A recent story in the Chicago Tribune notes that Illinois’ NCLB waiver plan sets lower targets for certain student subgroups, including minority and low-income students. This, according to the article, means that “Illinois students of different backgrounds no longer will be held to the same standards," and goes on to quote advocates who are concerned that this amounts to lower expectations for traditionally lower-scoring groups of children.
The argument that expectations should not vary by student characteristics is, of course, valid and important. Nevertheless, as Chad Aldeman notes, the policy of setting different targets for different groups of students has been legally required since the enactment of NCLB, under which states must “give credit to lower-performing groups that demonstrate progress." This was supposed to ensure, albeit with exceedingly crude measures, that schools weren't punished due to the students they serve, and how far behind were those students upon entry into the schools.
I would take that a step further by adding two additional points. The first is quite obvious, and is mentioned briefly in the Tribune article, but too often is obscured in these kinds of conversations: Neither NCLB nor the waivers actually hold students to different standards. The cut scores above which students are deemed “proficient," somewhat arbitrary though they may be, do not vary by student subgroup, or by any other factor within a given state. All students are held to the same exact standard.
That may seem like a semantic point, but there is a big difference between varying schoolwide targets by student subgroup and establishing different passing scores for different students.
The second point I would like to make on this issue is perhaps even less frequently-acknowledged: The simple proficiency targets in the NCLB waivers get all the controversy, but setting different growth expectations for different students is also an inherent feature of most value-added and other growth models. Although these models do not necessarily need to include controls for student characteristics such as race/ethnicity and income (i.e., subsidized lunch eligibility), most of them do control for prior achievement (test scores in the previous year), and the models therefore do set different expectations for student growth, which are in turn used to assess teachers’ and/or schools’ contributions to that growth.
Now, many people draw a distinction between varying student expectations by race/ethnicity or income on the one hand, and, on the other hand, based on performance level in the previous year (or years). These are indeed qualitatively different approaches, but it’s worth noting that prior achievement is itself highly correlated with student characteristics such as income and race/ethnicity. Thus, in many important respects, it does the same thing as controlling for these characteristics, just in a more indirect manner.
So, is that a “bad thing?" I don’t think so. Both the targets set by states and growth models are less about expectations for student performance than schools’ contributions to that performance. As mentioned in a previous post, if you’re setting expectations for the performance of institutions, you must apply those expectations to measures that isolate, to the degree possible, that performance. To be clear, the measures employed by NCLB and NCLB waivers are so painfully simplistic and poorly-designed that they do not accomplish this goal. Growth models, though they have their own weaknesses, have greater potential to do so precisely because they adjust expectations in a manner designed (again, imperfectly) to account for schools' circumstances.
From my perspective, then, if anything deserves criticism here, it is not the fact that NCLB and waiver targets vary by student subgroup, but rather the fact that they are terrible measures of school performance, and we seem to be doubling down on them.
Furthermore, regardless of the measures employed, there is one more important underlying issue here: Narrowing discrepancies in performance between student subgroups will likely require schools serving traditionally lower-scoring students to significantly outperform schools in wealthier neighborhoods, whose students begin school much further ahead. If we really consider this goal a priority (and I think we should), it is going to take a lot more than just tinkering with targets. It’s going to take smart, substantial investment of resources. I for one would like to see a bit less talk about the proficiency rates school are expected to meet each year and a lot more talk about how they’re supposed to do so.
- Matt Di Carlo