Experiments June 2007
Purpose of the experiment:
To assess the effectiveness and reliability of NEAT-o-Games an experimental study was designed and conducted per the approval of the local Institutional Review Board. Eight participants (7 males, 1 female) were recruited from the University of Houston (UH) campus. Prior to beginning the experiment all participants were requested to sign a consent form, read the NEAT-o-Games manual and fill in a pre-test questionnaire form, which asked questions concerning height, weight, and % body fat measured. During the experiment, participants' activity levels sensed with the NEAT-o-Games telemetry devices were recorded in the SQL server. At the end of the experiment, all participants were requested to fill in a post-test questionnaire.
Number of Participants: 8
Pre-Experiment Participant Information:
From the consent form and pre-test questionnaire, useful information about the profile of the participants was gathered. Specifically, the statistical mean and standard deviation of age, height, weight, percentage of fat, and Body Mass Index (BMI) for the participants were collected and computed (see Table 1). The population sample was composed of primarily young people who were bordering the overweight category. According to WHO I classification, people with 19 < BMI < 25 are normal, while those with 25 < BMI < 30 are overweight.
Table 1: Statistics of Physical Attributes for UH Participants
Tabulation of other profile information from the participants' answers in the pre-test questionnaire is shown in Table 2. It includes information about computer savviness, computer game preferences, active/inactive lifestyle, work breaks, and initial attitude to the NEAT-o-Games concept. The participants were computer literate and played computer games occasionally. They also had a moderately active lifestyle, a normal working schedule, and a positive attitude towards NEAT-o-Games.
Table 2: Profile of UH Participants
Experimental Framework:
The experiment consisted of 4 sessions. Each session included one weekday and one weekend day.
Session 1: Baseline Session. During this session, the participants were asked to carry around the NEAT-o-Games set (PDA + sensor). The system recorded their usual physical activity levels and the baseline was established.
Session 2: Emulator Session. The NEAT-o-Race simulated avatar option was activated. The player was represented by an avatar competing with a computer animated avatar in a virtual race. The rate of animation of the player's avatar was controlled by accelerometer data. The more the player moved the higher the rate of animation for the avatar it represented him/her in the virtual race. The pace of the simulated avatar was set to a level slightly lower of the recommended daily physical activity for an average person. Therefore, for the player to win the race, he/she had to complete at least the average daily physical activity quota.
Session 3: Energy Race. The human to human competitive option was activated. In that session the competitive avatar in the virtual race represented an actual player ("buddy") from the player pool that participated in the study. For each duo, a daily winner was proclaimed based on the activity scores logged by the corresponding players.
Session 4: Sudoku. In that session, each participant played competitively against his/her buddy. However, the player had the option to spend activity points gathered during the daily race in exchange for help in the PDA-based Sudoku game. This helped the player to solve difficult Sudoku puzzles, but to make up spent points he/she had to be more physically active.
Experimental Results:
Based on the intensity threshold between low and high activity, we counted the time players exhibited high activity in every session and plotted the relative results in the graph of Figure 1. High activity time as a percentage of the total time has increased in Sessions 3 and 4, where the competitive human to human race game was played, with respect to the baseline Session 1. Session 2, where the competitive human to machine race game was played appears to contradict this trend. In part, this shows that the human element may act as a stronger motivator. Nevertheless, one has to take into account that Session 2 was the first time ever players had the chance to play some form of the NEAT-o-Games (emulator mode). A sort of "cultural shock" may have contributed to a less than spectacular start. It is also likely that the current design of the emulator contributed to boredom, as the computer avatar maintained a constant speed throughout the session, which was the amortized value of an average daily activity.
Figure 1: Percentage time of high activity during the UH experiment.
Based on the intensity threshold, we also computed the descriptive statistics for low and high activity values in the various experimental sessions (see Figure 2). It appears that the intensity of high activity progressively increased as NEAT-o-Games options added in the experiment including human to human race (Session 3) and NEAT-o-Sudoku (Session 4). This mean increase of high activity values was accompanied by a widening spread. In other words, players while maintained some of the regular NEAT activities (e.g., walking at normal pace), they also exhibited more intense activities (e.g., walking at forced pace) as the experiment unfolded. Regarding the intensity of low activity, it remained relatively stable throughout the experiment.
Figure 2: High activity box plot for UH experiment.
