If you’re a teacher and looking for ways to improve your classroom, you may have heard of student growth percentiles (SGPs) – a statistic that compares students’ raw test scores from previous years with the average performance of academically-similar peers. But what exactly do these numbers tell you? And how can you use them to support your teaching decisions? Read on to learn more.
The data sgp in this article are based on the sgpData spreadsheet provided by the NCLB/PDE Data Portal, which is available here. The first column of the spreadsheet, ID, provides a unique student identifier; the next five columns, GPA_2013, provide prior-year grades for each student; the last two columns, SGP_2013 and SGP_2014, show the student’s current SGP and expected SGP, respectively. The spreadsheet also includes a sixth column, Teacher SGP, which is the sum of the teachers’ classroom-level SGPs.
To create a SGP, students are grouped by grade level and compared with other students who have the same combination of prior-year grades. This allows for a fair comparison of students, regardless of their differences in achievement history. This method is also less sensitive to differences in classroom composition than VAM models, which only account for the average teacher’s class.
Despite these benefits, it’s important to keep in mind that SGPs are relative measures of student growth. They do not tell us whether a particular student’s growth is adequate or whether the educational system would consider it appropriate. SGPs can also be misleading if a student’s high raw score in a given test section is due to luck rather than her ability.
In addition, SGPs are not the only measure of student progress; they do not take into account other factors that influence academic achievement. For example, a student’s socioeconomic status, family life, and educational environment can affect her growth.
For these reasons, it is advisable to format data in the long format for all but the simplest, one-off analyses. In fact, if you’re planning on running SGP analyses operationally year after year, we recommend using the higher-level wrapper functions instead of the lower-level studentGrowthPercentiles and studentGrowthProjections. The higher-level functions are designed to work with LONG format data and can simplify the process of updating and managing operational analyses by appending new years of data on top of currently existing long sets. They also often assume the presence of state specific meta-data in the embedded sgpstateData column.