How to Prepare Data for SGP Analysis

Students’ standardized test score histories contain significant estimation errors that make them inaccurate measures of latent achievement traits. To overcome this shortcoming, statistical growth plots (SGP) can be created from longitudinal student data using multiple years’ worth of test scores to generate knots and boundaries that provide more accurate evaluations than using only a single year’s test results as indicators of student performance.

SGPs allow teachers to provide educators with a way of discussing a student’s relative growth over time in percentile terms that are easily understandable. SGPs measure students based on their MCAS scores as compared to students with similar prior test score histories (their academic peers), and represent the percentage of peers that experienced greater relative growth than themselves.

The SGP Data API enables schools to import SGP information for use with their systems and applications. Once imported into an app, users can utilize the sgp api to create SGP files and display SGP results in different formats. Furthermore, it offers functions to create student lists of student SGP results for all windows for any given school and year; these lists may then be used for teacher evaluation or reporting to parents/stakeholders/other stakeholders.

SGP analyses should be easy to use when used properly; any issues encountered typically stem from insufficient data preparation before running SGP analyses. Therefore, it’s essential that as much time is dedicated to this step before performing SGP analyses.

Beginning SGP data analysis can be daunting for newcomers. To ease the process, the sgp package provides lower level functions that facilitate SGP analyses as well as higher-level wrapper functions that facilitate operational SGP analyses source code.

As is typical with lower level SGP functions such as studentGrowthPercentiles and studentGrowthProjections, their lower level counterparts, studentGrowthPercentiles and studentGrowthProjections require WIDE formatted data; higher level functions like the sgpList function utilize LONG data instead. When undertaking anything beyond simple analyses it is highly advisable that LONG format is utilized because this provides numerous preparation and storage benefits over WIDE formatted data.

Example data required by sgpList(sgpData) include ID as the unique student identifier and 5 columns named SS_2013-SS_2017 representing grade level assessment scores over five years for each student.

The sgpList() function then uses this data to produce a student list that displays each student’s SGP for each window in percentile terms that are familiar to most teachers and administrators. Once created, these student lists can then be easily sorted by SGP so educators can quickly identify students making most (or least) progress over time, enabling educators to focus their attention on those that may need extra support in accelerating learning. Lastly, SGP lists can be exported directly into Excel for easy reporting of performance over time as well as being used as starting points when creating student growth targets.