MySchools - Decision Support
With seemingly unlimited options available among NYC schools and many factors to consider for each, families were overwhelmed.
While the city must provide complete information on all 1,866 schools, transparency itself can lead to inequality, as tech-savvy families download spreadsheets of hundreds of schools’ performance and admissions data to sort and compare, while families with limited means lack the time to sift through massive school directories. This caused some students to apply only to their closest school, missing out on options better suited to them just a subway ride away.
We sought to reduce stress while empowering users to easily compare hundreds of options.
In order for families to quickly home in on schools of interest to them, we needed to develop tools that would support each student’s unique criteria in their decision-making process. While our key differentiator committed us to tailoring eligibility and acceptance criteria to each student, families would need additional help narrowing down their options.
I developed research-backed features to streamline complex data, aid decision making, and advance users confidently through their applications.
Experiments by NYU, Princeton, Columbia, and Seton Hall University demonstrated that shorter lists of personalized options improve students’ choices — and more often lead to their acceptance at better schools.
Using the researchers’ findings, I generated multiple potential features to present personalized suggestions to users.
Perhaps the option “easiest” for families — showing an icon indicating the student’s likelihood of an offer to each program — would unavoidably suggest some programs were better options than others, which isn’t true — a good application consists of programs with a mix of likelihoods. In this iteration, popovers explain the factors at play.
Partnering with our UI design team, I tested various icons and text labels to achieve clarity without steering users toward or away from certain schools. Because green and red implied quality ratings, we also attempted a palette (far right) without such connotations. In the end, any representation of offer likelihood shown without more explanation was too risky.
While each program's likelihood factors could be averaged automatically for each student, doing so might risk steering students away from “reach” schools they like. So we considered providing an optional pop-up “worksheet” for users to do these assessments themselves. Ultimately, this didn’t provide enough of an improvement over the old paper worksheets.
We also tried departing entirely from representing offer likelihood while still helping families find interesting schools with lower demand. These wireframes depict an expandable panel highlighting under-the-radar schools worth considering. CMS tools would allow administrators to create and feature sets of schools based on their unique offerings.
Because families’ criteria for choosing suitable schools differs widely, I proposed the concept of a personalized “fit”. Before seeing the list of schools, users could select some criteria important to them. Then, each school would show a fit “score” based on how many of those preferences it meets. Users could sort the list of schools by fit, and edit their preferences at any time.
Unfortunately, given time constraints, this feature was postponed for future consideration. Filters offer an alternative means for families to find schools matching their criteria.
The project team weighed these options against the risk of leading families away from other schools they might prefer. Ultimately, it was decided to lead with the complete list of schools (sorted by distance from the student’s home) with filters for users to define their own criteria, as well as dynamic rules-based suggestions that encourage users to build applications consisting of schools with a mixture of acceptance rates.
The final set of features designed to help families narrow in on schools of interest and build well-balanced applications includes sorting by distance from home, a map view, filters for various school offerings, and alerts on one’s in-progress application that suggest ways to improve it.
Each school card uses principles of progressive disclosure to emphasize key details without overwhelming users, while providing complete information — personalized to each student — within expanding panels.
Expanding panels allow users to see an overview of the school and its programs at a glance, with the option to open each for more information.
I also introduced a dedicated step for users to view just the schools they saved. In effect, this serves as a personally curated shortlist which advances users toward making their final selections.
Situating this step within the progression of the user’s application allows them to compare a shortlist of the schools they’re most interested in before making their final selections. A notification encourages users to save at least 25 programs. Doing so positions them to move on to the next step and build a strong application with their top 12 choices.
The personalized search and saving features were so successful that the city was able to stop printing paper directories — saving taxpayers the cost of 50 million printed pages per year.