MCAS Student Growth Percentiles (SGPs)
Unlike traditional student assessment reports, which tell you the student’s achievement level, SGPs show how much progress (or growth) a student has made. SGPs are based on a comparison of students’ raw score performance to the performance of their academic peers who have taken similar assessments in previous years.
Student growth percentiles are calculated from a statistical model that looks at trends in the statewide performance data over time. The model takes into account all students who have valid and consecutive test scores for a given subject and grade. This allows for the calculation of a growth model for every student in each school. It also means that the student growth percentages can be compared between schools in a district or between students across all schools within the state.
SGPs can be viewed by teachers and administrators on the Star Growth Report and on the MCAS Results by Student Group page of School and District Profiles. This information can help teachers understand trends in their students’ performances, compare those trends with the averages for all students in their grade level, and explore how student SGPs are affected by various demographic groupsings including racial/ethnic status, special education and multilingual learning.
Q. How are growth models calculated and how should they be interpreted?
SGP models are calculated using a series of quantitative regression analyses, which examine the relationship between the student’s assessment score history and the number of points they would have had to score in order to be at or above their current test-taking level. The model then calculates the difference between the actual score and the expected score, resulting in the growth percentage. This percentage is then used to rank the student’s score relative to other students who have taken the same assessment.
The sgpData csv file provided above contains the raw data that can be used for these analyses. The first column, ID, provides the unique identifier for each student. The next five columns, SS_2013, SS_2014, SS_2015, SS_2016 and SS_2017 provide the scale scores for each of these assessment years. The sgpData file also includes a vignette that provides more detailed instructions on the use of wide-format data for SGP analyses. Adam Van Iwaarden has created a python script that can be downloaded from his GitHub repository for performing these types of analyses. The script uses the sgpData file and the vignette to provide an example of a typical SGP analysis. Using this script is straightforward, but the bulk of the work required to perform an SGP analysis involves data preparation. This is why the SGP team has developed the sgpToolkit. The toolkit includes a script for importing the sgpData csv files and another for creating a graphical user interface to perform an SGP analysis. These tools are designed to simplify the process of analyzing large-scale longitudinal assessment data and providing insights about student growth.