-operative education program in their undergraduate degree. This measure may provide some indication of the work values and career motivations of respondents (see Stolzenberg 1994). Students who attended a co-op program may be less likely to continue on into a graduate or professional program. They may have their sights set on entering the labor market right after graduation, and may seek out experience-based learning, and the labor force networking it brings with it, to enhance their employment opportunities upon the completion of their undergraduate education. This study additionally controls for regional effects in the analyses, as a set of four dummy variables tapping into respondents' region of undergraduate institution was also included. Finally, a measure of satisfaction with their undergraduate field of study was included. This variable indicates whether or not the graduate would select the same field if they were to do their undergraduate studies again. It is possible that those who were satisfied with their undergraduate field of study may be more likely to continue on to higher levels of education within that field.

The key response variable in the analyses measures whether respondents attended a graduate degree (Master's or Doctorate), first-professional degree (degree in law [LL.B.], medicine [M.D.], dentistry [D.D.S., D.M.D.], veterinary medicine [D.V.M.], optometry [O.D.]), or did not pursue graduate education by 2005 (five years after graduating with their bachelor's degree). In line with existing studies (Mullen et al. 2003), the analysis differentiates first-professional degrees from Master's and Doctorate programs. Unfortunately, this study could not further separate out types of graduate degrees (e.g., MBAs) due to issues relating to sample size. Thus, the analyses are unable to account for differences across graduate programs in terms of the commitment required of students or the varied rewards upon completion (see Mullen et al. 2003). The analyses also do not capture those individuals who enter professional programs prior to completing their undergraduate degree nor do they provide information on individuals who may eventually enroll in or complete a program after 2005.

RESULTS

Descriptives

Table 1 displays the means and proportions of all variables of interest in the NGS 2000 to 2005. Overall, graduates in the sample were predominately single, nonvisible minorities, did not have a dependent child, and were more likely to be female. In terms of parents' education, about 37 percent of graduates' parents did not have any postsecondary education, about 20 percent had a college or trades diploma, about 26 percent had a bachelor's degree, and about 15 percent of graduates had a parent who held a Master's or Doctorate degree. That is, over 40 percent of the recent university graduates had one or more parents who also held a university degree.

Most graduates reported not having taken out any government-sponsored student loans, while fully 22 and 21 percent had loans less than or exceeding $15,000, respectively. In terms of ability, about 60 percent of the sample did not hold an achievement-based scholarship in their undergraduate education. About 30 percent were awarded small scholarships ($5,000 or less), and nearly 10 percent obtained larger scholarships ($5,000 or more). When asked about where they thought they ranked in comparison to their classmates, about 29 percent thought they were in the top 10 percent of their class, 47 percent believe they were below the top 10 percent but in the top 25 percent, and about 24 percent felt they ranked below the top 25 percent. In terms of aspirations or intentions of attending a graduate school, about 37 percent of graduates expressed plans of attending a Master's program, and 20 percent were undecided or did not know.

Most graduates in the NGS obtained degrees in the Social and Behavioral Sciences (24 percent), Physical and Life Sciences (20 percent), Arts and Humanities (17 percent), and Business and Management (16 percent). Only 11 percent of the sample graduated from a co-operative education program. Not surprisingly given the higher concentration of postsecondary institutions, most graduates obtained their degrees from Ontario (40 percent), the Western Provinces and Territories (27 percent), and Quebec (24 percent). The vast majority of graduates were also satisfied with the field of study they graduated from, as over three-quarters of the graduates reported that they would select the same field again.

Finally, 81 percent of graduates in 2000 did not pursue graduate or advanced professional degrees beyond their bachelor's. Only 3 percent had entered professional degrees, and 16 percent had enrolled in a Master's or Doctorate degree.

Before turning to the multinomial logits, it is important to first consider the marginal effects of family background on post-undergraduate decisions. Chi-square tests reveal that all variables are significantly related to education continuation decisions (see Appendix). Parent education does show strong direct effects on graduate school attendance. That is, nearly 35 percent of graduates whose parents obtained a Master's or Doctorate degree entered a professional or graduate level program compared to only about 13 percent of graduates whose parents did not attend postsecondary education. As others have found in the United States (see Mullen et al. 2003), parent education among Canadian graduates also appears to have stronger effects on entry into Master's or Doctorate programs in comparison to its effects on professional program entry. Nearly twice as many graduates whose parents obtained a graduate degree entered a Master's or Doctorate program in comparison to graduates whose parents obtained a bachelor's degree. Additional family background effects emerge when examining undergraduate student loans. About 21 percent of graduates without government-sponsored student loans entered a graduate or professional program compared to only about 14 percent of graduates with loans above $15,000. These strong SES effects lend initial support to a situation of persisting inequality at the graduate level.

Multinomial Logistic Regressions

To examine the extent to which these strong SES effects hold once controlling for other factors, we turn to multinomial logistic regressions. The parameter estimates for the regressions of graduate school attendance on the explanatory variables are provided in Table 2. In Model 1, the response variable is regressed on only family background characteristics. While students' economic background (approximated by their level of government-sponsored student loans) does not show any significant effects, parental education shows strong, positive effects on graduates' odds of obtaining a degree beyond the bachelor's level. Graduates whose parents held a university degree (bachelor's or graduate degree) were significantly more likely to obtain a professional degree compared to those whose parents did not have postsecondary credentials (p < .01). Graduates with highly educated parents were also significantly more likely to obtain a Master's or Doctorate degree. This relationship was particularly strong for graduates whose parents held a Master's or Doctorate degree (p < .001). These findings are consistent with more recent studies in the United States (Mullen et al. 2003) and contrast older studies which found minimal family background influences (Mare 1980; Stolzenberg 1994). In other words, the findings lend additional support for the persistence of inequality (at least in terms of parent's education). Both of these trends can be seen when looking at the predicted probabilities for Model 1 in Figure 1.

[FIGURE 1 OMITTED]

In Model 2, sociodemographic control variables are added to the analyses. Several findings emerge from these models. Graduates who were married were significantly (p < .01) less likely to obtain an advanced professional degree. Older respondents were also less likely to obtain a first-professional degree (p < .01). Women were also significantly less likely (p < .01) to pursue graduate degrees in comparison to their male counterparts. Unfortunately, we are unable to separate out graduate degrees to further explore whether these gender effects are attributable to differences at the doctorate level. On the other hand, the odds of entering a first-professional program did not differ significantly for males and females, a finding that is much more encouraging than previous U.S. studies. Despite adding these controls, the effects of parent education weaken only slightly in predicting entry into professional programs. For entry into graduate programs, however, only those graduates whose parents hold Master's or Doctorate degrees are significantly more likely to do so (p < .001).

In Model 3, measures of academic achievement, aspirations, and academic confidence are added to the mix. Several findings are worthy of mention. All of the effects observed in the previous models hold with one exception. The gender effect on pursuing graduate studies is no longer statistically significant, once controlling for ability, aspirations, and confidence. Multiple parameter likelihood ratio tests indicate that scholarships, aspirations, and academic confidence all have strong effects (p < .001) on post-undergraduate decisions. Specifically, graduates who obtained scholarships above $5,000 during their undergraduate studies were more likely to pursue advanced professional degrees and graduate degrees compared to graduates with no undergraduate scholarships (p < .001). Further, graduates with moderate (p < .05) and high (p < .001) levels of confidence in their academic abilities were significantly more likely to enter a professional program, and those with high levels of confidence also had higher odds of attending a graduate program (p < .01). For aspirations, an interesting picture emerges. Graduates who had plans of pursuing a graduate degree (p < .01), and even those who were undecided (p < .001), were significantly more likely to do so, and less likely to obtain a professional degree (p < .01) compared to their counterparts with no plans.

The addition of these measures greatly reduces the log likelihood and increases the pseudo [R.sup.2] from 0.0534 to 0.4753. The ability of these academic indicators to explain such a substantial portion of the variance in graduate decisions lends support to the "liberation" perspective. Social origins have long been shown to have strong direct impacts on academic achievement and student aspirations. As a result, their importance in explaining differences across post-undergraduate enrollment may be evidence of the indirect effects of family background.

Finally, Model 4 includes a number of educational characteristics in the models. The addition of these parameters increases the pseudo [R.sup.2] from 0.4753 to 0.5004. The results indicate that whether or not students attended a co-op program in their undergraduate studies did not significantly impact their decisions to attend a graduate or professional program. However, field of study, satisfaction with field of study chosen, and the region of the undergraduate institution all show strong and significant effects on post-undergraduate decisions. It is important to note that parent education continues to show strong direct effects. Figure 1 shows the predicted probabilities across levels of parent education from Models 1 and 4. Interestingly, when academic variables are included in the model, the probabilities of entering graduate education no longer increase in a linear fashion. While graduates whose parents hold a Master's or Doctorate remain most likely to obtain a graduate degree, graduates whose parents did not attend postsecondary education show the second highest probability of attending graduate school. Still, the strong parental education effects across all models support theories of persisting inequality.

Even after controlling for all other factors in the models, field of study plays an important role in determining one's post-baccalaureate outcomes. Figure 2 displays the predicted probabilities of obtaining a professional or graduate degree by field of study. Our findings coincide with much of the work done previously on U.S. graduates (Bedard and Herman 2008; Goyette and Mullen 2006; Sax 2001; Zhang 2005). Graduates from the Physical and Life Sciences, Arts and Humanities, and Social Sciences show the highest probabilities of completing an advanced professional degree as well as a graduate degree. At the same time, graduates from Business, Engineering, Math, and Computer Science show significantly lower probabilities of pursuing professional or graduate studies.

[FIGURE 2 OMITTED]

DISCUSSION

This study investigates two main issues related to graduate and professional school attendance. First, this study empirically establishes who attends first-professional, Master's, and Doctorate degrees within five years of the successful completion of their first bachelor's degree. Second, this research uniquely contributes to the existing debates on the relationship between social origins and education continuation decisions by exploring recent trends in Canada's postsecondary system.

To what extent can graduate school attendance be explained by family background influences beyond academic ability, aspirations, and the characteristics of one's undergraduate career? In other words, does inequality persist to graduate school entry whereby SES maintains a strong, direct influence, or are graduates "liberated" from their social origins by the time they consider post-undergraduate education?

Overall, the results revealed that nearly 20 percent of bachelor's degree-holders from the 2000 cohort went on to attend some form of professional or graduate degree. Social background was indeed a contributing factor in determining who is selected into this privileged group of individuals. The results of this study show social origins may affect graduate school attendance both directly through parents' level of education and indirectly through student performance.

Despite recent improvements in access to undergraduate education for traditionally disadvantaged groups, significant inequality persists at the professional and graduate levels. The marginal relationships revealed strong and significant relationships between both parental education and income and graduate school attendance. In all of the regression models, however, only parents' education yielded strong and significant effects on one's chances of completing a graduate program. Most notably, graduates with highly educated parents (i.e., Master's or Doctorate degree-holders) experienced the greatest advantages in securing professional or graduate degrees. These graduates were least likely to end their educational careers at the bachelor's level. The lack of a strong income effect is somewhat surprising given the deregulation of tuition fees in many professional programs. Still, contrary to Mare's (1980, 1981) and Stolzenberg's (1994) work, this study did find strong parental education effects, even when controlling for educational achievement and experience factors. The findings provide some evidence that inequality may be "maximally maintained" even at the graduate level by virtue of parents' level of education.

Despite the strong and consistent parent education effect, this study also found strong evidence to suggest that students may be at least partially "liberated" from their family backgrounds by their academic abilities and undergraduate experiences. Social origins exert their strongest influences on post-undergraduate decisions indirectly through students' educational achievements and experiences. Being highly skilled, having high educational aspirations, and perceiving oneself to have a high academic standing greatly explain differences in graduates' choices. High-ability students were much more likely to complete professional and graduate degrees. And, those who aspired to complete a graduate degree during their undergraduate studies were significantly more likely to do so, and less likely to pursue a professional program than refrain from further studies. Graduates who believed they were in the top 10 percent of their class were significantly more likely to pursue professional degrees, and were no more or less likely to pursue graduate degrees.

In addition, the findings also suggest that field of study plays a significant role. In line with U.S. studies, field of study (or college major) represents an increasingly important intervening mechanism for indirect family background effects on attending graduate school (Bedard and Herman 2008; Mullen et al. 2003; Perna 2004; Sax 2001; Zhang 2005). As undergraduate enrollments continue to rise, undergraduate field of study has become an "effective" way of differentiating graduates and maintaining inequality (see Lucas 2001). Recent Canadian evidence supports this hypothesis, as social origins continue to exert their influence on undergraduate field of study decisions (Zarifa 2008b).

In this study, those who attended some undergraduate fields of study were much more likely to enter professional or graduate programs than others. Graduates from more general arts and sciences programs were most likely to continue their education, while those in more applied degrees (e.g., Business, Education, Math, and Computer Science) were less likely to attend similar programs. It is possible that some graduates from these applied fields are less likely to continue on to higher levels of education in their field simply because they are more likely to secure full-time employment, work in a job related to their field, and attain higher earnings with their undergraduate degrees (Bourdarbat and Chernoff 2009; Finnie 1999; Finnie and Frenette 2003; Yuen 2010; Walters 2004). Students in the more general arts and sciences programs may continue on to graduate school because of the perceived (or real) difficulties in securing a "good" job right after completing their undergraduate degree. Further exploration of student motivations and attitudes underlying graduate school attendance and how these may vary across disciplines, however, remains an important area for future inquiry.

One limitation of the current study is that the data measure graduates' actions within five years of completing their bachelor's degree. As it stands, the findings presented here may not capture those individuals who pursue graduate studies after spending some time in the labor market. Perhaps the strong parental education effects emerging in the analyses are in part attributable to this short time frame. Yet, it remains unclear if students who take more time off between university degrees are characteristically different than those who enroll in higher levels of education with shorter or no breaks at all. This issue remains important for future research. Should new data become available, it would be useful to follow graduates over a longer period of time. Appendix Percentage of Graduates Entering Graduate and Professional Schools across Graduate Characteristics in the 2000 to 2005 Cohort of the National Graduates Survey and Follow-Up of Graduates Survey in Canada First Master's No graduate professional or Variable education degree Doctorate All graduates 81.3 2.9 15.8 Marital status *** Single/previously married 79.8 3.6 16.6 Married/common-law 85.5 1.0 13.6 Gender *** Male 77.9 2.9 19.2 Female 83.6 2.8 13.5 Race *** Not visible minority 81.1 2.6 16.3 Visible minority 81.9 4.6 13.4 Children *** No dependent child 80.3 3.1 16.5 Dependent child 89.3 0.6 10.1 Parent education *** No postsecondary 87.2 1.6 11.2 College, trades, or 84.3 1.9 13.8 vocational diploma Bachelor degree 79.7 4.6 15.7 Master's or Doctorate 65.3 4.4 30.3 degree Undergraduate student *** loans No government-sponsored 79.1 3.5 17.4 loans Loans of $15,000 or less 82.7 2.1 15.2 Loans above $15,000 85.7 1.9 12.4 Undergraduate scholarships *** No achievement-based 85.9 1.6 12.5 scholarships Scholarships of $5,000 or 84.0 3.7 12.3 less Scholarships above $5,000 43.2 7.6 49.1 Aspirations *** No Master's degree plans 94.4 4.6 0.9 Master's degree plans 93.4 1.9 4.8 Undergraduate field of *** study Arts and Humanities 81.2 4.3 14.5 Social and Behavioral 77.5 2.8 19.7 Sciences Business and Management 88.8 1.7 9.6 Physical and Life Sciences 74.0 4.9 21.1 Engineering, Mathematics, 81.6 1.3 17.1 and Computer Science Other 92.5 0.3 7.2 Type of undergraduate *** education Not a co-operative 81.2 3.0 15.7 education program Co-operative education 82.7 1.3 16.0 program Satisfaction with undergraduate field of study Would not select the same 87.7 2.2 10.1 field Would select the same field 79.4 3.1 17.5 Region of undergraduate *** institution Atlantic Canada 83.6 2.6 13.8 Quebec 75.5 1.5 23.0 Ontario 81.9 3.3 14.7 Western Provinces and 85.2 3.5 11.3 Territories Self-assessed undergraduate *** academic rank Below top 25 percent 90.7 0.8 8.5 Below top 10 percent but 82.0 2.4 15.7 in top 25 percent Top 10 percent 68.1 4.6 27.3 n 7,325 Chi-square tests of significance are reported. * p <.05; ** p < .01; *** p < .001.

References

Ayalon, H. and Y. Shavit. 2004. "Educational Reforms and Inequality in Israel: The MMI Hypothesis Revisited." Sociology of Education 69:82-101.

Ayalon, H. and A. Yogev. 2005. "Field of Study and Students' Stratification in an Expanded System of Higher Education: The Case of Israel." European Sociological Review 21(3):227-41.

Bedard, K. and D. Herman. 2008. "Who Goes to Graduate/Professional School? The Importance of Economic Fluctuations, Undergraduate Field, and Ability." Economics of Education Review 27:197-210.

Betts, J., C. Ferrall and R. Finnie. 2000. "The Role of University Quality in Determining Post-Graduation Outcomes: Panel Evidence from Three Recent Canadian Cohorts." Ottawa: Statistics Canada.

Boudarbat, B. and V. Chernoff. 2009. "The Determinants of Education--Job Match among Canadian University Graduates." IZA Discussion Paper No. 4513, Institute for the Study of Labor, Bonn, Germany.

Bourdieu, P. 1984. Distinction: A Social Critique of the Judgement of Taste. Cambridge, MA: Harvard University Press.

Bourdieu, P. and J.-C. Passeron. 1990 [1977]. Reproduction in Education, Society and Culture. Newbury Park, CA: Sage.

Breen, R. and J.O. Jonsson. 2000. "Analyzing Educational Careers: A Multinomial Transition Model." American Sociological Review 65:754-72.

Buchmann, C., T.A. DiPrete and A. McDaniel. 2008. "Gender Inequalities in Education." Annual Review of Sociology 34:319-37.

[CAGS] Canadian Association for Graduate Studies. 2004. The Completion of Graduate Studies in Canadian Universities: Report and Recommendations. Ottawa, Canada: CAGS.

[CAGS] Canadian Association for Graduate Studies. 2006. A Profile of Master's Degree Education in Canada. Ottawa, Canada: CAGS.

Davies, S. and N. Guppy. 1997. "Fields of Study, College Selectivity, and Student Inequalities." Social Forces 73:131-51.

Desjardins, L. and D. King. 2011. Expectations and Labour Market Outcomes of Doctoral Graduates from Canadian Universities. Ottawa, ON: Statistics Canada, Cat. No. 81--595-M--No.89.

Dwyer, M. 2009. "B.C. Grad Students Tell Us What They Think about Their Programs." Spotlight on Graduate Studies--Macleans OnCampus. Retrieved May 15, 2011 (http://oncampus.macleans.ca/education/2009/06/22/spotlight-on-graduate- studies/).

Eide, E., D.J. Brewer and R.G. Ehrenberg. 1998. "Does It Pay to Attend an Elite Private College? Evidence on the Effects of Undergraduate College Quality on Graduate School Attendance." Economics of Education Review 17(1):73-82.

Ethington, C.A. and J.C. Smart. 1986. "Persistence to Graduate Education." Research in Higher Education 24:287-303.

Finnie, R. 1999. "Fields of Plenty, Fields of Lean--A Cross-Cohort Longitudinal Analysis of Early Labour Market Outcomes." Canadian Journal of Higher Education 31(1):141-76.

Finnie, R. and M. Frenette. 2003. "Earnings Differences by Major Field of Study: Evidence from Three Cohorts of Recent Canadian Graduates." Economics of Education Review 22(2): 179-98.

Frenette, M. 2000. "Overqualified? Recent Graduates and the Needs of Their Employers." Education Quarterly Review. November. Ottawa, ON: Statistics Canada. Cat No. 81-003XIE.

Fox, J. 2008. Applied Regression Analysis and Generalized Linear Models. 2d ed. Thousand Oaks, CA: Sage Publications, Inc.

Gerber, T.P. and S.Y. Cheung. 2008. "Horizontal Stratification in Postsecondary Education: Forms, Explanations, and Implications." Annual Review of Sociology 34:299-318.

Gluszynski, T. and V. Peters. 2005. Survey of Earned Doctorates: A Profile of Doctoral Degree Recipients. Ottawa, ON: Statistics Canada. Cat. No. 81-595-MIE2005032.

Goyette, K.A. and A.L. Mullen. 2006. "Who Studies Arts and Sciences? Social Background and the Choice and Consequences of Undergraduate Field of Study." The Journal of Higher Education 77(3):497-538.

Hearn, J.C. 1991. "Academic and Nonacademic Influences on the College Destinations of 1980 High School Graduates." Sociology of Education 64(3):158-71.

Karen, D. 2002. "Changes in Access to Higher Education in the United States: 1980-1992." Sociology of Education 75(3):191-210.

Lang, D. 1987. "Stratification and Prestige Hierarchies in Graduate and Professional Education." Sociological Inquiry 57:12-31.

Long, J.S. 1997. Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks, CA: Sage.

Long, J.S. and J. Freese. 2000. Regression Models for Categorical Dependent Variables Using Stata. College Station, TX: Stata Press.

Lucas, S.R. 2001. "Effectively Maintained Inequality: Education Transitions, Track Mobility, and Social Background Effects." American Journal of Sociology 106(6):1642-90.

Mare, R.D. 1980. "Social Background and School Continuation Decisions." Journal of the American Statistical Association 75(370):295-305.

Mare, R.D. 1981. "Change and Stability in Educational Stratification." American Sociological Review 46:72-87.

Mullen, A.L., K.A. Goyette and J.A. Soares. 2003. "Who Goes to Graduate School? Social and Academic Correlates of Educational Continuation after College." Sociology of Education 76(2): 143-69.

Perna, L.W. 2004. "Understanding the Decision to Enroll in Graduate School: Sex and Racial/Ethnic Group Differences." The Journal of Higher Education 75(5):487-527.

Raftery, A.E. and M. Hout. 1993. "Maximally Maintained Inequality--Expansion, Reform, and Opportunity in Irish Education, 1921-1975." Sociology of Education 66:41-62.

Sax, L.J. 2001. "Undergraduate Science Majors: Gender Differences in Who Goes to Graduate School." The Review of Higher Education 24(2):153-72.

Shavit, Y., R. Arum and A. Gamoran, eds. 2007. Stratification in Higher Education: A Comparative Study. Stanford, CA: Stanford University Press.

Shavit, Y. and H.P. Blossfeld, eds. 1993. Persistent Inequality: Changing Educational Attainment in Thirteen Countries. Boulder, CO: Westview Press.

Statistics Canada. 2010. "University Tuition Fees." The Daily, September 16, 2010. Retrieved April 5, 2011 (http://www.statcan.gc.ca/daily-quotidien/100916/dql00916a-eng.htm).

Stolzenberg, R.M. 1994. "Educational Continuation by College Graduates." American Journal of Sociology 99:1042-77.

Tsai, S.-L. and Y. Shavit. 2007. "Taiwan: Higher Education--Expansion and Inequality of Educational Opportunity." Pp. 140-64 in Stratification in Higher Education: A Comparative Study, edited by Y. Shavit, R. Arum and A. Gamoran. Stanford, CA: Stanford University Press.

van de Werfhorst, H.G., N.D. de Graaf and G. Kraaykamp. 2001. "Intergenerational Resemblance in Field of Study in the Netherlands." European Sociological Review 17(3):275-93.

Waiters, D. 2004. "A Comparison of the Labour Market Outcomes of Postsecondary Graduates of Various Levels and Fields over a Four-Cohort Period." Canadian Journal of Sociology 29(1):1-27.

Waiters, D. and D. Zarifa. 2008. "Earnings and Employment Outcomes for Male and Female Postsecondary Graduates of Coop and Non-Coop Programmes." Journal of Vocational Education & Training 60(4):377-99.

Wiggers, R., M.C. Lennon and K. Frank. 2011. Expanding Opportunities for Graduate Studies: The Recent Experience of Ontario. Toronto: Higher Education Quality Council of Ontario.

Williams, G. 2005. Doctoral Education in Canada: 1900-2005. Ottawa, ON: Canadian Association for Graduate Studies.

Yuen, J. 2010. Job-Education Match and Mismatch: Wage Differentials. Perspectives, April 2010. Ottawa, ON: Statistics Canada. Cat. No. 75-001-X.

Zarifa, D. 2008a. "Emerging Forms of Stratification in Higher Education: Comparing Canada and the United States." Unpublished doctoral dissertation, Department of Sociology, McMaster University.

Zarifa, D. 2008b. "Field of Study Choices in an Expansionary Era: Comparing Two Cohorts of Baccalaureate Degree-Holders in the United States and Canada." Paper presented at the annual meetings of the American Sociological Association, July 31--August 4, Boston, MA.

Zarifa, D. 2011. "Social Background and College Selectivity in an Expansionary. Era." Paper presented at the annual meetings of the American Sociological Association, August 20-23, Las Vegas, Nevada.

Zarifa, D. and D. Waiters. 2008. "Revisiting Canada's Brain Drain: Evidence from the 2000 Cohort of Canadian University Graduates." Canadian Public Policy 34(3):305-20.

Zhang, L. 2005. "Advance to Graduate Education: The Effect of College Quality and Undergraduate Majors." The Review of Higher Education 28(3):313-38.

DAVID ZARIFA

Nipissing University

(1.) A recent survey of the 2000 to 2001 and 2003 to 2004 cohorts of Master's and doctoral graduates from the University of British Columbia, the University of Northern British Columbia, Royal Roads University, Simon Fraser University, and the University of Victoria found 60 percent of respondents attended graduate programs to enhance their career opportunities, and only 40 percent stated scholarly and research-oriented reasons for pursuing their graduate education (see Dwyer 2009).

(2.) For example, the average tuition fees for an MBA, medicine, dentistry, law, and pharmacy degree were $21,118, $10,244, $14,701, $8,697, and $9,250, respectively (Statistics Canada 2010).

(3.) Unfortunately, the 2005 NGS does not yet have a five-year follow-up to chart more recent graduate completion trends.

(4.) Unfortunately, the author was unable to identify institutions in the analyses, which prevented an exploration of institutional effects (e.g., school mission, school type, size of the school, selectivity, or ranking) as in some U.S. studies (see Mullen et al. 2003).

(5.) This variable was derived by the author from NGS variables indicating the highest level of education for respondents' mother and father.

(6.) Many advanced degree programs in Canada require graduates to write a standardized entrance examination (e.g., law school, medicine, dentistry). Unfortunately, these test scores are not available in the NGS.

(7.) Ethington and Smart (1986) used a similar measure of academic confidence in their analyses. Their self-rating of academic ability was coded as follows: (11 lowest 10 percent, (2) below average, 131 average, (4) above average, and (5) highest 10 percent.

(8.) Unfortunately, the institutional identifiers are not currently available at Statistics Canada's Research Data Centres. As a result, the analyses were unable to examine institutional effects, estimate multilevel models, or even adjust the standard errors for clustering in the data.

David Zarifa, Department of Sociology, Nipissing University, 100 College Drive, North Bay, Ontario, Canada P1B 8L7. E-mail: [email protected] Table 1 Descriptive Statistics for Variables from the 2000 to 2005 Cohort of the National Graduates Survey and Follow-Up of Graduates Survey in Canada Mean/ proportion Variable (SE) Marital status Single/previously married 0.72 (0.803) Married/common-law 0.28 (0.009) Age 25 (0.125) Gender Male 0.39 (0.011) Female 0.61 (0.011) Race Not visible minority 0.84 (0.010) Visible minority 0.16 (0.009) Children No dependent child 0.88 (0.007) Dependent child 0.12 (0.007) Parent education No postsecondary 0.37 (0.011) College, trades, or vocational diploma 0.20 (0.009) Bachelor degree 0.26 (0.010) Master's or Doctorate degree 0.15 (0.008) Undergraduate student loans No government-sponsored loans 0.56 (0.011) Loans of $15,000 or less 0.22 (0.009) Loans above $15,000 0.21 (0.009) Undergraduate scholarships No achievement-based scholarships 0.62 (0.011) Scholarships of $5,000 or less 0.28 (0.010) Scholarships above $5,000 0.09 (0.003) Aspirations No Master's degree plans 0.43 (0.011) Master's degree plans 0.37 (0.011) Do not know 0.20 (0.009) Undergraduate field of study Arts and Humanities 0.17 (0.009) Social and Behavioral Sciences 0.24 (0.011) Business and Management 0.16 (0.008) Physical and Life Sciences 0.20 (0.009) Engineering, Mathematics, and Computer 0.11 (0.006) Science Other 0.11 (0.006) Type of undergraduate education Not a co-operative education program 0.89 (0.006) Co-operative education program 0.11 (0.006) Satisfaction with undergraduate field of study Would not select the same field 0.22 (0.009) Would select the same field 0.76 (0.009) Region of undergraduate institution Atlantic Canada 0.09 (0.003) Quebec 0.24 (0.007) Ontario 0.40 (0.013) Western Provinces and Territories 0.27 (0.008) Self-assessed undergraduate academic rank Below top 25 percent 0.24 (0.009) Below top 10 percent but in top 25 percent 0.47 (0.011) Top 10 percent 0.29 (0.010) Type of graduate program No graduate education 0.81 (0.008) First-professional degree 0.03 (0.004) Master's or Doctorate degree 0.16 (0.007) n 7,325 Table 2 Multinomial Logistic Regression Models of Post-Undergraduate Choices by 2005 for the 2000 Cohort of University Graduates in Canada (Coefficients shown) First Professional Model 1 Model 2 Model 3 Model 4 Constant -3.785 17.391 11.940 11.325 (0.308) (7.466) (6.055) (5.512) Parent *** *** *** *** Education No -- -- -- -- postsecondary College, trades 0.179 0.023 -0.071 -0.037 or vocational (0.477) (0.483) (0.472) (0.479) diploma Bachelor degree 1.065 ** 0.867 * 0.855 * 0.824 * (0.386) (0.381) (0.405) (0.409) Master's or 1.219 ** 0.936 * 0.863 * 0.815 * Doctorate (0.399) (0.389) (0.405) (0.416) degree Undergraduate Student Loans No government- -- -- -- -- sponsored loans Loans of -0.458 -0.355 -0.305 -0.228 $15,000 or (0.348) (0.350) (0.399) (0.407) less Loans above -0.409 -0.145 -0.023 -0.006 $15,000 (0.509) (0.507) (0.536) (0.554) Marital Status * ** * Single / -- -- -- previously married Married/ -0.856 -0.944 -0.806 - common-law (0.322) (0.307) (0.304) Children No dependent -- -- -- child Dependent child 0.196 -0.090 0.050 (0.468) (0.502) (0.480) Log(age) -6.702 -5.298 -5.136 " (2.399) (1.943) (1.730) Gender * Male -- -- -- Female -0.037 -0.157 -0.326 (0.266) (0.281) (0.275) Race Not visible -- -- -- minority Visible 0.534 0.544 0.439 minority (0.315) (0.320) (0.337) Undergraduate *** *** scholarships No achievement- -- -- based scholarships Scholarships of 0.382 0.343 $5,000 or (0.365) ***(0.412) *** less Scholarships 1.416 *** 1.313 *** above $5,000 (0.350) (0.371) Aspirations No Master's -- -- degree plans Master's degree -1.110 ***-1.164 ** plans (0.410) (0.395) Don't know -0.909 -0.898 (0.568) (0.619) Academic *** *** Confidence Below top 25 -- -- percent Below top 10 1.101 * 1.052 * percent but (0.435) (0.444) in top 25 percent Top 10 percent 1.809 *** 1.769 *** (0.405) (0.417) Type of Undergraduate Education Not a -- co-operative education program Co-operative -0.875 education (0.682) *** program Undergraduate Field of Study Arts and -- Humanities Social and -0.015 Behavioral (0.405) sciences Business and -0.700 Management (0.447) Physical and 0.314 Life Sciences (0.371) Engineering, -0.903 Mathematics, and Computer Science (0.560) Other -2.179 (0.550) Satisfaction with Under graduate Field of Study Would not -- select the same field Would select the 0.409 same field (0.318) Region of *** Undergraduate Institution Atlantic Canada -- Quebec 0.171 (0.384) Ontario 0.291 (0.407) Western 0.306 Provinces and Territories (0.277) n 7,325 7,325 7,325 7,325 Log likelihood -52,488.6 -51,460.4 -28,526.9 -27,161.8 Pseudo 0.0345 0.0534 0.4753 0.5004 [R.sup.2] Deviance 104977.3 102920.8 57053.77 54323.64 Master's or Doctorate Model 1 Model 2 Model 3 Model 4 Constant 1.847 0.197 -2.606 -3.027 (1.851) (1.790) (1.999) (1.974) Parent *** *** *** *** Education No -- -- -- -- postsecondary College, trades 0.236 0.148 -0.291 -0.279 or vocational (0.184) (0.174) (0.263) (0.271) diploma Bachelor degree 0.383 * 0.290 -0.279 * -0.314 (0.156) (0.150) (0.244) (0.250) Master's or 1.240 *** 1.126 *** 0.922 * 0.931 *** Doctorate (0.177) (0.171) (0.266) (0.259) degree Undergraduate Student Loans No government- -- -- -- -- sponsored loans Loans of -0.112 -0.089 -0.002 -0.015 $15,000 or (0.127) (0.127) (0.181) (0.169) less Loans above -0.229 -0.195 0.084 0.200 $15,000 (0.201) (0.196) (0.298) (0.289) Marital Status * ** * Single / -- -- -- previously married Married/ -0.062 -0.083 -0.057 common-law (0.131) (0.212) (0.224) Children No dependent -- -- -- child Dependent child -0.217 -0.047 -0.006 (0.232) (0.426) (0.437) Log(age) -0.569 -0.722 -0.769 (0.559) (0.576) (0.580) Gender * Male -- -- -- Female -0.364 -0.373 -0.400 (0.128) (0.211) (0.213) Race Not visible -- -- -- minority Visible -0.203 -0.052 0.035 minority (0.221) (0.272) (0.289) Undergraduate *** *** scholarships No achievement- -- -- based scholarships Scholarships of -0.229 -0.021 $5,000 or (0.216) ***(0.218) *** less Scholarships 1.082 *** 1.188 *** above $5,000 (0.215) (0.238) Aspirations No Master's -- -- degree plans Master's degree 1.501 ** 1.508 ** plans (0.532) (0.534) Don't know 5.394 *** 5.481 *** (0.509) (0.529) Academic *** *** Confidence Below top 25 -- -- percent Below top 10 0.659 0.601 percent but (0.345) (0.347) in top 25 percent Top 10 percent 0.903 ** 0.835 * (0.331) (0.327) Type of Undergraduate Education Not a -- co-operative education program Co-operative -0.091 education (0.395) *** program Undergraduate Field of Study Arts and -- Humanities Social and -0.048 Behavioral (0.298) sciences Business and -0.748 Management (0.404) Physical and -0.194 Life Sciences (0.231) Engineering, -0.371 Mathematics, and Computer Science (0.310) Other -1.277 (0.269) Satisfaction with Under graduate Field of Study Would not -- select the same field Would select the 0.423 same field (0.207) Region of Undergraduate Institution Atlantic Canada -- Quebec 1.126 (0.207) Ontario 0.434 (0.268) Western -0.128 Provinces and Territories (0.200) n 7,325 7,325 7,325 7,325 Log likelihood -52,488.6 -51,460.4 -28,526.9 -27,161.8 Pseudo 0.0345 0.0534 0.4753 0.5004 [R.sup.2] Deviance 104977.3 102920.8 57053.77 54323.64 No post-undergraduate education is the reference category for the response variable. Multiple parameter likelihood ratio chi-square tests are reported for sets of dummy regressors. Robust standard errors are in parentheses. * p<.05; ** p<.01; *** p < .001.