The data sgp is an analytical tool for district staff to view student growth paired with achievement. The tool was created by Clare-Gladwin and Macomb ISDs to help educators use the data available to them in their districts.

SGP analyses use a statistical technique called quantile regression to compare students’ assessments scores with those of their academic peers. The academic peers are identified based on their grade and content area and include all demographic groups including student eligibility for specialized programs (e.g., sheltered English immersion and special education).

A student’s assessment score is compared to the average of their academic peers and the resulting percentile rank is reported to the teacher and student. The teacher and student can then use the percentile rank to compare their performance to others in their school, content area and across the state.

This year, 2024, the SGPs displayed in the dashboard include up to two years of test score history including the Badger Exam. The Badger Exam’s statewide performance levels were quite different from those of the WKCE and Forward exams that preceded it. SGPs are a method of evaluating a student’s performance over time and provide an alternative to the more commonly used mean value-added growth model.

As with all models, there are pros and cons to the SGP model. The biggest advantage is that the SGP model allows comparisons of performance between teachers and schools over a very large range of students, which makes it more realistic and useful than a simple mean value-added calculation. The drawback is that the SGP model is computationally intensive and requires a significant amount of memory to compute.

To run an SGP analysis, the sgp package, available for the free R software environment, must be installed. Fortunately, R is available for all major platforms and can be downloaded for Windows, Mac and Linux from CRAN. The SGP package contains lower level functions that can be conducted individually, but operational SGP analyses are often conducted using higher level wrapper functions such as abcSGP and updateSGP.

These higher level functions simplify the 6 steps to conducting a SGP analysis to only 5 lines of code. These functions also support the inclusion of SGPstateData meta-data if it is available and provides a faster and easier way to manage SGP data.

For example, to generate a student’s SGP percentile with the abcSGP function one must have the following information available in their data set:

The first variable, ID, is the unique identifier for each student. The next variables, VALID_CASE, CONTENT_AREA, YEAR, SCALE_SCORE, GRADE and ACHIEVEMENT_LEVEL are used as inputs to the quantile regression model that places the student’s assessment score on a normative scale.

The final result, the student’s percentile on the Normative Scale, is then displayed in the results table in the SGP Dashboard. For more information about the SGPs shown in the dashboard, refer to the SGP vignette. There are many ways to interpret and display SGP percentiles and projections, see the vignette for more details.