Everything you want to know about EVAAS, but were afraid to ask

  • We are not trying to have our students reach a score on a test, but instead to help a student grow by one academic year no matter where they start from.
  • Any student can grow even if they are not proficient.
  • If we concentrate on growth, proficiency will come.
  • All students matter and all students can grow.
  • We need to meet students at their level and help them grow from there.

Proficiency is not the only goal for teachers, getting your students to have at least one year’s worth of growth should also be a goal. Teachers should take their students and move them forward. It is true that proficiency is important and we should care about proficiency. Think of the relationship between proficiency and growth this way. If a student is behind, how else are we going to get them to be proficient than to help the student grow academically. If a student comes to class, already proficient then our goal is to get them to grow. It is not fair to ask a student who is ahead of their classmates to wait while everyone else catches up.

Given a student’s testing history, across subjects, what is the student likely to score on an upcoming test, assuming the student has an average schooling experience, an average years’ worth of growth.

Projections/predictions are not about predicting the future. They are about assessing students’ academic needs today.

Using the predictive model is like using all the times a runner receives on all of their races and using them to predict how they will score on their first half marathon.

It is not just the tests from the subject matter or just the past 3 years of tests, but it is really all test from all years and all subjects are used to determine a student’s true academic ability. Each test a student takes in NC is another piece of the puzzle in understanding that student and what they are capable of doing.

Why use NCE?
  • NCEs are used instead of percentiles when the data set is not normally distributed.
  • NCEs allows the use of tests from different years, subjects and grades because they are rescaled to 0-100.
  • Along the distribution of student performance, NCEs are even intervals.
  • A 1 NCE change is the same, no matter where on the distribution it is.
  • An NCE of any number is the same for each grade or subject.
NCEs are like the student’s place in line as compared to the counterparts. As long as a student ends the school year at the same place in line as compared to their counterparts as they started the school year, that is an average year’s worth of growth. If a student moves up or back a place in line as compared to their counterparts, that is either positive or negative growth.

Before the school year starts, every student is placed in line compared to each other on how they are expected to score. At the conclusion of the year they line those students up again according to how they actually did compared to their counterparts. The differences is growth.

It is important to understand that with NCE, it is not trying to get a student to make a particular score on a test, but instead to do as well as expected in comparison to their counterparts. How did students taking the same test the same year do as compared to how they expected to do.

Standard Error is a way of normalizing all of these differing sets of data for comparison purpose.

First let’s discuss what goes into the standard error.
1. The size of the data set, the number of tests that can be attributed to a teacher.
2. The history of the data set, how much we know about each student. This comes from the number of prior tests each student has coming to a teacher.

Holding everything else constant, the larger the data set or the more prior tests the students have the more accurate the predictions and therefore the smaller the standard error.

Teacher estimate is the combined value added of all the students attributed to that teacher in that school year. Students who do not have enough prior tests to have a prediction/projection and any student not enrolled for 140 days for a year long class or 70 days for a semester long class, will not be used for analysis. These students’ scores will show up in data display, but they are only there for learning purposes.

The teacher estimate should never be used to compare teacher or to make determinations of teacher effectiveness.

To determine teacher index, the same as a teacher’s Standard 6, the teacher should divide the teacher estimate by the standard error. The resultant is the teacher index.

The teacher index, Standard 6, is how teachers are compared and teacher effectiveness is determined.

For example, a teacher may have a teacher estimate of -1.5 and a standard error of 1.0. This does not truly tell us that teacher’s effectiveness. We need to determine how far from an average year’s growth that is. To do this we take the teacher estimate of -1.5 and divide it by 1.0. The resultant is a teacher index of -1.5. This teacher index is with 2 to -2 window of meets growth for the state.

Another example. We use the same teacher estimate of -1.5, but now we have a standard error of .5 We determine the teacher index in the same way and get -3.0. This number is below -2.0 and therefore is significantly different from the average teacher in the state.

In these 2 examples, both teachers had a teacher estimate of -1.5, but only the second teacher is significantly different from the average teacher in the state.

Standards 6 & 8

Status Update