SGP (Student Growth Percentiles)

data sgp

SGP (Student Growth Percentiles) measures student learning over time by ranking students against their peers who have taken the same prior year of tests. Using prior test scores provides more reliable and valid measures of student learning than simply examining raw achievement levels. Additionally, comparing students with peers who have the same combination of prior test scores allows for the removal of bias due to unobserved differences in the backgrounds of the students. These benefits have contributed to the increasing use of SGPs in the United States and beyond.

sgp tabel hari ini live tercepat

In practice, SGP models are typically built on the basis of more than one year of prior test scores. Current SGP is calculated for each student by calculating the difference between the student’s test score on the most recent assessment and their test score on an earlier assessment that is a part of the same testing window. The most recent test is considered the “recent” assessment even if it occurs outside of the school year.

The data set sgpData contains the test records for each student. Each record has a unique ID for each student and 5 columns that provide information about the student’s performance in each of the past five years of assessment. The columns SS_2013, SS_2014, SS_2015, SS_2016, and SS_2017 provide the scale scores for each of these assessments. The data set also contains a teacher-student lookup table that links each student to their instructor. The teacher-student lookup tables sgpData_INSTRUCTOR_NUMBER and sgpData_INSTRUCTOR are used in the higher level SGP functions studentGrowthPercentiles and studentGrowthProjections.

SGP is most commonly interpreted as an indicator of teacher effectiveness, but there are several important limitations to this interpretation. First, it is possible that aggregated SGP may reflect a source of bias that would be eliminated in a value-added model that regresses student test scores on teacher fixed effects, prior test scores, and latent variables.

Another potential source of bias is that teachers may be sorted into schools and classrooms that vary systematically with respect to the covariates we measure. For example, if more effective teachers tend to teach in schools with better facilities for math instruction, then their average true SGPs are likely to be higher than the average true SGPs of less effective teachers.

Lastly, it is possible that the relationships between true SGPs and student covariates are influenced by unmeasured factors not captured by our explanatory variables. This is a possibility that could be investigated using additional explanatory variables.

In these cases, it may be best to use an additional model to estimate the relationship between student covariates and true SGPs. Fortunately, the SGP package includes several high-level functions that simplify this process, particularly for operational analyses. These functions are described in the SGP package vignette and documentation.