Skip to main content

Table 4 Variables’ definitions

From: Does early tracking affect learning inequalities? Revisiting difference-in-differences modeling strategies with international assessments

Individual variables Definition
Population under study
 Natives Children with at least one parent born in the country
Social background
 Books at home Ln(n° books at home)
Children report the number of books at home, based on pictures depicting different numbers of shelves
Classification in PIRLS is 0–10; 11–25; 26–100; 101–200, > 200
Classification in PISA is 0–10; 11–25; 26–100; 101–200, 201–500, > 500
The last two classes in PISA have been aggregated, so the two classifications are now identical. We have considered the central value in each class (500 in the highest class)
In practice we use the following values:
Ln(5) = 1.61; Ln(13) = 2.56; Ln(63) = 4.14; Ln(150) = 5.01; Ln(500) = 6.21
 Parents with tertiary education At least one parents with tertiary education = 1
No parents with tertiary education = 0
Control variables
 Age Country-specific quartiles’ dummy variables (1°–4°)
We consider age in classes to allow for non-linear effects. The effect of age on test scores is unlikely to be linear. On the one side, the literature reports consistent evidence that older children tend to perform better (for example, in systems where regular children enter first grade in a given calendar year, children born in January tend to perform better than children born in December). On the other side, older children might be weaker. In some countries, there is flexibility in the age of first entry at school, so immature children might enter later, In other countries, poor performing children may be forced to repeat the school year, so older children are likely to be children who have experienced a grade failure
Quartiles are country-specific. This is particularly relevant for PIRLS,
as regular age and age variability of 4th grade children varies
substantially across countries
 Gender Female = 0, Male = 1