“To Clean or Not to Clean?” Reducing Daily Routine Hotel Room Cleaning by Letting Tourists Answer This Question for Themselves

Abstract

Changing default settings has proven to be a powerful approach to influencing consumer decisions without denying consumers the possibility of choosing freely. This is only the second study investigating the effectiveness of defaults in tourism, and the first testing also the combined effect of default changes and pro-environmental appeals in the context of changing room cleaning defaults in hotels from automatic daily cleaning (with the choice of opting out) to no daily routine cleaning (with the choice of opt-in and requesting a free room clean every day). 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. Adding a pro-environmental appeal to the change in defaults did not further reduce room cleaning overall, but has an effect on certain segments of hotel guests.

Keywords

choice architecture defaults sustainability pro-environmental behavior hotel room cleaning

Introduction

Tourism is the fourth most polluting industry globally. Tourism contributes around 8% of greenhouse gas emissions, a contribution that is expected to increase to 12% by 2025 (Lenzen et al. 2018). The industry produced yearly more than 35 million tons of solid waste (UNEP and WTO 2012). The accommodation sector alone used 1.3 cubic kilometers of water in 2005 (Gössling 2005), much of it in water-scarce regions (Gössling 2002), with the average tourist using 300 liters of freshwater every single day (UNEP and WTO 2012). Given that tourism is among the fastest-growing and biggest economic sectors in the world (UNWTO 2018), it is likely that the negative environmental impacts of the tourism industry are much higher today. The number of international tourists doubled over the past two decades, reaching 1.2 billion people per annum in 2015, with a predicted 1.8 billion in 2030 (UNWTO 2018). Adding domestic tourism brings the number of tourist trips to between 6.2 billion and 7.2 billion annually (UNWTO 2016).

While the importance of changing consumer behavior in climate change mitigation is generally acknowledged (Girod, van Vuuren, and Hertwich 2014), changing consumer behavior plays an even more critical role in tourism because governments have little incentive to restrict tourism activity given the substantial contribution of the tourism industry to national GDPs. The tourism industry does not have any incentive to take pro-environmental action, unless such action reduces the operating cost. Where pro-environmental action risks reducing tourist satisfaction, and with it potentially demand, tourism industry is unable to reconcile pro-environmental initiatives with their mission of maximizing profits. It is the tourists themselves, therefore, who represent promising targets of climate change mitigation action, they “have the largest adaptive capacity of elements within the tourism system” (Gössling et al. 2012, p. 36).

Hotel room cleaning is one of the areas where hotels are trying to change their service provision. While these initiatives may be driven primarily by cost-saving potential, they still imply a significant reduction in the environmental burden of hotel operations. A number of low-cost hotels have already changed the way they operate: instead of providing routine daily room cleaning—which uses some 35 liters of water, 100 ml of chemicals, and 1.5 kWh of electricity in a four-star hotel (Dolnicar, Cvelbar, and Grün 2019a)—they do not by default clean rooms daily, but ask hotel guests whether they want their room cleaned at an additional fee. Even luxury hotels are making attempts to reduce daily hotel room cleaning. These hotels entice their guests to actively opt out and waive room cleans. Sheraton Hotels, for example, reward hotel guests with points in their Starwood Loyalty program “Make a Green Choice” and beverage gift cards worth $5 if they opt out of their daily room clean. Similar incentive-based programs have since been rolled out in a number of luxury hotels.

The idea of trying to entice guests to waive their room clean voluntarily is not new. What has not been studied to date, however, is the effect of changing the default without any attempts of enticing or rewarding tourists for the cheaper and more environmentally friendly option. It is also unclear whether—as would be expected based on theoretical models of pro-environmental behavior—the effect of changing defaults can further be increased by adding a pro-environmental appeal. This is where the knowledge contribution of the present study lies: we determine the effect of changing the default hotel room cleaning system from automatic daily cleaning (with the option to opt-out) to no cleaning (with the option of opting in at no cost). In addition, we test whether the effect of this default change can be strengthened by using pro-environmental arguments when explaining the hotel’s room cleaning system to guests. Our hypothesis is that the number of room cleans will drop significantly when the default changes from gray to green, despite the fact that requesting a room clean in the green default option does not cost the guest anything. We also hypothesize that the pro-environmental appeal will further significantly reduce the number of hotel cleans above and beyond the default change effect. The resulting insights contribute to knowledge on how to change tourism to be more environmentally sustainable. They also offer immediate practical guidance to hotels on how to organize their room cleaning and how to communicate it to their guests.

This study also represents the starting point for future work investigating other opportunities to change defaults in hotels, which are currently leading to environmentally harmful tourist behavior, including default provision of soaps, shampoos, conditioners, and lotions in one-way containers, availability of large numbers of towels in hotel pool areas, use of luxurious thick cotton serviettes at breakfast tables (Dolnicar, Cvelbar, and Grün 2019b), and generous buffet breakfast offerings leading to food waste (Juvan, Grün, and Dolnicar 2018).

Literature Review

The theoretical concept at the center of this study is that of defaults. The default option is defined as behavior requiring “no action with regard to particular choice opportunity” (Davidai, Gilovich, and Ross 2012, p. 15201). Defaults represent the status quo. Defaults are easy to follow because they require less physical, cognitive, and emotional costs (Johnson and Goldstein 2003). The case of organ donation rates provides a good illustration of the effectiveness of defaults: in Austria, 99% of people are potential donors of organs; in Germany, only 12% are (Johnson and Goldstein 2003). This difference in organ donation between two European countries with similar cultural and economic background is due to defaults. Austria assumes that people will donate their organs in case they die. If they do not want to donate their organs, they need to declare this preference proactively. Organ donation is the default. Germany assumes that people do not wish to donate their organs unless they give explicit consent. People have to opt in proactively (Abadie and Gay 2006). The difference in defaults explains the 87 percentage points higher organ donation level in Austria.

Causing environmental damage in the tourism context is, in some ways, similar to donating organs. When we drive a car, organ donation is not top of our mind. When we go on vacation, we do not primarily think about climate change and how we can keep environmental damage to a minimum. Tourists do not make active decisions to harm the environment. Rather, some of the harm done to the environment by the tourism industry is only a consequence of the defaults offered to tourists (Sunstein and Reisch 2013). Changing default options can change human behavior (Leonard 2008) and has been specifically recommended as a promising climate change mitigation strategy (Girod, van Vuuren, and Hertwich 2014). Empirical proof of the effectiveness of changing defaults was provided in a number of empirical studies across a range of human behaviors, including, as mentioned above, organ donation (Abadie and Gay 2006; Johnson and Goldstein 2004; Leonard 2008), money donations to charities (Everett et al. 2015), selection of health insurance plans (Samuelson and Zeckhauser 1988), and suing as a consequence of a car accident (Johnson et al. 1993).

The concept of the default stems from libertarian paternalism. Libertarian paternalism is a philosophy introduced to the economic literature by Sunstein and Thaler (2003). Paternalism is “interference with a person’s liberty of action justified by reasons referring exclusively to the welfare, good, happiness, needs, interests, or values of the person being coerced” (Dworkin, 1971, p. 108). Sunstein and Thaler (2003) added “libertarian” to emphasize that these choices are “enlisted in the interest of vulnerable third parties” (2003, p. 1162), advocating that public and private institutions should direct people’s choices to increase their welfare without eliminating their freedom of choice. This can be achieved through choice architecture. Choice architecture organizes the situational context in which people display specific behaviors (Thaler and Sunstein 2008).

Defaults are believed to be effective for three reasons (Smith, Goldstein, and Johnson 2009): implied endorsement, cognitive bias, and effort. Implied endorsement means that people interpret defaults as recommended action by policy makers (McKenzie, Liersch, and Finkelstein 2006). They assume that most people will therefore choose this option (Sunstein and Thaler 2003), making it a good one. Cognitive bias means that people take the defaults for granted and worry they may experience loss by not behaving in line with the default (Smith, Goldstein, and Johnson 2009). Effort means that people tend to take the path of least resistance. The default option is, by definition, the path of least resistance, requiring minimum effort (Smith, Goldstein, and Johnson 2013).

Using changes in defaults to direct people’s behavior has been criticized as limiting freedom of choice (Mitchell 2004). Concerns have also been raised about policy makers using changes in defaults without increasing the well-being of individuals, thus violating consumer autonomy through manipulation (Smith, Goldstein and Johnson 2009). Google Play is one example: Google Play dominates the market of Android app stores. Google allowed Android phone and tablet manufacturers to put the Google Play app on their devices under the condition that they include Google search as the default search option, making Chrome the default browser. Consumers were not forced to use Chrome, but only 1 in 10 users downloaded an alternative browser and only 1 in 100 downloaded an alternative search app. The European Commission fined Google in 2018 for abuse of a dominant market position and misuse of default options (EC 2018).

Whether used to improve or reduce welfare, modifying default options changes human behavior (Leonard 2008), and represents a promising approach for climate change mitigation (Girod, van Vuuren, and Hertwich 2014). Green default options can increase human welfare and help save the environment for future generations. An example of a green default is German consumers’ increased willingness to pay for expensive green energy when it is the default: 99% of residents in Schönau in Germany—where green energy is supplied by default—use electricity from green sources. In a similar city with a gray energy supply as default, only 1% of citizens chose green energy (Kaenzig, Heinzle, and Wüstenhagen 2013).

In tourism, green defaults are not the norm. Some of the negative environmental consequences of tourism result from industry defaults (Sunstein and Reisch 2013). In hotels, for example, daily room cleans still represent the default. Hotels provide guests with soaps, shampoos, conditioners, and lotions in one-way containers by default. Hotels offer large breakfast buffets by default, and they stack large numbers of pool towels by default, suggesting that guests are welcome to use as many towels as they please.

To the authors’ knowledge, only one academic study has investigated the effect of defaults in tourism to date. Araña and León (2013) tested whether an opting-in or an opting-out model of carbon offsetting for flights is more effective. Typically, passengers are informed about the CO₂ emissions their flight causes, and invited to voluntarily pay to offset them (opt-in default). Alternatively, the carbon offset amount can be built into the price, with passengers having the possibility of opting out to avoid paying the additional cost (opt-out default). These two default conditions were compared from 2009 to 2011 for 1,680 attendees of conventions and conferences in Gran Canaria. Carbon offsets were higher under the opt-out condition, suggesting that current voluntary carbon offsetting schemes are not optimally designed.

Two other studies provide support that default changes are likely to work, but neither of those studies tested true default manipulation. The study proving that reducing the default buffet breakfast plate size by 3 cm leads to 20% less plate waste (Kallbekken and Sælen 2013) was not, strictly speaking, a study on defaults, because hotel guests had no choice. They were either provided with the large or with the small plate. If the default was the small plate, guests had no possibility to get a large plate. This intervention, therefore, can be classified as a change in infrastructure rather than in choice architecture. A study testing the effect of placing recycled paper serviettes on breakfast buffet tables instead of—less environmentally friendly—thick cotton serviettes (Dolnicar, Cvelbar, and Grün 2019b) showed that the change in the service setting reduced cotton serviette use by 95%, but hotel guests were only able to choose both serviette types in one condition. When the cotton serviette was provided, recycled paper serviettes were unavailable, again restricting the full choice of hotel guests. Despite neither of those two studies being fully compliant with the definition of defaults, they do suggest that substantial effects can be achieved by changing service settings, indicating that the study of defaults in the tourism context may be promising.

Much past research has studied the effectiveness of pro-environmental appeals. Such appeals can be verbal or nonverbal communications aiming at motivating people to display a behavior that does not harm the environment (Bolderdijk et al. 2013; Evans et al. 2013). Pro-environmental appeals are designed to activate norms and values or change attitudes, typically using as their theoretical basis norm-activation theory (Schwartz 1977) or value–belief–norm theory (Stern 2000). According to norm activation theory (Schwartz 1977), people’s evaluation of right and wrong is driven by moral considerations. Personal norms mediate situational and personality factors, which, together, drive behavior. According to Stern’s (2000) value–belief–norm theory, environmentally friendly behavior is a consequence of people’s environment-related values, their sense of responsibility for protecting nature, and their personal norms. This theory identified two key beliefs suitable for the development of interventions aimed at behavioral change: awareness of consequences and ascription of responsibility. But Stern’s (2000) theory has not been developed for behavioral contexts which are driven primarily by the pursuit of pleasure. While tourists state that they want to reduce the environmental footprint of their vacations, there is not much empirical evidence that they are willing to compromise their holiday enjoyment in view of environmental consequences of their vacation choices (Miller et al. 2010). Consequently, attempts of convincing tourists to behave more environmentally friendly have not proven to be very effective in changing behavior (Dolnicar, Cvelbar, and Grün 2017). In everyday life, many pro-environmental behaviors are habitual (Lavelle, Rau, and Fahy 2015). These habits typically cannot be transferred to the vacation context (Barr, Shaw, and Coles 2011) because the required infrastructure may not be available, and because vacations are fundamentally about maximizing pleasure, not saving the environment. As a consequence, the alignment of people’s norms and values with their behavior may be weaker in the vacation context than it is in everyday life; tourists can forgive themselves for behaving in environmentally unsustainable ways (Juvan and Dolnicar 2014) that are “socially and morally acceptable by peers of those travelling” (Buckley 2011, p. 1180). As a consequence, people tend to give up habits resulting in positive environmental consequences (Dolnicar and Grün 2009). Not surprisingly, therefore, successful interventions of behavioral change developed for the everyday living context (Bolderdijk et al. 2013; Evans et al. 2013; Taufik, Bolderdijk, and Steg 2015; Van der Linden 2015) fail in tourism (Dolnicar, Cvelbar, and Grün, 2019a), and tourists are not able to accurately report their pro-environmental vacation behaviors, typically substantially overestimating it (Karlsson and Dolnicar, 2016). To the authors’ knowledge, no studies in the tourism context have proven that altering beliefs influences actual behavior with environmental consequences. Even in the everyday living context, a recent meta-analysis of 171 academic studies concludes that beliefs are only moderately associated with climate change–related behavior (Hornsey et al. 2016).

Changing default behavioral options available to tourists—and thus not having to rely on tourists’ beliefs and values to trigger environmentally friendly behavior—may therefore represent a promising avenue to inducing behavioral change for more sustainable tourism. Defaults are what happens when people do not request a different option (Brown and Krishna, 2004). The combined effect of both changing the default and pro-environmental appeals has not been studied to date. Therein lies one of the contributions of our study. Finally, we build on prior work that has identified differences across market segments with respect to pro-environmental behavior on vacations (Knezevic Cvelbar, Grün, and Dolnicar 2017). We check if behavioral changes vary across market segment in response to the interventions in this study.

Method

We ran a quasi-experimental investigation in a city hotel located in the center of Ljubljana. Ljubljana is the capital of Slovenia, which has been experiencing significant tourism growth over the last few years. In 2016, Ljubljana was declared the Green Capital of Europe, and the United Nations named Slovenia the cleanest country in the world in 2018. We conducted the study in the three-star Hotel Park, which has 192 rooms. The hotel guest structure is representative of the guest mix of comparable three-star city hotels in Ljubljana.

We measured room cleaning rates under three study conditions: a control condition corresponding to the status quo and two experimental conditions. The control condition—the gray default—involved automatic daily room cleaning. Guests could opt out from room cleaning every day by using the “Please do not clean my room” door sign. Experimental condition 1 (EC1)—the green default without pro-environmental appeal—involved cleaning only if the hotel guest placed the “Please clean my room today” sign outside their room door. Information about the room cleaning practice in the hotel was provided to guests at their arrival. We expect—in line with theory on the functioning of defaults—that the green defaults will lead to a significantly lower room cleaning rate than the gray default (hypothesis 1). The exact wording of this information was as follows:

Experimental condition 2 (EC2)—the green default with pro-environmental appeal—involved the same procedure, but with a different information message, a pro-environmental appeal. The information message was based on Stern’s value–belief–norm theory of environmentalism (Stern 2000) and emphasized awareness of consequences and ascription of responsibility. In EC2, hotel guests were not only informed about how room cleaning works in the hotel, but were also provided with an environmental argument for not cleaning the room when it is not necessary. We expect that this pro-environmental appeal will further reduce the number of requests for room cleaning (hypothesis 2) because two mechanisms are at work: the default and activation of key beliefs known to drive pro-environmental behavior. The exact wording was as follows:

We are testing a new room-cleaning program and will be cleaning the rooms upon request in July and August 2017. This means that we will not automatically clean your room every day. But if you would like us to clean your room, we are happy to do so. All you need to do is to place the “Please clean my room today” sign on the outside handle of your door before 10 am. Please note that every time we clean a room we use 100 ml of chemicals, 35 l of water and 1.5 kWh of electricity, which is not good for the environment. You can make a difference and reduce the environmental burden of your stay by having your room cleaned upon request. Please help us make a difference to the environment.

Professionally designed information materials were available in English, German, Italian, and Slovenian. All reception and cleaning employees—who remained the same during the study period—were trained before the study commenced to ensure all guests received the intended information.

The experimental conditions were implemented sequentially, with changeovers occurring on Mondays. Only guests classified as leisure or business guests were included; people who were part of an organized tour group were excluded. The reason for excluding tour groups was that routine room cleaning was a part of the contract with the tour operator. The room cleaning default therefore could not be changed for these groups. Excluded from analyses were also guests who stayed for one night only.

The university’s human ethics committee approved the study under approval number 2017000985. A debriefing was not required, but the committee did require that guests sign a consent form acknowledging the anonymous inclusion of the data collected from them in this study. In the experimental conditions, the receptionist informed the hotel guests—after check-in—about the room cleaning procedures, and gave them written information material including a flyer, and a door sign that allowed hotel guests to express their room cleaning preference. Hotel guests could choose every single day whether or not they wanted their room cleaned.

Data were collected for 27 days in July and August, during the peak tourist months in Ljubljana, and the majority of hotel guests are leisure tourists. We have intentionally conducted the study during those months to ensure the lowest possible variation in guest mix across study conditions. The hotel provided—in anonymized form—information about the registered guest’s age, check-in date, check-out date, length of stay, type of guest (leisure, business), number of adults in the room, and number of children in the room. Room cleaning data were merged with these guest data. These data were initially analyzed descriptively, considering only room cleaning rate by length of stay (two nights vs. longer stays) and study condition. The room cleaning rate is determined for each guest party by dividing the number of room cleans by the length of stay, after excluding the last night and room clean from this calculation.

Next, we fitted a mixed effects logistic regression model using each potential room clean as unit of measurement. The binary indicator variable if the room is cleaned or not was included as the dependent variable; the study condition as independent variable; and single occupancy, type of guest, length of stay, and age as additional covariates. We included the guest party identifier to fit a random effect model accounting for the repeated measurements for guest parties staying longer than two nights. We fitted the model using maximum likelihood estimation to determine the point estimates for the regression coefficients, with standard errors based on standard asymptotic theory (Bates et al. 2015). We compared models including different sets of covariates and interactions using likelihood ratio tests. Predicted mean room cleaning rates (with 95% confidence intervals) are calculated for each study condition in combination with different guest party characteristics.

Data relating to 616 guest parties are available for analysis. These guest parties accounted for 989 observed overnight stays where they either had their room cleaned or not, that is, they could either opt out of room cleaning in the gray default (the control condition) or opt in to have their room cleaned in the green default (experimental conditions EC1 and EC2). On average 89% of these room cleans took place under the gray default, whereas only 32% of these room cleans were made under the green default.

Table 1 summarizes, for the total sample and the control and the experimental conditions, basic characteristics of guest parties: length of stay in days, age of the registered guest in years, single occupancy of the room, and type of guest (business or leisure). The average length of stay is 2.9 days (standard deviation 1.6 days), with the length differing considerably between control and experimental conditions (Kruskal-Wallis rank sum test: χ² = 87.2, df = 2, p value < 0.001). The average age of the registered guest for the room is 43.5 years (standard deviation 15.1 years) with the median age being similar across study conditions (Kruskal-Wallis rank sum test: χ² = 3.4, df = 2, p value = 0.19). The percentage of guest parties consisting of a single adult is 28.4% in the total sample, with substantial variation across study conditions (χ² = 11.0, df = 2, p value = 0.004). The overall percentage of business travelers is 14.9%, with again substantial variation across study conditions (χ² = 56.9, df = 2, p value < 0.001). This initial analysis suggests that guest composition across experimental conditions is only comparable for age, but not for length of stay, occupancy, and guest composition. Results comparing the study conditions where these covariates are not taken into account therefore need to be interpreted with care and are of exploratory nature. This also indicates the need to fit the regression model including these covariates as controls.

Table 1.

Descriptive Characteristics of Guest Parties Included in the Analysis.

	Total	Control	EC1	EC2
Number of guest parties	616	102	221	293
Length of stay (days)
Mean	2.9	3.8	2.9	2.5
Standard deviation	1.6	2.2	1.8	1.1
Age (years)
Mean	43.5	43.2	42.1	44.6
Standard deviation	15.1	14.7	14.6	15.6
Single occupancy, %	28.4	20.6	36.2	25.3
Business traveler, %	14.9	6.9	29.4	6.8

Results

Figure 1 contains the distribution of room cleaning rates for each guest party by study condition (Control, EC1, and EC2) split by guest parties staying only for two nights (short stay vs. long stay). The distribution is visualized by a histogram where the room cleaning rates are binned and for each bin the percentage of observations falling into this bin is determined. The x-axis shows the room cleaning rate and the y-axis the percentage of guest parties. Each rectangle corresponds to one bin. The position on the x-axis indicates the room cleaning rate; the height indicates the percentage of guest parties with this room cleaning rate for the subgroup of study condition and length of stay considered.

Figure 1.

Distribution of room cleaning rate by length of stay (short stay of two nights, long stay of more than two nights) and condition.

Guests who stay only for two nights (bottom row in Figure 1) only have one room clean, which they could potentially waive. Thus, for short stays the only possible values for the room cleaning rate are zero or one. The percentage of guest parties asking for their room to be cleaned in the two experimental conditions is 22% and 28%, respectively. In the control group, 57% of guest parties have their rooms cleaned.

The differences between the gray and the green defaults are even more extreme for guest parties staying longer than two nights (top row in Figure 1): nearly all guest parties (98%) have their rooms cleaned all the times in the control condition. In the experimental conditions, three main groups emerge: (1) guests who always waive their room clean (42% in EC1 and 53% in EC2), (2) guests who always have their rooms cleaned (16% in EC1 and 12% in EC2), and (3) guests who have their room cleaned about half the time (26% in EC1 and 28% in EC2). Overall, this exploratory analysis provides empirical support for hypothesis 1.

The mixed effects logistic regression model is fitted with the study conditions as independent variable, as well as the covariates single occupancy, type of guest, short stay and age, and their interaction with the study conditions. A stepwise procedure eliminates covariates with insignificant contributions and insignificant interactions. This results in a model containing the main effects of the study conditions, the main effects of short stay and type of guest, and their interactions with the study conditions and the main effect of single occupancy. The model comparison based on the likelihood ratio test indicates an equally good fit of this reduced model (χ² = 3.9, df = 5, p value = 0.56). Table 1 provides the estimated regression coefficients together with the standard errors and the asymptotic z tests for this smaller model. The intercept captures the effect in the control condition for leisure travelers traveling with others and staying for more than two nights. The regression coefficients indicate how much the log odds change. The sign indicates if the experimental condition or the covariate value leads to an increase (positive sign) or decrease (negative sign) in the room cleaning rate. This model is also compared to the reduced model, including only the same main effects, but without the interaction effects between study conditions and the other covariates. This model provides a significantly worse fit (χ² = 58.9, df = 4, p value < 0.001), but would indicate that the effect on reducing the number of room cleans is the same for EC1 and EC2 (estimated difference = −0.341, z value = −1.215, p value = 0.22). This highlights that the effect of environmental appeals needs to be investigated for guest segments separately. Hypothesis 2 is not empirically supported at the aggregate level, but does hold for some segments.

Figure 2 visualizes the predicted probability of room cleaning for each study condition and covariate value combination together with 95% confidence intervals to ease interpretation of the effects estimated by the regression model. The figure consists of four panels where each panel contains the predictions for a combination of having a short stay with being a business or leisure traveler. Within each panel predictions are for single travelers versus larger guest parties and study condition. The values are joined for the different study conditions to better indicate how the predicted probabilities change. Predicted probabilities of a room clean are joined by full lines for single travelers, while dashed lines are used for larger guest parties.

Figure 2.

Predicted probability of room cleaning (with 95% confidence intervals) by experimental condition, being a business or leisure traveler, single occupancy, and length of stay (with a stay of two nights being classified as short stay).

Single travelers have a significantly lower probability of having their room cleaned than people traveling with others. This is clear from Figure 2 because the solid black line for singles is consistently lower than the dashed line for nonsingles. In addition, the estimated coefficient for being single in Table 2 is equal to −1.105, implying that the odds of having the room cleaned are reduced by a factor of 0.331 regardless of the study condition as no interaction effect is included. Our model predicts that only 1.2% of room cleans are made for single business travelers on a short trip in the control condition, compared to 3.5% for those not traveling alone. Under the green default without pro-environmental appeal, 2.8% of room cleans are predicted to take place for singles and 8.0% for those not traveling alone under those same conditions. And under the green default with pro-environmental appeal, 15.7% of room cleans for singles and 36.0% for those traveling with company in this guest segment are predicted to be made.

Table 2.

Regression Analysis Results.

	Estimate	Standard Error	z Value	p Value
Intercept	5.701	0.872	6.541	<0.001
Single traveler	−1.105	0.294	−3.756	<0.001
Condition EC1	−5.592	0.925	−6.049	<0.001
Condition EC2	−6.767	0.953	−7.102	<0.001
Short stay (≤2 nights)	−6.268	1.044	−6.003	<0.001
Business traveler	−2.743	1.488	−1.843	0.065
Interaction condition EC1 and short stay	5.085	1.083	4.697	<0.001
Interaction condition EC2 and short stay	6.265	1.095	5.724	<0.001
Interaction condition EC1 and business traveler	1.367	1.551	0.882	0.378
Interaction condition EC2 and business traveler	3.222	1.598	2.017	0.044

Short stays also reduce cleaning significantly in the control condition. This can be seen in Figure 2 because the points for the control condition in the left two quadrants are lower than those in the right two quadrants. The estimated coefficients (Table 2) indicate that in the control condition the log odds are reduced by −6.268, which corresponds to a reduction in odds by a factor of 0.002. This effect of short stays is reduced for the experimental conditions where room cleaning rates are rather comparable regardless of length of stay.

In the control condition, business travelers generally do not require room cleaning as much as leisure tourists. In Figure 2, this is visible as generally lower points for the control condition in the bottom two quadrants. According to the estimated coefficient given in Table 2, the log odds are reduced by −2.743, which corresponds to a reduction of odds by a factor of 0.064. Again, this effect is reduced for the experimental conditions, even though this reduction is not significant in case of experimental condition 1.

The largest guest segment—leisure travelers on long stays sharing the room with someone else—accounts for 43% of the potential room cleans in the sample. For this segment, the predicted room cleaning rate under the gray default (the control group) is 99.7%, under the green default without pro-environmental appeal (EC1) 52.7% and under the green default with pro-environmental appeal (EC2) 25.6%. The second largest guest segment contains 24% of the potential room cleans in the sample and consists of leisure travelers on short stays sharing the room with someone else. For this segment, the predicted room cleaning rate under the gray default (the control group) is 36.2%, under the green default without pro-environmental appeal (EC1) 25.4% and under the green default with pro-environmental appeal (EC2) 25.7%.

Overall the model predicts a room cleaning rate of 72% if all guests in the sample had been assigned to the gray condition (the control group) compared to 34% if all were assigned to the green default without pro-environmental appeal (EC1) and 22% if all were assigned to the green default with pro-environmental appeal (EC2). This means that changing the defaults from gray to green substantially reduced the amount of room cleaning in the hotel, despite the fact that hotel guests do not need to pay for the room clean in either of those two defaults (lending empirical support to hypothesis 1). The additional effect of the pro-environmental appeal varies substantially across guest groups, with two groups most influenced by the pro-environmental appeal: long-stay leisure tourists sharing a room with someone else being followed by the guest segment of long-stay leisure tourists traveling alone. These findings suggest, overall, hypothesis 2—that pro-environmental appeals significantly influence pro-environmental behavior in the tourism context—is not empirically supported at the aggregate level, but does hold for some segments.

Conclusions

Results from a quasi-experimental study conducted in the field lead to the conclusion that changing gray defaults to green defaults in the hotel room cleaning context represents a promising strategy to reducing the environmental harm resulting from accommodation service provision without reducing guest satisfaction. Defaults imply consumer choice. In our gray default condition, hotel guests had the choice of taking no action and having their room cleaned on a daily basis, or taking action to prevent their room from being cleaned on selected days. In our green default conditions, hotel guests had the choice of taking no action and not having the room cleaned, or taking action to request room cleaning on selected days. Importantly, and in contrast to how the green default is implemented in low-cost hotels, the hotel guests did not have an advantage and did not suffer any disadvantage as a consequence of their choice; they did not have to pay to have their room cleaned. The change from the gray to the green default reduced the percentage of cleans to 32% of room cleans being requested when averaging across both green default conditions. This finding is strong evidence of the value of changing defaults in tourism service settings, supporting our hypothesis about the effect of defaults.

Our hypothesis in relation to the additional beneficial effect of using pro-environmental appeals when communicating the green default hotel cleaning model was not generally supported. Overall, the difference in predicted cleaning rates under those two conditions was insignificant. Some market segments responded to the pro-environmental appeal; in other cases, the green default without pro-environmental appeal was more successful in reducing cleaning rates. This is a theoretically important finding that supports conclusions by Dolnicar, Cvelbar, and Grün (2017) that the effectiveness of changing pro-environmental beliefs—postulated by most theories of environmentally significant behavior to serve as key antecedents to pro-environmental behavior—may be overrated at present. If this is indeed the case, hotels should move away from using pro-environmental appeals (such as requests to reuse towels and bedlinen to protect the environment) and instead explore opportunities to change service defaults.

Our study has one key limitation: we conducted the study in one single hotel with very specific features: a three-star rated hotel in a city setting. It is possible that results would be different in other contexts. Future work should therefore replicate this study across additional contexts to determine if findings generalize to other hotels. There are also ample opportunities for important future work in identifying other hotel service defaults, and defaults in the tourism industry more generally, that could be changed without reducing the level of service provision to guests. Those should be identified and systematically studied in field experiments.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We thank the Australian Research Council for support under grant DP180101855, Slovenian Research Agency (ARRS) and Austrian Science Fund (FWF) for support under grant J5-1783 and Slovenia Research Agency (ARRS) for support under grant P5-0128.

ORCID iDs

Ljubica Knezevic Cvelbar

Sara Dolnicar

Author Biographies

Ljubica Knezevic Cvelbar is an Associate Professor of Tourism at School of Business and Economic, University of Ljubljana in Slovenia. Her research is in tourism economics and sustainable tourism development in general and lately in triggering pro-enviromnetal behavior in tourism.

Bettina Grün is an Associate Professor of Applied Statistics at the Johannes Kepler University Linz in Austria. Her research in statistics focuses on model-based clustering working on methodological as well as computational issues and the implementation in statistical software, but she also enjoys contributing as applied statistician to empirical research in different areas of application including tourism, economics and business.

Sara Dolnicar is a Professor of Tourism at The University of Queensland in Australia. She is best known for her work improving market segmentation methodology, but is currently working on developing and experimentally testing interventions that make tourists behave more environmental friendly.

References

Abadie

Gay

2006. “The Impact of Presumed Consent Legislation on Cadaveric Organ Donation: A Cross-Country Study.” Journal of Health Economics 25 (4): 599–620.

Araña

J. E.

León

C. J.

2013. “Can Defaults Save the Climate? Evidence from a Field Experiment on Carbon Offsetting Programs.” Environmental and Resource Economics 54 (4): 613–26.

Barr

Shaw

Coles

2011. “Times for (un) sustainability? Challenges and Opportunities for Developing Behaviour Change Policy. A Case-Study of Consumers at Home and Away.” Global Environmental Change 21 (4): 1234–44.

Bates

Mächler

Bolker

Walker

2015. “Fitting Linear Mixed-Effects Models Using lme4.” Journal of Statistical Software 67 (1): 1–48.

Bolderdijk

J. W.

Steg

Geller

E. S.

Lehman

P. K.

Postmes

2013. “Comparing the Effectiveness of Monetary versus Moral Motives in Environmental Campaigning.” Nature Climate Change 3 (4): 413–16.

Brown

C. L.

Krishna

2004. “The Skeptical Shopper: A Metacognitive Account for the Effects of Default Options on Choice.” Journal of Consumer Research 31 (3): 529–39.

Buckley

2011. “20 Answers: Reconciling Air Travel and Climate Change.” Annals of Tourism Research 38 (3): 1178–81.

Davidai

Gilovich

Ross

L. D.

2012. “The Meaning of Default Options for Potential Organ Donors.” Proceedings of the National Academy of Sciences 109 (38): 15201–05.

Dolnicar

Cvelbar

Grün

2017. “Do Pro-environmental Appeals Trigger Pro-environmental Behavior in Hotel Guests?” Journal of Travel Research 56 (8): 988–97.

10.

Dolnicar

Cvelbar

Grün

2019a. “A Sharing-Based Approach to Enticing Tourists to Behave More Environmentally Friendly.” Journal of Travel Research 58 (2): 241–52.

11.

Dolnicar

Cvelbar

Grün

2019b. “Changing Service Settings for the Environment: How to Reduce Negative Environmental Impacts without Sacrificing Tourist Satisfaction.” Annals of Tourism Research 1–30. doi.org/10.1016/j.annals.2018.08.003.

12.

Dolnicar

Grün

2009. “Environmentally Friendly Behaviour: Can Heterogeneity among Individuals and Contexts/Environments Be Harvested for Improved Sustainable Management?” Environment and Behavior 41 (5): 693–714.

13.

Dworkin

1971. “Paternalism.” In: Morality and the Law, edited by R. A. Wasserstrom. Belmont, SA: Wadsworth.

14.

EC (European Commission). 2018. Press release. http://europa.eu/rapid/press-release_IP-18-4581_en.htm (accessed December 12, 2018).

15.

Evans

Maio

G. R.

Corner

Hodgetts

C. J.

Ahmed

Hahn

2013. “Self-interest and Pro-environmental Behaviour.” Nature Climate Change 3 (2): 122–25.

16.

Everett

J. A.

Caviola

Kahane

Savulescu

Faber

N. S.

2015. “Doing Good by Doing Nothing? The Role of Social Norms in Explaining Default Effects in Altruistic Contexts.” European Journal of Social Psychology 45 (2): 230–41.

17.

Girod

van Vuuren

D. P.

Hertwich

E. G.

2014. “Climate Policy through Changing Consumption Choices: Options and Obstacles for Reducing Greenhouse Gas Emissions.” Global Environmental Change 25:5–15.

18.

Gössling

2002. “Global Environmental Consequences of Tourism.” Global Environmental Change 12 (4): 283–302.

19.

Gössling

2005. “Tourism’s Contribution to Global Environmental Change: Space, Energy, Disease and Water.” In Tourism Recreation and Climate Change: International Perspectives, edited by Hall and Higham. Clevedon, UK: Channel View.

20.

Gössling

Scott

Hall

C. M.

Ceron

J. P.

Dubois

2012. “Consumer Behaviour and Demand Response of Tourists to Climate Change.” Annals of Tourism Research 39 (1): 36–58.

21.

Hornsey

M. J.

Harris

E. A.

Bain

P. G.

Fielding

K. S.

2016. “Meta-analyses of the Determinants and Outcomes of Belief in Climate Change.” Nature Climate Change 6 (6): 622–26.

22.

Johnson

E. J.

Goldstein

2003. “Do Defaults Save Lives?” Science 302 (5649): 1338–39.

23.

Johnson

E. J.

Goldstein

D. G.

2004. “Defaults and Donation Decisions.” Transplantation 78 (12): 1713–16.

24.

Johnson

E. J.

Hershey

Meszaros

Kunreuther

1993. “Framing, Probability Distortions, and Insurance Decisions.” Journal of Risk and Uncertainty 7 (1): 35–51.

25.

Juvan

Dolnicar

2014. “The Attitude–Behaviour Gap in Sustainable Tourism.” Annals of Tourism Research 48 (1): 76–95.

26.

Juvan

Grün

Dolnicar

2018. “Biting off More Than They Can Chew: Food Waste at Hotel Breakfast Buffets.” Journal of Travel Research 57 (2): 232–242.

27.

Kaenzig

Heinzle

S. L.

Wüstenhagen

2013. “Whatever the Customer Wants, the Customer Gets? Exploring the Gap between Consumer Preferences and Default Electricity Products in Germany.” Energy Policy 53 (1): 311–22.

28.

Kallbekken

Sælen

2013. “‘Nudging’ Hotel Guests to Reduce Food Waste as a Win–Win Environmental Measure.” Economics Letters 119 (3): 325–27.

29.

Karlsson

Dolnicar

2016. “Does Eco Certification Sell Tourism Services? Evidence from a Quasi-Experimental Observation Study in Iceland.” Journal of Sustainable Tourism 24 (5): 694–714.

30.

Knezevic Cvelbar

Grün

Dolnicar

2017. “Which Hotel Guest Segments Reuse Towels? Selling Sustainable Tourism Services through Target Marketing.” Journal of Sustainable Tourism 25 (7): 921–34.

31.

Lavelle

M. J.

Rau

Fahy

2015. “Different Shades of Green? Unpacking Habitual and Occasional Pro-environmental Behaviour.” Global Environmental Change 35:368–78.

32.

Lenzen

Sun

Y. Y.

Faturay

Ting

Y. P.

Geschke

Malik

2018. “The Carbon Footprint of Global Tourism.” Nature Climate Change 8 (6): 522–28.

33.

Leonard

T. C.

2008. “Richard H. Thaler, Cass R. Sunstein, Nudge: Improving Decisions about Health, Wealth, and Happiness.” Constitutional Political Economy 19 (4): 356–60.

34.

McKenzie

C. R.

Liersch

M. J.

Finkelstein

S. R.

2006. “Recommendations Implicit in Policy Defaults.” Psychological Science 17 (5): 414-20.

35.

Miller

Rathouse

Scarles

Holmes

Tribe

2010. “Public Understanding of Sustainable Tourism.” Annals of Tourism Research 37 (3): 627–45.

36.

Mitchell

2004. “Libertarian Paternalism Is an Oxymoron.” Northwestern University Law Review 99:1245.

37.

Samuelson

Zeckhauser

1988. “Status Quo Bias in Decision Making.” Journal of Risk and Uncertainty 1 (1): 7–59.

38.

Schwartz

S. H.

1977. “Normative Influences on Altruism.” Advances in Experimental Social Psychology 10:221–79.

39.

Smith

N. C.

Goldstein

D. G.

Johnson

E. J.

2009. “Smart Defaults: From Hidden Persuaders to Adaptive Helpers.” INSEAD Business School Research Paper, 3.

40.

Smith

N. C.

Goldstein

D. G.

Johnson

E. J.

2013. “Choice without Awareness: Ethical and Policy Implications of Defaults.” Journal of Public Policy & Marketing 32 (2): 159–72.

41.

Stern

2000. “Toward a coherent theory of environmentally significant behaviour.” Journal of Social Issues 56 (3): 407–424.

42.

Sunstein

C. R.

Reisch

L. A.

2013. “Green by Default.” Kyklos 66 (3): 398–402.

43.

Sunstein

C. R.

Reisch

L. A.

2014. “Automatically Green: Behavioral Economics and Environmental Protection.” Harvard Environmental Law Review 38:127.

44.

Sunstein

C. R.

Thaler

R. H.

2003. “Libertarian Paternalism Is Not an Oxymoron.” University of Chicago Law Review 70 (4): 1159–202.

45.

Taufik

Bolderdijk

J. W.

Steg

2015. “Acting Green Elicits a Literal Warm Glow.” Nature Climate Change 5 (1): 37–40.

46.

Thaler

R. H.

Sunstein

C. R.

2008. Nudge: improving decisions about health, wealth, and happiness. New Haven, CT: Yale University Press.

47.

UNEP and WTO (United Nations Environment Programme and World Tourism Organization). 2012. Tourism in the Green Economy—Background Report. Madrid: UNWTO.

48.

UNWTO (United Nations World Tourism Organization). 2018. UNWTO Tourism Highlights, 2017. Madrid: UNWTO.

49.

Van der Linden

. 2015. “Intrinsic Motivation and Pro-environmental Behaviour.” Nature Climate Change 5 (7): 612–13.