Detection and selection of high and low efficiency schools based on PISA 2015 results

Selection and initial characterisation of schools with high and low effectiveness: Results by Autonomous Region PISA 2015 results, Results
Once the models for the selection of schools with high and low effectiveness were applied, their identification was carried out. First of all, an analysis of the selected schools by autonomous region can be carried out, with the following chart showing the results.

Source: Adapted from Gamazo, Martínez-Abad, Olmos-Migueláñez & Rodríguez Conde (2018)

It is observed that certain autonomous regions have a clearly higher proportion of schools with high effectiveness (high residual) in relation to the selection of schools with low effectiveness, such as La Rioja, Navarra, Murcia, Madrid, Aragon and, especially, Castile and León. In the case of selected schools with low effectiveness (low residual), we find several regions that make important contributions: Asturias, Canary Islands, Extremadura, Andalusia and, especially, the Basque Country.

The case of the Basque Country is worthy of more in-depth study since, despite it being an autonomous region with one of the highest socio-economic levels in Spain, most of the schools selected were determined to have low effectiveness. This autonomous region’s own linguistic educational models and the language in which the PISA performance tests were carried out may have had a spurious effect on these results.
Selection and initial characterisation of schools with high and low effectiveness: Results by school type PISA 2015 results, Results
Among the schools selected from the analysis of high and low effectiveness, 19 were private (5.2%), 102 grant-maintained (27.9%) and 245 state (66.9%), a breakdown that is almost exactly the same as the original breakdown of Spanish schools available in PISA 2015.

It should be noted that prior to the application of the statistical techniques for the selection of schools, a filtering of the database was carried out in which only schools that obtained at least partial results from 20 or more pupils in PISA 2015 were selected. In an initial stage, schools with a small sample of pupils were disregarded, leaving an initial breakdown of 897 schools (66.6% state, 28.2% grant-maintained and 5.2% private).

A comparison of independence confirms that the breakdown of schools with high and low effectiveness obtained cannot be considered different to the initial breakdown of schools, and that therefore, the selection of schools is representative of the initial sample. When these results are given, we can continue working on the next stage of the study with certain confidence.

Contextual variables associated with academic performance (Reading, Mathematics and Science) in the final models calculated PISA 2015 results, Results

After the application of hierarchical linear models to the PISA 2015 test data to isolate the part relating to the average performance of schools not attributable to the effect of these contextual variables and obtain an average residual per school (what we understand in the project as school effectiveness), the following significant variables were found.

Group	Variable	Effects
Pupils	Gender	This is one of the most influential variables: While boys obtain higher results than girls in mathematics and science, in terms of reading comprehension, girls achieve a higher average level of performance.
	Family socio-cultural level	Pupils who have a greater variety of cultural and digital resources in their homes and whose parents have higher levels of education and jobs associated with a higher academic/economic status achieve higher average performance in the 3 areas.
	Repeaters	Repetition of an academic year is a significant factor in the three areas assessed, with repeating pupils showing lower levels of performance.
	Number of school changes	The number of school changes that pupils experience throughout their academic careers has an inverse effect on school performance: the greater the number of changes, the lower the level of performance in the 3 areas.
	Migratory status	While 1st-generation immigrant pupils are handicapped compared to natives in terms of performance, these differences dissipate with 2nd-generation immigrant pupils.
Schools	Socio-cultural level of the school	It seems that the average socio-economic level of the families whose children attend the school is the most influential variable at school level, significant in the 3 areas. Schools whose pupils come from families with a higher socio-cultural level achieve higher average performance.
	School size	In the case of Science, the size of the school has an inverse effect: Larger schools tend to show slightly lower average performance in Science subjects.
	Repeater rate	The repeater rate in secondary Year 4 of the school has a significant influence on performance in Science and Reading Comprehension. Classrooms in which there is a higher rate of repeaters show lower performance.
	Percentage of female pupils	The percentage of female pupils in a school appears to have a positive effect on performance, not only in Reading Comprehension, but also Science.

Effective Schools

Detection and selection of high and low efficiency schools based on PISA 2015 results