Experimental Design Details
We conduct a survey experiment among tourists in a natural tourism setting to study the research questions. Tourists are invited to join a survey about their views on the topics related to their visits by the enumerators. The survey contains largely three sections. In first section, we elicit respondent’s attitudes and perceptions with regard to hedonic motivations and environmental concerns, as well as individual characteristics and trip information. In the second section, we introduce the pro-environmental messaging that calls for pro-environmental behavioral changes during their trips. We randomly assign respondents to one of the two version of pro-environmental messaging, one with hedonic framing and one with normative framing. In the third section of the survey, we elicit respondents’ pro-environmental behavioral intentions and policy support as self-reported outcome variable.
1. Pre- intervention attitudes and perceptions
The pre-intervention questions are to obtain basic information on the respondent and understand their “natural” attitudes and perceptions towards vacations and environmental issues, with a focus on hedonic goals and motivations. This section provides baseline information that will be used to explore effect mechanisms and heterogenous effects.
Specifically, we elicit respondents’ hedonic motives using a set of questions. We also elicit respondents’ hedonic goals using a question where respondents are asked state their goals of the trip.
We also include questions on awareness of environmental issues, perceived outcome efficacy, ascribed responsibility and personal norm to understand and control for individual heterogeneity.
Because literature has evidence that people often act differently in hedonic context such as vacation from in everyday life, we also ask respondents about their pro-environmental behavioral at home with a set of behavioral frequency questions.
To avoid demand effect stemmed from being overly focused on environmental issues in the questions, we also included the “decoy” questions on other issues.
2. Interventions: pro-environmental messaging
The intervention occurs as a message in the middle of the service. The intervention design had already been described above. Beside the two treatment conditions, we have a control group where respondents only see the pictures in the interim between the first and third survey sections.
We use the between-subject design, where one respondent will only experience one treatment arm.
Because tourists tend to travel in groups, to reduce contamination across treatment arms, we randomize the treatment assignment at group level.
3. Post-intervention stage: outcome variables
The primary outcome variable in our study is the tourist’s actual pro-environmental behavior. Because the pro-environmental messaging aims to encourage tourists to engage in general pro-environmental behavior later during their visits, ideally, we would like observe tourists’ pro-environmental after arrival as they go on to visit various places. However, due to the difficulty in tracing tourists after they get off the ferry, in this study, we measure tourists’ pro-environmental behavior using their decisions on what to consume upon the arrival of the ferry, when they are still on-board.
Specifically, we present the consumption decisions to the respondents as compensations for participation in the survey. We offer the respondents a gift card that can be redeemed for a refreshing drink at the terminal café at Vung Tau. We ask the respondents to order their drinks in advance, so that the drinks can be prepared in advance and save waiting time upon arrival. Each respondent can choose one fruit tea and up to two cakes from the menu, as well as a bag of chips. There are four options for fruit tea and four options for cakes. More important, on the menu, there are additional options to choosing accessory items, such as straws, teaspoons, cake spoons, cup holders, plastics bags and tissue papers. These accessory options are framed as a way to facilitate the preparation of the drinks and speed up the process. The menu is printed on the back of the gift card. Respondents place their order by checking the correspondent boxes on the menu for tea and cakes and fill in the number of each accessory items they want to have, with the default (not filling in anything) to be zero.
We use the choices of these items as the measure for pro-environmental behavior. Given the relatively weak waste management system in Vietnam, and that most of these items are made of plastics, using more accessory items typically implies larger environmental impacts. In this way, we can obtain a measure of actual behavior with environmental consequences in a natural tourism setting. This is the main outcome variable in this study.
Each gift card consists of two sections: the main body for passengers to select their options and a stub on the side for enumerators to record these choices. Each gift card includes three pieces of information to link it to the survey: the survey date (determined by the gift card’s expiration date), the seat number, and the gift card ID. Because the consumption behavior is measured separately from the survey.
Beside the pro-environmental behavior measure described above, in the survey, we also elicit tourists’ behavioral intentions, which are widely used as outcome variables in the literature. We ask the respondents about how likely they are going to adopt a set of common behavior with environmental consequences, on a 7-point Likert scale. The pro-environmental behavior including reducing waste, picking up litter, avoiding using plastic bags, which are closely related to our behavioral outcome measures.
To mitigate demand effect stemming from being too focused on pro-environmental behavior in the questionnaire, we also include a few behaviors in other domains including donating to protection the parks, donating to preservation of historical and cultural buildings, purchasing traditional handicrafts to support local artisans and supporting organizations advocating for a ban on animal abuse in tourism.
Beside the outcome variables, we also elicit respondents’ subjective valuation of the pro-environmental messaging and their view of the survey to validify the manipulation.
4. Hypothesis
Pro-environmental messaging can improve tourists’ pro-environmental behavior intention and actual pro-environmental behavior. (Control vs. Hedonic & Normative)
Hedonic framing of pro-environmental messaging is more effective than normative framing in the context of tourism. (Hedonic vs. Normative)
Hedonic framing is more effective for respondents with stronger hedonic motives. (Control vs. Hedonic, interaction with hedonic motives)
Normative framing is less effective for respondents with stronger hedonic motives. (Control vs. Normative, interaction with hedonic motives)
Potential attitude-behavior gap: positive effects on behavioral intentions, but no effect on behavior. If the gap exists, we expect heterogeneity in the gap. Those who hold strong hedonic motives will shower larger gap. Perhaps similar effects on intention, much small effect on behavior.
5. Data collection
We carried out the study during Nov. - Dec. 2024. Our sampling frame is all the passengers travelling on the Greenline DP’s ferry trips from HCMC to VT. We choose passengers on the one-way trip because visitors, particularly foreigner visitors, usually arrive in HCMC first and travel from HCMC to visit different sites in the neighboring regions. Our polit suggests that passengers on the return trip from VT to HCMC are often tired and more difficult to approach. We also want to conduct the study before their visits so that the interventions are more likely to induce actual behavioral changes during their visits.
The ferry has a variable schedule based on the expected flow of passengers. On average, an enumerate can travel with the ferry for two round trips between HCMC and VT per day during our studied period. During each trip, the enumerator invites all passengers one-by-one on the ferry to participate in our study. The passengers are also informed of the study with a leaflet placed in the seat pocket. In case of group travelers, the enumerator will first confirm group members on the spot, and then invite all adult group members. To randomize at group level, the enumerator would generate a group ID on the spot with a roll of dice. The random allocation of treatment arms is based on the group IDs. All respondents from the same group are asked to fill in the group ID to start survey.
The enumerator would make it clear to the passengers that the survey is part of an academic project and reassure the passengers that the survey will not collect any information other than their answers in the survey and they are free to not answer certain questions.
In case that passengers refuse to participate, the enumerators would try to record the seat numbers and some observable individual characteristics (gender, domestic/foreign, group/alone) to evaluate the potential non-response bias.