Data SGP is a collection of classes, functions and data used to compute student growth percentiles (SGPs) and percentile growth projections/trajectories from large scale longitudinal education assessment data. Quantile regression is utilized as a statistical technique to analyze each student’s achievement history before using these coefficient matrices as projection bands that indicate amounts of growth needed in order for students to become proficient on forthcoming state tests.
SGPs measure how a student has performed on an MCAS test relative to their academic peers based on past performance. Simply put, SGPs reflect the difference between a student’s score and that of all his or her classmates who share similar MCAS score histories statewide; such groups may include gender, income or educational programs such as sheltered English immersion or special education.
The SGP calculation incorporates an additional prior score which serves to stabilize and minimize measurement error on pretest scores, providing valuable baseline information about an ideal academic peer group for the student. Furthermore, an additional prior score provides context for why two students may experience comparable growth despite having vastly differing posttest and pretest scores.
In order to use the SGP calculator, districts must first access a copy of OSPI’s student data file available in BAA Secure Site. These files contain lists of students with their assessment scores from each of the last five years of MCAS testing; please refer to sgpData for more details on working with these WIDE formatted datasets.
Although sgpData is the recommended format for SGP calculations, analyses may also be performed with other data formats. Most importantly, however, is having access to software and hardware capable of running R – the language utilized for all SGP analyses. SGP is available free on all operating systems and can be downloaded from CRAN (Comprehensive R Archive Network). Using SGP requires basic knowledge of R as well as familiarity with its syntax. More details on SGP analysis and an introduction to R are provided by our technical resources on our website in the form of the sgpData vignette and technical resources. In order to conduct calculations using SGP models, however, a computer with access to the internet must also be present. There are various means of accessing the internet, but when analyzing large datasets it is highly advised that a high speed connection be utilized. This is especially important when working with school district calculations that involve downloading and prepping data prior to running actual calculations – this can take significant amounts of time depending on their size. A faster machine will make this process go more quickly while decreasing overall analysis time.