Behavior Analysis
Industrial Design & User Experience
Client: Chartlytics, Inc., 2018, Boston MA
Background
Chartlytics is a company formed from a single obsession: behavior hacking. Their team of certified Behavior Analysts works with groups of special needs children to teach basic skills, often in high-risk, high-conflict environments. Accurate data collection is a serious challenge, and with the number of special needs children rising, the team can’t afford to slow their pace.
Behavior Analyst: a board-certified practitioner who assesses patients with behavior problems, studies how changes in the environment effects behavior, and implements plans to correct behavior in a scientific manner.

The Problem (why)
Problem I: Accurate Data Collection
When a Behavior Analyst is on the job, their attention is constantly divided. Proctoring lessons and tests while simultaneously monitoring how a child is responding can be overwhelming. Did he just throw a spitball? Why is he screaming? When did he bite his brother? Did he get that question right? Having to track this information by conventional means often leads to major mistakes.
Problem II: No Timestamp for Action 
“Time is the main element in behavior tracking. Capturing exactly when a behavior happens, and how long, allows us to determine its trajectory and understand if our interventions are successful,” - Amy, BCBA
The above photo shows conventional (if unprofessional-looking) methods of data collection for Behavior Analysts and Technicians. These methods of data collection don’t capture a timestamp, or sequential information about which order actions occurred.
Problem III: No Course Correction In-Field
Behavior Analysts submit data in a lump sum of results at the end of the session. If a child is under-performing compared to previous weeks, the Analyst does not know to adjust their technique until the next week’s session. This loses valuable weeks of progress, ultimately undermining the success of the program.
The image above shows a traditional method of charting behavior progress. The data is plotted manually onto graphs called Standard Celeration Charts, and course adjustments are only made after the session is completed.
Problem IV: Understanding the Motivation
Reliable data to Behavior Analysts means capturing three core elements: 1) the action; 2) the response; 3) the motivation. By capturing a clean set of data for each action, the Analyst can determine which actions are reinforcers for the response. It sounds simple, but often, it isn’t.
Scenario One:
During quiet times, child screams in class. Attention does not lesson screaming. Highly disruptive.
Action: a quiet period of activity, such as testing, reading, napping.
Response: screaming
Motivation: Attention? Boredom? There is not a clear motivation behind the behavior.
Behavior analyst notices through experimentation, that when taken out of the classroom the screaming stops. Analyst introduces child to a “break pass” that they hold each time they take a break. After several weeks of association, the child now raises a break pass during class when they need to visit the restroom. Screaming lessens, and eventually stops.
The User (who)
User Interviews
Over the course of the project, I interviewed three Board-Certified Behavioral Analysts who led me through their routine.
User Journey
From this information, I created a compiled a User Journey map to understand how the BCBA conduct their research. By recording observations during the sessions, I was able to extract key takeaways for each painpoint identified during the BCBA's day. 
Design (how)
Design Criteria
Based on first-person research, at minimum the device needs to monitor one child at a time, be easy to switch between children’s profiles, and record data reliably in realtime. Additional features could include: a microphone for making verbal notes, a light that can be triggered for behavioral reinforcement, a buzzer, and multiple attachments for clipping and holding.
The companion application serves to connect the device with the online Chartlytics platform in realtime. The app should be able to access the child’s profile, charts, and configure the buttons associated with the specific chart.
Part I: UX/UI Process
User Flow 
The existing user flow, created by Chartlytics for user testing, had poor usablity and an unclear route of initiating interaction. Because the user had to access multiple parts of the UI to initiate a session, many actions were up to 6 clicks deep (the number of clicks needed to perform an action).
In the revised flow, I envisioned a simplified navigation, making a new recording session only two or three clicks away. Because hundreds of charts can be accessed through the UI, filtering and grouping charts is paramount to helping the user sort through the data in a meaningful way.
Decision Matrix 
By building a decision matrix, I was able to develop a clear path for user flow through the app. The beginning path is (a) a filter, to locate the correct chart (b) a map, to assign keys their function (c) a reference, to guide the user and prevent initial mistakes.
By grouping items based by function, I was able to get an understanding of which UI elements should be placed together or sequentially. After several iterations, a streamlined selection process was developed for testing.
Hand Drawn Wireframes
Hand drawing the wireframes allowed me to explore physical button sizes and placement, while still being low-fidelity and low-commitment. Physically walking through the UI helps to slow me down, preventing mistakes which can easily be made digitally when rushing to complete the UI. It also allowed me to focus on what elements of the UI were the most important, to start developing a hierarchy of information.
User Experience Testing (validation)
After reviewing the decision matrix with behavior analysts, screen elements were placed into a wireframe for testing.
Insight I: Classrooms
From testing, I found Behavior Analysts wanted to create groups of charts together for quick access. Adding this functionality, which I named Classroom, allowed users to place all the charts they needed for each day in an easy to access location. Adding buttons in the app to cycle to the next test meant users could switch charts on the physical device, without having to access the companion app. 
Insight II: Mapping Buttons
Users were confused with the original dropdown menu and had a hard time spatially connecting buttons based on name alone. I proposed a new method of visual mapping that allowed users to press the button they wanted, placed over an image of the device. The new method was well received.
Insight III: A reference guide
While the behavior counter is meant to be used without looking, starting with a new chart or switching frequently can lead to confusion. From this insight, I decided to display the mapped buttons as a reference when the tracker is in use. If the app stays open, the reference image will appear on the lock screen even after a period of inactivity.
Part II: Physical Design
Device Ergonomics
As the form began to evolve, basic ergonomic constraints of a hand-held device began to emerge. Using Henry Drefuss’s The Measure of Man, I was able to create a simplified form that the 97.5% percentile of users could reach.
Physical Mockups
3D printing and foam core studies assisted with the design selection process. By letting users physically hold the model, I received feedback about how comfortable the shape was and which buttons were out of reach. Watching users try to press the buttons without looking created valuable insights for how the physical form could guide the function.
User Testing Insight: Blind Use
By watching users attempt to use the models behind their backs, I realized the form had to guide the hands with tactile clues. A dish was tested to guide the user's hands in a circular pattern through the available buttons, and alternating textures and shapes of buttons helped differentiate them.
Visual Refinement
In Adobe Illustrator, I created quick renderings to convey ideas for the user interface. Discussing the options with the client helped me understand their aesthetic and manufacturing goals.
Physical Design: Features
Concave Dish
An indented dish channeled the finger around the buttons, allowing the user to feel for them without looking. The indentation also protects buttons from accidental presses while in a pocket.
Button Placement
Placing the buttons at clock hour intervals helps analysts associate actions with the position of the buttons. Differing textures, and concave and convex buttons, help prevent mistakes.
Visual Reinforcing 
A light can be configured to shine with certain button presses, to positively reinforce behaviors. This method of behavioral reinforcement has been well documented since the 1950's, and was a must-have feature for Chartlytics.
Overview: The Chartlytics Behavior Tracking System
Solution I: Accurate Data Collection
The Behavior tracker allows easy, vision-free data capture that allows analysts to remain in the moment with their patient, while still collecting accurate data. Because the device includes color coded lights, a companion reference app, and undo button, accidental presses can be undone with a single press. Each button press is recorded in realtime, and a held button acts as a timer, recording the duration of the action and showing it in the Chartlytics chart.
Above: The Chartlytics Celeration Chart Platform (desktop view)
Solution II: An Accurate Timestamp
Because the Behavior Tracker records in realtime, actions can be plotted and responses overlaid to create a very clear map of behavioral response. Additional duration measurements (taken by pressing and holding) can measure how long actions take place (such as kicking or screaming), allowing a new depth of duration data to be considered. When multiple actions are recorded, their graphs can be overlaid to determine triggers and probable cause. All this information is new to the field of Behavioral Analytics, and will assist Analysis in better diagnosis and progress in their patients.
Solution III: Course Correction In-field
Because of realtime data collection, results can be plotted instantaneously and feedback given to the Behavior Technician during the session. A small vibration, known as haptic feedback, alerts the Technician of data trends. If a patient isn’t improving, a short buzz alters the Technician that an alternative approach should be applied. If the results are trending well, a long haptic “purr” is emitted from the device.
Solution IV: Understanding the Motivation
To understand the motivation, the Analysit must accurately record 1) the action, and 2) the response. The Chartlytics Behavior tracker allows up to eight variables to be recorded in realtime, meaning the Analyst has a breadth of options for how to record data. While the motivation must be interpreted by the Analyst, the Behavior Tracker allows more accurate data collection of actions and responses, leading to more informed analysis and quicker probable cause hypotheses.
Thank you for reading!

You may also like

Back to Top