How Data SGP is Calculated
For most educators, data sgp is an important tool in measuring student growth. These percentiles provide valuable information about the performance of individual students and groups of students, and they are useful for analyzing the effectiveness of instruction. They are also a critical part of state accountability systems. However, the calculations behind these percentiles can be complicated, and it is important to understand how they are derived before using them for assessment and decision making.
To calculate an SGP, a teacher’s student data is compared to the performance of academic peers with similar previous test scores (academic peers are identified by their MCAS results in grades 8, 6, and 7). The results are then placed on a normative scale using a statistical technique called quantile regression. This allows us to compare the performance of students statewide regardless of their initial test score and ensures that we can identify the highest and lowest performers for each subject.
SGPs are also averaged to analyze the typical growth for schools, districts, and subgroups within schools or districts. However, these averages involve a much smaller sample of available data than the individual data used to compute the SGPs, and therefore may be more volatile.
The data used to generate SGPs is a combination of several different files. The first file, sgpData, contains a unique ID for each student that is associated with the MCAS assessment results from the past five years. The next five columns, GRADE_2013, GRADE_2014, GRADE_2015, GRADE_2016, and GRADE_2017, provide the scale scores for each year of testing.
A student’s SGP is a number between 1 and 99 that describes her growth compared to the 85th percentile of her academic peers. The SGP is not a measure of a student’s achievement, as a low score on the state assessment can show high growth, and a high score can demonstrate lower growth. The SGP model is described in detail in the technical resources on the Student Growth School and District Resources webpage.
The process for conducting SGP analyses is complex and requires a lot of source code. Fortunately, the SGP Package has wrapper functions, abcSGP and updateSGP, that simplify the source code and speed up the execution of these analyses. Using wide-format data like sgpData with these wrapper functions is straightforward, but it is always best to consult the SGP Data Analysis Vignette for more comprehensive documentation on using sgpData and WIDE data formats in general for SGP analyses. In particular, the vignette provides detailed instructions for generating SGPs using the exemplar LONG data set, sgpData_LONG and INSTRUCTOR-STUDENT lookup files, sgpData_INSTRUCTOR_NUMBER. The vignette also discusses the different approaches to SGP analyses and provides examples of how they work. SGPs are often compared between two different years, but this is a dangerous practice. Generally, only differences of 10 points or more should be considered significant. The most reliable comparisons are between the same year of SGPs for a given student. This is a more accurate reflection of the actual increase in achievement.