Responses to the attitude probes

In general, the attitudes toward automation in the subject community of this study match closely the attitudes observed by Wiener a decade ago [9]. This was somewhat surprising, since the participants in Wiener's study worked for a different airline, were flying what was for most their first automated airplane, and had on average less than half the automated flight experience of the participants in the present study (The median hours in type at phase two of Wiener's study was 1,100. Median hours in type for the present population is 2570.) This indicates a fairly stable structure of attitudes in the pilot population at large. Our subjects seemed slightly more positive about automation than Wiener's subjects. This is consistent with the general trend to view automation more positively with increasing experience of it as discussed below.

See appendix 3 for the response histograms discussed in this section.

The scaled responses of the pilots were coded with values from 1 for strongly disagree to 5 for strongly agree. The scores reported below are simple averages of these numeric values. A score of 3.0 indicates a neutral response. Scores above 3.0 are positive on the probe, and scores below 3.0 indicate negative responses to the probe. For each probe, the mean and standard deviation are reported in the following format (mean/std dev).

Probe 1, FlyingSkills: In general, our respondents agree with this probe (3.43/1.15). The distribution of responses across agreement levels looks very much like that in Wiener's 1989 study except that more of our subjects agree strongly and fewer are neutral. The distribution is slightly bi-modal.

Probe2, WorksGreat: Our respondents agree with this probe (3.62/0.90). They are more in agreement with this probe than Wiener's subjects were.

Probe 3, KnowMode: Our pilots agree strongly with this probe (3.72/0.98). Their responses are virtually identical to those observed by Wiener. The distribution is very slightly bi-modal.

Probe 4, CompanyPressure: Our pilots did not agree with this probe (2.38/0.88). Their responses are virtually identical to those observed by Wiener.

Probe 5, FreeToManage: Our pilots agree strongly with this probe (3.79/0.91). The distribution of responses across agreement levels shows more pilots agreeing with this probe than did so in Wiener's study.

Probe 6, Surprises: Our pilots agree with this probe (3.37/1.04). They tend to agree slightly less than did the subjects in Wiener's study. The distribution is slightly bi-modal.

Probe 7, FewerErrors: Our pilots are neutral on this probe. The distribution approximates a normal distribution centered on 3.0. (3.02/0.97).

Probe 8, AheadOfAirplane: The pilots agree with this probe (3.51/0.96). The distribution is uni-modal and more skewed toward agreement than was observed by Wiener.

Probe 9, SetupAndManage: The pilots are on average neutral on this probe (3.17/1.13), but the distribution is strongly bi-modal. The bi-modality in our case is skewed toward agreement, while in Wiener's study a similar structure of bi-modality was skewed toward disagreement.

Probe10, NotReduceWorkload: The pilots disagree with this probe (2.69/1.11). The distribution is moderately bi-modal and skewed toward disagreement. Wiener observed a very strong, nearly symmetrical, bi-modality on this probe.

Probe 11, ConsultAnnunciator: Pilots agree with this probe (3.61/0.99). The distribution has a very slight bi-modality. There was no corresponding probe in Wiener's study.

Probe 12, AdequateTraining: The pilots agree with this probe (3.46/1.13). The distribution shows a slight bi-modality. Wiener's results showed a similar, but more pronounced bi-modality.

Probe 13, HelpDoJob: This probe received the strongest agreement levels (3.84/0.76). Our distribution is more strongly skewed toward agreement and has lower variance than Wiener's.

Probe 14, AltitudeBust: This is the probe with which the pilots agree least (2.27/2.67). Our distribution is unimodal, Wiener's was very slightly bi-modal.

Probe 15, ButtonPusher: The pilots do not agree with this probe (2.67/1.11). The distribution shows a moderate bi-modality. Wiener observed a uni-modal distribution skewed toward disagreement. [Some group of our guys feel more like button pushers than Wiener's pilots] Probe 16, MisunderstoodModes: The pilots disagree with this probe (2.48/1.15). We observe a strongly bi-modal distribution, as did Wiener. More of our pilots strongly disagree with the probe than did in Wiener's study.

Summary: In terms of grand means, HelpDoJob received the strongest agreement (and had the smallest variance), with FreeToManage, KnowMode, ConsultAnnunciator, and WorksGreat close behind. AltitudeBust was most strongly disagreed with, along with CompanyPressure and MisunderstoodModes. MisunderstoodModes and FlyingSkills showed the highest variance. Every possible response (from strongly agree to strongly disagree occurred at least once for every attitude probe.