Education

Growth Models: Not as Simple as They Appear

March 13, 2009 2 min read

When experts talk about accountability under NCLB, they agree on one thing: The future lies in growth models. Discussions usually end there, never delving into the complexities of what makes a good growth model, how to design one, or whether they accomplish what NCLB sets out to do.

Charlie Barone jumps into the morass and reports on some of the technical problems and design flaws with Tennessee’s growth model. In a report for Education Sector, he writes that the Tennessee model doesn’t measure whether the state’s students are going to meet NCLB’s ultimate goal: universal proficiency by the 2013-14 school year. Instead, under the statistical methods in the Tennessee model, a student is considered proficient if he or she is on track to being proficient in three years, based on the trajectory of past test scores.

In short, proficiency is a 'moving target,' always three or more years away, in perpetuity. There is nothing to ensure that, over the long run, [a student] moves ahead toward a cumulatively higher level of performance over successive years.

The end result is that a school may make AYP in the 2014 even if “a large proportion” of students aren’t proficient, Barone writes.

LINKS:
On his blog, Barone reinforces another point made in the Ed Sector report. “More statistical sophistication means less transparency,” he writes there. He also calls out the statisticians who created the Tennessee model for failing to fully explain the mathematics behind it.

Ed Sector’s Chad Aldeman explains Zeno’s Paradox— the mathematical theory that describes why a student could be considered proficient without ever getting there under a model that uses a three-year trajectory.

Oklahoma City teacher/blogger John Thompson says that the discussion of the paradox “can offer no insight into what should be expected of a teacher in a high poverty neighborhood school.”

Former teacher Sam Rosaldo cheers Tennessee’s growth model for favoring accuracy over simplicity.

For academic research on growth models, go to this page.

FINAL NOTE:
In a comment on my latest post on the Title I setaside, Barone light-heartedly complains that I reported that he wrote the administration “is” giving in to public school lobbyists. He actually wrote the administration “may” be doing so. As someone who has fought over whether a law says grant recipients “shall” or “may” do something, Barone understands the importance of word choice. I apologize for the error, and I thank him for being good-natured about it.

As long as I’m giving away links to Charlie’s blog, check out the exchanges between him and Thompson on the Title I setasides. By the end, I was nodding my head, thinking “Yeah, that was surreal.”

Related Tags:

A version of this news article first appeared in the NCLB: Act II blog.