Under the heading “What to Expect When Visiting the Terra Pavilion,” the official Expo's website (EXPO 2020 DUBAI UAE, 2021b) highlights the following items:

Terra consists of 4,192 mono-crystalline photovoltaic cells mounted on a 130-m wide-angled, oval-shaped smart canopy following the sun-like sunflowers and 18 Energy Trees scattered around the pavilion, as shown in Figure 1. It can generate up to 4 GWh of electricity per year, sufficient to power the energy consumption of 370 average households. Besides being a smart roof system for electricity generation and creating a thermally isolated underground exhibition space, the main canopy also helps collect rainwater and dew at night to produce clean water and recycle graywater.

The Terra display has two immersive tracks for guests to choose from: one exhibits the vivid secret world beneath the forest, as shown in Figure 2, whereas the other shows the fully diversified life beneath the sea, as shown in Figure 3. Following these trails, visitors are confronted with the sad reality that people are harming the lives of other creatures as a result of their actions.The dissolving theater display as shown in Figure 4, for example, depicts how ocean acidification and warming, both of which are the result of human-caused environmental degradation, will harm sea animals. Figure 5 shows the gnasher production machine. The gnasher uses nature as an input and produces items as an output. The gnasher also depicts the enormous amount of goods that are produced but not consumed. After comprehending how human activities degrade two rich ecosystems, the exhibition offers realistic, sustainable solutions from several countries.

The Terra pavilion targeted environmental issues as well as potential remedies. On the other hand, the pavilions of the countries were more concerned with sustainable solutions. Both the pavilions' constructions and exhibitions featured these solutions. This section overviews five country pavilions where sustainable solutions were prioritized. These pavilions were chosen using a peer-to-peer voting procedure. Staff from thirty country pavilions were asked to recommend five pavilions, other than their own, that focused on sustainability. This section gives snapshots of Germany, Italy, Spain, Singapore, and the Netherlands' pavilions.

Although environmental awareness as a construct was first used over 50 years ago, experts have no consensus about measuring it (Fu et al., 2020). However, scholars (e.g., Maloney and Ward, 1973; Dembkowski and Hanmer-Lloyd, 1994) have accepted that environmental awareness is a higher-order construct (HOC). The environmental awareness construct is formed from Cognitive (e.g., knowledge), affective (e.g., attitudes), and conative (e.g., behavioral intentions or willingness) constructs (e.g., Maloney and Ward, 1973; Maloney et al., 1975; Kollmuss and Agyeman, 2002; Zsóka, 2008; Zsóka et al., 2013). Several researchers (Kollmuss and Agyeman, 2002; Zsóka et al., 2013) have also included environmental concern (EC) in the environmental awareness construct, while others considered EC as part of the environmental attitude (Dunlap and Jones, 2002). The questionnaire used in this work assumes a three-dimensional construct to measure environmental awareness similar to that used by earlier researchers. Thus, the questionnaire operationalizes and measures the lower-order constructs (LOCs) of knowledge, attitude, and willingness.

Environmental knowledge affects a person's thinking, values, attitudes, and behavior toward environmental issues (Dembkowski and Hanmer-Lloyd, 1994). Fryxell and Lo (2003) defines environmental knowledge as “knowledge of facts, concepts, and relationships concerning the natural environment and its major ecosystems.” Zsóka et al. (2013) extended this definition such that environmental knowledge is not only about knowing environmental problems but also about knowing the solutions to these problems.

Environmental knowledge can be divided into objective and subjective knowledge (e.g., Maloney et al., 1975; Pagiaslis and Krontalis, 2014). On the one hand, objective knowledge measures the extent of accurate information about the environment and its problems stored in an individual's long-term memory (Vicente-Molina et al., 2013). On the other hand, subjective knowledge measures what or how much a person knows about specific pro-environmental actions (Pagiaslis and Krontalis, 2014). Researchers typically rely on either type of knowledge when measuring environmental Awareness (Vicente-Molina et al., 2013). It is argued in Pagiaslis and Krontalis (2014) and Fu et al. (2020) that measuring subjective environmental knowledge is more suitable for studies that measure the effects of environmental awareness on pro-environmental behavior.

The second LOC that measures environmental awareness is attitude. Attitude generally concerns a person's belief regarding the consequences of undertaking a specific behavior as a function of the person's valuation of the consequences (Ajzen, 1991). According to Milfont and Duckitt (2010), environmental attitude “is a psychological tendency expressed by evaluating the natural environment with some degree of favor or disfavor.”

Willingness to act is the third LOC needed to measure environmental awareness. According to Stone et al. (1995), willingness to act defines behavioral intentions that are assumed to capture the motivational factors that influence a given behavior (Ajzen, 1991). Maloney et al. (1975) measures willingness by measuring the verbal commitment to act.

Multiple scholars have investigated the influence of mega-events on environmental perspectives. Hao et al. (2011) and Deng et al. (2014) analyzed the effects of Expo 2010 on the air quality of Shanghai, China. Deng et al. (2014) investigated how effective planning would improve urban life in the city, while Hao et al. (2011) analyzed how Expo 2010 affected the air quality of Shanghai. Regarding Expo 2015 Milan, Guizzardi et al. (2017), and Prayag et al. (2021) investigated the impact of environmental certification on Milan's residents, not the visitors.

Previous research on Expos and environmental concerns focused solely on the environmental effects of the Expos on the host cities, such as urban planning and air quality. These studies did not study how these events have affected their visitors. Thus, it is essential to determine whether or not these mega-events achieve their goals of raising the environmental awareness of visitors.

This paper also documents the tools used to raise environmental awareness. These tools are not limited to exhibitions and showcases but also the pavilions' constructions. The effect of employing construction to raise environmental awareness was not tested before.

Since the Expo had two goals: (i) to shed light on environmental problems and (ii) to demonstrate that there are solutions to these problems, both goals were assessed through the survey questions. So, the survey included two questions about environmental problems and two other questions about solutions to environmental problems when assessing knowledge and attitude (Zsóka et al., 2013). All questions were adopted from the literature, possibly slightly modified, to avoid deviating from the literature, and still serve the purpose of this study. As illustrated in Figure 6, a reflective-formative HOC to assess awareness is used.

Questions EK1-EK4 in Table 1 are comparable to those in Pagiaslis and Krontalis (2014). Pagiaslis and Krontalis (2014) focused on biofuels; hence, they evaluated the consequences of pollution and how biofuels might resolve these issues. However, this study is not concentrating on a single environmental problem and its solution; hence, the questions adopted from Pagiaslis and Krontalis (2014) were modified to make them more suitable to the information presented in the key pavilions. As previously indicated, questions EK1 and EK2 test the environmental knowledge of visitors regarding environmental problems, whereas questions EK3 and K4 assess their environmental knowledge regarding solutions to these problems.

Dunlap et al. (2000) have created a set of environmental attitude-measuring questions, the revised new ecological paradigm (NEP-R) (Milfont and Duckitt, 2010), from which four questions are selected. Questions EA1 and EA2 in Table 1 inquire about the severity of environmental problems. The remaining two questions, EA3 and EA4, address potential fixes to the challenges. The verbal commitment subscale created by Maloney et al. (1975) examines what a person is willing to do about environmental issues, such as sorting rubbish or using public transportation. Questions EW1–EW4 in Table 1 assess willingness to follow the guidelines in Maloney et al. (1975). Moreover, these issues are compatible with what the respondents have experienced in the Expo.

Before collecting data, the questionnaire was validated using the procedures described by Bang et al. (2000). Seven professionals in the field assessed the questionnaire, one of whom has peer-reviewed articles on environmental awareness assessments. Face validation of the questionnaire and recommendations for question revisions were solicited. The questionnaire was then administered to twenty individuals who were questioned about the questions' clarity and ambiguity. Before sending the questionnaire to the respondents, questions were adjusted based on the inputs received from experts and the pilot test. The replies were measured using a five-point Likert scale, with responses matching the question type.

The LOCs in our model are reflective because the used indicators are manifestations of these constructs; however, environmental awareness is a formative construct because each LOC defines one of its facets. This distinction in causation between reflective and formative notions requires distinct methods for assessing their reliability and validity (e.g., Hair et al., 2011, 2012; Sarstedt et al., 2019).

The values of Cronbach's alpha(α), composite reliability (CR), and average variance extracted (AVE) are presented in Table 2 for the LOCs before and after attending the Expo. Cronbach(α) and CR values over 0.7, according to Cronbach (1951) and Hair et al. (2011), respectively, indicate internal consistency. Table 2 demonstrates that all LOCs surpass these threshold values in surveys done before and following the Expo. In addition to internal consistency, indicator reliability is approved if the indicator loading associated with each question exceeds 0.6, as discussed by Chin (1998) and Henseler et al. (2009). Before and after the Expo, the lowest indicator loading in both models was 0.646%. Thus, the LOCs are deemed reliable.

AVE must be more than 0.5 to establish convergent validity, as mentioned in Fornell and Larcker (1981). Table 2 demonstrates that all LOCs have met this criterion. According to Fornell and Larcker (1981), the Fornell–Larcker criterion for evaluating discriminant validity demands that the AVE of each concept be greater than the squared correlations with all other constructs. This relation holds for all LOCs in both models.

In our approach, the LOCs serve as environmental awareness indicators, a formative HOC. Petter et al. (2007) suggest that examining the reliability of formative constructs is not essential because each dimension assesses a different aspect of the construct. Consequently, the significance of each LOC on the HOC and the relationship between LOCs are evaluated (Henseler et al., 2009). The relevance of each LOC is determined by comparing the weights of the indicators to a threshold value of 0.1 (Andreev et al., 2009), where a bootstrapping approach determines the weights (Henseler et al., 2009; Hair et al., 2011). Table 3 demonstrates that all weights surpass 0.1 and are statistically significant based on the P-values associated with each LOC. Table 3 also displays variance inflation factors (VIFs) employed to quantify the degree of multicollinearity between the indicators (Hair et al., 2011). According to Diamantopoulos and Siguaw (2006) and Hair et al. (2011), indicators are deemed independent if their VIF values are less than 5 or 3.3, respectively. In our models, the VIF values are less than 3.3.

Each research question is associated with an appropriate hypothesis. Conclusions made from testing these hypotheses answer RQ1 and RQ2. Since measures are taken before and after visiting the Expo for the same individuals, paired tests are used to examine the differences in environmental awareness before and after the Expo. Wilcoxon-signed-rank test (Taheri and Hesamian, 2013) is used to answer RQ1, which is associated with Hypothesis 1, where μ~Before and μ~After are the population medians of environmental awareness before and after the Expo, respectively. The Wilcoxon-signed-rank test is non-parametric and is more appropriate for hypothesis resulting from Likert-scale data, which could violate the assumptions of normality needed to use parametric testing (Carifio and Perla, 2008; De Winter and Dodou, 2010).

Hypothesis 1:

In addition to testing environmental awareness before and after the Expo, the three LOCs are also tested, as shown in Hypothesis 2. In this hypothesis, q refers to one of the three LOCs components used to measure environmental awareness.

Hypothesis 2:

Using a confidence level of 0.01, i.e., α = 0.01, the four null hypotheses are rejected because the P-values for all the tests are less than 0.01. Thus, it is concluded that environmental awareness and all its dimensions were raised after visiting the Expo.

RQ2 is about identifying the environmental awareness component that was most increased because of Expo attendance. As shown in Hypotheses 3A-5B, six tests are conducted to rank the environmental awareness components based on how much they were raised. The null hypotheses assume the equality of the median changes, while the alternative, one-sided hypotheses check which of the tested components has been raised more. The Δμ^component represents the change in medians of the environmental awareness of the tested component. By component, we mean the three LOCs of knowledge, attitude, and willingness.

Hypothesis 3A:

Hypothesis 3B:

Hypothesis 4A:

Hypothesis 4B:

Hypothesis 5A:

Hypothesis 5B:

The P-value of hypothesis 3A is almost 1, while it is < 0.01 for hypothesis 3B. Thus, it is concluded that environmental attitude has been raised more than environmental knowledge. Hypotheses 4A and 4B conclusions show that environmental knowledge has increased more than environmental willingness, while Hypotheses 5A and 5B conclusions show that environmental attitude has increased more than environmental willingness. Consequently, the results of these three hypotheses show that environmental attitude was raised the most, followed by environmental knowledge and environmental willingness.

Table 4 highlights the respondents' demographic information. It also provides information regarding the number of pavilions visited by respondents and statistics regarding the Terra pavilion visits. Table 4 shows the gender neutrality of our sample since 52% of respondents are male, and 48% are female. However, the sample was biased in terms of age and education. Most respondents are 18–35 years old and hold bachelor's degrees. The Terra pavilion was visited by 84% of the survey respondents. Respondents are clustered into various groups based on the criteria indicated in column 1 of Table 4. Last, respondents came from 84 different counties; consequently, nationality is excluded from any further analysis.

Table 4 displays the results of five hypotheses tests to determine whether Expo attendance had an equivalent impact on all groups within each criterion. The tested hypotheses are shown in column 5 of Table 4, while the p-values of each test are shown in column 6 of Table 4. The Kruskal–Wallis test, the non-parametric version of the analysis of variance (ANOVA) test, is used. It is evident from the analysis that gender did not influence how much environmental awareness was raised. Similarly, the frequency of trips to the country pavilion did not affect environmental awareness, as seen by the p-value of the last hypothesis. The reason behind the high p-value of the last hypothesis is: (1) the majority of visitors, 84%, have visited Terra, and (2) all respondents have visited at least one of the listed pavilions.

Visiting Terra undoubtedly increased visitors' awareness since the fourth hypothesis in Table 4 was rejected. This significant outcome demonstrates the need for a dedicated pavilion to highlight environmental issues and their solutions. Furthermore, different age groups were affected differently regarding their environmental awareness. Young visitors' environmental awareness increased significantly, whereas middle-aged visitors' environmental awareness increased the least. Lastly, education affected the visitors' environmental awareness. Visitors with advanced degrees experienced the greatest increase in environmental awareness, followed by those with high school education.