A variation of the hypotheses tested in the individual experiments is that the changes in questionnaire reports will occur when data is pooled from all subjects. As such, analysis was performed by taking the set of all subjects who used sensors and comparing them to the set of all subjects who did not.

Wilcoxon tests comparing these two groups reveal that significant differences occur in several of the variables. Pooled participants with sensors reported that the situation was more of a hindrance (p=.02) than those without sensors. However, pooled participants with sensors also reported the situation to be less suspicious (p=.05), more fair (p=.05), and expressed a preference for using sensors (p=.02). A trend also occurred with pooled participants with sensors reporting the situation as more ethical than pooled participants without sensors (p=.06).

This pooling however groups interviewers, disadvantaged Visible Card participants, and poker opponents along with others who did not experience sensors. Perhaps what is needed is something which is finer-grained pooling like the set of quiz takers, interviewees, and poker players who used sensors vs. quiz takers, interviewees, and poker players who did not use sensors.

Wilcoxon tests performed with these new groups show only a significant difference in preference to use sensors (p=.002), with those who had been assigned to sensor conditions preferring situations "with sensors that collect information about emotion." Additionally, trends occurred in these groups with those who used sensors reporting the situation to be less respectful (p=.09) but more comfortable (p=.07).

Another interesting pooling is to ignore the Poker subjects since it can be argued that they were playing a game and may have not felt as strongly about the experiment (which might explain the lack of significant results for that experiment). If instead only participants from the Quiz and Interview experiments are included in the analysis such that we have the set of Quiz and Interview participants who used sensors vs. Quiz and Interview participants who did not use sensors then the following results occur. Wilcoxon tests find that those who used sensors found the situation less suspicious (p=.05) and more fair (p=.05). Performance varied significantly as well, with sensor participants performing better (p=.007), although the Quiz and Interview experiments are not scored the same way. Trends also occur showing that those with sensors reported the situation to be more ethical (p=.07), more of a hindrance (p=.14), and preferred situations "with sensors that collect information about emotion" (p=.09).

A final pooling groups quiz takers and interviewees who used sensors vs. quiz takers and interviewees who did not use (and whose partners did not use) sensors.

Wilcoxon tests between these groups reveal that those who used sensors found the situation to be less respectful (p=.04) but also expressed a preference for situations "with sensors that collect information about emotion" (p=.01). A trend also occurred showing that participants who used sensors found the situation to be more comfortable (p=.06).