Determinants of students academic performance the case of mekong river data

The OLS estimation results show that 5 out of 8 variables are chosen have significant influence on learning outcomes of student and the rest variables have no significant effect on stude

UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM

HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES

VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

DETERMINANTS OF STUDENT’S ACADEMIC

PERFORMANCE:

THE CASE OF MEKONG RIVER DELTA

BY TRAN CHI NGUYEN

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, NOVEMBER 2016

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

HO CHI MINH CITY THE HAGUE

VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

DETERMINANTS OF STUDENT’S ACADEMIC PERFORMANCE:

THE CASE OF MEKONG RIVER DELTA

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By TRAN CHI NGUYEN

Academic Supervisor:

DR PHAM KHANH NAM

HO CHI MINH CITY, NOVEMBER 2016

ACKNOWLEDGEMENT

I would to thank the dedicated assistance and supports of my supervisor, Dr PHẠM KHÁNH NAM, during thesis time The valuable comments and advices play a key role

in order to help to finish my thesis

Besides, I sincerely thank all lecturers - officers in Vietnam – Netherlands Program, especially Prof NGUYỄN TRỌNG HOÀI and Dr TRƯƠNG ĐĂNG THỤY, for their dedicated instruction, for all the knowledge from lectures and the supporting for my thesis during the course

I also would like to thank my classmates, especially Mr ĐẠT ANH, and my friend Ms THIÊN KIM, who has helped and supported me to complete my work

Finally, I wish lecturer team all the best and hope they will achieve many success in future path

ABBREVIATION

GSO: General statistics offices

UNESCO: The United Nations Educational, Scientific and Cultural Organization

OLS: Ordinary Least Square

VIF: Variance Inflation Factor

EPF: Education Production Function

VHLSS: Vietnamese Household Living Standard Survey

GPA: Grade Point Average

IV: Instrumental-Variable

BLUE: Best Linear Unbiased Estimator

IQ : intelligence quotient

EQ : emotional quotient

ABSTRACT This paper focus on examines factors that influence student academic performance in Mekong River Delta, Vietnam Basing on a survey that is conducted by the author in

2015, 297 students of Can Tho University have filtered to examine The Education production function of Bowles is applied and identified by a model includes eight variables, which are related academic performance of student The OLS estimation results show that 5 out of 8 variables are chosen have significant influence on learning outcomes of student and the rest variables have no significant effect on student academic performance in this study Hence, from the estimation results of this thesis, the suitable policies will be suggested in order to improve learning outcomes of student who experienced under performance group

TABLE CONTENTS

CHAPTER I: INTRODUCTION 1

1.1 Problem statements 1

1.2 Research objectives 3

1.3 Research questions 3

1.5 Research scope 3

1.6 Structure of the thesis 4

CHAPTER II: LITERATURE REVIEWS 5

2.1 Review of key concepts 5

2.2Review of theory 6

2.3Review of Empirical studies 8

2.3.1 Effect of family backgrounds on student academic performance 8

2.3.2 Effect of demographic factors on student academic performance 11

2.3.3 Effect of geographic factors on student academic performance 13

2.3.4 Effect of self-motivation of student on their academic performance 14

2.4Conceptual framework 15

CHAPTER III: DATA ANALYSIS AND RESEARCH METHODOLOGY 16

3.1 Model specification 16

3.1.1 Model specification 16

3.1.2 Measurement of variables 17

3.1.2.1 The dependent variable 17

3.1.2.2 The independent variables 18

3.1.3 Regression models 22

3.2 Data 24

3.2.1 Data source 24

3.2.2 Sampling strategy 24

CHAPTER IV: EMPIRICAL ANALYSIS AND DISCUSSION 26

4.1 Overview of education system in Vietnam and Mekong river delta 26

4.2 Data description 27

4.3 Regression results 33

4.3.1 Results 33

4.3.2 Test 37

4.3.2.1 Testing for goodness of fit 38

4.3.2.2 Testing for Heteroscedasticity 38

4.3.2.3 Testing for multicollinearity 39

4.3.2.4 Testing for normality 40

4.4 Discussion 41

CHAPTER V: CONCLUSION AND POLICY IMPLICATIONS 43

5.1 Conclusion 43

5.2 Policy implications 44

5.3 Research limitations and suggestion for further studies 45

References 47

Appendix 54

APPENDIX

Appendix 1: Matrix of correlation between variables 54

Appendix 2: Wald’s test of overall significant of the model 54

Appendix 3: Heteroscedasticity test 55

Appendix 4: Multicollinearity test 55

Appendix 5: Skewness/Kurtosis test of residuals and National exam score 55

Appendix 6: PCA result 56

Appendix 7: OLS estimation result 57

Appendix 8: Questionnaire 58

AP LIST OF TABLES

Table 3.1: List of variable and expected sign in relevance with student academic

performance .21

Table 4.1: The observation proportion by provinces 28

Table 4.2: Statistical description of factors on student academic performance 29

Table 4.3: Correlation of dependent variable and independent variables 33

Table 4.4: Regression result of OLS estimation 35

Table 4.5: Result of Wald’s test 38

Table 4.6: Result of Breusch-Pagan / Cook-Weisberg test 39

Table 4.7: Result of Multicollinearity test .40

Table 4.8: Result of Skewness/Kurtosis test for exam score .41

Table 4.9: Result of Skewness/Kurtosis test for residual 41

LIST OF FIGURES

Figure 2.1: Conceptual framework of the study .15

Figure 4.1: Distribution of student’s score 30

Figure 4.2: Relationship of gender and learning outcomes of student 30

Figure 4.3: Relationship between health and learning outcomes of student 31

Figure 4.4: Relationship between geography and learning outcomes of student 32

1.1 Problem statements

Nowadays, at macro level, education help a country to strengthen the human resource of the economy and occupies a very important role in the development of a country Besides, at micro level, education has long been perceived as a main factor help people to get the job, to get wealthy and promote earning of individual and family In the nineteenth century, it is generally believed that there is nothing can do about what education cannot do Recently, many researchers declared education plays an important role in the economy of every nation around the world Especially, developing countries such as Vietnam schools regarded as main element in the infrastructure The problem here is the determinant of student performance In general, people think that student outcomes are affected by intelligent, environment, effort, etc

In development economics, beside the sustainable development goals, education and health are two importance problems due to the influence of them in the long term The average education level of the society represents the level of development and achievements as well as the future prospects A community with high educational level will be more favorable to improve living standard and based on that the national economy is also benefited Besides, enhancing intellectual level of the people, well-trained labor plus advanced technology will help any countries to improve their productivity in both short term and long term

From the very beginning step, education is an essential part of human life; hence, the exact measurement of children's learning ability is very important The evaluation demonstrates ability, cognitive level and reaction of children Base on that, schools and parents can make the right decision for each child in order to maximum the limited human resource Another aspect, children learning ability also illustrates the advanced and efficient level of education system relied on those policy makers can apply reasonable policies

Vietnam and the people of Vietnam was recognized the essential role of education to the country Hence, the government has made a very strong push for the improvement of this field According to Le (2003), from 1991 to 2003, he mentioned that more than 120 tertiary institutions was built and this education field encountered highest growth rate in the last 10 year In the same period, Pham (2002) discovered that the number of student who studied postsecondary raised approximately six times In addition, the amount of student who studying upper secondary in Mekong river delta is about 377.099 students

There are some researchers focus on the determinant of education achievement in Vietnam In particular, Le (2000) has determined the factors of Vietnam's secondary education and he found that there is a difference in the enrollment rate between males and female students at upper secondary level Moreover, Le (2003) using the data of batch 26 students of HCMC University of Economics, he examined the factors between urban and rural students who studying in HCM University of Economics, the result shows that there is a gap between provincial and city students

Mekong River Delta has long been perceived as a remote region, the geographic location of Mekong Delta has many rivers and canals and the infrastructure, road system is underdeveloped which creates an obstacle for student to access to school Moreover, the youth literacy rate of Mekong River Delta is the third lowest in comparison with other regions in Vietnam and this rate is 0.80 percentage points lower than National youth literacy rate The adult literacy rate in Mekong Delta is

83.10 percent and it is about 6 percentage points lower than the National rate Therefore, more attentions should be paid for the low education level region such as Mekong River Delta

In fact, there are a few researches examine about the case of Mekong River Delta This thesis is the first research which examines the influence of the determinants on the individual student’s academic performance with the dependent variable is the individual score of students from National High School Graduation Examination in

2015 for the Mekong River Delta

In brief, the intellectual level of people in Mekong River Delta is quite low in comparison with the whole country A large part of population still living in the countryside, and the accessibility to education is very limited Therefore, the researches is about solving the education problems in Mekong River Delta is very necessary Hence, a method in order to identify determinants of student academic performance and the magnitude of the influence of those factors will be necessary This study will propose determinant factors which influence the student academic performance and base on that, the policy recommendations will be suggested to policy makers in order to have suitable policies Especially for developing country like Vietnam where the middle-income trap is becoming national problem

1.2 Research Objectives

There are two key research objectives of this paper:

(1) To examine the determinants of student’s performance

(2) To give policy implications to enhance student academic performance

1.3 Research Questions

In order to match the objectives in the section above, this study must concentrate on these questions:

(1) What are the determinants of student’s academic performance?

(2) Do these determinants affect academic performance of student?

1.4 Research scope

The scope of this research is Mekong River Delta of Vietnam, including 12 provinces

An Giang, Bac Lieu, Ben Tre, Ca Mau, Can Tho, Dong Thap, Hau Giang, Kien Giang, Soc Trang, Tien Giang, Tra Vinh, Vinh Long All statistic and data in this study are collected from a survey of the author at Can Tho University in 2015

1.5 Structure of the thesis

This study contains five chapters:

Chapter 1 is the introduction part of this study, which mentions about background and foundation of this study Chapter 2 provides fundamental definition about learning outcomes, theoretical reviews and some empirical studies relevant to determinants of student academic performance Moreover, there is also a recap table

of literature reviews and analytical framework the end of chapter two Chapter 3 will present the methodology, data and regression model for this study

Chapter 4 contains the data descriptive statistics and regression result of this study Conclusion of chapter 4 Chapter 5 base on the regression result in chapter 4 and gives a recap of estimation result Make recommendation to policy maker on main findings of the thesis then points out thesis limitations and suggestion for further research

CHAPTER II:

LITERATURE REVIEWS This chapter contains the literatures about theoretical, analytical and relevant empirical researches on learning outcomes of students This chapter first begins with some basic definitions about learning outcomes of student Then, the second part introduces the main theory on education achievement, which applied for this study The last part presents the review of empirical studies about determinant of learning outcomes and it shows the analytical framework for this study

2.1 Review of key concepts

Before investigate relevant theories, it is essential to look at some definitions and key words have connection with the issues, foundation of deeper analysis is to understand about fundamental notions clearly

Learning Outcomes

There are several definitions about learning outcomes in the world These definitions are different because of standpoints and aspects However, the similarity is learning outcomes are records that represent important and necessary knowledge that learners have achieved, and accurately display at the end of a process In other words, learning outcomes label what knowledge the student will get and can do at the end of

a program or course

Suskie (2009) asserts that learning outcomes are objectives that demonstrate how learner will be different because of a schooling or leaning experience More precisely, learning outcomes are the ability, skills, information, mindset and attitudes that learner will carry with them from schooling or learning experience

University of Warwick (2011) concludes the definition of learning outcomes as “The ability, knowledge learners will acquire upon successful completion a process of learning or a course.”

University of Manchester Institute of Science and Technology defines “learning outcomes are the acquisition of person's understanding, ability, knowledge or mindset that is the desired outcome of a learning process and will be demonstrate at the end

of that process or program” (UMIST, 2001)

In a compilation from other resources, Vlãsceanu et al (2004) states that student learning outcomes are the statements of what student is anticipated to figure out, recognize, and/or can explain after completed a learning process as well as peculiar practical or intellectual ability earned and showed by the successful fulfillment of a program or a course Learning outcomes, along with evaluation criteria, decide definitely the minimum requirement for the reward of credit, while the evaluating is based on achievement above or below the minimum requirements for the reward of credit Moreover, the learning outcomes are different from the goals of schooling in that they are involved with the attainments of student rather than with the general purposes of teachers

Learning outcomes score

Learning outcomes score is a worldwide-accepted score or a standardized-test score with highly reputation and are most precise criteria for ranking student

There are many grading systems around the world; each country uses their own grading system to measure varying levels of achievement in a course Grades can be assigned as a range (for example 1 to 10), as letters (typically A, B, C, D, F), or as a percentage of a total number of correct answers and so forth In Vietnam, the main grading systems is from 1 to 10 and 10 is the highest score ("10" - Excellent, "9" - Very good, "8" - Good, "7" - Acceptable, "6-5" - Satisfactory, "4-3-2" - Insufficient, "1" Fail) Typically, the lowest passing grade is 5 and it depends on each school

2.2 Review of theory

Educational production function theory

Historically, the production function was applied as an economic concept to the field

of education and it is called educational production function (EPF) The original idea of educational production function came from Coleman (1996), which had been carried forward by Bowles (1970) and Hanushek (1986) This EPF has received a lot of attention because education clearly captures an important place in every major economy of the world

In general, this function reflects the relationship between factors affecting student learning and learning outcomes of student These factors include schools, families, peers, neighborhoods, etc Bowles (1970) notes that this method makes the educational production process is probably become a specific value both in normative investigations and in descriptive studies of human capital formation to identify optimum educational resources distribution In addition, if there are some evidence show that schooling greatly affects labor productivity or earnings, it is necessary to know how schooling affects the development of human cognitive skills and attitudes

in school by tracking those evidences Moreover, a production function associating inputs of school to the advancement of productive capacity can provide a better explanation of why the more educated the better certified for productive aspects According to Hanushek (1986), the most common measure of schooling in previous studies are years of schooling completed However, a big problem with this popular measure of outcomes is that it assumes the same amount of student's achievement,

or skills are produced in a year of schooling This is simply counts the time people learn in schools without mentioning what actually happens in schools - therefore, it does provide an incomplete or inaccurate picture of outcomes

Hanushek (2007) showed the academic achievements of individual students is correlated to inputs that both controlled by policy makers (such as the characteristic

of schools, curricula, teachers and so forth) and not under-controlled as friends and families and innate intellectual abilities and learning capacities of learners Furthermore, educational process is cumulative, current levels of attainment affected

by inputs applied in the past To be more specific, he pointed out those demographic characteristics such as parental income, education and size of family usually used to identify family background of student Peer inputs typically are attainments for a school (or classroom) or aggregates socio-demographic characteristics of students School inputs commonly include teacher background, school organization (such as class sizes, administrative expenditure, facilities and so forth), and location or community factors (e.g., average expenditure levels) Most of empirical researches (except Coleman Report) based on data for other functions hence statistical analysis (commonly some form of regression analysis) is applied to conclude what particularly affects achievement

socio-According to Bowles (1970), an educational production function is defined as:

A = f(X1, ,Xm, Xn, ,Xv,Xw, ,Xz) Where:

A = measure of school output – for example, score on a standardized test;

X1,…,Xm = variables determining the school environments (such as teaching

quality, school facilities, and length of time student is exposed to these inputs)

Xn,…,Xv = variables representing environmental influences on learning outside

the school ( e.g., the parent’s income, education and so forth)

Xw,…,Xz = variables representing the student’s abilities and the initial level of

learning attained by the student prior to entry into schooling question

2.3 Review of Empirical studies

2.3.1 Effect of family backgrounds on student academic performance

There are many literatures have aimed on influence of parental backgrounds on their children's outcome, including education Among varies characteristics of parents,

parental education and income are two determinant affect significant to children's education

Glick and Sahn (2000) examined the influence of parental education and income on learning of boys and girls in West Africa, depending on utility theory, applying binary and ordered probit models Accordingly, they separate life of family into two stages The first stage is to work and raise their children The second stage is the retirement

of parents In the first stage, current income of the parents take responsibility for all the household spending There are two kinds of spending at that moment, including spending for education expenditure or education of their children and other consumption for living The education expenditure, in turn, determines the income levels of household in forthcoming time, the second stage when parents retired Thus, parents have to look at the tradeoff between the two expenditures The result depicted that rising in household earnings lead to the increasing on education achievement of girls but none influence on the boy children Furthermore, they also asserted that the positive significance is not only the achievement but also displays grade achievement Besides, the research declared that parents' education has positively association with children learning In specific, the higher parents’ education level is, the higher achievement rate of children earns Moreover, mother occupies a crucial role in the schooling determination of their children Undoubtedly, mother who is lacking education or is traditional women likely to forbid their girls taking part

in secondary school

Halle et al (1997) concluded that mother with high level of education tend to have higher expectation on their children academic achievement and these expectations were correlated with better performance of children in Math and Reading, using a dataset of an outnumbered group of families with low-income Similarly, Corwyn and Bradley (2002) declared that maternal education had a direct effect on their children cognitive and learning outcomes

Similarity, Tansel and Bircal (2006) also pointed out that family with high levels of income invest more money for education of their children rather than low-income household

Another research created by Blanden and Gregg in 2004 verified that the relationship between family income and education achievement was very close in the United Kingdom and it has been enhancing over time They also declared that earnings do have a causal effect on educational outcomes Besides, raises 1000$ in annual family income from 2 to 5 years leads to the growth of student achievement by 6% standard deviation (Duncan, Morris and Rodrigues, 2011)

Moreover, Hashimoto (1995) asserted that the average income elasticity is 1.72; applying OLS model to predict the elasticity of income of education in Japan That is, when income increase one percent, the spending of education will raise one point seventy-two percent This elasticity is much higher than that of the other consumption spending Similarity, Huston (1995) indicate the same outcome, which means that when family earnings increases, education spending ratio, which is determined by education spending over total family spending, is raised

In Vietnam, there is a tight relationship between household earnings and children's education Behrman and Knowles (1999) created the study, which aimed at demonstrating the relationship between educational cost, the spending of parents directly paid to school, and income of household Actually, it linked three implications

to assess the relationship The implications were the education cost directly paid to school, school’s quality and quantity of school The outcome indicated that there are higher parents’ earnings there is more money for spends on education Indeed, the elasticity is approximately 0.65 To be more specific, when households’ earnings raise one percent, the spending on education also increase 0.65 percent Besides, it also asserts that income of household positively influenced enrollment rate

In brief, there are many researches focusing on the influence of parental background

on their children's education Undoubtedly, education and earnings of parents positively affected learning of their children

2.3.2 Effects of demographic factors on student academic performance

Demographic factor is explained as family size, gender factor and health condition Smits and Hosgor (2006) examining the influence of demographic factors on education attendance in Turkey, declared that gender factor, number of children, profession of parents and parents' educational level are fundamental factors determining the school achievement rate in Turkey Particularly, it verified that there

is the difference in the attendance rate of the girls and boys Actually, the attendance

of the boys much higher than the girls

According to Mandilaras (2002), women tend to have better achievement than men

do in economics course Besides, the Department for education and skills in Great Britain also publishes a statistic summary and it shows that girls tend to do better in majority of subjects than boys However, there are some other researches show that males seem to have higher learning outcomes in economics (Choudhury 2001, Kane and Spizman, 1999) In addition, Ellis, Durden and Gaynor (1998) proved that there is

no relationship between gender of student and academic performance

Jæge (2006) discovered that the influence of family size, parent resource distribution

to the outcome of children, applying data form Wisconsin Longitudinal study, with instrumental variable (IV) approach and OLS Accordingly, this research predicted the influence of parental resources and other inputs including social, time, economic, and relational with their children The outcomes of this study implied that when number

of children in the family who at schooling age increases the total volume of parental resources and other inputs accessible for children increase but decline to each child Ray (2000) joining the employment information, demographic aspect, applying data of Employment survey in Rural and Urban in India implemented in 1993, examined the effect of the size of the family on benefit of children in India To evaluate, the research

connected the relationship between family size and gender of breadwinner to the economic status of the family Accordingly, the study indicated that the root of poverty is relevant to family size in India As a result, this brought to the negative effect on benefit of children, education achievement As a matter of fact, it illustrated that children from bigger size families are tending to receive less schooling chance than those from smaller families

Huston (1995) conducting the study on influence family size, which is denoted as number of children who at schooling age, to educational spending ratio The outcome

of the research indicates that when family size rise, educational spending ratio (which

is determined as ratio between educational spending over total household spending) increase as well In other words, when number of children who at schooling age increases, the proportion of educational spending on total spending of family be apt

to rise Absolutely, the relationship between size of family and children's academic achievement is negative

Poterba (1997) pointed out that increases in number of children who at schooling age

in the family leads to the decline of educational spending for each child, the study applied panel date of the State of United State from 1960 to 1990 proposes the same outcome that demographic structure influences the schooling expenditure Similarity, Tilak (2002) also confirmed that the amount of education expenditure for children raises when the number of schooling children increase but it declines for each child of family in India Additionally, Downey (1995) demonstrated that if the children from large and small family have the same level of parental resources, the children from large family gain less benefit compared to the children from small family from the parental resources

There are several studies apply socioeconomic surveys of behavioral data and found that there is a strong positive relationship between health of children and educational attainment (Florencio, 1988; Moock and Leslie, 1986; Gomes-Neto et al, 1992) These studies also point out that learning outcomes of a child is negatively correlated with

the chronic malnutrition status Moreover, Leslie and Jamison (1990) the lack of nutrition and poor health have a negatively influenced on school outcomes and school attendance

Briefly, there are many researches on influence of gender factor, size of family and health condition on education of children The typical outcome is that gender of children and health condition also occupy an important role in the learning decision Furthermore, when number of children who at schooling age rises, the education expenditure for each child is decreased but it brings a rising in total amount of education expenditure for children

2.3.3 Effects of geographic factors on student academic performance

Ersado (2005), pointed out that there is a difference between city and countryside regions connected to children's education achievement, applying the multinomial logit framework, combining with data from Zimbabwe, Nepal and Peru, insufficient determinants effect on decision for schooling in countryside and city Indeed, the education achievement rate in countryside is lower than that of city regions In the countryside zone, if the family earnings are not enough, children will join the labor market instead of taking school The schooling decision also relies on the possibility to approach to commercial bank Actually, there is a positive influence on schooling by approaching the bank Moreover, the education achievement will rise when the country infrastructure is enhanced, and the number of schools is raised

Similarity, Park (2008) indicated that there are several differences in education between countryside and city in China The first characteristic is the school admission rate, the ratio in countryside, from 84 percent to 90 percent is lower than that in city, from 93 percent to 95 percent The second characteristic is the school's quality in these regions Indeed, countryside household cannot support for their child in order

to finish 9 years necessary in accordance with education law

Another study created by Jeynes (2007) discovered that the effect of parents on their children education in the city regions of the United State Accordingly, the outcomes

of the research suggested that there is a positive influence of parental involvement on their children education in the city areas

In Vietnam, Glewwe and Patrinos (1999) asserted that families living in city regions allocate more than 79 percent on education than that those living in the countryside

of Vietnam, applying the household data survey in years 1992 - 1993 In addition, Booth and Le (2010), investigating the urban-rural discrimination in Vietnam, adopting Vietnam Living Standards Surveys in the period of 1993 - 1998, recommended that there is a big gap between countryside and city regions in Vietnam Actually, the basic cause leads to the gap is age, education and ethnic Accordingly, the outcomes of the research indicated that benefits from education, measured as real per capita spending between two regions, differs from each other In specific, the education in the urban regions pays off better than that in the rural

In brief, there are several studies indicating the differences between countryside and city regions, including education Accordingly, the common outcome from those researches that education achievement in the rural lower than that in the urban regions

2.3.4 Effects of self-motivation of student on their academic performance

Many studies have evaluated the impact of self-motivation, which is commonly denoted as number of hours a student spends studying outside of class, on the student outcomes

Keith (1981) examined the causal influence of homework time on the high school student attainments, applying the enormous High School and Beyond data set The result indicated that study time of student positively affected student achievements, the direct effect of study time only less significance than that of the intellectual ability

To be more specific, the longer time student invests in learning the higher-grade student can get Moreover, the research also recommended that increased the demands on homework and tighter grading standard might raise both student confidence and attainments in school

Similarity, Wolf (1979) concluded that there is a significant relationship between the number of hours of homework per week and student achievement, applying the Evaluation of Educational Achievement surveys Furthermore, the research suggests that homework consistently among the best estimating variable within these blocks

In short, there are some researches pointing out the impact of study time of student

on student achievement Indeed, the number of hour student spends on homework positively affected the student outcomes

2.4 Conceptual framework

The figure below shows how did the regression model was built

Figure 2.1: Conceptual framework of the study

CHAPTER III:

DATA ANALYSIS AND RESEARCH METHODOLOGY There are three main sections in this chapter The first comes with the analytical framework and research methodology used for estimation Then, the second section looks at the rational of variable selections and variables also explained in detail The last section of this study introduces the description of the dataset used in this research Accordingly, the effect of determinants on the learning outcomes of student was estimated by OLS regression with data collected from a small survey of the author

There are some models available for the purpose of this study, but the most suitable model is from Hanushek (1972) First of all, this model is based on the educational production function Second of all, this model also fitting the desire to research about tertiary students In addition, this model examines the influences of family and student characteristics, which are relevant for the goal of this thesis

Hence, the model of Hanushek (1972) is chosen Then,

Ait = f(Bi(t), Pi(t), Si(t), Ii) (1) This model stated that learning outcomes is influenced by family characteristics, peers

pressure, school inputs and innate abilities Replacing the Ait, B, P and I by variables in the equation (1) above, we have a new specify model about determinants of student learning outcomes Equation (2):

Yi = β 0 + β 1incomelvli + β 2medui + β 3childreni + β4genderi + β 5healthi + β 6priorityi +

β 7houri + β 8a-quintilei + €I (2) Where:

Yi is the score in the National Exam of student i;

incomelvli is the family income level of student i;

medui is the mother education level of student i;

childreni are the number of children in the family of student i;

genderi is the gender of student i;

priorityiis the geographic location of student i;

houri is the number of hours for study at home and homework of student i

a-quintilei is the index about family assets of student i

3.1.2 Measurement of variables

According to empirical researches and theoretical framework, it is predicted that parent's education and income, family's wealth, health condition of student, religious, gender and distance travel from home to classroom have direct relation with the learning outcomes of students Therefore, this section is committed to supply deeper information about these determinants

3.1.2.1 The dependent variable

Student learning outcomes is portrayed by factors such as GPA, standard test score Some other researches, standard test score is used as dependent variable to determine factors of student learning outcomes in one subject (King, 2002; Mitchell and Collom, 2001; and Weyner, 2002) The standard test score become more popular when the concept of student outcomes is employed at a wider level of understanding There are many studies about student academic achievement use them as dependent variable (Checchi et al, 2000; Camara and Schmidt, 1999; Stinebrickner and

Stinebrickner, 2000; Stinebrickner and Stinebrickner, 2001a; Sullivan and Slayton, 2003)

3.1.2.2 The independent variables

The independent variables compose four categories

Family background variables

Parent Income: The relationship between children attainment and parental income is very tight Actually, parental income positive affects the learning outcomes of children (Hashimoto, 1995; Glick and Sahn, 2000; Blanden and Gregg, 2004 and Tansel and Bircan, 2006) In addition, Stinebrickner and Stinebrickner (2001a) asserted that family earning has a positive influence on student attainment Whereas, children in low level income families are more likely to discontinue of university before completion, even other determinants are under controlled

Mother education: According to Becker (1993), the influence of parent, especially mother, on their children education is very obvious Indeed, the expectation of high education level parents is that their children get higher educational attainments (Berhman and Rosenzweig, 2002) Moreover, Glick and Sahn (2000) declared that mother's education has a positive influence on their children's academic achievement Demographic variables

Gender: In some particular places in the world, there are some differences in performance between boys and girls Some researchers concluded that male students likely to have better learning achievement in economic sectors (Kane and Spizman, 1999; Choudhury, 2001; and Botelho, Pinto, Portela and Silva, 2001) However, a research made by Ellis, Durden and Gaynor (1998) indicated that the relationship between educational achievement and gender is not significant, despite the fact that male students perform better than females in key economic classes In addition, Burrus, Durmas and Graham (2001) came to an end with the same outcome, which showed that gender has no significant influence on learning outcomes Whereas,

another study created by Mandilaras (2002) pointed out that women have higher chance of gaining a better degree in economics Hence, the relationship between learning attainment and gender is still ambiguous In Vietnam, examining the VHLSS data of Vietnam 1997 - 1988, Le (2000) showed that boys have higher probability of participating in secondary school and upper, however, when having a chance, female students likely to perform better than males in term of educational performance Furthermore, Nguyen (2001, p 39-40) asserted that male students, in average, have lower learning attainment than females because the number of males in private school is higher than that in public school, as students learn at private school have lower learning outcomes than that of students in public school

Health: Better health is positively correlated with children educational attainment, which is measured by better performance in test score Indeed, Leslie and Jamison (1990) the lack of nutrition and poor health have a negatively influenced on school outcomes and school attendance There are some other researchers conclude that health and nutrition affect the learning outcomes of children in school (Levinger, 1992; Myers, 1992; Pollitt, 1990) Another study created by Jere R Behrman in 1996 asserted that better health and nutrition lead to the improvement in educational outcomes of poor children in the developing countries

Family size: As mentioned above in the empirical studies, there are several researches assert that the number of children at schooling age in a family has a positively affected on the children educational attainments (Huston, 1995; Poterba, 1997; Ray, 2000; Tilak, 2002)

Geographical variable

Geographic factor: A report from the 2009 Vietnam Population and Housing Census created by GSO claims that the gap in education between city and countryside still existed to the 2009 Indeed, the imbalance in education between city and countryside sectors was enlarging Moreover, a calculation of Hoang in 2014 from the Vietnam - Household Living Standards Survey in 2012 indicated that there are some disparities in

admission rate between countryside and city sectors To be more specific, the admission rate in countryside (75%) lower than that in the city (85%) Furthermore, there are other researches support the similar result about the disparity in education between rural and urban (Glewwe and Patrinos, 1999; Ersadi, 2005, Jeynes, 2007; Park, 2008)

Effort variable

Self-motivation: Some researchers asserted that a student with high study motivation more likely to spend more time on homework reparation and class attendance (Tay, 1994), which are crucial determinants for having better schooling outcomes There are some relevant studies to this effect (Husén, 1972; Coleman et al., 1966; Wolf, 1979; Keith, 1982) In Wolf’s (1979), for example, declares, "The number of hours of homework per week is substantially related to achievement" (p 321)

The following table demonstrates all the variables for this study, including dependent variable and independent variables In addition, it contains the definition of all variables and their expected sign to learning outcomes of student

Table 3.1: List of variable and expected sign in relevance with student academic

performance

sign Dependent variable

Y Total score of 3 subjects in the national

graduation exam Independent variables

Family background

variables

medu Highest number of schooling year of

mother in the family

health Times use hospital or medical station

services in 12 months before the exam

(after time in classes)

Positive Calculating the quintile assets of family

Step 1:

Prepare a sub-data set from the original dataset, filter out the asset variables At the same time, encrypt the variable to zero if there is no asset and other particular numbers for the number of assets are in possession of the family

Step 2:

Run the Principal Components Analysis

The value in the Component 1 column is the index score for each family based on the number of assets they are possessed

Step 3:

Identify the number of sections should be divided (5 sections in this case) After that, from all the family, pick the highest index to minus for the lowest index and divide it

to 5 (the obtained value is a)

Step 4: Family classification under sections by:

Family with lowest quartile will located from min of index score to min +a of index score

Family with second quartile will range from min +a of index score to min +2a of index score

Family with third quartile will range from min +2a of index score to min +3a of index score

Family with fourth quartile will range from min +3a of index score to min +4a of index score

Family with fifth quartile will range from min +4a of index score to max of index score Eliminating the out-layers if they exist

As a result, family is classified into 5 groups from the lowest to highest

3.1.3 Regression models

As well as the descriptive method, which is applied to illustrate the big picture of sample 297 students Ordinary Least Square regression is used to estimate the influence of determinants on learning achievement In statistics, OLS (ordinary least squares) is used in a linear regression model as a method to estimate the unknown

parameters With the given dataset, this method help to minimize the sum of squares

of differences between the observed responses

Real value: Yi = β 1 + β 2 X i + ei Estimated value: Y’i = β 1 + β 2 X i

With: β 1 = Y’ – β 2X’ β 2 = ∑ x i y/ ∑x 1 2

X’ = ∑X/n x i = X i – X’

Y’ = ∑Y/n y = Y i – Y’

There are several assumptions which have to be imposed to make OLS method show meaningful outcomes

 Correct specification: The linear functional form is accurately stated

 Strict exogeneity: In the regression, the error term should have conditional mean equals zero

In econometrics, F-test is a statistical test which is used to specify the model is best fits for the dataset or not Next, the Breusch-Pagan test is test which applied to check heteroscedasticity (the variance of the error term in the regression is constant or not)

Moreover, the variance inflation factor (VIF) will help to identify how much

multicollinearity (between estimators) in regression and the Skewness/Kurtoris test is used in order to check for whether this data follows normal distribution or not

Admittedly, there is a problem with endogeneity in the OLS regression model of this study and there are some solutions for solving this endogeneity problem and the most suitable solution for this study is to apply Instrumental Variable (IV) However, the collected dataset for this study is not enough for a good IV for that reason this endogeneity problem cannot be treated with this data

3.2 Data

3.2.1 Data source

Data for this study is a primary data which originated from a small survey of the author The survey is conducted at Can Tho University which is located in Mekong River Delta in 2015 This survey, covers 297 students from 77 districts of 12 provinces

in the Mekong River Delta and it includes detailed data on student Family background data states information on assets, parent's education and income, family's size Individual data observes information on health, learning outcomes (test score), religious, gender and distance travel from home to school

The questionnaire used for this study contains 4 components: The first component is the general information questions in order to get basic information of student and their characteristics The second component is the main part of this questionnaire This section contains questions for the data which is used for the regression such as income, family size, parent's education, and hours of homework

3.2.2 Sampling strategy

This study employed 320 randomly selected students (300 observations come from face-to-face interview and 20 observations from the online internet survey) who took part in the university entrance exam in 2015 The survey is conducted inside Can Tho University area and students who are chosen come from random areas such as

canteens, dormitories, classroom blocks and Learning Resource Center Inside Can Tho University, a number of students who is asked if they were participated in the university entrance exam or not, if they say yes then each of them is requested to complete a detail questionnaire In order to complete the whole questionnaire each student must spend about 10 minutes To reducing the time, 5 students is asked to answer the questionnaire at a time or they can leave their email for the online survey After collecting answer, there are only 297 students, accounting for 92.81 percent of total observations, have fulfilled the requirements and gave enough information for regression