You should not compare apples with oranges. But what about schools? As it turns out, comparing schools with schools can be just as problematic.
School composition can vary significantly between different deciles and within the same decile. Some of these differences include prior student academic achievement, the student’s parents’ level of education, and the student’s parents’ average income.
We know from the vast amounts of education research that these student characteristics (which are outside the control of the school) are significant factors in a student’s academic success.
So how do we objectively compare schools?
The United States, the United Kingdom, and Australia have all built contextualised value-added (VA) models. These models estimate how much a school (and teacher) adds to a student’s education, accounting for the socioeconomic background of the students and the characteristics of the school (roll size, single-sex, rural, etc) (i.e. the context of the VA model).
The estimates from a contextualised VA model indicate a school’s contribution to its students’ learning, relative to the contribution of the average school (after adjusting for differences in student and school characteristics).
These VA estimates are not about student achievement levels like NCEA level 1, 2 and 3. They are about demonstrating the effectiveness of a school.
The VA measures do not tell us why some schools add more value than others. However, they give us a starting point by identifying the effective schools from which we can learn from.
What are we doing in New Zealand?
Using Statistics New Zealand’s Integrated Data Infrastructure (IDI), The New Zealand Initiative is building New Zealand’s first contextualised VA model.
By combining administrative data in the IDI from the Ministry of Education, the Ministry of Social Development, New Zealand Police, the Department of Corrections, and Immigration New Zealand, our contextualised VA model will be able to adjust for each student’s socioeconomic background as well as the characteristics of each secondary school in New Zealand.
The goal of our contextualised VA model is to demonstrate that schools in New Zealand can be objectively compared. It will allow us to show the range of school effectiveness among secondary schools in New Zealand.
We suspect that once you control for each student’s socioeconomic background the variation in school outcomes (NCEA achievement, progression to tertiary education, and employment) is much smaller than what we see in school league tables.
The data will tell us, and you, the state of school effectiveness in New Zealand.