Data sgp is a statistical analysis framework for large scale longitudinal education assessment data. The package utilizes quantile regression to estimate conditional density (and covariance) for each student in a system and then calculates growth percentiles, projections, and trajectories with this information.
Contrary to Value-added Models (VAMs), which are more dependent on classroom composition, data sgp’s models are general and apply equally across classes of students. Furthermore, data sgp provides various useful statistical analyses, including being able to compare student groups or examine variance in scores across time or grade levels.
SGPs (Student Growth Percentiles, or SGPs for short) measure where academically similar students stand in their distribution. SGPs function similarly to percentile ranks (i.e. lower SGPs indicate lower growth while higher ones suggest greater). For instance, an SGP of 45 indicates that they performed better (or at least as well) as 45 percent of academically similar students.
Students need at least two tests from different testing windows in order to obtain their current SGP score. Usually, one from autumn and one from spring is used, with current scores being calculated using “score match.”
If a held-back student has been successfully transitioned into their appropriate grade level during this school year, an SGP will be calculated using their test results from their original grade level and by comparing this student’s scores on recent tests with scores on tests from prior school years’ exams.
SGPs are widely utilized in New Jersey teacher evaluation. The New Jersey Department of Education uses district course roster submission data with teachers’ mSGP scores for evaluation purposes. A conversion chart converts SGPs from 1-99 scales into scores between 1-4 for use on NJ SMART evaluation system.
Teachers can use mSGP scores to measure their performance relative to other teachers in their subject area and with historical mSGP scores for themselves, helping identify areas for improvement while creating a plan on how they will do this.
SGP research typically utilizes small datasets in comparison to other applied statistics fields (for instance analyzing global Facebook interactions) – this practice has come to be known as medium data analytics.
mSGP conversion charts and tools assist teachers in understanding how their mSGP scores convert to teacher evaluation ratings of 1-4. Teachers can access these charts through either their evaluative system dashboard or by viewing individual student mSGP reports online in an evaluative system. Lower level functions that perform calculations – studentGrowthPercentiles and studentGrowthProjections – require wide formatted data; higher-level wrappers of these lower level functions like studentMeanSGPTrajectories require long formatted data.
