Rong Deng 2021 China S Higher Education Expansion and Its Impact On Equality in Educational Opportunity

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Article

International Journal of Educational

China’s Higher Education Reform


2022, Vol. 31(1) 79–97
© The Author(s) 2021
Expansion and Its Impact Article reuse guidelines:
​sagepub.​com/​journals-­​permissions
on Equality in ​DOI: ​10.​1177/​1056​7879​2110​26667
​journals.​sagepub.​com/​home/​ref
Educational Opportunity

Liying Rong1 ‍ ‍and Feng Deng2

Abstract
By making use of a nationally representative dataset collected by China Family Panel
Studies, this article discusses how the higher education expansion policy influences
the equality of higher education access opportunity. Moreover, this study applied
the multilevel cross-­classified model in order to control for inequality in chances
of enrollment caused by the fixed admission quota for each province. The results
indicate that the expansion has interrupted the continuously enlarged gap of access
opportunities between regions; the expansion policy did not increase rural and ethnic
inequalities and narrowed the rural inequality within provinces; it also shows that
gender gap has been significantly reduced.

Keywords
higher education, expansion policy, equal access

Introduction
With its sustained and rapid economic development, China has seen an increasing
demand for highly educated, talented professionals. Since China’s reform and open-
ing-­
up policy, higher education has enjoyed steady growth. Following the

1
College of Education, Capital Normal University, Beijing, China
2
School of Humanities and Social Sciences, Beijing Institute of Technology, Beijing, China

Corresponding Author:
Liying Rong, College of Education, Capital Normal University, No. 23, Baiduizijia Street, Beijing 100037,
China.
Email: ​liying_​rong@​126.​com
80 International Journal of Educational Reform 31(1)

implementation of higher education expansion policies in 1999, this field developed


rapidly in its transition from elite to mass higher education. With expansion came
new challenges, notably how to ensure that wider and more disadvantaged sections of
society could benefit from new mass system—reflecting a belief that education could
perform a transformational role in society. “In today’s society where wealth and
power are distributed unevenly, education, as an important path to social mobility,
was trusted with a role to break the intergenerational transmission of knowledge”
(Huimin, 2010).
Unsurprisingly, therefore, the effects of higher education expansion on equal access
to education have received widespread attention in society. In this article, this chal-
lenge is explored through two related theoretical frameworks regarding the effects of
educational expansion on inequality in educational attainment and access to higher
study. The first is the hypothesis of maximally maintained inequality (MMI), which
states that educational expansion does not necessarily change educational inequalities
among the social strata, and only when the privileged stratum reaches saturation at a
given level of education will further expansion of that level reduce inequality (e.g.,
Raftery & Hout, 1993). The second, the effectively maintained inequality (EMI) the-
ory, posits that social strata continue to exist in higher education. Once that level of
schooling becomes nearly universal, the socioeconomically advantaged would use
their position to secure quantitatively similar but qualitatively better education. In this
case, education inequality would be effectively maintained (Lucas, 2001).
While coming to broadly the same conclusion, these theories are, in fact, looking at
different dimensions of the expansion phenomenon: quantitative and qualitative.

In essence, both theories believe that education expansion itself will not necessarily re-
duce inequality in educational opportunities. Their difference lies in that, MMI explores
equality in the qualitative sense while the EMI focuses on the qualitative disparity through
analysis of educational participation of various quality and types. (Xiaohao, 2006)

Quantitative equality may be easier to achieve, while qualitative equality, as a


higher goal, is more difficult to realize.

Literature Review
Higher Education Expansion and Inequality: Different Research Approaches
A common paradigm in this regard is to examine the distribution of higher educational
opportunity in society. For example, Feng (2012) distinguished between the “multi-
plying expansion” and “equivalent magnification” in higher education. By comparing
the changes in higher education enrollment rates in urban and rural areas before and
after the education expansion policy, he found that higher education enrollment in
rural areas increased rapidly. However, due to the great disparities in urban areas, the
real increase in rural areas was actually small. Other Chinese scholars also use
Rong and Deng 81

enrollment rates, the share of rural students in total freshmen, and stratification mobil-
ity rates to evaluate educational equality in urban and rural areas. According to their
research, the urban–rural education attainment gap was narrowed thanks to the educa-
tion expansion. In 1989, rural students only accounted for 43.4% of total educational
attendance rates, but in 2000, this figure rose to 48%, and in 2012, it reached 59.1%.
Generally speaking, the proportion of rural students going to higher education was
increasing.
Another research approach in this field is to divide the social background into sev-
eral factors (e.g., the cultural and political capital of parents) and analyze their effects
on access to higher education using regression models. Based on the 2003 China
General Social Survey (CGSS) data, Jinming (2006) employed the Cox proportional
hazards model to conduct a systematic examination of the higher education inequality
changes. He concluded that if higher education stratification is excluded, the advan-
tages enjoyed by parents with other occupational/educational status, compared with
manual laborer parents with low education levels, have been noticeably weakened
since the education expansion (Jinming, 2006).
More studies show, however, that higher education expansion actually increased
inequality. Based on the data from the China Health and Nutrition Survey (CHNS),
Guo and Wu (2008) examined the trend of educational inequality from 1991 to 2006.
They found that there was a substantial increase in inequality between urban and rural
areas and among social strata (Guo & Wu, 2008). In other research, Chunling (2010),
using data from a 1% population sampling survey in 2005 and a logistic regression
model, concluded that university expansion did not reduce the educational opportunity
gaps between different social strata, ethnic groups, and genders, but actually worsened
educational equality between urban and rural areas (Chunling, 2010).
On the other hand, Ge and Guangzhou (2011) questioned this finding since it
ignored the selection bias of unmatched populations when applying household mem-
bers matched data (e.g., less-­educated girls in rural areas who left their own family at
a young age, or rural families who became urban residents in further studies), thereby
potentially weakening their conclusions. Therefore, Ge and Guangzhou (2011) reex-
amined the research of Chunling (2010) by improving the matching precision of
household members matched data and reestablishing a research model. Their research
showed that after the higher education expansion, the influence of paternal education
and occupation on their children’s higher education opportunity was reduced, the gaps
between genders and ethnic groups were also narrowed, but the urban–rural gap was
widened (Ge & Guangzhou, 2011). Researchers also urged further improvements in
basic data collection and statistical methods for future studies. Chao and Qifa (2012)
used the CGSS data in 2006 and conducted a quantitative analysis on urban–rural
difference of higher education opportunities through Oaxaca-­Blinder decomposition
of binary choice. The empirical results showed that the urban–rural difference of
higher education opportunities is noticeably different, and this difference has been
widening over time (Chao & Qifa, 2012). Similarly, Yuxiao (2013) used a logistic
regression model based on the CGSS data in 2008, and found that after 1999,
82 International Journal of Educational Reform 31(1)

inequality of higher education attainment caused by urban–rural gaps and number of


siblings had grown significantly. Different research paradigms thus can lead to contra-
dictory findings on higher education equality, especially with respect to the urban–
rural gap, making it difficult to offer an accurate assessment of the impact of higher
education expansion or to provide suggestions for the improvement of education
equality. Some important variables in the regression model, if missed, may result in
biased errors.
Nevertheless, many empirical studies have shown that family background has a
direct and profound bearing on children’s higher education opportunities. Better edu-
cated people with prestigious occupations and high income have more advantages in
cultivating their children to become social elites. However, it will be argued that such
family influence is a result of a certain social system. Therefore, the education system
itself should not be overlooked in analyzing the causes of educational access.

China’s Enrollment and Quota System


In order to examine the education system effect on the process of educational expan-
sion and educational inequality, it is important to take into consideration China’s “pro-
vincial quota and enrollment score” policy in college admission systems and compare
the educational attainment disparity between different provinces using various
models.
“Provincial quota” is a policy where the state allocates higher education opportuni-
ties to different provinces and municipalities according to certain rules. In this system,
the number of candidates to be admitted and their distribution in universities and
majors are decided prior to the entrance exam. However, the allocation of quotas is not
decided by the number of candidates in the provinces.
The current college entrance examination system in China was established in 1952
when the Ministry of Education clearly stipulated that most of the higher education
institutions would take part in the national unified entrance exam, except for several
schools specially approved by the Ministry of Education. Thus, the students’ enroll-
ment regulations combined unified leadership and operation of the provinces, munici-
palities, and autonomous regions, leading to a unified college entrance examination
system. The unified college entrance examination enrollment plan can be divided into
the annual enrollment scales of higher education institutions and the plan of enroll-
ment sources for different provinces and majors (The Ministry of Education, 1951).
Since 1952, the enrollment scale of higher education institutions has been stipu-
lated by the Ministry of Education, according to the required national cadres amount
and category. The annual total scale of enrollment of higher education institutions has
been jointly decided by the Ministries of Education, Development, and Reform
Commission. The various higher education institutions formulate an annual enroll-
ment plan for different majors, according to their development scale, major settings,
teachers, teaching equipment, classrooms, dormitories, and other capacities, and then
Rong and Deng 83

report to the appropriate department of the school for audition. This department would
then report to the Ministry of Education for approval.
The second aspect is the enrollment sources plan. Enrollment sources refer to the
regions from which new students of higher education institutions come. The plan sets
the enrollment amount of higher education institutions in every province, municipal-
ity, and autonomous region. This plan, originating from the national unified exam in
1952, had the purpose of completing the national enrollment plan for higher education
institutions. Student flows were much lower at that time, and the task of enrollment at
higher education institutions in the major administrative regions did not match regional
sources of students, student ambitions, or the comprehensive national plan. In order to
complete the enrollment of higher education institutions, the Ministry of Education
allocated major higher education institutions’ student enrollment amounts, increasing
the flows of students in a planned way. At the same time, the national college enroll-
ment committee would assign the flows into certain universities and majors according
to the national enrollment plan, referring to the students’ conditions.
In 1977, the college entrance examination enrollment system was resumed (The
Ministry of Education, 1977). The enrollment of higher education institutions was
divided into three categories. The first was that higher education institutions and the
majors facing the whole country would enroll students throughout the country. The
second is that higher education institutions and the majors facing certain regions would
enroll students on a regional scale. The third is that higher education institutions and
the majors facing the provinces, municipalities, and autonomous regions would enroll
students from where they were located. In 1983, the Ministry of Education put forward
the plan of enrollment sources for the first time in official files (National Education
Commission, 1983), stipulating that enrollment sources of higher education institu-
tions belonging to the state council would be aggregated and balanced by the Ministry
of Education. The provinces, municipalities, and autonomous regions would thus need
to implement the plan. With the goal of ensuring the quality of student applicants, the
enrollment sources plan and graduate allocation plan would have to be properly com-
bined. The national key colleges and universities were allowed to have freedom and
rights to a certain degree for adjustment; that is, if the previous plan cannot guarantee
the basic students’ quality, the scale of adjustment should be controlled to under 20%.
The Household Registration System for the National Higher Education Entrance
Examination is a policy that requires candidates to take the entrance exam where their
residence is registered. It is not a move initiated by the Gaokao reform, but is a result
of the sprawling Household Registration System that controls mobility in all aspects
of society including education (Liying, 2012). In this respect, the “provincial quota”
caused a great provincial disparity in higher education enrollment. Students could
have different educational opportunities in part due to their registered residence.
Therefore, research studies should distinguish the educational opportunity disparity
between provinces, urban and rural areas, social strata, genders, and ethnic groups. It
is also noteworthy that the great institutional gap caused by provincial quotas cannot
be overcome solely through the individuals’ own efforts. While, the exam score for the
84 International Journal of Educational Reform 31(1)

entrance examination is the decisive factor in higher education enrollment, students


are selected based on the province-­based enrollment scores in accordance with the
provincial quota. Candidates will have to compete for the already distributed quotas
using their exam score, thus leading to the education opportunity disparity among
individuals. The gaps between urban–rural, social strata, ethnic groups, and genders,
as discussed by scholars in previous studies are, in essence, the inequality of outcome
based on equal opportunity.
Only when we take into consideration the provincial disparity of educational oppor-
tunity can we arrive at a more accurate conclusion. Over a decade has passed since the
expansion phase of higher education, so it is necessary to dive deeper into the issue of
unequal access to higher education using more sophisticated research methods based
on nationally representative data.
This research study employs the cross-­classified multilevel model with statistics
from the 2010 Chinese Family Panel Studies to examine the education inequality in
different provinces and years. It can help us estimate the disparity of educational
opportunity among different provinces more accurately and can show the net effect of
social strata, urban–rural, ethnic groups and genders, and numbers of siblings on edu-
cation equality since the higher education expansion in 1999.

The China Family Panel Studies: Data and Variables


The China Family Panel Studies (CFPS) is a nationally representative, annual longitu-
dinal survey of Chinese communities, families, and individuals conducted by the
Institute of Social Science Survey (ISSS) of Peking University, China. The 2010 CFPS
data cover 25 provinces and autonomous regions (Tibet, Qinghai, Xinjiang, Ningxia,
Inner Mongolia, Hainan, Hong Kong, Macao and Taiwan not included) with a nation-
ally representative sample of 21,822 adults. The survey employed three stage cluster
sampling with unequal probability, namely, province, district, and village.
This study focuses on the changes in equal access to education from 1977 when the
college entrance exam was resumed to the year of 2008. Most of the previous surveys
did not provide the exact time when individuals attended university, so their college
years have to be roughly estimated using their birth year.
In the CFPS, however, the individual’s education background was recorded in full
detail, such as whether they received any higher education, for how long, and when
they left college. This enables us to estimate the exact year that they entered a higher
education institution. The study selected 1524 individuals who entered full-­time higher
education institutions from 1977 to 2008, and 11,293 university matriculation age
individuals (born between 1959 and 1990) who did not receive any higher education
during the same time period.
Individual characteristics include gender (male = 1, female = 0), ethnicity (minority
= 1, Han = 0), and siblings (single child, child with one–two siblings, three or more
siblings). Previous studies usually divide the urban and rural samples by paternal reg-
istered residence. This is, however, a time-­varying variable and is not necessary the
Rong and Deng 85

same as the child’s registered residence when he/she participated in the college
entrance exam. CFPS asked in detail the individuals’ residence at birth, then at the
ages of 3 and 12 and in 2010, as well as their household types (urban or rural) and
current registered residence.
The study divides university matriculation age individuals by their household types
at the age of 12 into two groups: urban candidates and rural candidates. This is because
individuals at this age are unlikely to change their household types for military service,
recruitment, or promotion (except for a few cases such as land expropriation).
Furthermore, candidates are can only attend the college entrance examination in the
province where their residences are registered. Therefore, we try to ascertain the indi-
viduals’ registered residence at the age of 12. For those whose birthplaces are the same
as their current registered residences, their current registered residences can also be
seen as unchanged at the age of 12. For a few individuals who change from rural resi-
dents to urban residents and whose current registered residences are different from that
before adulthood, we will regard their residence at the age of 12 as the residence where
they took the college entrance exam. By determining the residence of individuals
before their adulthood, the impact of province on the individual’s access to higher
education can be explored.
According to the standard released by the National Bureau of Statistics in 2003, the
individual’s residence is divided into China’s Eastern, Central, and Western provinces
as well as municipalities. A common practice in the past was to divide society into ten
social strata by “ownership of organizational, economic and culture resources based
on occupational classification” (Xueyi, 2002). However, criteria such as occupational
classification and ownership of organizational and economic resources are time-­
varying elements, except for cultural resources, which cannot be changed in the short
term.

Research Method and Model


Approaches of Previous Studies
Previous studies on educational inequality changes mainly employed logistic regres-
sion models with different age cohorts, where admission to higher education is set as
the dependent variable, and factors influencing the admission are independent vari-
ables. The year-­on-­year changes of educational inequality are measured through the
two-­way interactions between age cohorts and other independent variables. This
research also uses this model to examine the educational inequality changes before
and after educational expansion. The measurements are modeled as follows:
( )
Pi
‍
Log = 1−Pi = β0 + β1 ∗ Expansion + Σk βk Xik + Σk β1k Xik ∗ Expansion + ei
‍ (1)

where Pi represents the individual’s probability to enter a college; Xik represents vari-
ous independent variables influencing the college admission (including gender,
86 International Journal of Educational Reform 31(1)

ethnicity, registered residence, numbers of siblings, parental education); and βk is the


regression coefficient of each variable, reflecting the influence of each variable on the
probability of university attainment. “Expansion” is the dummy variable representing
higher education expansion policy. “Expansion*Xik” represents the interactions of
expansion policy with other controlling variables, while β1k represents the impact of
education expansion on educational inequality.
However, the above model fails to consider the educational inequality among dif-
ferent provinces and will lead to a biased estimation of educational attainment odds. In
addition, China’s higher education has been expanding since the resumption of the
college entrance examination and the implementation of the higher education expan-
sion policy in 1999 gave further momentum to its rapid development. Therefore, the
year-­on-­year gap in higher education attainment should be taken into consideration.
The samples in this research study were collected from 25 provinces over a time span
of 32 years, which means individuals entered universities between 1977 and 2008.
This investigation adopts an approach similar to that of Wensheng (2008) by introduc-
ing 24 dummy variables representing different provinces and 31 dummy variables
representing different years to control the university attainment gap between different
provinces and different years:
[ ]
Log = pi /(1 − pi ) = β0 + β1 ∗ Expansion + Σk βk Xik + Σk β1k Xik ∗ Expansion
(2)
‍ +p1 + p2 ........... + p23 + p24 + y1978 + y1979 ......... + y2007 + y2008 + ei ‍

However, there are several flaws inherent to formula (2). First, the changing trends
of educational attainment in different provinces and years are set to a constant, but, in
fact, the unbalanced development of higher education has resulted in an uneven edu-
cation expansion in different provinces. The growth rate of educational enrollment
varies from year-­to-­year and from province to province. To examine the impact of
different provinces and years, we should add 799 variables as interactive dummy vari-
ables representing different provinces and years, but this will make the model too
complicated. In addition, MMI theory reveals that different factors have different
impacts on education opportunity during different time periods. For provinces where
higher education is popularized, the urban–rural gap will decline along with the satu-
ration of urban educational needs. For provinces where higher education is only
enjoyed by social elites, urban students still have obvious advantages in competing for
higher education opportunity. Even when we add the interactive dummy variables
representing different provinces and years in this function, the urban–rural variable
will remain as a constant. We cannot evaluate it in the random effects of provinces and
years.
Their registered residence and the time when a candidate attends the college
entrance exam are the two macro factors determining their odds in higher education.
The particular year of the exam determines the total amount of higher educational
quotas and the particular province decides their distribution in different provinces.
Rong and Deng 87

From this we can see that individuals are nested with other units such as provinces and
enrollment years.

The Cross-Classified Multilevel Model


Hierarchical linear models (HLMs) can efficiently analyze nested coefficients. Here,
as an individual sample is nested with a particular province and a year, we will use the
cross-­classified multilevel model (Raudenbush & Bryk, 2002), an advanced form of
the HLM-­multilevel interaction model. In this model, we will set provinces in the
horizontal row and years in the vertical columns, and every interacting group has its
own educational attainment probability and filtering mechanism, which means that an
individual has to compete with his peers in the same province for a seat in university.
Moreover, as the dependent variable (receiving higher education or not) in this research
is a dichotomous variable, we will extend this model into a hierarchical generalized
linear model (HGLM).
In the cross-­classified multilevel model, the total variance of dependent variables is
differentiated into the following three groups: level 1 province-­year interacting groups,
level 2 provincial variables in the horizontal rows, and level 2 yearly variables in the
vertical columns. Researchers must thus use the corresponding variables in different
levels to explain the results. Here we introduce a set of dummy variables representing
provinces into level 2 horizontal rows, and the variable representing higher education
expansion into level 2 vertical columns. Since the variation of year and province is
separate from each other on an individual level, we can also get the net effect of indi-
vidual and family variables on higher education access. We can use the cross-­classified
multilevel model as well to examine the random effects of individual-­level variables.
In this research study, the impact of urban–rural registered residence on higher educa-
tion attainment is set to vary in different provinces and years. Next, we will discuss the
interaction of level 1 variables and level 2 variables, that is, the impact of education
expansion policy on individual characteristics, and the impact of family background
on higher education attainment. The multiple variable intercept and variable slope are
modeled as follows:
( ( ))
pijk pijk
Level − 1 model : Log 1−p ijk 1−pijk
=
π0jk + π1jk (Ruralijk ) + π2jk (Genderijk ) + π3jk (Minorityijk ) +
‍ π4ij ∗ (Sibling) + π5jk (PEduijk ) + eijk ‍
where:
pijk is the probability of individual i in province k in the year of j;
π0jk is the intercept of province k and/in the year of j;
πpjk is the regression coefficient of independent variable related to province k in the
year of j;
eijk is the random effect of individuals i in level 1, that is, the deviation of individual i
from the mean of interacting group jk.
88 International Journal of Educational Reform 31(1)

Level − 2 model : π0jk = θ0 + b00j + c00k + β01 (Expansionj ) + γ0p (Areak )


π1jk = θ1 + b10j + c10k + β11 (Expansionj ) + γ1p (Areak )

‍ πpjk = θp + βp1 (Expansionj ) ‍


where:
θ0 is the mean of all samples intercepts;
θ1 is the urban–rural educational opportunity gaps of all samples;
β01 and βp1 are the impact of higher education expansion on educational attainment
and on educational inequality, respectively;
γ0p and γ1p are the provincial higher educational opportunity gaps and urban–rural
gaps in this regard; bp0j and cp0k are the random effects corresponding to years and
provinces.

Findings
This research uses the logistic regression model to examine the effects of educational expan-
sion on equal access to higher educational institutions. Model 1 examines the interaction of
the enrollment expansion and individual-­level variables. In Model 2, we add a set of dummy
variables representing different provinces. In Model 3, we introduce the interacting group of
provincial dummy variables and the expansion variable. We use the Logit Model to see how
the expansion influence the equity of higher education access (Table 1).

From Table 1, we can see Model 1 shows the interaction effect between expansion
variable and individual level variables. Model 3 shows noticeable gaps in educational
opportunity in urban and rural areas before the educational expansion, which are wid-
ening even further after the expansion policy. It is not, however, statistically signifi-
cant. Before higher education expansion, girls had far less opportunity than boys. This
situation was improved by the expansion policy, after which the gender gap declined
significantly. In terms of siblings, single children and children with one or two siblings
had noticeably greater opportunities to receive higher education before higher educa-
tion expansion. After the expansion, the impact of numbers of siblings declined.
Regarding the role of parental education, before expansion, the children of the more
educated parents were more likely to receive higher education, but after the expansion
policy such influence was reduced. In terms of provincial gaps, before higher educa-
tion expansion, candidates from municipalities had the most opportunity to access
higher education compared to others and those from Eastern and Central China also
had far more opportunity than those from the Western region. After the expansion, the
gap between municipalities and China’s Western region widened even more, while the
gap between Central and Western China was greatly narrowed.

To predict the probability of individual access to higher education using the cross-­
classified multilevel model, we have to construct a null model and separate the effects
Rong and Deng 89

Table 1. Effect of Higher Education Expansion on Equal Educational Access: An Analysis


Based on Logit Model.
Model 1 Model 2 Model 3
Independent variable Coefficient Coefficient Coefficient
Intercept −2.606*** −2.836*** −2.965***
Expansion 1.915*** 1.924*** 2.164***
Rural residence −1.340*** −1.304*** −1.296***
 Expansion × Rural −0.153 −0.148 −0.180
Female −0.403*** −0.402*** −0.405***
 Expansion × Female 0.428*** 0.427*** 0.433***
Ethnic minority −0.044 0.159 0.222
 Expansion × minority −0.303 −0.289 −0.310
Single child 0.388*** 0.327** 0.377**
 Expansion × Sgl child −0.141 −0.158 −0.254
1–2 siblings 0.341*** 0.325* 0.342***
 Expansion × 1–2 siblings −0.425 −0.436 −0.455**
Parental primary edu 0.896*** 0.885*** 0.859***
 Expansion × P prim edu −0.263 −0.253 −0.207
Parental secondary edu 2.460*** 2.470*** 2.441***
 Expansion × P s edu −1.101*** −1.127*** −1.071***
Parental higher edu 3.291*** 3.271*** 3.242***
 Expansion × P hig edu −0.529 −0.494 −0.412
Municipality 0.798*** 0.618**
 Expansion × Municipal 0.437
Eastern regions 0.361** 0.407**
 Expansion × Eastern −0.080
Central regions 0.179* 0.440**
 Expansion × Central −0.512**
Minority regions −0.777** −0.663*
Expansion × minority −0.208
Cox & Snell 0.188 0.190 0.192
Nagelkerke 0.363 0.368 0.373

of province and year on individuals from that of family-­related factors. The null model
reveals great differences in the individual’s probability of receiving higher education
in different years (τb00 = 1.491***) and provinces (τc00 = .748***).
90 International Journal of Educational Reform 31(1)

( )
Pi
Level − 1 Model Logijk 1−Pi = π0jk + eijk (1.0)
‍ ‍
‍ Level − 2 Model π0jk = θ0 + b00j + c00k ‍ (2.0)
The results form cross-­classified multilevel model can be seen from Table 2. In
Model 1, we insert the dummy variable representing education expansion into the func-
tion of the horizontal row and insert a set of dummy variables representing provinces
(Eastern, Central China, and municipalities, with Western China as the control group)
into the function of the vertical column. We also allow a random variation of provincial
educational attainment differences in different time periods in the horizontal row (c0pk).
Results show that the 1999 education expansion policy greatly increased samples’
opportunity to higher education (β01 = 1.946***). Specifically speaking, in terms of
provinces, candidates in municipalities (γ01 = 1.690***) and Eastern China (γ02 = .522*)
enjoyed far more opportunities to higher education than those in Western China. A
residual analysis of random effects thus indicates that education expansion policy has
been very effective in controlling provincial gaps in higher education attainment.
Taking educational attainment in municipalities for example, the residual is a measure
of deviation of the approximation from the exact solution, that is, the deviation of the
municipal-­western actual attainment gap coefficient in a certain year from the
γ01(1.690). The residual in 1977 is −0.52, which means that the coefficient of the
municipal-­western gap in higher education opportunity is 1.17 (1.690, 0.52).

The Logit Model 3 assumes that the provincial disparity of educational opportunity
in each year prior to education expansion is a fixed value, and that after the expansion
it is another fixed value. A simple comparison between two fixed values cannot
describe very well the effects of education expansion on educational opportunity in
different provinces.
‍ Level − 1 Model Logijk = π0jk + eijk ‍ (1.1)
‍ Level − 2 Modelπ0jk = θ0 + b00j + c00k + β01 (Expansionj ) + (γ0p + c0pk )(Areak )‍ (2.1)
In Model 2, we inserted feature variables representing individual and family into
the Level-1 Model and allowed for the variation of the urban–rural attainment gap in
different provinces and years. The result shows that urban–rural residence, gender, and
parental education have significant influence on higher educational attainment.
Individuals with one–two siblings have the lowest odds in higher educational attain-
ment. The random effect reveals a significant gap of urban–rural educational attain-
ment in different years (τb10=0.103***) and provinces (τc10=0.332***), and the latter
exerts a greater influence than the former.
Level-1 Model
( )
Pi
Logij = π0jk + π1jk (Ruralijk ) + π2jk (Genderijk ) + π3jk (Minorityijk ) + π4jk (PEduijk ) + eijk
‍ 1−Pi ‍ (1.2)
Rong and Deng 91

Table 2. Effect of Higher Education Expansion on Equal Educational Access: An Analysis


Based on HGLM Model.
Fixed effect Null model Model 1 Model 2 Model 3 Model 4
*** *** *** ***
Intercept,θ0 −2.245 −2.254 −2.674 2.665 −2.701 ***
 Expansion, β01 2.093 *** 2.253 *** 2.391 ***
 Municipality, γ01 1.765 *** 1.257 ** 1.256 **
 Eastern, γ02 0.581 ** 0.446* 0.454 *
 Central, γ03 0.360 0.253 0.260
 MINORITY −0.597 −0.823* −0.822*
Rural residence, θ1 −1.518 *** −1.616 *** −1.578 ***
 Expansion, β11 0.104 0.052
 Municipality, γ11 1.316*** 1.306***
 Eastern, γ12 0.753*** 0.775***
 Central, γ13 0.727*** 0.739***
 MINORITY −0.776** −0.648**
Female, θ2 −0.225 *** −0.233 *** −0.290 ***
 Expansion, β21 0.583**
Ethnic Minority, θ3 −0.117 −0.108 −0.105
 Expansion, β31 0.053
Single child,θ4 0.037 0.051 0.078
 Expansion, β41 0.222
1–2 siblings,θ5 −0.171 * −0.175 * −0.176
 Expansion, β51 0.195
Parental primary edu, θ4 0.620*** 0.595*** 0.581***
 Expansion, β41 −0.195
Parental 2ndary edu, θ5 1.614*** 1.570*** 1.715***
 Expansion, β51 −1.308***
Parental higher edu, θ6 2.860*** 2.835*** 2.881***
 Expansion, β61 −0.987**
Random effect Variance
comp
Level 2 horizontal row
variance
 Year, τb00 1.491*** 0.369*** 1.328*** 0.605** 0.573
 Rural residence, τb10 0.103*** 0.086** 0.085
Level 2 vertical column
variance
 Province, τc00 0.748*** 0.426*** 0.354*** 0.219*** 0.214
 Rural residence, τc10 0.332*** 0.099* 0.081
92 International Journal of Educational Reform 31(1)

Level − 2 Modelπ0jk = θ0 + b00j + c00k


(2.2)
‍ = π1ij = θ1 + b10j + c10k ‍
In Model 3, we introduced level 2 variables and level 1 variables in the model and
focused on their influence on the urban–rural gap and in different provinces. Results
show that the influence of education expansion on the urban–rural gap is not signifi-
cant (β11 = .104). The urban–rural gap is the greatest in minority area (γ14 =−0.776**)
Western China (θ1 = −1.616***), followed by that of Central China (γ13 = .727***), and
Eastern China (γ12 =0.753***), while the gap in municipalities is the lowest (γ11 =
1.316***).
Level-1 Model
( )
Pi
Logijk 1−P i
= π0jk + π1jk (Ruralijk ) + π2jk (Genderijk ) + π3jk (Minorityijk ) +
‍ π4jk (PEduijk ) + eijk ‍
(1.3)
Level − 2 Modelπ0jk = θ0 + b00j + c00k + β01 (Expansionj ) + γ0p (Areak )
(2.3)
‍ = π1jk = θ1 + b10j + c10k + β11 (Expansionj ) + γ1p (Areak ) ‍

In Model 4, we introduced the interaction of the “Expansion” variable and other


level 1 variables in our final model. Results show that before education expansion, the
gender gap of educational attainment is large (θ2 = −0.290***). As a result of the educa-
tion expansion, the gender gap was reduced noticeably (β21 = .583**). In addition, the
influence of parental education on children’s educational attainment was reduced after
the higher education expansion, but the expansion has no significant effect on reducing
the urban–rural gap (β11 = .052).

Discussion
The random effect of null model in the cross-­classified multilevel model shows that
there is significant disparity between provinces in terms of access to higher education.
The result of Model 4 suggests that students in municipalities and Eastern and Central
China have 2.9, 1.7, and 1.2 times more opportunities to receive higher education than
their counterparts in Western China.
The provincial quota policy means that students in different provinces or cities
never enjoy equal access to higher education because of the Household Registration
System. Different enrollment requirements in different provinces during a time of uni-
fied national entrance examinations have raised extensive attention from all quarters of
society. The gap in province-­based enrollment scores in some years exceeded even
100 points in different regions, which is considered to constitute the biggest inequality
in education (Zhong, 1999).
Our research study found that expansion helps to improve equality in access to
higher education by reducing the differences in access between regions. The random
Rong and Deng 93

effect of the provincial educational gap in different years shows that the provincial gap
was widening year-­on-­year since the resumption of the college entrance examination.
Such gaps were reduced shortly after the launch of the education expansion policy.
However, the situation worsened again with further implementation of expansion.
Jinming (2007) uses the Theil index and Gini coefficient to analyze the changes in
access to higher education between provinces and regions from 1998 to 2006. The
result shows that unequal access between provinces declined rapidly during the first 3
years when expansion was at its peak. The declining momentum of inequality began
to fade after 2002 when the government began to regulate the expansion scale.
Inequality increased slightly in 2006. Also, inequality between regions first dropped
and then increased.
One of the issues that this study has investigated is how expansion influenced
urban–rural access to higher education. Previous research based on national represen-
tative micro-­data shows that inequality in access between urban and rural areas has
increased since education expansion. By adopting the often-­used logit model, we also
found that urban–rural inequality in access actually increased after expansion. Yu
(2010), when talking about inequality in China, pointed out that it was influenced by
collective mechanisms, such as regions. These types of structural differences could not
by removed by individuals.18 In the cross-­classified multilevel model, we distinguished
the expansion variable from provincial gaps and found that expansion itself did not
worsen urban–rural inequality. The real cause is the unbalanced development of higher
education due to historical, financial and geographical distribution reasons, and so on.
During the period of expansion, it was the local government and market that promoted
higher education. Most of the expansion quota needed to be digested by the province
itself. The education expansion scale depended on the ability to enroll at the provincial
colleges and universities, which was financially supported by the provincial budget
and economic environment.
In fact, due to insufficient investment in education from local governments, expan-
sion was slow in central and western regions where most rural students came from. As
a result, qualified rural students from these regions, where higher education resources
were lacking, could not compete with students from municipalities and eastern regions
where education was fully developed. However, the effect of the provincial quota on
urban–rural equal access within certain regions should not be ignored. The higher
enrollment rate a region has, the smaller the urban–rural gap will be within that region.
In other words, municipalities and the eastern region have already led the nation in
creating more equal urban–rural educational access.
As for the effect of the number of siblings on an individual’s access to higher edu-
cation, the result of the HLM model shows that it has no significant influence on
access to higher education before or after expansion. The logit model meanwhile
shows that the number of siblings was negatively associated with access to higher
education before expansion and that expansion undermined the influence of the num-
ber of siblings. HLM and logit models produce different results because the logit
model fails to control for the educational disparity in different years and provinces. A
94 International Journal of Educational Reform 31(1)

single child enjoyed more opportunity to access higher education, but it does not
necessarily mean that families with fewer children enjoyed more education resources.
Most single children were born in more advanced provinces after the opening up
policy. The reasons why individuals with siblings were underprivileged in terms of
access to higher education include the incidence of their birth before the one-­child
policy and life in central and western regions with small enrollment plans. The fact
that they had more siblings is not associated with their low level of higher
education.
The HLM model and logit model show that parental education, genders, and eth-
nicity have roughly the same impact on access to higher education before and after
expansion. We discovered that expansion weakened the transmission of parental cul-
ture capital between generations. China’s current social structure remains unhealthy
with a pyramid-­shaped model, instead of an olive shape common to modern society.
Due to the small share of the privileged class in China’s population, even if their chil-
dren’s need for higher education is met, they would only account for a small part of the
expansion. Thus, middle and lower class children would have more opportunity. Take
individuals whose parents have received higher education for example, where they
only account for 4% of the total sample. Their enrollment rate before expansion was
already 60% and reached 85% after expansion. The need of higher education for this
social stratum has reached saturation.
Concerning gender disparity, we found that significant differences between genders
were disappearing after expansion. According to the Education Statistics Yearbook of
China, the proportion of female college students to total numbers was increasing. In
1990, it was 33.4%; in 1997, 37.3%; in 2000, 40.9%; and in 2005, 47.1%. A possible
reason for the narrowing gap between genders in terms of educational access is the
one-­child policy, which undermined the traditional idea of favoring males to receive
better education. Another possible reason is that females were more eager to receive
education than males because higher education could help females to get equal treat-
ment in a discriminatory labor market.
With regard to ethnicity, this study found no obvious differences between Han
people and ethnic minorities in access to higher education before expansion. The
Government has always adopted some favorable policies when enrolling minority
students. In the early years after the founding of new China, minority ethnicities
had a weak educational foundation. Accordingly, the Government made favorable
plans for minority students from remote, underdeveloped areas to ensure that they
enjoyed equal rights to higher education. The result of the HLM model shows that
minority students and Han students had roughly the same access to higher educa-
tion after expansion. The Government continued to improve higher education
enrollment policy for minority students after expansion. Since 2000, when making
enrollment plans for central ministries, the Ministry of Education has requested that
admission units and departments in charge should “increase the year-­ on-­year
enrollment number of students from border areas and western underdeveloped
areas.”
Rong and Deng 95

Conclusion and Proposals


This research shows that the current provincial quota policy contributed to the dispar-
ities in access to higher education between provinces and regions. It also has had an
indirect influence on urban–rural education gaps. The National Medium- and Long-­
term Education Reform and Development Plan (2010, 2020) proposes to “explore
measures to narrow regional gaps in access to higher education” in order to promote
social equality. Nevertheless, there is rationality to the provincial quota policy. Even
though some advanced provinces with sufficient educational resources enjoy greater
enrollment quotas, the need of less strong provinces where education is underdevel-
oped is not ignored when allocating quotas. Against the current background of unbal-
anced education development, the only plausible method is to make more reasonable
provincial quota policies based on certain rules. When determining the quota for col-
leges and universities, enrollment should be based on the number of candidates.
Compensatory measures should be directed to protect minority areas and economi-
cally weak areas.
We conclude that education expansion has played an important role in promoting
equal access to higher education. First, expansion stopped the growing provincial gap
in access to higher education. Second, the gender gap started to disappear after expan-
sion. The transfer of family cultural capital between generations was also reduced to a
certain degree. When expansion is separated from unbalanced development of higher
education, we find that the urban–rural gap did not widen after expansion, nor did the
ethnic disparity.
Research in England, Wales, and Japan shows that early stages of expansion may
bring about higher inequality and that unequal access for certain classes may only be
improved at later stages (Halsey et al., 1980; Kaneko, 2005; ). The reason why expan-
sion did not worsen higher education inequality is associated with the college entrance
exam system in China. Its enrollment policy centers on academic ability. Children
from unprivileged classes can overcome the disadvantages of coming from a poor
family and weaker education background by studying and working harder. The rela-
tively equal entrance exam system plays a major role in maintaining education equal-
ity and social justice, and in improving social mobility from lower classes to upper
classes.
The disparities in access to higher education, to a great extent, reflect the differ-
ences in basic education quality. Rural students receive poorer basic education, which
puts them in a disadvantaged position for the national college entrance exam. This is
why differences in access to higher education between urban and rural areas did not
improve after expansion. It is the government’s responsibility to provide equal educa-
tion resources according to the law. The National Medium- and Long-­term Education
Reform and Development Plan (2010, 2020) requests that

Efforts to bridge the education development gap between urban and rural areas shall
be accelerated. A framework shall be brought forth to integrate urban and rural areas in
96 International Journal of Educational Reform 31(1)

compulsory education development, with preference given to rural areas in fiscal fund-
ing, school construction, and teachers’ allocation. Coordinated urban and rural develop-
ment shall be realized first in counties before it is promoted in broader scope.

Declaration of Conflicting Interests


The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, authorship,
and/or publication of this article: This research was supported by Beijing Social Science Fund,
Young Academic Leaders Program.

ORCID ID
Liying Rong ‍ ‍https://​orcid.​org/​0000-​0002-​4019-​4167

References
Chao, C., & Qifa, X. (2012). Research on urban-­rural difference of higher education opportu-
nities: Empirical results through O-­B decomposition of binary choice. Statistics and Infor-
mationa Forum, 4.
Chunling, L. (2010). Expansion of higher education and inequality in opportunity of education:
A study on effect of ‘Kuozhao’ policy on equalization of educational attainment. Sociolog-
ical Studies, 3.
Feng, D. (2012). Higher education expansion and rural-­urban educational access. Research on
Higher Education, 8.
Ge, Y., & Guangzhou, W. (2011). On the misusage and adjustment of husehold members
matched data: A discussion with the paper expansion of higher education and inequality in
opportunity of education. Sociological Studies, 3.
Guo, M., & Wu, X. (2008). School expansion and educational stratification in China, 1981–
2006 [Paper presentation]. Paper presented at the American Sociological Association
Annual Meeting, MA: Boston.
Halsey, A. H., Heath, A. F., & Ridge, J. M. (1980). Origins and destinations: Family, class, and
education in modern Britain. Clarendon Press.
Huimin, Q. (2010). Equity in access opportunity: A most topical issue on equality in higher
education in China. Chinese Educational Law Review, 8.
Jinming, L. (2006). Expansion of higher education in China and inequality in entrance opportu-
nities: 1978–2003. Society, 3.
Jinming, L. (2007). Examination of regional disparities of higher education. Peking University
Education Review, 4.
Kaneko, M. (2005). Equality of opportunity in education—concept and reality. Tsinghua Jour-
nal of Education, 5.
Rong and Deng 97

Liying, R. (2012). The relations between college entrance examination and household regis-
tration system: A study of China’s college entrance examination policy. Journal of Capital
Normal University (Social Sciences Edition), 4.
Lucas, S. R. (2001). Effectively maintained inequality: Education transitions, track mobility,
and social background effects. American Journal of Sociology, 106(6), 1642–1690. https://​
doi.​org/​10.​1086/​321300
National Education Commission. (1983). Interim provisions on the preparation of the enroll-
ment sources plan for the general higher education institutions. National Education
Commission.
Raftery, A. E., & Hout, M. (1993). Maximally maintained inequality: Expansion, reform, and
opportunity in Irish education, 1921-75. Sociology of Education, 66(1), 41–62. https://​doi.​
org/​10.​2307/​2112784
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data
analysis methods. Sage Publications.
The Ministry of Education. (1951). The central people’s government issued the provisions of
high school freshmen enrollment in the summer of 1951. 105.
The Ministry of Education. (1977). Opinion on the enrollment work of colleges and universities
in 1977, the ministry of education. The Ministry of Education.
Wensheng, L. (2008). The equity of entrance to the China’s higher education. Peking University
Press.
Xiaohao, D. (2006). Expansion and equality in Chinese higher education. Peking University
Education Review, 4, 24.
Xueyi, L. (2002). A report on contemporary Chinese social strata. Social Sciences Academic
Press.
Yu, X. (2010). Understanding inequalities in China. Society, 3.
Yuxiao, W. (2013). Educational opportunities for rural and urban residents in China, 1978–
2008: Inequality and evolution. Social Sciences in China, 34(3), 58–75. https://​doi.​org/​10.​
1080/​02529203.​2013.​820555
Zhong, H. (1999). Enrollment scores of inequality. Beijing Observation, 12.

Author Biographies
Liying Rong is an associate professor at the College of Education, Capital Normal
University in Beijing, China. Her study focuses on higher education policy, education
resources allocation among regions, and so on.

Feng Deng is an associate professor at the School of Humanities and Social Science,
Beijing Institute of Technology. His study interest is education economics.

You might also like