The data sgp package provides two key functions for preparing and conducting SGP analyses: studentGrowthPercentiles and studentGrowthProjections. Higher level wrapper functions such as studentGrowthTrajectories and stateSGPData are designed to utilize these two lower level functions. The sgpdata package also contains five exemplar datasets utilized by the SGP package for testing and demonstration purposes. The sgpdata package is available as a long data set (L) or wide data set (W). We recommend that you use the L format for operational analyses. The management of data in the L format is significantly simpler than with the W format. For example, when adding a new year of data to an existing long data set, the additional year of data is simply appended onto the bottom of the currently existing long data set. In addition, many of the higher level functions in the SGP package utilize the embedded state specific meta-data in the long data set.
The sgpdata package also includes a number of utility functions that are intended to assist with managing the data set and providing more descriptive information about the data. These include a plotting function, a summary function and an import function. The plotting function allows you to create graphs of data with both the current and historical values for a given variable. The summary function summarizes the current value of a variable by providing a list of the values for that variable in an easy to read table format. The import function can be used to quickly bring in data from a variety of sources.
A student’s growth percentiles and projections are based on their MCAS scaled score histories and how they relate to those of their academic peers from previous MCAS administrations. Thus, students with similar score histories will have very similar student growth percentages and trajectories. This makes it quite easy for teachers to determine if their students are performing well or if they need to focus on improving student performance.
However, it is important to remember that these trajectories do not reflect actual student performances. They are only a snapshot of what a student might expect to achieve on future assessments.
Another point is that student growth data for a given student will vary by school. For example, a student in Macomb or Clare-Gladwin may have very different growth percentiles than students in other schools because of the unique achievement levels of their student population. Finally, it is important to realize that student growth percentiles across the state do not follow a typical bell-shape curve. Instead, they are clustered around the middle with approximately equal numbers of students in each decile or grouping of 10 percentiles.
In general, sgpdata and the associated higher level functions are very simple to use and will provide you with valuable insight into student performance. Please be sure to take the time to fully understand the data before using it for educator evals. As stated by the Michigan Department of Education, it is highly recommended that districts do not apply these trajectories to high stakes educator evaluations until 2018-19 when the growth data will have had three years to stabilize.