DATA SGP is a set of aggregated measures that provide educators and parents with insight into student achievement and progress. It includes individual-level measures like test scores and growth percentiles as well as aggregated measures at a school/district level that can be used to compare student performance across schools or districts. The Department of Education (DPI) provides a variety of reports and dashboards to help districts, schools and teachers utilize these measures to support instructional decisions and school improvement efforts.

There are many different approaches to SGP analyses and many factors that drive which ones make sense for any given situation. For example, some may find it easier to conduct the lower level SGP functions, such as studentGrowthPercentiles and studentGrowthProjections directly whereas others will prefer to use higher level wrapper functions (abcSGP and updateSGP) that “wrap” the lower level functions and perform all of the complex calculations required for operational SGP analyses. The SGPdata package installed with the SGP application contains exemplar WIDE and LONG formatted data sets, sgpData and sgpData_LONG, to assist in setting up these types of operational analyses.

The sgpData_INSTRUCTOR_NUMBER data set is an anonymized lookup table providing instructor details associated with each student’s test record. This table is particularly useful for teachers who wish to assign multiple instructors to a single student across several content areas for one academic year.

Student growth percentiles are calculated by comparing the current assessment score of a student to the student’s prior test score on any other testing window in a previous year. A student’s prior test score does not have to be a state assessment or even the same testing window as the current test; however, both assessments must have been administered within the past 18 months.

A student’s true SGP is a number between 1 and 99 that indicates how far the student’s current assessment score moved him/her relative to 85 percent of students with similar academic backgrounds. For example, a student with a high SGP will have demonstrated a lot of learning since his/her last test while a student with a low SGP will have shown little or no learning.

SGPs are a good measure of student achievement because they tend to correlate more with other student performance measures, such as grade level and subgroup test scores, than with demographic factors. They are also more stable than medians, which can be influenced by a small number of students who either cannot keep up with their classmates or who start the school year behind their peers in terms of knowledge, skills and ability. This stability and consistency in SGPs is why the Department uses them as part of its federal accountability system and not mean test scores. This article specifies a model for latent achievement attributes, defines true SGPs using this model and demonstrates that they are related to covariates in student background characteristics.