One question that may interest readers was how subject behavior and performance interacted with reports of ethical opinions. Specifically, subjects' performance would be improved by deceptive behavior such as inflating scores in the Quiz Experiment or dishonesty about previous employment experience in the Interview Experiment. Subjects who performed "better" were thus more likely to have engaged in ethically questionable behavior. Would such subjects have different opinions about the ethical implications of their situation than those who were likely to have been honest?

Consider participants in the Interview Experiment who were in the Control motivator (viz. from "Good Company" so they had no need to lie) and also who were successful in getting hired. How did the successful "Good Company" candidate reports differ from those in the other motivators (viz. from "Bad Company") who also were hired?

Using a Wilcoxon test, significant differences were observed on the variables measured: the immoral-moral and unfair-fair variables. Participants from Bad Company reported the situation as more Immoral (p=.05). In addition, participants from Bad Company reported their situation as more unfair (p=.04). A trend also occurred along the uncomfortable-comfortable axis with those from Bad Company reporting their situation as more uncomfortable (p=.14).

One possible explanation for these differences is that the participants from Bad Company were placed in a difficult situation where they needed to be dishonest in order to maximize their reward. One plausible conjecture is that these sorts of situations influence view of the ethical acceptability of the systems tested. In support of this conjecture, consider that there was a significant difference (p=.05) in statements of how honest interviewees were. Those in the control condition and from "Good Company" reported being most honest followed by those from "Bad Company" in the Charity Loses and Charity Gains conditions.

Participants in the Quiz Experiment in some cases were give opportunities to cheat and inflate their score by misrepresenting the number they had gotten right. How do the results of those who were in the control condition (which was constructed to reduce opportunities to cheat) compare with those who reported the highest scores from the other conditions?

If we take the top quartile in terms of performance from the three non-control motivators (where people had an opportunity to cheat) and compare it with all those in the control condition (who had a decreased opportunity to cheat) then many significant differences emerge. Differences were seen on the following variables: ethical-unethical (p=.02), uncomfortable-comfortable (p=.01), hindrance-help (p=.02), trustful-suspicious (p=.02), and unfair-fair (p=.009). The control condition was reported to be significantly less ethical, more uncomfortable, more of a hindrance, less trustful, and less fair. A trend also occurred in reports along the respectful-invasive variable (p=.08) where the control was less respectful.

Surprisingly, here it seems that those with highest scores, who were not visibly surveilled by the experimenter, reported more positive opinions than did the group who were visibly surveilled. One possible explanation for this might be that the frequent presence of the experimenter (who entered the room two times during the control to examine their work vs. zero in the non-control conditions) was actually more uncomfortable than the sensors. It is also possible that subjects (lacking feedback) did not stop to consider what the attached sensors were doing, or how that information could be used in an adversarial way.