− −
Examining the effects of school environmental factors on
pupil’s learning achievement in Ugandan primary schools
Albert Byamugisha
(Of ce of the Prime Minisiter, Uganda / Graduate School of International Cooperation Studies, Kobe University)
Abstract
In this study, the author employed a multistage analysis procedure to examine the effects of school environmental factors that in uenced reading and mathematics achievement among Grade 6 primary school pupils in Uganda. The data for this study were collected in 2007 as part of the Southern African Consortium for Monitoring Education Quality (SACMEQ) that sought to examine the quality of education offered in primary schools in the SACMEQ countries. The study focuses on Uganda with the aim of evaluating the in uence of home characteristics in raising pupil’s academic achievement as well as establishing the extent to which the school system characteristics in uence pupil’s achievement, and lastly investigating how the home-school characteristics in uence the pupil learning achievements. Results of the study show that; for the home context factors, the age of the pupil, size of house hold, sex of the pupil and whether the pupil speak English at home are important factors in the prediction of achievement in reading and mathematics. Also the education of the mother and having electricity at home are important factors that in uence pupil’s achievement in reading but not mathematics. For the school context factors, results indicate that; pupil having lunch at school, the type and location of the school, school resources and head teacher tertiary education are important factors in the prediction of achievement in reading and mathematics. For the home-school context factors, the number of days a pupil is absent, repetition of pupils, teachers meeting parents and parents paying extra tuition, characteristics are important factors in the prediction of achievement in reading and mathematics at Grade 6 level in Uganda. From the results of the analyses, several policy suggestions have been made to improve the facets of education that seem to be worthy of action.
1. Introduction
Uganda faces the challenge of meeting the growing demand for an educated, skilled and competitive work force as well as ensuring equity in access to quality education to meet the dual goal of sustained economic growth and poverty reduction. Recent ndings that cognitive achievement is statistically important in determining workers’ productivity suggest that pupils’ achievement has important implications for economic growth. In pursuit of national growth and development, increasing access and quality of education has been- evidenced from the last Poverty Eradication Action Plans (PEAP) of 1997 and 2004, and continues to be a national priority as articulated in the 2010/11-2014/15 National Development Plan (NDP). However, Government expenditure on education as a share of gross domestic product is still low for
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− − instance in 2009 it stood at 3.45% (World Bank 2009).
Due to budgetary pressure to expand investments that primarily drive growth (energy, roads, and other sectors), the education budget share dropped to 17.5 percent in 2007, with 65 percent allocated to primary education and 23 percent to secondary education. Government commitment to education is also re ected in the two mass education reforms-Universal Primary Education (UPE) and Post Primary Education and Training (UPPET) which were launched in 1997 and 2007 respectively. As a result, Uganda registers over 8 million children in the 7-year cycle of primary education, with a net enrolment ratio of over 90 percent for either sex. Similarly, enrolment in secondary education has increased from 814,087 in 2006 (45% girls) to 1,165,350 (46% girls) in 2009 following up to the commissioning of the UPPET reform: which saw a rise in secondary education Net Enrolment Ratio from 18.6% in 2006 to 24.4% in 2008 (MOES 2008). In spite of all the above investment, the sector continues to register low learning outcomes . School and individual characteristics have been pointed out to be enormous problems in Uganda and a major cause of poor performance of primary school pupils (Nannyonjo 2007)
Typically, therefore, the performance of primary education attracts policy debate both in terms of the systems coverage at this level of schooling and its ability to produce student learning. The performance of students on achievement tests administered within many of these countries suggests that academic achievement is often very low (Byamugisha & Ssenabulya 2004). The relative poor school performance may be partly explained by the school environmental characteristics represented by home, school and a combination of home and school variables. In light of the increased expenditure on the key school inputs for the Ugandan primary schools, it is important to understand why knowledge and skills acquisition has not registered similar improvements and, understanding what factors and investments most ef ciently improve pupil learning is of crucial importance. Given this background, it is of particular interest in this study to address the above concerns for Uganda by constructing several variables and test them for providing causal explanations of pupils’ achievement.
Therefore, the study attempts to provide answers to the following three main research questions:
a. How do the home characteristics in uence on the pupil’s academic achievement?
b. To what extent do the school characteristics influence on the pupil’s academic achievement?
c. To what extent does the home- school variables influence on the pupil’s academic achievements?
2. Study Objectives
The overall objective of this study is to examine the effect of school environmental factors
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
(categorised as home, school system and home-school) in raising learning achievements of pupils and the study uses the Multistage Ordinary Least Squares (OLS) regression analysis to investigate this relationship. Therefore, the study fall under three main objectives: First, the study evaluates the in uence of home characteristics in raising pupil’s academic achievement. Second, the study establishes the extent to which the school system characteristics in uence pupil’s achievement. Third, the study investigates how the home- school context characteristics influence on the pupil learning achievement. Numerous studies on school environmental factors and learning achievements in Uganda, with a conceptual focus on the link between environmental factors and pupil’s performance, have been conducted in the past and a quantitative study that combines school characteristics and family background of pupils has been largely ignored. However, to the best knowledge of the author no previous studies had examined the effect of inter-relationship that exit between the distinct environmental factors of the home, the school and home-school contexts on learning achievement in primary schools.
The current debates on education quality in Uganda also provide another reason for undertaking this empirical investigation. Uganda provides a good case for investigating issues of school quality for other reasons as well. First, as earlier mentioned, current policy debates casts doubt on whether supply side factors such as teachers, classrooms, and textbooks are the most important factors for improved learning achievement given the fact that increased education budgetary resources that have coincided with reduced learning achievement. Finally, although the focus of this current study is Uganda, the implications of the findings can be extended to other Sub-Saharan countries implementing UPE programs or even post primary education and training.
3. Literature Review
Many studies have been undertaken in an effort to identify the main determinants of scholastic achievement. Some have analyzed a large number of school based studies and agree that school matters and can have powerful effects on achievement. Other researchers have established that home background variable is an important determinant of levels of achievement. But the growing consensus is that the home environment and school environment are so bound up together that it is imperative to have a real understanding of one without the other. A brief summary of such studies is here by attempted.
The controversy as to which school factors contributed to school achievement was sparked in the United States of America (U.S.A.) in the late 1960’s with the Coleman Report (Coleman et al. 1966) that concluded that family background characteristics and community level variables accounted for more variance in student achievement than school resource variables like pupil-teacher ratios, per pupil expenditures or teacher characteristics. The stronger contribution of home factors to student achievements has also recently been confirmed by (Woesman 2005) in a study of primary students reading achievement in Colombia, Argentina,
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
and East Asia (Hong Kong, Japan, South Korea, Singapore and Thailand) using TIMSS data to analyze achievement in mathematics and science. Using ordinary least squares and multilevel models, Baker et al. (1997) confirmed that the predominant role of family background on achievement was similar across nations, regardless of national income.
A critical survey of literature has shown that educational facilities available at home (Alexander & Entwisle 1988; Majoribanks 2004), play an important role in the academic achievement of pupils and could thus positively in uence the learning performance of pupils and possible achievements of pupils. This positive link between parent’s academic attainment and children learning at school success was strengthened by the results of Epstein (1997).
On contrary, a study in Uganda by Heyneman (1979) finds that the social economic background of students particularly primary pupils do not matter in raising pupils’ achievement at all while an analysis of National Assessment for Progress in Education (NAPE) results of 2003 conducted by Nannyonjo (2007) found a positive relationship between numbers of books at home, language spoken at home a combination of English and vernacular, and pupil’s performance. Additionally, the analysis found a negative relationship between pupil’s age, and family size on pupil’s test scores. However, as stated in her report (p.2), the study was limited to a few variables and suggested further in-depth study to be carried out (p.xiii). As a point of departure, this study has employed multi-stage analysis to isolate effects attributed to home, school and home-school context variables. This analytical technique is critical in obtaining the variables that are fundamental in explaining variations in pupil performance outcomes, and also in decision making and policy targeting.
A number of research studies conducted in developing countries categorized problems relating to quality of primary education in three categories, namely, inputs, facilitating inputs and will to act – and identi ed a number of factors under each of these three categories. Fuller (1987) in his review of 26 multivariate studies with focus on input factors observed that quali cation of teachers have in uence on learner’s achievement and is a major factor amongst the input factors. The effect of the teachers’ experience yields results that are roughly similar to ndings for the United States. Although 35 percent of the studies (sixteen out of forty-six) display signi cant positive bene ts from more teaching experience (the analogous gure for the United States is 29 percent), the majority of the studies – twenty-eight out of forty-six-found this input statistically insigni cant. The results for teacher education, on the other hand, diverge in relative terms from those seen in the U.S. studies, with a majority (thirty- ve out of sixty three) supporting the conventional wisdom that more education for teachers improves student performance. In the U.S. studies, teachers’ education was the least important of all inputs. Although these results are still surrounded by considerable uncertainty (twenty-six estimates out of thirty estimates are insignificant and two display significantly negative effects), they do suggest a possible differentiation by stage of development and general level of resources available.
Also Ehrenberg et al. (1991) analyzed the causal relationship between student and
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
teacher absenteeism and the resulting impact on test scores in USA and Pakistan, respectively. The former found that higher student absenteeism was associated with poorer performance by students on tests scores in USA. More recent studies of the effect of inputs on pupil achievement include those by Hanushek (1995) and Glewwe and Jacoby (1999) among others. Therefore, availability of instructional materials (class resources, class resource chalk, wall charts and writing board) provides more motivating conditions for learning achievements. This is especially true for countries in which the level of school resources is already high. Fuller and Clarke (1994) reinforce this conclusion taking into account the cross-count differences in socio- economic and cultural settings even within developing countries. Other studies conducted by Heyneman and Loxley (1983) in Paraguay and El Salvador in science showed positive results in raising student’s achievement. Fuller (1987) reviewing 80 multivariate studies conducted in the third world, states that the developing countries present theoretically different conditions – the school is often a novel institution, operating in social settings where written literacy and formal socialization are relatively recent phenomena.
Actual counts of textbooks in Uganda in a comprehensive examination also revealed significant influence on pupil achievement (Heyneman 1989). The research indicated a moderate effect of textbooks and instructional material on achievement. Part of a study similar to this one for Uganda using NAPE results also showed that between 50-60% of the variation in test scores was determined by school xed effects (Nannyonjo 2007). The study further noted that results for pupil performance and teacher quali cations appeared to be mixed, in particular for Mathematics where scores appear to clearly decrease with increase in teacher quali cation except for teachers with tertiary education.
As regards to home-school, there are indications that nowadays, in at least one European country, for instance Germany, parents have a right to take part in educational policy decisions at school and regional levels. In other countries, parents are involved in curriculum decision making (Denmark, Norway). In the Netherlands, public schools (which are supported by local communities) and private schools (mainly church-related) have a tradition of strong parent committees at school and community levels.
The latest study on ‘School Quality and Student Learning’ conducted by Varghese (1993) in one of the educationally advanced states, namely, Kerala establishes that type of management does not seem to be an in uencing factor in learners achievement. However, it is internal management of the school, irrespective of the type of management the school has, is very important in influencing learner achievement. The study further shows that the level of school infrastructure and variations in the availability of teaching materials seem not to be closely associated with the levels of learning. Quali cations of teachers have also been found to be positively correlated with learner’s achievement so as the frequency of homework and class work in all grades. Similarly, achievement scores have been found to be positively associated with help for homework at home. An effective parent-teacher interaction is essential for smooth functioning of a school and there should not be communication gap between the school and the
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− − parents (Epstein 1994).
Another important variable is to care about student attendance because a child’s later success depends up a concrete educational background which depends partly on regular school attendance. A handful of authors have included a measure of school attendance as a right-hand-side variable in education production function. For instance, Fuller and Heyneman (1989) also uses this variable to measure achievement in Brazil but the point estimates were insigni cant. A recent study based on a randomized evaluation in Northern Uganda nds that take home rations conditional on school attendance boost math scores for primary school children (Adelman 2008), thus indicating that school feeding is an important factor that in uences pupil learning achievement.
4. Methodology
4.1. Conceptual framework
The causal relationship identified the malleable (i.e., attributes that are under the control of government and parents) and non malleable (home and individual characteristics) associated with home, school and pupil that influence learning achievements. In this case, a methodological framework is needed to identify the appropriate type of causations that can estimate and judge the credibility of the estimation methods. To study the impact of home and school related attributes on pupils’ learning achievement, the conceptual framework used in this study borrows the works of Glewwe and Kremer (2005) who employed causal relationships and behavior models in investigating education production functions that can be depicted as:
A = a(Y , Q , C , H , I) [1]
Where A is skills achieved and in this case are the performance grades, Y is years of schooling of the pupil, Q is a vector of school and teacher characteristics, C is a vector of pupil characteristics, H is a vector of household characteristics and I is a vector of school inputs under the control of parents, such as ensuring children’s daily school attendance and purchase of pupil learning materials. While at school, the pupil may require different inputs to produce different outputs suggesting that equation [1] can be regarded as multiple input-output functional model. Besides, the elements of H (household attributes), Q (school and teacher characteristics) and C (pupil attributes) are taken to be exogenous because they are determined outside the model framework and have no associated prices. The vector prices P1 are costs associated with years of schooling and P2 are costs associated with purchase of textbooks and other school supplies, such that the price vector has an indirect effect on learning, through the decisions and choices made on the endogenous variables Y and I.
For the household to maximize utility based on the years of schooling (Y) and school inputs by parents (I), we assume that only one school is available to each household and that nothing parents can do to change the characteristics of that school, teacher and pupil, and prices
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
because they are xed or given (i.e. exogenous). Then the household will choose Y and I to maximize its utility subject to the above constraints, thus:
Max U = u(Y , I) subject to:
Y = f(Q , C , H , P1) [2] I = g(Q , C , H , P2) [3]
Where U is the utility and u is a utility function of Y and I Substituting equations [2] and [3] into [1], we obtain,
A = a(Q , C , H , P1 , P2) [4]
From equations [2] and [3] households maximize utility with respect to each schooling choice conditional on the school that leads to the highest utility arising from years of schooling (Y) and scholastic inputs costs (I) by the household. Equation [4] is a fundamental equation upon which model formulation for this study is built, that stipulates that for the pupil to achieve utility maximizing skills (i.e. performance grades) at school, the school, and home characteristics plus associated school costs play a signi cant role. Besides, the environment within which the above attributes operate can relatively contribute to the education outcomes of the learning achievements.
The three types of environment mentioned in section 2, when examined together form a complex net of forces acting on the individual to change level of learning achievement at school. This net of forces is represented by a device called the conceptual framework. The conceptual framework in Figure 1 displays the proposed set of interrelationships between pupil achievements and its determinants. It has been developed by the author and is supported by the works of Keeves (1972), Marjoribanks (2004) and Glewwe and Kremer (2005).
As can be seen in Figure 1 three models represented by blocks of variables are developed. The variables are designed to identify which of the factors within blocks are responsible for the pupil learning achievements. The variables are grouped based on the malleability and the environment within which the pupil is being subjected to and evaluated as illustrated in the subsequent sections. Another factor guiding the grouping of the variables is the hierarchical nature of the education policy and implementation processes in Uganda and the possible interactions between the school and home casual effects.
Block ‘A’ constitutes a set of home context variables which constitute the predictors of model 1, and with Block D attributes as the outcomes/dependent variables, and the variables therein (Block A) are hypothesized to have direct relationship with pupil performance outcomes. From the onset, a pupil is born and groomed from home during his/her toddler stages up to the time he/she is exposed to formal education. The second of block (Block B) of factors are summarized in the school context variables and most of the variables in this block
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
are regarded as malleable factors (i.e. can be in uenced by government policy) and has a direct relationship with performance outcomes. A combination of Block ‘A’ and Block ‘B’ factors constitute the independent variables of model 2 (and Block D variables constitute the dependent variables) in that as a pupil leaves parental home nds another home which is the school.
The third most relevant block (Block C) of the variables is the home-school context variables that combine the roles of the home and school environments together and has a direct and strong relationship. For effective results and proper up-bringing of the pupil, the home and the school have to co-exist and share responsibilities towards this cause. Generally, the combination of Blocks A, B and C variables constitute the independent variables of model 3, thus the name multi-stage regression analysis.
Figure 1: Conceptual framework of school environmental factors of pupils learning achievements
Block A Home Context variables 1. Pupil age 2. Household size 3. Pupil’s sex
4. Pupil speak English at home 5. Pupil stays at home with parents 6. Mother and Father’s education 7. Having electricity at home 8. Pupil attending pre-primary 9. Total possessions at home index
Block D
Performance outcomes - Reading scores - Mathematics scores Block B
School context variables 1. Pupil having lunch at school 2. Pupil use of computer 3. School type & location 4. Education of class teacher & head teacher’s qualification and training 5. School resources index 6. Classroom index 7. Teacher absenteeism 8. Years of service and special training programs of the Head teachers
Block C Home-school context variables 1. Pupil repeating grades six 2. School distance from home 3. Pupil borrows books from the library 4. Pupil given homework at school 5. Pupil helped with homework at home 6. Teachers meeting parents at school 7. Number of days pupil is absent 8. Parents paying tuition for extra- coaching
Block E
Government: To finance and
influence policy Model 2
Model 3
Model 1
Source: Created by Author Based on Keeves (1972) and HSRC (1999)
Note: (i) imply direct and strong relationship. This represents model 1, model 2 and model 3 (ii) Implies reversal but weak relationship. Performance of a school will indirectly
attract parents to be very active in school activities.
(iii) Implies direct and unquantifiable relationship. The government can influence the home-school activities by developing policies like automatic promotion, placing books in the hands of pupils, pupils starting school at the correct age, distance from to school (iv) Block A=Model 1; Block A + Block B=Model 2; Block A + Block B + Block C=Model 3; Block D are performance outcomes and Block E are intervening government policy variables
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
4.2. Hypotheses
Based on the meaning of school environmental factors on learning achievements and the notion from literature that some factors affect pupil performance, the following hypotheses are formulated based on the research questions and objectives of the study and guide our investigation. From the conceptual framework and model, it was hypothesized that;
a. Home characteristics will have positive effects on pupil’s academic achievement. b. School system characteristics will have positive effects on academic achievement c. The home-school characteristics will have a positive effect on pupils’ academic
achievement
4.3. Model
To test the hypotheses mentioned in section 4.2 the study adopts the multi-stage regression function as used by Marjoribanks (2004). In the rst stage of the model, the home characteristics were considered. The home context attributes are regarded as control variables that provide a means of understanding casual effects in a manner that adjusts for the circumstances in which the pupil live in and thus the reduced form of the model is speci ed as:
Tis = Xi'1 + i [1]
Where Tis stands for test score of the ith pupil in subject s. Xi contains home related attributes,
I is a vector of constants to be estimated and i is the error term that follows Classical Linear Regression Model assumptions of zero mean and unity variance.
The vector of the predictor constitutes the variables in Block A. The second stage of the analysis involves adding another block of the variables successfully to the original regression model. The empirical strategy set out in this study requires additional attributes besides those stated in model [1]. The school system context variables are of interest to government and parents because of the interventions in the school system that can impact on both the school and/or classroom levels. This study therefore adds the school attributes which are considered relevant in the learning environment for the pupils and also this investigation has reduced the level of disaggregation. Therefore the study adopts the second modi ed structural regression equation which is stated as follows:
Tis = Xi'1 + Yi'2 + j [2]
Where Xi are the home related attributes and Yi are the school characteristics, ’1 and ’2 are the vectors of constants to be estimated. The error term j represent the un-captured attributes associated with home and school environments. The school related attributes considered for the study is given in Block A and B. At the third stage of regression model analysis, it
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
is hypothesized that the pupil education achievements are controlled by the attributes in the home context, school system context and home-school context. The third dimension related to home-school context attributes is of great interest because these variables describe elements of educational environment of home and school that are amenable through interventions made by parents, teachers, and school heads. The three block variable model to measure the effect of pupil learning achievements is stated as:
Tis = Xi'1 + Yi'2 + Zs'3 + k [3]
The Xi is the vector of home context variables, Yi comprises a block of school systems context attributes and Zs is a vector of home-school context attributes. The error term k is regarded to be uncorrelated with the independent variables and the vectors of coef cients ’1, ’2 and ’3 are to be estimated recursively using the OLS methods. The three equations as outlined above constitute multi-stage regression analysis to isolate home related effects, school contexts effects and home-school context effects. Goldstein (1995) raises concerns of possible statistical problems associated with modelling pupil outcomes without clustering the pupils to isolate the possible effects within schools, pupils and geographical location. However, the multi-stage analysis used in this study has isolated effects attributed to home background characteristics, school context and home-school context attributes. Additionally, in the Ugandan context, there is low correlation between pupil outcomes with in schools and classes, and as such, the multi- stage results were tested to be equally robust as the HLM results. The attributes related to home-school context are shown in Block C
4.4. Data
Sample: The study utilized SACMEQ data collected in 2007, which tested 5307 primary six pupils in 263 primary schools in Uganda. The data were collected using a strati ed two-stage cluster sample design. At the rst stage, schools were selected within regions with probability proportional to the number of pupils in the de ned target population and the second stage used a simple random sample of 20 pupils picked within each selected school.
Measures: The outcome variables of interest in the SACMEQ III project were pupil scores (on Rasch scales) in reading and mathematics tests at Grade 6 (in Uganda called primary 6) and were developed after careful curriculum mapping by a panel of subject specialists drawn from the school systems to identify those elements of curriculum outcomes that were considered important and ensured that they conformed to the national syllabi of SACMEQ countries. In addition, during the process of test development and before the tests were administered they were eld-tested in all SACMEQ school systems and their psychometric characteristics were examined using classical item analysis and Rasch analysis. The scale is such that the international mean was 500 with a standard deviation of 100. The independent variables included pupil (individual, and family)-level and school-level characteristics derived from the
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
questionnaires. The variables examined in this study are those variables identi ed as potential predictors of academic achievement following preliminary analyses, sound reasoning and research ndings from previous analyses of the SACMEQ I data, HSRC (1999) and SACMEQ II data Byamugisha (2004), and other researchers).
Data Analysis: In order to assess the effect of school environmental factors on learning achievements, the study undertook the analysis at two levels namely: Univariate and multivariate analysis. Under the univariate analysis, the frequencies or the distribution of the pupils were established across various disaggregates such as region, subject, gender and school location or type among others. Besides, the distributive properties of the variables used in the multivariate analysis were also established with the view to understand the underlying multivariate estimation techniques to be adopted. The multivariate analysis involved employing the multi-stage recursive models that established the possible causations between the pupil learning outcomes and the school environmental factors. The study used the STATA (version 10) for both univariate and multivariate analyses. In order to take care of multi-collinearity problem as alluded to by Gujarati (1988), the step wise estimation procedure was adopted. This analytical technique was used to retain the variables that were purely exogenous in order to facilitate model analysis and also produce statistically meaningful results upon which inference can be made.
5. Results
5.1. Performance outcomes
The results of the multi-stage OLS regression analyses are presented in Tables 1 and 2 for reading and mathematics models respectively, and test the effect on school environmental factors on learning achievements of primary schools in Uganda. All the three models for each subject considered have controlled for home related factors in order to rule out any possible unmeasured effects arising from background characteristics. According to reading results, model 1 control for only home related variables to evaluate their effect on pupil achievements. Model 2 controls for both school system context and home related attributes to evaluate their effect on pupil learning outcomes. The third model then controls for all the three learning environments of the pupil and thus becomes the basis for the inference and statistical hypotheses because it captures the net effects of all the forms of pupil behavioural approaches and support to the pupil to facilitate his or her learning achievements.
Results from model 1 indicate that the model explains 17 percent of the total variation implying that the home related variables contribute 17 percent towards learner’s achievement. The rest is attributed by other factors other than the home attributes. Model 2 controls for home related variables and the school system variables and explains 21 percent of the total variation. Model 3 controls for, in addition to home related attributes, school system context variables and home-school context variables and explains 25 percent of the variation.
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
Like the reading model results, results for mathematics are also represented by 3 models namely; home related variables, school related variables and both home – school related variables. Model 1 results indicate that only 9 percent of the variation is explained by the data. Model 2 introduces an additional school related variables as control to the model. As expected, R-squared increased to 12 percent implying that the total variation explained by the data has increased by 3 percent compared to the rst model. Like the reading results model 3 controls for home related factors, school system context attributes and school-home related variables gives a prediction of 14 percent.
5.2. Home related characteristics on learning outcomes
Only the results of the variables which are signi cant are explained below and nally discussed. Results indicate that pupils with 11 years scored 26.39 grade points in reading and 14.15 points in mathematics more as compared to those who were more than 11 years respectively. With regards pupil sex, results indicate male pupils perform better in reading and mathematics than girl pupils by 5.79 and 11.78 point scores respectively. An examination of the relationship between family size and pupils’ performance was undertaken. Results from this study indicate that pupils with smaller families had higher test scores for both reading and mathematics and had a reduction in performance in reading and mathematics by 0.73 and 0.93 point scores respectively. With regards to speaking English at home results demonstrate that pupil who often speak English at home perform better in reading and mathematics by 15.20 and 18.37 grade points as compared to those who do not, respectively.
With regards to parents education results indicate that pupils whose mothers have secondary level of education improved their academic performance in reading by 12.64 point scores as compared to mothers with no education. However, results were not significant for mathematics. And also the results for father’s education were not signi cant for both reading and mathematics. The above finding is supported by the works of Teachman, (1997) who indicated that the high levels of maternal education may be helpful to pupils as long as mothers work less than full time, thus allowing them more time at home in support of pupil learning achievements. Pupils were asked to indicate one of the main sources of lighting by which they can read in the place (home) where they stay during the school week and the options included,
re, candle, paraf n or oil lamp, gas lamp, electric lighting and no lighting at all. The results indicate that the effect of pupils using electricity at home while doing homework and revising improves on their reading performance by 9.87 grade points as compared to those who use other sources of energy. However, the impact of using electricity is positive and has no signi cant impact on mathematics grades.
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
− −
Table 1: Reading: Standardized OLS coefficients for relationships among home background, school context, home-school contexts and pupil’s learning achievement
3 - l e d o M 2
- l e d o M 1
- l e d o M s
e l b a i r a v r o t c i d e r P
Home related variables
*
*
* 9 3 . 6 2
*
*
* 1 8 . 8 2
*
*
* 2 0 . 1 4 )
s r a e y 1 1 ( l i p u p e h t f o e g A
*
*
* 7 7 . 4 1
*
*
* 0 1 . 9 1
*
*
* 5 9 . 7 2 )
s r a e y 2 1 ( l i p u p e h t f o e g A
*
*
* 0 5 . 8
*
*
* 6 1 . 0 1
*
*
* 7 4 . 4 1 )
s r a e y 3 1 ( l i p u p e h t f o e g A
Household size (number of siblings at home) -1.13*** -0.88*** -0.73**
*
*
* 9 7 . 5
*
*
* 5 9 . 5
*
*
* 3 0 . 7 )
1
= e l a M ( x e s l i p u P
*
*
* 0 2 . 5 1
*
*
* 1 1 . 6 1
*
*
* 9 7 . 9 1 )
1
= s e y ( e m o h t a h s i l g n E k a e p s l i p u P
8 0 . 0 5
0 . 0 6
0 . 0 e
m o h t a s k o o b f o r e b m u N
Pupil stay at home with parents 6.04** 5.22 4.79 0 1 . 0 8
2 . 0 - 2
4 . 1 - r
e h t o m f o n o i t a c u d e y r a m i r P
*
*
* 4 6 . 2 1
*
*
* 0 9 . 0 1
*
*
* 5 2 . 9 r
e h t o m f o n o i t a c u d e y r a d n o c e S
4 1 . 5 8
3 . 3 6
1 . 6 r
e h t o m f o n o i t a c u d e y r a i t r e T
9 3 . 2 - 9
8 . 1 - 4
6 . 4 - r
e h t a f f o n o i t a c u d e y r a m i r P
1 0 . 4 4
4 . 5 3
0 . 1 r
e h t a f f o n o i t a c u d e y r a d n o c e S
7 3 . 8
* 5 9 . 8 7
3 . 7 r
e h t a f f o n o i t a c u d e y r a i t r e T
*
* 7 8 . 9
*
* 5 0 . 0 1
*
*
* 4 8 . 2 3 )
1
= s e y ( e m o h t a y t i c i r t c e l e g n i v a H
5 7 . 2 - 2
5 . 2 -
*
* 7 1 . 4 )
1
= s e y ( y r a m i r p - e r p d e d n e t t A
0 3 . 0 5
4 . 0
*
*
* 1 1 . 2 e
m o h t a s n o i s s e s s o p l a t o T
School system context
3 7 . 4
* 3 7 . 5 )
1
= s e y ( l o o h c s t a h c n u l g n i v a h l i p u P
1 5 . 2 - 6
4 . 1 - )
1
= s e y ( r e t u p m o c a d e s u l i p u P
*
*
* 5 8 . 2 3 -
*
*
* 0 9 . 9 2 - )
1
= t n e m n r e v o g ( e p y t l o o h c S
*
*
* 0 0 . 1 2 -
*
*
* 7 2 . 4 2 - )
1
= l a r u r ( n o i t a c o l l o o h c S
Lower secondary education of school head 5.87 4.11
*
*
* 7 1 . 9
*
*
* 9 9 . 8 d
a e h l o o h c s f o n o i t a c u d e y r a i t r e T
Lower secondary education of class teacher 9.63*** 10.29***
4 0 . 0 1
0 . 0 - r
e h c a e t s s a l c f o n o i t a c u d e r e p p U
*
*
* 8 7 . 2
*
*
* 6 2 . 3 s
e c r u o s e r l o o h c S
Reading teacher access to teaching materials -0.70 -0.03
Days in a month class teacher has been absent -16.43*** -16.64*** 1 0 . 0 2
1 . 0 r
e h c a e t d a e h e h t f o e c i v r e s f o s r a e Y
Special training program of the head teacher (yes=1) 9.35*** 8.79*** School-home context
*
*
* 1 7 . 1 1 - )
1
= s e y ( x i s e d a r g g n i t a e p e r l i p u P
2 5 . 2 - e
m o h m o r f l o o h c s f o e c n a t s i D
Pupil borrows books from school library (yes=1) 4.09*
Pupil given home work at school (yes=1) 14.88***
Pupil helped with home work at home (yes=1) 0.52
*
*
* 2 3 . 1 4 y
l r a e y - s t n e r a p t e e m s r e h c a e T
*
*
* 4 0 . 5 2 y
l m r e t - s t n e r a p t e e m s r e h c a e T
*
*
* 9 2 . 9 1 y
l h t n o m - s t n e r a p t e e m s r e h c a e T
*
*
* 6 5 . 1 - t
n e s b a s i l i p u p s y a d f o r e b m u N
Parents paying tuition for extra coaching (yes=1) 10.26***
*
*
* 1 9 . 1 4 4
*
*
* 1 4 . 8 6 4
*
*
* 9 5 . 6 2 4 t
n a t s n o C
2 4 0 4 7
0 2 4 2
9 0 5 N
5 2 . 0 1
2 . 0 7
1 . 0 d
e r a u q s - R
) 0 0 . 0 ( 6 0 . 8 5 s
c i t s i t a t s -
F 35.79 (0.00) 35.59 (0.00)
Source: Estimated by Author (2010)
Note: 1. The dependent variable is mathematics.
2. Asterisks denote significance level; *** = 1percent, ** = 5 percent, * = 10 percent. 3. To test for consistency of results as alluded to by Vignoles et al. (2000, p.9), the OLS and
HLM results have been tested to be robustly the same as described by Goldstein (1995).
3Bࣇࣜ࢝ᩍ⫱◊✲SGI
