Master’s Thesis
My master's thesis (in a few words) is an examination of the Ecological Interface Design framework, and whether it can be a benefit to users in environments where there are two simultaneous tasks the user must engage in, such as driving or flying.
You can download the entire thesis here if you'd a more in-depth look.
I began working on my thesis at the end of my first year of graduate school. At first, my idea was to test Ecological Interface Design (EID) in an automotive diagnostic situation, where a driver would have to determine if and what a problem might be with a car they were driving. However, as I began to go further into the project, I realized that I needed to abstract the idea some more. Kim Vicente and others had been looking at using EID in aviation environments, but no one had done much research on if EID helped operators in those environments. The issue, as I saw it, was that EID should be able to help operators detect and diagnose faults in a dual-task system if it were to show any benefits in the 'real world'. Eventually, I even dropped the measurement of diagnosing the faults, and concentrated on the detection of faults.
To measure the detection of system faults I created the Dualsim (with help from programmer David Clark) . The Dualsim uses the DUal REservoir System Simulator (DURESS) as the fault detection task, while using a visual-psychomotor task (the ball task) as a second task. Below is a picture of the workstation setup, showing both tasks. The Ball Task is on the left monitor, DURESS is on the right.
Participants in the experiment were told to use either an EID interface or a standard interface and press the space bar when they saw a system fault. While doing this, they also had to press the arrow keys to catch falling balls in the ball task. To make it more difficult, the speed the balls fell faster on some trials.
My hypothesis was, that that participants in the EID condition would perform better at detecting faults than those in the standard display condition. Unfortunately, I did not find many conclusive results. Most of my results, while not significant actually showed participants did better using the standard interface.
Overall, I learned a great deal about conducting experiments and experimental design while working on this project. One of the most interesting things that came out of this project was a new way to look at signal detection accuracy. Dr. Gugerty (my advisor) and I were trying to find something relevant in signal detection literature that applied to this experiment, and we could not. Most signal detection experiments have very short trials where a signal is either present, or it's not. In my experiment, the trials were 3 minutes long, and the fault could happen at any time, or not at all. So, we had to develop a way to measure how good someone was at detecting faults, while not making false alarms. What we came up with was a new variable we called "speeded sensitivity". Basically speeded sensitivity takes into account the amount of time a fault was not present, and the amount of time after a fault was present. The greater the amount of time a fault wasn't present, the better, and the lower the amount of time a fault was present, the better.
I learned a lot about project management, too. With my advisor out of the state quite a lot during my time at Clemson, it was up to me to keep focused and on track. Completing my Master's Thesis is one of the most educational experiences in my life (which is, I think, the way it should be).
