Skip to article
AHFE AHFE Discoveries AHFE International Open Access Journal
AHFE OPEN ACCESS OPEN ACCESS CONFERENCE PROCEEDINGS PAPER

Verification of the Effects of Personalized Evacuation Alerts using Behavioral or Location Information with the Sense of Urgency in a Disaster

Ritsumeikan University 1-1-1, Noji-Higashi, Kusatsu-City, Shiga, JAPAN ↗
Publication year
Digital Object Identifier10.54941/ahfe1001701

Abstract

Delayed evacuation from a huge disaster is one of the factors that increase human damage. It has been pointed out the difficulty in urging an early evacuation after a disaster because people tend to underestimate information that is inconvenient for them, which is called normalcy bias. The most effective way to urge evacuation is to contact each of the evacuees directly, such as calling them. That is, we consider that we should provide information to evacuees that makes them feel as if it is being said to them directly.In this study, we focus on disaster alert services which send disaster information to personal mobile devices (i.e., smartphones) and consider textual expressions of alert that generate a sense of urgency for disaster. Since people always carry their smartphones with them, we expect that personal expression of alert will be possible using the information available on their own smartphones. As for information obtained from a device, in this study, we focus on the location information acquired from GPS and the behavioral information of the user acquired from the accelerometer. We assume that if an alert with textual expression that seems to identify the individual user by using these information is provided, the user receives it will feel as if it is being said to him/her directly. As for the degree of identification of individuals, three degrees of expression are set for location and behavioral information, respectively. Concretely, for location information, 1. the first indicates that a river has flooded, and 2. the second one gives the name of flooded river, and 3. the last shows the distance from the flooded river to user’s current location. For behavioral information, 1. the first includes no information about user’s behavior, 2. the second indicates that the user is reading the alert message, and 3. the last points out how the user is operating the smartphone (e.g., a smartphone is in user’s hand or on user’s desk). We set nine patterns of alert expression by combining the expression based on location and behavioral information.To verify the effectiveness of the alert expressions, we conducted a comparative verification experiment for them. In this experiment, we had the collaborators perform the specified daily task alone in a dark room hearing rain. While the collaborators were performing the task, the one evacuation alert of the nine patterns was sent to them. After receiving the alert, the collaborators confirmed it and then answered the questionnaire. Each of the collaborators repeated this process five times. In the questionnaire, we asked questions about how much they felt a sense of urgency, and how much they thought the alert was directed at them and so on. As a result of this verification, we found that textual expression using location information tend to be more effective as they feel that the message is being said to them. We will describe the detailed experimental results in the camera ready manuscript.

INTRODUCTION

Delayed evacuation from a huge disaster is one of the factors that increase the impact on people. It has been identified that it is difficult to urge early evacuation after a disaster because people tend to underestimate information that is inconvenient for them; this is called normalcy bias. The most effective way to urge evacuation is to directly contact each of the evacuees, such as by calling them. That is, information to evacuees should be provided in a manner that makes them feel as if it is being said to them directly.

In this study, we focus on disaster alert services that send disaster informa- tion to personal mobile devices (i.e., smartphones) and consider the textual expressions of alerts that generate a sense of urgency. Because people always carry their smartphones with them, we expect that personal expression of alerts will be possible using the information available on their own smartph- ones. Studies have examined the ways in which textual expressions improve evacuation awareness. For example, Tanaka et al. (2011) showed that using unfamiliar adverbs in relatively concise Japanese textual expressions encou- rages rapid evacuation. In another study, textual expressions considering psychological factors were proposed, and the effects of enhancing evacuation

© 2022. Published by AHFE Open Access. All rights reserved. 140

Verification of the Effects of a Personalized Evacuation Alerts 141

awareness were compared (Yasui et al. 2020). The psychological factors con- sidered in their study were fear-arousing communication (Janis et al. 1953), majority synching bias (Takayama et al. 2014), and impatience. However, no studies have examined textual expressions that focus on what makes evacuees feel that it is a personal message addressed to me.

In this study, we focus on the location information acquired from GPS and the behavioral information of the user acquired from the accelerome- ter on the device. We assume that if an alert with textual expression that seems to identify the individual user by using the information is provided, the user receiving it will feel as if it is being said to them directly. We set nine patterns of alert expressions by combining expressions based on location and behavioral information. To verify the effectiveness of the alert expressi- ons, we conducted a comparative verification experiment for the nine textual expressions.

TEXTUAL EXPRESSION OF EVACUATION ALERT In this study, we assume flooding caused by heavy rain and consider the textual expression of evacuation alerts for the flooding. For information obtained from a device, we focus on the location information acquired from the GPS and the behavioral information of the user acquired from the accele- rometer. That is, by receiving a message with a text expression that includes detailed position information and specific state information, a user will feel as if it is being said to them directly.

Regarding the degree of identification of individuals, three degrees of concreteness of expression were set for location and behavioral information, respectively. Specifically, for location information, the first indicates that the river has flooded, the second one gives the name of the flooded river, and the last shows the distance from the flooded river to the current location of the user. For behavioral information, the first includes no information about the user’s behavior, the second indicates that the user is reading the alert message, and the last mentions how the user is operating the smartph- one (e.g., the smartphone is in the user’s hand or on a user’s desk). We set nine patterns of alert expressions by combining the expressions based on location and behavioral information. Table 1 shows the nine expression patterns. In the table, the columns show the concreteness of the location information, and the rows show the concreteness of the behavioral information. The text in the table includes the river name (e.g., Kusatsu River) and the distance from the river (e.g., “2.7km”) applied in the experiment (explained in the next section).

COMPARATIVE EXPERIMENT To verify the effect of the text expressions shown in the previous section, we conducted a comparative experiment. The nine textual expressions were evaluated in terms of awareness of evacuation, directness of the message to a receiver, and a sense of urgency.

Table 1. Textual expression of evacuation alert for flooding.

Table available in the original PDF paper.

[1] Only river [2] River name [3] River name and distance [A] Non [1-A] You are in danger near the river. Please evacuate.

[2-A] You are in danger near the Kusatsu River. Please evacuate.

[3-A] You are in dan- ger at 2.7 km from the Kusatsu River. Please evacuate. [B] Reading state

[1-B] The area near the river is dangerous. Ple- ase evacuate if you are reading this message.

[2-B] The area near Kusatsu River is dan- gerous. Please evacuate if you are reading this message.

[3-B] The location 2.7 km from the Kusatsu River is dan- gerous. Please evacuate if you are reading this message. [C] How a user operates

[1-C] The area near the river is dangerous. Ple- ase evacuate if you are reading this message holding your phone..

[2-C] The area near Kusatsu River is dan- gerous. Please evacuate if you are reading this message holding your phone.

[3-C] The location 2.7km from the Kusatsu River is dan- gerous. Please evacuate if you were reading this message holding your phone.

Experimental Procedure We first explained the flow and contents of the experiment to the participant and obtained their consent to participate in the experiment. Then, we had the participant perform the specified daily task, English translation, alone in a dark room hearing rain sounds. While the participant was performing the task, they received one of the nine evacuation alerts on their phone. The participant confirmed it and then completed the questionnaire. After comple- ting the questionnaire, they resumed the English translation. Each participant repeated this process five times. The time interval for receiving the alerts was set randomly at 3, 4, and 5 min each time. During the experiment, the expe- rimenter left the room and observed the participant using a camera located in the room. Then, the experimenter selected the five messages from the nine messages shown in Table 1 and presented them manually in random order .

Results and Discussion

We had 18 participants, and each alert expression was evaluated by 10 participants. In the questionnaire, we asked the following questions: Q1. Do you want to evacuate after reading this message? Q2. Did you think that a disaster was happening close to you when you read this message? Q3. Did you think this message was directed at you? Q4. Did you feel a sense of urgency when you read this message?

For each question, the participants answered on a seven-point scale (1 = strongly disagree, 7 = strongly agree).

Tables 2, 3, 4, and 5 present the average scores for Q1, Q2, Q3, and Q4, respectively. For all questions, comparing the differences depending on the

Verification of the Effects of a Personalized Evacuation Alerts 143

Table 2. Questionnaire results for Q1 (awareness of evacuation).

Table available in the original PDF paper.

[1] [2] [3] Average [A] 3.90 3.10 4.20 3.73 [B] 2.60 4.00 4.40 3.67 [C] 3.80 4.00 3.80 3.87 Average 3.43 3.70 4.13 3.76

Table 3. Questionnaire results for Q2 (degree of closeness to the disaster).

Table available in the original PDF paper.

[1] [2] [3] Average [A] 3.90 4.10 4.90 4.30 [B] 3.40 4.20 5.50 4.37 [C] 4.70 4.20 4.90 4.60 Average 4.00 4.17 5.10 4.42

Table 4. Questionnaire results for Q3 (directness of the message).

Table available in the original PDF paper.

[1] [2] [3] Average [A] 4.20 3.80 5.50 4.50 [B] 3.20 4.60 5.50 4.43 [C] 3.90 5.00 3.70 4.20 Average 3.77 4.47 4.90 4.38

Table 5. Questionnaire results for Q4 (a sense of urgency for disaster).

Table available in the original PDF paper.

[1] [2] [3] Average [A] 4.10 3.10 4.40 3.87 [B] 2.60 3.80 5.00 3.80 [C] 3.80 3.80 3.90 3.83 Average 3.50 3.57 4.43 3.83

specificity of the location information, it tends that there are positive answers in expression 3-A, 3-B, and 3-C, which show the river name and distance. In the case of using only the river name (i.e., case [2]), some of the questions had positive answers, for example, Q2 in Table 3 or Q3 in Table 4. Textual alerts with no river information were evaluated as negative in many cases. Comparing the differences in behavioral information, there is no common trend in these results for the questions. For each question, the most highly rated behavioral information for each location was different. However, when looking at the combination of the two types of information, expression 3-B was the most highly rated for all questions.

Two-way ANOVA was conducted for location and behavioral information. Table 6 shows the p-values obtained from the analysis. There are no signi- ficant differences in the factors of behavioral information. For the location factor, there are significant differences or trends, except for Q1. In addition, the results show the interaction effects for Q3 and Q4.

Table 6. Results of a two-way ANOVA.

Table available in the original PDF paper.

Q1 Q2 Q3 Q4 behavioral information 0.887 0.802 0.831 0.988 location information 0.241 0.048** 0.093* 0.053* behavioral * location 0.169 0.585 0.096* 0.095*

*:p<0.1, **:p<0.05

By increasing the concreteness of the location information presented, the average values of the answers tended to be positive. In particular, the rated values were high for the text that added distance information to the river name; there is a significant difference in the questions about the degree of closeness to the disaster. For the question about the directness of the mes- sage and a sense of urgency, we also found a significant tendency in terms of location information. In this experiment, the river name and specific distance values ⣋⣋were presented for the “distance from the flooded area,” so it is considered that it was easy to feel a sense of urgency. However, we could not see the effect of the concreteness of behavioral information. The reason for this can be found in the text expressions of “you are reading the message”and “you are reading this message holding your phone.” These expressions seem to specify the state of the receiver of the information, but this is a natural behavior of the receiver reading the message. Therefore, many participants may not have interpreted the message as their actions were identified and the message was directed at them. Thus, although we cannot see the effect of concreteness of the behavioral information, it is shown that the concreteness of the location information tends to increase the sense of urgency and the sense of directness of the message.

CONCLUSION

In this study, we discussed the text expressions of alerts that generate a sense of urgency in a disaster or that make the evacuees feel as if they are being addressed directly. Assuming flooding of the river due to heavy rain, this study presented nine patterns of text expressions using location and behavio- ral information obtained from mobile devices. Although we did not observe the effect of concreteness of the behavioral information in the comparative verification, it was shown that the concreteness of the location information tends to increase the sense of urgency and sense of directness of the message. Our future work will examine other expressions of behavioral information and verify their effects.

ACKNOWLEDGMENT

This study was supported in part by JSPS KAKENHI (grant number JP20K11911).

Verification of the Effects of a Personalized Evacuation Alerts 145

REFERENCES

Janis, I. L., Feshbach, S. (1953). “Effects of fear-arousing communication”, The Journal of Ab-normal and Social Psychology, 48(1), 78–92. Takayama, Y., Hirosaki, M. (2014). “Quick Evacuation Method for Evacuation

Navigation System in Poor Communication Environment at the Time of Disa- ster”, Proc. in the 2014 International Conference on Intelligent Networking and Collaborative Systems, 415–429. Tanaka, K., Kato, T. (2011). “Perceived danger can be influenced by how the emergency information is expressed”, The Japanese journal of cognitive psychology, 9(1), 1–7. Yasui, T., Kitamura, T., Izumi, T., Nakatani, Y. (2020). “A Validation of Textual

Expression about Disaster Information to Induce Evacuation”, Proc. in the 22th International Conference on Human-Computer Interaction (HCI’20), 289–301.

Cover of the proceedings volume Usability and User Experience

Explore the complete digital volume

Usability and User Experience

Read this individual research paper free through the original publisher, or explore the full digital proceedings book.

ISBN 978-1-958651-15-5

Explore this ISBN in the AHFE Bookstore Listed in the July 2026 bookstore catalogue; search by ISBN to confirm current availability · Digital book purchase optional

Explore Related Research

Related research 8 matches

More papers from this volume

← Back to Research ExplorerBrowse the full library →