Factors influencing the academic achievement of the Turkish urban poor

International Journal of Educational Development 29 (2009) 17–29 Contents lists available at ScienceDirect International Journal of Educational Development journal homepage: www.elsevier.com/locate/ijedudev Factors influencing the academic achievement of the Turkish urban poor Cennet Engin-Demir * Middle East Technical University, Department of Educational Sciences, 06531 Ankara, Turkey A R T I C L E I N F O A B S T R A C T Keywords: Primary school Urban poor Academic achievement Gecekondu This study estimates the individual and combined effects of selected family, student and school characteristics on the academic achievement of poor, urban primary-school students in the Turkish context. Participants of the study consisted of 719 sixth, seventh, and eighth grade primary-school students from 23 schools in inner and outer city squatter settlements. The findings indicated that the set of variables comprising student characteristics, including well-being at school, scholastic activities and support, explained the largest amount of variance in academic achievement among the urban poor. Although the effect sizes are small, family background characteristics and school quality indicators were also found to be significantly related to academic achievement. The implications of this study for improving primary schools in urban poor neighborhoods are discussed. ß 2008 Elsevier Ltd. All rights reserved. 1. Introduction and purpose Empirical evidence clearly shows that education plays a significant role in influencing an individual’s economic and social circumstances, with formal schooling playing an important role in the enhancement of economic growth (Barro, 1997; Krueger and Lindahl, 2001). By increasing economically productive knowledge and skills (e.g. literacy, numeracy and problem solving skills), education increases individual productivity and thereby individual earnings (Psacharopoulos, 1994). Education is considered as a basic need that supports the fulfillment of other basic needs such as shelter, food, clothing and security and helps steady improvement of quality of life. Through its positive effects on earnings (Psacharopoulos, 1994) and on housing, water, sanitation, utilization of health facilities and women’s behavior in terms of decisions related to fertility, family welfare and health (Lleras-Muney, 2005), education has been regarded as an instrument for poverty reduction (Tilak, 2002). In this context, the increasing importance of educational experiences and achievements in shaping people’s opportunities, especially their ability to secure decent work, has significant implications for social policies in many countries (Machin, 2006). Learning is a product not only of formal schooling, but also of families, communities and peers. Social, economic and cultural forces affect learning and thus school achievement (Rothstein, 2000). A great deal of research on the determinants of school achievement has centered on the relative effects of home- and * Tel.: +90 312 210 4038; fax: +90 312 210 7967. E-mail address: cennet@metu.edu.tr. 0738-0593/$ – see front matter ß 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijedudev.2008.03.003 school-related factors. Most findings have suggested that family background is an important determinant of school outcomes, whereas school characteristics have minimal effects (Brooks-Gunn and Duncan, 1997; Coleman et al., 1966; Heyneman and Loxley, 1983); however, debates continue regarding the relative importance of family and school inputs (Chevalier and Lanot, 2002; Schiller et al., 2002). Various studies have shown that both home and school environments have a strong influence on the performance of children, especially at the primary-school level (Carron and Chau, 1996; Griffith, 1999; Mancebon and Mar Molinero, 2000). In addition to influences related to home and school, academic achievement is also affected by students’ preexisting human capital, which includes their unique way of interacting with each type of educational ‘‘institution’’, namely, family, community, school, peer group, the economy and the culture (Rothstein, 2000). Individual characteristics such as attitude towards school, perceptions of the school environment, involvement in scholastic activities and level of motivation have also been found to influence academic achievement (Connolly et al., 1998; Ma, 2001; Veenstra and Kuyper, 2004). A review of the literature reveals that studies investigating determinants of school achievement have focused on the relative importance of home- and school-related factors, whereas scholastic activities, student well-being in school, attitude towards school, family characteristics and school characteristics are rarely examined in the same study. The present study differs from earlier studies, especially those conducted in Turkey, in that it focuses simultaneously rather than separately on how home-, student- and school-related factors affect the school achievement of the urban poor. Therefore, the purpose of this study is to examine the relative importance of home-, individual- and school-related factors in 18 C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 accounting for academic achievement of poor, urban primaryschool students in Turkish context. The specific research questions in the study are as follows: 1. How well each set of home-student and school-related factors explain the variance in academic achievement over and above the other sets? 2. To what extent the combination of home-student and schoolrelated factors accounts for the variance in academic achievement? 2. Theoretical background 2.1. Family characteristics Research has consistently shown that family background characteristics such as socio-economic status (SES) as measured by parental education level, parental occupation and family income have an influence on school achievement (Coleman et al., 1966; Häkkinen et al., 2003; Heyneman and Loxley, 1983). Among SES indicators, parental level of education has been found to be the most significant source of disparities in student performance (Chevalier and Lanot, 2002; Fuchs and Wößmann, 2004; Guncer and Kose, 1993; Parcel and Dufur, 2001; Schiller et al., 2002; Willms and Somers, 2001; Yayan and Berberoğlu, 2004). Using PISA results, Fuchs and Wößmann (2004) concluded that the effects of parental education on reading achievement of 15-year-old students are greater than on math and science achievement. PISA 2000 results indicated that students whose mothers had completed their upper-secondary education achieved higher levels of performance in reading than other students in all participating countries (OECD, 2001). Hanushek and Luque (2003) reported that family background exerts a very strong effect on 9and 13-year-old students’ performance. Specifically, they found that students from disadvantaged families and families where parents had less education have systematically performed worse on the Third International Mathematics and Science Study (TIMSS) tests than other students. Schiller et al. (2002) have argued that regardless of national context, parents who have more education appear better able to provide their children with the academic and social support important for educational success when compared to parents with less education. Parents with higher levels of education also have greater access to a wide variety of economic and social resources (e.g. family structure, home environment, parent–child interaction) that can be drawn upon to help their children succeed in school (Coleman, 1988, 2006; Gregg and Machin, 1999; McNeal, 1999). Looking at Turkish TIMMS data, Yayan and Berberoğlu (2004) found that when parental education levels and numbers of books at home increased, eighth grade student achievement in mathematics also increased. Similarly, Thompson and Johnston (2006) used PISA results from 20 OECD countries to explore the relationship between non-school factors and student achievement. They found that students at the highest SES levels, as measured by the number of books in a student’s home, had an educational advantage over students at the lowest SES levels. A sizable body of evidence (Currie, 1995; Gregg and Machin, 1999) exists that indicates educational achievement is significantly lower among children from disadvantaged backgrounds characterized by poverty, low levels of parental education, negative parental attitudes and negative neighborhood characteristics. Furthermore, a number of studies have found that a child’s home environment – specifically, the existence of opportunities for learning, the warmth of mother– child interactions and the physical conditions of the home – accounts for a substantial portion of the measured effects of family income on cognitive outcomes of children (Brooks-Gunn and Duncan, 1997; Majoribanks, 1994). Higher family income is associated with higher student achievement in most of the studies (Hanushek, 1992; McLanahan and Sandefur, 1994; Peters and Mullis, 1997). However, whether the income effect is causal, or merely reflects the correlation of income and some observable characteristics of parents such as parental education, occupational status and parent–child interaction remains unclear in some studies (Chevalier and Lanot, 2002; Ganzach, 2000; Mayer, 1997). Smits and Gündüz-Hoşgör (2006) found that education of both parents, the number of brothers, whether or not mother was able to speak Turkish and father’s occupation were the major variables affecting educational participation of 9–11 years old Turkish girls. Parental education and income of the household had significant positive affect on educational participation of both boys and girls. Several studies have demonstrated that increased numbers of children in the family leads to less favorable child outcomes, presumably through the mechanism of resource dilution (Blake, 1989; Patrinos and Psacharopoulos, 1995). Resource dilution refers to the quantity of time and material resources that parents are able to invest in their children (Teachman et al., 1996); when the number of children increases, parents can offer fewer resources per child. Under such conditions, all forms of family capital – financial, human and social – are more finely spread across the children (Coleman, 1991). Again, empirical evidence supports these claims: children from larger families have been found to have less favorable home environments and lower levels of verbal facility (Parcel and Menaghan, 1994) as well as higher rates of behavior problems and lower levels of educational achievement (Downey, 1995). Cross-national research on the relative effects of home and school has indicated that the relationship between a child’s social background (e.g. parent’s education, family structure) and his or her academic achievement is stronger in developed nations than in developing nations, whereas school-related factors have been found to be more important than out-of-school factors in explaining achievement variance in developing countries (Coleman et al., 1966; Fuller and Clarke, 1994; Heyneman and Loxley, 1983). In contrast, Simmons and Alexander (1978) concluded that the determinants of student achievement appear to be basically the same in both developing and developed countries. Similarly, recent cross-national studies found variations in national levels of economic development had no affect on the relationship between children’s social background and their academic achievement (Baker et al., 2002; Hanushek and Luque, 2003). 2.2. Individual student characteristics Studies have shown that individual student characteristics such as student well-being, perception of the school environment, motivation, involvement in scholastic activities and student effort, gender, work and students’ perception of parental support and involvement all have significant effects on a student’s school achievement. In their School Well-being Model, Konu and Rimpelä (2002) conceptualized well-being in school as a four-dimensional phenomenon: school conditions, social relationships, means for self-fulfillment and health status. Research has shown that student well-being in school affects both their behavior and their examination results (Hoy and Hannum, 1997; Sabo, 1995). Well-being of students in school depends on many factors, including their opinions about school rules and their relations with teachers and schoolmates (Veenstra and Kuyper, 2004). Student well-being may also affect other student-related characteristics, such as achievement, motivation, attitude towards study and effort (Samdal, 1998; Veenstra and Kuyper, 2004). For C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 instance, late elementary school children who perceive their teachers to be fair and caring are more likely to have positive attitude towards school and increased motivation to achieve (Babad, 1996). Children’s perceptions of teacher support and teacher expectations have also been found to be positively related to achievement (Connolly et al., 1998). Moreover, children who are accepted by their peers have been found to be more likely to enjoy school (Verkuyten and Thıjs, 2002). Scholastic activities and individual effort are also important for achievement. Keith et al. (1986) showed that regardless of intelligence, as students spend more time on homework, their grades improve. In an international comparison study of science achievement in 23 countries, Postlethwaite and Wiley (1992) found that the average science achievement was high in those countries where students reported spending large amounts of time on homework. Amount of time spent on homework has also been found to be strongly related to a student’s motivation to achieve and the positive feelings associated with achievement, which themselves have been shown to have an affect on actual achievement (Steinberg et al., 1992). In addition, school attendance has been shown to be highly correlated with individual achievement (Hanushek et al., 1996). Gender is another important variable to be considered in explaining variations in school achievement. In a sequential regression analysis of gender differences, Farkas et al. (1990) concluded that when all other variables are held constant, girls receive higher course grades than boys. Evidence has also been found to indicate girls score higher than boys in terms of student well-being, achievement motivation and effort (homework time) (Connolly et al., 1998; Veenstra and Kuyper, 2004). On the other hand, girls have been shown to perform lower than boys in terms of math and science scores (Berberoğlu, 2004; Wößmann, 2003). The relationship between school performance and work is generally perceived to be negative. Balancing the demands of work and education places physical and psychosocial strain on children and often leads to poor academic performance (Binder and Scrogin, 1999; Heady, 2003). Akabayashi and Psacharopoulos (1999), found a child’s reading and math ability decreased with additional hours of work, whereas they increased with additional hours of school attendance and study. Ray and Lancaster (2003) investigated the effect of work on the school attendance and performance of children in the 12–14 year age group in seven countries particularly in terms of the relationship between hours of work and school attendance and performance. They concluded that hours spent at work had a negative impact on education variables with the marginal impact weakening at higher levels of work hours. An exception to this was in the case of Sri Lanka, where a weekly work load of up to (approximately) 12–15 h a week contributed positively to the child’s schooling and to his/her study time. Students’ perceptions of parental support and involvement are also considered as influential factors on their achievement motivation (Grolnick and Slowiaczek, 1994). Children’s perceptions that their parents are involved and interested in school and encourage them to do well are positively related to academic achievement (Wang and Wildman, 1995). Through their involvement, parents convey the message that school is important and provide their children with positive emotional experiences in relation to school. Children, in turn, internalize their parents’ positive expectations toward school and reflect them in their own school attitudes (Connolly et al., 1998). Fuchs and Wößmann (2004) observed that students performed significantly worse in reading, math and science in schools whose principals reported that learning was strongly hindered by the lack of parental support. Some research, however, has shown most aspects of the relation- 19 ship between educational support of parents and scholastic achievement of children to be negative (Muller and Kerbow, 1993; Sui-Chu and Willms, 1996). It may be that parents, in an effort at what might be termed ‘‘crisis-intervention’’, increase help with homework, discuss grades more often and contact teachers more frequently when their child’s school achievement drops or when their child has a discipline problem (Veenstra and Kuyper, 2004). 2.3. School characteristics Findings reported in the literature regarding the relationship between school resources and student achievement measurements are inconsistent. While some research has suggested that more resources do not necessarily yield performance gains for students (Hanushek, 1997; Hanushek and Luque, 2003; Häkkinen et al., 2003), other research provides clear evidence that variations in school characteristics are associated with variations in student outcomes (Card and Krueger, 1996; Greenwald et al., 1996; Lockheed and Verspoor, 1991). Parcel and Dufur (2001) have demonstrated that attending a school with a better physical environment is associated with increased math scores. Willms and Somers (2001) reported a significant positive effect on schooling outcomes associated with student–teacher ratio, instructional materials, size of the library and teacher training in 13 Latin American countries. Large-scale studies in low-income countries have emphasized the importance of human and material resources in achieving better schooling outcomes, including such factors as school infrastructure, class size, teacher experience and qualifications and availability of instructional materials (Fuller and Clarke, 1994; Heyneman and Loxley, 1983). However, in a study examining the role of schools and proxies for school quality in explaining increases in student achievement levels in a developing country, Bacolod and Tobias (2005) concluded that school characteristics explained only six percent of total variations in school achievement. Among the various school characteristics, class size has been the most widely examined variable in educational policy studies; however, findings regarding the effects of class size on school achievement are inconsistent. Contrary to expectations, Wößmann (2003) found that smaller class size was significantly related to inferior student performance in math and science in 39 countries participating in the 1994/1995 TIMSS, whereas Lindahl (2005) found that some minorities and economically disadvantaged groups in Sweden benefited from smaller classes. Lindahl’s results are in line with those of Angrist and Lavy (1999) for Israel and Krueger (1999) for the United States, but they conflict with those of Hoxby (2000), whose longitudinal study of 649 elementary schools in the U.S. indicated that class size had no statistically significant effect on student achievement. Another recent study conducted by Rivkin et al. (2005) with three through eighth grade students in Texas demonstrated that the effects of class size on math and reading achievement growth, while statistically significant, were modest and declined as students progressed through school. However in seventh grade students did not get any significant benefit from smaller classes in mathematics and reading. Although some studies have shown a positive effect of class size on academic achievement, especially in lower grades (Rivkin et al., 2005; Lindahl, 2005; Krueger, 1999) there has been no critical class size suggested by the researchers that increased academic achievement (Ehrenberg et al., 2001). In addition to class size, teacher–student ratio is another variable widely used as an index of school quality; however, results of studies looking at the effects of teacher–student ratios on school achievement are also ambiguous (Hanushek et al., 1996; Fuchs and 20 C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 Wößmann, 2004). International comparisons have failed to show any significant improvements in academic achievement as a result of smaller teacher–student ratios (Wößmann, 2003). Recent studies have highlighted the contribution of teacher quality to academic achievement (Darling-Hammond, 2000; Fetler, 2001; Rivkin et al., 2005). As determined by DarlingHammond and Sykes (2003), key teacher quality components are verbal ability, subject-matter knowledge, knowledge of teaching and learning and ability to use a wide range of teaching strategies adapted to student needs. Wiseman and Brown (2002) found that teacher education levels are positively related to student performance. That is, as teachers attain higher levels of education, meet standards for certification in their communities and spend more of their pre-service preparation time actually teaching, their students tend to score higher on standardized achievement tests. On the other hand, other studies have shown teacher degree level to be a relatively unimportant predictor of school outcomes (Goldhaber and Brewer, 1996; Rivkin et al., 2005). 3. Method 3.1. Main characteristics of urban poor This study was conducted in urban areas (population 20,001+) within the Greater Ankara Municipality. As the capital and the second largest city in the nation, Ankara attracts large numbers of migrant families from Central and Eastern Anatolia; however, it lacks the employment potential to adequately absorb these incoming families. The subsequent mismatch between population and urban physical infrastructure and basic services has led to the ‘gecekondu’ phenomenon, i.e. the construction of illegal squatter housing built by migrants along the outskirts of major cities. Most of the urban poor in large cities live in such gecekondu settlements, while some live in similar squatter housing in inner city neighborhoods. Gecekondu inhabitants can be distinguished from the established city population by the following distinctive characteristics: ‘‘stronger ties to the village in comparison with established urbanites; membership of the city’s lower classes (low-income, low skilled jobs and low education levels, informal housing); and lives centered around the communities that they have formed in the city’’ (Erman, 2001, p. 995). It is these characteristics of gecekondu dwellers, researchers argue, that have created a sub-culture within the city (Erder, 1995; Kongar, 1999). As part of their everyday existence, the urban poor live with certain basic problems such as inadequate and unsanitary living conditions, including a lack of clean water and air pollution (Keleş, 2000). There is no doubt that children of poor families grow up in circumstances worse than children whose family income level is sufficient to meet their daily needs. While one might believe the availability of free schools in poor neighborhoods provides children of the urban underclass an opportunity to receive an education and thus improve their possibilities for the future, the fact is that education that is free in theory in practice requires families to pay for school uniforms, notebooks and lunches and contribute funds solicited by school administrations, thus preventing some poor families from enrolling their children in school (Bugra and Keyder, 2003). Moreover, schools serving poor children in villages and in urban gecekondu neighborhoods tend to be lowperforming and staffed with less-experienced, poorly trained teachers with low morale and low expectations. Classrooms may be inadequately supplied with reference material, books and other learning material, and the buildings themselves may be characterized by unsafe environments that lack adequate lighting or even functioning toilets. While better endowed schools may exist outside these neighborhoods, poor children do not have the opportunity to enroll in them. (World Bank, 2005). Yet, despite the poor quality of education offered to gecekondu residents, studies have indicated that almost all of them have positive attitudes and high aspirations regarding their children’s education; they assume their children will eventually go on to complete a university-level education, thereby securing improved living conditions in better neighborhoods (Erder, 1995; Kongar, 1999). 3.2. Turkish education system Children in the Turkish education system may enter noncompulsory pre-school education at the age of three. Compulsory primary education begins at age six or seven and comprises eight years of schooling, after which children may attend secondary education (four-year general high school or vocational high school or technical high school) or vocational education (apprenticeship training). Secondary education was extended from three to four years in 2005. Although efforts are being made, within the framework of European Union membership negotiations, to extend the length of compulsory education to 12 years, secondary education is currently not mandatory (Ministry of Education (MONE), 2007). The government runs and finances free public schools, and MONE dictates the educational curricula of both public and private pre-school, primary and secondary schools. Primary and some of the secondary school teachers (e.g. English language teachers, Computer education and instructional technology teachers) are trained in four-year faculties of education. Secondary school teachers are trained in two main programs: First, the five-year undergraduate program for the students of faculties of education. Second, a master of science program without thesis offered by faculties of education for students who graduate from four-year science and letters and other relevant faculties. 3.3. Participants The data used in this study are part of a larger research project on light work and schooling. ‘‘Light work’’ is defined as work that does not interfere with schooling and it is not exploitative, harmful or hazardous to a child’s development (International Labor Organization (ILO), 2002). For the main study a Multi-stage Stratified Systematic Random Cluster Sampling method was used in sample selection of schools, children who combine school and work, children who are currently in school and not working. Initially a total of 200 schools were selected from within the six districts of Greater Ankara Municipality based on the published information and opinions of an urban sociologist and an expert from ILO Ankara office about SES level of dwellers. It was assumed that the number of children who combine school and work might be higher in inner city and other gecekondu neighbourhoods. Each of the selected schools was contacted by phone and requested to provide researchers with a list containing information on the approximate numbers of working children and total number of male and female students at their schools. These lists were used to aid in the selection of schools from among the six districts to serve as the second stage sampling frame which was used in the selection of 25 schools based on the number of working children using probability random selection. A listing form containing questions on children’s sex, age, work status, family socioeconomic status and neighbourhood developmental level form was administered to all sixth, seventh and eighth grade students in selected 25 schools. Information collected from 10,040 students through listing form was compiled and used as the sampling frame for the third stage of sample selection. Two schools were excluded due to the very low number of working children enrolled, as indicated by the listing study. At the final stage of sample selection, C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 50 students were selected randomly from each school with more than 50 working children using stratification criteria. In schools with fewer than 50 working children, all working children were selected. Students who were currently in school and not working were selected randomly from each stratum. A total of 1095 students (672 working and 423 non-working) were selected from 23 schools. For the present study 296 working students from among 672 were selected by using proportional stratified systematic sampling and all of the non-working students were included in the sample. Study participants consisted of 719 (441 male, 278 female) sixth-, seventh- and eighth-grade primary-school students from 23 schools in inner city squatter settlements and other gecekondu neighborhoods within the Greater Ankara Municipality. The number of male and female students was not equal in the sample due to the low proportion of female students among working children. 3.4. Procedure To ensure reasonable and accurate information capable of generating sufficiently valid conclusions, face-to-face structured interviews were conducted with interview schedules (questionnaires) developed by the research team. Both working and nonworking students were asked same questions. However, additional work related questions such as type of work, weekly hours of work, reason of starting to work were asked to working children. The common questions included in the Student Questionnaires were related to student perceptions regarding scholastic activities, peer support, teacher support and parental support as well as questions related to household characteristics and household possessions. Principals were interviewed using a questionnaire developed by the research team and designed to collect information on school-quality indicators. The questionnaire comprised questions related to different school characteristics, including number of students, teachers and classrooms; teacher–student ratios; class size; availability of a school counselor; teacher characteristics, such as number of teachers who participated in MONE in-service training during the previous year, number of teachers teaching outside their subject area, number of teachers with a graduate or post-graduate degree or currently enrolled in a graduate program; whether or not the lack of teacher availability prevented any class from being offered; and questions related to school facilities. The Turkish language, math, and science scores as well as attendance records for each student interviewed were obtained from the school records by the interviewers. A pilot study was conducted to test the validity of the questionnaires and to assess the data collection procedures in two schools that were not included in the main sample. As a result of the knowledge and experience gained from the pilot study, several changes were made to improve the survey instruments and to finalize a work plan for field implementation of the data collection for the actual study. For example, it was observed that some students had difficulty in answering the questions in the listing forms and the questionnaires because they lacked competency in certain basic skills such as reading and writing. Therefore, it was decided to use face-to-face structured interview method to ensure the validity and accuracy of the data. In addition, questions in the initial listing form were reworded and the listing form was redesigned to make it more attractive to children. Questions on the Student Questionnaires were also revised to improve clarity and coherence. Interviewers were selected from among postgraduate and senior sociology students at Ankara University. Interviewers were 21 requested to attend a one-day training session prior to data collection. The training included information about the purpose of the study, data collection techniques and procedures, characteristics of school settings, characteristics of adolescents between the ages of 12–14 and the purpose of each question asked in the data collection instruments. A detailed field implementation work plan was prepared in advance and given to the interviewers. At the end of each day the interviewers brought the completed questionnaires to the co-ordinator’s office and discuss any difficulties encountered and presented suggestions as solutions. Coordinator also conducted random checks of questionnaires providing feedback to the interviewers regarding the quality of data collected and, if necessary requesting repeat interviews. Data collection process lasted about 40 days. 3.5. Measures 3.5.1. Academic achievement A weighted composite of mathematics, Turkish and science scores (weighted .35, .35 and .30, respectively) from the same semester obtained from school administrative records was used to measure academic achievement. This decision was based on the fact that, Turkish, math and science are allocated the greatest number of weekly hours (6 h/wk for Turkish, 4 h/wk for math and science) in the curriculum and because a strong (Field, 2005) and significant correlation was observed between these three scores. The correlation coefficients between Turkish and math, Turkish and science and math and science are .61, .60 and .59, respectively (p  .001). In Turkish schools grading ranges from 1 to 5 (1 = Fail; 2 = Pass; 3 = Moderate; 4 = Good; 5 = Excellent). Students’ overall GPA could not be obtained because of the time of the study. Data were collected during December and students were asked to indicate whether they engaged in any work activity during the previous (reference) month. 3.6. Independent variables Independent variables in this study were selected based on both theoretical and empirical considerations, including findings from previous research, data availability, data comparability. Three sets of independent variables – family characteristics, individual student characteristics and school characteristics – drawn from the Student and Principal Questionnaires were used in this study. Dummy coding was used to represent the categorical independent variables. Definitions of variables used and descriptive statistics are presented in Table 1. The first set of variables comprising family characteristics (family background) included variables for father’s education level, mother’s education level, home ownership, household size and household possessions (Table 1). The decision to include a variable for household size was based on the strong correlation (r = .75, p  .001) between number of siblings and total number of household members. Since the high percentage of urban poor working in the informal sector and the large number of family members working temporary jobs made it difficult to obtain reliable information on family income from students, income was not included as an indicator of family SES. Instead, home ownership and household possessions were used as indicators of family income. The second set of variables comprising individual student characteristics included variables related to some background characteristics such as grade, gender, work status and attendance level, as well as variables related to student well-being at school (perceptions of teacher support and peer support), scholastic 22 C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 Table 1 The variable definitions, percentages, means and standard deviations (N = 719) Variables Definitions Academic achievement A weighed composite of the math. Turkish and science achievement scores of related semester (weighted .35. .35, .30, respectively). Grading range: 1 = fail, 2 = pass, 3 = moderate, 4 = good, 5 = excellent Family characteristics Father education Mother education Household size Home ownership Household possessions a Student characteristics Grade 7 Grade 8 Work status Gender Student’s perception of treatment by teachers Number of friends at school Participation in extracurricular activities Time spent on studies Time spent on leisure activities Level of homework completion Getting help with studies outside school Parents level of follow-up Attendance level School quality indicators Teacher degree level Teacher in-service training Teacher–student ratio Classize School’s facilitiesb Percent Mean S.D. – 2.1846 1.0633 1 = has a secondary level education or higher, 0 = otherwise 1 = has a secondary level education or higher, 0 = otherwise Total number of people living in the dwelling Where do you live? 1 = our own home. 0 = otherwise A summary of value was composed by summing the number of 1 s over each area (with a range of 1–13), available = 1, not available = 0 17.1 7.2 – – 5.0250 – 9.4214 – – 1.3648 – 1.7597 Students is at grade 7 = 1, 0 = otherwise (grade 6 is reference) Students is at grade 8 = 1, 0 = otherwise (grade 6 is reference) Whether the child combine school and work (1) or not (0) Male = 1 How do your teachers treat you? 1 = very good, 2 = good, 3 = moderate, 4 = bad and very bad How many friends do you have in school? 1 = many, 0 = otherwise Do you participate in any extracurricular activities at school? 1 = yes, 0 = no 33.5 34.5 41.2 61.3 – – – – 3.1238 – – – – – – Total hours spent on studying including time spent during the school week and on weekends Total hours spent on sports/play and other leisure activities including time spent during the school week and on weekends How often do you do your homeworks? 3 = often, 2 = sometimes, 1 = seldom and never Is there anyone who helps you with your studies after school? 1 = yes, 0 = no Do your parents visit your school to ask about your progress? 3 = often, 2 = sometimes, 1 = not at all Number of days child not attended school during the first semester Number of teacher with a master or Ph.D. degree or currently undertaking graduate study Number of teachers participating in a MONE in-service training event during the past year Division of school enrollment by the number of teachers Principals perception of average class size in school A summary of value was composed by summing the number of 1 s over each area (with a range of 1–13), available = 1, not available = 0 54 – 70 79.3 15.165 23.022 54.5 2.7830 – 1.8748 .55853 7.4165 11.873 .45425 – .65574 2.114 2.8701 2.8108 6.7761 30.695 35.489 7.3032 1.6531 9.4575 14.447 6.6827 2.6128 a Bathroom, toilet-inside house, kitchen, central heating, running water, electricity, TV set, refrigerator, dishwasher, oven, washing machine, vacuum cleaner, personal computer. b Library, sport facilities, science lab, computer lab, audiovisual materials, cafeteria, medical room, workshop or art room, school clubs, language lab, Internet connection, playground for students, transportation. activities, and perceptions of parental support (Table 1). Work was measured in relation to a reference month during the school year. One hour of work during the reference month was considered sufficient for classifying a child as engaged in economic activity. Definition of work encompasses all market production (paid work) and certain types of non-market production (unpaid work). For example, children engaged in unpaid activities in a marketoriented establishment operated by a relative living in the same household were considered to be working in an economic activity. The third set of variable school characteristics (school-quality indicators) included variables for teacher–student ratio, class size, teacher in-service training, teacher degree level and school infrastructure (Table 1). The decision to include a variable for teacher–student ratio was based on the strong correlation (r = .70, p  .001) between school size and teacher–student ratio. may be limited in their generalizability. The present study was conducted in a single metropolitan area in Turkey. Thus the extent to which results apply to other cities is not known. Therefore, conclusions need to be verified by conducting similar studies across other large cities in Turkey. Third, academic achievement was measured using grades given by subject teachers in language, math and science in the same semester; in order for these results to be generalized with confidence to academic achievement in general, this research needs to be extended to include standardized achievement test scores. Finally, as with any cross-sectional studies such as this one the results should be viewed with caution. The data collected in this study did not include students’ academic history. 4. Limitations of the study Multiple regression analysis was conducted to examine how well each set of variables – family, students and school characteristics – predicted academic achievement over and above the other sets (Green et al., 2000). The models also suggested how academic achievement was affected when family, students and school characteristics were combined. The significance of a set of independent variables was tested by examining the increment of R2 for the set over and above the R2 for those sets entered earlier. Several study limitations deserve further attention in future research. First, because the data utilized were derived from a larger study and not created specifically for this type of study or to answer all the questions raised within the framework of this research, not all the variables necessary were available; therefore, there is a potential for omitted variable bias. Second, study results 5. Results 23 C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 Table 2 Standardized regression coefficients for relations between family background characteristics, student characteristics, school quality indicators and academic achievement Predictors Academic achievement Model I Family background characteristics Father education Mother education Household size Home ownership Household possessions Model II b t .143 .000 .009 .103 .098 3.573*** .002 .238 2.787** 2.473* Student characteristics Grade 7 Grade 8 Work status Gender Student’s perception of treatment by teachers Number of friends at school Participation in extra-curricular activities Time spent on studies Time spent on leisure activities Level of homework completion Getting help with studies outside school Parents level of follow-up Attendance level b .23*** .048 .054 .054 b t .106 .039 .010 .105 .051 2.788** 1.053 .288 3.004** 1.364 .089 .048 .006 .087 .035 2.372* 1.314 .188 2.499* .893 .049 .013 .024 .179 .138 .101 .014 .093 .031 .059 .084 .024 .076 1.211 .312 .640 4.623*** 3.766*** 2.919** .413 2.504* .862 1.596 2.409* .702 2.115* .029 .001 .029 .173 .133 .097 .037 .075 .028 .060 .081 .025 .073 .734 .024 .789 4.563*** 3.716*** 2.859** 1.072 2.059* .800 1.653 2.356* .724 2.067* .132 .118 .144 .035 .054 3.615*** 2.835** 3.704*** .989 1.439 School quality indicators Teacher degree level Teacher in-service training Teacher–student ratio Classize School’s facilities Multiple R Adjusted 100R2 Effect size (R2) R2 change Model III t .45*** .18 .21 .15 .50*** .22 .25 .043 N = 719. * p < .05. ** p < .01. *** p < .001. Table 2 shows the standardized regression coefficients and t-test statistics for relations between family, student and school characteristics and academic achievement. The causal priority of three sets of variables – variables related to home, student and school – in explaining academic achievement was determined considering the literature review as follows: (a) Family SES influence student characteristics such as work status, their values and attitudes toward school. (b) Students’ perceptions of school environment influence school quality indicators. Simultaneous-entry approach – all sets of explanatory variables were entered into the model regardless of significance levels – was selected because of the study aim of developing a comprehensive picture of the different factors contributing to the explanation of the variances in the academic achievement of urban poor primaryschool students in the Turkish context (Green et al., 2000). The assumptions of the regression model were checked. As the zero-order correlations coefficients among independent variables are all less than .34, the Variance Inflation Factors (VIF) values changed between 1.008 and 1.131 and tolerance statistics changed between .884 and .992, there is no evidence to suggest the final model specification suffered from any multicollinearity that would challenge the findings. That is to say, there is no strong correlation between two or more predictors in the regression model. The Durbin–Watson statistic is also between one and three (1.620) implying that errors in regression are independent (Tabachnick and Fidell, 1996). Standardized residuals were examined to detect the availability of outliers. Two cases were determined as having standardized residuals 3.3 and 3.0. As none of those two cases had a Cook’s distance – a measure of the overall influence of a case on the model – greater than one and sample size is large none of them was having undue influence on the regression model (Field, 2005). The assumptions of normality, linearity and homescedasticity were checked by looking standardized residuals scatterplots to examine whether residuals are normally distributed about the predicted achievement scores, the residuals have straight line relationship with predicted achievement scores and the variance of residuals about predicted achievement scores is the same for all predicted scores. It was found that all the assumptions were met (Tabachnick and Fidell, 1996). The first equation included as predictors only those variables associated with family background characteristics. Regression analysis showed family characteristics were significantly related to school achievement (R2 = .054; adjusted R2 = .048; F(5,713) = 8.204; p < .001. The multiple regression correlation coefficient of Model I was .23. Standardized regression coefficients in Model I indicated that father’s level of education, home ownership and household possessions independently had significant effects on academic achievement, whereas household size and mother’s level of education independently did not have significant effects (Table 2). In the second equation, the second set of variables – student characteristics – was added to the set of family background 24 C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 characteristic variables. Regression analysis showed the linear combination of family and student characteristics to be significantly related to school achievement (R2 = .21; adjusted R2 = .18; F(13,700) = 10.172; p < .001). The set of student characteristic variables accounted for 15 percent of the total variation in academic achievement (R2 change = .15). Standardized regression coefficients in Model II indicated that student perceptions of treatment by teachers, student perceptions of number of friends at school, sex (being male), total hours spent on studies per week, getting help with studies outside school and total number of days absent from school independently had significant effects on academic achievement of urban poor primary-school students. In contrast, this study found that whether or not the student combine school and work, grade level, student’s participation in an extracurricular activity, level of homework completion, time spent on leisure activities, and student’s perception of his/her parents level follow-up did not contribute significantly to variations in academic achievement. The absolute t-values associated with Model II revealed that gender was the most substantial predictor of academic achievement. Boys perform on average .17 more poorly than do girls. Student’s perception of treatment by teachers was also found to be an important predictor of academic achievement (Table 2). In the third equation, the third set of variables – school-quality indicators – was added to the sets of family and student characteristic variables. Regression analysis showed the linear combination of family, student and school characteristics to be significantly related to academic achievement (R2 = .25; adjusted R2 = .22; F(5,695) = 7.881; p < .001). The set of school-related variables accounted for 4.3 percent of the total variation in academic achievement (R2 change = .043). The multiple regression correlation coefficient of Model III between combined family, student and school predictors and academic achievement was .50. Standardized regression coefficients in Model III indicated that teacher–student ratio, in-service teacher training and teacher’s level of education independently had significant effects on academic achievement, whereas school facilities and class size did not have significant effects. The absolute t-values associated with Model III indicated that teacher–student ratio was the most substantial predictor of academic achievement. Teacher degree level was also found an important predictor of student achievement (Table 2). Analysis indicated that when entered into the regression equation with student characteristics and school quality indicators the contribution of the variable ‘‘household possessions’’ became non-significant. The final R of .50, an index of the goodness-of-fit of the final regression model (Model III), demonstrated that the overall fit of the academic achievement model to data was acceptable. In sum, the most important set of predictors was found to be the set of student characteristics, which accounted for 15 percent of variance in academic achievement. The set of family background characteristics accounted for 5.4 percent of variance and the set of school-quality indicators accounted for 4.3 percent. The combination family, student and school characteristics (Model III) accounted for .25 percent of variance in academic achievement. 6. Discussion 6.1. Family characteristics Family background accounted for 5.4 percent of the total variations in academic achievement among urban poor students, which, although small, is still significant. Among the family background variables examined in this study, educational level of fathers (at least a secondary education) as one of the SES indicators had statistically significant and positive unique effect on variations in academic achievement among the urban poor. That is, students whose fathers have at least secondary or higher level of education tend to have higher academic achievement. This result confirms those of other studies on the effects of family SES on educational outcomes of children (Fuchs and Wößmann, 2004; Guncer and Kose, 1993; Parcel and Dufur, 2001; McEwan and Marshall, 2004; Schiller et al., 2002; Willms and Somers, 2001; Wößmann, 2003; Yayan and Berberoğlu, 2004). Another study in the Turkish context by Guncer and Kose (1993) found that family background as measured by father’s educational level accounted for more variance than school-related factors on the academic achievement of Turkish high school students. The finding that family income as measured by home ownership and household possessions had a significant and positive effect on academic achievement was also consistent with the results of other studies that found both parental education levels and home ownership had significant effects on children’s academic achievement in Turkey (Tansel, 2002; Tansel and Bircan, 2006) and other countries (Al-Nhar, 1999; Bacolod and Tobias, 2005). One interesting finding from this study with regard to family characteristics is that educational level of the mother (at least a secondary education) did not contribute significantly to explaining variations in academic achievement. This conflicts with PISA 2000 results, which suggested that students whose mothers have completed upper-secondary education attain higher levels of reading performance in all participating countries (Fuchs and Wößmann, 2004). It is, however, in line with a study by Ma (1997), which found that educational levels of mothers in the Dominic Republic, a developing country like Turkey, did not have a statistically significant effect on math achievement. It should be noted that levels of education are in general quite low among urban poor women in Turkey (according to the Student Questionnaire, 53.5 percent of mothers had completed only five years of primary school and 12.8 percent were illiterate), as are the rates of women wage-earners (General Directorate of Women’s Status and Problems, 1999). Thus, it may be argued that the finding of a lack of correlation between academic achievement and mother’s educational level is due to the similar SES among the mothers of students included in this study. This, in turn, implies that mother’s educational level affects academic achievement through the mechanism of SES, or, in other words, that mother’s education level is strongly moderated by mother’s SES. In light of this finding, future studies utilizing regression equations should include variables related to employment and occupation of mothers to examine their effects on school achievement. Results of this study revealed that household size did not have a significant effect on academic achievement, suggesting that variation in the number of siblings does not affect the amount of attention provided by parents per individual child with respect to school work. This result is consistent with results of case studies conducted in developing countries that found that number of siblings does not contribute substantially or has a positive effect on children’s educational outcomes in developing countries (Buchmann and Hannum, 2001). In contrast, studies conducted in the United States and some European countries (Downey, 2001; Parcel and Menaghan, 1994) have consistently found a negative association between large numbers of siblings and individual educational outcomes. 6.2. Student characteristics Results of this study revealed that the set of variables categorized as student characteristics significantly accounted for 15 percent of the variation in the academic achievement of urban C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 poor primary-school students in the Turkish context. Perceptions of treatment by teachers had strong and positive unique effects on academic achievement from among the set of student-related variables (Table 2). Students who describe treatment of their teachers as ‘very good’ and ‘good’ tend to have higher achievement scores than students who describe treatment of their teachers as ‘bad’ and ‘very bad’. Student perceptions of number of friends they have in school also had a significant and positive unique effect on student achievement. Considering these two variables are related to student well-being at school, these findings are in line with the literature. As mentioned earlier, well-being at school includes student perceptions of relationships with teachers and schoolmates as well as opinions regarding school rules (Hoy and Hannum, 1997; Samdal, 1998; Samdal et al., 1999). Researchers have stated that student well-being at school affects behavior and exam results as well as work attitude, achievement motivation and effort, and thus school performance (Babad, 1996; Connolly et al., 1998). Other studies have also reported great positive effects of school climate, especially the teacher–student relationship, on student achievement and attitude towards school in schools with disadvantaged students, including economically disadvantaged students (Schouse, 1996; Griffith, 1999). One possible explanation of this result is that a teacher–student relationship and school environment characterized by warmth and care may meet the urban poor student’s unmet needs for intimacy, warmth and support. This study showed gender to have the strongest unique effect on student achievement levels. The finding that girls had better average achievement scores than boys is consistent with the literature, which shows that although boys outperform girls on math and science tests (Farkas et al., 1990), girls receive higher course grades than boys (Sammons, 1995; Veenstra and Kuyper, 2004; Van Houtte, 2004). This trend has also been observed among Turkish students (MONE, 2005). It is possible to argue that the continuing perception among villagers and the urban poor of education as a privilege for girls also increases its perceived value for girls, thus increasing their achievement motivation and effort. In line with expectations, total hours spent on studies per week were found to have a strong positive effect on academic achievement. However, student characteristic ‘‘homework completion’’, was not found to be statistically significant. This result contradicts the results of other research which found the amount of homework completed by students to be an important predictor of school achievement (Postlethwaite and Wiley, 1992; Keith et al., 1986; Simmons and Alexander, 1978). A significant negative relationship was found between academic achievement and availability of someone to offer help with homework. This result is consistent with a study by Epstein (1987) that found a negative relationship between pupil achievement in math and reading test scores and parental help with homework. This negative relationship might be an indication that rather than helping regularly, parents or older siblings help only when a child needs help. Muller (1998) points out that parental response to adolescent needs in terms of schoolwork is context-specific, that is, parents tend to respond when a child needs help. It may also be argued that because of their own low levels of education, gecekondu parents may lack the knowledge and skills needed to assist their children with homework and stimulate them intellectually. When these factors are taken into consideration, it becomes clear that more detailed information is needed regarding the nature and extent of the help a child receives from family members. In terms of perceived parental support and involvement, student perceptions of the level of parental follow-up regarding a child’s academic progress was not found to contribute significantly to variations in academic achievement. This result 25 is consistent with the findings of Fine and Cook (1993), showing parental involvement in parent–teacher organizations (PTOs) in low-income communities had no significant impact on student achievement. Similarly Aypay (2003) concluded that parent– school relations did not have a significant effect on achievement and school choice of 14-year-old Turkish students. Research seems to indicate that the effects of parental involvement on student academic achievement depend on both school characteristics and the nature of parental involvement. For instance, Lee (1993) found personal contacts between parents and teachers to be associated with poor student performance and behavioral problems. When students are having trouble with school, their parents are more likely to become involved by maintaining contact with the school (McNeal, 1999). In Turkey, especially among the urban poor, parental involvement, including checking on student progress, is quite low. Only 16 percent of students interviewed in this study reported that their parents visited school often. One reason for this that has often been mentioned by principals and teachers in Turkey is that parents are frequently asked by school administrators to make financial contributions they are unable to afford, ostensibly to cover school expenses, despite the fact that education is provided free by the state. Parents’ own low levels of education may also lead them to leave the education of their children up to the schools; research has shown that parents with higher educational levels frequently manifest greater confidence in their ability to support their school-age children academically (Harmon et al., 1997 cited in Sukon and Jawahir, 2005, p. 554). Another expected finding was that student attendance level, measured by total number of days a student was absent during the semester, had a significant negative effect on academic achievement. The other student characteristics such as work status, grade level, total hours spent on leisure activities per week and level of participation in extra-curricular activities, were not found to have any significant effect on student academic achievement. Contrary to the literature regarding the relationship between work and academic performance (Akabayashi and Psacharopoulos, 1999; Heady, 2003) whether the student combine school and work was not found to have significant effect on school achievement. This might be due to the nature of work engaged by working students in this study. That is, nearly two-thirds of working students interviewed were unpaid family workers (e.g. working in family owned shop and other economic activities at home). Working in a family owned shop may limit the hazards associated with work outside home, allow students to attend the school more regularly and spent more time for their studies. 6.3. School characteristics The findings of this study revealed school characteristics to have a significant independent effect on the academic achievement of the urban poor. Variations in school quality as measured by school facilities, teacher–student ratio, class size, in-service teacher training and teacher education levels were found to account for 4.3 percent of the variation in academic achievement among students. Interestingly, teacher–student ratio was positively correlated with achievement scores. The finding that achievement scores increase as the number of students per teacher increases was in line with Fuchs and Wößmann (2004), whose analysis of PISA 2000 results showed a significant positive correlation between teacher– student ratio and student performance, but in conflict with other research showing a significant negative relationship between teacher–student ratio and school achievement (Guncer and Kose, 1993; Willms and Somers, 2001). A significant positive relationship 26 C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 between teacher–student ratio and academic achievement may be due to the greater availability of resources, such as better teaching aids and physical equipment and facilities, at larger schools. With regard to this study, it should be noted that the average teacher– student ratio (31) of the participating schools is above the figure suggested by educators for an effective school. Although this study found teacher–student ratio to have a significant effect on student academic achievement, class size (35) was not found to have a significant effect. This result is consistent with Hanushek and Luque’s (2003) findings regarding the effects of class size on performance in various countries. According to these researchers, the estimated effects of reduction in class-size are not systematically larger on science and math performance in poorer countries. In another study Rivkin et al. (2005) concluded that although class size had little impact on the achievement of children not from low-income families, it had a positive effect on the math and reading achievement of low-income children in the fourth and fifth grades. In light of this conclusion, the authors suggested that policies aimed at reducing class size for the fourth, fifth and sixth grades for non-low-income children would not be cost-effective. Fuchs and Wößmann (2004) have argued that while better equipment and instructional material and better-educated teachers always result in higher achievement, smaller class size does not. Darling-Hammond (2000) has also mentioned that reducing class size, when accompanied by the hiring of well-qualified teachers, contributes to student learning. The results of all these studies make it clear that differences in teacher quality have much greater affects on variations in achievement than differences in class size. It may be argued that the finding of a lack of significant correlation between class size which is 35 and can be considered as large (Brewer et al., 1999) and academic achievement in this study due to similarity in the characteristics of all public schools in terms of limited instructional materials (World Bank, 2005) and teachers’ wide use of deductive instructional strategies such as lecture and recitation in all classes (Önür and Engin, 1996; World Bank, 2005). Although it is not statistically significant, there is still a negative relationship between class size and academic achievement of urban poor primary-school students (Table 2). A closer examination of the study results revealed that the effects of the number of teachers with graduate or post-graduate degrees or currently engaged in graduate studies and the number of teachers who participated in an in-service training program during the previous year were negative and statistically significant. This result confirms the results of Parcel and Dufur (2001) study that found attending a school where a high percentage of teachers held graduate degrees had a negative effect on changes in math achievement. Goldhaber and Brewer (1996) found a statistically insignificant association between the percentage of teachers with at least an MA degree and student achievement. Fuchs and Wößmann (2004) found that students performed significantly better in schools where teachers had a higher than average level of education. This is especially true for degrees in pedagogy in the respective subject and for holding specific teaching certificates. However, in line with most past research, Rivkin et al. (2005) found no evidence that holding a master’s degree improved teacher skills. The lack of clarity regarding the estimated relationship between student achievement and higher education of teachers brings into question the widespread existence of pay scales that reward such teacher characteristics, as with the newly instituted teacher promotion system in Turkey, which awards teachers who obtain at least a master’s degree a title and a salary increase. The counterintuitive finding of this study that teacher level of education correlated negatively with student achievement may be related to the unavailability of necessary teaching materials in schools. According to PISA 2003 results, 80 percent of Turkish school principals reported that student learning is hindered by a lack of quality instructional materials (MONE, 2005), a problem that is particularly acute in primary schools in poor neighborhoods. It can thus be argued that knowledge, competencies and proficiencies possessed by teachers do not necessarily translate into teaching practice. Essentially, a teacher may know what to do and how to do it, but may be unable to put this knowledge into practice in the classroom. Finally, school facilities seemed to have a negative impact on student achievement, although the effect was not statistically significant. This is understandable, considering the PISA 2003 results of Turkey, in which more than two-thirds of school principals reported that an inadequate school infrastructure adversely affected math achievement of primary-school students. This adverse effect was well above the average adverse effect of OECD countries (MONE, 2005). Furthermore, students who reported using computers in primary schools were found to have lower cognitive performances on average than those who reported having no access to computers at their schools. Similar findings have been determined with regard to other facilities, such as libraries (Berberoğlu, 2004). A lack of books and a shortage of teachers in Turkish schools were also mentioned more frequently as concerns among poorer households when compared to wealthier ones (World Bank (WB), 2005). It is possible to conclude that merely equipping schools with such facilities is not enough to raise student achievement, rather, what matters most is whether these facilities are utilized properly. In this regard, much remains to be learned as to how principals and teachers mobilize and organize scarce instructional materials. 7. Conclusions and implications This study investigated the relative importance of selected family-, individual- and school-related factors in accounting academic achievement of urban poor primary-school students. The findings indicate that the set of variables comprising student characteristics, including grade, gender, work status, well-being at school, scholastic activities and parental support, explained the largest amount of variance (15 percent) in academic achievement among the urban poor. Family background characteristics and school quality indicators accounted for 5.4 and 4.3 percent of academic achievement, respectively. Variables related to student well-being, such as perceptions of treatment by teachers and number of friends in school, clearly had significant positive effects on academic achievement. This suggests that educational policymakers take into greater consideration the affective characteristics of teachers, such as their attitudes and expectations from students, in order to improve school achievement among the urban poor. Teacher training should equip candidate teachers with the necessary knowledge and skills as well as attitudes to work successfully with disadvantaged students, including those who are economically disadvantaged. It may be expected that good teachers – those who can adapt a wide range of teaching strategies to the specific needs their students and who have high expectations from their students (Darling-Hammond, 2000; Darling-Hammond and Sykes, 2003; Samdal et al., 1999; World Bank, 2005) – can go a long way towards increasing achievement among gecekondu children. This, in turn, would transform schools into more attractive places for children and could thus be expected to increase student attendance, which the present study found to be a significant factor in academic achievement. Time spent on studies and the availability of someone to offer help with studies were also found to have significant effects on children’s academic achievement. Given the low level of parental C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 education, especially that of mothers, as well as the absence of learning materials at home, schools in poor neighborhoods could increase academic achievement by providing students with afterschool remedial sessions under a teacher’s guidance. Results of previous studies suggest that after school remedial sessions on school subjects are related to student academic performance (Philips, 1997; Schouse, 1996). Awareness of the fundamental role of education in producing social and economic disadvantages (Machin, 2006), coupled with the understanding that children who experience favorable material conditions at home and at school achieve greater academic success (Baharudin and Luster, 1998; Coleman, 1991), suggest that policies implemented in gecekondu areas to stimulate improvements in human capital should target not only schools, but households as well. In this regard, instruction, materials and learning experiences may be offered to parents to enable them to provide a supportive learning environment at home. Such an approach would also increase parental involvement and enhance the school–community partnership. As Haghighat (2005) has argued, strengthening community–school networks and creating positive school environments represent a significant means of improving academic achievement in schools located in neighborhoods in which poverty is concentrated. A close examination of the study results indicate that while school facilities and class size do not have a significant effect on student achievement, teacher–student ratio and teacher training do have a strongly significant effect. This is due to the fact that as public schools in poor urban neighborhoods, and therefore subject to Ministry of Education rules and regulations with regard to resource allocation, all the schools studied were equipped with more or less the same quality facilities and physical resources. In view of the homogeneity of the participating schools, the findings of this study should be interpreted with caution when comparing them with findings of other studies that have included public and private schools from middle- and high-SES neighborhoods, where even state schools are well known to benefit more from informal financial and in-kind contributions than schools serving poorer families. The relationship between family background and academic achievement of primary school children brought to light through this study may be typical of other developing countries as well. A number of cross-cultural studies have found family background to be less important than school characteristics in determining academic outcomes among students in less developed countries (Coleman et al., 1966; Fuller and Clarke, 1994; Heyneman and Loxley, 1983). Here, it is important to note that while this study found student characteristics to account for greater variance in academic achievement than either family or school characteristics, some individual variables included within the set of student characteristics – student perceptions of treatment by teachers, and student perceptions of having many friends – are closely related to school climate variables in that they represent additions to school quality from a student perspective (Johnson et al., 1996; Hoy and Hannum, 1997). In light of this understanding, it is possible to conclude from the findings that school characteristics, especially those involving social and human capital (e.g. teacher–student relationships, teacher degree level, teacher–student ratio) that contribute to a school’s ability to provide a positive environment for pupils are, in fact, more important than family characteristics as predictors of academic achievement among the urban poor in Turkey. Such an analysis supports the investment in educational resources, especially those that would improve teacher quality and school environment, in order to increase student academic achievement. Finally, it should be emphasized that the most comprehensive regression model (Model III) employed in this study could only 27 account for approximately one-fourth of the variance in academic achievement registered among students. In other words, the majority of the variation in academic achievement among the urban poor remains unexplained by the selected independent variables, which strongly suggests that further conceptual and empirical efforts be focused on identifying the additional independent variables related to academic achievement. Future studies may examine additional factors related to student characteristics (e.g. differences in achievement levels upon entering school, self-esteem), family characteristics (e.g. home environment, family–child interaction) and school characteristics (e.g. classroom processes, school climate, teacher motivation) to maximize the amount of variance explained in academic achievement among the urban poor. In addition, this study suggests that structural equation modeling techniques can be used to explore the complex and reciprocal relationship among family-, individual-, and school-level variables related to academic achievement of urban poor primary-school students. References Akabayashi, H., Psacharopoulos, G., 1999. The trade-off between child labour and human capital formation: a Tanzanian case study. The Journal of Development Studies 35 (5), 121–140. Al-Nhar, T., 1999. Determinants of grade eight achievement in Jordan: a multilevel analysis approach. Educational Psychology 19 (1), 37–45. Angrist, J.D., Lavy, V., 1999. Using Maimonides’ rule to estimate the effect of class size on scholastic achievement. The Quarterly Journal of Economics 114, 533– 575. Aypay, A., 2003. The tough choice at high school door: an investigation of the factors that lead students to general or vocational schools. International Journal of Educational Development 23, 517–527. Babad, E., 1996. How high is ‘‘high inference’’? Within classroom differences in students’ perceptions of classroom interaction. Journal of Classroom Interaction 31, 1–9. Baharudin, R., Luster, T., 1998. Factors related to the quality of the home environment and children’s achievement. Journal of Family Issues 19 (4), 375–403. Bacolod, M., Tobias, J.L., 2005. Schools, School Quality and Academic Achievement: Evidence from the Philippines. Retrieved July 3, 2006, from http://www. socsci.uci.edu/mbacolod/cebu_schlrank.pdf. Baker, D.P., Goesling, B., Letendre, G.K., 2002. Socioeconomic status, school quality and national economic development: a cross-national analysis of the ‘‘Heyneman-Loxley Effect’’ on mathematics and science achievement. Comparative Education Review 46 (3), 291–312. Barro, R.J., 1997. Determinants of Economic Growth: A Cross-country Empirical Study. MIT Press, Cambridge, MA. Berberoğlu, G., 2004. Student Learning Achievement. Paper Commissioned for the Turkey ESS. World Bank, Washington, DC. Binder, M., Scrogin, D., 1999. Labour force participation and household work of urban schoolchildren in Mexico: characteristics and consequences. Economic Development and Cultural Change 48 (1), 123–146. Blake, J., 1989. Family Size and Achievement. University of California Press, Berkeley. Brewer, D.J., Krop, C., Gill, B.P., Reichardt, R., 1999. Estimating the cost of national size reductions under different policy alternatives. Education Evaluation and Policy Analysis 21, 179–192. Brooks-Gunn, J., Duncan, G.J., 1997. The effects of poverty on children. The Future of Children 7 (2), 55–71. Buchmann, C., Hannum, E., 2001. Education and stratification in developing countries: a review of theories and research. Annual review of Sociology 27, 77–103. Bugra, A., Keyder, C., 2003. New Poverty and the Changing Welfare Regime of Turkey. United Nations Development Program. Ajans Turk, Ankara. Card, D., Krueger, A., 1996. School resources and student outcomes: an overview of the literature and new evidence from North and South Carolina. Journal of Economic Perspectives 10, 31–40. Carron, G., Chau, T.N., 1996. The Quality of Primary Schools in Different Developing Contexts. UNESCO Publishing, Paris, France. Chevalier, A., Lanot, G., 2002. The relative effect of family characteristics and financial situation on educational achievement. Education Economics 10 (2), 165–181. Coleman, J.S., Campbell, E.Q., Hobson, C.J., McPartland, J., Mood, A.M., Weinfield, F.D., 1966. Equality of Educational Opportunity. National Center for Educational Statistics, Washington, DC. Coleman, J.S., 1988. Social capital in the creation of human capital. American Journal of Sociology 94, 95–120. Coleman, J.S., 1991. Parental involvement in education. In: Policy Perspective, Office of Educational Research and Improvement, US Department of Education, Washington, DC. Coleman, J.S., 2006. The adolescent society. Education Next 6(1) 40–43. 28 C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 Connolly, J.A., Hatchette, V., McMaster, L.E., 1998. School Achievement of Canadian Boys and Girls in Early Adolescence: Links with Personal Attitudes and Parental and Teacher Support for School. Retrieved May 6, 2006, from http:// www.11.sdc.gc.ca/en/cs/sp/sdc/pkrf/publications/research/1998-002344/ 1998-002344.pdf. Currie, J., 1995. Welfare and the Well-Being of Children, Fundamentals of Pure and Applied Economics No. 59. Harwood Academic Publishers, Switzerland. Darling-Hammond, L., 2000. Teacher quality and student achievement: a review of the state policy evidence. Education Policy Analysis Archives 8 (1), 1–30. Darling-Hammond, L., Sykes, G., 2003. Wanted: a national teacher supply policy for education: the right way to meet the ‘Highly Qualified Teacher’ challenge. Educational Policy Analysis Archives 11, 33–50. Downey, D.B., 1995. Bigger is not better: family size, parental resources, and children’s educational performance. American Sociological Review 60, 746– 761. Downey, D.B., 2001. Number of siblings and intellectual development: the resource dilution explanation. American Psychologist 56 (6), 497–504. Ehrenberg, R.G., Brewer, D.J., Gamoran, A., Willms, J.D., 2001. Class size and student achievement. Psychological Science in the Public Interest 2 (1), 1–30. Epstein, J.L., 1987. Parent involvement: what research says to administrators. Education and Urban Sociology 19, 277–294. Erder, S., 1995. Yeni kentliler ve kentin yeni yoksulları [The new urbanites and new poor of the city]. Toplum ve Bilim 66, 106–119. Erman, T., 2001. The politics of squatter (gecekondu) studies in Turkey: the changing representations of rural migrants in the academic discourse. Urban Studies 38 (7), 983–1002. Farkas, G., Sheehan, D., Grobe, R.P., 1990. Coursework mastery and school success: gender, ethnicity, and poverty groups within an urban school district. American Educational research Journal 27 (4), 807–827. Fetler, M., 2001. Student mathematics achievement test scores, dropout rates, and teacher characteristics. Teacher Education Quarterly 2001 (Winter), 151–168. Field, A., 2005. Discovering Statistics Using SPSS, 2nd ed. Sage Publications, London. Fine, M., Cook, D., 1993. Parent involvement: reflections on parents, power and urban public Schools. Teachers College Record 94 (4), 29–39. Fuchs, T., Wößmann, L., 2004. What accounts for international differences in student performance? A re-examination using PISA data. CESifo Working Paper No. 1235. Fuller, B., Clarke, P., 1994. Raising school effects while ignoring culture? Local conditions and the influence of classroom tools, rules and pedagogy. Review of Educational Research 64, 122–131. Ganzach, Y., 2000. Parent’s education, cognitive ability, educational expectation and educational attainment. British Journal of Educational Psychology 70, 419–441. General Directorate of Women’s Status and Problems (Başbakanlık Kadının Statusu ve Sorunları Genel Mudurlugu), 1999. Calısmaya hazır isgucu olarak kentli kadın ve degisimi [Urban women as labour force and change]. Başbakanlık Kadının Statusu ve Sorunları Genel Mudurlugu, Ankara. Green, S.B., Salkind, N.J., Akey, T.M., 2000. Using SPSS for Windows: Analyzing and Understanding Data, 2nd ed. Prentice Hall, New Jersey. Greenwald, R., Hedges, L.V., Laine, R., 1996. The effect of school resources on student achievement. Review of Educational Research 66, 361–396. Gregg, P., Machin, S., 1999. Childhood disadvantage and success or failure in the labour market. In: Blanchflower, D., Freeman, R. (Eds.), Youth Employment and Joblessness in Advanced Countries. National Bureau of Economic Research, Cambridge. Griffith, J., 1999. School climate as ‘‘social order’’ and social action’’: a multi-level analysis of public elementary school student perceptions. Social Psychology of Education 2, 339–369. Goldhaber, D.D., Brewer, D.J., 1996. Evaluating the effect of teacher degree level on educational performance. Accessed, June 1, 2006, http://nces.ed.gov/pubs97/ 97535l.pdf. Grolnick, W.S., Slowiaczek, M.L., 1994. Parents’ involvement in children’s schooling: a multidimensional conceptualization and motivation model. Child Development 65, 237–252. Guncer, B., Kose, R., 1993. Effects of family and school on Turkish students’ academic performance. Education and Society 11 (1), 51–63. Haghighat, E., 2005. School social capital and academic performance. International Journal of Sociology of Education 15 (3), 213–235. Häkkinen, I., Kirjavainen, T., Uusitalo, R., 2003. School resources and student achievement revisited: new evidence from panel data. Economics of Education Review 22, 329–335. Hanushek, E.A., 1992. The trade-off between child quantity and quality. Journal of Political Economy 100 (1), 84–117. Hanushek, E.A., 1997. Assessing the effects of school resources on student performance: an update. Educational Evaluation and Policy Analysis 19 (2), 141–164. Hanushek, E.A., Rivkin, S.G., Taylor, L.L., 1996. Aggregation and the estimated effects of school resources. The Review of Economics and Statistics 78 (4), 611–627. Hanushek, E.A., Luque, J.A., 2003. Efficiency and equity in schools around the world. Economics of Education Review 22, 481–502. Heady, C., 2003. The effect of child labour on learning achievement. World Development 31 (2), 385–398. Heyneman, S.P., Loxley, W.A., 1983. The effect of primary school quality on academic achievement across 29 high- and low-income countries. American Journal of Sociology 88 (6), 1162–1194. Hoxby, C.M., 2000. The effects of class size on student achievement: new evidence from population variation. Quarterly Journal of Economics 115 (4), 1239–1285. Hoy, W.K., Hannum, J.W., 1997. Middle school climate: an empirical assessment of organizational health and student achievement. Educational Administration Quarterly 33 (3), 290–311. ILO, 2002. Every Child Counts: New Global Estimates on Child Labour. ILO, Geneva, Switzerland. Johnson, W.L., Johnson, A.M., Zimmerman, K., 1996. Assessing school climate priorities: a Texas study. The Clearing House 70 (2), 64–66. Keith, T.Z., Reimers, T.M., Fehrmann, P.G., Pottebaum, S.M., Aubey, L.W., 1986. Parental involvement, homework, and TV time: direct and indirect effects on high school achievement. Journal of Educational Psychology 5, 373–380. Keleş, R., 2000. Kentleşme politikası [Urbanization Policy]. Imge Yayinlari, Ankara. Kongar, E., 1999. 21. Yuzyılda Turkiye: 2000’li yıllarda Turkiye’nin toplumsal yapısı [Turkey in 21st Century: The social structure of Turkey in 2000s], 16th ed. Remzi Kitapevi, Istanbul. Konu, A., Rimpelä, M., 2002. Well-being in schools: a conceptual model. Health Promotion International 17 (1), 79–87. Krueger, A., 1999. Experimental estimates of education production functions. The Quarterly Journal of Economics 114, 497–532. Krueger, A., Lindahl, M., 2001. Education for growth: why and for whom? Journal of Economic Literature 39, 1101–1136. Lee, S.A., 1993. Family structure, effects on student outcomes. In: Schneider, B., Coleman, J.S. (Eds.), Parents, Their Children, and School. Westview Press, Boulder, pp. 43–75. Lindahl, M., 2005. Home versus school learning: a new approach to estimating the effect of class size on achievement. Scandinavian Journal of Economics 107 (2), 375–394. Lleras-Muney, A., 2005. The relationship between education and adult mortality in the United States. Review of Economic Studies 72, 189–221. Lockheed, M.E., Verspoor, A., 1991. Improving Primary Education in Developing Countries. World Bank/OUP. Ma, X., 1997. A multiple regression analysis of mathematics achievement in the Dominican Republic. International Journal of Educational Development 17 (3), 313–321. Ma, X., 2001. Stability of socioeconomic gaps in mathematics and science achievement among Canadian schools. Canadian Journal of Education 26 (1), 97–118. Machin, S., 2006. Social disadvantage and education experiences. OECD, Social, Employment and Migration Working Papers No. 32. Mancebon, M.J., Mar Molinero, C., 2000. Performance in primary schools. Journal of the Operational Research Society 51, 843–854. Majoribanks, K., 1994. Families, schools and children’s learning. International Journal of Educational Research 21, 439–555. Mayer, S., 1997. What Money Can Buy. Harvard University Press, Cambridge. McLanahan, S., Sandefur, G., 1994. Growing up with a Single Parent: What Hurts, What Helps. Harvard University Press, Cambridge. McEwan, P.J., Marshall, J.H., 2004. Why does academic achievement vary across countries? evidence from Cuba and Mexico. Education Economics 12 (3), 205– 217. McNeal, R.B., 1999. Parental involvement as Social Capital: differential effectiveness on science, achievement, truancy, and dropping out. Social Forces 78, 117–144. MONE, 2005. PISA 2003 Projesi: ulusal nihai rapor [PISA 2003 Project: National Final Report]. Milli Eğitim Basımevi, Ankara. MONE, 2007. Turkish Educational System, Retrieved January 31, 2008, from http:// digm.meb.gov.tr/uaorgutler/BM/turkish_education_system.pdf. Muller, C., Kerbow, D., 1993. Parent involvement in the home, school, and community. In: Schneider, B., Coleman, J.S. (Eds.), Parents, Their Children and Schools. Westview Press, Boulder, pp. 13–46. Muller, C., 1998. Gender differences in parental involvement and adolescents’ mathematics achievement. Sociology of Education 71, 336–356. OECD, 2001. Knowledge and Skills for Life: Results from PISA 2000, Retrieved June 13, 2006, from http://oecdpublications.gfi-nb.com/cgi-bin/OECDBookShop. storefront/1755../962001141P. Önür, A., Engin, C., 1996. Is civic education in Turkish schools conducive to education for democracy? In: Paper Presented at the First Balkan Civitas Conference, June 9–15, Primorsko, Bulgaria. Parcel, T.L., Dufur, J.M., 2001. Capital at home and at school: effects on student achievement. Social Forces 79 (3), 881–911. Parcel, T.L., Menaghan, E.G., 1994. Early parental work, family social capital, and early childhood outcomes. American Journal of Sociology 99, 972–1009. Patrinos, H.A., Psacharopoulos, G., 1995. Educational performance and child labour in Paraguay. International Journal of Educational Development 15 (1), 47–60. Peters, E., Mullis, N., 1997. The role of family and source of income in adolescent achievement. In: Duncan, G., Brooks-Gun, J. (Eds.), Consequences of Growing up Poor. Russell Sage Foundation, New York, pp. 340–438. Philips, M., 1997. What makes schools effective? A comparison of the relationship of communitarian climate and academic climate to mathematics achievement and attendance during middle school. American Educational Research Journal 34 (4), 633–662. Postlethwaite, T.N., Wiley, D.E., 1992. The IEA Study of Science II: Science Achievement in Twenty-three Countries. Pergamon Press, Oxford. Psacharopoulos, G., 1994. Returns to investment in education: a global update. World Development 22, 1325–1344. Ray, R., Lancaster, G., 2003. Does Child Labour Affect School Attendance and School Performance? Multi Country Evidence on SIMPOC Data. Unpublished Report. ILO/IPEC. C. Engin-Demir / International Journal of Educational Development 29 (2009) 17–29 Rivkin, S.G., Hanushek, E.A., Kain, J.F., 2005. Teachers, schools and academic achievement. Econometrica 73 (2), 417–458. Rothstein, R., 2000. Finance Fungibility: Investing Relative Impacts of Investments in Schools and Non-school Educational Institutions to Improve Student Achievement. Center on Educational Policy Publications, Washington, DC. Sabo, D.J., 1995. Organizational climate of middle schools and the quality of student life. Journal of Research and Development in Education 28, 150–160. Samdal, O., 1998. The School Environment as a Risk or Resource for Students’ Health-related Behaviors and Subjective Well-being. Research Center for Health Promotion, Faculty of Psychology, University of Bergen. Samdal, O., Wold, B., Bronis, M., 1999. Relationship between students’ perceptions of school environment, their satisfaction with school and perceived academic achievement: an international study. School Effectiveness and School Improvement 10 (3), 296–320. Sammons, P., 1995. Gender, ethnic and socio-economic differences in attainment and progress: a longitudinal analysis of student achievement over 9 years. British Educational Research Journal 21, 465–485. Schiller, K.S., Khmelkov, V.T., Wang, X.Q., 2002. Economic development and the effects of family characteristics on mathematics achievement. Journal of Marriage and Family 64, 730–742. Schouse, R.C., 1996. Academic press and sense of community: conflict, congruence and implications for student achievement. Social Psychology of Education 1, 47–68. Simmons, J., Alexander, L., 1978. The determinants of school achievement in developing countries: a review of the research. Economic Development and Cultural Change 26 (2), 341–357. Smits, J., Gündüz-Hoşgör, A., 2006. Effects of family background characteristics on educational participation in Turkey. International Journal of Educational Development 26, 545–560. Steinberg, L., Lamborn, S.D., Dornbush, S.M., Darling, N., 1992. Impact of parenting practices on adolescent achievement: Authoritative parenting, school involvement, and encouragement to succeed. Child Development 63, 1266–1281. Sukon, K.S., Jawahir, R., 2005. Influence of home-related factors in numeracy performance of fourth-grade children in Mauritius. International Journal of Educational Development 25, 547–556. Sui-Chu, E.H., Willms, J.D., 1996. Effects of parental involvement on eight grade achievement. Sociology of Education 69, 126–141. Tabachnick, B.G., Fidell, L.S., 1996. Using Multivariate Analysis, 3rd ed. HarperCollins College Publishers, New York. 29 Tansel, A., 2002. Determinants of schooling attainment for boys and girls in Turkey: individual, household and community factors. Economics of Education Review 21, 455–470. Tansel, A., Bircan, F., 2006. Demand for education in Turkey: a tobit analysis of private tutoring expenditures. Economics of Education Review 25, 303–313. Teachman, J.D., Paasch, K., Carver, K., 1996. Social capital and dropping out of school early. Journal of Marriage and the Family 58, 773–783. Tilak, J.B.G., 2002. Education and poverty. Journal of Human Development 3 (2), 191–207. Thompson, G.H., Johnston, J.S., 2006. Variation in the Relationship Between Nonschool Factors and Student Achievement on International Assessments. National Center for Education Statistics, Institute of Education Sciences, NCES, 2006-014. Van Houtte, M., 2004. Why boys achieve less at school than girls: the difference between boys’ and girls’ academic culture. Educational Studies 30 (2), 159–173. Veenstra, R., Kuyper, H., 2004. Effective students and families: the importance of individual characteristics for achievement in high school. Educational Research and Evaluation 10 (1), 41–70. Verkuyten, M., Thıjs, J., 2002. School satisfaction of elementary school children: the role of performance, peer relations, ethnicity and gender. Social Indicators Research 59, 203–228. Wang, J., Wildman, L., 1995. An empirical examination of the effects of family commitment in education on student achievement in seventh grade science. Journal of Research in Science Teaching 32, 833–837. Willms, D.J., Somers, M.A., 2001. Family, classroom, and school effects on children’s educational outcomes in Latin America. School Effectiveness and School Improvement 12 (4), 409–445. Wiseman, A.W., Brown, D.S., 2002. Does teacher preparation really matter? The influence of teacher preparation on student achievement in the United States and abroad. In: Paper Presented at the Conference of the Oklahoma Association of Teacher Educators, Fall, 2002, Edmond, OK. World Bank, 2005. Sustainable Pathways to an Effective, Equitable, and Efficient Education System for Preschool Through Secondary School Education. Education Sector Study, Turkey Report No. 32450-TU. Wößmann, L., 2003. Schooling resources, educational institutions and student performance: the international evidence. Oxford Bulletin of Economics and Statistics 65 (2), 0305–9049. Yayan, B., Berberoğlu, G., 2004. A re-analysis of the TIMSS 1999 mathematics assessment data of the Turkish students. Studies in Educational Evaluation 30, 87–104.