The Student Growth Percentile (SGP) is a percentile rank of a student’s current academic achievement among students who share similar prior academic achievement. The SGP is a preferred measure of student growth in two respects: First, it provides a clear and consistent scale that can be used to evaluate individual student progress. Second, it is a more relevant and interpretable indicator of educator effectiveness than unadjusted achievement levels.
The SGP varies from teacher to teacher, depending on the background characteristics of the students a teacher teaches. The variation may arise from different ways that true SGPs are related to the student covariates in a teacher’s classroom, or from the relationship of true SGPs to latent achievement attributes. These relationships can range from simple, linear effects to more complex, nonlinear interactions.
For example, in Table 3 below, we find that for math and ELA, true SGPs are on average 5.1 percentile points lower for students who participate in the FRL program, compared to students who do not. In contrast, Asian and Other Home Language groups have true SGPs that are on average 9 to 10 percentile points higher than other students.
These differences are attributed to the unequal distribution of student covariates across classrooms. The covariates tend to have a greater contribution to the variance of student achievement for students with higher grades than for those with lower grades, and they are more important for students who receive more individualized instruction.
If the goal is to use aggregated SGPs to examine teacher effectiveness, then this source of variance should be removed. This is easy to do in a value-added model that regresses student test scores on teacher fixed effects, prior test scores, and student background variables.
Another approach is to define terminal groups for each cohort, allowing for a greater degree of variance across covariates. For example, the X-SGP records in sgpData below specify TERMA, TERMB, and TERMC as individual terminal groups and GROUP1 as the set of all three terminal groups.
Using data from an administrative database, such as the NCLB DATA FILES in NCES, to calculate a student’s SGP is straightforward. It involves defining an X-SGP record ID for each terminal group. In the SOURCE field of a user’s logonid record, you provide the record ID of an X-SGP record and the name of the terminal group or set of terminal groups it defines.
Once you have defined the X-SGP records, you can use the SGP function in the data package to calculate student SGPs. This function assumes that you have a unique identifier for each student in your data, and that the ID is associated with one or more of the 5 years of standardized test scores. In the case of math and ELA, you can also indicate the grade level of the assessment score in each year.
Depending on the analyses that you plan to run, you should decide whether to use WIDE or LONG data formats. The WIDE data format is generally used for lower-level functions such as studentGrowthPercentiles and studentGrowthProjections, while the LONG data format is better suited for higher-level wrapper functions. The decision to use a specific data format should be made based on your needs, the size of the analysis, and the frequency of repeating analyses.