Learning and Educational Achievement in Punjab Schools (LEAPS) - 2003
Whether one is in favor of private education or not, it is here to stay and there is a critical need to understand this new environment. Unfortunately, little is known about the private sector and what its growth implies for the provision of education. There are important questions we need to answer before engaging in productive debate about how education can be best provided in the Pakistani context. For instance:
a. Where are private schools setting up? Are they only being established in urban areas and only for the elite?
b. What is the quality of education in private sector schools? How does it compare to public schools?
c. Are the poor being left out? Is the private sector creating two classes of people in Pakistan—those who can afford private education and those who cannot?
d. What is the effect of private schools on government schools?
Kind of data
Public Release Version 2.0
Rural Punjab, Pakistan
Unit of analysis
How much a child learns depends on teachers, parents and the child herself. How these three coordinate and work together also depends on the head-teacher and the educational institutes that support the delivery of education. Our research strategy reflects this belief. We survey both schools and households and test children to assess how much they are learning. Here is a brief overview of the survey structure:
a. Teachers: Rosters with basic information for all teachers in the schools and detailed interviews with the class-teachers of the children tested.
b. Head-teachers: A detailed questionnaire with the head-teacher with basic information about his/her background and teaching experience.
c. Schools: We also collected information on schools, the children who come, the fees charged (private schools) and their current needs.
d. Children: For a random sample of 10 children from those tested the survey collects basic information on their households (parental education, assets and brothers and sisters), how far they travel to get to school, and their height and weight.
e. Households: We complete household surveys for 16 households in every village, with information on what parents know and what parents do with regard to their children’s education. In addition, these surveys contain the basic information on expenditures, assets, education and health that will allow us to look at the relationship between these factors and educational performance. For instance, when the mother is sick, does the child perform worse in school?
Producers and sponsors
Tahir Andrabi (Pomona College), Jishnu Das and Tara Viswanath (World Bank), Asim Ijaz Khwaja and Tristan Zajonc (Harvard University)
The sample comprises 112 villages in 3 districts of Punjab-Attock, Faisalabad and Rahim Yar Khan. The districts represent an accepted stratification of the province into North (Attock), Central (Faisalabad) and South (Rahim Yar Khan). The 112 villages in these districts were chosen randomly from the list of all villages with an existing private school. This allows us to look at differences between private and public schools in the same village. Although these villages are thus bigger and richer than average villages in these districts, we believe this is a forward-looking strategy and the insights earned here will soon be applicable to a significant fraction of all villages in the country.
Deviations from sample design
The attrition has been remarkably small, averaging 3-4 percent in each year.
The LEAPS survey was designed using a two-stage sampling strategy. First, mauzas were selected with equal probability from the universe of all rural mauzas with at least one private primary school. Mauza selection was stratified at the district level -- 46 mauzas were selected from each district. Several mauzas were subsequently dropped because their private schools were found to have closed, and a few very large mauzas with more than 20 schools were also dropped for logistical reasons. This resulted in a final sample of 112 mauzas, 37 of which are in Attock, 43 in Faisalabad, and 32 in Rahim Yar Khan.
Following the selection of mauzas, a sample of sixteen households were chosen from a census of all households within each sample mauza. This selection was also stratified -- 12 households were randomly selected from the population of households with at least one child enrolled in class three, and 4 households were randomly selected from the population of households with children between 8 and 10 years old, none of whom were enrolled in school. Households without children between the ages of 8 and 10 were not included. This sampling strategy was designed to provide the counterpart to our survey of class three children at the school level.
Depending on the empirical question of interest, it is sometimes important to use the included household probability weights. Households with enrolled children and households with no enrolled children are weighted differently, and without weights estimates of population means like enrollment rates for class three eligible children will be inaccurate. Note that the use of weights does not allow researchers to recover all the possible parameters of interest in a village. Households without children between 8 and 10 years old are not included in the sample, so while the LEAPS data gives accurate estimates of class three enrollment rates, it cannot be used to estimate the enrollment rates of other populations. Furthermore, because mauzas were selected from the population of mauzas with at least one private school and were not selected with probability proportional to population, findings from these mauzas cannot be used to make generalizations about the district in which they reside.
Dates of collection
Data collection supervision
Tracking Children: In 2004 we tested 12,000 children in 838 public and private schools in Grade 3 and re-tested them in Grades 4 (2005) and 5 (2006). In 2006, we also included Grade 3 children, increasing the total number of children tested to
25,000. These children all need to be tracked through the year to find out where they are the next year. The table below shows for instance, the status of child-tracking as they moved from Grade 3 to Grade 4. In the transition children could (a) drop-out (b) remain in the same school and be promoted; (c) remain in the same school and not be promoted; (d) switch schools within the village and be promoted (in which case they would be tested in another school) and be promoted; (e) switch to schools within the village and not be promoted and (f) switch to schools outside the village or leave the village all together. Although close to 1800 children out of 12,000 were no longer in the same class-school combination that they would have been if they did not switch schools and were promoted, we were able to determine the status of all except 500. In one previous study, authors could track only 10 percent of the children compared to our rate of above 95 percent.
Surveying Households: We have now created a longitudinal dataset of 1800 households across the 3 years, and in each of the years there are close to 750 children on whom we also have information on learning from the school testing exercise. This is the first database in low-income countries that combines detailed household information (including consumption aggregates) with school-level inputs and learning.
The LEAPS project consists of a variety of questionnaires distributed to different groups in each village in order to obtain a complete picture of the educational environment.
School Survey: Head teachers and school ownders were asked a variety of questions about about infrastructure, prices, costs and other facilities available in the neighborhood of the school.
Teacher surveys: The LEAPS project administered three sets of teacher surveys. A shorter roster was administered for all teachers in the school and for all teachers who had left the school in the previous two years. This roster yields information on above 5000 teachers in the LEAPS project schools. A longer questionnaire was administered to the teachers of the tested children. This questionnaire includes detailed socioeconomic information about the teacher and yields data on just above 800 teachers. In addition, a questionnaire was also administered to the head-teacher (where the head-teacher was different from the class teacher) with questions on management practices and bonus schemes, along with other modules.
Child Tests: All children in Class 3 (approximately 12,000) were tested in the LEAPS project schools with specially designed tests in Urdu, Mathematics and English. These tests were administered by the LEAPS team to ensure impartial test circumstances. Further, for a sample of 10 randomly selected children in every class (roughly 6000 in total), a short questionnaire was administered to the child with information on parental literacy, family structure and household assets (in classes with less than 10 children, all children were chosen).
Household surveys: Information on the educational inputs that children receive from home, a full-fledged household questionnaire was fielded for 1800 households in the sampled villages, with a special focus on covering those households with a child enrolled in class 3. To ensure that we could compare the activities of enrolled with out-of-school children we also sampled households with eligible kids who were not in school in a stratified fashion.
Andres Yi Chang
In order to access the data, you will be required to register. Registration is a simple one-step process and is only designed to allow us to keep a track of the users of this data.
Use of the dataset must be acknowledged using a citation which would include:
- the Identification of the Primary Investigator
- the title of the survey (including acronym and year of implementation)
- the survey reference number
- the source and date of download
Disclaimer and copyrights
To the best of our knowledge and ability, the data has been purged of identifying variables and errors of coding.
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.