We invited 39 students from the University of Tokyo to participate in the experiment including pre-and after- experiment session. Out of the 39 students, 30 were Mainland Chinese (76.9%), 3 were Taiwanese (7.7%), 4 were Indonesians (10.3%), and 2 were Japanese (5.1%). The participants included 24 males (61.5%) and 15 females (38.5%). Their ages ranged from 20 to 30 (92.3%) years, and only 3 participants were over the age of 30 (7.7%). The experiments were conducted in 6 parts and lasted from 40 to 50 minutes. The participants were asked to participate in the survey to evaluate their satisfaction (subjective utility) as tourists in the virtual tourist attractions. They were paid 1500 yen for their participation. We also asked them to state their personal preferences about the different characteristics of tourism. Most participants reported frequent previous travel experience. For example, 17 out of the 39 participants said that they are very experienced travelers, in or out of Japan (43.6%), while 18 (46.2%) said that they frequently traveled in Japan. The remaining 4 (10.4%) reported that they were not frequent travelers.
Having four destinations/attractions in each experiment resulted in a 4 × 3 × 2 combination, which generated 24 different possibilities of visiting or reviewing sequence in the experiments at both levels. Each participant was assigned to one of the sequences at each level, and each sequence was followed by at least one participant. The participants finished the experiment in 30 to 40 minutes.
Analysis method and notations
In this study, the contrast bias was confirmed by comparing the satisfaction levels of two attractions that were successively visited. The following two distinctions of successive attractions were considered in the data analysis. However, the attractions at the beginning of visiting sequences were not considered in this analysis because they had no previous attractions.
During the pre-experiment examination, we explored the original preference of the participants and their ratings about the tourism destinations they visited in the formal test. We calculated the semantic distance of the virtual destinations based on their ratings on each characteristic, and grouped them based on their proximity to one another.
Based on the distance between a pair of destinations, we defined Everest-lake, beach-waterfall as the same type of destination/attraction; and shop-Everest, shop-lake, waterfall-gallery, park-gallery, and beach-gallery as different types of destinations/attractions. In the experiments, if the two destinations/attractions that the participants successively visited were of the same type, the rating of the latter was marked as “low contrast of type”. If the successfully visited destinations/ attractions were different, the latter rating was marked as “high contrast of type”. We also defined those destination pairs with significantly different ratings as “high contrast of quality” and others as “low contrast of quality”. In addition, if the better destination was previously visited before the worse one, the rating of the latter was marked as “negative contrast of quality”. If the better destination was visited after the worse one, the former was marked as “positive contrast of quality”.
Experimental result
Contrast bias at the expectation level
The following figures present the results obtained from statistical processing (error bar: 5% of the total scale).
Contrast bias is evident in the expectation test (Figure 4). Satisfaction with an attraction is higher when the participants have visited different attractions than when they are visiting the same type. Similarly, satisfaction when under high contrast of type is higher than the original with no contrast. The independent t-test showed the difference is significant between high contrast and original for shop (t=2.42, p<0.05), Everest (t=2.484, p<0.05), and lake (t=1.83, p < 0.05), and between high and low contrast for Everest (t=2.48, p<0.05) and lake (t=2.25, p<0.05). However, the difference between low contrast and original is not significant for Everest (t=−1.319, sig.=0.22) and lake (t=−0.49, sig.=0.63).
For the contrast bias of quality, the effect on satisfaction is not very significant at the expectation level, except for spa and shop. Based on the independent t-test, the difference is significant between negative and positive contrast for shop (t = −2.575, p < 0.05) and spa (t=−3.617, p<0.01).
Contrast bias at the experience level
In the experience test, contrast bias on the type of destination shows that satisfaction with an attraction is higher when tourists visit a different attraction than when they visit a similar attraction or the original score. Figure 5 shows the results of the independent t-test in which the difference is significant between high contrast and original for gallery (t=3.003, p<0.05), waterfalls (t=2.320, p<0.05), and beach (t=3.704, p<0.05), and between high and low contrast for waterfalls (t=2.627, p<0.05) and beach (t=2.081, p<0.05). However, the difference between the scores of low contrast and original is not significant.
For the contrast bias of type on information learning at the experience level, the tourists reported high learning about the destination/attraction after they recently visited a different attraction more than after visiting a similar one. The independent t-test showed the difference is significant between high and low contrast for waterfalls (t=1.79, p<0.05) and beach (t=3.01, p<0.01).
A pattern of contrast bias on the quality of attraction implies that satisfaction with an attraction is higher when tourists visit an attraction rated low, and lower when they visit an attraction rated high as shown in Figure 6. Significant is indicated in the results of the independent t-test between negative and positive contrast for gallery (t=−4.431, p<0.01), park (t=−3.167, p < 0.01), waterfalls (t=−3.56, p<0.01), and beach (t=−3.617, p<0.01).
The ability of the data obtained from the VE to reflect the tourist satisfaction assessment in the real world is another key issue in validating the result of the experiment. In the experiment, the participants were requested to evaluate the similarity of their decision/evaluation processes with those they would make in their actual travels. The results are as follows: The average score was 5.35 (SD=1.317) with the minimum score of 3 and the maximum of 7 on the 1 to 7 scale (with 1 indicating “fundamentally different” and 7 indicating “no differences at all”).
The difference between participants’ decision-making in the VE and in reality is marginal and not fundamental.