School Beat: The Affluent Parents Index, a.k.a The Academic Performance Index (Part 1)

by K.C. Jones on May 26, 2005

This week’s School Beat column is the first of a two-part overview of the main statistical measures of California’s public schools. Today’s piece will provide a quick overview of those measures. Next week’s column will look at what the data is for schools in the San Francisco Unified School District. The data used for these two columns can be found at http://api.cde.ca.gov/

Our kids have been subjected to another year’s worth of mandated testing. Now it’s time for the numbers to be crunched. I figure it’s a good time to take a close look at how their test scores are translated into the schools’ Academic Performance Index (API), School Rank, and Similar School scores we see and hear so much about. Parents shopping for schools can’t help but pay attention to them. It’s hard not to pay attention to your kids’ school scores too – even if you’re fundamentally opposed to high-stakes or mandated testing. They’re just so tidy and simple and easy to compare. But what is the API? And what are the School Rank and Similar Schools scores all about?

SCI, API, and all that
The Academic Performance Index (API) is the most prominent measure of a school’s academic performance. The API is combined with other data about the demographic traits of the children attending a school. The demographic traits are combined into the School Characteristic Index, or SCI. This combination of API and SCI data produces the infamous School Ranks and Similar Schools scores.

A school’s API score is based on the standardized test scores for all kids in the school. API scores range from 200 to 1000. The goal for every school is to score at least 800. Schools not meeting that 800 goal are assigned API targets to be measured the following year. A complex formula is used to combine the individual results of the separate tests children take –CAT/6, CST, CAPA and CASHEE–into an API score.

The School Rank score is computed from the school API data. All schools’ APIs are sorted into ten buckets. The School Rank is based on which of the ten buckets a school’s API score falls into.
The School Characteristics Index is pretty obscure to most parents, but it is arguably the most important. The SCI is a statistical regression of the following demographic factors:
* Pupil mobility
* Pupil ethnicity
* Pupil socioeconomic status
* Percentage of teachers with full credentials
* Percentage of teachers with emergency credentials
* Percentage of English Language Learner, or ELL, pupils
* Average class size
* Multi-track year-round programs
Pupil mobility is measured as the percentage of students who are new to the school. Pupil ethnicity is African American, American Indian, Asian, Pacific Islander, Filipino, Hispanic, or White. The other factors appear to be relatively simple to understand and measure.

There are many details about how the regression of these data is computed, what the criteria are for determining when an SCI is valid or not, etc. But the basic idea is to toss these data into a regression to compute a single SCI using relatively simple statistical regression methods. Note that separate regressions are computed for elementary, middle and high schools, so SCI data can only be compared among schools of the same type.
All this hand waving results in a nice, tidy SCI number. Is it valid? Does it mean anything? Who cares? It’s a useful number that does, by definition, correlate with the relative educational advantage or disadvantage of different schools. Another way to look at it is the SCI attempts to take into account significant factors affecting student performance that are cannot be controlled by the school itself.

With the API and SCI in hand, the Similar Schools computation is easy. All schools in California are sorted by type then by SCI. Each school’s closest 100 neighbors in that list are deemed to be that school’s similar schools. Finally, this list of 100 similar schools is sorted by their school API score and then divided into 10 buckets, or deciles. The decile that spans the school’s API is then that schools Similar Schools Rank.

Those three numbers: API, Rank, and Similar Schools are the most visible data for each school. There are lots of other data related to compare year over year results for each school. As new test results become available, the API computation changes. The SCI regression changes too. The most salient thing to know is that Base API reports refer to the results of the year in question while the Growth reports are an attempt to recast the previous year’s results in a way that will be comparable to the next year’s results.

KC Jones is the parent of a child at Clarendon Elementary School and the co-founder and co-moderator of the Yahoo! Group “sfschools,” which is an active public schools online discussion group. Go to http://groups.yahoo.com/group/sfschools to join the sfschools list.

Filed under: Archive