The data sgp project is assembling unprecedented amounts of geochemical data for Earth history research. The scale of the data set is a significant step up from previous work but, in comparison to analyses of global Facebook interactions or planetary meteorological data sets, is relatively modest for ‘big’ data analytics.
As a result, the bulk of our time and effort is spent on data preparation and not on running calculations. We are working to make these preparations as easy and straightforward as possible. Once the data is prepared correctly, we believe that most SGP analyses can be run with very little technical support.
In SGP analyses, the goal is to determine what students will need in order to reach or maintain proficiency (become “growth standards”). The most useful information comes from comparing a student’s current performance with an estimate of their prior achievement level. To do this, the SGP algorithm compares a student’s observed scale score with a distribution of several years worth of compiled test results. The algorithm determines the ‘knots’ and ‘boundaries’ of this distribution (i.e., where the 20th and 40th quantiles are located) and then compares the students’ actual scale score with these boundaries.
We also use the sgptData_LONG data set for this purpose which contains 8 windows of assessment data (3 windows annually) in LONG format for three content areas. This data set contains the same demographic/student categorization variables as sgptData but also includes additional information necessary to perform SGP analysis.
A common misconception is that SGPs can be used to evaluate educator effectiveness. However, there are several problems with this interpretation. First, the differences in expected SGPs between teachers are likely due to differences in the backgrounds of the students each teacher serves rather than differences in the educators themselves. Second, SGPs are typically aggregated to the teacher level so that the differences between teachers reflect the differences in their classroom contexts. This can lead to a false positive perception of teacher effectiveness, as the average true SGP for an individual teacher will be correlated with the background characteristics of the students they serve.
Lastly, SGPs can be misleading if they are used to evaluate educator effectiveness because they measure a student’s performance on one specific exam in the future. While we believe that SGPs can be valuable indicators of educator effectiveness, they are best used when combined with other data on teacher practice such as observations or classroom evaluations.
For all of these reasons, we believe that it is important to continue our efforts to improve the usability of our SGP data and to provide assistance to users of this data. To this end we have created this wiki as a resource that is constantly being updated to incorporate new questions and answers, links to useful resources, and SGP documentation. We hope that the wiki will be a valuable resource for those interested in using data sgp. Please feel free to contact us with any additional questions you may have.