• Federal Highway Administration >
• Publications >
• Research Publications >
• 96147 >
• Sec4.Cfm >
• Development of Human Factors Guidelines for Advanced Traveler Information Systems and Commerical Vehicle Operations: Components of the Intelligent Transportation Systems: Designs Alternatives for In-Vehicle Information Displays

Development of Human Factors Guidelines for Advanced Traveler Information Systems and Commerical Vehicle Operations: Components of the Intelligent Transportation Systems: Designs Alternatives for In-Vehicle Information Displays

CHAPTER 4. CONCLUSIONS AND DESIGN IMPLICATIONS

Figure 17 summarizes the ATIS and driver characteristics that this experiment considered. The effects of these characteristics on driving safety and warning compliance were analyzed using a range of intervening variables and direct measures. The results of the analysis have both general and specific implications for the design and evaluation of ATIS.

GENERAL CONCLUSIONS

DESIGN IMPLICATIONS

GENERAL CONCLUSIONS

A general issue facing ATIS designers is the concern that ATIS warning messages may go unheeded by drivers. A critical element of ATIS design concerns designing an ATIS device that makes information easily accessible and compelling so the drivers comply with the warnings. The results show converging evidence that ATIS warnings can generate a greater compliance compared to road signs; however, the effects on trust and self–confidence show that certain ATIS designs may undermine drivers' relationships with the ATIS device, leading to overreliance and reduced safety. The results also show that ATIS design characteristics can be manipulated to affect the level of driver compliance.

Another general issue that faces ATIS design is its potential to undermine driving safety. Based on the information processing and mental workload paradigm, many have suggested that an improperly designed ATIS device could jeopardize driving safety by overloading drivers. Multiple–resource theory predicts that this will be particularly critical for devices that force drivers to share their visual resource between reading ATIS warnings and the driving task. Measures of workload, driving performance, and situation awareness all suggest that multiple–resource theory and the mental workload paradigm does not explain safety decrements associated with the design characteristics of the ATIS.

This investigation hypothesized another safety concern. An improperly designed ATIS device might jeopardize safety by leading drivers to favor in–vehicle information sources and ignore critical roadway information. The results of this experiment show that ATIS devices can undermine driver performance by fostering an overreliance on ATIS information. The effects associated with workload, situation awareness, and driving safety measures all support this assertion. The results show little evidence that information overload undermines safety. Instead of workload–related safety problems, it seems that the ATIS may induce "complacency" as discussed in Singh, Molloy, and Parasuraman (1993). It seems that ATIS information may lead drivers to become complacent, focusing on in–vehicle information while disregarding important, out–of–vehicle information. More specifically, complacency may reflect inappropriate cue utilization (Hammond, 1966). The ATIS messages, particularly the command messages, may appear as a particularly salient cue that is weighted more heavily than more important roadway information. The results also show how particular ATIS design characteristics can exacerbate the overreliance and its negative effects on driving safety.

Not surprisingly, driver age emerged as an important variable that moderates the effectiveness of the ATIS. Although the overall driving performance of older drivers was worse than that for younger drivers, the negative effects on safety of the ATIS messages are less pronounced for older drivers compared to younger drivers. One explanation for this effect is the extensive driving experience of older drivers. The more extensive experience may provide older drivers with better strategies for sampling the environment and combining information from the ATIS and the roadway. Older drivers may use their experience to weight in–vehicle cues more appropriately than younger drivers. In addition, older drivers seem more likely to trust the capabilities of the ATIS, even when ATIS information is not consistently available. These results build upon those described by Kantowitz, et al. (1997). Their findings suggest that a different process governs the trust and self–confidence of older and younger drivers. They suggest that younger drivers use the ATIS based on their subjective feelings, while older drivers' use of the system alters their feelings. This suggests a greater inertia in older drivers' level of trust. Our data support this conjecture because the trust of older drivers was generally higher and did not drop when ATIS information was unreliable. This result is counter to previous research, but it is consistent with the process proposed by Kantowitz, et al. (1997). If drivers automatically receive information from an ATIS, older drivers learn to trust it more than younger drivers. If an ATIS device requires drivers to request information, then younger drivers' initial trust will lead them to use it more and older drivers may not learn to trust it.

Gender interacted with driver age and message style. Several converging effects suggest that younger women assimilate ATIS notification messages more effectively than command messages. The opposite is true for older women who assimilate command messages more effectively. In general, men assimilate notification messages more easily than command messages. These results show that complex sociological trends might complicate the design of ATIS devices.

Top

DESIGN IMPLICATIONS

The implication for design guidelines is summarized briefly, followed by a more detailed description of the more important outcomes of this investigation.

Specific Design Guidelines and a Suggestion for Future Experiments

These results can be summarized with several recommendations:

• Where possible, ATIS information should be paired with redundant roadway information.
• Command–style messages without redundant roadway information should be used for only high–criticality information that requires the driver's immediate response.
• Command–style messages should be used for medium–criticality information only when they can be paired with redundant roadway information.
• Notification messages should be paired with redundant roadway information and used for low–criticality information.
• Complex interactions with age and gender suggest that systems may require options that can be tailored to the specific requirements of the driver.
• Younger drivers should receive additional training so that they do not become over reliant on ATIS information.
• Text–based messages should be used to preserve drivers' trust where ATIS information is not consistently available.
• Command messages should be used as infrequently as possible because they undermine drivers' trust in the system and self–confidence in their abilities to extract the information from roadway sources.
• Subjective estimates of situation awareness are not well–correlated with objective measures, suggesting that future studies should include objective measures of situation awareness when possible.

An important design implication concerns the implementation of ATIS devices relative to the infrastructure of standard and changeable–message road signs. Figure 18 shows the effects on safety and compliance for the different levels of ATIS and road–sign information. This figure shows that providing drivers with only ATIS information leads to a high level of compliance, but it can also compromise safety. The figure also shows that providing ATIS information with redundant road–sign information generates a high level of compliance without the associated decline in safety. When no ATIS information is available, compliance is relatively low.

Message style emerged as a critical ATIS design characteristic. This design characteristic has not been widely studied, but the results suggest that it has a more powerful effect on driver behavior than more commonly studied characteristics, such as display modality. Message style influences both compliance and safety. Figure 19 shows the design trade–offs for message style and the redundancy of roadway signs. Given the consequences for safety and compliance, command messages should be reserved for safety–critical situations. This is particularly true for situations where redundant roadway information is not provided. Specifically, the effect of message style in figure 19 suggests that the importance of the messages needs to be considered when choosing a message style. The four quadrants of figure 19 link the importance of the message to the style in which it should be presented. High–criticality messages should use a command style, while low–criticality messages should use a notification style with redundant roadway information.

Command and notification message styles also had a great effect on drivers' trust in the ATIS and self–confidence. Figure 20 shows that the command message style undermines both trust and self–confidence. This situation will leave drivers in a double–bind situation, where they feel uncomfortable using the ATIS information, but have little alternative because they do not have confidence in their abilities.

Message modality and location had little consistent effect on safety or compliance and so the results do not suggest any guidance beyond those that currently exist for these characteristics.

The design implication generated by this experiment addresses in–route IVSAWS and ISIS messages. These implications may also apply to route–guidance messages, but this should be validated. Furthermore, these conclusions are based on the results of a single experiment and further investigation is needed to confirm them and establish the limits of their application in ATIS guidelines. The Battelle automotive simulator is a relatively low–fidelity simulator, which can be considered as a "microworld" (Brehmer, 1990). As a microworld, the simulator presents the driver with many of the same demands and circumstances faced in actual driving scenarios; however, the limited field of view, low resolution of images, and the novelty of the ATIS messages limit generalizability of the results. Studies using more realistic driving simulators and on–road driving will help validate the findings of this study.

Top

FHWA-RD-96-147