Ohio will use the URM approach for the new science reports Necessary since science OAA is not administered during consecutive years First time Ohio has used the URM approach; non-SOAR users will require conceptual help understanding this model and the URM science V-A report High school value-added analysis will always be done in with the URM approach. So, when Ohio rolls out it’s new assessments in several years, they’ll be looking at a URM analysis at the high school level for value-added purposes. The reason is because students don’t take consecutive testing in the same subject from year to year.
All users (including SOAR users) will have a single EVAAS® account All LEAs (including SOAR districts) will receive the state's value-added results for math and reading grades 4-8, and science grades 5 & 8
Science will not be included in state accountability Inclusion of historical data will result in increased report accuracy due to fewer gaps in the data Report views containing all tested students ’ data are more useful for diagnostic purposes than WKC students’ report views
Science will not be included in state accountability Inclusion of historical data will result in increased report accuracy due to fewer gaps in the data Report views containing all tested students ’ data are more useful for diagnostic purposes than Where Kids Count (WKC) students’ report views
Re-rostering: Teachers will appreciate the ability to see reports containing their current students ’ data Projection reports/pie charts: 1-year projections valuable for diagnostic purposes 2-year projections used for accountability purposes (transparency to the AYP Growth Measure for Ohio) Add student deadlines: July 22, September 16 and January 13 http://www.ode.state.oh.us/GD/Templates/Pages/ODE/ODEDetail.aspx?page=3&TopicRelationID=117&ContentID=75219&Content=110636
The AYP Growth Measure is not value-added analysis. It’s looking at individual students. It’s a more simplistic growth model. Less than a dozen states use a growth model for AYP.
Two standard errors increases statistical certainty to 95% Exception: 2011 only : requires 3 years minus (red) to go down When discussing the LRC, it’s often asked if high achievement performing districts can show growth. Here’s a stat: In 2010, of the 15 highest performing Ohio districts based on Performance Index (all had 108 or above), their value-added composite (1 standard error in 2010) was as follows: Above Expected Growth (Green) – 10 Met Expected Growth (Yellow) – 2 Below Expected Growth (Red) – 3
RttT provides teacher-level value-added reports statewide. Innovation dollars in RttT to explore value-added analysis in other grades/subjects. ODE has a frequently asked question document on the RttT web page regarding HB 153, Evaluations, Performance-based compensation and seniority.
Value-Added Toolkits: Unlike the existing toolkits the updated version will focus on the diagnostic use of V-A reports as opposed to providing introduction to and help with interpreting V-A reports Useful for augmenting training Problem-based approach How much growth was produced grade level-by-grade level across your school? Which students benefited most (and least) from your school ’s curriculum & instruction? What other schools in your district or across the state performed well with particular groups of students? (Teacher-level) How can members of your team use each teacher ’s individual strengths to improve the performance of the entire team? Key reports organized by audience District, Building, Teacher Fall distribution Refreshed On-line Courses: Focused on using data to solve problems and set goals Formative instruction-type questions embedded throughout the courses with immediate feedback
Educators have no control over the prior achievement levels of the students who enter their classrooms.
At a minimum, we should expect each student to progress one academic year in one year ’s time. Students who are academically behind, must grow more than one academic year in one year ’s time to close the gap.
Value-added provides a picture of student growth regardless of students ’ achievement levels. Value-added can help us understand whether high-achieving students are making enough progress to sustain or even improve their achievement levels. Value-added can help us understand whether low-achieving students are making enough progress to close the proficiency gap. Talk about this slide from a perspective of a school with Jacob-like students or Adam-like students since value-added is a group effect versus individual student results. In this slide, Student A is currently above the proficiency bar, but is losing ground relative to proficiency. In this slide, Student B is not yet proficient, but is closing the gap on the proficiency bar.
This is the Power of Two, Achievement & Progress information combined
This is a scatter plot of poverty (horizontal) with Achievement as measured by the Performance Index (vertical). Each dot represents one school. School performance in terms of achievement continues to be correlated with the level of poverty in the school. This result is consistent with other data we have comparing student background factors and achievement in schools. It is also consistent with most national studies. This is based on a regression equation and resulted in a negative slope.
This is a scatter plot of poverty (horizontal) with a measure of Growth - the Value-Added Gain Score (vertical). Each dot represents one school. There is no correlation between the two. School performance in terms of effects of schooling is largely independent of the level of poverty in the school. This results has been shown each of the 3 years for which we have value-added data. This is based on a regression equation.
This chart shows data from a variety of schools, and portrays a more complete picture of schools ’ effectiveness. Schools D, G and K are low-achieving and low-progressing. These schools are not likely to change their results without a change in curriculum, school processes, personnel or instruction. Schools F and B are currently above the “Standard”, but students, in general, are not making a year’s growth in a year’s time. Schools A and H have not yet reached the “Standard” passage rate, but because they are high-progressing, they are making up ground on the proficiency bar. Chances are good that the students in these schools will soon become proficient and meet the standard. Schools C, E and J are high-achieving and high-progressing.
This is a simple analogy of the value-added growth measure. However, the metric is much more complex. Value-added is a measure of present (observed score) achievement minus past (baseline score) achievement. Because of the measurement error inherent in testing situations, statistical procedures must be used to produce a more reliable and valid observed score and baseline score. Student mobility, incomplete testing records and measurement error are all addressed in Sanders ’ value-added methodology.
For tests to be used in the EVAAS value-added calculation they must exhibit three properties. Tests must be highly correlated, but not necessarily perfectly correlated to curricular objectives or standards. Tests must have stretch. Stretch is determined across the series of tests a student has taken, not only by the stretch in a single test. There may be a few isolated cases where students who top out on a test, but these cases do not typically rise to the level of statistical significance. Finally, tests must be reliable and produce consistent results for students.
For tests to be used in the EVAAS value-added calculation they must exhibit three properties. Tests must be highly correlated, but not necessarily perfectly correlated to curricular objectives or standards. Tests must have stretch. Stretch is determined across the series of tests a student has taken, not only by the stretch in a single test. There may be a few isolated cases where students who top out on a test, but these cases do not typically rise to the level of statistical significance. Finally, tests must be reliable and produce consistent results for students.
This is an example for teaching purposes only. EVAAS calculations are more statistically sophisticated to ensure that all students are included in the analysis and that confidence intervals reflect the entire history of student testing. The EVAAS methodology also allows future data to refine past data estimates for more accuracy. NCEs are Normal Curve Equivalents. The NCE scale enables longitudinal data connections and the definition of a growth standard that does not change from year-to-year. This model links student data from one year to the next. Obviously this is representation of a low performing group of students
BACKGROUND INFORMATION: A normal curve equivalent (NCE), indicates a student's rank compared to other students on the same test (similar to Percentile): NCEs run from 1 to 99 with 50 at the center of the base year distribution. BUT: Normal curve equivalents convert scaled scores to an equal-interval scale Since NCEs are represented on an equal-interval scale, scores can be averaged and compared longitudinally The NCE scale enables longitudinal data connections and the definition of a growth standard that does not change from year to year NCEs represent where a student ’s score would place that student relative to student performance in the state’s base year 2010 for OH See conversion table at: http://www.cal.org/twi/evaltoolkit/appendix/nce2percentile.pdf A normal curve equivalent (NCE), indicates a student's rank compared to other students on the same test (similar to Percentile): NCEs run from 1 to 99 with 50 at the center of the base year distribution. BUT: Normal curve equivalents convert scaled scores to an equal-interval scale Since NCEs are represented on an equal-interval scale, scores can be averaged and compared longitudinally The NCE scale enables longitudinal data connections and the definition of a growth standard that does not change from year to year. NCEs represent where a student ’s score would place that student relative to student performance in a state base year 2010 for OH
In a URM Analysis: looking at the relationship between how students typically score on a test to how they actually perform A student’s test history acts as their own control in creating the predicted score
This is one of the most helpful reports for teachers Diagnostic reports are especially helpful in identifying student patterns Students are placed in the state quintile subgroups based on the average between their current and prior year scores. This is an approximation of their prior achievement level. Minimum of 5 students is needed in any subgroup in order to display a bar for that subgroup.
There is no interpretation difference between MRM and URM Diagnostic Reports for utility purposes. Based on prior achievement level, students are placed in subgroups based on how they are predicted to score in reference to the entire pool. Minimum of 5 students is needed in any subgroup in order to display a bar for that subgroup.