Student Growth Percentile (SGP) Analysis

Gambling Blog Jan 13, 2024

The data sgp package, which is installed when one installs the SGP software, provides the capability to analyze longitudinal (time dependent) student assessment data. The package includes exemplar WIDE and LONG format data sets, sgpData and sgpData_LONG, that can be used to conduct SGP analyses. Both formats allow each case/row to represent a unique student and columns to represent variables that are associated with the student at different times.

A student growth percentile (SGP) describes a student’s performance as compared to other students with similar prior test scores (“their academic peers”). SGP is an estimate of a student’s progress and can be used to evaluate a teacher’s effectiveness. SGP calculations can be complex and difficult to interpret, but they are important for assessing student learning and understanding performance trends over time.

SGPs are calculated using a statistical algorithm that compares a student’s current performance with an estimate of their previous performance. These estimates are based on the likelihood that a student’s actual score would be in the same range as their estimated score, taking into account factors such as test difficulty and time spent in the grade level. While SGPs are calculated from probability distributions, they can be interpreted using descriptive statistics such as mean and median.

An SGP is calculated by dividing the difference between a student’s current assessment score and their estimated prior test score by the student’s overall mean or median score on all tests taken in their grade level. The SGP calculation also takes into account the overall standard deviation of each assessment and a student’s ability to perform well on tests taken in their grade level.

The sgpData data set, which is installed when one installs the data sgp package, contains longitudinal assessment data for five years of a single student’s performance. The first column, ID, provides a unique student identifier and the remaining columns, GRADE_2013, GRADE_2014, GRADE_2015, GRADE_2016, and GRADE_2017, provide the student’s assessment score for each of these years. sgpData_LONG, which is also provided with the package, contains the same longitudinal data but in a long format where each row represents a year of student testing and each column represents a variable that was measured at each year.

The sgpData_LONG data set can be used with the SGP analysis tools to calculate an SGP and a number of other descriptive statistics. The SGP package includes vignettes that detail how to use the sgpData_LONG data and the SGP analysis tools. These vignettes are available online. The sgpData_LONG vignette is particularly helpful for those who are new to longitudinal data analysis and SGP. This vignette includes a step-by-step example of calculating an SGP for a single student. It also includes an introduction to the SGP analysis tools and a description of the logic behind each of them. SGP packages also include lower level functions that help simplify operational calculations for users. For example, the SGP package’s sgpMetrics function provides the means to compute student growth percentiles using simple formulas.