Geoscience Reference
In-Depth Information
and frequented school. The result depends on the relative localization of each pupil
and her/his school;
- how many schools have an underprivileged social profile? Such a query
consists of selecting schools according to a summary of the students who frequent it.
For instance, selection can be done using the % of students in the “underprivileged”
social category in comparison with the same percentage at the level of the region;
- what are the schools whose share of underprivileged pupils has increased
during the period? This query requires an analysis of the evolution of the targeted
indicator's trajectory.
Queries can multiply. It can be seen that the complexity increases quickly when
the queries question evolutions that mobilize composition or spatial relationships,
which are themselves evolving.
2.3.4.3.
Step 3: to analyze the relationships between school success the students'
and the schools' characteristics
Simple methods of statistical bivariate analyzes (correlation, variance analysis,
chi2 testing and following the qualitative or quantitative nature of the variables at
stake) allow examining the relationships between differentials in school success and
social inequalities. At the student level, the point is to test whether there is a
significant relationship between their social category and success (measured, for
example, from their success or not at the brevet (French high school exam), or from
the grade obtained). At the school level, the main interest lies in the relationship
between the success rate and the social profile of the school (derived from the
proportion of pupils whose parents belong to certain well-off social categories or on
the contrary underprivileged, or whether from the first component of a principal
component analysis on the % associated with the different social categories).
The next step consists of developing multivariate models (multiple regression or
logit models) that help to explain the “success” in function of several explanatory
variables. At the student level, a logit model allows, for example, showing that the
social category of the student and the social profile of the school she/he attends are
similarly discriminating in the public sector, whereas the social profile of the school
is more discriminant in the private sector [FRA 14a]. At the school level, a multiple
regression model allows showing that
everything being equal
with regard to the
social profile, the interschool differentials in terms of success at the brevet (national
exam), are related to the importance of the migratory turbulence of the school (i.e.
the importance of incoming and outgoing flows of pupils during high school years).
The analysis of the residual map of such a model allows highlighting the local
specificities, with varyingly important success rates than what was expected from
the model. Figure 2.14 shows the results for public sector schools as an example. It
