How to Analyze Student Growth Percentiles (SGP)

When analyzing student growth data, one of the most important things to know is how your students are growing compared to their peers. The best way to do that is through student growth percentiles (SGPs) that measure relative student performance in the most recent assessment compared with their prior performances on previous assessments. These percentiles range from 1 to 99, with higher numbers representing higher relative growth. For example, a student with an SGP of 45 means that they out-grew the 45th percentile of their peer group.

The SGP package has a set of lower level functions (studentGrowthPercentiles and studentGrowthProjections) and also high-level wrapper functions (abcSGP and updateSGP) that perform these analyses in a more concise way. These higher-level functions work with LONG formatted data whereas the lower-level functions require WIDE formatted data. It is generally easier to run operational SGP analyses using the LONG versions of these functions unless a specific reason exists for needing a more compact version of the analysis.

This article will walk through the steps needed to run a basic SGP analysis with exemplary data sets provided by the DESE for the state of Massachusetts. These exemplars include an anonymized, teacher-student lookup table (sgpData_INSTRUCTOR_NUMBER) and a student assessment record file (sgpData_STUDENT).

While most SGP analyses focus on comparing students with their peers within the same year, it is also possible to compare students across years. These comparisons are known as multi-year SGP analyses and allow educators to establish official achievement targets for their students based upon state-wide growth standards (e.g. a 2 year target). Then they can use the SGP methodology to quantify how much student progress is required in order for a student to reach their goal, and what the student’s current rate of growth is in terms of percentage of their peers.

A key thing to note is that, in order for a growth percentile to be calculated, the student must have valid test scores from two of the previous assessments on their profile. This is why a student’s SGP is not calculated in grade 3, because science is only tested in grades 5 and 8, and it is why the growth percentile for a grade 10 student does not compare their 10th grade MCAS score with their 8th grade test score since there is no MCAS testing at the grade 9 level. The next section will take a closer look at the calculations that go into producing these growth percentiles for a student. Then it will discuss the two most common uses of this information: window specific SGPs, which allow teachers to quickly compare a student’s progress between distinct time windows, and current SGPs that provide a quick check-in on a student’s progression over the course of an academic year. In both cases, the bulk of the SGP analysis is in the data preparation and preprocessing steps. If the data is prepared correctly then the subsequent analyses should be relatively straightforward.