Analytics as a Planning Tool

• By William Lazarus
• 01/01/11

As school districts face tougher and tougher budget decisions, many are struggling to determine how best to allocate resources. Uneven population growth and changes in residential settlement patterns have created significant geographical imbalances in many areas, leaving a number of school districts with unbalanced school utilization. These demographic shifts are creating a need to close or repurpose schools in some areas, while other schools in the same district may be overcrowded and rely on costly portable classrooms. At the same time, transportation costs continue to escalate, burning scarce resources that are better devoted to teaching students.

Recognizing these inefficiencies, many districts have begun to consider changing school boundaries to increase efficiencies and provide a richer educational experience for all students. However, determining new boundaries and consolidating existing facilities are challenging tasks with significant emotional and political stakes. School boards and administrators must be able to support these tough decisions with clear thinking and solid evidence.

New Planning Tools
School districts have achieved success with these issues through the use of management science tools to drive efficiency and better business decisions. The application of some of these tools in student assignment planning procedures was pioneered by Hillsborough County (Fla.) Public Schools, the nation’s eighth largest school district, and Tampa-based research firm SeerAnalytics. An extension of the methodology was subsequently used in Portland Public Schools in Portland, Ore.

Applying evidence-based analytics, these tools offer solutions for districts of all sizes to maximize resource utilization. At the same time, the use of rigorous, fact-based approaches provide district administration and school boards with a credible way to communicate and justify difficult and sometimes contentious decisions.

For districts that are faced with resource allocation challenges, exploratory tools can provide an initial, low-cost way to understand “big picture” issues and trade-offs. As part of a long-range planning process, Hillsborough County Public Schools used a pro forma “as-built vs. minimum transportation” analysis to visualize over and under utilization of buildings, ethnic and socio-economic diversity and “best case” transportation scenarios. The process involved creating a set of pro forma spatially compact “neighborhoods” defined by the actual location of existing high schools and then analyzing the implications of hypothetical “optimal” boundaries.

While it would be impractical to implement these mathematically defined boundaries in the real world, Hillsborough County Public Schools was able to use the analysis as a way to understand the real opportunities and constraints they faced in optimizing efficiency. The exercise was so illuminating that the school board decided to conduct a similar study as part of a long-range plan to create new boundaries for the district’s middle schools.

With results outlined in a straightforward maps and graphics report, this type of sophisticated analysis is easy to understand. Specific schools can also be removed from the analysis to determine the impact a school closing would have on the school utilization, travel costs and demographics of the remaining schools.

Long-Term Planning Methods
Conducting a pro forma analysis is just the first step in long-range student assignment planning. Districts that are ready to engage in a process to change school boundaries and reallocate resources should begin by determining which criteria and decision rules should be used to evaluate boundary options.

When Hillsborough County Public School set out to establish new boundaries for its high schools in 2008-2009, district staff, school board and community members agreed that school utilization balance, transportation costs and socioeconomic diversity would be the criteria used to evaluate boundary options. By setting decision rules based on community values and not drawing maps until later in the process, the project team was able to illustrate the fairness of the method, demonstrating to the sometimes skeptical community members that potential boundaries would be generated without considering specific communities and households.

Because multiple decision rules were set in the Hillsborough County Public Schools project, the team chose to conduct the analysis using a multi-objective geo-spatial optimization model. The analysis generated thousands of possible boundary solutions, which gave different weights to each of the decision criteria.

The project team identified 79 of the best boundary options and presented them to district staff and board members to review with full documentation on school-specific utilization, transportation and diversity trade-offs. After thorough review, staff and board members chose four potential solutions that were considered most effective in balancing these factors.

Maps were introduced only after these four scenarios were chosen. These maps were presented and discussed at a set of community meetings to address concerns and gain support for the selected boundaries. At the final meeting to approve the new boundaries, not one parent or community member spoke against the boundaries, an unprecedented result in Hillsborough County history.

Using this analytical method, the district was able to balance school utilization with the maximum difference between “full” and “empty” schools at only six percentage points. When the boundaries were applied, the district was able to reduce annual transportation costs by hundreds of thousands of dollars.

Grouping Schools and Justifying Closings
Hillsborough County Public Schools is not the only district that has found success with these analytical tools. A similar analytical process was used in support of student assignment planning for Portland Public Schools in Portland, Ore. For Portland Public Schools, maintaining feeder patterns, which allow all of the students from a given elementary school to move together to the same high school, was a critical decision rule.

Working with Portland staff, analyzers determined that a different analytical technique was required for this project. Instead of a multi-objective trade-off model, a simulation approach employing cloud computing was used to create K-8-to-high school grouping options.

In addition to maintaining feeder patters, Portland Public Schools set additional decision criteria to:

• keep enrollment in each school within a set minimum and maximum band;
• minimize the number of students reassigned to new high schools;
• optimize student travel time using available public transportation; and
• maximize school socio-economic diversity.

After providing the potential high school feeder pattern groupings, the project rank ordered all high school configurations in terms of the four other decision criteria. The analysis offered all possible scenarios for consolidating the number of high schools by eliminating or re-purposing one or two existing schools. This revealed that configurations that included two specific schools consistently under-performed all of the other configurations. Using this information, Portland Public Schools approved a plan to close one school and repurpose another.

While school closings are always difficult, Portland Public Schools’ staff felt confident that they had made the correct decision because of the evidence provided by the analysis. They used the findings to communicate the reasons for the closings to the community.

Although these examples come from urban school districts, these analytical methods are also applicable to smaller districts seeking cost effective planning tools to improve efficiency. Many districts can benefit from using management science tools in their student assignment planning process. By employing evidence-based analytics to generate optimal school boundary solutions and resource allocation recommendations, school districts of all sizes can now determine how best to employ resources and make better business — and educational — decisions.

Dr. William Lazarus, Ph.D., is president and CEO of SeerAnalytics, a consumer research company that specializes in market analysis and predictive modeling. Contact Lazarus at blazarus@seeranalytics.com.