How DESE Calculates Student Growth Percentiles (SGP)

data sgp

Student growth percentiles (SGP) are norm-referenced statistics that tell us how a student’s current year MCAS test score compares to the scores of other students with similar prior scale scores. For example, a student whose growth percentile is at the 90th percentile has made more progress than 90% of the students who performed similarly in the past.

SGPs are important for several reasons. They help us understand how much a student has learned in one year and can help guide our instructional decisions to ensure that students are making the best possible progress. They also allow us to make fairer comparisons between students because they account for the fact that students enter school at different levels and are unlikely to have identical previous MCAS scores.

DESE calculates SGPs in ELA and math for students in grades 4 through 8, and in grade 10. The SGPs provide a measurement of relative progress, which is intended to complement the traditional measures of academic achievement (i.e., test scores and class rank).

To calculate SGPs, DESE takes a sample of student data from the state and compares it to a standard population. This is done using the latest available MCAS data from each school. Using the same process each year, we create a distribution that is used to compute the SGPs for all students in each grade. This distribution is based on a distribution from multiple years of compiled data so that any fluctuations in individual year data are smoothed out.

The sgpData_INSTRUCTOR_NUMBER data set contains an anonymized, student-instructor lookup table that provides insturctor information associated with each students test record. The data set is a key piece of the SGP calculation and is an integral part of the dataset that is provided to schools. The sgpData_INSTRUCTOR_NUMBER is used to calculate both baseline-referenced and cohort-referenced SGPs.

A limitation of baseline-referenced SGPs is that they are sensitive to changes in the tests, particularly with respect to the number and type of questions. For example, a change in the number of test items can have a significant effect on the median baseline SGP for a given student.

A further limitation is that baseline-referenced SGPs can be systematically related to prior year scale scores in some grades. In particular, for students with high prior year scale scores in both 4th and 6th grade math, the baseline SGPs were consistently lower than the cohort-referenced SGPs for these students. We are working on ways to address these limitations. Please contact us if you have any suggestions or would like to discuss them further.