Students’ standardized test score histories provide invaluable information about their academic development; however, deciphering this data can be complex. One way of making analysis easier is creating statistical growth plots (SGPs). These graphs compare each student’s current assessment score against its prior assessments score, showing how their progress compares against similar test history data sets of other students and helping teachers assess whether students are meeting predetermined growth standards.
To create these graphs, students’ test history data are compared with that of students from their school who took identical tests at the same time. The resulting SGP shows a percentage that indicates how well each student did relative to his/her peers – with higher numbers representing improved performance vs lower numbers denoting worse performance.
The data sgp package provides classes and functions that calculate student growth percentiles and projections/trajectories using large scale longitudinal educational assessment data. Quantile regression estimates conditional density associated with an achievement history while coefficient matrices provide information on required percentile growth goals for reaching them.
The sgpData exemplar data set contains seven columns for each student in its database: ID is an unique student identifier; then five more columns; SS_2013, SS_2014, SS_2015, SS_2016 and SS_2017 offer numerical assessments scores from 2013-17 for every assessment attempt administered during that year and one more called SS_TEST_DATA gives details about how many MCAS tests the student took during that same time frame.
Data to be processed using SGP analyses must be formatted either WIDE or LONG format for analysis by lower level functions such as studentGrowthPercentiles and studentGrowthProjections within the SGP package; while higher level wrapper functions that generate aggregated student level reports and charts require LONG formatted data.
For operational SGP analyses, it is highly recommended that data be formatted in LONG format as this format takes advantage of many features built into the package. Please refer to the sgpData vignette for assistance in setting up this data format.
Assuming your data has been formatted correctly, performing SGP analyses should be straightforward. Most errors encountered during SGP analyses stem from improper preparation; so the first step should always be checking that everything has been set up properly.