Education Outcomes National Panel Survey (NPS) 2002-2008

The Education Outcomes Panel Survey (NPS) was designed as a panel survey based on a subsample of households interviewed in the 2002/03 Inquérito aos Agregados Familiares (IAF), a national household income and expenditure survey conducted in all provinces of Mozambique from July 2002 to June 2003. The NPS data collection took place from September 2008 to February 2009 and it was performed by a contractor in Mozambique (KPMG), with World Bank and UNICEF field supervision.

The NPS sampling frame consists of enumeration areas (EA) that were drawn to correspond to a particular set of months of the 2002/03 IAF, namely March to May 2003, since it is expected that the IAF has a nationally representative subsample of EAs assigned each quarter. It is important to highlight that the NPS data is nationally representative at the rural and urban areas, but not representative below this level. The main reason is that the IAF sample was clustered to maximize efficiency in the data collection process across a 12 month period, while the NPS sample, due to costs constrains, includes only 3 months. Therefore, the NPS sample does not have enough geographic dispersion to be representative at the province level or below.

All IAF households in the enumeration areas during the months of March-May were included in the NPS sample, resulting in 221 EAs and 2,234 households. This sampling strategy was chosen to reduce the effect of seasonality in the panel analysis when comparing the 2002/03 IAF data to the 2008 NPS data for the same sample households. Originally it was planned to interview all the IAF sample households in these EAs during the same month in which they had been interviewed for the 2002/03 IAF. However, because of delays in the survey planning process, the data collection for the NPS was postponed took place from September 2008 to February 2009.
The survey was designed to target eligible children/student (i.e. children aged 0-17 y.o. in 2003 or members enrolled in school in 2003) from the IAF sample. The households in the NPS sample were divided into 2 categories based on their status in 2003:

A. Target 2003 households. These are households that meet at least one of the following criteria:
· Households that had at least one child 0-17 years-old in 2003 (see question a13 in the questionnaire )
· Households that had someone in primary or secondary school in 2003, in spite of age (see question a14 in the questionnaire)
B. Alternate 2003 Households (14% of original NPS sample)

For the households that did not have any children or student in 2003 but were part of the IAF sample and were in the NPS enumeration area, the following two questions were asked to the first person who was found in the alternate household in 2008:
· Does this person's 2008 household currently have anyone who is between 5 and 17 years of age? (see question a15 in the questionnaire)
· Does this person's 2008 household currently have anyone who attending primary or secondary school? (see question a16 in the questionnaire)

If the answer was YES to either question (a15 or a16), the interviewer proceeded with the entire questionnaire. If the answer was NO to both questions, the interviewer stopped the interview.

In sum, target households are the source for the panel of children, while alternate households were included to supplement sample size.

There were two types of tracking in the NPS, that of households and that of children/students who split from the original 2003 household and joined new households in 2008. If the entire 2003 household moved in 2008, the field team would gather their new contact information with local leaders, neighbors, friend, etc and follow and interview the household at their new location, provided the household moved within the district (the survey only followed households/children that moved within the district level). New members of the household were also included in the interview.

If the 2003 household was split in 2008 and the members who moved out were a target member (children/student in 2003) who had moved within the district, then the team followed the individuals and interviewed both the original household (if a target member still lived there) and the split household.

The screening for tracking are in section B1 of the questionnaire. A member would be tracked if b100a =1 (this variable is an indicator of whether the member was target member, i.e. less than 17 y.o. in 2003 or attending school in 2003), and b108=2 (the member no longer lives in the household), and b111 <=2 (the member moved to the same village or district). If all these conditions were met, questions B112 (should the member be tracked?) should be 1 (YES) and the household should be followed. The variable "sp" indicates whether the household was the original (sp=0) or a split household (sp>=1).

In case all target members (b100a=1) moved out of the household, the interviewer should end the interview with the original household at question B114.