I have the pleasure of working on a very exciting project designed to deliver new data insights to K-12 school district leadership. Our project focuses on using a combination of the right people, processes, and simplified technology to deliver meaningful insight and actionable information to school districts. We are currently using a combination of sFTP, a Microsoft Azure data lake, custom API to source the data, transform it, and provide secure access via our custom API. We then attach any visualization tool to the API end-points. We are currently using Tableau Desktop during our prototyping work with a school district. In a short few months, we’ve gone from understanding the district’s pain points and goals all the way to delivering useful data visualizations and data insight. The comparative analysis between the different data sets is providing more value to the district beyond simple analyses with one data set. Cloud technology has truly enabled the team to be agile in terms of ingesting data and analyzing it quickly.
Whenever we create data visualizations, there are basically two paths we can take: either explore the data in the visualization tool to see what “jumps” out at you or begin with a specific problem or question that can guide the analysis. One of the problems that was identified while working with this school district was that a very low percentage of students in 3-8th grade were not reaching the minimum acceptable standard on a state assessment (SBA – or Smarter Balanced Assessment). The school district knows this and has as one of its goals for this year is to have at least 50% of all 3-8th graders meet the state standard on the assessment. A few of the grades have very low percentages. It is a district that has a high number of immigrants and very large Hispanic community. English language arts (ELA) has been identified as a focus area for these students who are not being successful in developing their language skill.
Our data services team has identified what this school district calls the “bubble” students – or those students who are within +/- 10% of the appropriate scale score on the SBA assessment for their grade. The purpose of the analysis is to identify those students who are very close to the cut score for the ELA standard on the SBA, which is a cohort of students that teachers and administrators can plan interventions. For example, third-graders who took the ELA portion of SBA needed to reach a minimum scale score threshold of 2432 to achieve level 3, which is “met standard”. Anything below this value indicates that the student did not meet the ELA standard. The students that are really close to meeting the standard can be targeted relatively easily so that teachers can plan interventions.
Of course, in protecting the school district and student names, we came up with the following visualization that can depict this bubble. Note that this is only for 3rd graders, but the same analysis can be done for each grade.
Along the x-axis, we have the range of possible scores within the particular “bubble” for 3rd graders who took the SBA assessment. Along the y-axis, we have individual masked IDs for each of the students. When we combined the SBA and ELPA21 data, we identified some students who were also within the appropriate bubble for ELPA writing. Thus, the real power of this analysis is when we integrate different datasets into a single data visualization! This means that on both assessments, these particular students were very close to the cut scores. What can districts do with this data, then?
So, visualizations are great – when they are easily digestible and we can take action based on them. It’s important to create goal-aligned visualizations — that is, creating visualizations that are specific to the questions and problems that were identified during the beginning phases of the project. But creating and sharing visualizations is really only there to enable decision-making in terms of creating interventions for students who need them.
A student that falls within these bubble scores is the “low-hanging fruit” with which we can intervene. Identifying and creating interventions for these students could mean growth in achievement for these students and show growth in overall student achievement across the district. We can add more information to the viz in terms of which school the student attends, and whether or not the student was in a “bubble” on ELPA21 or DIBELS as well.
This type of analysis only scratches the surface of the things we can do to analyze student-related data — all for trying to improve student achievement and success. I’m very excited about the things we can do in terms of data science and predicting students who may be successful on one test, but not on another, and so forth. We are just scratching the tip of the iceberg in terms of the types of analyses we can do with data in the education space.
Here are a few additional resources and case studies of how some school districts are using data and analytics more effectively to address student achievement challenges.