« Are Lexus Lanes a HOT Idea? | Main | Friends for Life »

September 02, 2005

High Stakes Testing: There is No Free Lunch

The newest Washington Assessment of Student Learning--the tests that the state uses to gauge students' academic achievement, as well as schools' success or failure under the US "No Child Left Behind Act"-- was just released.  And as I was perusing this Seattle Times data supplement, I couldn't help but notice the correlation between income and achievement -- namely, that elementary schools with a large share of students signed up for free or reduced-price school lunches (a sign that the school serves a low-income population) also have a small share of 4th graders who pass the test.

Free_lunch_vs_waslThis should come as no real surprise; by now, it's been pretty conclusively established that poverty impedes academic performance.  But just to amuse myself, I played with the numbers a bit this morning -- and was astonished by just how strongly incomes predict school performance.

Take a look at the graph to the right: on average, every 10 point increase in the share of students in a given school who receive free or reduced lunches correspondsto a 6 point decline in the share of students who passed all sections of the WASL.  In other words, the poorer the student body, the worse the school does on WASL.

Just to be clear:  the graph uses data from all 358 Puget Sound elementary schools covered by the Seattle Times data supplement.  Each dot represents about 36 of those schools, ordered by the share of students in each school who receive free or reduced-price lunches.  Students who are eligible for those programs come from needy families.  And what the numbers demonstrate, with abundant clarity, is that schools that serve relatively needy communities also tend to have relatively poor performance on the WASL.

Of course, there are some schools that have above-average WASL pass rates (at least this year), even though they serve relatively poor communities; and some that have relatively low pass rates, even though few of their students are signed up for lunch assistance.

That said, school lunches turn out to be an extremely powerful predictor of school performance: among Seattle and Puget Sounds elementary schools, between half and two-thirds of the variation in school performance on the WASL could be explained based solely on the share of students whose families need help buying them lunch.  That doesn't leave much room for other factors, like the quality of teaching, the effectiveness of the curriculum, and the like.

I know little about educational policy, but this sort of thing can only undermine my confidence in the use of high-stakes tests to gauge how schools are actually doing. Obviously, we want schools to be effective, and we hope that each year's crop of fourth graders does better in school than the previous year's; and tests like WASL probably play a useful role in identifying trends, and clarifying which educational strategies seem to be working. 

But when you combine all of the factors over which a school has little control -- the socio-economic status of its students, shifts in community demographics, random variation in the level of academic interest in the school population, and the fact that students who take the same test twice can get different scores -- it seems, well, sort of silly, or at least misguided, to treat the WASL as an accurate, unerring gauge of the inherent quality of any particular school.

Posted by ClarkWD | Permalink

TrackBack

TrackBack URL for this entry:
https://www.typepad.com/services/trackback/6a00d834573a7069e200d8345e9eaa69e2

Listed below are links to weblogs that reference High Stakes Testing: There is No Free Lunch:

Comments

Wow. When one looks for indicators, the more an indicator encapulates, the better an indicator it is.

You found a good 'un, sir.

Posted by: Dan Staley | Sep 2, 2005 5:07:14 PM

I taught high school for 30 years and your statements certainly reflect my experiences.

I entered teaching from case working and was focused on helping those kids who most needed help, but the reality is that generally kids who didn't need help continued to do far better than those that needed help because of their family situation.

Posted by: Loren | Sep 2, 2005 8:53:12 PM