The Relationship Between Work, School Performance and School Attendance of Primary School Children in Turkey

tarix	27.06.2016
ölçüsü	165.5 Kb.

Work, school performance and school attendance

The Relationship Between Work, School Performance and School Attendance of Primary School Children in Turkey*
Cennet Engin Demir^a

Erol Demir^b

Sevil Uygur^c
Paper presented at the European Conference on Educational Research, University of Geneva, 13-15 September 2006

Abstract
This study is a part of larger research project designed to investigate the effect of light work on school attendance and school performance of children in Turkey. The purpose of this paper is to examine the relationship between child work and school performance, child work and school attendance from the perspective of children. Children between 12-14 years of age who combine school and work formed the target population of the study. Children currently in school and not working formed the control group. 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. A total of 652 working children, 423 non-working children from six districts and 23 schools in urban areas of the capital-Ankara participated in the study. Face-to-face structured interviews were conducted by the trained interviewers to collect the data. Results revealed that there is a significant difference in the school performance and school attendance of working and non-working children. The means test scores and attendance of working children are lower than non-working children. The determinants of working and non-working children’s school performance and school attendance were also examined and results were presented in the paper.
Keywords: Light work, school performance, school attendance, primary school

Introduction
The relationship between school attendance, school performance and work is generally perceived to be negative. Work interferes with schooling because it requires too much of children’s time (Heady, 2003). Balancing the demands of work and education places physical and psycho-social strain on children and often leads to poor academic performance and dropping out. Work may demand extensive physical energy, so that the child lacks the energy required for school attendance or effective study. As a result of fatigue and a lack of leisure activities to support physical, social and emotional development, the child will experience very little mental stimulation and will end up neglecting his or her studies (Binder & Scrogin, 1999). Akabayashi & Psacharapoulos (1999), for example, found that a child’s reading and mathematics ability decreased with additional hours of work, whereas they increased with additional hours of school attendance and study. In their study, Ray & 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 the 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 hours a week contributed positively to the child’s schooling and to his/her study time.

Children who perform poorly in school are more likely to enter the labour market at an early age. Researchers have suggested that even limited amounts of work adversely affect a child’s learning, as reflected in a reduction in the child’s school attendance rate and length of schooling (Ray & Lancaster, 2003; Heady, 2003). However, it should be acknowledged that, in some cases, work enables children to afford schooling by providing additional income for families (Binder & Scrogin, 1999).

In Turkey, school attendance seems to be the major deterrent to market work. The 1999 Child Labour Survey of the State Institute of Statistics indicated that 1.6 percent of children enrolled in school were engaged in economic activity, 27.3 percent in some sort of household work and 71.1 percent in neither.
Findings of a study conducted to investigate the working and living conditions of migrant child workers in the cotton industry; the effects of work on their physical, psychological and educational development (Gulbucuk et al., 2003) indicated that seasonal work has particularly adverse effects on the schooling of children. In another study focusing on children in street work, work was also shown to impact negatively on children’s schooling (Aksit et al., 2001).
Although education authorities are aware of the problems working children face in attending school in general, there is still insufficient understanding of the exact nature of the impact of work on the school attendance and educational performance of children. Most available data on work and education has been gathered through household and labour force surveys. As a result, information on children’s educational activities is generally limited to whether or not they enrolled in school. School-based surveys can provide an opportunity for obtaining more detailed information on the amount of time children spend in school and doing homework, how often they miss school due to work, their academic progress in relation to other students, their ability to join extracurricular activities, and the direct cost of schooling. This type of data should prove invaluable in developing effective tools for the retention of working children in the education system and in understanding the school performance and development of children in general.

Purpose of the Study
This study is a part of larger research project designed to investigate the effect of light work on school attendance and school performance of children in Turkey. The purpose of this paper is to examine the relationship between child work and school performance, child work and school attendance from the perspectives of children.
Method
Participants
Children between 12-14 years of age who combine school and work formed the target population of the study. Children currently in school and not working formed the control group. The selection of the target group was based on the Turkish Government’s ratification of ILO Convention No.138 on minimum age, which sets the minimum age for employment at 15 and which permits children ages 12-13 to engage in “light work” in countries with insufficiently developed economies and educational facilities (Art.2).
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.
Stage 1: Based on published information and expert opinion, six administrative districts within the Greater Ankara Municipality with relatively high numbers of working children were selected (Districts: Kecioren, Mamak, Cankaya, Etimesgut, Altındag, Yenimahalle) Purposive sampling was used as the sampling methodology at this first stage. The sampling frame used was the number of primary schools in each district, the number of female and male students in each school and the total number of primary school students in Ankara
Stage 2: A total of 200 schools were selected from within the six districts based on the ratio of working children using Probability Proportional to Size (PPS) with Measure of Size (MoS). A school principal or guidance counselor at each of the selected schools was contacted by phone and requested to provide researchers with a list containing information on the numbers of working children and male and female students at their respective schools. These lists were used to aid in the selection of schools from among the six districts selected in Stage 1 to serve as the second stage sampling frame. This sampling frame was used in the selection of 25 schools based on the number of working children using Probability Random Selection.
Stage 3: Systematic Random Sampling was used to identify children in the selected schools in Grades 6, 7 and 8 who combine school and work by administering a listing form in each classroom. Listing forms compiled data on children’s sex, age, work status, family socio-economic status and neighbourhood developmental level as implicit stratification criteria for selecting the final sampling unit. Information collected through this listing study was compiled and used as the sampling frame for the fourth stage of sample selection. Two schools were excluded due to the very low number of working children enrolled, as indicated by the listing study.
Stage 4: At the final stage of sample selection, 50 students were selected 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. A total of 652 working children, 423 non-working children were interviewed from six districts and 23 schools in urban areas of the capital-Ankara .
Data collection procedure
Data was collected by trained interviewers using face-to-face interviews with a structured interview schedule (questionnaires) developed by the researchers which comprised of mostly close-ended questions with some open-ended questions. Same questions were asked working and non-working children. However, additional work related questions were asked to working children. Data on students attendance and mid-term examinations scores in Mathematics, Turkish Language and Science courses were obtained from school record’s. Interviewers were trained and a pilot study was conducted to test the validity of questions and to assess the data-collection procedures.

Data Analysis

The following statistical analyses were employed to examine the work- and school-related characteristics of working and non-working children:

Cross-tabulations and descriptive statistics to describe some background characteristics of working children and the nature of work done.
Linear Regression Analysis to estimate the determinants of working and non-working children’s school performance and school attendance. The following regression equation was used:

Where
Y denotes the average of test scores in Turkish, science and mathematics for school performance and number of days a student attended school for school attendance;

X denotes quantitative explanatory variables; and

B denotes qualitative explanatory variables.
Discriminant Analysis was used to determine the effect that different amounts of time spent on economic activity and household chores per week had on the school performance and attendance of economically active children.

Results

The nature and duration of work

This study found the majority of working children in the selected schools to be boys from large families of low socio-economic status. The majority of both working girls and boys interviewed were twelve-year-olds, followed by 13-year-olds and 14-year-olds. Due to the low rate of female working children in the urban population in Turkey, the number of girls in this study sample was also low.
More than two-thirds of the 652 working students interviewed engaged in household chores in addition to economic activity, and all of the female working students engaged in household chores in addition to economic activity. Nearly two-thirds of the working students interviewed were unpaid family workers. Girls engaged in unpaid family work at higher rates than boys. Following unpaid work in a family business, the next most frequently engaged in economic activity was selling low-value items on the streets. While working in a family-owned shop may limit the hazards associated with work outside the home, working on the streets can have serious adverse effects on a child’s physical, psychological and moral development. Because the hazards to children stem from the nature of the street environment itself, this is true even for children who spend relatively few hours at work on the streets.
The majority of children who combined school and work spent more time at both unpaid family work and paid work on the weekends than during the school week.

School Performance

Results indicated that there is a significant difference in the school performance of working and non-working children (0.05≤significance level). The mean test scores of non-working children (M=2.27; SD=1.17) were higher than mean test scores of working children (M=2.00; SD=1.04). Mean test scores were also higher for working girls (2.52) than for working boys (1.87).

Determinants of school performance of economically active children

A full list of the variables used in Regression Analysis is provided in Appendix 1.

In order to identify the variables that determine the school performance of economically active children, 41 linear regression models were tested and the optimum model selected. The following factors were considered in the selection of the optimum model:
Multiple R² and Adjusted R²,

Number of explanatory variables in the model,

F test results and significance level,

Plot charts by predicted value and actual value for related variables.

The following equation was found to be the optimum model in determining the school performance of economically active children:

Where,
EF denotes the number of household members with at least a secondary education;

GKGS denotes the number of days a student was late for school during the first semester;

OKG denotes the school quality indicator;

GDS denotes the number of days a student attended school during the first semester;

DOS denotes the total hours per week engaged in studies;

EOS denotes the average hours per day engaged in paid work;

DB3 denotes a child’s perception of his/her school performance as “moderate”;

DB1 denotes a child’s perception of his/her school performance as “very good”;

DC denotes being male

DY denotes availability of someone to help child with studies outside school; and

DYM1 denotes eating only one meal per day.
The following formula was obtained using the model’s coefficients:
Y= -1,389 -0,392 DB3 +0,543 DB1 -0,345 DC+ 0,158 EF -0,08GKGS -0,038OKF -0,035 GDS+ 0,015 DOS- 0,172 DY- 0,199 DYM1+ 0,035 EOS
Table 1 summarizes the statistics for the variables used in the optimum model of determinants on the school performance of economically active children. As can be seen from the Table 1 Linear Regression Analysis revealed that child’s perception of school performance as “good” ; number of household members with at least a secondary education; school quality indicator; average hours per day engaged in paid work; number of days a student attended school; child’s perception of school performance as “moderate”; being male; eating only one meal per day; availability of someone to help child with studies outside school; number of days late for school had significant effect on school performance of economically active children.
Table 1 Determinants of school performance (economically active children)

Variable		t H	P value
Child’s perception of school performance as “moderate”	-0,392	-4,956	0,000
Child’s perception of school performance as “very good”	0,543	4,753	0,000
Being male	-0,345	-3,682	0,000
Total number of household members with at least a secondary education	0,158	2,944	0,003
Number of days late for school during first semester	-0,08	-2,730	0,007
School quality indicator	0,038	3,193	0,001
Number of days a student attended school during first semester	0,035	3,036	0,002
Total hours per week engaged in studies	0,015	2,775	0,006
Availability of someone to help with studies after school	-0,172	-2,286	0,023
Eating only one meal per day	-0,199	-2,265	0,024
Average hours per day engaged in paid work	0,035	1,991	0,047
R²	0,247
Adjusted R²	0,234
Model’s F test	19,107
Significance Level	0,000

Determinants of school performance of non-economically active children
In order to identify the variables that determine the school performance of non-economically active children, 30 linear regression models were tested and the optimum model selected. The following equation was found to be the optimum model in determining the school performance of non-economically active children:

Where
EF denotes the number of household members with at least a secondary education;

GKGS denotes the number of days a student was late for school during the first semester;

EVOS denotes the average hours per day spent on chores;

DB1 denotes a child’s perception of his/her school performance as “very good”;

DB2 denotes a child’s perception of his/her school performance as “good”; and

DC denotes being male.
The following formula was obtained using the model’s coefficients:
Y= 1,850 +1,505 DB1 + 0,453 DB2 -0,326 DC +0,159 EF -0,151 GKGS +0,179 EVOS
Table 2 summarizes the statistics for the variables used in the optimum model of determinants on the school performance of non-economically active children.

Table 2 Determinants of school performance (non-economically active children)

Variable		t H	P value
Child’s perception of his/her school performance as “very good”	1,505	10,529	0,000
Child’s perception of his/her school performance as “good”	0,453	3,963	0,000
Being male	-0,326	-3,227	0,001
Total number of household members with at least a secondary education	0,159	2,418	0,016
Number of days late for school during the first semester	-0,151	-2,147	0,032
Average hours per day engaged in household chores	0,179	2,067	0,039
R²	0,288
Adjusted R²	0,278
Model’s F test	28,029
Significance Level	0,000

Linear Regression Analysis revealed that there is a significant relationship between the variables child’s perception of school performance as “very good”; child’s perception of school performance as “good”; number of household members with at least a secondary education; average hours per day engaged in chores; being male and number of days late for school and the school performance of non-economically active children

Comparison of determinants of school performance for economically active and non-economically active children
As shown above, there are 11 significant variables in the optimum model for economically active children’s school performance and only six significant variables in the optimum model for non-economically active children’s school performance. Moreover, only four variables were found to effect the school performance of both economically active children and non- economically active children:

Child’s perception of school performance as “good”
Number of household members with at least a secondary education
Being male
Number of days late for school

The school quality indicator was found to effect the school performance of economically active children but not non-economically active children. The Adjusted R² was also found to be higher for non-working children.

School Attendance

Attendance records indicated that there were significant differences between the school attendance of working and non-working children. The mean attendance of non-working children (M=88.3 out of 90 days, SD=2.11 days,) was higher than the mean attendance of non-working children (M=87.3 out of 90 days, SD=3.22). Attendance of working girls was higher than attendance of working boys.

School attendance of economically active children
In order to identify the variables that determine the school attendance of economically active children, 35 linear regression models were tested and the optimum model selected.
The following equation was found to be the optimum model in determining the school attendance of economically active children:

Where,
A denotes age of child

DORT denotes average test score;

HHB denotes household size;

FAAL4 denotes number of days per week engaged in Economic Activity 4;

FAAL1 denotes number of days per week engaged in Economic Activity 1;

OKG denotes school quality indicators;

ETOP denotes total hours per week engaged in paid work;

DC denotes being male;

DOD1 denotes child’s completion of homework “often”; and

DY denotes availability of someone to help child with studies outside school.

The following formula was obtained using the model’s coefficients:

Y= 96,237 - 0,539 A -1,176 DC + 0,878 DOD1 +0,809 DY +0,361 DORT -0,255 HHB -0,171 FAAL4 -0,211 FAAL1 +0,101 OKG -0,026 ETOP
Table 3 summarizes the statistics for the variables used in the optimum model of determinants on the school attendance of economically active children.
Table 3 Determinants of school attendance (economically active children)

Variable		t _H	P value
Age	-0,539	-3,107	0,002
Being male	-1,176	-3,766	0,000
Child completes homework “often”	0,878	3,231	0,001
Availability of someone to help child with studies outside school	0,809	3,369	0,001
Average test scores	0,361	3,019	0,003
Household size	-0,255	-2,777	0,006
Number of days per week engaged in Economic Activity 4	-0,171	-3,204	0,001
Number of days per week engaged in Economic Activity 1	-0,211	-2,248	0,025
School quality indicators	-0,101	-2,585	0,010
Total hours per week engaged in paid work	-0,026	-2,150	0,032
R²	0,162
Adjusted R²	0,149
Model’s F test	12,354
Significance Level	0,000

As showed on Table 3 Linear Regression Analysis revealed that the completion of homework “often”; availability of someone to help child with studies outside school; average test score; being male; age; household size; number of days per week engaged in Economic Activity 1; number of days per week engaged in Economic Activity 4; school quality indicators; total hours per week engaged in paid work had a significant effect on school attendance of working students.

School attendance of non-economically active children
In order to identify the variables that determine the school attendance of non-economically active children, 25 linear regression models were tested and the optimum model selected.
The following equation was found to be the optimum model in determining the school attendance of non-economically active children:

Where
GKGS denotes the number of days student was late for school during first semester;

DOS denotes the total hours per week engaged in studies;

DORT denotes average test score; and

DO3 denotes completion of homework “seldom”.
The following formula was obtained using the model’s coefficients:
Y= 87,455 -0,496 GKGS +0,033 DOS -3,304 DO3 +0,204 DORT
Table 4 summarizes the statistics for the variables used in the optimum model of determinants on the school attendance of non-economically active children.
Table 4 Determinants of school attendance (non-economically active children)

Variable		t _H	P value
Number of days that student was late during first semester	-0,496	-3,095	0,002
Total hours per week engaged in studies	0,0326	2,150	0,032
Child’s level of homework completion (seldom)	-3,304	-2,461	0,014
Average of test scores	0,204	2,097	0,037
R²	0,073
Adjusted R²	0,064
Model’s F test	8,246
Significance Level	0,000

Linear Regression Analysis revealed that variables such as average test score; total hours per week engaged in studies; completion of homework “seldom”; number of days late for school in first semester had a significant effect on school attendance of non-working children.

Comparison of optimum models for working and non-working children’s school attendance
There are 10 significant variables in the optimum model for working children’s school attendance and four significant variables in the optimum model for non-working children. Average test score is the only variable that exists in the optimum models for school attendance of both economically and non-economically active children.

Effect of hours of work on school performance and school attendance

Discriminant Analysis was used to determine the effect that different amounts of time spent on economic activity and household chores per week had on the school performance and attendance of economically active children. Results indicated that there is no consistent trend in average test scores of economically active children according to total hours spent for work/chores. For example, the child who spent the least time per week engaged in work/chores had a low test score of 1.42, whereas a child who spent 94 hours per week engaged in work/chores had a test score of 1.33 – a difference of only 6.33 percent.

Discussion and Conclusions

The background characteristics and nature of work
This study found the majority of working children in the selected schools to be boys from large families of low socio-economic status. The majority (53%) of both working girls and boys interviewed were twelve-year-olds, followed by 13-year-olds and 14-year-olds. Due to the low rate of female working children in the urban population in Turkey, the number of girls in this study sample was also low. The majority (59.3%) of working children came from households with five to seven members which is above the average size for both Ankara and Turkey.
Nearly two-thirds of the working students interviewed were unpaid family workers. Girls engaged in unpaid family work at higher rates than boys. Following unpaid work in a family business, the next most frequently engaged in economic activity was selling low-value items on the streets.

Work and school performance

Results indicated that there is a significant difference in the school performance of working and non-working children. Girls had higher test scores than boys, a result that is consistent with the Regression Analysis finding that “being male” negatively affect the school performance. It can be concluded from the results that combining school and work leads a poor academic performance at school.

Linear regression analysis was able to determine the factors that impacted both positively and negatively on children’s school performance. The two factors found to have the most significant positive effect were a child’s perception of his/her school performance as “good” and at least one member in a child’s household having a secondary level of education or higher. Factors found to have a significant negative effect were a child’s perception of his/her school performance as “moderate”, sex (being male), eating only one meal per day, lack of availability of someone to help with studies outside school and being late for school.
While school quality was expected to have a significant influence on the school performance and school attendance of working and non-working children, survey findings showed it had only a slight positive effect on the school performance of working children and no effect on the performance of non-working children. It is possible that this finding may be related to the similarity in characteristics of the selected schools, all of which are public schools located in urban areas of low or lower-middle socio-economic status in Ankara. However, the similarity of class size, facilities and teacher characteristics among the schools would still not explain why school quality would have an affect only on working children.
Of all the variables examined, only four were found to have an effect on the school performance of both working and non-working children. These were: a child’s perception of his/her school performance as “very good”, being male, at least one household member with a minimum of a secondary education and lateness for school.

Effects of work hours on school performance
According to the results of Discriminant Analysis, there is no consistent trend in average test scores of working children according to total hours spent for work. For example, the child who spent the least time per week engaged in work/chores (one hour) had a low test score of 1.42, whereas a child who spent 94 hours per week engaged in work/chores had a test score of 1.33 – a difference of only 6.33 percent. Overall, based on test scores, it can be argued that work has a negative affect on school performance; however, variations in the time spent on work do not seem to have an affect.

Work and school attendance

Attendance records indicated that there were significant differences between the school attendance of working and non-working children. Attendance of working girls was higher than attendance of working boys. Results indicated that combining school and work had a negative impact on school attendance of primary school children.

Results of linear regression analysis conducted to investigate the determinants of school attendance of working children indicated that the availability of someone to help with schoolwork outside of school, frequent completion of homework and higher tests scores had positive impact on the school attendance of working children. However, being male; increases in age, household size and number of days engaged in Economic Activity 1 (shining shoes, selling items on the street) and/or Economic Activity 4 (working in family owned shop and other economic activities); and decreases in school quality had negative impact on school attendance. Therefore, older boys from large families who work on street and in family owned shop are more likely to not attend school regularly.
Regression analysis showed that average test score was the only variable to have a significant effect on the school attendance of both working and non-working children. The study also found that total hours spent for work activity did not affect the school attendance of working children.
Recommendations
Based on analyses of the study findings, the following recommendation can be made for educators (policy-makers, teachers and principals) to improve the school attendance and performance of working children and/or to support the removal of children from work.
To ensure that what is taught is appropriate for the conditions of individual schools and responsive to the needs of working children, teachers need to be permitted flexibility in adapting the standard curriculum to their local conditions. In support of this, members of the Board of Education responsible for determining curriculum should be encouraged to participate in MONE training programmes designed to raise awareness on the needs of economically and socially disadvantaged children in general and working children in particular. Moreover, the schools in which the number of working children is higher may provide some remedial help for working students to increase their test scores.
Although this study found school quality to contribute minimally to the performance of working children, it is possible that this was due to the similar school and teacher characteristics of all schools included in the study, all of which are public schools located in low socio-economic neighbourhoods in urban areas in Ankara. In this regard, it is important to note that despite the significant difference found in the school performance of working and non-working children, the mean test scores of both groups were low. This suggests a need for improvement in some aspects of school quality, especially the indicators of teacher quality, class size, facilities and instructional materials, in order to positively contribute to the school performance of both working and non-working children.
References
Akabayashi, H. & G. Psacharopoulos (1999). The trade-off between child labour and human

capital formation: A Tanzanian case study. The Journal of Development Studies, 35 (5),

pp. 121-140
Akşit, B., Karancı, N., Gündüz-Hoşgör, A. (2001). Turkey Working Children in Three Metropolitan Cities: A Rapid Assessment. Geneva: ILO/IPEC.
Binder, M. & D. Scrogin (1999). Labour force participation and household work of urban school children in Mexico: Characteristics and Consequences, Economic Development and Cultural Change, Vol. 48, No. 1, pp.123-146.
Gülbuçuk, B., Karabıyık, E., Tanır, F. (2003). Research on Worst Form of Child Labor in the

Agriculture Sector of Turkey. Unpublished research report, ILO Ankara.
Heady, C. (2003). The effect of child labour on learning achievement, World Development, Vol. 31, No. 2, pp. 385-398.
Ray, R. & G. Lancaster (2003). Does Child Labour Affect School Attendance and School Performance? Multi Country Evidence on SIMPOC Data. Unpublished Report. ILO/IPEC.
State Institute of Statistics & ILO (2001). Child Labour in Turkey 1999. Ankara: SIS Printing Division, December, 2001.

Appendix 1
List of the variables used in Regression analysis and the ways in which they were formed:
Quantitative Variables

Total number of household members with a secondary-level education or higher.
Total number of household members who were employed last month: Total number of household members 12-years-old or older who were employed during the month prior to the survey.
Time since child started work to present: Child’s current age, minus the age at which the child started to work.
Total hours spent on household chores per week: Total hours spent on household chores, including time spent during the school week and on weekends.
Average hours spent on household chores per day: Total hours spent on household chores per week, divided by number of days per week child performed chores.
Total hours spent on household chores for the first semester: Total hours spent on household chores per week, multiplied by 12 (total number of week in the first semester).
Total hours spent on unpaid family work per week: Total hours spent on unpaid family work, including time spent during the school week and on weekends.
Average hours spent on unpaid family work per day: Total hours spent on unpaid family work per week, divided by number of days per week child engaged in work.
Total hours spent on unpaid family work the first semester: Total hours spent on unpaid family work per week, multiplied by 12 (number of weeks in the first semester)
Total hours spent on paid work per week: Total hours spent on paid work, including time spent during the school week and on weekends.
Average hours spent on paid work per day: Total hours spent per week on paid work, divided by number of days per week child engaged in work.
Total hours spent on paid work in the first semester: Total hours spent on paid work per week, multiplied by 12 (number of weeks in the first semester).
Total hours spent on studies per week: Total hours spent on studying, including time spent during the school week and on weekends.
Total hours spent on sports/play per week: Total hours spent on sports and play, including time spent during the school week and on weekends.
Total hours spent on other leisure activities per week: Total hours spent on other leisure activities, including time spent during the school week and on weekends.
Number of days child attended school: Total number of school days in the semester (90), minus number of days child was absent from school.
Economic Activity 1: Number of days per week spent engaged in shining shoes, selling low-value items on the streets.
Economic Activity 2: Number of days per week spent engaged in collecting discarded materials, washing car windscreens, carrying groceries at an outdoor market.
Economic Activity 3: Number of days per week spent engaged in domestic work in another household, as an apprentice at a local business, garage or other repair shop, as a bell-boy/dishwasher, housepainter, etc.
Economic Activity 4: Number of days per week spent engaged in work at a family-owned workplace; other economic activities.
School quality indicator: [(the value of indicator related to school characteristics)*0.40 + (the value of indicator related to teacher characteristics)*0.30 + (the value of indicator related to facilities in school)*0.30] (Formation of this variable was in line with the literature, which, groups school quality indicators into three categories as pertaining to either schools, classrooms or teachers.)

Qualitative Variables

Socio-economic status of household (Dummy)
Sex (Dummy)
Availability of computer at home (Dummy)
Child’s engagement in household chores, unpaid family work and paid work (Dummy)
Present job is first (or not) (Dummy)
Child’s satisfaction/dissatisfaction with both studying and working (Dummy)
Child’s satisfaction/dissatisfaction with attending school (Dummy)
Availability of someone to offer child help with studies outside school (Dummy)
Child’s perception of his/her school performance (Dummy)
Child’s treatment by teachers (Dummy)
Number of friends child has at school (Dummy)
Child’s treatment by peers (Dummy)
Child’s level of participation in extra-curricular activities
Child’s level of homework completion
Parents’ level of follow-up with school regarding child’s progress
Child’s attitude towards dropping out of school
Number of meals child eats per day
Number of hours child sleeps per day

*Paper presented at the Annual Conference of the European Educational Research Association, (ECER), September, 2006, Geneva.

The project was supported by ILO, Geneva.

^aMiddle East Technical University, ^bAnkara University ^cState Institute of Statistics

Contact:cennet@metu.edu.tr