For phase one due to its duration, only one participant was able to go through the experiment. The average value for all four scenarios was roughly the same. These results although their analysis was not as in depth as Mikellides results portray the same level of arousal is maintained throughout all four scenarios.
For phase two, 18 participants were tested (7 female and 11 male) although only 17 of the results were usable since the data recorded for participant 7 had a system error. Phase 2 results showed main changes in the EDA results with a common trend of starting high then decreasing, with the majority the lowest point being in the high saturation green (G100) then increasing at the low saturation blue (B50). The slow decrease in arousal from the start till the green could be considered due to participants getting used to the experience but the trend of an uprise in the data when getting to the B50 suggests that the colours are having a physiological impact.
The ranges of values acquired for each participant varied. For example p1 had a larger range and base level of skin conductance than p8 but there were still variations between the different stages. Therefore if the actual values for all participants were overlaid it would be hard to distinguish a clear trend. This is why to graph all participants results for EDA in one graph, the values were normalised. When plotting the normalised EDA changes, it was found that there was a clear trend of a decreasing slope and a rapid change in the last stage of B50.
The results of colour a high colour saturation having a bigger effect than colour hue have not been reflected in the results of phase 2. When comparing the average of each colour and compiling them per colour (testing hue) and per saturation, the colour Red has the largest average (50.1%) followed by blue (32.46%) and green (28.60%). While the difference in saturation is slightly higher value for high saturated colours it’s not a very large difference only 5.15%.
The ECG data analysis needs to the worked on since there was a lot of noise in the signals. This caused the final values to be very varied and not fully correct. Table 9 shows a graph of the normalised values although the data has to be cleaned further to be able to produce a readable graph.
TABLE 7. EDA results graphed
Error with data
TABLE 8. EDA combined results graphed
TABLE 9. Colour hue versus saturation results
TABLE 10. Beats per Minute chart