Experiments January 2007

Purpose of the experiment:
These studies aimed to evaluate the robustness and usability of the system as well as obtain a first indicator of its behavioral impact.

 

Number of Participants: 4

 

Pre-Experiment Participant Information:
The participants lived a mostly sedentary lifestyle, they were in their twenties and thirties, and on average they were overweight (BMI=26, WHO I classification). All of them were PDA savvy. Each user completed three sessions on different days. Sessions were scheduled post-lunch or pre-dinner times when people often take walks or watch television.

 

Experimental Framework:
Each session lasted 45 minutes and completed at a different day, but at consistent times.

Users played this in pairs. Each user was equipped with a system (PDA and a waist activity sensor) and selected each other to play the game through the user buddy list. Each system was communicating the data recorded by its sensor to the competitor's system through the cellular broadband. At the end of the session a winner was declared.

For the two-user session, the two users were up to 15 miles apart. Every user completed pre- and post-study questionnaires based on the SUS usability instrument.

 

Experimental Results:
As shown in Figure 1, addition of the computerized avatar greatly increased activity versus baseline. Moreover, activity increased further with the human opponent. All user activity was generated simply by walking in or around their offices and homes.

Figure 1. Comparison of cumulative (integral) energy expenditure averaged across 4 pilot users. The bottom line represents no motivation. The middle line represents playing against the computer avatar. The top line represents playing against another human user.

The post-session questionnaire focused on general usability questions, such as complexity and intuitiveness using a 5-point Likert scale as well as open ended questions. Based on a Kruskal-Wallis test there was no significant difference among the 4 users (p=0.8649), so results were combined. Scores for most questions were positive (i.e., 3.25 - 4). The only question with a negative response was whether the system could be used without any technical support (mean 2). However, this difference was not statistically significant. The open-ended questions revealed that the users loved the idea of NEAT-o-gaming, felt it would be helpful in a weight-loss program, and wanted more.