Data SGP – Preparing and Analyzing Student Growth Percentiles

Data sgp is an important tool for assessing student learning, providing educators with insight into how well students are progressing. However, it can be difficult to interpret and understand the results of SGP analyses. This article provides an overview of what SGPs are, why they are useful, and how to prepare and manage data for analyzing student growth percentiles (SGPs).

SGPs describe how much a student’s performance improved over time relative to other students with similar achievement histories. They are useful for evaluating student progress, identifying students who need additional support, and informing educational decisions. This article describes how to prepare data for analyzing SGPs and introduces a set of user-friendly functions for performing such analysis.

A common method for estimating SGPs is to use current and prior test scores from each student. These scores are noisy measures of the underlying latent achievement trait, however, because they are subject to measurement error and other sources of bias. In this article, we demonstrate how to use SGP-specific statistical methods to reduce the noise and bias in estimated SGPs. We use SGPs based on the math and ELA assessments from the AIR Data Explorer to illustrate this approach.

There are a number of important considerations when preparing data for SGP analysis, including data format and time dependent variables. The SGPdata package installed with the SGP software includes exemplar WIDE and LONG formatted data sets (sgptData_WIDE and sgptData_LONG, respectively) to help you get started.

For most operational analyses, we recommend that you use the LONG formatted data set. Managing data in long format is easier than managing it in wide format and most of the higher level SGP functions (e.g., studentGrowthPercentiles and studentGrowthProjections) are designed to work with long formatted data. Additionally, if you plan to update your analyses with additional years of data, it is usually more convenient to append the new year to an existing long data set rather than creating a separate wide data set each time.

In addition to SGPs based on standardized test scores, there are a number of other ways to analyze data with the SGP package. The functions available in the SGPdata package include support for calculating student growth percentiles using other types of assessment and other data sources, such as PITTs, teacher observations, and enrollment. The SGP package also includes an extension to create student growth percentiles from a matrix of state-level achievement data. This feature allows users to create SGPs from a state-level data matrix without requiring access to the underlying state data. See the SGP package documentation for more information on using this extension. SGPs based on standardized test data are also subject to the same estimation errors that affect SGPs based on other types of assessment data. These errors will cause the calculated SGPs to be inaccurate and should be interpreted with caution. Nevertheless, they provide a useful way to examine how student achievement changes over time and across schools in a state.