The limited impact of positive cueing on pro-environmental choices

A B S T R


Introduction
Positive cueing has been presented as a social marketing technique to promote environmentally sustainable consumer behavior: When common environmentally beneficial behaviors (e.g., not littering, reusing grocery bags, sorting trash, turning off the light) are cued as proenvironmental, consumers increasingly view themselves as concerned with the environment, which subsequently results in more environmentally friendly choices (Cornelissen, Pandelaere, Warlop, & Dewitte, 2008).This reasoning is in line with classic work on self-perception (Bem, 1972) and consistency (Cialdini, Trost, & Newsom, 1995;Festinger, 1957;Freedman & Fraser, 1966).The positive cueing technique leads people to infer attitudes from observations of own overt behaviors (Bem, 1972) and as such leads to the self-perception as a certain type of person.The salience of this identity may in turn lead people to act consistently and as a result make additional pro-environmental choices, a "positive spillover" effect (Truelove, Carrico, Weber, Raimi, & Vandenbergh, 2014).
Other research, however, has shown inconsistency effects.For instance, people who exhibit environmentally friendly behavior at home are often the most likely to use modes of transportation with higher carbon footprint on holidays (Barr, Shaw, Coles, & Prillwitz, 2010).More recently, it has also been shown that purchasing green products leads to decreased intentions to engage in subsequent environmentally friendly behaviors (Meijers, Noordewier, Verlegh, Willems, & Smit, 2019).Thus, when people use (the environmentally friendly) public transport (instead of the environmentally unfriendly car) to commute to work, this choice can lead them to choose the plane (i.e., the non-sustainable choice option) over the train (i.e., the sustainable choice option) for a city trip.
Two different psychological mechanisms may explain this "negative spillover" effect.The first mechanism is moral licensing: Previous "good" behavior provides a boost in a moral, positive self-concept, which then increases the preference for an indulgent "bad" option (over a "good" option) since the commitment to previous virtuous behavior removes the negative self-attributions associated with the indulgent behavior (Khan & Dhar, 2006).As such, Miller and Effron (2010) point out that moral licensing explains how acting in one way (e.g., having a pro-environmental lifestyle) may give people the moral credentials to act in an inconsistent way (e.g., taking the plane instead of the train for a city trip).The second mechanism is motivated reasoning: People engage in motivated reasoning and search for reasons in support of their preferred option (Kunda, 1990).That is, people prefer the environmentally unfriendly option (e.g., taking the car instead of the bike) upfront and search for arguments to do so.In line with this, research on compensatory green beliefs has shown that consumers use green behavior (at a given time and in a given domain) as an excuse to engage in non-sustainable behavior at another time and/or in another consumption domain (e.g., using public transport can make up for flying abroad on holiday; Kaklamanou, Jones, Webb, & Walker, 2013).Thus, as previous environmentally friendly behavior may instigate the preference for non-sustainable choices, the positive cueing technique may backfire.Highlighting the pro-environmental nature of habitual behavior might provide the moral credentials and so the license to prefer non-sustainable choice options (cf.moral licensing) or turn out to be a tool that provides excellent excuses not to behave in a pro-environmental way if people are not immediately able to justify their choices (cf.motivated reasoning).

Two contrasting hypotheses
Summarized, based on a consistency perspective, positive cueing will lead people to see themselves as more environmentally conscious and aim to behave consistently with this self-perception, which results in an increase in the preference for sustainable choices (a positive spillover effect).However, based on either a moral licensing or motivated reasoning perspective, the positive cueing condition allows people to reduce the potential negative self-associations that come along with the non-sustainable behavior and may thereby decrease the preference for sustainable choices (a negative spillover effect).Formally, this leads to the following two contrasting hypotheses: Hypothesis 1. Positive cueing will increase the preference for a more sustainable choice.
Hypothesis 2. Positive cueing will increase the preference for a less sustainable choice.
While the different perspectives presented above either predict positive or negative spillover effects, it is also plausible that positive cueing does simply exert no influence on (non-)sustainable choices.First, although positive cueing originally resulted in a positive effect on sustainable choices which was attributed to consistency (Cornelissen et al., 2008), the original choice study suffered from a very small sample size and took place in one specific decision context (i.e., choices among a limited set of household items).Moreover, it has already been reported decennia ago that it is difficult to reproduce (classic) consistency effects from some seminal papers in social psychology (Cialdini et al., 1995).Second, the same applies for negative spillover effects.For instance, moral licensing effects of green behavior on both subsequent within-domain (pro-environmental) or out-of-domain (dishonest) behavior could recently not be replicated (Urban, Bahník, & Kohlová, 2019;Urban, Braun Kohlová, & Bahník, 2021).Therefore, to allow for a fair test of our hypotheses, we aim to make use of different studies with large sample sizes and use different dependent measures in different studies while keeping the positive cueing procedure constant.

Substitutability
The second aim of this paper is to provide insight into the specific conditions that lead either to a positive or negative spillover effect.Whether a spillover effect can be replicated may depend on some crucial factors.We posit that substitutability between the sustainable and nonsustainable choice options may be an important moderator for the presumed cueing effect, be it a positive or negative spillover effect.Substitutability is the similarity between choice options in their ability to reach their end goals (Goodman & Selin, 2012;Ratneshwar & Shocker, 1991).For instance, hunger can be reduced by either eating bread or eating pasta and therefore both pasta and bread can be seen as substitutes to reach the goal of reducing hunger.We present two contrasting hypotheses about the potential role of substitutability depending on the potential psychological mechanism at work (as presented above: consistency for a positive spillover effect; motivated reasoning for a negative spillover effect).

Substitutability and positive spillover
From a consistency perspective, a positive spillover effect may be more pronounced for more substitutable choice options if it is driven by the aim to behave consistently with an environmentally conscious selfperception (induced by the cueing).When an environmental conscious self-perception is activated, a non-sustainable choice may result in dissonance arousal, given that the behavior is inconsistent with this selfperception (Priolo et al., 2016).One way to either prevent or get rid of dissonance induced arousal is by the possibility to justify the inconsistent choice.As Aronson (2012, pp. 219-237) writes: "… the greater the personal commitment or self-involvement implied by the action and the smaller the external justification for that action, the greater the dissonance and, therefore, the more powerful the need for self-justification."(p.225).The arguments to justify an inconsistent choice are likely to be more readily available when both choice options are not seen as real substitutes (e.g., buying a sportier vs. a fuel-efficient car), but are much more difficult to come up with when choice options are substitutable (e. g., buying vegetables transported by plane vs. ship).Therefore, to avoid any dissonance, positive cueing may result in especially decreased preferences for non-sustainable choice options when a sustainable substitute is available (and thereby increase the choice for the sustainable option).Thus, following this reasoning that the positive self-perception induced by positive cueing may result in cognitive dissonance when making a non-sustainable, especially substitutable choice, we formally predict: Hypothesis 3. The increased preference for sustainable choice options in response to positive cueing is especially pronounced when substitutability between choice options is high.

Substitutability and negative spillover
While moral licensing and motivated reasoning both predict negative spillover effects, the different perspectives lead to different predictions regarding the potential moderating role of substitutability.We do not expect any moderating role of substitutability if such effects can be explained by a moral licensing perspective.Moral licensing presumes a boost in positive self-perception (Kouchaki, 2011) which allows people to reduce the negative self-associations that come along with the non-sustainable behavior and thus it should not matter if choices are substitutable or not.
From a motivated reasoning perspective, however, substitutability does matter.One may argue that people seek justification for choosing non-sustainable choice options, which positive cueing may provide.As people need a good argument to prefer one alternative over another (Shafir, Simonson, & Tversky, 1993), it is more difficult to justify a non-sustainable choice when both choice options are more substitutable.Since non-sustainable choices are more difficult to justify when they can be more easily substituted for their sustainable counterpart, the effect of positive cueing on the preference for non-sustainable choices should be more pronounced for more substitutable choice options because in this case positive cueing provides an excuse for the hard-to-justify non-sustainable behavior (e.g., "Hey, see I've already engaged in pro-environmental behaviors").In contrast, a non-sustainable choice can always be more easily justified when both choice options are less substitutable (e.g., when choosing between a luxurious hotel room and a basic eco-lodge).In that case, if positive cueing is a tool to justify a non-sustainable choice it should not exert much influence anymore.Formally, this brings us to the hypothesis that: Hypothesis 4. The increased preference for non-sustainable choice options in response to positive cueing is especially pronounced when substitutability between choice options is high.

Overview of studies
To test our contrasting hypotheses, we set up three experiments in which we investigate how the positive cueing technique influences preferences for (non-)sustainable choice options.Importantly, our dependent measure always consisted of one choice within a specific K. Millet et al. decision dilemma (contrasting a more vs.less sustainable choice option).We did not adopt a (more frequently used) dependent measure consisting of different sequential choices to prevent any spillover effects to emerge within this set of choices (and so reduce noise in the data).All studies were conducted in accordance with the Ethical guidelines of the American Psychological Association (APA).The data for all studies can be found on Research Box (https://researchbox.org/329).In our first preregistered study (N = 166) we provide evidence for a positive spillover effect (Hypothesis 1) when making use of an (adapted) item of the Pro-Environmental Behavior Task, a computerized paradigm to assess actual pro-environmental decision making (Lange, Steinke, & Dewitte, 2018).In our second preregistered study (N = 1615) we aim to generalize study 1 findings to an externally validated set of 81 choice pairs developed based on Moran et al. (2020) which cover various consumption domains.Each choice pair includes a sustainable and non-sustainable option and the different choice pairs vary on the extent to which the two choice options are substitutable.In this study, we do not find any support for one of our hypotheses as we do not observe a significant positive cueing effect nor an interaction with substitutability.We also do not observe a main effect of substitutability on preferences, suggesting that this factor does not matter when sustainable and non-sustainable choices are weighed against each other.In our final study (N = 708), we again find support for a positive spillover effect of positive cueing in one specific decision context (choosing between beverages).In partial support of Hypothesis 3, this effect emerges for consumers who have used both the sustainable and the non-sustainable option before (and for whom the two options thus may be seen as more substitutable).In summary, we do not obtain any evidence for a negative spillover effect and the empirical evidence for a positive spillover effect is only observed in particular decision contexts.We conclude that the beneficial effect of positive cueing probably exists, but that its practical relevance may be limited given that the effect is only found for specific choices and under specific conditions (discussed in more detail later).

Study 1
The aim of the first study is to test the first two contrasting hypotheses against each other.Based on a consistency perspective, positive cueing will lead people to see themselves as more environmentally conscious and aim to behave consistently with this self-perception.Therefore, positive cueing will increase the preference for sustainable choices (Hypothesis 1).However, positive cueing may also decrease the preference for sustainable choices (Hypothesis 2) by providing the moral credentials and so the license to prefer non-sustainable choice options (cf.moral licensing) or providing an excellent excuse for not behaving in a pro-environmental way if people are not immediately able to justify their choices (cf.motivated reasoning).

Materials and procedure
The experiment consisted of a cueing procedure and a choice task.We randomly assigned participants to two between-subjects conditions (positive cueing vs. control).In both conditions, participants were asked to indicate, on a seven-point scale ranging from "strongly disagree" to "strongly agree," whether they usually engage in each of five behaviors which were presented in a random order.In the positive cueing condition, the question was formulated as follows: "I usually engage in the following pro-environmental behavior," and the behaviors were five green behaviors that people regularly engage in: turning off the light, reusing grocery bags, not littering, sorting trash, and using reusable water bottles/cups/coffee mugs.The positive cueing items were constructed from a pretest and a pilot study detailed in Appendix A. Importantly, we followed the procedure as described in Cornelissen et al. (2008) to select these items.In the control condition, the question was worded as follows: "I usually engage in the following behavior," and the behaviors were five neutral, common behaviors that have no clear relation to environmentalism: watching Netflix series, listening to music, reading news articles, chatting with friends, and cooking.
Following the cueing manipulation, participants completed an adapted item from the Pro-Environmental Behavior Task (Lange et al., 2018). 3Participants were asked to choose between a car or a bike for going on a trip.They were told that after making their choice, they would need to wait for the trip to be completed before being able to proceed to the next survey question.The trip would take 5 s by car (and cause 16.25 g of CO2 emissions) or it would take 65 s by bike (without causing additional CO2 emissions).Participants indicated their choice by selecting "I'll go by bike" (with a stylized picture of a bicycle) or "I'll go by car" (with a stylized picture of a car).

Results
We run a chi square test to examine if the preference for the more (less) sustainable choice option depends on condition (positive cueing vs control).The results show a positive spillover effect of positive cueing.People in the positive cueing condition were more likely to choose the more sustainable option over the less sustainable alternative (76.25% chose the bicycle over the car) compared to the non-cueing condition (58.14% chose the bicycle over the car; chi 2 (1) = 6.136, p = .013,Cramer's V = .192;Odds Ratio = 2.312, 95% CI = [1.183,4.517]).A post hoc power analysis (using G*power; Faul, Erdfelder, Lang, & Buchner, 2007) suggests that a similar study with the currently observed effect size and sample size would have a power of 70.44%.

Discussion
In line with Hypothesis 1 (and in contrast to Hypothesis 2), the results in this study indicate a positive spillover effect of positive cueing.If participants first rate how frequently they perform five very common environmentally friendly behaviors, they are more inclined to select the more sustainable mode of transport (bicycle) over a less sustainable mode of transport (car) compared to participants who rate common behaviors that are not environmentally relevant.
The current study replicates the positive spillover effect first demonstrated by Cornelissen et al. (2008) in a different choice context (choosing a mode of transportation).However, it is not clear to what extent this finding is limited to the specific participant pool, which consists of Dutch students.In the Netherlands, traffic infrastructure is organized so that the bike is a convenient and generally accepted mode of transport and one of the things the Dutch are famous for is their culture to bike (Stoffers, 2012).For many citizens a bicycle is one of the regular ways of getting around and this holds especially for students (Plazier, Weitkamp, & van den Berg, 2017).Therefore, it seems likely 1 While we invited 260 students from our school's subject pool (the largest possible sample we had access to), we expected between 160 and 260 to participate (see preregistration).
2 86 participants in the positive cueing condition, 80 in the control condition.
3 While we used a dependent measure that was adapted from the Pro-Environmental Behavior Task, it is important to be aware that we only had 1 trial instead of 72 in the original task and that our task did not consume any energy (in contrast to this original task in which different trials consumed different amounts of energy and resulted in different levels of CO2 emissions).
K. Millet et al. that for many (if not most) participants in this experiment a bike offers a viable alternative for a car and thus may be a worthy substitute in this specific population.Given this specific choice context in the current study, we aim to generalize the positive spillover effect of positive cueing to different product contexts and to test the potential moderating role of substitutability in study 2.More specifically, we aim to test if this positive spillover effect is especially pronounced when substitutability between choice options is high (Hypothesis 3) and if it holds over several product categories (not only transport, but also food, building, etc.).While in study 1 the bike and car are viable alternatives, it is also possible that the temptingness of both options (rather than substitutability) matters.After all, both transport modes are not just substitutable options in study 1.The bike is even the preferred mode of transport (as reflected in the overall preference for the bike over the car in both conditions).But what may happen if the non-sustainable alternative is more tempting than the sustainable option?While not the focus of the manuscript, we aim to explore this in the next two studies.Recent research on hedonic food consumption has shown that temptingness matters in the justification for indulgent choices.The more an indulgent food choice (e.g., chocolate) is seen as tempting, the more reasons consumers seek to justify its consumption (De Witt Huberts, Evers, & de Ridder, 2014), and the more acceptable any reason (e.g., a prior good deed) becomes to give in to the temptation (van de Ven, Blanken, & Zeelenberg, 2018).This (temptation-based) motivated reasoning process suggests that positive cueing may result in negative spillover effects for more (vs.less) tempting non-sustainable choices.On the other hand, from a moral licensing perspective, we do not expect any moderating role of temptingness.According to moral licensing, people may reduce negative self-associations that come along with any non-sustainable choice because of a prior boost in positive self-perception (Kouchaki, 2011), irrespective of the temptingness of this non-sustainable choice.
In study 2, we disentangle substitutability and temptingness to explore if any of these factors is of relevance to observe spillover effects of positive cueing.

Study 2
The first aim of study 2 is to test if the positive spillover effect in study 1 generalizes to product choices across several domains.Recently, Moran et al. (2020) modeled the potential impact of a portfolio of 90 green behavior changes and showed that reasonable levels of adoption of these behaviors by consumers in the EU could lead to a substantial reduction of the CO2 footprint.The behaviors cover several domains of household consumption choices, including food, transport, building, equipment, textiles, furniture, paper, and plastics.The behavior changes typically contrast a sustainable option to a non-sustainable option.Therefore, we used a large pool of such choice dilemmas adapted from Moran et al. (2020).
The second aim of study 2 is to investigate if and how substitutability moderates a potential positive spillover effect of positive cueing.If the reasoning holds that consistency is driving a positive spillover effect, non-sustainable choices may be seen as more inconsistent with the selfperception of being a pro-environmentally person when they can be more easily substituted by their sustainable counterpart.Therefore, positive cueing may lead to an increased preference for the sustainable alternative especially when substitutability between choice options is high (Hypothesis 3).On the other hand, if positive cueing is just used as a tool to justify a non-sustainable choice (in line with motivated reasoning) it should especially result in an increased preference for the non-sustainable alternative when substitutability between options is high (Hypothesis 4) since these choices are more difficult to justify than when substitutability is low.
Finally, the procedure we apply in study 2 not only enables us to disentangle substitutability and temptingness, but also provides the opportunity to test if temptation-based reasoning (see discussion of study 1) also applies to the choice between a sustainable and non-sustainable alternative.

Pretest
As a preparatory step, the authors each independently reformulated the behavior changes identified by Moran et al. (2020) into pairs that each contain a sustainable and a non-sustainable option, which from now on will be referred to as respectively "sustainable option" and (non-sustainable) "distractor" in our method and results sections.Based on a point-by-point review and discussion, we reformulated all the behavior changes into 90 choice pairs to be used in the subsequent data collection.Through two pretests, we assessed how these choice pairs are perceived by consumers in terms of environmental friendliness (i.e., is the distractor effectively perceived as less sustainable than the sustainable alternative?),substitutability (i.e., to what extent are the distractor and the sustainable alternative perceived to be fulfilling the same consumption goal?) and temptingness (i.e., to what extent is the distractor perceived to be more tempting than the sustainable alternative?).Out of the 90 choice pairs, 81 were useable for our main study.Details of the choice pairs and pretests are included in Appendix B.

Participants
The main study was preregistered (https://aspredicted.org/9t55h.pdf).We requested 1700 respondents on Prolific and 1703 respondents completed the survey.As preregistered, we removed participants who failed the attention check, leaving a sample of 1615 (57.5% women, M age = 38.7,SD = 14.2).This is in line with the envisioned sample size of 1620 (see preregistration).

Materials and procedure
The experiment consisted of a cueing procedure and a choice task.We randomly assigned participants to two between-subjects conditions (positive cueing vs. control 4 ), and participants completed the same cueing procedure as in study 1.Next, participants received a choice task.They were randomly assigned to one of the 81 pretested choice pairs (i.e., the second between-subjects factor in the design).Note that the 81 behavior pairs (see Appendix B) will not be treated as fixed categories in the analysis, but as instantiations of specific levels of substitutability and/or temptingness.Participants were asked to choose between a sustainable option and a non-sustainable distractor 5 (e.g., use recycled toilet paper vs. use non-recycled toilet paper).The two options were displayed vertically (one above the other) and presentation position (sustainable option on top or bottom) was randomized across respondents and behavior pairs.Participants indicated their preferences on a 5-point scale, with response labels 'Definitely [option A]', 'Rather [option A]', 'Indifferent', 'Rather [option B]', 'Definitely [option B]'.All responses were recoded from 1 to 5, so that higher scores indicate a more sustainable ("ecological") choice; this variable is henceforth called 'ecochoice.'

Results
The current data have a multilevel structure, with respondents (N level 1 = 1615) nested in choice sets (N level 2 = 81, with average cluster size = 19.94observations).The dependent variable, eco-choice, is the rating that expresses respondents' relative preference for the sustainable option (vs. the distractor).Eco-choice has an intra-class correlation ICC 4 798 participants in the control condition; 817 in the positive cueing condition. 5In Study 2 and 3, we use the term 'distractor' to refer to the non-sustainable option for reasons of readability, without aiming to imply an additional psychological mechanism.
K. Millet et al. = 0.354, indicating that a substantial proportion of variance in ecochoice is situated at the between choice sets level (level 2), in support of the use of a multi-level specification.We specify the variable substitutability (and later also temptingness) at level 2, the experimental manipulation (POSCUE, a dummy that takes on a value of 1 in the positive cueing condition, 0 otherwise) at level 1.All variables except for the experimental manipulation are grand mean centered before analysis.We test the effects of interest in a multilevel model with a random intercept for eco-choice and a random slope for the experimental effect (at level 1), with the slope being moderated by choice set substitutability (at level 2), resulting in a cross-level interaction.The fixed component of the effect of POSCUE on eco-choice is small and not statistically significant (Est.= 0.064, 95% CI = [− 0.035, 0.163]).Further, the effect of POSCUE on eco-choice is not significantly moderated by substitutability (Est.= − 0.016, 95% CI = [− 0.160, 0.127]).There is also no significant main effect of substitutability on eco-choice (Est = − 0.116, 95% CI = [− 0.435, 0.203]).A post-hoc power analysis (see Appendix D) indicates that the current sample is sufficient to detect a small effect size.
For purely exploratory purposes (and not adopted in our preregistration), we estimated the model with temptingness and substitutability as the cross-level moderators separately for each of the following categories: food, transport, buildings, equipment and textiles (i.e., those categories that have sufficient choice sets for the second-level model to be identified and estimable).In this model, substitutability and temptingness of the non-sustainable option (at level 2) act as antecedents of eco-choice and as moderators of the effect of POSCUE on eco-choice.The results are shown in Table 1. 6To summarize, for none of the categories a main effect of positive cueing, or interaction with temptingness or substitutability emerges.

Discussion
The positive spillover effect observed in study 1 did not generalize over product domains.There is no evidence of any positive cueing effect, neither unconditionally nor conditionally (as a function of substitutability and/or temptingness of the non-sustainable option), and substitutability does not seem to affect the preference for the sustainable alternative (also not within any category, as suggested by the exploratory analyses).This preference is, however, strongly negatively impacted by the temptingness of the non-sustainable option.

Study 3
In the third study we aim to further explore the potential moderating roles of substitutability and temptingness by means of a traditional

Table 1
Regression results exploratory follow-up analyses per category (Study 2).K. Millet et al. experimental approach.While this has the disadvantage of focusing on only one specific choice context, it allows to manipulate substitutability and temptingness of the alternatives in the same product domain.In study 2, choice sets were non-overlapping, in that each of the 81 choice sets consisted of a unique sustainable option and a unique nonsustainable distractor.By contrast, in the current study (study 3), the sustainable alternative is kept constant across conditions, while the nonsustainable distractor is manipulated to represent varying levels of substitutability and temptingness.In particular, we studied consumers' choice for a given sustainable alternative, contrasted with one of three types of non-sustainable options: a non-tempting close substitute, a tempting close substitute, or a tempting alternative that is only a distant substitute.Note that substitutability and temptingness are not fully crossed because non-tempting non-substitutes are not viable choice alternatives.

Participants
We requested 800 respondents on Prolific based on a power analysis (using G*Power 3.1; Faul et al., 2007) for an F-test (effect size η p 2 = 0.01, power 1 − β = 0.80, required N = 779) and 806 participated in the study.We removed participants who failed the attention check (n = 60) and participants who never drink tap water (n = 38), leaving a sample of N = 708 participants (55.36% female, M age = 36.28,SD = 13.58).

Materials and procedure
The study consisted of a two (positive cueing vs. control) by three (distractor: tempting close substitute, non-tempting close substitute, tempting distant substitute) between-participants design. 7Participants were randomly assigned to conditions.To develop stimuli for this experiment, we finetuned choice pairs based on the large-scale pretests for study 2 (see Appendix B).We started from the observation that, compared to tap water (the eco-option), a warm drink like tea and coffee is generally seen as a tempting distractor that is not a close substitute, whereas bottled water is seen as a close substitute that is not tempting (see appendix C, 'water v warm' in the food related panel, and 'no bottled water' in the panel related to paper, plastics, chemicals).We conjectured that a cooled plastic bottle of water with a consumer's favorite flavor would offer a tempting close substitute.We conducted two pretests to verify our selection of stimuli, details of which are included in Appendix E.
Participants first completed the cueing procedure as used in previous studies but since the subsequent choice task involved beverages, we replaced "using reusable water bottles/cups/coffee mugs" in the positive cueing condition with "not wasting food" to avoid interference with our dependent measure.These two cueing items are comparable based on the pretest and pilot study used to select the cueing items (see Appendix A).Participants were then asked to make a choice: "If you are thirsty and want to drink something now, which option would you choose?"Participants were presented with two options: tap water (the sustainable option), and one of the three pretested non-sustainable distractors: plastic bottled water (a non-tempting close substitute), plastic bottled water with your favorite flavor from the fridge (a tempting close substitute), or coffee in a disposable cup (a tempting alternative that is only a distant substitute).The two options were displayed vertically (one above the other) and presentation position (sustainable option on top or bottom) was randomized across respondents and behavior pairs.Participants indicated their choice on a five-point scale ranging from 1 = "definitively [option A]" to 5 = "definitely [option B]".All responses were recoded from 1 to 5, so that higher scores indicate a more sustainable (ecological) choice; this variable is called 'eco-choice.'Lastly, we measured distractor user status by asking participants to indicate (yes/no) whether they sometimes drink each of the following beverages: bottled water, plastic bottled water with your favorite flavor from the fridge, and coffee in a disposable cup.Based on this information we created a binary variable to be used in an exploratory follow-up analysis: Participants who responded yes to all questions were coded as distractor users and those who did not were coded as distractor nonusers.
Closer inspection of the data indicated that for some respondents a positive cueing effect might not emerge because the non-sustainable option is not truly a distractor, in that they simply do not use the product in their daily lives.To address this possibility, in an exploratory follow-up analysis, we extended our model by treating the study as a two (positive cueing vs. control) by three (tempting close substitute, nontempting close substitute, tempting distant substitute) by two (user status: distractor user vs. distractor non-user) between-participants design.Here, user status is added as an observed variable with two levels: user vs. non-user of the distractor options, indicating whether or not a given consumer is at least an occasional user of the distractor options used in the experiment (i.e., coffee, bottled water, flavored bottled water).The latter variable accounts for the fact that even though some products can be defined as substitutes at the aggregate market level (where substitutes are products that are able to fulfill the same consumption goal, e.g., quenching thirst), they may not be viable alternatives for consumers who do not use them (i.e., they are not substitutes at the individual level).Note that we take into account all three products to compute this variable to ensure that it is orthogonal to the experimental manipulation.
We run an ANOVA with eco-choice as the dependent variable and testing all main and all interaction effects of the experimental factors positive cueing (vs.control), distractor type (a distractor that is a nontempting close substitute, a tempting close substitute, or a tempting alternative that is only a distant substitute, operationalized by respectively bottled water, cooled and flavored bottled water, and coffee) and user status (distractor user vs. non-user).Results are shown in Table 2.These results show that there is a positive cueing main effect that is however moderated by distractor user status, as depicted in Fig. 1.For participants who do not use the distractor products in daily life, ecochoice is high because they are likely to not really take the alternative 7 121 participants in the control, non-tempting close condition; 122 in the control, tempting close condition; 119 in the control, tempting distant condition; 117 in the positive cueing, non-tempting close condition; 111 in the positive cueing, tempting close condition; 118 in the positive cueing, tempting distant condition.into consideration (regardless of condition).For participants who take both options already into consideration in their daily life, positive cueing increases the likelihood they opt for the sustainable alternative.

Discussion
In line with Study 1, we replicate a positive spillover effect of positive cueing in a beverage packaging setting (i.e.choosing between drinking tap water or bottled water).However, in line with Study 2, we do again find no support for the propositions that substitutability or temptingness of the non-sustainable alternatives may alter the positive cueing effect.On the other hand, exploratory analyses show that the positive spillover effect only emerges when both alternatives are already used before positive cueing takes place, suggesting that substitutability is important if defined at individual, but not at product level.

General discussion
Positive cueing has received considerable attention in marketing and psychology literature since it may help to nudge people towards more pro-environmental choices.Whereas positive cueing originally resulted in a positive effect on sustainable choices (Cornelissen et al., 2008), it was only studied in one specific choice setting with a limited number of participants.Whereas the procedure to select items for the positive cueing technique in the current set of studies is the same as in the study of Cornelissen et al. (2008), a different dependent measure was used to capture pro-environmental preferences.Therefore, the current studies are conceptual (rather than identical) replications.More interestingly, the current studies show that the specific operationalization of the measure to capture pro-environmental preferences matters: Whereas a positive spillover effect of positive cueing is replicated on specific choices for transport mode (study 1) and beverage packaging (study 3), this effect cannot be generalized over domains (study 2).Moreover, the results of study 2 and 3 indicate that substitutability (i.e., the matter in which two choice options fulfill the same consumption goal) at product level does not interact with positive cueing or even predict sustainable choices (independent of positive cueing).Exploratory analyses in study 3, however, do indicate that substitutability may matter if defined at the individual level.A positive spillover effect of positive cueing only emerges for people who truly balance the choice options against each other.

Theoretical and practical implications
Our research makes several contributions to previous literature.First, it explores if positive cueing leads to positive or negative spillover effects.Despite the major societal changes in the last 15 years with regard to sustainability and its potential impact on how we perceive (non-) sustainable choices, we replicate the originally documented positive spillover effect in two specific decision contexts (Study 1: choice between bike or car; Study 3: tap vs. bottled water).However, the effect sizes in these two decision contexts remain (very) small (in contrast to the original study from Cornelissen et al., 2008) and the spillover effect seems to be very context-specific since it cannot be generalized over domains and choices (Study 2).Importantly, the absence of an effect in this generalizability study is in line with several other studies that used comparable cueing interventions which were unable to replicate a positive spillover effect as well (e.g.van der Werff, Steg, & Keizer, 2014a, 2014b;Lacasse, 2015Lacasse, , 2016;;Noblet & McCoy, 2018;Urban et al., 2021;Van der Werff & Steg, 2018).As such, the results suggest that the impact of positive cueing on sustainable choices is limited and point to the limited practical relevance of this technique to stimulate pro-environmental behavior in general.At the same time, these results are important for theory development since they suggest that the difficulty to replicate spillover effects may be due to idiosyncratic properties of the decision context.More research is needed to investigate which characteristics are of importance to observe positive cueing effects, and why so.
Second, while substitutability at product level (as operationalized in two different ways in study 2 and study 3) does not exert any moderating influence on positive cueing, exploratory analyses of study 3 indicate that it matters if defined at individual level.That is, positive cueing will only result in a positive spillover effect if both the sustainable and nonsustainable alternative are used in daily life and thus are both viable options.
Third, we do not find an association between substitutability and preferences for sustainable options (study 2) and people in general seem to have difficulty to consider non-sustainable and sustainable alternatives as true substitutes (pretest results, Appendix B).Even though it is important to develop high quality sustainable substitutes to move towards a more sustainable society, these results suggest that policy makers should not only rely on the development of these alternatives to shift consumer behavior to more sustainable choices.We believe this observation corroborates some recent evidence on food consumption, suggesting that non-users of meat substitutes (non-vegetarians) are even less likely to accept such substitutes in 2019 than in 2004 (Elzerman, Keulemans, Sap, & Luning, 2021), despite the increased similarity in texture and taste of some recently developed plant-based mock burgers (Vandenbroele, Slabbinck, Van Kerckhove, & Vermeir, 2021).Our research at least suggests that merely the development of close substitutes is not enough.
Fourth, we did not obtain any evidence in support of a moral licensing or motivated reasoning perspective (both predicting negative Note.Type III Sum of Squares.spillover effects).This raises the question to what extent morality is involved in the motivation for (non-)sustainable choices.Since the publication of the initial study on the positive cueing effect in 2008, more and more viable sustainable alternatives became available on the market.Therefore, non-sustainable choices may be seen as more immoral than they were 10 years ago and therefore justification processes of non-sustainable choices may have become more important than ever (leading to a negative spillover effect).However, we did not find any evidence that suggests people need justification of nonsustainable choices (also independent of positive cueing, as substitutability does not matter too).Therefore, future research may explore when and to what extent choice dilemmas in which consumers weigh sustainable alternatives against non-sustainable options are morally charged and investigate how such perceptions may influence sustainable choices and justification of non-sustainable choices.The observation that non-sustainable choices do not seem to be morally charged in the current set of studies is in line with recent work indicating that people recognize sustainability problems, but do not perceive them as moral problems (Wakeman, Tsalis, Jensen, & Aschemann-Witzel, 2021).Fifth, we extend recent findings on the importance of temptingness as a driver of preferences.Whereas the relative temptingness of two equally unhealthy and comparable food options (i.e., two types of donuts) has been shown to influence preference (van de Ven et al., 2018), the present research illustrates that temptingness of a non-sustainable option is also an important driver of preferences in a decision context that exists of two different choice options (i.e.sustainable and non-sustainable alternatives).There is a strong negative main effect of temptingness of the non-sustainable alternative on the choice for the sustainable option.In other words, for choice pairs that consist of a sustainable and non-sustainable option, the chance that people opt for the sustainable option tends towards zero the more tempting the non-sustainable option is (relative to the sustainable one).The existence of this effect is not surprising, but what is surprising is its strength, especially in light of the used method: relative temptingness of the non-sustainable option (vs. the sustainable option) was measured independently of the eco-choice variable in another, independent sample of respondents (a pretest to study 2, see Appendix B).This indicates that temptingness is something that can be conceptualized not only at the level of individual consumers (as a function of personal preferences) but also to an important degree at the aggregate market level.Clearly, if the goal is to make consumers opt for sustainable alternatives, these alternatives need to be as tempting as the options they aim to replace.In many instances, this might be challenging, as illustrated by for instance the disgust that many consumers feel in response to insect-based alternatives to meat (Koch, Bolderdijk, & van Ittersum, 2021).Further research needs to further clarify what makes sustainable alternatives more or less tempting.
Sixth, from a methodological perspective, we adopted a choice paradigm in Study 2 that may be useful to study the impact of other social marketing techniques as well.The current method allows to examine the robustness of a phenomenon in a neat way.Furthermore, this specific method may turn out to be useful in disentangling different moderators (in our study substitutability and temptingness), while at the same time leaving the opportunity to use the data as input for future research (as we make our data publicly available).Other research teams may also "pretest" the choice dilemmas on other factors and use these scores as input for multilevel analyses on the current set of data.For instance, since we find in study 3 that the positive cueing effect is limited to sustainable alternatives that are already used before positive cueing takes place, one may pretest all choice dilemmas on familiarity and/or experience with the sustainable alternative and thereafter test on the current set of data if the positive cueing emerges for sustainable alternatives that people are more familiar with.

Limitations
Finally, despite the contributions of the current set of studies to gain more insight into the effect of positive cueing on pro-environmental choices, we want to mention some limitations of our experimental procedures that deserve attention in future research.Whereas we tested the technique in different decision contexts, the preference for the sustainable choice option was intentional and not consequential.That is, only in the first study participants experienced some (quite limited) consequences of their sustainable choice.When they preferred the sustainable option (the bike instead of the car), they had to wait 60 s longer before they could continue the task.Interestingly, the effect size was also the largest in study 1, even though it was the only study in which the sustainable choice was somewhat consequential.Therefore, despite the small effect sizes and difficulties to replicate positive cueing effects, it remains important to test the technique in the field to explore if it has any potential in increasing sustainable behavior in real life.Moreover, whereas the effect does not generalize to different decision contexts in Study 2, some of these decision contexts in the study may have been experienced as far-fetched and as such have increased noise in the data.Although we did not observe any effect of positive cueing, also when we explored for specific categories separately, it remains worthwhile to explore to what extent people need to be familiar with the specific choice dilemma to observe a spillover effect.

Conclusion
In conclusion, the set of findings presented contribute to the understanding of the impact of positive cueing on pro-environmental choices.If anything, the positive spillover effect seems to be very limited as it is only effective for limited choice settings in which consumers already use or have used both the sustainable and non-sustainable options in their daily lives.From a practical perspective, the findings suggest that policy makers should be careful in expecting much from using this technique as part of their social marketing plans.Finally, we would like to highlight the observation that sustainable options remain difficult to be seen as true alternatives for their non-sustainable counterpart, which is another important element policy makers need to be aware of.

Fig. 1 .
Fig. 1.Positive cueing is moderated by user status (Study 3).Note.Error bars represent the 95% CI.Different superscripts indicate that the difference between means is significant (p < .05).