Data SGPs provide an essential window into a student’s academic progress and serve as an indicator of future achievement targets. SGPs are calculated by taking into account both recent assessment results as well as prior tests from that same student to determine their SGP score, then using quantile regression techniques such as conditional density matrices to generate percentile growth projections or trajectories that represent their estimated path to their goals.
Data used to calculate Student Growth Percentiles is drawn from MCAS assessments administered in Massachusetts public schools since 2005-06. In order to receive an SGP, students must first take at least one valid and consecutive assessment in their subject and grade; only then can SGPs be calculated accurately; otherwise they cannot be obtained. SGPs cannot be generated for science, writing, EOC biology and EOC math tests.
Each year, state growth percentages (SGPs) fluctuate to reflect student performance trends on an aggregate level. While such movements may be noticeable (e.g. during a Covid-19 pandemic) they can also be subtler depending on individual students’ progress compared to academic peers. Changes in SGPs for individual students reflect their relative academic growth compared to peers within their cohort.
Students determine their Student Growth Progressor by comparing their current MCAS test score with those of academic peers who have scored similarly on prior MCAS exams in that subject area. Peers are identified through quantile regression and include students from all demographic groups and educational programs (such as sheltered English immersion and special education). Because Students A and C have had different MCAS score histories and thus, different academic peer groups, their SGPs may vary.
Student B and C demonstrate different SGPs because of different prior scores used to calculate their current SGPs. Timing of prior scores can have an enormous effect on how much growth each student shows each year.
SGPs can be studied at three different levels: state, school and district; however they’re most useful when examined within specific groups like schools, districts or subgroups of these – where average SGPs tend to fluctuate due to having smaller sample sizes for analysis.
The sgptData_LONG dataset is an anonymized, panel data set that offers assessment results from 8 windows (3 windows annually) in LONG format for Early Literacy, Mathematics and Reading content areas. Included are 7 required variables for SGP analyses such as VALID_CASE, CONTENT_AREA, YEAR, ID, SCALE_SCORE GRADE ACHIEVEMENT LEVEL. In addition there are demographic/student categorization variables used by summarizeSGP to create student aggregates by creating student aggregates using summarizeSGP function; contact us if interested utilizing this dataset before its availability for download in Spring 2022.