Newsweek on Wake's move away from socioeconomic diversity

The federal guidelines for accountability models state:

"The accountability model must establish high expectations for low-achieving students, while not setting expectations for annual achievement based upon student demographic characteristics or school characteristics."

There's lots of information about accountability models online and NC's was approved by the federal government. EVAAS is the deployment of that model.

Models are not allowed to use student demographics as INPUT because it skews results. Dr Sanders didn't say RESULTS from the model couldn't be examined and grouped. If factors like FRL have a correlation to student achievement, this correlation will be apparent in the results.

Here's an excerpt from a document written by Dr Sanders. Although he is referring specifically to SES, the same is true of any demographic characteristic:

Unlike our advocacy that socioeconomic (SES) variables should not be explicitly included in value-added modeling efforts, it is indeed a fact that many other proposed value-added models include controls for a variety of student-level and higher-level (classroom, school, community) demographic and socioeconomic variables. The following are the reasons for our advocacy that these variables should not be included.

1.
At the student level, if the entire multivariate, longitudinal data vector is fitted, then the inclusion of SES variables is not needed to insure fairness. To the extent that SES factors persist over time, then these influences are already contained in the observational vector. This has been empirically confirmed by Ballou, et al. (2004) and also more recently by Lockwood and McCaffrey (2007) who conclude:
“William Sanders … has claimed that jointly modeling 25 scores for individual students, along with other features of the approach, is extremely effective at purging student heterogeneity bias from estimate teacher effects … The analytical and simulation results presented here largely support that claim.”
Additionally, by including these variables, one is directly assuming that there will be different expectations for two students with the same prior achievement pattern who come from different SES communities.

2.
Whether to include adjustment for SES variables at the group level is much debated among serious investigators of value-added modeling. We recommend that these adjustments not be made. To the extent that teacher assignment patterns to schools are related to some degree to teacher effectiveness, then adjustment for group SES factors will over-adjust the estimates and can camouflage the fact that students in certain schools are not getting an equitable distribution of the teaching talent. It has been documented in many studies that novice teachers are less effective than veteran teachers; it has also been documented that schools with a higher concentration of poor and minority students also get a disproportionate number of beginning teachers (see National Center for Education Statistics, Monitoring Quality: An Indicators Report, December 2000.) This is one example of why an adjustment for group SES factors may hide influences that policy makers will need to address.

-------------------------------------------------
End of excerpt.

And another excerpt about subgroup reporting:

-------------------------------------------------
Once a multivariate, longitudinal data structure is completed, there is a wealth of positive diagnostic information available for educational decision makers— teachers, principals, curricular specialists, superintendents, school board members, etc. The use of summative value-added measures as one component of accountability systems is important; but in our view, the diagnostic information is of greater importance. For those districts for which we are providing analytical services, a series of reports for each school are produced. These are delivered via the web and can be accessed only by individual educators who have authorized passwords.

Some of these reports are:

For each grade and subject, presentation of progress rates of students by prior achievement level, either for the whole school or any demographic subset with comparisons with previous cohorts. This enables educators within each school to ascertain which subset of students is not making the appropriate progress.

Projections for each student to various academic endpoints. This enables local educators to identify which students, say, are not on trajectories to meet high school graduation requirements with sufficient time to plan different curricular and instructional strategies for these at-risk students, or to identify students who are meeting all proficiency requirements yet are not on a trajectory to be prepared for a more technical college major.

Some principals and superintendents have learned to use the flexible projection reports to plan for the number of ‘seats’ that will be required to accommodate all students who are on trajectories to be successful in Algebra as 8th graders, based on the students’ projections at the end of 6th grade. It has been found that in some schools, the number of ‘seats’ available is less than the number of students who could benefit from a more rigorous course.
 
Some teachers are finding it to be helpful to have all of the prior testing information available in an intuitively understandable web interface for each student as they enter their classrooms.

-------------------------------------------------
End of excerpt.

And in Wake County, guidance counselors still don't have access to these reports.