Professional Documents
Culture Documents
This Content Downloaded From 201.150.98.4 On Mon, 11 Jul 2022 20:40:15 UTC
This Content Downloaded From 201.150.98.4 On Mon, 11 Jul 2022 20:40:15 UTC
disadvantage
Author(s): George Frempong, Xin Ma and Joseph Mensah
Source: Higher Education , January 2012, Vol. 63, No. 1 (January 2012), pp. 19-32
Published by: Springer
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/41343604?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Springer is collaborating with JSTOR to digitize, preserve and extend access to Higher
Education
Abstract While access to postsecondary education in Canada has increased over the past
decade, a number of recent studies demonstrate that youth from disadvantaged socioeco-
nomic backgrounds are vulnerable to some degree of exclusion from postsecondary edu-
cation. These studies tend to emphasize the lack of financial resources and social capital as
the main sources of this vulnerability. Our paper employs multilevel framework to explore
the extent of the impact of schools on access to postsecondary education, especially for
youth from disadvantaged background. Our analyses revealed that: (1) for youth with
similar financial constraints who attend schools with relatively similar quality, those from
disadvantaged socioeconomic backgrounds who attend schools with high concentration of
low SES students are particularly vulnerable to exclusion from university education, and
(2) a substantial portion of the SES effect operate through the impact of high school
academic achievement and postsecondary education expectation on access to postsec-
ondary education.
Introduction
'Will my child go to university? And can I afford it?' These are important questions for
most families across Canada, as postsecondary education becomes increasingly critical in
determining individuals' career (and, consequently, socio-economic) success. While access
G. Frempong (И)
Human Science and Research Council (HSRC), Pretoria, South Africa
e-mail: gfrempong@hsrc.ac. za
X. Ma
University of Kentucky, Lexington, US
J. Mensah
York University, Toronto, Canada
â Springer
topostsecondary education in C
studiesdemonstrate that youth f
some level of exclusion in our p
emphasize the cost of tuition an
nerability. Research involving a
assumption that financial constr
education, and that, loans, grants
to the problem.
Over the past three decades, st
for the US - point to the import
comes. In a review of studies in
Wolfe (1995) demonstrated that
related outcomes well before th
studies, including Frempong an
(1999), lend some support to th
status (SES) of students and the
With the current emphasis on qu
call for more research on vuln
backgrounds. The important qu
advantage? Our paper employs
Survey (YITS) to explore the ext
education, especially for youth
specifically, the paper addresses t
of the impact of SES on access
quality of secondary schools aff
Our analysis deals mainly with u
more difficult for youth from l
constitutional responsibility for
the years differences in the pos
students attend postsecondary
secondary education. However,
"CEGEP", or Collège d'enseigne
education college) after Grade 1
vocational or pre-university co
university and 3 years for most
traditional college attendance in
education in Canada that includ
Literature review
Ö Springer
Using data from YITs, Frenette (2007) employed an econometric framework that
decomposed the gap in university access associated with family income into an explained
and unexplained components in an attempt to identify the most important characteristics
that accounts for the gap. His analysis indicated that financial constraint accounted for only
12% of the gap while "factors such as standardized test scores in reading obtained at age
15, school marks reported at age 15, parental influences, and high-school quality account
for 84% of the gap". He used his findings to suggest a shift from the financial constraint
argument to focus on cognitives abilities and schooling of youth from disadvantaged
socioeconomic backgrounds. We contribute to this line of research by employing more
complex statistical models (multilevel models) that allow us to focus on the SES effect on
access within and between schools and address the question - can schools compensate for
socioeconomic disadvantage in access to university education? We hope that our analysis
Springer
Sample
Data for this analysis came from the Canadian sample in the Programme for International
Student Assessment (PISA) in 2000. The Canadian sample was also linked with the Youth
in Transition Survey (YITS), a national longitudinal survey designed to collect information
about major transitions in young people's lives, particularly those involving education,
training, and work. In 2000, a total of 29,687 students at the age of 15 took part in both
PISA and YITS in Canada. These students were selected through a stratified random
sampling procedure from 1,117 schools across the 10 provinces (our analysis excluded
very small schools with fewer than 15 students). Later on, the same sample of students
participated in 2 more cycles of YITS surveys in 2002 and 2004. The current analysis drew
data mainly from student and school questionnaires used in PISA and YITS to obtain
information describing students' access to postsecondary education as well as student,
family, and school characteristics. Students' scores on reading literacy as measured in
PISA were also used in the data analysis.
Measures
Our dependent variable depicted students' access to postsecondary education. The survey
interviewed students in April or May 2000, followed with a second interview in February to
May, 2002, and a third interview in Febrary to June 2004. Our analysis focuses on students
who were Grade 10 on December 31, 1999 (most students in this grade were 15 years old)
and got a high-school diploma on December, 31, 2003. Our dependent variable is university
participation by December, 2003. This dependent measure was coded as a dummy variable
denoting access to postsecondary education compared with lack of access to postsecondary
education. Our independent variables described student and family characteristics (i.e.,
student-level variables) as well as school characteristics (i.e., school-level variables). These
variables are presented in Tables 1 and 2. Table 1 shows that a total of 9 student-level
variables were selected, including SES (socioeconomic status), minority status, family type,
family size, self-esteem, postsecondary education expectation, reading literacy, high school
GPA (grade point average), and financial barrier expectation. In the form of scales, some
variables were continuous (SES, family size, self-esteem, and reading literacy); whereas
other variables were categorical (minority status, family type, postsecondary education
expectation, high school GPA, and financial barrier expectation). The SES is a composite
measure of parental occupational prestige (based on students' response to their parental
occupation), parental education level (coded as years of schooling), and family wealth which
was derived from students' reports on: (a) the availability, in their home, of a dishwasher, a
room of their own, educational software, and a link to the Internet, (b) the number of cellular
phones, television sets, computers, motor cars and bathrooms at home. A standardized
(mean = 0, standard deviation = 1) measure of SES was used in all the analysis.
Categorical variables were dummy coded for data analysis. Minority status compared
minority students with non-minority students (in a population sense). Family type com-
pared students of dual parents with students of single or no parents. Postsecondary
Ô Springer
803
803
â Springer
Variables Models
(Logit coefficienst w
12 3 4
Socioeconomic status (SES) 0.58( 1 .79) 0.59( 1 .80) 0.59( 1 .80) 0.27( 1.31)
Family type (two biological parents vs. other) 0.25(1.28)
Minority 0.45(1.57)
Number of siblings -0.06(0.94)
University education expectation
Self 2.11(8.24)
Parent 0.81(2.25)
Self esteem 0.05(1.05)
Reading literacy 0.0 1 ( 1 .0 1 )
High school GPA (in percentage)
90 versus 60 2.73(15.3)
80 versus 60 2.1 1(8.24)
70 versus 60 1.07(2.91)
Financial barrier expectation -0.36(0.70)
SES of school (MeanSES) 0.32( 1 .38) 0.57( 1 .77) 0.57( 1 .77) 0.3 1 ( 1 .36)
Private school 0.27(1.31) 0.17*(1.19) 0.10*(1.1 1)
Urban school 0.07*(1.07) 0.08*(1.08) 0.03*(1.03)
Province
â Springer
Analysis
Multilevel analysis was chosen as the primary statistical technique in the current analysis
because the PISA and YITS data are multilevel in nature. The data exist at the levels of
students (including families) and schools (including provinces) (i.e., students nested within
schools). This data hierarchy must be taken into account in any statistical analysis (Rau-
denbush and Bryk 2002). Therefore, our multilevel models included two levels, students at
the first level and schools at the second level. Because the dependent measure is dichot-
omous, our two-level models took the form of multilevel logistic models. The first-level
model could be expressed as:
prob(event) = ^ ^ z
At this first level or student level, our focus was on examining the effects of student and
family characteristics Xpij on the likelihood of students' access to postsecondary education.
The intercept ß 0j estimated the average likelihood that students within a school gained
access to postsecondary education, and this school average likelihood could then be
modeled by school characteristics at the second (school) level:
Springer
that estimated the effects of student-level and school-level variables on students' access to
postsecondary education, i/0j was important in the current analysis because as a measure of
the variation in students' access to postsecondary education among schools it indicated
whether schools could make a difference in promoting or preparing students to gain access
to postsecondary education.
Results
A series of multilevel models were performed to examine the likelihood of students' access
to postsecondary education under different conditions or adjustments. We started modeling
with what is often referred to as the "null" model that contained no variables at either
student or school level. The null model was mainly used to estimate the variation (vari-
ance) in the likelihood that students coming from different schools gained access to
postsecondary education. We found that this variation was statistically significant at the
school level (see Table 4), indicating that students from different schools demonstrated
different likelihood of access to postsecondary education. This finding is a good indication
that schools made a difference in promoting or preparing students to gain access to
postsecondary education.
We present the findings of the rest of our models in Table 3. Model 1 introduced
socioeconomic measures at the student and school levels into the null model to build what
is often referred to as the "contextual" model that examines the impact of family and
school socioeconomic context on a dependent measure. Results of this model showed that
students with high SES were more likely to gain access to postsecondary education than
students with low SES. Consider two students with their SES one standard deviation apart,
the one with higher SES was 1 .79 times more likely to access postsecondary education than
the one with lower SES. In the multilevel framework, a coefficient of a variable such as
SES can be allowed to vary at the school level to determine in our case whether the effects
of SES varied across schools. When we allowed the impact of SES to vary across schools,
we found that the variance of this impact was not statistically significant. This means that
the impact of SES on students' access to postsecondary education is relatively the same
across schools.
Model 1 also showed that students attending schools with high average SES were more
likely to gain access to postsecondary education than students attending schools with low
average SES. Specifically, consider two schools with their school mean SES one standard
deviation apart. Students from the school with higher average SES was 1.38 times more
likely to access postsecondary education than students from the school with lower average
Models
0 12 3 4
Springer
We would like to come back to the statistically significant variance in the likelihood of
students' access to postsecondary education across schools. Again, students in some
schools showed significantly better likelihood of access to postsecondary education than
students in other schools. Even after student-level and school-level socioeconomic back-
ground was adjusted, variation in the likelihood of students' access to postsecondary
education was still statistically significant among schools. The importance of schools in
promoting or preparing students to gain access to postsecondary education, therefore,
remained. To foreshow, this conclusion was sustained throughout all models tested in the
current analysis. This consistency represented a strong argument that schools did make a
difference in students' access to postsecondary education beyond the impact of student-
level and school-level characteristics (or even after adjustment over student-level and
school-level characteristics). To save space, the importance of schools was not repeatedly
reported in the upcoming interpretation of various models.
Model 2 aimed to examine provincial variation in the likelihood of students' access to
postsecondary education over and above student-level and school-level socioeconomic
impacts. Results of this model indicated that provinces indeed differed in preparing their
students to gain access to postsecondary education. Recall that Ontario was used as the
base-line category against which other provinces were compared. We found that students
in Newfoundland schools were 1.90 times more likely to gain access to postsecondary
education than students in Ontario schools, and students in Prince Edward Island schools
were 2.16 times more likely to gain access to postsecondary education than students in
Ontario schools. Furthermore, students in both Nova Scotia and New Brunswick schools
were 2.29 times more likely to gain access to postsecondary education than students in
Ontario schools.
There were also smaller provincial differences in the likelihood of students' access to
postsecondary education. Students in Manitoba schools were 1.51 times more likely to gain
access to postsecondary education than students in Ontario schools, and students in British
Columbia schools were 1 .09 times more likely to gain access to postsecondary education
than students in Ontario schools.
On the other hand, students in Ontario schools were more likely to gain access to
postsecondary education than students in Quebec and Alberta schools. For convenience of
interpretation, throughout this report we interpreted its reciprocal when an odds ratio was
smaller than 1 . With this practice, we found that students in Ontario schools were 4.5 times
more likely to gain access to postsecondary education than students in Quebec schools, and
students in Ontario schools were 1.7 times more likely to gain access to postsecondary
education than students in Alberta schools. Finally, students in Saskatchewan and Ontario
schools were equally likely to gain access to postsecondary education.
We note that these provincial differences remained even in the presence of school type
(private vs. public) and school location (urban vs. rural). In fact, school location did not
have any statistically significant impact on the likelihood of students' access to postsec-
ondary education. But, students attending private schools were 1.31 times more likely to
gain access to postsecondary education than students attending public schools.
â Springer
Of
course, provincial differences
reflect the differences in the oppo
provinces. We note that postseco
individual provinces providing t
alternatives to formal postseconda
dents attending university. If ma
studies may consider a three-level
provinces. Province-level variabl
provinces in students' access to p
present analysis.
Model 2, as one may notice, emph
added school climatic background
climatic variables showed statisti
access to postsecondary education
standard deviation apart, students a
1 .25 times more likely to gain acc
the school with poorer disciplinary
pressure one standard deviation apa
pressure were 1.25 times more l
students attending the school with
apart between two schools in stud
with better student-teacher relat
postsecondary education than stud
relationship.
Other school-level variables (teacher educational support, students' sense of belonging
to school, school material resources, school instructional resources, student-teacher ratio,
student behaviors, and teacher participation in school decision making) showed no sta-
tistically significant effects on the likelihood of students' access to postsecondary educa-
tion. In the presence of statistically significant school climatic variables, only school mean
SES remained as a statistically significant school contextual variable. Most provincial
differences remained except that students in British Columbia and Ontario schools became
no different in the likelihood of access to postsecondary education once school climatic
variables were added to school contextual variables.
Model 4 was what is often referred to as the "full" model that contained variables at
both student and school levels. Specifically, the full model in the current analysis intro-
duced a wide range of student-level variables in addition to SES (already in the model) to
examine whether student and family characteristics (student-level variables) and school
characteristics (school-level variables) simultaneously affected the likelihood of students'
access to postsecondary education. All student-level variables showed statistically sig-
nificant effects on the likelihood of students' access to postsecondary education. The SES
effect reduced from an odd ratio of 1.80-1.31. We also found that students with dual
parents were 1 .28 times more likely to gain access to postsecondary education than stu-
dents with other forms of parental conditions. From a population perspective, minority
students were 1 .57 times more likely to gain access to postsecondary education than non-
minority students. Students with one fewer sibling were 1.06 times more likely to gain
access to postsecondary education than students with one more sibling.
Postsecondary education expectation demonstrated substantial effects on the likelihood
of students' access to postsecondary education, with student expectation showing con-
siderably stronger effects than parent expectation. Students who expected themselves to go
^ Springer
Ö Springer
â Springer
Policy implications
Ö Springer
References
Card, D. (2001). Estimating the return to schooling: Progress on some econometric problems. Econometrica,
69(5), 1127-1160.
Carneiro, P., & Heckman, J. J. (2002). The evidence on credit constraint in post-secondary schooling. The
Economic Journal, 772(482), 705-734.
Coleman, J. S. (1966). Equality of educational opportunity. Washington, DC: US Dept. of Health, Edu-
cation, and Welfare, Office of Education.
Eccles, J. S., Adler, T. F., Futterman, R., Goff, S. В., Kaczala, С. M., Меесе, J. L., et al. (1983). Expec-
tancies, values and academic behaviors. In J. T. Spence (Ed.), Achievement and achievement motives.
San Francisco: W. H. Freemen.
Ellwood, D. T., & Kane, J. L. (2000). Who is getting college education? Family background and the
growing gap in enrollment. In S. Danziger & J. Waldfogell (Eds.), Securing the future: Investing in
children from birth to college. New York: Russell Sage Foundation.
Finnie, R., & Mueller, R. (2009). The backgrounds of Canadian youth and access to post-secondary edu-
cation: New evidence from the youth in transition survey. In R. Finnie, R. Mueller, A. Sweetman, &
A. Usher (Eds.), Who goes? Who stays? What matters? Access to and remaining in post-secondary
education in Canada. McGill-Queen's University Press: Montreal-Kingston.
Finnie, R., Eric, L., & Arthur, S. (2005). Who goes? The direct and indirect effects of family background on
access to post-secondary education. Analytical studies branch research paper series. Catalogue
no. 1 1F0019MIE No. 237. Ottawa: Statistics Canada.
Frempong, G., & Willms, J. D. (2002). Can school quality compensate for socioeconomic disadvantage? In
J. D. Willms (Ed.), Vulnerable children: Findings from Canada's longitudinal study of children and
youth (pp. 277-303). Edmonton, AB: University of Alberta Press.
Frenette, M. (2007). Why are Youth from lower-income families less likely to attend Universty? Evidence
from academic abilities, parental influences, and financial constraints. Analytical studies branch
research paper series. Catalogue no. 11F0019 No. 295. Ottawa: Statistics Canada.
Haveman, R., & Wolfe, B. (1995). The determinants of children's attainments: A review of methods and
findings. Journal of Economic Literature, 33 , 1829-1878.
Human Resources Development Canada (1998). High school may not be enough , Human Resources
Development Canada catalogue no. SP-105-05-98E, Ottawa.
Junor, S., & Usher, A. (2004). The price of knowledge: Access and student finance in Canada Millenium
research series. Ottawa: Canada Millenium Scholarship Foundation.
Kane, J. T. (1994). College entry by Blacks since 1970: The role of college cost, family background, and the
return to education. Journal of Political Economy, 702(5), 878-91 1.
Ma, X., & Klinger, D. (2000). Hierachical linear modelling of student and school effects on academic
achievement. Canadian Journal of Education, 2(1), 41-55.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models : Applications and data analysis
methods (2nd Ed.). Newbury Park, CA: Sage.
Willms, J. D. (1999). Quality and inequality in children's literacy: The effects of families, schools, and
communities. In D. Keating & C. Hertzman (Eds.), Developmental health and the wealth of nations:
Social, biological, and educational dynamics (pp. 72-93). New York: Guilford Press.
Wilson-Relyea, B. J. (1997). Influences on the level of mathematics achieved by female adolescents: A test
of a model of academic choice. Unpublished doctoral dissertation, University of Memphis. Disser-
tation Abstracts International, 58, 1 194.
â Springer