Saturday, March 14, 2009

School Correlations

We discuss instances where correlation does not imply causality a great deal at Chez Badmom. A year ago, Sandra Tsing Loh asked me to write up my observations comparing Iris' school 3 blocks east of us to the school she would have attended if we had bought a house 3 blocks west of us. It's a bit late, but she's not paying or grading me.
(I've noted previously that I am not a stranger to late assignments.)

It helps if you read what she has to say about school statistics. is the easy-to-use, test-driven school listings resource visited by one in three U.S. K-12 families (30 million users). While the compilation of data is incredible, has an unfortunate “one number” rating system, based (essentially) on a school’s API score as averaged over the entire student body. This number which correlates most strongly with homogeneous populations of affluent families and high real estate values (in the same way SAT scores are strongly linked with family income). Parents tend to wrongly conclude they must move to an expensive suburb (where starter homes are $1.2 million) to find quality teachers for their children.
As luck would have it, the 2008 average home prices in 90266 and 90278 (as estimated by ESRI) are $1,188,703 and $845,655. Don't faint. Our neighborhood is still a bit rough around the edges and some people won't even visit it after dark. LA is just plain expensive.

Let's call the school in the expensive suburb school A. gives it a 10 (out of 10) rating and school B, Iris' school, gets an 8. The academic performance index or API, is another one number school summary. The two schools receive 943 and 885 respectively.

Which peer group would you like to buy for your kids? (You are not buying an outcome, only a peer group.) The one with the higher achieving kids? Before you shell out $1.2 million for a shack near school A, take a closer look at the numbers.

School A School B
zip code 90266 90278
median income $141,734 $87,447
avg SFR price $1,188,703 $845,655
Greatschools rank 10 8

CA API 2008 943 885

Parent Educ %

HS dropout 0 5
HS grad 2 13
some college 7 23
college grad 36 39
postgrad 51 21

CST Math, Kids with Postgrad Parents
Grade 2
453.2 448.3
Grade 3 452.2 478.2
Grade 4 430.6 451.0
Grade 5 439.2 *

* No comparison can be made for 5th graders because the sample size at school B is too small.

More than half the parents in school A hold graduate degrees. If you compare apples to apples, and only the kids whose parents hold graduate degrees, the results are flipped. The kids in school B have higher Math test scores. (Which is not the same as saying they are better at Math, but that is a subject better left for another time.)

I am posting details for only the Math scores because the English scores don't show such a strong pattern. Kids at school A score slightly better in English, but I suspect it is because no students in school A are English learners. The average English scores for grades 2-4 are 412 and 409 respectively.

3 points is in the noise, but a math gap of 26 points is significant.

The difference is even more striking if you know that school B allows any child who will turn 5 before the December 2 cutoff date to enroll in Kindergarten. School A requires that any child born in the Spring, Summer or Fall be evaluated for 'Kindergarten readiness'. Kids in school district A are a half year older than kids in school district B. (One September birthday child from school district A attends school B because the principal insisted she repeat Kindergarten until she gained more reading fluency!) Kids in district A are also more likely to have attended academic preschools, have hired tutors and a quiet place to study.

If you rely solely upon, you will never learn this stuff. Go visit your neighborhood schools. Talk to some kids, parents and teachers about what they like and don't like about their school.

I am not advocating that you try to enroll your kid in school B. Instead of chasing high test scores, I believe that concerned parents should channel their energy into helping to raise the quality of their neighborhood school. And you can't do that if you are killing yourself working around the clock to afford the most expensive neighborhoods.


  1. CST Math, Kids with Postgrad Parents
    Grade 4 451.0
    Grade 5 *

    * No comparison can be made for 5th graders because the sample size at school B is too small.

    It's kind of funny that the 4th grade number is reported based on 11 students, but the 5th grade number can't be reported because it is based on 10 students.

    I guess if you're going to draw a line, you have to draw it somewhere.

    Asterisk aside, it seems that something might be wrong in the 5th grade at School B. Test scores for the overall population and college-educated-parents population took a tumble relative to earlier grades.

  2. They didn't take a tumble; they aren't the same cohort of kids. These scores reflect neighborhood gentrification.

  3. Nope, that's not it.

    In 2007 the kids in 4th grade had a mean CST Math score of 407.8. 58% Advanced, 25% Proficient, 12% Basic... only 5% Below Basic.

    In 2008 the same kids in 5th grade had a mean CST Math score of 358.9 (49 point drop), and 23% of the students were rated Below Basic.

    How about the preceding cohort? In 2006 they were in 4th grade and had a mean CST Math score of 422.4. Only 4% were Below Basic. In 2007 those same kids in 5th grade had a mean score of 384.0 (38 point drop), with 19% rated Below Basic.

    It looks like a consistent problem to me.

    I checked a few other random elementary schools to see if they showed the same pattern. And one did show a somewhat similar pattern, but three others did not. So the explanation is not just that fifth graders are universally scored more harshly.

  4. Hmm, I hadn't looked at it in the same detail you did.

    No, the 3rd graders in one year are not necessarily the 4th graders the next. There is a substantial amount of student mobility in our area.

    Additionally, in 2006, the district switched from U of Chicago's Everyday Math series in favor of Houghton Mifflin Math. I prefer the EM math series, but that's considered an elitist thing to say. The rationale for the switch was that HM produces higher std math test scores because EM was too perplexing for a significant fraction of the kids. I never taught elem. age students so I don't know if those are valid reasons.

  5. The 2008 API scores are out and the school slipped from 10 to 9. We are busy high-fiving each other in anticipation of less overcrowded classrooms next year.

    We also found out that Iris' 4th grade teacher does not believe in coaching kids for the standardized tests. She found better uses for classroom time. She taught 4th graders rudimentary algebra such as solving linear equations for x.

  6. Ashley23:37

    I gladly take my kid to an 8 out if 10 school... but what if you start comparing 8 out if 10 to 3 out of 10s. this is the different between the school in my district from the school 1 block outside the district. its THAT issue that many of my mommy friends are facing that are causing them to move. some even have to settle for a 1 out of 10. in our cases even just a 6 will do. its sad theres even that big of a difference and kids cant get some quality education depending on where they live. and you have to go through hell applying for charter and magnet schools only to get let down.

    1. I just looked up one of my former MS. It is a 2 out of 10. Oddly, my sister and both remember some of the upsides. We had a fantastic orchestra teacher. The GATE classroom had only a handful of kids and we all had individually-tailored independent study curricula.

      It sucked when we missed recess due to gunfire in the neighborhood. When that happened, I could go to classrooms of my favorite teachers instead. So that wasn't as bad as it sounded.