The HBM was first proposed in the early 1950's by social psychologists (21), and has since been widely used in the health behavior industry to better understand health education and interventions (22). This model recognizes that personal health beliefs and the effects of those beliefs on attitudes toward preventive activity may be the first in a series of events leading to health promotion. Based on this, health educators can improve their risk communication based on a solid understanding of the psychological mechanisms (23). Though the HBM was created to understand patient practices in relation to specific diseases or their willingness to have early checkups for these diseases, this study proposes that the model can be used to explain the safety behavior associated with online MICE because this behavior can be viewed as a way to prevent or reduce the probability of contracting a disease (24, 25). According to the HBM, health beliefs play a significant role in preventive health behavior, and the ways of knowing and acting are founded on subjective schemata (26, 27). According to the HBM, perception variables such as the perceived safety threat, outcome expectations (a composite score based on perceived barriers and advantages), and self-efficacy can predict health or protection behavior (28). These beliefs are thought to be part of the cognitive mediation process (29). However, little research has been undertaken to investigate the effects of health attitudes on health risk avoidance behavior in the MICE context. This is despite the fact that there is undeniable proof that travel and tourism can hasten the spread of infection (30) and that ignoring the importance of protective health habits can lead to new outbreaks in local communities (31).

The IDT was created to explain why people choose to accept or reject a new technology based on their beliefs (32). While online meetings are not new, they are being utilized more frequently now than in the past, and to individuals participating in them for the first time, they often appear to be new technology (33). According to the IDT, comparative advantages, compatibility, complexity, observability, and trialability are all factors that influence whether or not an innovation is adopted (34, 35). A comparative advantage represents the superiority of a current innovation to similar previous innovations, while compatibility is the degree to which an invention is believed or perceived to be consistent with the adopter's values, needs, and previous experiences (36). Complexity is the degree to which an innovation is viewed or regarded as difficult to use (34), while observability is the degree to which the outcomes of the innovation are evident and trialability is the degree to which an innovation can be implemented in stages (37). The most widely identified explanations for a consumer's intention to adopt new technology have been comparative advantages and trialability (38). Therefore, this study focuses on examining the influence of these two factors on attendees' behavioral intention to participate in online MICE events.

Based on these two theories, this study offers a modified theoretical model that incorporates the HBM and IDT to examine the antecedents influencing attendees' behavioral intention to attend online MICE. This model is presented in Figure 1.

This study used an anonymous cross-sectional survey of current users (adopters) and non-users (non-adopters) of online MICE technology to test the hypotheses. The information obtained by cross-sectional survey is collected at a given point in time (61). The ‘point in time’ that data was collected in this study was during the COVID-19 pandemic. The non-adopters included in the survey are considered potential users of the online MICE technology.

The survey was conducted in China because exhibitors increasingly prefer to hold events in Asia–Pacific countries such as China, rather than the traditionally strong markets in Europe and the United States (10). By consulting experts and scholars in the MICE field, we compiled a questionnaire for online MICE technology.

The survey consisted of three sections. In order to ensure that all respondents (especially non-adopters) were fully aware of how online MICE operated, Section 1 introduced the working mechanisms and main features of online MICE technology; this description was illustrated with relevant images. Section 2 asked questions about the perceptions, attitudes, and intention to adopt online MICE technology, while Sect ion 3 collected demographic information about the respondents, including their gender, age, education, and income.

To collect a representative sample, this study conducted the online survey using Wenjuanxing, a professional data science company (62). To ensure that the participants were motivated to complete the survey, they were told they would receive a small digital gift from the researchers at the end of the survey. From May 28 to June 19, the link for the questionnaire was posted to the WeChat group of company employees and the WeChat group of MICE attendees and 505 responses were received. After eliminating surveys in which the screening question was answered incorrectly (a question that asked the respondent to answer “strongly disagree” was added in the second section) or that were not sufficiently complete, 439 valid questionnaires were used for subsequent analysis.

The survey contained 26 items designed to measure seven variables: perceived safety threat, comparative advantages, trialability, outcome expectations, attitude, cues to action, and behavioral intention (Table 1). Each variable was measured using items from the following sources: perceived safety threat from Yuen et al. (31), comparative advantages from Wang et al. (45) and Shih and Fang (47), trialability from Wang et al. (45), outcome expectations from Yuen et al. (32), cues to action from (63), attitudes from Shih and Fang (47) and (63), and behavioral intention from Agag and El-Masry (35). Many studies have shown that seven-point Likert scale show higher reliability than any other number of options (64, 65). The items were evaluated using a seven-point Likert scale ranging from 1 (“strongly disagree”) to 7 (“strongly agree”).

In Table 2, the proportion of male (243) and female (196) respondents was 55.4 and 44.6%, respectively. A total of 8 (1.8%) respondents were under 19 years old, 116 (26.4%) were 20–29 years old, 164 (37.4%) were 30–39 years old, 121 (27.6%) were 40–49 years old, and 30 (6.8%) were over 50 years old. In terms of education, 188 (42.8%) graduated from a university or college, 51 (11.6%) had studied at graduate school, and 168 (38.3%) had graduated from high school. The monthly average income had the following distribution: 126 respondents earning 0–4,999 yuan (40.3%), 170 earning 5,000–9,999 yuan (38.7%), 52 earning 10,000–14,999 yuan (11.8%), and 40 earning over 15,000 yuan (9.1%).

A confirmatory factor analysis was conducted to evaluate the overall model fit and the reliability and validity of the scales. The goodness-of-fit index (GFI), adjusted goodness-of-fit index (AGFI), comparative fit index (CFI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR) were selected to evaluate the fit of the model (Table 3) (66). The ratio of the Chi-square value to the degrees of freedom (χ²/df) was 1.460 (p < 0.05). The model fit indices (CFI = 0.977, TLI = 0.973, RMSEA = 0.032, and SRMR = 0.035) all passed their respective minimum cut-off points proposed by Hu and Bentler (67). Given the number of indicators, the overall quality of the measures was supported by the statistics (68).

This study uses three criteria to assess the convergent validity. First, the standardized path loading (λ) should be statistically significant and larger than 0.70 (69). Second, the composite reliability (CR) of each construct must be higher than 0.70. Third, the average variance extracted (AVE) for each construct should exceed 0.50 (70, 71). As presented in Table 3, the standardized path loadings were all significant and higher than 0.70. Moreover, the CR exceeded 0.79 for all constructs, and the AVE for each construct was >0.54. Therefore, the convergent validity for the constructs was supported.

By comparing the square root of the AVE for each construct with the correlations between the target construct and other constructs, the discriminant validity of the measurement model was determined. Discriminant validity was assumed if the square root of the AVE was larger than the correlations between the target construct and other constructs (31). The square root of the AVE for each construct exceeded the correlations between the target construct and the other constructs (Table 4), confirming the discriminant validity.

The hypotheses proposed in this paper were tested using a structural equation model (SEM). SEMs are often used to assess how well the structure of the proposed model or the construction of the hypotheses is explained by the collected data (72). The effect of the control variables (age, education, and income) on attendees' behavioral intention to adopt online MICE technology was determined. The reliability of experimental data was determined by a p < 0.05 (73). The hypotheses were tested based on the significance of the construct and the correlation of the standardized estimates. In addition, squared multiple correlations (R²) were calculated to determine the latent variables explained by the percentage variance. The results are presented in Figure 2.

Figure 2 shows that the fit of the structural model was sufficient (χ²/df = 1.726, CFI = 0.953; TLI = 0.947; RMSEA = 0.041; SRMR = 0.069). Significant positive relationships were found between attitude and the perceived safety threat, comparative advantage, trialability, and outcome expectations (0.337, 0.104, 0.104, and 0.231, respectively, p < 0.05). Figure 2 also shows that outcome expectations had a positive effect on cues to action (β = 0.527, p < 0.05). Therefore, H1 to H5 were all supported. Attendees expressed a more positive attitude toward online conference technology when they perceived greater benefits than risks. According to the HBM, the perceived safety threat reflects attendees' perception of the risks associated with attending online MICE events during the COVID-19 pandemic, whereas the outcome expectations suggest that attendees perceive net positive functional and economic benefits. The perceived comparative advantages and trialability of online MICE technology was related to attendees' attitudes toward online exhibitions, with attendees who recognize the comparative advantages of attending online MICE events and believe that online exhibition technology is feasible are more likely to hold a positive attitude toward online MICE. The positive attitudes of attendees toward online MICE technology and the perceived comparative advantages together encourage attendees to adopt online MICE technology intention instead of traditional offline MICE technology. In addition, attendee expectations about an online MICE event can influence their cues to action.

The perceived safety threat, comparative advantage, trialability, and outcome expectations explained 59.6% of the variance in attendee attitude (R² = 0.596), and the outcome expectations explained 27.8% of the variance in cues to action (R² = 0.278). According to the classification of R² values by Chin et al. (74), perceived safety threat, comparative advantage, trialability, and outcome expectations had a high explanatory power for attendees' attitudes, while outcome expectation had a medium explanatory power for cues to action. These results suggest that attendees had a more positive attitude if they perceived a lower safety threat, a greater comparative advantage, high trialability, positive outcome expectations, and more positive cues to action.

The significantly positive effect of cues to action on behavioral intention (β = 0.451, p < 0.05) supported H6. Additionally, the correlation between cues to action and attendee attitude was 0.165 (p < 0.05), while attitude had a significant impact on behavioral intention (β = 0.377, p < 0.05); therefore, H7 and H8 were accepted. Attendees receive information about the health risks of an outbreak and recognize the benefits of following positive leads, increasing their readiness to engage in a health behavior (54). Therefore, cues to action are required for attendees to form a positive attitude toward the use of online MICE technology. In this study, cues to action and attitude were both positively correlated with the attendees' behavioral intention of online MICE (R² = 0.503), which is consistent with the HBM. This suggests that attendees with a positive attitude are more likely to employ online MICE technology during or after the COVID-19 pandemic. This is consistent with previous studies (7, 75). Cues to action are events that guide exhibitors to exhibit safe behavior. These cues may be internal, positive, or negative experiences of exhibiting or not exhibiting safe behavior. They may also be external cues from colleagues and the media. The results showed that attendees who were given more frequent cues were more likely to engage in safe behaviors.

To test the mediating influence of attitude on perceived safety threat, comparative advantage, trialability, and outcome expectations, behavioral intention was examined. The mediating influence of attitude on cues to action and behavioral intention was validated, and separate analyses were performed using Baron and Kenny's (76) approach. The impact of the exogenous variables on the endogenous variables was examined (Table 5). In the theoretical model presented in Figure 2, attitude fully mediates intention and the perceived safety threat, comparative advantage, trialability, and outcome expectations, while attitude partially mediates cues to action and behavioral intention.

In terms of the direct effects, outcome expectations had a direct effect on cues to action (a41 = 0.527). The main predictor of attitude was perceived safety threat (a12 = 0.337), followed by outcome expectations (a42 = 0.231), trialability (a32 = 0.201), cues to action (a52 = 0.165), and the comparative advantage (a22 = 0.104). Finally, cues to action had a greater direct impact on behavioral intention (a53 = 0.451) than did attitudes (a63 = 0.377).

In terms of the indirect effects, outcome expectations were the only direct exogenous variable for attitude (b42 = 0.087). The indirect influence of outcome expectations on attitude was transmitted through a single intermediary, cues to action. In addition, outcome expectations had the greatest indirect effect on behavioral intention (b43 = 0.358), followed by perceived safety threat (b13 = 0.127), trialability (b33 = 0.076), cues to action (b53 = 0.062), and the comparative advantage (b23 = 0.039). Thus, the indirect effects of the perceived security threat, comparative advantage, trialability, and outcome expectations on behavioral intention were mediated by attitude as a single mediator.

It was found that cues to action had the greatest total effect on behavioral intention (c53 = 0.513), followed by attitude (c63 = 0.377), which was traced to its direct effect on behavioral intention (a63), outcome expectation (c43 = 0.358), perceived safety threat (c13 = 0.127), trialability (c33 = 0.076), and comparative advantage (c23 = 0.039).

Finally, no statistically significant link was found between the control variables (age, education, and income) and behavioral intention, even though previous studies have shown that higher income levels, higher education levels, and people under the age of 55 have significantly higher levels of new technology and Internet access (77, 78). Regardless, this data supports the theory that theoretical concepts are more accurate predictors of attendees' behavioral intention to attend online MICE events than are demographic characteristics.

This study makes a number of implications to existing theories. First, this study enriches the research literature on the online MICE industry in the context of COVID-19 and other public health crises. Since the end of 2019, the widespread COVID-19 pandemic has forced the convention and exhibition industry to integrate resources and has shifted offline conventions and exhibitions online, which represents a developmental opportunity for this industry. Since the emergence of COVID-19, the most of study's on the MICE industry has focused on marketing theories that explain the subsequent changes and developmental trends within the industry, while rarely analyzing consumer psychology empirically. Therefore, this research acts as an important reference for theoretical research of online MICE events.

Second, this study combines the HBM and IDT to develop a new theoretical model to understand the reasons why consumers use online MICE technology during a public health crisis. Through empirical analysis, this research found that, during COVID-19, the perceived threat and outcome expectations felt by consumers directly affected the intention to use online technology, which is also an important reason for the development of the online MICE industry. In addition, the comparative advantages and trialability of online MICE technology had an impact on consumer intentions to use the technology. This model combines consumer perception and industry characteristics to provide a comprehensive analysis of online MICE.

Third, this study enriches the HBM and IDT. Attitude and cues to action were used as intermediary variables to evaluate their influence on behavioral intention. Cues to action is an intermediary variable in the HBM, and it was verified that it plays an intermediary role between outcome expectations and behavioral intention. At the same time, attitude was proposed as an intermediary variable in the HBM and the IDT.

The findings of this study provide practical implications for MICE organizers and exhibitors in terms of enhancing their attendees' behavioral intention to attend online MICE events. First, the results indicate the need to clearly recognize that the COVID-19 pandemic poses a threat to the safety of attendees and affects how they choose to attend events. In order to overcome the security threat posed by the epidemic and to improve the outcome expectations of the attendees, MICE companies should actively innovate their exhibition strategies and extensively use new online exhibition platforms.

Second, in order to improve the comparative advantage of online exhibitions and to encourage the acceptance of online MICE technology, a digital exhibition mode is required. For example, using AR, VR, 5G technology, cloud computing, and AI technology to simulate face-to-face communication and produce targeted smart exhibition galleries could increase product sales. In addition, exhibitors can use live online broadcasts and short videos to release and promote new products online and can offer online negotiation, procurement, and docking services for exhibitors, which would not only reduce costs for exhibitors but also enable domestic and foreign companies to engage in business transactions during the COVID-19 pandemic.

Third, a positive attitude toward online MICE services will encourage attendees to participate in online MICE events. The implementation of various marketing and publicity activities on social platforms via community marketing can enhance the emotional value for customers and create a pleasant experience. This target customer may become transmitters, facilitating community fission marketing.

Finally, given the significant intermediary impact of cues to action, external cues to action such as media campaigns, alerts from health workers, and advice from others can encourage people to behave safely during the COVID-19 pandemic. The most effective way to create online MICE events is to promote positive word-of-mouth among attendees. Therefore, exhibitors should take advantage of the technical advantages of social media to segment attendee groups with different characteristics so that customers can better understand MICE brand services and that word-of-mouth marketing among attendees can be achieved.

This study has several limitations that should be considered when interpreting the results. First, this study analyzes the factors that affect attendees' behavioral intention of online MICE technology using the HBM and IDT, thus it does not consider other theoretical lenses. Future studies may thus consider incorporating the technology acceptance theory (35), planned behavior theory (79), and trust theory (32) in the model development process and to analyze the intention to attend online MICE events. With COVID-19 prevention and control in China showing a positive trend, cities across China are gradually resuming work and production, and exhibition organizers, venues, and exhibitors are also preparing for the event. Many exhibitors responded to the call for national policies and actively carried out online MICE events, which provided a good example for the digitalization of the global MICE industry. However, this study was conducted solely on a population of Chinese consumers, so the results of this study may not apply to the populations of other countries.