PERVADE

PERVADE (Pervasuve Data Ethics for Computational Research) consists of a team of researchers with hopes to build interactive decision support tools for data scientists to help guide ethical decision making.

The project has produced a large amount of data on user expectations for big data reuse in various contexts, data science norms and controversies, and regulation, as well as guiding principles for researchers. They would now like to build tool that makes use of that data to help guide researcher and industry decision-making.

Working tool coming soon!

Focus

PERVADE

Team

Jana Hijji

Sarah Snider

Niloo Alavi

Eden Metzger

When

January 2022- May 2022

Role

  • Strategist (requirements gathering, refining project goals)

  • Project Manager (sprint planning, presenting to clients)

  • User Experience Researcher (expert interviewing, user testing, data analysis, competitive analysis)

  • User Experience Analysis (user flow, affinity mapping)

  • Visual Designer (sketching, iterating, prototyping, mockups, etc.)

Tools

  • Figma

  • Miro


High-Level Goals & Users

Goals

  • Increase user awareness of data ethics

  • Encourage users to incorporate, engage with, and explore ethical considerations in the course of their work

  • Facilitate continued engagement with data ethics by creating a platform that elevates the learning experience from a baseline "must fulfill" compliance activity to a rewarding and fun activity

Key User Base

  • Students just getting started with the field of data research and ethics,

  • Educators teaching those students about the field of data research and the role of ethics in that field, and

  • Researchers active in the field of data ethics (both industry and academic) who are looking to enhance their knowledge and mastery of data ethics.

Stakeholders

  • PERVADE team

    • a group of researchers that come from diverse backgrounds and academic fields with a common goal to address privacy and ethics for big and pervasive data

  • Users

    • Data Research Students (undergraduate and beyond)

    • Educators

    • Data Researchers

Who are we designing for?

User Groups

Students

Beginner -Intermediate

Instructors

Expert

Industry Professionals

Intermediate - Expert

Academic Researchers

Intermediate -Expert

Although we started off focusing on academic researchers, we quickly realized that we had more user groups than we imagined. Our client wanted to make ethical decision making part of the data research process industry wide, and that meant including students and the instructors who teach them as well as industry professionals and academicians. To hone in on the specific needs of users, we created three personas to clarify pain points and design considerations.

Strategy and process

Sprint Breakdown

For this project, we utilized the design sprint framework developed at Google. Over two semesters, we did a total of five sprints which ranged from four to six weeks in length. Each of these sprints was focused on a different problem related to the overall project scope, which allowed us to break the complexities of the tool down into manageable chunks.

This was beneficial not only to us as designers, but also to our client, as it meant each sprint they were able to give us feedback on a specific facet of the tool. In addition, it allowed us to tailor our expert interviews and user testing sessions to specific user groups, giving each one or more chances to demonstrate their particular pain points.

This resulted in a more user-centered final product, with specific customizations to reflect the needs of each user group.

Map: Evaluate the Problem

Set goals for the sprint, conduct expert interviews, and map the big ideas that the team wants to focus on this sprint



Sketch: Getting Ideas on Paper

Team members divide and conquer the ideas that they want to see In the interface, sketch their designs


Decide: Distilling the Ideas

The team votes on their favorite concepts from the sketching phase to include In the prototype



Prototype: Giving Form to the Design

High fidelity prototyping and iteration occurs



Test: Finding the Pain Points

User testing is conducted and data is ingested and analyzed so that new goals can be determined for next sprint



What We Did

  • Met with 10+ experts in the field to gain insight into critical areas of development

  • User testing: 15+ group and individual interviews as well as surveys

  • Performed competitive analysis of other products in the space

  • Researched critical components of the app, including gamification and e-learning



Key Findings

  • In most classrooms, data ethics is a secondary subject - usually presented in a conversational way rather than as its own module

  • Many users view ethics in data science as a “checklist” activity - a minimal baseline solely for compliance to IBRs

  • Little centralized information is available, and there is no universally recognized authority for data ethics

  • Data researchers fall into a wide variety of user bases and have varying needs and expertise when it comes to data ethics - creating an “all-in-one” tool that serves all users equally is a constant tension

Weighing Our Options

We wanted to design the tool to the best of our ability, but with time and resource constraints we had to consider which features were best fit user needs

High Value/Low Complexity

Toggles and Tooltips

Ethical Data Lifecycle

Video Modules

Low Value/Low Complexity

Customizability

Badge System + More Gamification

High Value/High Complexity

Toggles and Tooltips

Ethical Data Lifecycle

"Evaluate My Scenario" Journey

Low Value/High Complexity

Interactive Decision Tree

Design Thinking Process

Balance

Balance users’ desire for lots of information with need to have a clean, intuitive interface

DESIGN CUES: negative space in design, clean lines, modern type

Familiarity

Align tool’s framework with familiar concepts in data science research

DESIGN CUES: use of ethical data lifecycle as framework, video modules similar to existing e-learning modules

Presentation

Present information in a format that encourages user engagement while minimizing cognitive overload

DESIGN CUES: badge system, customizable dashboard

Flexibility

Create a tool that works equally well for all user groups

DESIGN CUES: toggles for features within the app, ability to choose level and experience

Mission

Align design to existing PERVADE branding and mission statement

DESIGN CUES: bold color palette and fonts shared with existing branding

Accessibility

Consider accessibility throughout to allow the best user experience for everyone

DESIGN CUES: fonts checked for clarity, color ramp tested for color blindness and contrast, tooltips and alt text

Drawing Users with Familiarity

To ease users into our tool, we wanted to use an initial framework that was clear and understandable

for our users.


We spoke with experts and finally decided to incorporate the ethical

data lifecycle since it was conceptually familiar to all of our user bases; although finding one universally accepted version was a bit more challenging than we thought!


After iterating through several versions and incorporating user feedback

until we arrived at our final version seen here:


Initial Features

Initially, our tool was geared exclusively towards researchers and, after our initial expert talks and rounds of sketching and

voting, we decided to implement a few key features such as:

  • a landing page based on the ethical data lifecycle

  • a flowchart to outline the various parts of each section of the lifecycle

  • video modules and a searchable video catalogue that aligned with the parts of the lifecycle

  • links to key journals and research to fill in the gaps


Broad Horizons

After rounds of design, iteration, and

user testing, we decided to expand our target user groups to students and professors as well.


This meant that we needed to also broaden the tool's feature set in order to accommodate growing user needs.


The PERVADE tool needed to give these users means to:


Aesthetics MAtter

Branding Considerations

We had our prototype - but we needed to make sure it aligned with who PERVADE was as a brand. We asked ourselves some questions:

  1. What key brand descriptors did we need to keep in mind?

  2. What existing design elements were already in place?

  3. What was the group’s brand “voice”?

Key Brand Descriptors

  • energetic

  • modern

  • hopeful

Existing Design Language

  • logo

  • font: Roboto

Brand / Product Voice

  • trustworthy

  • bright

  • educated

Logo


Pervade's existing logo was a solid start - it aligned with their overall brand image, and provided a jumping off point for a color palette. In fact, the colors were already well-differentiated, and passed accessibility testing with only a few minor tweaks to increase contrast.


Color Palette and Accessibility

Testing was done on our color palette for the eight types of color-blindness to endure the palette used was fully accessible.

Typography

Pervade uses Roboto as their main font, and although we chose not to use the same exact font for the tool, we kept it in the neighborhood with a similar sans serif, modern font. We chose Poppins. with its rounded curves, it lent a playful edge to the tool, but it still echoed the brand values of modernity, hopefulness, and energy.

96px

35 px

20 px

Headline

Subtitle

This Is the paragraph text.


Final Prototype

After months of iterations and multiple sprints, we came up with our final design for the tool - linked below:


Link to the Final Design!


Lessons Learned

Challenges

  • Creating a tool that is fun and engaging, but still carries weight in a linear, data-driven field is difficult.

  • Convincing users that ethics is worth incorporating into their research as more than a “checklist item” is (also) difficult!

  • Ensuring that the tool’s effectiveness wasn’t being compromised in an effort to cater to all user groups was a constant tension.

Surprises

  • Gamification and hard science can coexist!

  • Data researchers care a surprising amount about design (color, dark mode, fonts, etc.)

  • There is a surprising absence of competing products in the space - users were just glad PERVADE is developing the tool at all