The data sgp package offers an efficient means of organizing longitudinal (time dependent) student assessment data into statistical growth plots. It offers two common formats for time-dependent data: WIDE and LONG. In the WIDE format each case/row represents a single student and the columns represent variables associated with that student at different times; the LONG format spreads the time dependent variables across multiple rows per student. The package includes sample data sets in both formats to help users set up their own data sets.
The lower level functions in the SGP package that do the actual calculations, studentGrowthPercentiles and studentGrowthProjections, require the WIDE formatted data whereas the higher level wrapper functions, that provide user-friendly interfaces to these functions, utilize the LONG data format. For all but the simplest, one-off analyses, we strongly recommend formatting your data in the LONG format since there are numerous preparation and storage benefits of doing so over using the WIDE format.
A growing number of education agencies are using growth percentile analyses to inform their teacher evaluation systems and to communicate student academic progress in ways that are more meaningful to teachers and parents. Unlike traditional test score measures, such as standardized test scores and grade point averages, growth percentiles compare a student’s current year’s test scores with those of students who have taken tests in prior years and have similar achievement histories, referred to by the SGP methodology as a student’s “academic peers”.
Our recent work has demonstrated that the estimation errors inherent in standardized testing can lead to large differences between true and estimated teacher effects (see Akram & Erickson, 2013; Lockwood & Castellano, 2015; Monroe & Cai, 2015). This creates a significant problem if aggregated estimates of teacher effectiveness are to be used as performance indicators for educators because it implies that educators with equally effective students will tend to produce slightly different SGPs when compared at the school or district level. This problem is easy to avoid in a value-added model that regresses teacher fixed effects on prior test scores, student background variables and other individual-level factors.
In this article we build upon the work of Dr. Damian Betebenner to extend the SGP methodology to allow for multi-year growth projections based upon official state achievement targets/goals. This allows districts to stipulate what future growth is required for a student to meet a target and then track the progress needed to get there. The SGP methodology also provides a framework for reporting student growth in terms of the percentage of students who meet or exceed a given target in a given subject and grade level. These results, illustrated in the figure below, can be reported to teachers and parents in a way that is easy to understand and meaningful.