When Growth Goes Nowhere

Dr. Matthew Strom

Dr. Matthew Strom presented on limiting the inferences based on student growth percentile calculations.  The presentation covered issues with inferences at a student-level, inferences related to “one year’s growth in one year’s time,” inferences related to what is enough growth, and inferences related to closing achievement gaps.  The data collected for the presentation included over 20,000 student test records over the course of the last four years of AIMS testing.  Dr. Strom spoke extensively to limiting the use of the student-growth metrics for high-stakes decision making.

The presentation was divided into three activities.  The first activity included developing a better understanding of student growth percentiles and what influences the output for the growth metric.  This part of the presentation was built to help attendees understand that data from several years of spring testing has an influence on the current year growth metric.  Ultimately, Dr. Strom showed that there is quite a bit of variance in student-growth measurements even though students might have moved across the scale score spectrum at the same rate from the previous year to the current year.

The second activity included building a better understanding of what is enough growth, as measured by SGP, in order to get students to move across a criterion-reference measurement like a performance indicator.  Dr. Strom showed that only very high student growth percentiles are likely to result in a student moving across performance indicators and this supports the idea that growth often measures movement within a performance indicator as opposed to across a performance indicator.  He also spoke to the idea that consistent growth measurements that were above average were no better than a coin flip at resulting in a student moving across a performance indicator.

The final activity brought everything back to the original purpose of standardized assessment as prescribed by NCLB and linking this idea to SGP measurements at a school-level.  NCLB was designed to be the educational civil rights act of the 21^st Century.  In reality, when looking at SGP measurements aggregated at a school-level there is no relationship between closing achievement gaps and high growth schools.  So, while accountability desires to ensure a certain degree of educational equity we must be careful when inferring that high growth schools, as measured by SGP, are schools that close achievement gaps.

Overall the theme of the session was to be cautious when using district-level SGP data to make any high-stakes decisions.  Decisions involving teacher evaluation, principal evaluation or program evaluation should seek to use the data in a descriptive and qualitative fashion as opposed to quantitatively setting desired quotas.  Dr. Strom cited the principal designer of SGPs quite frequently, Dr. Betebenner, and he quoted him at the end of the presentation when Betebenner wrote in 2009, “The annually performed, normative growth percentile results are blind to any changes in the efficacy of the education system over time.”

Terry Locke

Chandler Unified School District