Data sgp is a data file that contains the student-level assessment scores and growth data for each student in each year. These data are used to estimate a student’s growth rate and generate predictions about their future achievement levels. It is important to note that the SGP predictions are not absolute, but instead are estimates of the probability that a student will achieve a certain level. This is why it is critical that the prediction model be accurately calibrated to the actual student growth data available for each year.
Data for a student is generated from multiple sources and in many different formats. The best source is the official school report, which contains a variety of data including student assessments, grades, and demographic information. These reports are often distributed to teachers and parents in a spreadsheet format. Another source of data is the SGP aggregation file, which contains the student-level results from all schools in a district. This data is also provided to parents and teachers in a spreadsheet format, but it is not as detailed as the official school report.
The data sgp package provides tools for extracting, transforming, and processing the student-level assessment data to prepare it for SGP analyses. The lower level functions that do the calculations, studentGrowthPercentiles and studentGrowthProjections, require WIDE formatted data whereas higher level functions (wrappers for the lower level functions) support both WIDE and LONG data. The higher level functions also provide more sophisticated functionality than the lower level functions, making them easier to use in operational SGP analyses.
Getting started with data sgp is fairly straightforward. However, it is recommended that you consult the SGP vignette and the data analysis chapter of the user guide for more comprehensive documentation. There are several ways to conduct SGP analyses, but we generally recommend starting with an exemplar data file, sgpData_LONG, and the INSTRUCTOR-STUDENT lookup file, sgpData_INSTRUCTOR_NUMBER, which can be created using the prepareSGP function.
After preparing the input data, the next step is to transform it into an SGP prediction file. This can be done by using the sgpPredict_Wide and sgpPredict_Long functions of the SGP package. The sgpPredict_Wide function takes the exemplary data file, sgpData_LONG and creates an SGP prediction file with all of the required parameters. The sgpPredict_Long function takes the exemplary data file, a set of INSTRUCTOR-STUDENT projection files and creates an SGP prediction file with the projections and forecast values for all of the students in the sample.
After the transformation and prediction processes are complete, it is time to run the SGP analyses. The SGP package includes a number of pre-built models for assessing student growth. These models are based on a variety of statistical and mathematical approaches. Choosing the correct model for a particular problem is not an exact science, and in general it is a good idea to review the predictions from each of these models prior to finalizing the decision. Any errors that are encountered in this process usually revert back to the initial steps of data preparation and analysis.