Social Selectivity in Higher Education: A Case Study of Canada and the Czech Republic

Sean Hellingman1, Petr Mazouch2, Kristýna Vltavská2, Samuel Ryckenberg2
1 Wilfrid Laurier University, 2 Prague University of Economics and Business
Abstract
This work examines the choices of individuals with respect to higher education in
the Czech Republic and Canada. Specifically, how do the students’ socioeconomic
backgrounds influence their study decisions. Data from the Czech edition of The
European Union Statistics on Income and Living Conditions survey and the Canadian
Survey of Labour and Income Dynamics was used to identify influences of students’
entry to university. Individuals from households with higher socioeconomic status
were more likely to enter university than people who were less well off in both
countries. Social selectivity is much more present in higher education in the Czech
Republic than in Canada.

Keywords: Higher Education, Czech Republic, Canada, Social Selectivity, Logistic
Regression, Ordered Statistics

Social Selectivity in Higher Education
The analysis of education is crucial, as higher education increases the potential for economic
growth and an improvement in the quality of life for members of society. Education should be
accessible to any member of a society with the ability and motivation to participate. Despite this,
historically it has been proven that this is not the case (Belley et al., 2011; Chevalier et al., 2013;
Frenette, 2007; Savas, 2016). Families with lower educational and financial status are at risk of
social exclusion in education. In other words, they are at risk of being victims of social selectivity
in higher education. In most cases, the upbringing of individuals, specifically the actions of parents,
have a profound impact on the way children live their lives (Frenette, 2007). A lot of these parental
influences cannot be quantified in the sense of a cause and effect on their children’s decisions,
specifically when discussing education. Fortunately, key variables like household income and
parental education can be quantified and used in analysis. Considering this, the impact of a
predisposition of certain family members to study will be compared across two different systems
of higher education from the Czech Republic and Canada. Other research, that will be outlined in
the literature review, has proven that the socioeconomic background of students has an influence,
however, the strength of this impact is debated. Furthermore, much of the research has focused on

the United States or comparing the United States with Canada (Belley et al., 2011; Frenette, 2007;
Savas, 2016).
The direct comparison of two nations, one with tuition fees and the other without, motivates
the selection of the Czech Republic and Canada. The presence of tuition fees in Canada may reduce
accessibility to higher education for individuals from families with lower incomes. This research
will examine the accessibility of higher education for people from different socioeconomic
backgrounds in Canada and the Czech Republic. Historically, the presence of social selectivity in
education has been proven in Canada and the United States; but this phenomenon hasn’t been
deeply examined in the seemingly more egalitarian, tuition free, Czech Republic. Despite this,
there are definitely some signals that higher education is influenced by social status of family and
as the social status is connected with education level of parents, there is also correlation with this
which was observed in the survey EUROSTUDENT V and VI (Fischer & Vltavská, 2017).
Data from the Czech EU-Statistics on Income and Living Conditions (EU-SILC) from 2009 to
20151 and the 2011 Canadian Survey of Labour and Income Dynamics (SLID)2 surveys contain
information about the presence of a student in the household and the level, and if they were
studying at the secondary/high school or university level. This newer Canadian data will provide
insight into the direction of social selectivity in Canadian higher education. Furthermore,
information about other members of the household is also quantified in the data sets. The education
attainment levels of members of the household, material deprivation, and income variables are
found in both sets of data. These key comparable socioeconomic variables provide a context for
comparability and overall selection of these datasets. Binary logistic regression estimated using
the maximum likelihood method will be applied in the case of dichotomous explanatory variables
to estimate the significance and influence of explanatory variables included in the estimated model.
Using Binary logistic regression and comparative statistical tests, the data from both nations
will be analyzed. The existence of social selectivity with regards to entry to university will be
evaluated and compared to the pre-existing literature. Furthermore, conclusions from this analysis
about the similarities and differences in the impacts of the variables from both nations will be
drawn. The conclusions drawn from this analysis will be conducted within the framework of
previous research. Comments will also be made addressing the existence of tuition as a potential
catalyst for social exclusion in higher education. Ideas regarding the causes and potential impacts
of the results will also be addressed and explored. Construction of analysis such as this has never

1

been completed between the two nations in this field; the results obtained here could provide a
platform for further comparisons of this nature. Using newer Canadian data and different variables
than previous research, this work contributes valuable insight about the Czech Republic and
Canada to the conversation about social selectivity in higher education.
Literature Review
Human capital is defined as an investment in school, training, or medical care with the purpose
of adding to the person’s knowledge or wellbeing (Becker, 2009). Furthermore, Becker (2009)
argued that education and training are the most important investments in human capital as higher
education leads to higher average income. Increased education has a positive influence on earning
potential in the labour market for individuals from The Organization for Economic Co-operation
and Development (OECD) countries (OECD, 2020a.). Research into the attendance of tertiary
education is being done for many countries throughout the world. Tertiary education, mainly
university, refers to all education classifications above “Post-secondary non-tertiary education”,
including: “Short-cycle tertiary education”, “Bachelor’s, Master’s, Doctoral studies, or their
equivalents” (Eurostat, 2016). The attainment levels of tertiary education in many countries have
been increasing in recent history as measured by OECD (OECD, 2020b). The OECD is an
organization made up of 35 countries including Canada and the Czech Republic; “to promote
policies that will improve the economic and social well-being of people around the world” (OECD,
2020a). Therefore, an increase in education and potential earnings are good examples of measures
available to improve the socioeconomic wellbeing of an individual. With this in mind, it seems
intuitive that students would wish to venture beyond basic secondary school education. Identifying
factors of influence that may limit the accessibility of higher education to students could be a good
first step towards ensuring equal access to students from all backgrounds, leading to more
opportunities to improve the overall socioeconomic wellbeing of individuals.
Accessibility of Higher Education
Even though education is a valuable tool for improving socioeconomic status, the accessibility
of higher education appears to be unequally distributed across socioeconomic classes (Belley et
al., 2011). According to regression models constructed for data from Canada, it was found that
students whose parents’ income is in the highest quartile are 20 percentage points more likely to
go to university than the students who come from families in the lowest quartile of income (Belley
et al., 2011). The results of social selectivity were even stronger when examining data from the
2

United States. Students whose parents’ income is in the highest quartile are 45 percentage points
more likely to go to university than the lowest quartile (Belley et al., 2011). In other words, students
from poorer backgrounds are the least likely to attend university in Canada and the United States
(Belley et al., 2011). Accordingly, similar results were obtained when examining students in the
United Kingdom (Chevalier et al., 2013). These results were obtained through the construction of
regression models using the parental income and parental education of the students as explanatory
variables. Both variables proved to be significant in the models where male students with
uneducated parents were most likely to drop out of school in the United Kingdom (Chevalier et
al., 2013). In these OECD nations the parental income of students seems to influence their decision
to attend university. These results do not bode well for young people from lower income families
who would like to surpass their parents’ labour market earnings, as an increase in education
attainment generally leads to an increase in labour market earnings (OECD, 2010).
From a research and a policy perspective it is important to continue to monitor the impacts of
socioeconomic class on accessibility to higher education. For example, the data used for the
analysis on Canadian students by Belley et al. (2011) was created in 2000. Though providing
important insight into the restriction at the time, it is important to examine potential changes since
then.
Prokop (2020) found that in the Czech Republic there is an intergenerational transmission of
poverty which is visible in the proportion of children attending university. Prokop (2020)
compared the proportion of these at-risk children with the levels of education of their parents.
Children (up to age 30) from parents with an education level below post-secondary non-tertiary
education (level 4, ISCED classification) are two times more likely to be at risk of poverty than
children with higher educated parents (Prokop, 2020). One third of these children are in the poorest
10% of society and have no savings of their own (Prokop, 2020). Furthermore, in the Czech
Republic, it was shown that few children with parents that have low levels of education attended
university (Prokop, 2020).
This cross section of literature strongly suggests that an individual’s socioeconomic
background directly impacts their entry into higher education. Understanding this impact and
identifying ways of eliminating it are key to creating an equally accessible higher education
system.

3

Canada and the Czech Republic
Beyond the comparison of two OECD countries with and without tuition fees there are several key
similarities and differences between Canada and the Czech Republic. When focusing on higher
education, examining the age range of 25–34-year-olds, both nations’ educational attainment levels
are higher than the OECD average for individuals with education higher than education level 2
(below upper secondary) (OECD, 2020b). Moreover, 39% of the 25–34-year-old age range in
Canada have completed a bachelor’s, master’s, or doctoral degree and 33% of this age range in the
Czech Republic have completed a bachelor’s, master’s, or doctoral degree which are both below
the OECD average of 40% (OECD, 2020b). In other words, both nations on average have higher
secondary school attainment levels than the OECD average but lower university education (levels
6,7 and 8) attainment than the OECD average.
Notably, most individuals with tertiary education in Canada in the 25–34-year-old age range
have a bachelor’s degree where the majority of the 25–34-year-old individuals with tertiary
education in the Czech Republic have a master’s degree (OECD, 2020b). Furthermore, over 20%
of individuals with tertiary education in Canada in the 25–34-year-old age range have completed
short-cycle tertiary education where less than 1% of their Czech counterparts have this level of
educational attainment (OECD, 2020b). There is direct comparability in the education attainment
levels of the two countries.
Tertiary education appears to be more advantageous in the Czech Republic than in Canada. In
the Czech Republic the average increase in relative earnings by obtaining tertiary education is 58%
(OECD, 2020b). In Canada the average increase in relative earnings by obtaining tertiary education
is 39% in Canada (OECD, 2020b). Although the financial returns of higher education seem to
differ between the nations, the attainment levels are similar.
Educational Costs
Equal opportunity in education can be achieved if certain barriers are removed. One of the
largest barriers to entry is monetary and without external support the poorer students may be unable
to afford higher education. Results of a study conducted on British data indicate that family
financial constraints influence students’ choice of university (Callender & Jackson, 2008). In many
OECD nations tuition is not constant; in other words, tuition may differ between universities and
across study fields (OECD, 2010). Intuitively, this means that universities of a perceived higher
standard with sought after programs may cost more than their less prestigious counterparts.
4

Therefore, people from more modest backgrounds may not have the possibility or may be more
hesitant to spend large amounts of money on tertiary education or other educational resources.
Student performance during compulsory school also influences a student’s decision to attend
post-secondary education (Frenette, 2007). Examining students at 15 years of age and then the
same students at 19 years of age, it was found that students from lower income families had lower
standardized test scores and were less likely to go to university; supporting the argument that the
socioeconomic backgrounds of students have influences very early on in their academics (Frenette,
2007). Using logistic regression on data from the United States, it was shown that the
socioeconomic background of the students and their grade school results influence their entry to
university (Savas, 2016). Specifically, students with lower grade school results and with a lower
socioeconomic status were less likely to enter university (Savas, 2016).
Conversely, other work has shown that the financial constraints of a family do not have a strong
influence on the ability of the student to afford university; however, it does limit their access to
academic support earlier in life (Frenette, 2007). This academic support comes in the form of
museum trips, additional books, and even the ability to move to a neighbourhood with better
schools (Frenette, 2007). Before needing to overcome the possible burden of tuition costs it appears
that many students are already at a disadvantage due to their socioeconomic backgrounds. These
findings suggest that the presence of tuition is not the primary driving factor in limited access for
certain individuals to higher education.
Costs associated with attending university include housing and general living expenses,
transportation, books, and tuition in some countries. The average tuition cost in Canada for
undergraduate students in the 2015-2016 academic year was 7,049 Canadian Dollars (CAD;
Statistics Canada, 2020). Whereas, the average cost of tuition in the United States for the 2014–
2015 academic year was 7,792 American Dollars (USD) for public universities and 24,928 USD
for private universities (United States Government, 2020). Unassisted, these tuition fees represent
an additional cost to higher education that does not exist in countries that do not charge tuition.
Increases in tuition led to a decline in enrollment at the university level in the United States (Hemelt
& Marcotte, 2008). More specifically, regression models were estimated quantifying that an
increase of tuition by 100 USD forces a decline in enrollment of around 0.25% (Hemelt &
Marcotte, 2008). It was also observed that the impact of tuition increase was more noticeable on
the attendance of top research universities and less noticeable on the liberal arts and comprehensive

5

universities; the tuition increases forced students to rely more on financial aid (Hemelt & Marcotte,
2008). In Canada, the effects of tuition were examined between Québec and Ontario due to a tuition
freeze put in place in 2007 in Québec. (Bastien et al., 2014). With a 1,000 CAD increase in tuition,
there is little to no decrease in overall enrollment (Bastien et al., 2014). Further manipulation of
the model yielded that an increase in tuition of 1,000 CAD causes a decrease of 19% in enrollment
of students whose parents had no postsecondary education and a 7% decrease for students with
parental postsecondary education/training but no university education (without tertiary
education;Bastien et al., 2014). Moreover, the enrollment increased by 10% for children of parents
with tertiary education thus making the argument that both tuition and parental education influence
the decision to study at university (Bastien et al., 2014). As a consequence, the poorest families
are the hardest hit.
There are countries within the OECD that do not charge tuition at their public universities;
some of these countries include: The Czech Republic, Denmark, Finland, and Sweden (OECD,
2010). In 2007, seven out of sixteen German states introduced tuition fees (Achelpöhler et al.,
2007). This was used as an opportunity to examine the effects of the implementation tuition fees
and enrollment rates using the assumption that students prefer to attend university in the state in
which they graduated secondary school (Hübner, 2012). As one might expect, the introduction of
tuition fees in these German states led to a reduced probability of student enrollment for the states
charging tuition (Hübner, 2012). This law has since been removed and Germany is a tuition-free
nation for students at the bachelor’s level (Veselinovic, 2014). Other research indicates that tuition
does not play a significant role in a student’s decision to enroll. Time series data from the
Netherlands was used to show this to be the case, as the students viewed the cost as an important
investment in human capital (Canton & De Jong, 2005). Moreover, the existence of financial aid
for students in the Netherlands has a positive relationship with enrollment rates and may negate
any adverse effects of tuition on enrollment (Canton & De Jong, 2005). The cost of tuition is just
one of the costs facing students when they choose to attend university.
This work adds to the narrative surrounding tuition by comparing one country that charges
tuition for public education with one that does not. Given the mixed results surrounding the impacts
of tuition, developing a better understanding of this phenomenon is required.

6

Gender
Gender is another aspect that influences students’ tertiary education attendance. Female
students are more likely to attend tertiary education than their male counterparts (OECD, 2010).
This indicates that gender is an important factor for the decision of students to attend university in
OECD countries. In contrast, this was not always the case; in 1947 undergraduate men
outnumbered women 2.3 to 1 in the United States (Goldin et al., 2006). By the 1980s parity had
been reached and by the early 2000s the enrollment ratio was 1.3 female students for every 1 male
student (Goldin et al., 2006). The increase in labour force participation by females throughout
recent history has paved the way for a higher share of females in higher education than males
(Goldin et al., 2006). Furthermore, female students outperform their male counterparts on average
during their secondary school studies, based on United States data; allowing female students to
have an easier time with university as a whole (Goldin et al., 2006). Accordingly, females are
almost twice as likely to go to university than males in the United States (Savas, 2016). Although
gender is not a key focus of this work, it is important to understand as many influential factors
regarding the entry into higher education as possible.
Objective
This work set out to examine the existence of social selectivity in higher education in the
Czech Republic and Canada. Using the newer data, a more up-to-date picture of the social
selectivity in Canada will be drawn. Further intentions focused on identifying new influential
variables not discussed in previous literature, examining the presence of tuition and making new
comparisons between two countries with different educational systems. The existence of social
selectivity in higher education has been shown in previous literature. What remains in question is
the strength of the effects of certain socioeconomic variables such as parental education and
parental income. Initially, the expectation that entry to university is partially driven by the
socioeconomic background of students will be tested. If found significant, the magnitudes of these
influences will be compared across the two nations. As a result of the literature mentioned above
the following hypothesis will be evaluated: Entry into university is influenced by students’
socioeconomic backgrounds. Furthermore, if social selectivity appears to be more present in
Canada than the Czech Republic the argument for reducing or eliminating tuition as a method of
increasing accessibility to higher education can be made.

7

Data
The two datasets examined come from the Czech Republic and Canada. The Czech data
comes from the European Union—Statistics on Income and Living Conditions survey (EU-SILC).
The Canadian data comes from the 2011 Survey of Labour and Income Dynamics (SLID) public
use microdata file. The EU-SILC data is collected yearly as a joint effort between Eurostat and the
National Statistical Institutes of the European Union, European Free Trade Association, and
candidate countries (Mack, 2016). The main objective of the EU-SILC is to provide data for people
at risk of poverty or social exclusion, income distribution and monetary poverty, living conditions
and material deprivation (Eurostat, 2013b). The Canadian SLID is an annual voluntary survey
covering the population of the 10 Canadian provinces with the exception of Native reservations,
residents of institutions, and military barracks (Statistics Canada, 2011). The SLID was designed
to be, first and foremost, a longitudinal survey, with primary focus on labour and income and the
relationships between them and family composition (Statistics Canada, 2011). Then, the decision
was made to extend the objectives of SLID to be the primary source of cross-sectional household
income data (Statistics Canada, 2011).
EU-SILC
EU-SILC survey data collected during the years 2009 to 2015 from the Czech Republic
were compiled into one larger dataset. The data collected over this time period were reduced to
only households containing a student in secondary school or university. This reduced dataset now
included as many as four observations for each household as the EU-SILC survey is a panel survey
where each respondent is interviewed for four consecutive years (Eurostat, 2013b). This was
corrected by including only the most recent observations where the student still remained in the
household; in other words, repeated households were removed keeping only one observation for
that particular household. Without removing all but one of the repeated households, the regression
results could have become biased due to the non-independence of the sample. This data reduction
rendered 3748 usable observations with 1778 households containing at least one student only in
secondary school and 1970 households containing at least one student in university.
Canadian SLID
The sample size of this dataset is split into four categories: person file, economic family, census
family, and key files (Statistics Canada, 2011). The data used in this analysis focused primarily on
the economic family with some variables from the person file which will be explained in detail
8

further on. This economic family file contained 25109 observations (Statistics Canada, 2011).
Furthermore, the Canadian data is also weighted to reflect the structure of the survey population
(Statistics Canada, 2011). Further deconstruction of this data was conducted to only include
households with students in secondary school or university in the same way as the Czech EU-SILC
data as this research is only focused on access to higher education.
The SLID 2011 data is from one year and includes the 10 Canadian provinces and excludes the
population that lives in the territories (Statistics Canada, 2011). This data is a public use microdata
file and includes less detail and fewer observations than the complete original dataset (Statistics
Canada, 2011). The educational status of each respondent was used to identify households with
individuals who studied during the reference year. This educational variable was used to create an
indicator variable containing two options: Households with students only in secondary school and
households with at least one student in university. As mentioned earlier, one should note that all
Native reservations, residents of institutions, military barracks, and Canadian territories are not
included, resulting in estimates that only reflect the survey population. This data does contain many
socioeconomic variables such as income levels and educational attainment levels as laid out in the
objective of the survey.
Methods
Ordered Statistics
Non-parametric Mann-Whitney U Tests based on ordered statistics were conducted on the
EU-SILC data and the SILD data to test for differences in distributions of explanatory variables.
In this case, differences between households containing only secondary/high school students were
compared to households with at least one university student. The Mann-Whitney U test sometimes
called the Mann-Whitney-Wilcoxon test is used to test the rank homogeneity between two
independent ordered samples (Bagdonavičius et al., 2013). This test was chosen instead of a t-test
because the Mann-Whitney U Test does not assume normality of the compared distributions (Fay
& Proschan, 2010). The application of this test compares the mean values of the ordered samples.
In other words, the means of the ranked samples (mean rank; Fay & Proschan, 2010). A chisquared test may be employed in place of this test, but the Mann-Whitney U Test was chosen under
the assumption that at least some of the distributions are not the same to begin with. This version
of the test was conducted due to its flexibility, the shape of the two distributions need not be the

9

same, while still allowing for conclusions to be drawn about the sample means (Van den Berg,
2018).
Regression
Binary logistic regression was used to model the odds of households containing only
secondary school students compared to households with at least one university student given a set
of explanatory variables from the EU-SILC and the SLID datasets. Logistic regression is defined
by the use of the Bernoulli response distribution and the logit link function (De Jong & Heller,
2008). Logistic regression ensures that the estimates for the probabilities are on the interval (0,1)
where the estimates may be outside of this interval with linear regression (De Jong & Heller, 2008).
This regression predicts which of two categories a person is more likely to belong to given certain
other information (Field, 2009). Given the binary nature of the dependent variable, logistic
regression was deemed to be an appropriate method for modelling this data. These models were
estimated using the maximum likelihood method within the context of generalized linear models.

Analysis and Results
Binary logistic regression and Mann-Whitney U tests were conducted on the Czech EUSILC and Canadian SLID datasets. The two household groups were separated with the assumption
that only some of the students in secondary school will go on to enter university. In other words,
the households with students in high school represent the whole population and the households
with university students represent a subpopulation. Identifying significant variables and their
effects on the likelihoods of households having a university student was the primary goal of this
analysis.
EU-SILC Variables
The dependent variable for the analysis conducted on the Czech EU-SILC data was a binary
variable that indicated if the household contained only students in secondary school or if the
household contained at least one student in university. Explanatory variables used in this analysis
were included to determine factors that influence the likelihood that a household contains at least
one university student.
1. The household income was examined by creating a variable “income per capita” expressed
in hundreds of thousands of Czech koruna (CZK). All households without reported income
10

and with only one person were removed from the analysis, as in this case the one person
would be the student and considered economically inactive. One hundred-thousand CZK
represents the approximate difference between the lower and upper quartiles.
2. The material deprivation is outlined as the number of items the household can’t afford from
the following list of nine items: to pay their rent, mortgage, or utility bills; to keep their home
adequately warm; to face unexpected expenses; to eat meat or proteins regularly; to go on
holiday;to own a television set, a washing machine, a car, or a telephone (Eurostat, 2013a).
3. The education level of the head of the household, either having no university education or
being university educated was included. This variable was derived from a larger set of
responses including various levels of education.
4. Another included variable contained two categories: Employed and unemployed. This
variable determined the economic status of the head of the household with all pensioners
removed from the analysis as they are again economically inactive.
EU-SILC Mann-Whitney U Tests
Due to the non-normality of the data, Mann-Whitney U tests with α = .05 significance were
conducted on some of the explanatory variables to verify differences in the means of the ranked
values; the null hypothesis of this analysis is that the mean of the ranks are the same.
The first of the Mann-Whitney U tests were conducted on the income per capita of the
households. The distributions of the income per capita separated between secondary school and
university households are found in Figure 1. In both cases, the distribution is positively skewed
and appears to be leptokurtic. When comparing the two sample distributions it seems as if the
university households shown in green, have more observations in the higher income sections.
Visually, it also seems that households with only secondary school students are more concentrated
at the income level just over zero. The results of Mann-Whitney U test in Figure 1 yield that
significantly, the null hypothesis is rejected in favour of the alternative that there is a difference in
ranked means.
The differences in material deprivation were next compared across the two groups.
Examination of the sample distributions found in Figure 2 seem to indicate that secondary school
households have a higher level of material deprivation than university households. Both
distributions are heavy-tailed with many households able to afford all of the items on the list

11

although it appears that many more university households than secondary school households are
without deprivation. The results of the Mann-Whitney U test are found in Figure 2; the null
hypothesis is rejected in this test.
EU-SILC Logistic Regression
The matrix of the Pearson correlations of the included variables can be found on Figure 3.
This was examined to prevent the possibility of collinearity within the model estimations.
The results of the binary logistic regression estimation including: variables, parameter
estimates, significances, and confidence intervals are shown in Table 1. Prior to the construction
of the model outlined in Table 1, smaller regression models were estimated to understand and
validate possible inclusions of variables in the final model, the results of these estimates will not
be discussed further. As shown in Table 1, not all of the variables are significant in this estimated
model. The explanatory variable describing the employment status of the head of the household
was not significant in the model. Significantly, with an increase of one hundred-thousand CZK in
income per person a household is 1.270 times more likely to have a student in university compared
to just a secondary school student. Furthermore, with an increase in one item of material
deprivation, the odds of the household having a university student decreases with odds of 0.838.
Households with a university educated head were estimated to be 1.771 times more likely to have
a student in university than just in secondary school. In the context of the overall model the
socioeconomic backgrounds of Czech students do influence their entry to university.
SLID Variables
Dependent variable “high school or university” was the indicator variable separating
households containing only students in high school and households with at least one student in
university. The motivation for this analysis is exactly the same as for the Czech EU-SILC with the
only difference being the term high school instead of secondary school. By the ISCED
classification secondary school and high school are considered to be the same level (Eurostat,
2016). Using similar methodology to the analysis of the Czech data; explanatory variables used in
this analysis were included to determine factors that influence the likelihood that a household
contains at least one university student.
1. The variable “income per capita” was expressed as thirty-thousand CAD. This thirty
thousand CAD represents the approximate difference between the lower and upper quartiles

12

of the income per capita of the Canadian households. All households with zero reported
income and one person were removed from the analysis. Any household with one person in
this analysis would be the student and therefore economically inactive. Households with zero
reported income were also removed due to the high probability of this being misreported.
2. The material deprivation is outlined as the number of items the household can’t afford from
the following list of ten items: to eat fresh fruit and vegetables every day; able to get dental
care if needed; eat meat, fish, or a vegetarian equivalent at least every other day; able to
replace or repair damaged appliances such as a vacuum or a toaster; have appropriate clothes
for job interviews or other special occasions; able to get around your community, either by
having a car or by taking the bus or an equivalent mode of transport; able to have friends or
family over for a meal at least once a month; dwelling free of pests and insects such as mice,
bedbugs, or cockroaches; able to buy some small gifts for family or friends at least once a
year; or have a hobby or leisure activity (Statistics Canada, 2011).
3. The educational background of the other members of the household was included in the
analysis in the form of a variable indicating if a member of the household that wasn’t
identified as the student in the analysis had completed tertiary education. In other words,
whether or not the student is living with someone who has completed tertiary education.
4. The province of residence was included in the set of explanatory variables.
5. A variable indicating if someone in the household had received social assistance within the
reference year was also examined for potential influence on students.
SLID Mann-Whitney U Tests
Mann-Whitney U tests with α = 0.05 significance were conducted on some of the
explanatory variables of the Canadian SLID dataset to verify differences in the means of the ranked
values. The null hypothesis of this test is that the mean of their ranks is the same.
The distributions of income per capita were compared between households with only high
school students and households with at least one university student. Figure 4 shows the
distributions between the two groups and the results of the Mann-Whitney U test conducted. The
null hypothesis of the Mann-Whitney U test, that is that the ranked means are the same, is rejected
when comparing these two distributions. With the mean income per capita for households with
only a high school student at 28,512 and 29,775 for households with a university student.
13

The distributions of the material deprivation between the two groups can be seen in Figure 5.
In both cases most of the households are without any kind of material deprivation. It appears,
however, that households with only high school students have a slightly higher level of material
deprivation than households with at least one university student. This observation is validated
based on the results of the Mann-Whitney U test also found in Figure 5; the null hypothesis is
rejected in favour of the alternative hypothesis, that is the ranked means are significantly different.
SLID Logistic Regression
Binary logistic regression models were constructed to examine the significance and effects of
included Canadian variables. Regression models were constructed with one or two explanatory
variables to evaluate their potential for inclusion in a larger complex overall model and will not be
mentioned further. The matrix of the Pearson correlations of the included variables can be found
in Figure 6. This was examined to prevent the possibility of collinearity within the model
estimations.
All previously significant explanatory variables were encompassed in the overall model. Table
2 contains parameter estimates, significance levels, and confidence intervals of the estimates. The
set of all explanatory variables were significant at the α = 0.05 level. With an increase in thirty
thousand CAD per capita the odds of a household having a university student are increased by
1.026. Moreover, the odds of a household having a university student decrease with increasing
material deprivation and if someone in the household had received social assistance with odds of
0.916 and 0.908 respectively. Households containing a member who has already completed tertiary
education are more likely to contain a student studying in university. The province of residence is
significant however uninformative. The construction of this model suggests that socioeconomic
factors also influence Canadian students’ entry into university.

Discussion
As a form of human capital, higher education attainment has a positive influence on earning
potential in the labour market (OECD, 2010). With this in mind, it would be imperative to identify
what influences such decisions to be made. Parental income and parental education influence the
entrance into university (Belley et al., 2011; Chevalier et al., 2013; Frenette, 2007). There have
been mixed results regarding the influence of tuition on university attendance (Bastien et al., 2014;
Canton & De Jong, 2005; Hemelt & Marcotte, 2008).
14

Within both sets of data, conclusions are drawn that the socioeconomic statuses of the
households do influence entry into university in both countries.
The Czech Republic
When examining the results from the EU-SILC data, the income per capita, material
deprivation, and the education level of the head of the family are all significant in influencing the
entry to university. These findings validate the results obtained in other research (Belley et al.,
2011; Chevalier et al., 2013; Frenette, 2007; Savas, 2016). The results show that households within
this sample with higher socioeconomic statuses are more likely to contain a university student. As
there is a relationship between educational attainment and earning potential found by Finnie and
Frenette (2003) in the labour market, one could argue that the explanatory variables used in this
analysis should be highly correlated and redundant. This is not the case, as shown in Figure 3.
Social selectivity is present in the Czech Republic and households with lower socioeconomic
means are the most at risk for exclusion from university.
Canada
The results from the Canadian SLID dataset also seem to weakly further validate other results
(Belley et al., 2011; Chevalier et al., 2013; Frenette, 2007; Savas, 2016). The income per capita,
material deprivation, tertiary education of a household member, and if a member of the household
received social assistance are all statistically significant in influencing entry into university. These
results, in all cases, indicate that households with higher socioeconomic statuses are more likely
to contain a student in university in Canada. Like the Czech explanatory variables, a correlation
matrix was constructed to examine the Pearson correlations. As can be seen from Figure 6,
although the correlations are significant, they are weak, warranting the inclusion of all the
explanatory variables in the model. The estimated odds ratios from this model are quite small,
notably when looking at income per capita. The estimated odds ratio of 1.026, although significant,
indicates that the income does not have a large influence on the entry to university. One could
argue that these estimates are only significant due to the large number of observations in the data
(Lin et al., 2013).
Comparing Results
The most comparable variables of the two sets of explanatory variables are the income per
capita and the material deprivation of the household. The material deprivation lists contain
different elements but both serve to quantify household deprivation levels. Both of these significant
15

variables indicate that households that have higher income are more likely to have a university
student. When making a direct comparison of the estimated odds ratios, it shows the magnitude of
the changes is a lot higher in the Czech Republic than in Canada, suggesting that the entry into
university in Canada is much less restricted than in the Czech Republic. Figure 7 and Figure 8
show the differences in income per capita per person between households with only secondary/high
school students and households with at least one university student for the Czech Republic and for
Canada.
Figure 7 shows an obvious discrepancy in the income per capita between the two groups.
Extreme observations have been removed from Figure 7 to focus on the main portion of the sample.
This visualization of the data reinforces the regression parameter estimates of the income per capita
variable. When comparing Figure 7 and Figure 8, it is quite noticeable that the differences between
secondary and university for the income per capita variable are quite a bit larger for the Czech
households than the boxplots for the Canadian data.
Figure 8 shows that the Canadian income per capita between high school and university
households are very similar. This is reflected in the estimated odds ratio of 1.026; stating that an
increase of thirty thousand CAD per person in the household only slightly increases the odds of
the household having a university student. Outliers were removed from Figure 8 to focus on the
main part of the survey population.
When examining all of the parameter estimates from the models, the existence of social
selectivity appears to be more present in the Czech Republic with regards to entry to university.
The odds estimates are larger in the Czech model than in the Canadian model for all of the
explanatory variables. As Canada charges tuition for public education, one would expect social
selectivity would be more present than in the Czech Republic (Bastien et al., 2014; Hemelt &
Marcotte, 2008; Hübner, 2012). Contrary to expectation, it seems that, although social selectivity
appears to exist in Canada, the magnitude of the selectivity is much larger in the Czech Republic.
Financial Aid
The existence of financial aid policies allows for students from lower socioeconomic
backgrounds to enter university (Canton & De Jong, 2005). Based on this analysis, the financial
aid policies are more successful in reducing social selectivity in Canada. All students may apply
for student grants and loans funded by their province of residence and the federal government
(Government of Canada, 2020) On the contrary, in the Czech Republic the only financial aid is
16

only aimed at the poorest part of population and in 2017 only supported 0.7% of all students at
university level (MŠMT, 2020). Also, this support is very limited, as it only covers around 43% of
the average expenditure of a typical Czech student (Fischer et al., 2016). In the 2013–2014
academic year the Canadian federal government spent 5.1 billion CAD on “Human Capital
Formation” including scholarships, grants, and financial aid to post-secondary students (PBO,
2016). Matěj et al. (2003) argued that a slow growth of educational opportunities in the post
communist Czech Republic has contributed to the higher apparent presence of social selectivity.
The public system in the Czech Republic isn’t able to support all students in need (Fischer et al.,
2016). A possible solution for this could be private loans issued by banks, similar to the ones issued
by major banks in Canada. In 2020, only three of the fifty banks operating in the Czech Republic
offer students loans and the maximum loan amount of three hundred-thousand CZK which covers
about three years of the student’s living and study costs for those studying at the bachelor level
(Fischer et al., 2016; Student Finance, 2016). The socioeconomic background of the students in
both nations has a significant, albeit differing, influence on their entry to university. It appears that
even without tuition, there are not enough resources available to Czech students to help them
overcome lower socioeconomic backgrounds.
Making a direct comparison between the two nations’ investments in education a clear source
of disparity emerges. In 2019 the estimated gross domestic product in purchasing power parity
(GDP in PPP) per capita in Canada was 49031.4 USD and 40314.2 USD in the Czech Republic
(The World Bank, 2020). Total expenditure on educational institutions as percentage of GDP (in
2017) was 5.9% in Canada and only 3.6% in the Czech Republic (OECD, 2020b). Focusing on
tertiary education only, the percentage of GDP expenditure in Canada was 2.3% and 0.9% in the
Czech Republic (OECD, 2020b). There is also a difference in Total expenditure on educational
institutions per full-time equivalent student on tertiary level of education. In 2019 in Canada
expenditure in PPP was 24,671 USD and it was less than half that in the Czech Republic at 11,484
USD (OECD, 2020c). Even without charging tuition, the Czech Republic is apparently not
providing sufficient support to help students from lower socioeconomic backgrounds enter tertiary
education.
Conclusions
An extensive amount of literature exists in examining influences on students’ decisions
regarding their studies. This literature has identified that family background, specifically the

17

socioeconomic background, influences students’ higher education decisions. Developments in this
area were made due to this analysis conducted on data from the Czech Republic and Canada. This
work provides important insight into the influences of social backgrounds on education in the
Czech Republic. Furthermore, the comparison of these two nations provides unique information
about the similarities and differences in these influences between countries with and without
tuition.
The hypothesis that the entry to university is significantly influenced by the socioeconomic
background of the students in the Czech Republic was strongly supported. The significance of key
variables in the Canadian model suggests that social selectivity may still exist to a smaller degree.
This social selectivity hinders the possible enrollment of many students from lower socioeconomic
classes. This phenomenon of social exclusion seems to be more prevalent in the Czech Republic
than in Canada. The overall magnitude of the odds ratio estimates for the Canadian data are far
smaller than the Czech estimates. Based on these findings, a hypothesis that the social selectivity
in a tuition-free nation is less than that of a nation charging tuition for public university cannot be
supported. Moreover, these results may be linked to the structure of the financial aid in place in
Canada or the lack of opportunities to enter higher education in post-communist era Czech
Republic (Matěj et al., 2003).
The idea of higher education being equally accessible to all who have the skill set and
motivation to study is very important to this work. Education is a means for individuals to improve
their own standard of living which is inherently critical for those who find themselves at the bottom
of the socioeconomic ladder. Understanding the influences of students’ backgrounds is an
important step in developing competent methods of ensuring open access to higher education for
all members of society. Social selectivity is prevalent in the entry to higher education in the Czech
Republic and Canada.
This work is limited by the inability to show the direction in which the social selectivity appears
to be going. In other words; this analysis would benefit from the inclusion of older data to see if
improvements are being made. Results from other work suggest that things have become more
egalitarian recently with regards to entry into higher education in Canada. Furthermore, no
information about the specific labour market’s educational demands and cultural educational
demands of each country at the time of data collection is included in the analysis. This could
provide some valuable additional context to why people are making their study decisions. The

18

surveys analyzed were not directly designed for this research, and therefore could be missing
additional explanatory variables of interest. As tuition fees are not constant across nations, it is
necessary to consider the magnitude of these costs, and how much government funding is provided
to education when applying these results to other nations. For example, public university tuition in
the United States appears to be comparable to Canadian tuition rates however, private universities
in the United States charge much higher tuition as outlined by the United States Government
(2020) and this may impact the social selectivity problem more.
Further analysis into the post-treatment effects of social inclusion policies and projects could
also be a direction in which to develop this research. As mentioned above, steps are being taken to
create more social equality in higher education in both nations. Moreover, the identification of
successful and unsuccessful attempts to develop this equality could be used to create more efficient
ways of achieving this goal. With this in mind, more in-depth analysis of the influences of tuition
seems to be in order.
Based on the comparison of the Czech Republic and Canada, tuition does not appear to be a
large issue for Canadians from lower social classes. It appears that the financial aid policies in
Canada are doing a better job than those in the Czech Republic. This may lead other nations to
examine and adopt some of the successful elements of the Canadian system helping to solve this
problem in many places.
This analysis does not provide enough empirical evidence to confirm or reject the influence of
tuition outright. Another way to develop this work would be to include more countries or a wider
time span. This would allow for further understanding of the effects of these socioeconomic
variables in a broader context. Continued monitoring of the social selectivity and equality
promotion projects in higher education is important in order to provide fair opportunities for people
to improve their own socioeconomic status.
Acknowledgements
This paper is prepared under the support of the project of the University of Economics, Prague—
Internal Grant Agency, project No. 29/2017 “The Impact of Social Status on the Attitude and
Behaviour of Students at Higher Education Institutions”.

References

19

Achelpöhler, W., Bender, K., Himpele, K., & Keller, A. (2007). The introduction of tuition fees
in Germany and the International Covenant on Economic, Social and Cultural Rights UN
ICESCR. https://download.eiie.org/Docs/IRISDocuments/Education/Higher%20Education%20and%20Research/200700288-01-E.pdf
Bagdonavičius, V., Kruopis, J., & Nikulin, M. S. (2013). Non-parametric tests for complete data.
John Wiley & Sons. https://doi.org/10.1002/9781118557716
Bastien, N., Chenard, P., Doray, P., & Laplante, B. (2014). Économie, société et éducation:
L’effet des droits de scolarité sur l’accès aux études universitaires au Québec et en
Ontario [Economy, society and education: The effect of tuition fees on the access of
university studies in Quebec and Ontario]. CIRST.
https://depot.erudit.org/bitstream/004078dd/1/2014-01.pdf
Becker, G. S. (2009). Human capital: A theoretical and empirical analysis, with special
reference to education. University of Chicago Press.
Belley, P., Frenette, M., & Lochner, L. (2011). Post-Secondary attendance by parental income in
the US and Canada: What role for financial aid policy? Working paper. National Bureau
of Economic Research. https://doi.org/10.3386/w17218
Callender, C., & Jackson, J. (2008). Does the fear of debt constrain choice of university and
subject of study?. Studies in Higher Education, 33(4), 405–429.
https://doi.org/10.1080/03075070802211802
Canton, E., & De Jong, F. (2005). The demand for higher education in the Netherlands, 1950–
1999. Economics of Education Review, 24(6), 651–663.
https://doi.org/10.1016/j.econedurev.2004.09.006
Chevalier, A., Harmon, C., O’Sullivan, V., & Walker, I. (2013). The impact of parental income
and education on the schooling of their children. IZA Journal of Labor Economics, 2(1),
8. https://doi.org/10.1186/2193-8997-2-8
De Jong, P., & Heller, G. (2008). Generalized linear models for insurance data. Cambridge
University Press.
Eurostat (2013a). Glossary: Material deprivation. https://ec.europa.eu/eurostat/statisticsexplained/index.php/Glossary:Material_deprivation

20

Eurostat (2013b). Methodological guidelines and description of EU-SILC target variables.
www.dst.dk/ext/620929312/0/ukraine/ENG_METHODOLOGICALGUIDELINESAND-DESCRIPTION-OF-EU-SILC-TARGET-VARIABLES-2014operation-(VersionSeptember-2013--pdf
Eurostat (2016). International standard classification of education.
https://ec.europa.eu/eurostat/statisticsexplained/index.php/International_Standard_Classification_of_Education_(ISCED)
Fay, M. P., & Proschan, M. A. (2010). Wilcoxon-Mann-Whitney or t-test? On assumptions for
hypothesis tests and multiple interpretations of decision rules. Statistics Surveys, 4, 1–39.
https://doi.org/10.1214/09-SS051
Field, A. (2009). Discovering statistics through SPSS:(and sex and drugs and rock’n’roll).
SAGE Publications. https://us.sagepub.com/en-us/nam/home
Finnie, R., & Frenette, M. (2003). Earning differences by major field of study: Evidence from
three cohorts of recent Canadian graduates. Economics of Education Review, 22(2), 179–
192. https://doi.org/10.1016/S0272-7757(02)00003-1
Fischer J., & Vltavská, K. (2017). Social selectivity of the Czech tertiary education: Comparison
between EUROSTUDENT V and VI. EUROSTUDENT. Researchers’ forum. Hannover,
Germany: DZHW—German Centre for Higher Education. https://www.dzhw.eu/
Fischer, J., Vltavská, K., Brázdilová, M., Lipovská, H., Mazouch, P., Ptáčková, V., & Šimková,
M. (2016). EUROSTUDENT VI Základní výsledky šetření postojů a životních podmínek
studentů vysokých škol v České republice [Basic results of a survey of attitudes and living
conditions of university students in the Czech Republic]. MŠMT.
https://www.msmt.cz/uploads/odbor_30/TF/Analyticke_materialy/Eurostudent/E_VI_zav
erecna_zprava.pdf
Frenette, M. (2007). Why are youth from lower-income families less likely to attend university?
Evidence from academic abilities, parental influences, and financial constraints.
Analytical Studies Branch Research Paper Series. Statistics Canada.
https://www150.statcan.gc.ca/n1/en/catalogue/11F0019M2007295
Goldin, C., Katz, L. F., & Kuziemko, I. (2006). The homecoming of American college women:
The reversal of the college gender gap. Journal of Economic Perspectives, 20(4), 133–
156. https://doi.org/10.1257/jep.20.4.133

21

Government

of

Canada

(2020).

Student

aid.

https://www.canada.ca/en/services/benefits/education/student-aid.html
Hemelt, S. W., & Marcotte, D. E. (2008). Rising tuition and enrollment in public higher
education. IZA Discussion Paper No. 3827. https://doi.org/10.1111/j.00427092.2007.00700.x
Hübner, M. (2012). Do tuition fees affect enrollment behavior? Evidence from a ‘natural
experiment’ in Germany. Economics of Education Review, 31(6), 949–960.
https://doi.org/ 10.1016/j.econedurev.2012.06.006
Lin, M., Lucas Jr, H. C., & Shmueli, G. (2013). Research commentary—too big to fail: Large
samples and the p-value problem. Information Systems Research, 24(4), 906–917.
https://doi.org/10.1287/isre.2013.0480
Mack, A. (2016). Data handling in EU-SILC. Koln, Germany: GESIS—Leibniz-Institut für
Sozialwissenschaften. https://doi.org/10.21241/ssoar.47123
Matěj, P., Řeháková, B., & Simonová, N. (2003). Transition to university under communism and
after its demise the role of socio-economic background in the transition between
secondary and tertiary education in the Czech Republic 1948–1998. Sociologický
časopis/Czech Sociological Review, 39(3), 301–324.
https://doi.org/10.13060/00380288.2003.39.3.02
MŠMT (2020). Monitorovací ukazatele [Monitoring indicators]. MŠMT.
www.msmt.cz/vzdelavani/vysoke-skolstvi/monitorovaci-ukazatele
OECD (2010). Education at a glance 2010: OECD indicators. OECD.
https://doi.org/10.1787/19991487
OECD (2020a). About the OECD. www.oecd.org/about/
OECD (2020b). Education at a glance 2020: OECD indicators. OECD.
https://doi.org/10.1787/69096873
OECD (2020c). Price level indices. https://data.oecd.org/price/price-level-indices.htm
Office of The Parliamentary Budget Officer (PBO) (2016). Federal spending on postsecondary
education. https://www.pbodpb.gc.ca/web/default/files/Documents/Reports/2016/PSE/PSE_EN.pdf
Prokop, D. (2020). O chudobě, vzdělávání, populismu a dalších výzvách České společnosti
[About poverty, education, populism and other challenges of Czech society]. Host.

22

Savas, G. (2016). Gender and race differences in American college enrollment: Evidence from
the Education Longitudinal Study of 2002. American Journal of Educational
Research, 4(1), 64–75. https://doi.org/10.12691/education-4-1-11
Statistics Canada (2011). User’s guide for cross-sectional public-use microdata file: Survey of
Labour and Income Dynamics (SLID) 2011. Catalogue no. 75F0002M (2014). Ottawa,
Canada: Statistics Canada. https://www.statcan.gc.ca/
Statistics Canada (2020). Canadian students, tuition and additional compulsory fees, by level of
study, annual (dollars). Table: 37-10-0121-01.
https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=3710012101
Student Finance (2016). Úvěry a půjčky pro studenty [Credits and loans for students].
https://student.finance.cz/finance-slevy/%20studentske-uvery/
The World Bank (2020). GDP per capita (current US$). Washington D.C., USA.
https://data.worldbank.org/indicator/NY.GDP.PCAP.CD?most_recent_value_desc=true
United States Government (2020). College affordability and transparency list explanation form
(2016–17). https://catalog.data.gov/dataset/college-affordability-and-transparency-listexplanation-form-201617
Van den Berg, R. (2018). SPSS Mann-Whitney test—simple example. SPSS Tutorials.
https://www.spss-tutorials.com/spss-mann-whitney-test-simple-example/
Veselinovic, M. (2014). Want a top notch college degree but can’t afford it? Head to Germany.
http://edition.cnn.com/2014/10/22/world/europe/want-education-cant-paygermany/index.html

Note

1 This analysis is based on the Czech Statistical Office Survey on Income and Living Conditions
Survey (EU-SILC) (2009-2015). All computations, use and interpretation of these data are entirely
that of the authors of this work.
2 This analysis is based on the Statistics Canada Public-Use Microdata File: Survey of Labour and
Income Dynamics (SLID) (2011). All computations, use and interpretation of these data are
entirely that of the authors of this work.

23

Figure 1
EU-SILC Income per Capita

Figure 2
EU-SILC Degree of Material Deprivation

24

Figure 3
EU-SILC Pearson Correlation of Explanatory Variables

Table 1
Czech Binary Logistic Regression
Secondary or University
Odds Ratio:

Confidence Interval:

Income per capita

1.270∗∗∗

(1.156, 1.395)

Material Deprivation

0.838∗∗∗

(0.784, 0.895)

Head of Family Education

1.771∗∗∗

(1.463, 2.143)

Head of Family Work Activity

0.647

(0.579, 1.404)

Constant

0.498∗∗

Note.∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

25

Figure 4
SLID Income per Capita

Figure 5
SLID Degree of Material Deprivation

26

Figure 6
SLID Pearson Correlation of Explanatory Variables

Table 2:
Canadian Binary Logistic Regression

High School or University
Odds Ratio:

Confidence Interval:

Income per capita

1.026∗∗∗

(1.021, 1.031)

Material Deprivation

0.916∗∗∗

(0.909, 0.922)

Tertiary Education of Household Member

1.040∗∗∗

(1.033, 1.048)

Province of Residence

1.001∗∗∗

(1.001, 1.001)

Family Member Social Assistance

0.908∗∗∗

(0.895, 0.921)

Constant

1.211∗∗∗

Note.∗p<0.1; ∗∗p<0.05; ∗∗∗p<0.01

27

Figure 7
EU-SILC Income per Capita Boxplots

Figure 8
SLID Income per Capita Boxplots

28