SGPs are calculated by comparing students’ current MCAS test scores with those of academic peers with similar score histories over a two year period, including special education students (such as sheltered English immersion) and low income groups. Growth is placed onto a normative scale using quantile regression; SGPs then allow comparison of relative performance on a common scale for use in student achievement evaluation, educator evaluations and school improvement planning.
The Student Growth Percentile Dashboard displays statewide growth results for all grades 4-11 students. A feature added this year enables districts to view individual student SGPs by downloading “Student Data File” under Reports on the BAA Secure Site, copy and pasting only data into one of several interactive tools and viewing student growth graphs with associated growth percentages for MCAS tests taken since fall 2015. These tools also enable users to see student graphs with growth percentages for MCAS exams taken since fall 2015.
While the SGP Demonstration enables districts to examine trends in student growth by demographic group and grade level, it should be remembered that its data are still not suitable for high-stakes educator evaluations. Currently, state is working on creating a process to ensure reliability of SGPs and associated student growth percentiles for use in high-stakes educator evaluations.
Before performing SGP analyses, it is recommended that you spend some time familiarizing yourself with R. For optimal performance, SGP requires at least 4GB of memory available on a computer to facilitate computations of both percentiles and projections; moreover, many higher level functions assume sgptData meta data is included within SGPstateData data sets.
Through the Student Growth and Achievement Package (see link above), numerous high quality analyses of SGP data are made available to users. These examples demonstrate how to utilize various functions from this package in order to build confidence intervals or perform advanced analyses on student growth and achievement data.
The sgptData_LONG dataset includes LONG format data for 8 windows of assessment (3 windows per year) covering three content areas. Seven variables are required for performing SGP analyses: VALID_CASE, CONTENT_AREA, YEAR, ID, SCALE_SCORE GRADE and ACHIEVEMENT_LEVEL (for constructing student growth percentiles).
The mSGP dataset is limited to grades with sufficient baseline performance data for SGP creation. This includes grades that administered both MCAS and NECAP tests this year: 6th, 8th and 11th grades. This data will be released as an early look preview so educators can become acquainted with it before being used in high-stakes educator evaluations.