An Introduction to the Student Growth Percentile (SGP)
When you’re a student, teacher, administrator or policymaker looking for ways to measure student growth, you probably know about the Student Growth Percentile (SGP) as one of the most widely used and reliable measures. However, you may be less familiar with how it works or how to use it to make informed decisions about instruction and improvement. This article will provide a quick introduction to SGP and its uses and will suggest some practical applications for it to help you improve your educational practices.
In general, SGP compares a student’s performance to that of his or her academic peers nationwide. These peers are students in the same grade with a similar achievement history on Star assessments. The goal is to establish an objective measurement of student progress that can be compared over time and across schools.
SGP estimates are based on statistical models that utilize data from several years of Star assessments to identify a student’s latent achievement traits and then estimate the extent to which those traits predict future performance. The estimated growth standard is then compared to the average student’s performance to determine how much a student needs to improve to meet an identified target for each year.
To minimize estimation errors, SGP analyses rely on least squares regression modeling and Bayesian inference to estimate student growth traits. In practice, these methods cannot guarantee the accuracy of any given estimate. As a result, it is important to understand the uncertainty associated with SGP estimates.
For more information on the methodology used in SGP, you can refer to the SGP documentation and FAQs. You can also download a free copy of the SGP software from the SDCC website.
The sgpData_LONG data set includes a LONG format version of the WIDE data set, plus an anonymized instructor lookup table sgpData_INSTRUCTOR_NUMBER. This table provides insturctor information associated with each student’s test record. It is important to note that a single student can have multiple instructors associated with their test records in a given year.
The sgpData_LONG format is preferred for SGP analyses because it provides numerous preparation and storage benefits over the WIDE format. It is also recommended that the SGP functions studentGrowthPercentiles and studentGrowthProjections be run using LONG format data if possible.