On the importance of field studies for testing theory-driven behavioral change interventions in (sustainable) tourism

Introduction

The world is “on a pathway to global warming of more than double the 1.5-degree limit” (United Nations, 2022). According to the most recent report of the Intergovernmental Panel on Climate Change, “changes to behavior can result in a 40-70% reduction in greenhouse gas emissions by 2050”. Behavioral change interventions, therefore, represent a key strategy for climate change mitigation across all industry sectors, including tourism. As opposed to most other industry sectors, tourism heavily depends on the pristine unspoiled nature and natural resources; it is, in fact, a key driver of tourist visitation (e.g., Kim, 1998; Garms et al., 2017). Yet, tourism also harms these very resources it existentially depends upon by using extensive amounts of fresh water (Gössling et al., 2012; Rico et al., 2020), emitting some 8% of total global CO2 emissions (Lenzen et al., 2017), generating substantial amounts of solid waste (Diaz-Farina et al., 2020) and food waste (Gössling et al., 2011; De Visser-Amundson, 2022) and by discharging toxic sludge (Moscovici, 2017). Ensuring that tourism will continue contributing to life satisfaction (Dolnicar et al., 2012) and the improvement of the living conditions at host communities (Uysal et al., 2016) requires finding ways to improve the environmental performance of tourism. Previous research has shown that only about 10% of global travelers consider environmental consequences in the tourism context (e.g., Karlsson & Dolnicar, 2016); most tourists fail to convert their positive environmental attitudes into environmentally friendly behavior when they go on vacation (Juvan & Dolnicar, 2014), especially in activities with high negative environmental impact, such as air travel (Gössling & Dolnicar, 2022).

Because of the urgent need to find effective ways of improving the environmental performance of the tourism industry through behavioral change interventions, researchers are increasingly directing their attention towards developing and testing theory-driven practical measures that could be deployed by tourism businesses and destinations to make tourists behave in more environmentally sustainable ways. Effective practical measures (interventions) include reduced the plate size at buffet meals (Kallbekken & Sælen, 2013), introducing stamp collection games for families at buffet meals (Dolnicar et al., 2020), asking hotel guests to commit to towel reuse and giving them a visible pin to publicly display their commitment (Baca-Motes et al., 2013) and changing the default to hotel room cleaning upon request only (Kneževič Cvelbar et al., 2021).

More such interventions need to be developed. Before interventions can be recommended for deployment by tourism businesses and destinations, their effectiveness needs to be proven. A common way of testing the effectiveness of interventions is to conduct survey experiments. Survey experiments allow causal conclusions about the effectiveness of an intervention with respect to changes in beliefs and behavioral intentions (Viglia & Dolnicar, 2020). Successful survey experiments do not necessarily imply that any given behavioral change intervention will also be effective in changing real tourist behavior in tourism businesses and at tourism destinations. Proof for the potential to change real behavior can only be derived from field studies, which remain rare in tourism research (Viglia & Dolnicar, 2020).

The aim of the present study is to investigate the extent to which relying solely on survey experiments can lead to misleading conclusions about the effectiveness of behavioral change interventions. We test empirically whether field studies are necessary, or whether it is sufficient to run survey experiments to accurately assess the effectiveness of newly developed behavioral change interventions to trigger more environmentally sustainable behaviors among tourists in the real tourism context. This represents a key methodological insight, which is critically important to guide future experimental studies testing the effectiveness of behavioral change interventions that have the potential of making material contribution to mitigating environmental costs of tourism.

Literature review

Methodological approaches to testing the effectiveness of behavioral change interventions

Self-report surveys and survey based experiments are the two most commonly used tools to test the effectiveness of behavioral change interventions in tourism (Viglia & Dolnicar, 2020). Surveys represent a low-cost and time-efficient approach to accessing information from a large population. Some behaviors such as past pro-environmental behavior on vacation and behavioral intentions can be accessed only by self-report survey studies. A well-designed survey is also able to measure whether a behavioral change intervention triggers the corresponding psychological constructs that are postulated to induce behavioral change (Sparks et al., 2013). As a result, survey studies are extensively used in tourism research to assess the effectiveness of behavioral change interventions, including promoting sustainable transport choices (Brög et al., 2009), inducing responsible water and electricity use (Tussyadiah & Miller, 2019) and adopting post-visit environmentally responsible behaviors (Ballantyne & Packer, 2011). Responses from survey studies, however, are prone to capturing several biases, including social desirability bias (Fisher, 1993). Social desirability bias occurs when respondents answer survey questions in a way that makes them look good in the eyes of the researchers, rather than reflecting the truth (Paulhus, 1984). Social desirability can inflate estimated environmentally sustainable tourist behavior by 74% (Juvan & Dolnicar, 2016). Additionally, the reliability of the measurements as indicated by item level test retest reliability can be very low (Dolnicar et al., 2022). When using a seven-point bipolar answer format in the questionnaire, for example, respondents only give the same response twice in a row without external conditions having changed in 47% of cases (Dolnicar, 2021). Finally, real behavior can never be measured in surveys. For behavioral change intervention survey experiments the dependent variable is typically a stated behavioral intention. Stated behavioral intentions relating to socially desirable pro-environmental behaviors are particularly prone to capturing social desirability bias (Juvan & Dolnicar, 2017), thus potentially undermining the external validity of findings derived from survey experiments.

An alternative way of testing the effectiveness of behavioral change interventions is to run laboratory or field experiments (Harrison & List, 2004). Laboratory experiments observe specific types of behaviors in a fully controlled environment increasing overall control by the researcher (Gneezy, 2017; Viglia & Dolnicar, 2020). Laboratory experiments, therefore, are less prone to capturing social desirability bias, but the behavior that can be observed is rarely the main behavior of interest. For example, the differences in attention attraction of different carbon offsetting appeals in encouraging travelers to commit to voluntary flight carbon offsetting (Babakhani et al., 2017) or of visual carbon labels on restaurant menus to encourage ordering of low-emissions menu items (Babakhani et al., 2020) can be tested validly with eye fixations as dependent variable in eye tracking studies. Eye fixations are a measure of attention and measuring attention is of interest because attention is a precursor to a carbon label or a carbon offsetting appeal influencing actual consumer decisions, but they do not correspond to consumer decisions. As such, laboratory experiments offer insights in terms of the dependent variable than survey experiments, but they still fall short of being able to claim that a newly developed behavioral change intervention has a high likelihood of being effective when deployed in real tourism businesses and at tourist destinations.

Alternatively, field experiments can overcome this limitation by testing behavioral change interventions in real-life contexts (Gerber & Green, 2012). Testing the effectiveness of behavioral change interventions using field experiments ensures high external validity; thus making the findings more reliable (Viglia & Dolnicar, 2020). True field experiments are almost impossible to conduct in real life tourism contexts, because they require random assignments of study participants to experimental conditions. In a hotel, for example, this would imply being able to control which hotel guests come to stay during the control period and which guests come to stay during the experimental period. Because this is not possible, field studies in tourism are mainly implemented as quasi-experimental field studies. Such studies have all features of a field experiment except random assignment of study participants. Even quasi-experimental field studies remain rare in tourism research because they are costly and time-consuming to implement. A recent review of field studies aimed specifically at improving the sustainability of tourism, concludes that, to date, only a small number of behaviors and intervention strategies have been extensively tested in the field (Demeter et al., 2023). Towel reuse and food waste reduction emerge as the most frequently targeted tourist behaviors. In terms of theoretical foundations, leveraging beliefs and social norms are the two most popular behavioral change mechanisms; this is despite the relatively low success rate of these approaches – 58% and 47% respectively (Demeter et al., 2023).

Successful behavioral change interventions for sustainable tourism

The study of sustainable tourism has a long and proud history (Bramwell & Lane, 1993). Yet surprisingly few tangible measures have been proposed that are proven to reduce the environmental footprint of tourism. Measures that have been proposed in the past cover a limited number of behaviors, predominantly focusing on reducing water use, food waste, and routine hotel room cleaning; directing online bookings to low-carbon options; encouraging carbon offsetting; using recycled paper instead of thick cotton serviettes and increasing towel reuse.

Towel reuse, for example, can be increased by explicitly asking guests for help and by leveraging descriptive social norms. In the context of a mid-sized, mid-priced hotel in the US, using message appeals with descriptive social norms increased the towel reuse rate by 44%. Adding descriptive norms on the same room identity added another 5% increase in towel reuse (Goldstein et al., 2008). A similar approach in Australian motels confirmed the effect of descriptive norms on towel reuse (Mair & Bergin-Seers, 2010): 87% of guests re-used towels under the descriptive norms condition. Of those, 6% reported doing so because of the intervention; other reasons stated include habit and no need for daily fresh towels. A study in a hotel in California demonstrated that social norms can best be leveraged by making tourist commitment to towel reuse visible to other hotel guests (Baca-Motes et al., 2013): asking guests to commit to towel reuse and giving them a wearable pin to signal their commitment led to 40% more towels being reused. At a typical see-sun-sand hotel in Spain, using comprehensive message appeals communicating declarative, procedural and effectiveness knowledge also reduced towel reuse by almost 7% (Gössling et al., 2019).

Results from a quasi-experimental study conducted in a three-star city hotel suggest that the change in defaults significantly reduced room cleaning, with only 32% of room cleans requested on average (Kneževič Cvelbar et al., 2021). Sharing savings from fewer room cleans in the form of a drink voucher that can be used at the hotel premises reduced unnecessary room cleaning by 42% (Dolnicar et al., 2019).

A study in a Scandinavian hotel chain shows that using 3 cm smaller plates at food buffets reduced plate waste by 22% and that using a table sign with a message inviting hotel guests to return to the buffet many times (rather than overfilling the plate on a first visit), reduced plate waste by 20% (Kallbekken & Sælen, 2013). In a typical sea-sun-sand hotel at the Adriatic coast, introducing a stamp collection game for families, which rewards children for not leaving any uneaten food behind at the end of the meal, reduced average plate waste per family member by 34% (Dolnicar et al., 2020).

Environmentally sustainable online booking behavior can be increased by including video content that triggers specific feelings of empathy for the future generation, leading to more travel packages with lower CO2 emissions being chosen (Araña & León, 2016). Providing instant feedback on hot water use in hotel showers reduced hot water electricity use by 11% (Tiefenbeck et al., 2019). Defaults have also proven effective in increasing voluntary carbon offsetting in the conference travel context to Grand Canaria in Spain (Araña & León, 2013). When the carbon offset is included in the price and conference attendees have to opt out to avoid purchasing the offset, the willingness to pay for the conference fee including the carbon offset was significantly higher. Similarly, placing recycled paper serviettes on breakfast tables and offering less environmentally friendly, thick cotton serviettes at the buffet only, led to an almost complete change-over to the more environmentally friendly option: 95% reduction in the use of the less environmentally sustainable large white cotton serviettes (Dolnicar et al., 2019b).

Overall, field experiments allowed to identify successful behavioral change interventions by empirically testing them in real world environments. However, due to their cost and time requirements only a limited range of behaviors and potential measures have been covered. Survey experiments would allow the testing of different interventions in a more cost and time efficient manner. Given that the reliability and validity of results obtained based on survey experiments only is doubtful, a comparison of outcomes from survey and field experiments is required to assess the suitability of replacing field experiments with survey experiments.

Methodology

We developed and tested in two separate studies two alternative interventions aimed at reducing buffet plate waste in a five-star hotel. One study is a survey experiment which also served as the manipulation check to determine whether the newly developed interventions triggered the psychological constructs hypothesized to drive behavioral change. Manipulation checks provide insights into changes in latent psychological constructs that are not observable (unless there is a possibility to survey tourists on site; Sparks et al., 2013; Viglia & Dolnicar, 2020). The other study tested the effectiveness of the interventions in a quasi-experimental field study, allowing us to determine the effect on actual tourist behavior – in this case not generating plate waste at buffets – in a real hotel.

We developed two different behavioral change interventions. The first intervention built upon the theory of planned behavior (Ajzen, 1991) and value belief norm theory (Stern, 2002) in leveraging environmental beliefs by providing factual information to guests about how food leftovers negatively affect the environment: ‘Please help the environment: don’t leave edible food leftovers on your plate at the end of your meal’ (see Figure 1a). The second intervention focuses on humor, which has been recognized as an effective tool to help people process information actively and enhance concentration in a hedonic context such as tourism (Pearce, 2009; Kahneman et al., 1999). We leverage the effect of humor by informing guests that the food leftovers negatively affect the health of the already overweight cat Marko who lives at the hotel: ‘Please help Marko with his diet: don’t leave edible food leftovers on your plate at the end of your meal’ (see Figure 1b). Both signs were designed in line with the hotel branding.

Figure 1. Table signs used in the experimental conditions

Study 1 – Survey experiment and manipulation check

We recruited 317 respondents via the online recruitment platform Prolific Scholar. We chose this survey method because self-administered online surveys are less influenced by social desirability bias when measuring self-reported environmental behavior and behavioral intention in comparison with the traditional face-to-face survey method (Heerwegh, 2009). Only respondents who have stayed at a hotel in the past five years and have eaten at a hotel buffet in the past five years qualified for the survey experiment.

All survey respondents were primed to imagine they are on a relaxing beach holiday and an all-you-can-eat dinner buffet is included in the hotel price. We informed them that plate waste is a common problem at buffets and that the average guest leaves about 100g of uneaten food behind on their plate for each meal. We then asked respondents to imagine they are about to have their buffet dinner during the beach holiday. We randomly assigned respondents to one of three conditions. The control group (n = 110) did not see a sign. Experimental group 1 (n = 102) saw the sign including the environmental beliefs. Experimental group 2 (n = 105) saw the sign with Marko, the cat. After seeing the signs, respondents answered the following question: ‘Which percentage of the food you take from the buffet would you leave behind uneaten? Please remember that the average guest leaves 100g of uneaten food behind at each meal.’ Respondents recorded their answers using a slider scale ranging from 0% to 100%.

We tested whether the interventions activated the intended emotions by asking: ‘Now please remember the sign you saw. How did it make you feel?’. Respondents chose ‘Yes’ or ‘No’ for each of the following ten emotions presented to them: wanting to eat up everything on my plate, interested, responsible, concerned, sustainable, guilty, amused, entertained, upset, annoyed. Based on value belief norm theory (Stern, 2000), we expect both interventions will make respondents feel guilty, concerned, responsible, sustainable, upset and wanting to eat up everything on their plate, while after seeing the sign with the cat respondents will feel, more entertained, amused, interested and less annoyed.

The stated behavioral intentions are summarized for each experimental condition using descriptive statistics (mean, standard deviation, skewness, excess kurtosis, median and quantiles) to assess differences in location as well as shape. A shift in location is tested for all three conditions using a Kruskal-Wallis rank sum test as well as a Wilcoxon rank-sum test for the comparison between control and each of the experimental conditions. The power of the Wilcoxon rank sum test for a sample size of 100 in each group and a significance level of 5% is assessed using sampling-based methods. Observations obtained in the control condition are bootstrapped using 10,000 replications with different positive shifts being added for the treatment condition while also taking the admissible range of values into account. The activated emotions are analyzed by visualizing the percentages of agreement for each of the two experimental condition across the ten emotions using bar plots. The differences in proportions between the two experimental conditions are tested using a chi-squared test for each emotion. The power of the chi-squared test is investigated for a sample size of 100 in each group and a significance level of 5% based on the normal approximation for the binomial distribution using different success probabilities for the two experimental conditions.

Study 2 – Quasi-experimental field study

We conducted a quasi-experimental field study to ensure we could draw causal conclusions about actual tourist behavior that are valid for the hotel context (Viglia & Dolnicar, 2020). Our study is a quasi-experimental field study because it is not possible to randomly assign tourists to a specific experimental condition when the experiment is conducted during normal hotel business operations. The field site for the experiment was a five-star hotel in a Slovenian coastal resort town, a popular tourist destination, especially during the summer months. The hotel is a high-end five-star hotel located centrally within the destination on a hill overseeing the sea.

The dependent variable was average plate waste generated at the breakfast and dinner buffets per person per day in grams. Plate waste was measured using a floor scale that was placed under the plastic bins used in the kitchen to dispose of the plate waste. These bins are not used for kitchen preparation waste. The bins include food and non-edible biodegradable items such as serviettes. The number of people who ate breakfast and dinner on any given day was provided by the hotel. The hotel has a staff member located at the entrance to the dining area. This staff member welcomes guests and notes their room numbers. In addition, we obtained de-identified daily guest mix data from the hotel.

We measured plate waste per person per meal for the control condition and the two experimental conditions. Both the control and the experimental conditions ran during the peak summer tourist season. The control condition ran from 1st to 21st of July 2019, and effectively represented the status quo of operations at the hotel. The first experimental condition was implemented between the 24th of July 2019 and the 11th of August 2019 and involved an information intervention providing facts about the negative environmental impacts of food waste (see Figure 1a). The second experimental condition – fielded between 12th of August 2019 and 2nd of September 2019 – involved the humor message intervention (see Figure 1b).

The intervention was presented as a cube-shaped table sign where each side contained the same information as shown in Figure 1 in a different language, covering the main languages used by guests in this hotel (Slovenian, English, German and Italian). The design of the table signs was in line with the hotel design.

To check whether the signs were noticed and triggered the intended emotions – make people giggle and increase knowledge about wastefulness and the environmental impacts of plate waste – we conducted a voluntary survey study, asking people at checkout to answer a few questions about their dining experience. Because this was a voluntary survey, the response rate was low: 22 tourists in the control condition completed the survey, 32 in the factual information condition focusing on environmental beliefs, and 24 in the humor condition. In both experimental conditions respondents indicated that they have noticed the table sign more often than tourists in the control group (control group: 32%, experimental group 1 – environmental beliefs: 69%, experimental group 2 – humor: 92%; Fisher’s exact test: p-value < 0.001), with a slight indication that the table sign might be noticed more frequently in the humor condition (Fisher’s exact test: p-value = 0.05). When inspecting which emotions were triggered by the table signs, 68% of respondents in the humor condition indicated that they were amused by the table sign, but only 14% did so in the environmental beliefs condition (Fisher’s exact test: p-value = 0.001). In the environmental beliefs condition 73% of respondents agreed to feel sustainable, while only 46% of respondents agreed to this statement in the humor condition, but this difference was not statistically significant (Fisher’s exact test: p-value = 0.12).

For each day and meal, we calculated the average plate waste per person in grams. Additional information available for each of these daily meal-specific observations consists of the number of guests, the percentage of guests having this meal for the first time at the hotel, the percentage of guests having this meal for the last time at the hotel, the percentage of guests being male, the percentage of guests with different countries of origin (categorized into Austria, Italy, Germany, Hungary, Slovenia, the remaining Eastern European and Balkans countries and all other countries of origin) and from different age groups (categorized into the following age groups in years: 0-6, 7-14, 15-24, 25-60, 61-100).

The additional information is summarized separately for each meal and experimental condition using the empirical mean, standard deviation, skewness and excess kurtosis. We assess differences in the number of guests across experimental conditions for each meal using F-tests comparing linear regression models controlling for weekday effects with and without experimental condition as independent variable. We assess differences in guest characteristics across experimental conditions for each meal using likelihood ratio tests comparing logistic regression models controlling for weekday effects with and without experimental condition as independent variable. A time series plot indicating the experimental conditions using color-coding visualizes the average plate waste per person (in g) for each meal. To assess differences in plate waste across experimental conditions, weighted linear regression models are fitted using the average plate waster per person (in g) as dependent variable, the number of guests as weights, the guest characteristics as control variables and the experimental condition as independent variable. The effect of the experimental condition results from comparing the model with and without the experimental condition as independent variable using an F-test and by determining the mean and the 95% confidence intervals for these effect estimates. Based on the daily control measurements of plate waste per person for breakfast and dinner, a power analysis is performed for a two-sample t-test using the observed standard deviation as well as different shifts in mean values for a sample size of 20 and 30 in each group.

The study received approval from the Ethics Committee in the authors’ university (2019/HE001609; 2022/HE001344).

Impact of COVID-19 on our research

Our primary data (field experiment) was collected before COVID-19. But the pandemic does not affect the implication and conclusions of our study because it studies fundamental mechanisms driving human behavior. Specifically, we study how two alternative behavioral change interventions – two table signs containing messages based on diametrically opposed social science theories – affect food waste generation at a hotel breakfast buffet. Although some food outlets changed their serving style during the pandemic, which could lead to potential plate waste reduction (Chang, 2022), evidence suggests that the tourism industry has largely focused on re-establishing the pre-pandemic status after the COVID-19 disruption because the dominant role of relational and normative expectations (Zhu & Dolnicar, 2022). We see no reason to believe, therefore, that either buffets would be abandoned in the tourism industry (which would make our study irrelevant) or that fundamental constructs such as beliefs or the desire to experience enjoyment would affect human behavior in a different way after COVID.

Results

Study 1 – Survey experiment and manipulation check

Table 1 summarises the stated behavioral intentions of the respondents in the survey across all three experimental conditions regarding the percentage of food left behind on their plate which they have taken from the buffet. The location measures (mean and median) indicate that these are highest for the control condition, followed by the humor and the environmental beliefs condition. The quantiles also indicate that the values are highest for the control condition with the environmental condition outperforming the humor condition, particularly for the extreme 10% and 90% quantiles.

The differences in median values are significant according to a Kruskal-Wallis rank sum test (χ2 = 8.31, df = 2, p-value = 0.02), with the environmental beliefs condition having a significantly lower median value than the control condition (W = 6,885; p-value = 0.004) whereas the difference in median values is not significant at the 5% level for humor versus control condition (W = 6,377; p-value = 0.186). The power analysis indicates that a sample size of n = 100 would not have sufficient power to identify a median shift of five (power = 0.72), while the power is already sufficiently high for a median shift of six (power = 0.97).

Table 1. Summary of the behavioral intentions for the three experimental conditions using number of observations (N), an empirical mean, standard deviation (SD), skewness (Skew.) and excess kurtosis (Kurt.) as well as the median and several other α-quantiles Qα.

Figure 2 visualises the results of the manipulation check assessing if the signs trigger the intended emotions. The plot contains the percentages of agreement for the respondents in the two experimental conditions for each of the ten emotions shown separately in panels with the panel label indicating the emotion. The emotions are ordered by overall agreement percentages across both experimental conditions.

Figure. 2. Percentages of agreement with the ten emotions being evoked by the signs of the two experimental conditions

The highest associated emotion overall is “Wanting to eat up everything on my plate”. This emotion is triggered significantly more by the environmental beliefs condition with 91% compared to the humor condition with 60% (chi-squared test: χ2 = 25.4, df = 1, p-value < 0.001). In both conditions, respondents indicate that they feel interested (75% and 69%), with the difference insignificant at the 5% level (χ2 = 0.9, df = 1, p-value = 0.34). A clear majority in the environmental condition also feels responsible, concerned, sustainable and guilty, while the levels of agreement with these emotions are significantly lower for the humor condition (all p-values < 0.001). This is reversed for amused and entertained (again all p-values < 0.001). Only few respondents express being upset or annoyed with the former being slightly higher for the environmental beleif condition (p-value = 0.003) whereas the low levels of annoyance are not significantly different (p-value = 0.32). The power analysis indicates that a sample size of n = 100 would not have sufficient power to identify a difference in proportion of 0.1 in case the proportions themselves vary between 0.1 and 0.9 (power between 0.29 and 0.51), while the power is already sufficiently high for a difference in proportion of 0.2 in case the proportions themselves vary between 0.1 and 0.9 (power between 0.81 and 0.94).

The results from this survey experiment suggest that the environmental message has the potential to significantly reduce plate waste at hotel buffets. In addition, there is very little risk of either message upsetting or annoying hotel guests. Based on these findings, researchers would recommend to hotels operating buffet meal services to make use of the environmental belief-based table sign in an attempt to prevent unnecessary food waste. The fact that the table sign with the cat triggers emotions like amused and entertained suggests that its effect on real tourist behavior in a hedonistic context should also be investigated and assessed in comparison to the stated behavioral intentions.

Study 2 – Quasi-experimental field study

Table 2 provides an overview of the plate waste data collected during the field study. For each meal and experimental condition, the table contains the number of daily observations as well as the empirical mean and standard deviation of the daily plate waste per person per day (in g) and the number of guests. The power analysis based on the control measurements indicates that for a relative decrease of 10% in average waste per person and day the power is insufficient for the sample size of n = 20 (power = 0.54 for breakfast and 0.58 for dinner), and n = 30 (power = 0.72 for breakfast and 0.76 for dinner). By contrast a relative decrease of 15% in average waste per person and day leads to a power above 0.8 for both sample sizes (n = 20: power = 0.87 for breakfast and 0.90 for dinner; n = 30: power = 0.97 for breakfast as well as dinner).

Table 2. Summary of plate waste per person and day (in g) and number of guests for each meal and experimental condition based on the number of observations (N), the empirical mean, the standard deviation (SD), skewness (Skew.) and excess kurtosis (Kurt.).

Complementing Table 2, Figure 3 visualizes the average plate waste per person in grams per day and meal for the three experimental conditions, separately for breakfast and dinner. The daily average plate waste is indicated by a bullet and these bullets are joined by lines between consecutive days and within each experimental condition. Colors indicate the different experimental conditions: control (red), environmental beliefs (green), humor (blue). In addition, the overall average plate waste per person (in g) per day and meal for each experimental condition is given by straight lines.

Figure. 3. Plate waste pattern for breakfast and dinner. (Red – control condition, green – environmental beliefs condition, blue – humor condition)

A few key findings can be derived from Table 2 and Figure 3. First, the average plate waste is much lower for breakfast than dinner. Second, plate waste fluctuation across days is substantially higher for dinner. Third, the control group does not, as hypothesized, generate the highest amount of plate waste.

Guest characteristics differ considerably across the experimental conditions as well as the two meal types – Table 3 shows the empirical mean and standard deviation for the daily number of guests and the guest characteristics for each meal type and experimental condition. These results indicate that accounting for differences in guest characteristics is warranted when estimating the effects of the experimental conditions. The guest characteristics are included as control variables in the regression models estimating the effects of the experimental conditions on the average plate waste.

Table 3. Summary of number of guests and demographic and guest characteristics for each meal and experimental condition based on the empirical mean and the standard deviation (in round parentheses). For the number of guests in square brackets also the skewness (first) and the excess kurtosis (second) are given. The remaining Eastern European countries and countries on the Balkans are combined in one category abbreviated as EEB. p-values are reported for likelihood ratio tests assessing differences between experimental conditions separately for each meal and variable / category while controlling for weekday.

Estimating average plate waste while controlling for the proportion of first day guests, last day guests, guests from different countries of origin, and guests of different age groups, confirms that neither of the two experimental conditions providing information to guests and asking them explicitly to reduce the plate waste they generate, significantly reduce the amount of plate waste generated (breakfast: F = 0.23, p-value = 0.79; dinner: F = 0.23, p-value = 0.79). These results are complemented by the mean and 95% confidence intervals for the effect estimates of difference in average plate waste for each experimental condition compared to the control group given in Table 4.

Table 4. Mean and 95% confidence intervals (indicated by their lower and upper bound) of the effect of the experimental condition compared to the control group on average plate waste per person (in g) for each meal estimated using weighted linear regression models with guest characteristics as control variables.

The results from this quasi-experimental field study indicate that neither of the two messages designed to entice tourists to eat up everything they have taken from a hotel breakfast buffet onto their plate are effective in reducing the food waste these tourists generate. This conclusion stands in direct contradiction to the findings from the survey experiment, suggesting instead that producing table sign with the environmental belief message and placing on dining tables would represent a waste of resources and a waste of precious dining table real estate.

Conclusions

The aim of this study was to determine whether conducting survey experiments leads to accurate estimates about the effectiveness of behavioral change interventions in real tourism settings. Based on the comparison of the results derived from a survey experiment and those derived from a quasi-experimental field study, we conclude that this is not the case. In the survey experiment, tourists were over-reporting their intention not to generate plate waste, leading to the incorrect conclusion that the environmental belief message has the potential to significantly reduce plate waste at real hotel buffets.

This finding has immediate implications for tourism researchers working on developing and empirically testing the effectiveness of behavioral change interventions for sustainable tourism. When such research aims to derive tangible managerial recommendations in terms of operational changes that can be implemented to entice tourists to behave in more environmentally sustainable ways, newly developed behavioral change interventions must be tested in field studies to be able to conclude with certainty that they indeed have a significant impact on tourist behavior.

The limitation of the present study is that it has been conducted only in the context of plate waste and has tested only two specific messages. It is possible that results would be slightly different for environmentally sustainable behaviors less affected by social desirability bias.

References

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179-211.

Araña, J. E. & León, C. L. (2013). Can defaults save the climate? Evidence from a field experiment on carbon offsetting programs. Environmental and Resource Economics, 54(4), 613-626.Araña, J. E., & León, C. J. (2016). Are tourists animal spirits? Evidence from a field experiment exploring the use of non-market based interventions advocating sustainable tourism. Journal of Sustainable Tourism, 24(3), 430-445.

Babakhani, N., Lee, A., & Dolnicar, S. (2020). Carbon labels on restaurant menus: Do people pay attention to them?. Journal of Sustainable Tourism, 28(1), 51-68.

Babakhani, N., Ritchie, B. W., & Dolnicar, S. (2017). Improving carbon offsetting appeals in online airplane ticket purchasing: Testing new messages and using new test methods. Journal of Sustainable Tourism, 25(7), 955-969.

Baca-Motes, K., Brown, A., Gneezy, A., Keenan, E. A., & Nelson, L. D. (2013). Commitment and behavior change: Evidence from the field. Journal of Consumer Research, 39(5), 1070-1084.

Ballantyne, R., & Packer, J. (2011). Using tourism free-choice learning experiences to promote environmentally sustainable behaviour: The role of post-visit ‘action resources’. Environmental Education Research, 17(2), 201-215.Bramwell, B., & Lane, B. (1993). Sustainable tourism: An evolving global approach. Journal of Sustainable Tourism, 1(1), 1-5.

Brög, W., Erl, E., Ker, I., Ryle, J., & Wall, R. (2009). Evaluation of voluntary travel behaviour change: Experiences from three continents. Transport Policy, 16(6), 281-292.

Chang, Y. Y. C. (2022). All you can eat or all you can waste? Effects of alternate serving styles and inducements on food waste in buffet restaurants. Current Issues in Tourism, 25(5), 727-744.

De Visser-Amundson, A. (2022). A multi-stakeholder partnership to fight food waste in the hospitality industry: A contribution to the United Nations Sustainable Development Goals 12 and 17, Journal of Sustainable Tourism, 30(10), 2448-2475.

Demeter, C., Fechner, D., & Dolnicar, S. (2023). Progress in field experimentation for environmentally sustainable tourism – A knowledge map and research agenda. Tourism Management, 94, 104633.

Diaz-Farina, E., Díaz-Hernández, J. J., & Padrón-Fumero, N. (2020). The contribution of tourism to municipal solid waste generation: A mixed demand-supply approach on the island of Tenerife. Waste Management, 102, 587-597.Dolnicar, S. (2021). 5/7-point “Likert scales” aren’t always the best option: Their validity is undermined by lack of reliability, response style bias, long completion times and limitations to permissible statistical procedures. Annals of Tourism Research, 91, 103297.

Dolnicar, S., Grün, B., & MacInnes, S. (2022). Assessing survey response stability: A complementary quality assurance protocol for survey studies in the social sciences. Social Sciences and Humanities Open, 6(1), 100339.

Dolnicar, S., Juvan, E., & Grün, B. (2020). Reducing the plate waste of families at hotel buffets – A quasi-experimental field study. Tourism Management, 80, 104103.

Dolnicar, S., Kneževič Cvelbar, L., & Grün, B. (2019). A sharing-based approach to enticing tourists to behave more environmentally friendly. Journal of Travel Research, 58(2), 241-252.

Dolnicar, S., Kneževič Cvelbar, L., & Grün, B. (2019b). Changing service settings for the environment – How to reduce negative environmental impacts without sacrificing tourist satisfaction. Annals of Tourism Research, 76, 301-304.

Dolnicar, S., Yanamandram, V., & Cliff, K. (2012). The contribution of vacations to quality of life. Annals of Tourism Research, 39(1), 59-83.

Fisher, R. J. (1993). Social desirability bias and the validity of indirect questioning. Journal of Consumer Research, 20(2), 303-315.

Garms, M., Fredman, P., & Mose, I. (2017). Travel motives of German tourists in the Scandinavian mountains: The case of Fulufjället National Park. Scandinavian Journal of Hospitality and Tourism, 17(3), 239-258.

Gerber, A. S., & Green, D. P. (2012). Field experiments: Design, analysis, and interpretation. W.W. Norton.Gneezy, A. (2017). Field experimentation in marketing research. Journal of Marketing Research, 54(1), 140-143.

Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint: Using social norms to motivate environmental conservation in hotels. Journal of Consumer Research, 35(3), 472-482.

Gössling, S., & Dolnicar, S. (2022). A review of air travel behavior and climate change. Wiley Interdisciplinary Reviews: Climate Change, e802.

Gössling, S., Araña, J. E., & Aguiar-Quintana, J. T. (2019). Towel reuse in hotels: Importance of normative appeal designs. Tourism Management, 70, 273-283.

Gössling, S., Garrod, B., Aall, C., Hille, J., & Peeters, P. (2011). Food management in tourism: Reducing tourism’s carbon ‘foodprint’. Tourism Management, 32(3), 534-543.

Gössling, S., Peeters, P., Hall, C. M., Ceron, J. P., Dubois, G., & Scott, D. (2012). Tourism and water use: Supply, demand, and security. An international review. Tourism Management, 33(1), 1-15.

Harrison, G. W., & List, J. A. (2004). Field experiments. Journal of Economic Literature, 42(4), 1009-1055.

Heerwegh, D. (2009). Mode differences between face-to-face and web surveys: An experimental investigation of data quality and social desirability effects. International Journal of Public Opinion Research, 21(1), 111-121.

Juvan, E., & Dolnicar, S. (2016). Measuring environmentally sustainable tourist behaviour. Annals of Tourism Research, 59, 30-44.

Juvan, E., & Dolnicar, S. (2017). Drivers of pro-environmental tourist behaviours are not universal. Journal of Cleaner Production, 166, 879-890.

Kahneman, D., Diener, E., & Schwarz, N. (Eds.). (1999). Well-being: Foundations of hedonic psychology. Russell Sage Foundation.

Kallbekken, S., & Sælen, H. (2013). ‘Nudging’ hotel guests to reduce food waste as a win-win environmental measure. Economics Letters, 119(3), 325-327.

Karlsson, L., & Dolnicar, S. (2016). Does eco certification sell tourism services? Evidence from a quasi-experimental observation study in Iceland. Journal of Sustainable Tourism, 24(5), 694-714.

Kim, H. (1998). Perceived attractiveness of Korean destinations. Annals of Tourism Research, 25(2), 340-361.

Kneževič Cvelbar, L., Grün, B., & Dolnicar, S. (2021). “To clean or not to clean?” Reducing daily routine hotel room cleaning by letting tourists answer this question for themselves. Journal of Travel Research, 60(1), 220-229.

Lenzen, M., Sun, Y. Y., Faturay, F., Ting, Y. P., Geschke, A., & Malik, A. (2018). The carbon footprint of global tourism. Nature Climate Change, 8(6), 522-528.

Mair, J., & Bergin-Seers, S. (2010). The effect of interventions on the environmental behaviour of Australian motel guests. Tourism and Hospitality Research, 10(4), 255-268.

Moscovici, D. (2017). Environmental impacts of cruise ships on island nations. Peace Review, 29(3), 366-373.Paulhus, D. L. (1984). Two-component models of socially desirable responding. Journal of Personality and Social Psychology, 46(3), 598-609.

Pearce, P. L. (2009). Now that is funny: Humour in tourism settings. Annals of Tourism Research, 36(4), 627-644.

Rico, A., Olcina, J., Baños, C., Garcia, X. & Sauri, D. (2020). Declining water consumption in the hotel industry of mass tourism resorts: contrasting evidence for Benidorm, Spain. Current Issues in Tourism, 23(6), 770-783.

Sparks, B. A., Perkins, H. E., & Buckley, R. (2013). Online travel reviews as persuasive communication: The effects of content type, source, and certification logos on consumer behavior. Tourism Management, 39, 1-9.

Stern, P. C. (2000). New environmental theories: Toward a coherent theory of environmentally significant behavior. Journal of Social Issues, 56(3), 407-424.

Tiefenbeck, V., Wörner, A., Schöb, S., Fleisch, E., & Staake, T. (2019). Real-time feedback promotes energy conservation in the absence of volunteer selection bias and monetary incentives. Nature Energy, 4(1), 35-41.

Tussyadiah, I., & Miller, G. (2019). Perceived impacts of artificial intelligence and responses to positive behaviour change intervention. In J. Pesonen & J. Neidhardt (Eds.), Information and communication technologies in tourism (pp. 359-370). Springer.

United Nations (2022). UN climate report: It’s ‘now or never’ to limit global warming to 1.5 degrees. Retrieved 08.11.2022, from <https://news.un.org/en/story/2022/04/1115452>.

Uysal, M., Sirgy, M. J., Woo, E., & Kim, H. L. (2016). Quality of life (QOL) and well-being research in tourism. Tourism Management, 53, 244-261.Viglia, G. & Dolnicar, S. (2020). A review of experiments in tourism and hospitality. Annals of Tourism Research, 80, 102858.

Zhu, O. Y., & Dolnicar, S. (2022). Can disasters improve the tourism industry? The role of normative, cognitive and relational expectations in shaping industry response to disaster-induced disruption. Annals of Tourism Research, 93, 103288.