Data sgp provides an efficient means of organizing longitudinal (time dependent) student assessment data into statistical growth plots. This package offers two formats for storing time dependent data: WIDE and LONG; the former keeps all rows of cases together while LONG spreads them over multiple rows per student over time. Furthermore, sample datasets in both formats are provided via tables sgpData_LONG that help users better understand these various analyses are performed.
Key to the SGP method is its focus on measuring student performance relative to other students rather than simply comparing raw scores with an established percentage of state students. Calculations for Student Growth Profile (SGP) work best when they incorporate multiple assessments over an academic year or multiple years, making it essential for educators to have access to SGP calculation results in terms of percentile ranks that are familiar to teachers and parents alike. The sgpData table presents several percentile rank results for individual students from five grades from 2013-2017 school year. The ID column contains unique student identifiers; subsequent columns (GRADE_2013 – GRADE_2017) provide assessment scores from each year of assessment; finally SS_2013 – SS_2017 lists scale scores associated with each assessment. An instructor number table named sgpData_INSTRUCTOR_NUMBER contains details regarding instructors assigned to test records.
Interpreting aggregated SGP scores and visualizing them using the graphing functions found within the data sgp package is made straightforward thanks to graphing functions in data sgp package. For instance, one may observe that math SGP for one student correlates positively to prior math test score while being negatively related to prior ELA test score; these relationships suggest teacher effects in SGP models may be affected by factors other than student background characteristics.
Value-added models provide an easy solution to this source of variance by regressing current test scores on teacher fixed effects, previous test scores and student background variables. Unfortunately, this reduction of variance allows for potentially large correlations between student characteristics and latent achievement traits and student achievements.
Although SGP estimates are imperfect, they remain valuable tools for providing information about how much a student must improve in order to reach an achievement target. For instance, sixth graders in Simon’s classroom might need to improve their scale score by 70 points in order to meet the state-mandated sixth grade ELA goal of 300; with SGP growth percentile analysis included with this package it can be seen that in order to make such progress they would need to reach 75th percentile or above; such information can assist both teachers and administrators when making decisions about how best support their students’ learning needs.