Replies

What California does is it crunches all the data in all schools, such as racial proportions (% white, % black, etc.), % qualifying for school lunch, % mobile (in school for 1 year o less) and 20-30 other data points. It uses a complex regression model (I can lecture for three days on this subject, unfortunately, after which you will pass AP stat and go on to work at the RAND Corp, or you will kill me in a rage) to assign weights to these factors insofar as they determine test scores - because they do, and how.

Based on this model they can say that a school belongs within a series of a dozen or so bands. Within a band the schools are considered to be dealing with a student population that is similarly difficult to teach. Simplified (way too simplified) example would be School 1 is 50% White and 50% black and 30% Free lunch, while School 2 is 40% Asian 60% Hispanic and 50% Free lunch (just BS examples off the top of my head). The model would consider them similar schools and their test scores are properly comparable. It is correct to say that if School 1 has an API (the “grade” for schools) of 700 and School 2 has an API of 600, School 1 is doing much better than School 2. This is further crunched into a similar school ranking, from 1-9. 9 is good, it means a school is doing very well given the material it has to work with.

School 3, Which has a population of 50% White and 50% Asian with school lunch 10% is NOT a similar school. Their API may be 850 but they cant be compared with School 1 or School 2, because frankly their kids would take the test and give the school an 850 API even if all the teachers spent the school year in a coma.

They are compared with THEIR similar schools, like School 4, 30% white 70% Asian 20% School Lunch which may have a 950 API. In which case it becomes clear that with just an 850 API, with the kids they have been given, they are really doing very badly. They would get a Similar Schools rank of 1, and the school board should get on their case.