Readers may also be interested in whether there were gender differences in the variables measured in the various experiments. This would relate to an ongoing discussion within ethics about the role of gender in the formation of ethical opinion.
Lawrence Kohlberg's work on the moral development of children early on argued "that girls on average reached a lower level of moral development than boys did" [wikipediaGilligan2005]. Subsequent work by Gilligan argues that Kohlberg's view of moral development was biased towards Justice and values that appealed to boys. Gilligan offered "ethics of care" as a voice which contrasted with the more male-centric "ethics of justice" voice [gilligan1993].
In an effort to at least begin the discussion of how gender might relate to ethical evaluation of affective systems, I performed an analysis comparing the opinions of the participants pooled by gender. As gender differences were not among my original hypotheses, it is important to emphasize that this analysis was performed following the analyses outlined in the results chapter.
When pooling subjects based on gender and comparing the female groups with and without sensors as well as the male groups with and without sensors many trends and significant differences were observed. The following table lists the p-values from Wilcoxon tests comparing females with sensors to those without (denoted as f=) and males with sensors to those without (denoted as m=).
Table 7.4. Gender pooled sensors vs. no sensors p-value summary
| Poker-Charity Gains | Poker-Charity Loses | Interview-Control | Interview-Charity Gains | Interview-Charity Loses | Quiz-Control | Quiz-No Effect on Charity | Quiz-Charity Gains | Quiz-Charity Loses | |
|---|---|---|---|---|---|---|---|---|---|
| Ethical-Unethical | f=.76, m=1.0 | f=.79, m=.10 | f=.43, m=.51 | f=.08, m=.19 | f=.46, m=.49 | f=.51, m=.22 | f=1.0, m=.38 | f=.11, m=.65 | f=.45, m=.12 |
| Respectful-Invasive | f=1.0, m=.79 | f=1.0, m=.08 | f=.63, m=.83 | f=.19, m=.39 | f=.60, m=.19 | f=18., m=.08 | f=.69, m=.86 | f=.15, m=.18 | f=.64, m=1.0 |
| Uncomfortable-Comfortable | f=.31, m=.35 | f=.46, m=.75 | f=.88, m=.36 | f=.14, m=.85 | f=.88, m=1.0 | f=.31, m=.83 | f=.38, m=.46 | f=.29, m=.06 | f=.47, m=.91 |
| Hindrance-Help | f=.05, m=.21 | f=.15, m=.26 | f=.10, m=.49 | f=.04, m=1.0 | f=.80, m=.48 | f=.82, m=.91 | f=1.0, m=.68 | f=.20, m=.10 | f=.71, m=.26 |
| Immoral-Moral | f=.10, m=.71 | f=.52, m=.47 | f=.82, m=.84 | f=.02, m=.10 | f=.84, m=1.0 | f=.24, m=.14 | f=1.0, m=.64 | f=.48, m=.50 | f=.32, m=.84 |
| Trustful-Suspicious | f=.66, m=.02 | f=.26, m=1.0 | f=.54, m=.43 | f=.04, m=1.0 | f=1.0, m=.34 | f=.65, m=.09 | f=.49, m=.94 | f=.05, m=.51 | f=.37, m=1.0 |
| Unfair-Fair | f=.78, m=.89 | f=.51, m=.66 | f=.64, m=.55 | f=.08, m=.06 | f=.68, m=.16 | f=1.0, m=.29 | f=.74, m=.44 | f=.27, m=.53 | f=.05, m=.73 |
| Sensors-No Sensors | f=.22, m=.89 | f=.22, m=.27 | f=.11, m=.54 | f=.06, m=1.0 | f=.36, m=.32 | f=1.0, m=.31 | f=.65, m=.29 | f=.81, m=.51 | f=.28, m=.37 |
| Performance | f=.49, m=.51 | f=.73, m=.57 | f=.84, m=.11 | f=.28, m=.47 | f=.94, m=1.0 | f=.35, m=.03 | f=.34, m=1.0 | f=.42, m=.91 | f=.44, m=.48 |
In the Poker Experiment a significant difference was observed with women in the Charity Gains motivator who had sensors reporting the situation to be more of a hindrance. Men in the Charity Gains motivator on the other hand reported that the situation was less suspicious when they had sensors.
In the Interview Experiment all of the significant differences occurred in the Charity Gains motivator. Women in that situation with sensors found them significantly less of a hindrance, more moral, and more trustful.
In the Quiz Experiment there were significant differences for men in the Control motivator: those with sensors performed significantly worse. For women, those with sensors viewed the situation as less trustful in the Charity Gains motivator. Women viewed the situation as more unfair in the Charity Loses motivator.
This early analysis of gender effects suggests that there are some interesting differences. However, more work ought to be undertaken to address the role that gender plays directly.