Page 336 - Contributed Paper Session (CPS) - Volume 4

CPS2253 Gabrielle Palermo
Figure 1: Summary of pupils moviments over time – Geres

Once observed, the additional pupils became members of the survey. From
wave 3 onward, the survey members that repeated the grade and continued
studying at a sampled school were tracked either, though their classmates
were not added to the survey. The members that moved to another school
which was not in the initial sample were not followed. Except for the case of
1
Rio de Janeiro, where 600 Geres pupils were transferred to non-sampled
schools in 2007, due to a local administrative decision (Brooke and Bonamino,
2011), so they were followed, and their new class peers were surveyed as well.

2. Methodology
Generalised Weight Share Method (GWSM) presented by Lavall´ee (1995,
2007) can be applied for either longitudinal and cross-sectional surveys.
Longitudinal surveys have their sample designed for the first wave. Some of
these surveys add new units to the original sample, and a few of them can be
from another population than the initial one, or it is quite complex to compute
their initial inclusion probability, considering the initial survey design.
A similar problem happens with some of the cross-sectional surveys, they
do not have a frame list to select a probabilistic sample, but their
correspondent target population has links with another framed population.
Thus, the subsequent waves population or unframed cross-sectional
population could be sampled through an indirect sampling when they are
somehow linked with a weighted sample.

1 Pupils from municipal schools.
325 | I S I W S C 2 0 1 9