Evaluating the Effectiveness of the Earthquake Early Warning Message in China: An Affordance Perspective Using Immersive Virtual Reality

Abstract

The early earthquake warning (EEW) system is essential for mitigating the effects of seismic incidents. However, in China, the design of EEW messages has not received much attention. This study employs affordance theory to examine the effectiveness of the EEW message generated by the Institute of Care-Life (ICL) in China, specifically by investigating four aspects of affordances: functional, cognitive, sensory, and emotional affordance. With 68 participants, we conducted an immersive virtual reality experiment. The results revealed that the ICL EEW message has a strong emotional affordance but inadequate functional, cognitive, and sensory affordance. These data provide recommendations for enhancing EEW messages, which could result in better interaction during earthquakes in China. This study investigated the viability of immersive virtual reality as a research tool for EEW. It increases understanding of the elements that determine the effectiveness of EEW communications, leading to better preparedness and response measures, reducing the impact of earthquakes and saving lives and property.

1. Introduction

An earthquake is the violent trembling of the Earth’s surface triggered by the abrupt displacement of a plate in the Earth’s crust. Earthquakes are inevitable and can result in significant loss of life and property. Numerous nations are emphasizing the development of earthquake early warning (EEW) systems due to the difficulty of predicting earthquake occurrence [1].

Before the ground begins to tremble, EEW can provide a few seconds to a few minutes of notice [2]. It is a revolutionary approach to decreasing disaster risk and enhancing urban earthquake resilience [3]. The primary purpose of EEW is to forewarn individuals and automated systems that anticipate seismic waves at a particular location, allowing them to take precautions to reduce the risk of injury [4]. As the demand for oil and gas resources increases, particularly in some big industrialized nations, EEW systems also play a crucial role in ensuring energy security [5,6,7]. The 2008 Wenchuan earthquake in China was the driving force behind the founding of the Institute of Care-Life (ICL), which currently covers 1,100,000 square kilometers of EEW. ICL is continuously improving EEW technology in a wide range of capacities to reduce risks and damage both domestically and internationally. Since 7 June 2011, over 10,000 earthquakes have triggered the ICL EEW without a false alarm (see http://365icl.com accessed on 20 December 2022). Massive improvements have increased the accuracy of EEW’s earthquake detection capabilities [8]. To effectively utilize the notifications, EEW must involve the public due to technological and social limitations [9]. The difficulty is convincing individuals to take the necessary precautions after perusing the message [10]. The content of ICL EEW messages includes attributes such as color, intensity, shape, icon, typesetting, etc. Each of them must be well designed.

The ICL EEW message on television was employed as a research target in this project. When an earthquake occurs, a pop-up window containing information about the upcoming quake appears on the television screen. The content of the message includes a countdown to the earthquake’s arrival, its epicenter, magnitude, and the user’s current location. According to several studies, information provided on shaking frequency or duration is neither required nor helpful [11]. Omori et al. [12] reached the opposite conclusion, stating that in order for individuals to be informed and initiate a safe and effective response, they require complete emergency information as soon as possible. It is also necessary to note that the ICL EEW message contains no information about the guidance of preventive actions. Other EEW messages, such as ShakeAlert, include human-based protective behavior recommendations (such as drop, cover, and hold on) [13]. Consequently, what specific details are necessary when an earthquake occurs? Can the current ICL EEW message’s users take the right precautions? The aim of this paper is to evaluate the effectiveness of the current ICL earthquake early warning message on television. It has been demonstrated that EEW plays a significant role in reducing earthquake casualties and losses [4], but what are the potential physical and psychological impacts on individuals? In other words, is there a relationship between the EEW message and people’s behavior when an earthquake occurs? In this study, it was proposed that there is a connection between the ICL EEW message and how individuals react during quakes. We aimed to test this hypothesis through a simulation experiment.

This paper is organized as follows. In Section 2, the idea of affordance theory is presented with an analysis of the viability of affordance-based methods as an investigation strategy for this study. Furthermore, Section 2 presents considerations for immersive virtual reality as a simulation setting. Section 3 continues an in-depth analysis of affordance and IVR and identifies the study topic for this paper. Section 4 presents the experimental design, along with instructions on how to use Unreal Engine to simulate earthquakes and EEW and how to perform movement in a virtual device. Section 5 describes the hardware, procedures, and data to be collected. Section 6 presents the experiment results, including the four aspects of affordance. Section 7 discusses the results of the experiment to identify the advantages and drawbacks of the ICL EEW message as well as the contributions and limitations of this paper. Section 8 concludes the article and clarifies the findings.

2. Literature Review

Previous studies have shown that by urging people to take precautions, warning messages have the ability to lower the number of fatalities during extreme occurrences like earthquakes [14]. In certain research, data on people’s responses was gathered in order to examine the impact of EEW messages. Sutton et al. [15] examined individual responses to EEW messages by simulating them on a screen. The results showed that informational icons have a limited effect on individuals’ protective behaviors. Fischer et al. [16] investigated people’s perceptions of warning messages by gathering the responses of participants after viewing the messages. The results showed that protective action guidance increased people’s response efficacy. However, people may act differently in an emergency than they would in a non-stressful experiment environment [17]. In this study, we hoped to find a method to efficiently study the design of earthquake early warning messages.

It was determined that the affordance-based method is the most effective method for analyzing this issue. From the standpoint of ecological psychology, affordance was utilized to define the actional properties of the environment and an actor. It was discovered originally by James Gibson [18]. In ecological settings, animals may instantly perceive the information provided by an object’s physical features, enabling them a variety of behavioral options. Norman [19] identified affordance as the perceptible and actual attributes of an object, particularly the fundamental attributes that determine how the object can be utilized. This definition is derived from the standpoint of product design. The affordance of a product can provide critical information about its function. For instance, a knob is for rotating, and a flat door handle results in thrusting; therefore, affordances are environmental factors that provide movement to an organism designed to perform a specific function [20]. Consequently, affordance-based methods may provide a new perspective for evaluating EEW messages by emphasizing the complementary relationship between human behavioral responses and EEW messages in earthquake scenarios.

Several methods, including interviews [21] and surveys [22,23], can be used to collect information on human responses to an EEW message in an actual earthquake. These data were primarily derived from individuals’ prior earthquake experiences [24], which could be affected by recall bias and lead to inaccurate results [25]. In addition to collecting behavioral responses to EEW messages elicited by actual earthquakes [15], specific research has utilized laboratory environments to collect behavioral responses. The data are less susceptible to recall bias and memory loss when real-time observations and surveys are conducted promptly after the experiment. Nonetheless, the lack of ecological validity is a major limitation of these investigations. Traditionally, ecological validity refers to the ability to generalize behavior observed in the laboratory to behavior observed in the wild [26]. As a result, research into the affordances of EEW messages requires a method that gathers data in real time while maintaining ecological validity. IVR (Immersive Virtual Reality) is a technology that generates a simulated environment using computers, applications, and ancillary devices [27]. IVR employs auditory and visual effects to establish an authentic setting for the user to experience. During IVR interactions, users shift from passive observers to active participants capable of generating personalized behavioral responses to the virtual environment [28]. We can attain greater ecological validity by simulating the residential environment during an earthquake using immersive virtual reality.

In a laboratory environment, we investigated the affordances of the ICL EEW message on television for this project. We recreated a virtual residence in IVR, simulating earthquake-induced shaking and displaying the ICL EEW message. We then measured the participants’ reactions to the message by collecting data on their avoidance area choices, their timing of avoidance, their recall of information, and their emotions. The results of this study offer guidelines for designing the ICL EEW message and increasing public awareness of its importance, which are crucial to the success of the ICL EEW system.

3. Methods

3.1. Affordance

Affordance has recently gained attention in a variety of disciplines, including product design [29], package design [30], social media intervention design [31], interaction design [32], app design [33], etc. Despite the fact that various studies have varying definitions of affordance, it offers a fresh perspective to enhance the design. Tsai et al. [34] suggested an innovative strategy and methodology to investigate how affordances affect smartphone usage. Karat et al. [35] concentrated on the concept of affordance and the effect of user perception on frequency modulation as it pertains to smartphone apps. Carattin et al. [36] demonstrated the affordance theory as a method for evaluating the design of the refuge signage system.

Numerous researchers have enriched and clarified the concept of affordance as its application has expanded into an increasing number of domains. Hartson [37] distinguished four types of affordances: cognitive, object, functional, and sensory. Gaver [38] developed the idea of technological affordance, underscoring the aggregation of time and space support that facilitates action through technological interaction. By elucidating the distinction between usability and utility, McGrenere et al. [39] clarified affordances in terms of design. The usefulness of a design is decided by what it provides, and the design’s usability can be enhanced by creating perceptual information that clearly specifies these affordances.

Expanding the concept of affordance enables a fresh perspective on EEW messages. Before making the message more functional, it is essential to evaluate its usefulness (what it offers and what it should offer in addition). Consequently, in this study, we accomplished this by analyzing the participants’ responses when they received the message in a virtual earthquake simulation (see Figure 1). Figure 2 showed the affordances of the ICL EEW message examined from the four perspectives, specifically, in functional affordance, cognitive affordance, sensory affordance, and emotional affordance. Functional affordance helps users achieve tasks; cognitive affordance allows users to engage in cognitive actions, such as thinking, understanding, and remembering; sensory affordance enables users to respond rapidly; and emotional affordance is an attribute that establishes an emotional connection among users.

3.2. Immersive Virtual Reality

Immersive virtual reality allows us to simulate authentic situations in a virtual environment. IVR has had an important impact on the investigation of emergency human behavior and is now widely used in research across multiple disciplines. Several studies have created emergency situations such as fires [40], cyclones [41], earthquakes [42], and gunfire [43] in IVR, thereby resolving the problem of extreme dangers in conventional experiments. Kwegyir-Afful et al. [40] created power station scenarios and simulated fires utilizing virtual reality technology in order to investigate personnel’s ability to detect fires and instigate evacuations while performing their assigned duties. Lovreglio et al. [43] investigated the effects of virtual reality gunfire exercises on people’s experience, revealing that VR is superior to traditional emergency drills. Gong et al. [44] proposed a SIGVerse-based virtual reality-based earthquake exercise simulation system. The outcomes demonstrated that the system could effectively simulate earthquakes and be utilized in earthquake exercises. Chen et al. [45] used virtual reality to assess the effect of crowd movement on personnel escape behavior in a subway station fire emergency. Nevertheless, previous studies did not use IVR to access the EEW message. We simulated an earthquake in virtual reality to examine the capabilities of the ICL EEW message.

4. Experimental Design

4.1. Software

A game engine is a software framework that includes 2D or 3D image rendering units, physics algorithms, audio systems, scripts, actions, artificial intelligence, and so on [46]. There are numerous game engines available for the creation of immersive virtual reality, including Unity 3D, Unreal Engine 5, and others. Unreal Engine 5 was chosen for building the simulated environment for this investigation because it is the most sophisticated and accessible real-time 3D creation tool in the world. It contains Nanite, a sophisticated polygonal graphics technology, and Lumen, a technology for building an illumination environment. UE5 is extensively utilized in the video game, advertising, and film industries [47]. Individual users can obtain and utilize UE5′s full functionality for free, making it popular among individual developers. Additionally, UE5 is compatible with Oculus Quest 2, HTC VIVE, and other VR devices, which provides VR software developers with great convenience. The engine’s functions can be implemented in C++ and blueprints, allowing developers to design procedures to satisfy the experiment’s requirements, controlling the implementation of changing circumstances and the user’s engagement with the surroundings. Figure 3 depicts the experiment scenario creation process.

4.2. Virtual Environment

This research utilized a 3D scene model to create a virtual environment and simulate an earthquake scenario on VR devices. It is a standard apartment with a living room, bedroom, kitchen, and bathroom. Well before investigation, the participants were told that this place would be their residence in the future. The model was imported into UE5 using the methods outlined in Figure 3. The location was furnished with items such as a television, bookshelf, lamp, etc. The EEW message was broadcast on television as part of the initiative. Participants were informed that this apartment was located on a high floor and that navigation outside the rooms was unavailable. In addition, light mapping was rendered in real time to enhance the authenticity of the virtual environment. Triggers were also used to record participant coordinates after controller-based movement within the scene.

4.3. Earthquake Simulation

The simulation of shaking in a virtual environment posed a significant challenge to the veracity of the experiment [48]. In this investigation, the tremors are presented both visually and physically. The visual tremors were simulated by compiling blueprints in an indoor Unreal Engine 5 scene. To provide physics effects for the subsequent physical simulation, we initially applied box collisions to each object in the scene. To simulate the earthquake, we added a Z-axis displacement using the set world location function and set a circular timeline for the floor model using the blueprint’s timeline function. At the specified time, the model moved vertically to simulate the vertical swaying caused by the P-wave at the onset of earthquakes. Simultaneously, physics simulation was enabled for the scene’s objects in the level blueprint, producing realistic physics effects as the floor model vibrates. For instance, vases on cabinets collapsed when the cabinets trembled, and chairs on the floor toppled when the floor shook (Figure 4). Several approaches have been used in research to incorporate the physical quivering element into the gaming experience [49,50]. In this study, we have chosen the game controller vibration option because it is the simplest to implement and has no location restrictions. The Unreal Engine 5 VR pawn blueprint now includes a haptic effect in which both controllers vibrate continuously during an earthquake. The participant acted for sixty seconds during the earthquake simulation.

4.4. Earthquake Early Warning Simulation

As indicated in the introduction, the ICL EEW message includes five pieces of information: the countdown time (10 s), the location, the epicenter, the magnitude, and the time of the earthquakes (see Figure 5a). The first step to simulating the ICL EEW in IVR was to construct a video for display in the virtual environment. It consisted of an advertisement and the ICL EEW message’s pop-up window. The advertisement lasted two minutes, with the ICL EEW message appearing one minute after the advertisement began. The FBX model of the television was mounted on the wall and scaled appropriately so that the participants could see the television screen in the VR device with clarity. Then, a cube of the same dimension as the screen in the scene was constructed, and the video was imported into Unreal Engine 5. Attaching media material to the cube allowed the video to be displayed on it (Figure 5b,c). The audio was played upon the occurrence of the alert, maximizing the effectiveness of the ICL EEW.

4.5. VR Locomotion

VR locomotion is the technology that enables movement in a virtual environment. Two categories of virtual locomotion techniques exist: controller-based and motion-based [51]. Motion-based VR locomotion enables users to freely move in the real world while the movement of their feet is traced and converted to action in the virtual environment. However, it is inappropriate for this research due to the danger of participant injury during earthquake simulation. Teleportation is among the more popular ways of moving in current VR games. It is a controller-based locomotion interaction within virtual reality in which the user can press a trigger on one of the controllers to teleport to the location they are gazing at (Figure 6) [52]. Although teleportation is less immersive than walking, it reduces motion sickness and increases the experiment’s success rate.

5. Evaluation

5.1. Hardware

A laptop computer running Windows 10 was used to host the IVR scenario in Unreal Engine version 5.0.3. The laptop computer carried an Nvidia RTX3060 graphics card and an AMD Ryzen 7 processor. The virtual scene was built for use with Oculus Quest 2, a headset with two controllers. The visual output of the VR device is simultaneously displayed on the laptop computer’s monitor. Oculus Quest 2 was connected to the computer via the Oculus Link. In addition, while an earthquake occurs within the IVR scene, the VR controller grips apply vibration feedback to the participants to increase the realism of the experiment (Figure 7).

5.2. Procedure

From 4 to 11 March 2023, the experiment was conducted in the Department of Industrial Design conference room at Beijing Forestry University. Participants were first apprised of the experiment’s procedure by reading the experiment’s information sheet. The participants were separated into two groups: the control group, in which the ICL EEW message was not displayed; and the experimental group, in which the ICL EEW message was displayed ten seconds prior to the earthquake simulation. The participants were randomly assigned to separate groups without prior notice. Then, the participants were asked to sign an informed consent form after perusing the information leaflet and agreeing to the experiment and the use of their data. After completing the consent form, the participants filled out a pre-test questionnaire containing their name, gender, age, and VR device experience. After that, the researchers gave a brief training session on how to use the VR equipment, including how to grasp the VR controllers, modify the VR viewpoint, and navigate in VR. The participants were given a maximum of five minutes to familiarize with the VR system and the layout of the virtual room by openly navigating the virtual environment. The investigation commenced once the participants had mastered the VR apparatus. The participants sat in chairs, donned VR equipment, and completed the investigation while wearing the headsets. The experiment ended once the participant chose a safe area to protect themselves. The participants were then given a post-test questionnaire for data analysis.

5.3. Data Collection

Participants’ activities within the virtual scenario were videoed and saved for behavior analysis. In the virtual reality environment, we compiled the participant-selected avoidance areas. Following the earthquake, we recorded the time it took participants to find the avoidance area. In addition to the data collected through virtual reality, we collected data through a questionnaire given after the VR experiment, including the participants’ recollections of the information and their emotional responses in both groups. The data will be analyzed to determine the functional, cognitive, sensory, and emotional affordances of the ICL EEW message on television.

6. Results

6.1. Participants

Four of the 72 participants who were tested did not finish the process, resulting in a final sample of 68 participants (31 males, 46%; 37 females, 54%). The age distribution of the participants is depicted in

Table 1. Participants’ information.

	n (Total n = 68)	Rate
Sex
Female	37	0.54
Male	31	0.46
Age Range
<18	1	0.01
18–22	53	0.78
22–30	12	0.18
>30	2	0.03

; their ages range from 9 to 45 years, with 78% between the ages of 18 and 22 and the average age being 20.97 years (SD = 4.34). This was due to the experiment’s location on a university campus and the fact that students were more likely to be interested in virtual reality experiments than middle-aged and senior individuals. Recruiting volunteers for the experiment was simpler among students, and 63% were from the northern region of China, while 37% were from the southern region.

Table 2. Participants’ experience with VR devices.

Frequency	n (Total n = 68)	Rate
Never	37	0.55
Once a year	7	0.10
More than once a year	24	0.35

reveals that 55% of respondents had no prior experience with VR devices, 10% used VR once per year, and 36% used VR more than once per year. In addition, 32 people were in the control group, which would not receive the ICL EEW messages during the investigation, while 36 participants were in the experimental group, which would receive the ICL EEW message.

6.2. Effect of the EEW Message on Avoidance Area Choices

To determine the functional affordance of the earthquake warning system, we had to determine whether the ICL EEW message had a significant impact on the participants’ choice of avoidance area. We collected the avoidance location coordinates (x, y) for the control and experimental groups and performed a normality test (

Table 3. Nonparametric test result of coordinates.

	x	y
Without EEW (n = 32)	550.000	880.000
With EEW (n = 36)	540.000	850.000
Mann-Whitney U test	496.000	551.000
Mann-Whitney Z test	−0.780	−0.088
p-value	0.435	0.930

). The results revealed that both x and y were statistically significant (p > 0.05), indicating that the various group samples for x and y were consistent and did not differ significantly.

The coordinates of the two categories of avoidance locations were then classified. A frequency analysis (

Table 4. Participants’ avoidance area choices.

Areas	n (without EEW)	Rate	n (with EEW)	Rate
Showerhead	15	0.47	16	0.44
Near sofa	6	0.19	4	0.11
Bedroom	5	0.16	5	0.14
Balcony	2	0.06	2	0.06
Near table	2	0.06	4	0.11
Wash basin	1	0.03	3	0.08
Others	1	0.03	1	0.03

) revealed that 47% and 44% of the two groups chose the bathroom, which was the most preferable location for avoidance by both groups. In addition, 19% and 11% selected the area near the sofa because of its proximity to the starting point and short avoidance time. The bedroom area was also chosen by some participants, accounting for 16% and 14%, respectively. A small number of participants also chose locations such as the balcony and under the table.

The coordinates of the two groups were marked on the floor plan of the experimental room in order to visualize the data and provide a more intuitive view of the risk avoidance decisions made by two groups. As depicted in Figure 8, the blue points represent the spots chosen by the control group, whereas the red points represent the spots chosen by the experimental group. The graph plainly demonstrates the propensity of the two groups to choose the area of avoidance during the earthquake simulation, with no overall significant differences. The preponderance of avoidance areas in the bathroom was under the shower head. The post-test interview revealed that the majority of participants in both groups believed the bathroom to be a relatively secure location in the event of an earthquake, as it offered both a water source and a solid triangular protection zone. The experiment revealed no significant difference in the risk-avoidance behavior of the participants between the two scenarios with and without EEW. We concluded, based on post-interviews and questionnaires, that the similarity between the two groups in their choice of avoidance location was due to the ICL EEW message’s lack of protective action information. Participants tended to choose to avoid earthquakes on the basis of their prior knowledge of earthquake prevention.

6.3. Recall of Information on the EEW Message

In this section, we investigated the cognitive affordance of the ICL EEW messages via a post-questionnaire that included inquiries about the epicenter, magnitude, and time of the earthquake. Following the experiment, we rated the participants’ knowledge of earthquakes and EEWs from −3 to 3. Cronbach’s alpha was 0.688, indicating that the survey was reliable. Participants generally had a greater understanding of earthquakes than EEW.

Table 5. Participants’ experience with earthquakes, EEWs, and earthquake drills.

Variables	n (Total n = 68)	Rate
Earthquake drills	63	0.93
EEWs	19	0.28
Earthquakes	15	0.22

displays the past experiences of the participants, with 93% having participated in an earthquake drill, 28% having received an EEW message, and only 22% having personally experienced an earthquake.

In the experimental group, a questionnaire was given out immediately after the experiment to collect data on how well participants remembered the message on a scale from −3 to 3. In total, 53% of participants remembered the countdown to the earthquake; 36% remembered nothing about the EEW; 14% remembered the magnitude of the earthquake; 11% remembered the epicenter; and only 2% recalled the time of the earthquake (

Table 6. Participants’ recollection of information on the EEW.

Variables	n (with EEW)	Rate
Countdown time	19	0.53
Don’t remember	13	0.36
Magnitude	5	0.14
Epicenter	4	0.11
Time	2	0.06

). The data indicate that participants required additional effort to read and comprehend the message immediately after receiving it. However, the majority of participants remembered the countdown well, indicating that they subconsciously found it to be the most arresting. Due to their lack of specific knowledge, participants mentioned in the post-interview that the magnitude and epicenter of earthquakes on the EEW message have a very limited impact.

In addition, we examined whether the participants’ knowledge of earthquakes and EEWs affected their ability to recall information after receiving an EEW. The data sets were therefore imported into SPSSAU for correlation analysis. The correlation between participants’ memory capacity, EEW knowledge, and earthquake knowledge was examined using Spearman’s correlation coefficient to determine the strength of the correlation (see

Table 7. Normality test result of knowledge and memory.

Variables	Shapiro-Wilk Test W	p-Value
EEW	0.882	0.001 **
Earthquake	0.852	0.000 **
Memory	0.816	0.000 **

** p < 0.01.

). The correlation coefficient between recollection and EEW is 0.228, which is close to 0, and the p-value is 0.181 > 0.05, indicating that there is no correlation between participants’ knowledge of EEWs and their memory capability of it. The correlation coefficient between memory and earthquake is −0.097, which is close to 0, and the p-value is 0.573 > 0.05, indicating that there is no correlation between participants’ knowledge of earthquakes and their memory regarding it (

Table 8. Spearman test result of knowledge and memory.

	Memory
EEW	0.228
Earthquake	−0.097

).

Due to the absence of EEW in the control group, we determined the participants’ need for EEW message information via a questionnaire.

Table 9. Participants’ expectations of information on the EEW message.

Variables	n (without EEW)	Rate
Action guidance	29	0.91
Countdown time	25	0.78
Time	25	0.78
Magnitude	20	0.63
Epicenter	13	0.41

reveals that, of the five options we provided, 91% of participants wanted the EEW to include information on protective guidance; 78% believed that the countdown and time of the earthquake were necessary; 63% believed that the magnitude of the earthquake would assist them in choosing the appropriate avoidance behavior; and 41% wanted information on the epicenter to be included in the warning. After the experiment, we interviewed participants about the EEW messages. The majority of participants reported that it was difficult to read and comprehend all of the message’s information in a short amount of time, so they wanted the EEW message to convey the most important information effectively.

6.4. Effect of the EEW Message on Timing of Avoidance

To access the sensory affordance, we collected the responses of participants who noticed the ICL EEW message. Firstly, we recorded the avoidance time for the control group and the experimental group separately. In both groups, the duration between the onset of the earthquake and the participants’ arrival at their designated evacuation location was recorded. The data presented in

Table 10. Participants’ time taken to find avoidance places.

	Average ± SD
Without EEW	24.040 ± 7.195 s
With EEW	21.836 ± 10.451 s

indicated that the earthquake warning shortened the participants’ avoidance time after the tremor, from 24.040 ± 7.195 s in the control group to 21.836 ± 10.451 s in the experimental group. The EEW message for the experimental group appeared 10 s prior to the seismic simulation, giving the experimental group participants 10 s more to react than the control group participants. However, many participants required several seconds to respond after receiving the ICL EEW message. Furthermore, 16 participants (44%) in the experimental group did not take immediate action to avoid the warning, according to

Table 11. Participants’ reasons for not taking actions.

Variables	n (with EEW)	Rate
I didn’t know what to do	9	0.28
I didn’t have enough time to think	8	0.25
I couldn’t react because of fear	4	0.09
I thought I was safe	4	0.09
I thought the magnitude was not high	2	0.06
I didn’t believe EEWs	0	0

. The purpose of the post-test questionnaire was to ascertain why they did not react quickly. According to the graph, 28% of participants did not know what to do, and 25% said they did not have enough time to consider their options. These two factors were the primary reasons why they were unable to act immediately. This result is consistent with the conclusion reached in Section 5.3, namely that participants had the greatest need for instructions on protective actions. It suggests that the cognitive affordance of the ICL EEW message to facilitate immediate risk avoidance responses must be enhanced.

6.5. Analysis on Emotion Scores in Simulation

In this section, we collected the participants’ emotions during the simulated earthquake using a post-test questionnaire, the details of which are outlined in the following table. On a scale extending from −3 to 3, we asked participants to rate their post-experiment self-emotions during the earthquake as scared, astonished, nervous, anxious, uncomfortable, excited, delighted, and relaxed. We utilized the S-W test because our data sample size of each group was less than 50. The test results are displayed in

Table 12. Normality test result of emotion scores.

Variables	Shapiro-Wilk Test W	p-Value
Scared	0.902	0.007 **
Scared (with EEW)	0.917	0.010 *
Astonished	0.855	0.001 **
Astonished (with EEW)	0.915	0.009 **
Nervous	0.870	0.001 **
Nervous (with EEW)	0.766	0.000 **
Anxious	0.909	0.011 *
Anxious (with EEW)	0.820	0.000 **
Uncomfortable	0.877	0.002 **
Uncomfortable (with EEW)	0.879	0.001 **
Excited	0.932	0.046 *
Excited (with EEW)	0.916	0.010 **
Delighted	0.894	0.004**
Delighted (with EEW)	0.934	0.032 *
Relaxed	0.929	0.038 *
Relaxed (with EEW)	0.931	0.028 *

* p < 0.05, ** p < 0.01.

, where it was determined that both data categories were significant (p < 0.05) and, therefore, lacked normality. For data that lacked normality and could not be analyzed using the t-test, we merged the two data categories into a single table. We conducted a non-parametric test on the two data groups using the group as the independent variable. The results indicate that the p-value for astonished and anxious is less than 0.05, indicating that there is a statistically significant difference between the two data groups for these two emotions.

The non-parametric test was used to examine the differences between groups for the eight variables of scared, astonished, nervous, anxious, uncomfortable, excited, delighted, and relaxed, as shown in

Table 13. Nonparametric test result of emotion scores.

Variables	Without EEW	With EEW	Mann-Whitney U	Mann-Whitney Z	p-Value
scared	1.000 (1.0, 2.0)	1.000 (0.3, 2.0)	549.000	−0.344	0.731
astonished	2.000 (1.0, 3.0)	1.000 (1.0, 2.0)	421.000	−1.990	0.047 *
nervous	2.000 (1.0, 3.0)	3.000 (1.3, 3.0)	428.500	−1.910	0.056
anxious	1.000 (1.0, 2.0)	2.000 (1.0, 3.0)	407.000	−2.146	0.032 *
uncomfortable	1.000 (0.0, 2.0)	1.000 (−0.8, 2.0)	539.000	−0.469	0.639
excited	0.000 (−1.0, 1.0)	0.000 (−1.0, 1.0)	522.500	−0.671	0.502
delighted	−1.000 (−2.0, 0.0)	−1.000 (−2.0, 0.0)	570.500	−0.069	0.945
relaxed	−0.500 (−2.0, 0.8)	−1.000 (−2.0, 0.0)	512.500	−0.799	0.424

* p < 0.05.

. The different group samples did not demonstrate difference (p > 0.05) for the six variables of scared, nervous, excited, and delighted. The samples from the other groups demonstrated difference (p < 0.05) for astonished and anxious.

In the sensitivity analyses, we adjusted the method in analysis of emotion scores. Paired Samples Wilcoxon test was used to study the variability of the two data groups. As shown in

Table 14. Paired Samples Wilcoxon test result of emotion scores.

Paired Groups	Median M Difference	z-Value	p-Value
scared	0.000	0.145	0.885
astonished	1.000	2.009	0.045 *
nervous	−1.000	1.917	0.055
anxious	−1.000	2.319	0.020 *
uncomfortable	0.000	0.143	0.887
excited	0.000	0.722	0.470
delighted	0.000	0.309	0.757
relaxed	0.500	0.886	0.376

* p < 0.05.

, of the total of 8 sets of paired data, 2 sets of paired data showed a difference (p < 0.05). Specifically, there was a 0.05 level of significance between two “astonished” groups (p = 0.045 < 0.05), as shown by the 0.05 level of significance between two “anxious” groups (p = 0.020 < 0.05) and the specific comparison differences. The result did not change materially.

We presented the data as a box plot to provide a more straightforward visual representation of the difference in mood scores between the two groups (Figure 9). The graph demonstrates that the mean anxious score was substantially higher when ICL EEW was present compared to when EEW was absent. The average score for astonished was significantly lower than the equivalent score without the ICL EEW. There were no additional significant distinctions between the remaining emotional scores. This indicates that the presence of the ICL EEW message caused stress and anxiety while diminishing the participants’ sentiments of surprise. In the post-test interview, the majority of participants reported feeling panicked and overwhelmed by the EEW message, primarily due to the countdown and alert. The participants had to read and comprehend the message in a short time, which made them anxious. In addition, they reported that the ICL EEW message allowed them to mentally prepare for the earthquake. Consequently, their astonishment was mitigated when the simulated earthquake occurred.

7. Discussion

This study used affordance theory to investigate the effectiveness of EEW messages from a new perspective. Functional affordance, cognitive affordance, sensory affordance, and emotional affordance formed the basis for our evaluations. We collected data from an earthquake simulation in IVR. This work broadens the current research methods on EEW messages. Combining the affordance-based method with immersive virtual reality yields guidelines for future research on the design of emergency messages. This study’s main purpose was to evaluate the usefulness of the ICL EEW message and bring attention to the design of EEW messages in China. The results may indicate the need for additional research in this area to enhance EEW messages.

Few researchers had previously attempted to simulate the EEW in virtual reality, which posed one of the greatest obstacles for this endeavor. Attention, behaviors, and perceptions of EEW messages displayed on a computer monitor were analyzed by Sutton et al. [15]. The fact that participants did not encounter any threatening conditions was a major limitation. In order to provide a more authentic scenario, we followed their research and simulated an earthquake in IVR. In such a circumstance, the reliability of participant responses to VR events increases [53].

This investigation has uncovered a number of findings. First, the analysis of the avoidance choices of two groups of participants revealed that the ICL EEW message had a negligible effect on the location where people chose to avoid peril during earthquakes. Without an indication of protection actions in EEW messages, participants typically relied on past earthquake exercise experiences and earthquake knowledge to determine where to seek safety and how to protect themselves. This indicates that the current ICL EEW message’s functional affordance is insignificant. Other EEW messages, such as ShakeAlert, provide direct instructions for protective behaviors like “Drop, cover, and hold on!”. The majority of participants indicated in the post-questionnaire that they expected to see instructions for protective actions in the first place, as this would allow them to protect themselves more swiftly. After considering the physical impacts of disasters, the science underlying those consequences, and the scope of the possible serious repercussions, Wood et al. [54] concluded that the emergency system should focus on the steps individuals should take to become better prepared. This finding is supported by the outcome of our experiment, which indicates that protective action information is crucial to the functional affordance of EEW messages.

The cognitive affordance of the ICL EEW message must also be enhanced, according to a significant finding. Allen et al. [12] argued that providing information regarding shaking intensity or duration is neither necessary nor preferable. In this study, we discovered that participants’ impressions of the information on the message were primarily focused on the earthquake countdown. Nevertheless, few participants had an idea of the earthquake’s magnitude, epicenter, or exact time. The majority of participants required only a few seconds to comprehend the text on the warnings. Still, participants had difficulty comprehending and remembering the message, and they even waited to respond in order to digest the information. The results may demonstrate that consumers do not desire earthquake-related information. It may even have a negative impact on users who take immediate action. To enhance the cognitive affordance of the ICL EEW message, it is necessary to eliminate superfluous information.

In addition, data collected after the experiment and presented in Section 6.3 revealed that although the average time required to locate the warning was shorter for participants in the experimental group than for participants in the control group, the majority of participants required a few seconds to read the message and act until the countdown ended. The majority of participants stated in post-interviews that they would prefer a more engaging and straightforward presentation of EEW messages, such as a larger font size and a more logical layout. Some participants mentioned that the blue color of the ICL EEW message prevented them from recognizing it and caused them to mistake it for an advertisement pop-up window. The visual decision reaction time was also shown to be considerably affected by color contrast, as indicated by Balakrishnan et al. [55]. To further enhance the sensory affordance of EEW signals, it is recommended that colors be studied in more depth in future research.

There was a statistically significant difference between the “anxious” and “astonished” responses of the experimental and control groups after receiving the ICL EEW message. The experimental group had significantly higher anxious scores while the control group had much lower astonished scores. This discovery suggests that participants were anxious but more psychologically prepared for the earthquake because of the ICL EEW message’s emotional affordance. Cognitive and affective capacities help in decision-making, as discovered by Samanez-Larkin et al. [56]. While EEW messaging may help individuals feel more emotionally prepared to make decisions in the face of an earthquake, further research is needed in this area. In conclusion, the study found that there is a lot of room for improvement in EEW messages. After the trial, almost all of the participants had a favorable impression of the ICL EEW message, saying that it had protected them from harm. Nevertheless, institutions and organizations also need to raise EEW message awareness, develop EEW message standards, and constantly improve the affordance of EEW messages.

Limitations

This research has several limitations. Initially, there is ecological validity. It was stated in the introduction that the disparity between IVR and the natural world is not insignificant. Even in immersive virtual reality, it is impossible to recreate or simulate the visual, auditory, and tactile sensations that a real earthquake brings to individuals. We do not know if the participants in the study would take the steps they indicated in a real-world circumstance because they were only presented with a simulated scenario. The objects had collisions, but there were no other options for interactivity (picking up, carrying, etc.). Even though there were dynamic effects of vibrations within the scenes and vibration feedback from the handles, the earthquake simulation did not completely replicate the physical sensations of an earthquake. Although participants could move freely throughout the virtual environment, not all objects could be interacted with. The ecological validity of the IVR study would be greatly enhanced by the addition of interaction. In future research, it could be considered to increase the authenticity of the earthquakes and the interactivity of the scenarios.

Furthermore, the samples for this research were primarily from university students, and the sample size is not large enough. Future related studies should enlist participants from a larger population, and a larger sample size that is more suitable for regression analysis is required. The authors believe that there will be some variation by age and gender, so future relevant studies will require a more diverse participant pool. Moreover, this research provides only a case analysis of a domestic scenario in an apartment. Future research should include more locations, such as campuses, retail centers, and offices. EEW in specific locations would be more pertinent, and future research may pursue this path by conducting experiments and analyzing crowded locations. With the development of EEW technology, mobile phones are becoming increasingly important for disseminating earthquake warning information [57]. Future research should pay more attention to mobile phones, computers, outdoor displays, and other media to enrich the study of EEW messages.

In addition, this study was limited to an evaluation of the ICL EEW. It identified the deficiencies of the current EEW but has not yet provided specific development suggestions. The authors believe that experimental analysis of specific content, such as color, typography, and the presence or absence of images, can be performed in the future to provide specific solutions for improving EEW messages.

8. Conclusions

Incorporating IVR and affordance theory is an alternative method for conducting EEW message research. This study conducted a virtual seismic scenario with the ICL EEW message in an immersive virtual reality environment. The application of affordance allowed the categorization and logical analysis of participant feedback. After receiving the EEW message on television, the majority of participants took several seconds to check the information. Upon recognizing the message, the majority of participants chose to seek refuge in what they believed to be a secure location. There was no significant difference between the experimental and control groups in the selection of secure locations. The presence of the ICL EEW message significantly increased their anxiety. Few participants had a clear impression and memory of the ICL EEW message after the experiment, and the majority could not recall any of the information. Our findings reveal the decision-making processes of those who have received EEW messages. We discovered that the current ICL EEW message on television is ineffective at spreading earthquake information and influencing people to avoid danger. Nevertheless, it has the potential to substantially affect individuals’ emotions during a natural disaster. These findings may assist future researchers in enhancing and expanding the effectiveness of EEW messages, resulting in better preparedness and response measures, reducing the impact of earthquakes, and saving lives and property.