Even as large data sets become more readily available, analyzing it remains challenging. A recent study analyzed every Facebook user worldwide over several months using traditional relational database tools and techniques – this was considered “big data” according to most standards but does not meet that standard yet.
Modern parlance often refers to data that exceeds traditional analytical tools’ capability of handling. Our SGP research, while large, does not meet this threshold and so we consider this work a medium-sized data analysis project.
Student Growth Percentile (SGP) is a relative measure of student performance that takes the form of comparing current assessment scores against all those that took similar subject-matter assessments in terms of test score distribution and performance.
SGPs are measured from 1-99; higher numbers indicate greater relative student growth. A score of 85 indicates a student has made more progress than 85% of her academic peers, providing teachers and administrators with information that allows them to assess whether students are progressing more or less than expected.
This information is available to students in grades 4-11 who have at least five years of test score histories, including Badger year test score histories. SGPs for each student will be calculated based on up to five years’ test score histories including Badger year; however, because SGPs are calculated annually they should not be seen as definitive values between years.
This dashboard presents SGP data gathered during the 2017-18 school year. Beginning fall 2019, all high school grades have access to SGP graphs; during this transition period they will also display each student’s growth from prior years alongside 2017-18 results.
To facilitate more in-depth analyses of student growth data, the sgpData database offers districts a tool for accessing and interpreting it for every student in their district. Similar to dashboard, this resource offers student growth percentiles and trends covering up to five years of test score history per student.
SGP analyses can be completed in various ways, but the general process involves two steps: prepping and processing data. The SGP package offers function wrapper files abcSGP and updateSGP that combine the lower level functions studentGrowthPercentiles and studentGrowthProjections into single function calls to make operational analyses simpler and quicker to conduct.
SGP analyses at the school, district and subgroup levels involve pooling all individual student growth percentiles to find the mean one for their group and reporting this as their SGP score. Unfortunately, DPI does not currently use SGPs as part of its federal accountability system – instead relying on value-added scores instead.