In the last 15 years, school districts have spent considerable time and resources collecting thousands of bytes and pieces of data about their students, often conducting retrospective analysis on student achievement. But with the advent of big data and the increasing power of algorithms, we should be using data predictively—to get a better idea about where our kids should be heading and what we can do to help them get there.
Data Analysis Versus Predictive Analytics
To understand the power of predictive analytics, you need to appreciate the root difference between predictive analytics and simple data analysis. Simple data analysis in school districts looks at raw data points and tries to spot trends and indicators that might suggest a better way of doing things. To move toward predictive analytics, you simply construct essential questions behind each data point.
Essential questions constructed around data points can provide clarity and precision, and help formulate ideas around future practices. As teachers review data, they respond to the essential question to be solved, inquire, sift the data to identify current practices, and construct new ways to use the data. A well-designed essential question is one that probes deeper into the meaning of the data to set the stage for further questioning-it fosters the development of a culture of inquiry. The essential questions surrounding data collection become the end points to drive future destinations, define new practices, and improve outcomes.
Constructing Essential Questions
To construct essential questions around data points, you must first recognize the top objectives of what the district and school are trying to change. Once these objectives are constructed, then data should be correlated with well-designed essential questions.
In Kansas, we provide districts the opportunity to participate in interim “predictive” state assessments. Predictive assessments are designed to analyze certain elements of data points to ensure curriculum is properly aligned to the depth of knowledge required when designing instruction. The process requires the analysis of a predicted score for each individual student. These interim assessments are given three times per year to measure student achievement against state standards in both math and English language arts. To establish district alignment to state standards, we review the data after each interim to determine a curriculum gap analysis by formulating essential questions behind the data, such as:
- Is there evidence of congruency of the curriculum with the performance outcome (as in the first predictive assessment in math and English language arts)?
- How can we determine the overlap between curriculum and student performance outcomes as it relates to mastery?
Once the essential questions are formulated behind the data points, our teachers use their professional learning community (PLC) time to look for the data that will best answer the questions, and they look for ways to access, organize, visualize, and interpret the interim assessment data. The results of the weekly grade-level PLC sessions are beginning to take hold. Teachers are finding new ways to blend predictive analytics into an art and science of data interpretation as each team participates in a culture of inquiry. A culture of inquiry drives educators to continually learn more about what they should be measuring in order to generate the data that will provide insight into students’ greatest needs.
It is also important to establish norms when breaking down the data to meet specific objectives. These proposed data review norms are helpful in scaffolding techniques to provide assurances for nonjudgmental conclusions among team members during interpreting sessions:
- Define the project objective by choosing the right indicators.
- Identify the most critical indicators, present the talking points behind the data, and formulate essential questions.
- Analyze data to statistical models to validate the essential questions, ruling out assumptions by setting appropriate targets.
- Create predictive models, estimating the relationship between dependent variables and various independent variables.
- Refine and construct new models of practice from resulting data to accurately predict future events while customizing data dashboards for transparency.
Custom-Designed Data Dashboards
A second element to consider when introducing predictive models is fashioning tools for teachers to easily access, monitor, and act on predictive analyses. Many districts today have extensive data warehouses which store all necessary student data, but if the data cannot be accessed in custom formats to meet the needs of an inquiry culture, then data has little or no relevance. Districts need to construct data dashboards—like an automobile dashboard that provides an array of information about school performance and practices—to sort and analyze the information.
As an information management tool, a data dashboard visually tracks, analyzes, and displays key performance indicators and key data points in order to monitor the overall condition of individual student progress, a school, or a specific subject. They are customizable to meet the specific needs of a school or district.
Just recently, our school faced a data design challenge. We wanted to access student assessment information as a method to assign interventions to individual students. We had to design and construct a real-time data dashboard to monitor the progress of individual students. Our district’s director worked with our school to pull customizable fields of data together into a user-friendly format. These data fields allowed us to establish data point ranges as they relate to both subjective and objective data. Our intervention teams meet quarterly to review the dashboard as we monitor the progress of individual students based on specific intervention criteria. To customize our Tier Intervention Data Dashboard, we had to:
- Determine the types of universal assessments to populate the dashboard; these assessments included state and local standardized assessments.
- Customize the dashboard to review each student independently as a way to document and record intervention sessions.
- Establish the frequency of reviews so data assessment fields would be available when tier placement teams meet.
- Resolve how data should be sorted into score ranges to include the following levels: below basic, basic, proficient, and advanced.
- Assign a point system to each range, which would automatically be calculated to determine overall score for each assessment administered.
- Include student subjective data fields for teachers to score learning behaviors based on a scale of the students’ support needs as a part of the overall score.
- Provide teacher training on ways to use the system and include the human factor of keeping it simple and allowing for teacher exploration.
Data by Design
The foundation for predictive analytics can be broken down in many different ways. Results-driven data (which many schools currently use) becomes the basis for which teachers form their decisions. School leaders should also support customizable data dashboards to make the information seamless when monitoring the progress of individuals and groups.
To advance to the next level of predictive analytics, standards must be aligned to improve instructional efforts tailored to individual students. With established data design factors, predictive analytics can identify at-risk students and assist in monitoring students’ progress over time, thereby providing the necessary support and intervention for those who need skill enhancements.
To be successful in predictive analytics, each district will need to define its own predictive model, as leaders define the foundations for a culture of inquiry. Leadership must focus on important essential questions characterized by purposeful data. These essential questions will, in turn, drive the foundations of improved practices.
This is a culture not just built on interpretation of data, but open to a spirit of curiosity and the use of questions, rather than a rear-view mirror approach of the past. For some school leaders, becoming “champions of data” will require more than data decision making; it will require broadening the perspectives of principal and teachers to understand the importance of predictive analytics. Data analytics can become a design factor while predicting various outcomes with articulated action steps.
Michael King is principal of Dodge City Middle School in Kansas and a former NASSP Digital Principal of the Year.