The data sgp package provides an open source set of functions to support the analysis of SGPs. The lower level functions sgpPercentiles and studentGrowthProjections require wide formatted data in order to run. The sgpData dataset, provided with the sgp package, is an example of this data format and can be used as a model to prepare data for these analyses.
SGPs use up to two years of historical MCAS test results to compare students’ current performance with the performance of academic peers in the same grade who have similar prior test scores. This is done through a statistical procedure called quantile regression that places the students’ previous test scores on a normative scale and makes it possible to identify their percentile rank.
A student’s SGP is a number between 1 and 99 that represents their relative growth, compared to the growth of other students with similar starting points. For example, a student with a score of 85 shows more growth than 85 percent of other students who started at the same point.
This is a measure of student achievement, not teacher evaluation. It is not unusual for a student’s SGP to move up or down during the course of one year. The reason for this is that the statewide trend in test results changes from year to year. This can cause a large percentage of students to experience more or less growth in their academic skills. A good example of this occurred in 2024 during the Covid-19 pandemic, where many students experienced less growth than usual.
In addition to evaluating individual student performance, SGPs can also be aggregated and used to analyze school or district level progress. For example, the graph below shows a comparison of student achievement (achievement scores on the M-STEP) and growth for students in a particular district. The data tools developed by Macomb and Clare-Gladwin ISDs allow users to download the sgpData dataset from the BAA Secure Site and create graphs that pair student growth with student achievement.
The sgpData dataset, provided by the sgp package, is an exemplar of the data format required for use with the lower level SGP functions. The dataset is structured with 5 columns that provide a unique student identifier, the grade level associated with each of the assessments in each year, and the student assessment scores for each year. The last five columns, SS_2013, SS_2014, SS_2015, SS_2016, and SS_2017, provide the scaled test scores for each of these years.
In the sgpData dataset, there is also an anonymized column, sgpData_INSTRUCTOR_NUMBER, that provides insturctor information associated with each test record. This allows the SGP functions to track which teachers a student was assigned to in each year of testing. This is important because in many cases, teachers will have multiple students with SGPs, and it is only the highest of these that will be used for teacher evaluation purposes. For this reason, it is important to ensure that the correct teachers are identified in each year of SGP calculations.