‘They Only Get the Feed That Grows on Our Farm’: A Survey Experiment on Government-Subsidized Greenwashing of Swiss Meat and Agricultural Policy

Abstract

Increased consumer concern about environmental issues has led to a surge in green advertising, including misleading forms known as greenwashing. Definitions of greenwashing have mostly focused on companies misleading consumers about products, and existing empirical analyses of greenwashing effects have mostly used fabricated ad materials referring to fictitious products. The objective of the present paper is to use ad material from an actual ad campaign to examine joint greenwashing, by an industry organization and a national government, of consumer products and public policies. Using an actual government-subsidized video commercial for ‘Swiss meat’ as an example, the present paper examines greenwashing targeting both consumers and citizens to influence their product choices and policy opinions. Based on an online survey of an age- and gender-representative sample (n = 637), we measured aspects of consumer knowledge that would enable the consumers to put the ad’s message in perspective, perceptions of the ad, and beliefs about Swiss meat production. Furthermore, we used a two-factorial experimental design to examine how the video commercial and objective product information affected policy opinions relative to control treatments. We find that the consumers strongly overestimated the true percentage of Swiss farms that used the environmentally friendly feeding system featured in the ad. Regarding perceptions, 52% of the respondents ‘rather’ or ‘fully’ agreed with the statement ‘the ad conveys a realistic image of Swiss meat production’, and 41% ‘rather’ or ‘fully’ agreed with the statement ‘the ad shows the production of Swiss beef as it is’. Regarding beliefs, 51% of the respondents ‘partly’, ‘rather’, or ‘fully’ agreed that ‘more meat production in Switzerland is good for the environment’. The video commercial did not influence beliefs about Swiss meat but changed policy opinions in the favor of government subsidies for the advertisement of Swiss meat. The information treatment influenced policy opinions in the opposite direction but did not decrease the effect of the misleading ad. Other significant predictors of support for increased meat subsidies included frequent meat consumption, low factual knowledge, and positive product beliefs. The results of the survey indicate that the ad campaign may have unduly influenced perceptions of Swiss meat production and public opinion about agricultural policy. The findings raise concerns about the current self-regulation of the communication industry and joint private and public advertising of Swiss agricultural products and policies.

1. Introduction

With increasing public attention to environmental issues and absent effective measures to solve them, many consumers wish to reduce environmental impacts through voluntary efforts at the individual level. Producers have responded to this desire by ‘green’ advertising, which has been described as a promotional message that may appeal to the needs and desires of environmentally concerned consumers [1]. In a recent definition, green advertising includes all advertising with an implication or statement about environmental benefits [2]. Misleading forms of green advertising have become known as ‘greenwashing’ [3,4], which has been defined as ‘the act of misleading consumers regarding the environmental practices of a company or the environmental benefits of a product or service’ [5]. Rapidly increasing research has shown how verbal and other cues can mislead consumers and how knowledge and other individual factors moderate effects [6].

While most definitions of greenwashing refer to companies that mislead consumers about the environmental benefits of products or services [7], research has also examined cases where greenwashing is applied to influence regulatory policy [8,9]. Greenwashing has therefore been identified beyond the domain of product marketing, and misleading messages may not only target consumers but also citizens and regulators. Lyon and Montgomery [10] reviewed and classified the varieties of misleading environmental communication that appear in organization theory, economics, marketing, and other disciplines. In the realm of politics, policymakers may engage in the greenwashing of national environmental regulation both internally to the national population and externally to other countries [11]. Public and private actors may engage in greening discourses that are applied strategically to legitimate existing practices, as examined in the context of European agricultural policy [12].

Existing experimental research on the effects of greenwashing and measures to counter greenwashing effects has used researcher-generated ad materials and experimental treatments involving fictitious companies or brands. The advantage of this approach is that verbal content, imagery, and product knowledge can be tightly controlled [8]. A disadvantage concerns the external validity of the findings with respect to specific products or firms and information environments. Laboratory-like experimental designs can identify potential mechanisms, but they cannot establish whether these mechanisms are effective and important in natural settings with pre-existing product knowledge and beliefs. In these settings, the effects of greenwashing may, for instance, depend on details of sophisticated messages. A similar issue concerns experimental treatments to counter greenwashing, such as warning messages [13]. In the field of advertising more generally, similar considerations have led to calls for more field research [14,15].

The objective in this paper is to experimentally examine the effects of greenwashing, and an intervention to counter these effects, using commercially produced ad material that aims to promote both consumer products and related public policies. Government-subsidized food-industry ads for Swiss agricultural products offer excellent material for the purpose of this study. A video commercial from a government-subsidized ad campaign to promote ‘Swiss meat’ was selected as an example of a misleading ad. The video commercial featured a farmer explaining his environmentally friendly but very untypical production system, followed by the vague and unsubstantiated claim ‘The fine difference. Swiss meat’ (cf. Section 2.1, Section 3.1, Section 3.2 and Section 4.2). In parallel, we examined a second video commercial that made a clear and verifiable statement about the advertiser’s entire product category: ‘beef’. The study approach is both observational and experimental. We measured how the respondents perceived the realism of the production presented in the ads. We then experimentally tested how the misleading information (about animal feeding in the showcase farm) and exposure to objective information about animal feeding in Switzerland affected the respondents’ product beliefs and related policy opinions.

Using a designed survey experiment with an age- and gender-representative sample of N = 637 participants, we examined the following specific questions. Regarding (1) consumer understanding, (a) is consumer knowledge about production systems sufficient to put an exceptional showcase production in perspective; (b) how do consumers perceive a showcase production; and (c) do consumers hold correct belief about the environmental performance of national production relative to other origins? Regarding the effects of greenwashing, (2) how does the presented video commercial affect (a) beliefs about products and (b) opinions about related policies, and (3) how do objective information, subjective knowledge, and education moderate the effects?

We see two main contributions of this study: First, using commercially produced ad materials from large-scale campaigns, we show that many consumers have too limited knowledge to put an untypical showcase production in perspective and may thus perceive a showcase production as conveying a realistic image of production more generally. Second, we extend the experimental study of greenwashing effects to a domain of joint advertising of consumer products and related government policy, and we show that misleading ads that target both consumers and citizens may influence policy opinion. Additional contributions include the novel focus on a ‘showcase’ advertising strategy which is frequently used in the domain of agriculture and food and a detailed analysis of relevant objective consumer knowledge.

The findings are relevant for debates about the regulation of greenwashing in Switzerland, as the examined ad had been the subject of complaints to the ‘Swiss commission of truth in advertising’—the self-regulatory organization of the Swiss communication industry. In 2022, the organization had decided that the examined ad is ‘not misleading’, without any reference to related theory or empirical evidence.

The remaining paper is structured as follows: Section 2 reviews the related literature and formulates the specific hypotheses. Section 3 and Section 4 describe the methods and present the results. Section 5 and Section 6 offer a discussion and the conclusions.

2. Background and Hypotheses

2.1. Consumer Understanding and Regulation of Green Claims

Rapidly increasing research has examined how green claims and other cues such as nature-evoking imagery are perceived [4], and how they can be effectively used to mislead consumers [5] (see Section 2.2). Empirical work has also specifically examined the role of consumer knowledge and found that topic knowledge (or the lack thereof) moderates greenwashing effects in expected ways [5,6] (see Section 2.3). Normative evaluations of greenwashing offered in the literature are unequivocal. The practices undermine trust in green products and are therefore harmful to consumers, genuinely sustainable firms, and society [16,17,18]. Some authors have also noted more fundamental issues with self-regulation and voluntary green consumption [8,9]. Parguel and Johnson [9] argue that producers may welcome the attention to greenwashing regulation as it may distract from more effective regulation.

Despite these findings and concerns, green advertising remains largely unregulated in most countries, or the regulation takes the form of mere guidelines that are rarely enforced [5,17,19]. In the European Union, a directive that aims to ensure that companies substantiate their environmental claims to combat greenwashing regulation is currently being prepared [20]. In the United States, Section 5 of the Federal Trade Commission Act prohibits deceptive acts and practices in commerce. A representation, omission, or practice is deceptive if it is likely to mislead consumers acting reasonably under the circumstance and is material to consumers’ decisions [21]. The Commission has also issued guidelines for green marketing which caution marketers [21]. These guidelines are interesting in the present context, as they reflect the commission’s perception of consumer understanding. They provide a reference for comparing and evaluating the perceptions of consumer understanding reflected in statements by the Swiss self-regulatory organization.

In the general principles, the Federal Trade Commission states that ‘qualifications and disclosures should be clear, prominent and understandable’ and claims ‘should not overstate, directly or by implication, an environmental attribute or benefit’. Comparative claims should be ‘clear and avoid consumer confusion about the comparison. Marketers should have substantiation for the comparison’. Regarding ‘general environmental benefit claims’, the guidelines state that, ‘in many cases, such claims likely convey that the product, package or service has specific and far-reaching environmental benefits and may convey that the item or services has no negative environmental impact’. Because it is ‘highly unlikely’ that marketers can substantiate all reasonable interpretations of these claims, ‘marketers should not make unqualified general environmental benefit claims’. Instead, they should ‘use clear and prominent qualifying language that limits the claim to a specific benefit or benefits’ [21]. The FTC thus recognizes that unqualified general claims and unclear comparisons likely mislead consumers and formulates the guidelines for green marketing accordingly.

The Swiss Federal Act on Unfair Competition (SR 241) likewise prohibits incorrect or misleading statements about goods and services. However, there have not been any greenwashing cases brought to court, and there is no government authority supervising and enforcing this regulation with respect to green advertising. Contrary to the US, there are no official guidelines for green marketing. To avoid more regulation, the advertising industry has set up its own ‘Swiss Commission for Truth in Advertising’ [22]. The SCTA mimics a legal institution by inviting individuals and organizations to ‘file complaints’, and three ‘chambers’ issue (non-binding) ‘decisions’ in which they either reject a complaint or issue a recommendation for an advertiser to stop or change an objected ad. The SCTA processed an increasing number of complaints about greenwashing in recent years, including two complaints concerning the showcase video commercial for “Swiss meat” examined in the present study. The organization argued that the commercial is not misleading, as ‘[t]he average consumer is probably aware that different forms of animal husbandry also exist in Switzerland’ ([23]; cf. [24]). Apparently, there was no concern that the ad may ‘overstate, directly or by implication’, that the claim may not be ‘clear and avoid consumer confusion’, or that the claim may fall in a category of ‘unqualified general environmental benefit claims’ against which the FTC guidelines warn.

Empirical analysis of how consumers perceive Swiss production and related green claims is therefore directly relevant for evaluating the current regulatory practice in Switzerland. More generally, consumer understanding of environment-related product characteristics in the domain of food and agriculture is relevant to understand whether citizens, both as consumers and as voters and taxpayers, are able to form correct beliefs about the environmental outcomes of agricultural policies. This cannot be taken for granted when consumers are informed mainly through food and agricultural industry ads and their regulation is left to the advertising industry (cf. Section 3.1).

We examine the following hypotheses about consumer understanding which may be relevant for regulation:

H1.1: 

Consumers have sufficient knowledge to put an exceptional showcase production in perspective.

H1.2: 

Consumers perceive a lack of ad ‘credibility’ and ad ‘realism’ when confronted with a ‘showcase’ ad promoting a much larger product category.

H1.3: 

Given the current information environment, consumers hold correct beliefs about the environmental benefits of Swiss meat production relative to other origins.

2.2. Greenwashing Effects

Empirical research on greenwashing effects has rapidly increased in recent years [25,26]. Many studies have focused on non-food products such as automobiles [5], air travel [27], cruises [28], or product packaging [8], while studies or agricultural or food products are relatively rare [18,29]. A bibliometric analysis of 351 articles and book chapters published between 1990 and 2020 identified 19 studies on greenwashing in agriculture, the food industry, or food retail [29].

Regarding the examined measures, researchers have examined perceptions of claims and products [4], beliefs about environmental outcomes [8], attitudes toward advertisements, products, and brands [6,30], purchase intentions [31,32], (hypothetical) product choices [18], greenwashing perceptions [33], or policy opinions about proposed regulations [8].

Regarding theoretical backgrounds, the research has mostly drawn on the theory of communication and persuasion and dual process theories. Parguel et al. [5] build on the Elaboration Likelihood Model (ELM) [34] to examine the effect of verbal claims and visual cues. Schmuck et al. [8] rely on the Affect–Reason–Involvement (ARI) model which distinguishes rational cognition and affect [35] to understand how product attitudes are formed. Polonsky et al. [36] draw on the heuristic–systematic model (HSM) of information processing [37,38] to understand how consumers perceive meaningful and meaningless claims. All of these theories highlight the roles of systematic vs. heuristic processing which can predict responses to greenwashing including, for instance, reactions to positive imagery vs. complex product information or the moderating effect of individual knowledge. The expanded HSM model incorporating motivations that go beyond accuracy concerns [38] is particularly helpful for the present study. In this model, the defense motivation arouses a desire to confirm and defend the validity of preferred attitudinal positions, and the impression motivation a desire to express attitudes that are socially acceptable. These motivations can engender selective information processing, both systematic and heuristic, and largely outside awareness. The model can therefore not only explain whether systematic or heuristic cognition dominate but also predict directions of biased cognition.

Although definitions of greenwashing typically refer to firms, similar practices may also involve public organizations. Political greenwashing has been considered in organization theory which examines mechanisms such as the disconnect between the structure and the activities or between the actions and goals of an organization (called ‘decoupling’), or the disconnect between promises and actions (‘symbolic management’) [10]. Political greenwashing may be used to defend the national image without any substantial costs [11], or discourse about environmental concerns and the involvement of environmental actors may be applied strategically to legitimate existing practices, as argued by Alons [12] in the context of EU agricultural policy. Liu et al. [39] find that companies with political connections exhibit a greater degree of greenwashing than those without.

Regarding the advertising materials examined, most studies of green advertising have used print materials or websites with fictitious or little-known brands, and these advertisements are presented in different versions with varying claims and/or image backgrounds [5,6,40,41]. An exceptional study is that of Timmons et al. [42] who used real advertisements involving familiar and unfamiliar brands to examine the effects of greenwashing inoculation interventions (cf. Section 2.3. Otherwise, there are, to the best of our knowledge, no experimental studies using unchanged original ads. The present experiment with commercially produced ad materials and complex environment-related product information contributes to this limited literature. By experimenting with using unaltered commercials, we also respond to claims that advertising research lacks realism and to calls for more field research [14,15].

Concerning strategies used in greenwashing, a widely cited report by the consulting firm Terrachoice (Toronto, Canada) distinguished seven strategies and identified ‘hidden trade-offs’ as the most frequent strategy [43]. Hidden trade-offs refer to claims suggesting that a product is green based on a narrow set of attributes without attention to other important environmental issues. Out of 1753 false or misleading environmental claims on 1018 products, the authors of the report find that 57 percent of the claims fell in this category. Related to this strategy, there is a small amount of recent literature on ‘halo effects’ which describes the phenomenon that one single characteristic of a person, place, or object may create an overall impression of that person, place, or object [44]. Several studies have demonstrated and examined the effect in the context of advertising [45,46]. Our study links to these strategies by examining an advertisement that presents a ‘showcase’ farm product and focuses on a single product characteristic to create an overly positive impression of a larger product category. While the domain of food and agriculture is underrepresented in the greenwashing literature [29], ‘showcase’ ads are frequently used in this domain, focusing on idyllic family farms, for instance, rather than the more industrial systems in which much of the food is produced.

The present study examines the effects of greenwashing involving both consumer products and government policy, commercially used rather than researcher-generated ad materials, and a ‘showcase’ advertising strategy in the domain of agriculture and food where products are heterogenous, and beliefs may be particularly important. We examine the following hypotheses:

H2.1: 

The Showcase Ad positively affects beliefs about environmental outcomes.

H2.2: 

The Showcase Ad positively affects policy opinions about meat subsidies.

In addition to the hypothesis tests and drawing on the expanded HSM model which integrates the role of motives [38], we also explore how the dependent variables are associated with individual factors that could be linked with motivated cognition, including subjective and objective knowledge and dietary habits. We expect that objective knowledge and education are associated with more realistic (i.e., less positive; cf. Section 3.1) beliefs about environmental outcomes and less favorable policy opinions, while frequent meat consumption may be associated with the opposite.

2.3. Moderators of Greenwashing Effects

Research on advertising has established that product or brand attitudes as well as purchase intentions or choices are affected by environmental beliefs [47] and beliefs about marketers’ goals, tactics, and effectiveness [48]. One important factor that may determine these beliefs is knowledge about the advertised products or production [49]. Several authors have therefore experimentally examined if measures of knowledge moderate the effects of greenwashing advertisements. Related to this hypothesis, a few studies have also experimentally examined whether relevant information, warning messages, or literacy interventions can be used to correct or eliminate greenwashing effects [5,6].

Schmuck et al. [6] examine how vague and false claims along with pleasant nature-evoking images affect perceptions of ads and brands, considering environmental knowledge and environmental concern as potential mediators. They find that vague claims did not enhance perceived greenwashing regardless of knowledge or concern, while false claims enhanced greenwashing perceptions. The latter effect was moderated by knowledge in one of two samples (and overridden by the effect of nature imagery). Parguel et al. [5] find that effects of nature-evoking imagery on brand image were moderated by consumers’ topic knowledge. They also allocated respondents to (orthogonal) experimental treatments with different environmental performance information. While raw performance numbers only partially corrected the greenwashing effect (among both experts and non-experts), information in the form of traffic light symbols eliminated the effect.

Political science research on greenwashing has examined perceptions of voluntary corporate environmental action and policy preferences related to environmental regulation [8]. Consumer perceptions and preferences were assessed under three conditions: (1) with information about industry claims of voluntary action and (2) with the industry claims plus greenwashing allegations (formatted as excerpts from magazine articles), and (3) a control treatment without any such information. The greenwashing allegation made respondents more skeptical of self-regulation. However, effects on policy opinions about regulation were only weakly significant (for reporting requirements) or absent (for direct regulatory action to reduce impacts). This study thus examines the perception of claims and experimentally tests the effectiveness of additional information on policy choices.

These moderators can also provide access points for strategies and interventions to counter greenwashing effects. Many recent studies experimented with warning messages and literacy interventions. Bingaman et al. [13] found that a pre-warning message before an ad made participants less likely to buy the product, and hence more likely to resist greenwashing. Naderer and Opree [50] and Fernandes et al. [51] found that advertising literacy interventions helped consumers spot deception. Timmons et al. [42] tested the effects of a literacy intervention on greenwashing perception, brand trust, and purchasing intent, using misleading and non-misleading commercially produced advertisements. The intervention induced an increased greenwashing perception with both misleading and non-misleading ads, implying that the intervention induced a general skepticism.

We examine the following hypotheses about moderators of greenwashing effects:

H3.1: 

Relevant contextual information eliminates the effect of the Showcase Ad.

H3.2: 

Objective knowledge and education reduce the effect of the Showcase Ad.

3. Methods

3.1. Study Region and Agricultural Policy Context

Meat production in Switzerland is characterized by (1) border protection for Swiss meat, (2) generous subsidies for meat producers including subsidies for generic advertisements to promote Swiss meat [52], and (3) the generous toleration of environmental impacts and related costs [53,54,55]. In the following, we expand on this context as seems relevant for the present study.

Farming contributes less than one percent to the gross domestic product of Switzerland [56]. However, the sector plays a prominent role in politics. At the time of the present survey study (end of 2023), the minister for economic affairs (including agriculture) was a farmer, and three out of the seven ministers had a degree in agronomy. Ten percent of the seats in the National Council were occupied by farmers [57]. A study that classified the subsidies according to their function and allocated them to product categories found that, in 2018, more than 80 percent of the budgetary subsidies for agriculture were allocated to animal products. The subsidies for beef and veal amounted to CHF 985 million [54]. The price support through tariffs for ‘beef and veal’ was Euro 518 million in 2022 [58]. Including the external costs, the consumers paid less than half of the costs of Swiss beef and of Swiss meat more generally, with taxpayers and society paying the rest [54,55].

One instrument of financial support for agriculture is a CHF 538 million scheme (in 2022) to promote the ‘production and sales’ of agricultural products [56]. In this scheme, there are CH 64 million for ‘quality and sales’ which is mainly support for marketing and communication measures for products and environmental services of agriculture [52]. Of these, about CHF 35 million were allocated to animal products [52]. The funds are used to cover 50 percent of the costs of selected generic marketing campaigns by industry organizations. Meat marketing campaigns were (in 2022) supported with CHF 5.75 million [56]. Ads with slogans like ‘Swiss meat—everything else is a side order’ have been run on Swiss television and elsewhere for many years.

Members of the parliament have repeatedly questioned the promotion of meat sales [59] due to conflicts with official environmental and nutritional targets [53,60,61]. The Federal Council has taken an industry-friendly perspective and argued that the support is not aimed at increasing meat consumption but ‘at creating a consumer preference’ relative to foreign meat and that the effect of the ads on the environment ‘could not be quantified’ [52,62,63]. Swiss import regulations, however, allow meat imports only to the extent that demand exceeds national production capacity. In 2021, for instance, 96 percent of beef and 84 percent of veal consumed in Switzerland were therefore produced nationally, although prices were more than twice as high as in neighboring EU countries [56]. The video commercial examined in the present study is therefore highly unlikely to target only consumers, as consumer preferences play only a limited role for Swiss meat consumption. What is more relevant for meat producers is the continued political acceptance of the industry-friendly agricultural policy with its heavy import restrictions for meat, generous direct and indirect subsidies for meat production, and a modest environmental performance.

Regarding the environmental performance of Swiss meat production, most subsidies for agriculture are now conditioned on basic environmental requirements. However, farming practices in much of lowland and lower montane Switzerland conflict with environmental law, especially concerning nitrogen and pesticide emissions [53]. Official evaluations found that much of the government support for environmental measures and food security did not have the intended effects [64,65]. To date, however, these findings have not led to significant changes in the support policies or their communication to the public. Regarding the problem of excessive nitrogen emissions [66], Switzerland imported 27 percent of the protein fed to animals in 2021 [67]. Ammonia emissions in 2021 were at 42,000 tons [56] instead of the legally acceptable 25,000 tons [68], and emissions per hectare were among the highest in Europe [69]. Government-mandated studies and official reports do not find or suggest any superior environmental performance of Swiss agriculture compared with imports [70] (p. 35). In the domain of animal production, official reports identify feed imports as the main cause of nutrient imbalance [70] (p. 26). The suggestion of the examined ad that Swiss meat is the environmentally friendly choice is therefore unsubstantiated and misleading.

3.2. Experimental Design and Treatments

The survey design included two experimental treatment factors. The first factor allocated the respondents to treatment groups with either one of two video commercials or a control treatment without a video commercial. A second, orthogonal factor allocated the respondents to treatments with and without information that placed the claim of the misleading commercial in perspective (

Table 1. Experimental design and sample sizes across treatments.

	Info	Control	Total
Showcase Ad	108 (143)	101 (140)	209 (283)
Retailer Ad	106 (146)	107 (147)	213 (293)
Control	109 (147)	106 (149)	215 (296)
Total	323 (436)	314 (436)	637 (872)

Note: Entries are numbers of participants in the screened and (in parentheses) the unscreened sample (see Section 3.4).

). The interaction of the two factors thus tests the hypothesis (H3.1) that any effects of the Showcase Ad are due to the lack of information that would place the showcase in perspective.

The first video commercial (‘Showcase Ad’) is an advertisement of the meat industry association Proviande. It highlights the high environmental standards in one Swiss beef farm by drawing attention to the farm’s animal feed. A farmer explains that his animals ‘only get the feed that grows on our farm’, ‘If we had more land, we might have more animals’, and ‘In other words, if we don’t have more grass, we simply don’t have more animals’. These statements are followed by the inserted words: ‘The fine difference. Swiss meat’. This feeding system is therefore presented as typical for ‘Swiss meat’. The video footage shows beautiful nature, healthy animals, and a friendly farmer. The commercial is misleading as the featured farm’s feeding ‘philosophy’ (as it was called in a longer version of the commercial) is not at all representative of Swiss beef farms and even less of Swiss meat production more generally (see Section 4.2). The message apparently responded to political debates about excessive animal stocks and feed imports in Swiss agriculture, which had long been criticized for their effects on nutrient surplus in Switzerland and on deforestation in exporting countries. The interpretation of the ad as ‘greenwashing’ is also supported by government studies and reports that document a modest environmental performance of Swiss agriculture precisely due to excessive amounts of imported animal feed and the resulting nutrient loads and imbalance [53,70]. Moreover, given limited own feed sources [71], any increase in meat sales due to the subsidized ad campaigns would necessarily rely on imported feed (cf. Section 3.1).

The second video commercial (‘Retailer Ad’) is an advertisement of the retailer Lidl. Its message is that the retailer’s ‘entire offering of beef meat comes from particularly animal-friendly animal housing systems that are BTS-certified’. Contrary to the Showcase Ad, the Retailer Ad therefore makes a clear and verifiable statement about the entire product category: ‘beef’ (referring to ‘BTS’ which is the abbreviation of a government-approved and subsidized animal welfare standard). The video shows a well-known Swiss wrestler delivering meat to a distribution center. Full transcripts of the video commercials are available in Appendix A. Both video commercials were run in the period 2021/2022. The video files are available from the authors on request.

The control treatment did not show a video commercial and instead referred to ‘meat ads you recently encountered’ in respective questions. Regarding the comparison between the treatments, it is important to note that the two video commercials, while selected to represent a misleading and an acceptable communication, also differ in many other aspects including the source and the message. The comparison between the two treatments is therefore not informative with respect to the effect of the misleading showcase design. The main interest is in the comparisons of each ad with the control treatment. The focal treatment of this study is the Showcase Ad. The Retailer Ad is useful, however, to see how the effect of the Showcase Ad relative to the control compares with the effect of another (very different) ad.

The information treatment relates directly to the content of the Showcase Ad. The treatment provided, on a single screen, the numbers to contextualize the showcase within the product categories ‘Swiss beef’ and ‘Swiss meat’. It was presented as the resolution to a quiz about animal feed in Swiss meat production (cf. Section 3.3).

3.3. Questionnaire

The online questionnaire started with screening questions about gender and age, followed by ten sets of questions which were divided into three sections by the two treatment interventions. The first section concerned the following topics and questions:

• Dietary styles: how the respondents would describe their dietary style, (if non-vegetarian) how often they ate meat, and what they associated with the production of meat in Swiss agriculture (open question for keywords);
• Beliefs I: whether they agreed that more meat production in Switzerland was good for animal welfare, the environment, and food security in times of crisis;
• Policy opinions I: whether subsidies for meat and agriculture in general should be increased or decreased.

Intervention 1 asked the respondents in the treatment groups: Showcase Ad and Retailer Ad to attentively watch the following video commercial. After the commercial, they were asked to choose which one out of four presented statements appeared in the video. The second section concerned perceptions of the ads, another set of questions about beliefs, and subjective and objective knowledge.

4.

Perception of (the) ads: (See Section 4.3 for questions.) The treatment groups were asked ‘To what extent do you agree with the following statements:’ For the control group, the question was ‘If you think of the ads for Swiss meat that you encountered recently, to what extent do you agree with the following statements:’ (An additional open question asked them whether they could remember a specific ad and, if they did, what the message was);

5.

Beliefs II: whether they agreed that Swiss meat production was more sustainable and animal welfare requirements were higher than in other countries;

6.

Subjective knowledge: how they rated their knowledge about agriculture and meat production in Switzerland and how intensively they had dealt with aspects of origin, environment, and animal welfare;

7.

Objective knowledge: how they estimated the percentage of farms in Switzerland that fed their animals only with the farm’s own feed (with adjustable levers for beef, dairy, pig, and poultry farms), how many percent of the animal feed were imported to Switzerland (with adjustable levels for cattle, pig, and poultry), and how many percent of the Swiss cropland were used for animal feed (adjustable lever). Each question ended with ‘your best estimate is sufficient’ and included a do not know option: ‘no idea’.

Intervention 2 provided the respondents of the treatment group: ‘Info’ with the actual numbers for the preceding knowledge questions (referring to data sources ‘Agroscope/Swiss Farmers Union/Federal Office of Agriculture’; cf. Section 4.2). The final section included another set of questions about policy opinions, followed by questions about socioeconomic characteristics and relations with agriculture:

8.

Policy opinions II: whether existing subsidies for the advertisement of Swiss meat and agricultural products in general should be increased or decreased;

9.

Socioeconomic characteristics: highest educational degree, current occupation, and urban/periurban/rural residence;

10.

Personal relationship with agriculture: whether respondents had such a connection and, if yes, which kind (with five options including ‘other’).

Questions about agreement with statements (about beliefs, perceptions, and policy opinions) were asked on 5-point Likert-type scales. Those in sets 2, 3, 4, 7, and 8 included a ‘don’t know’ option. The translated questionnaire is available in the Supplementary Materials. Figure 1 illustrates the survey process and the examined measures, questions, and hypotheses.

3.4. Survey Procedures and Sample

The survey was implemented in Qualtrics by the authors and administered to a commercial panel of respondents by the Swiss (Zurich) subsidiary of the market research firm Bilendi & Respondi (Paris). The Swiss panel consisted of 80,000 people recruited via various sources (mainly online) who had agreed to take part in surveys on a regular basis in return for a small incentive. The continuous panel management conformed to the quality standards of the industry (e.g., the identity check of panelists, regular and systematic panel cleaning, and controlled number of invitations sent per panelist). A representative composition of the sample was achieved in this study by setting quotas for the characteristics age and gender. A random sample was then drawn from the panel based on the quota characteristics. The last survey participants were specifically invited based on open quota characteristics. The survey was targeted at respondents between 18 and 75 years of age in German-speaking Switzerland. The stratification targets for age and gender were determined based on official statistical estimates for the population (see

Table 2. Sample and population.

	Full Sample		Screened Sample		Population a (%)	Chi2, Full Sample b	Chi2, Screened Sample b
Characteristic	N	%	N	%
Gender
female	436	50.0	316	49.6	50.0	χ2(5) = 3.79, p = 0.58	χ2(5) = 5.10, p = 0.40
male/other	436	50.0	321	50.4	50.0
Age
18–30	180	20.6	110	17.3	20.3	χ2(15) = 6.00, p = 0.98	χ2(15) = 7.99, p = 0.92
31–45	250	28.7	172	27.0	29.0
46–60	253	29.0	188	30.0	29.3
61–75	189	21.7	167	26.2	21.4
Education
Compulsory school	51	5.8	37	5.8	14.6	χ2(20) = 27.2, p = 0.13	χ2(20) = 20.5, p = 0.43
Secondary level II (vocational education)	381	43.7	279	43.9	36.1
Secondary level II (general education)	122	14.0	92	14.5	6.9
Tertiary level (higher vocational education and training)	147	16.9	111	17.5	14.6
Tertiary level (university)	163	18.7	117	18.4	27.8
Total	872	100.0	637	100.0	100.0

Notes: ^a Sources: [72] (data 2022); the distribution of educational levels in the population was calculated from the Swiss Labour Force Survey [73] (data 2022) which covers the age range from 25 to 74 years (rather than 18 to 75 years as in the survey). ^b Test for differences among the six treatment subsamples.

). The survey was conducted and completed in the days from the 15th to the 17th of November 2023. The respondents received a payment of CHF 1.40 for the completed survey.

The sample of completed responses comprises 872 observations. The responses to the questions about the content of the video commercial were used to examine if the respondents had attentively watched the videos. The analysis of correct/incorrect responses by the survey response time showed that the consistency increased sharply up to a response time of five minutes and remained constant at around 80% above that level. To account for the treatment effects on the survey duration, we trimmed each of the six subsamples at the first-quartile response time which was between 245 s (treatment Control/Control) and 304 s (treatment Showcase Ad/Info). The 25% fastest responses were thus excluded from the analyses. Furthermore, we excluded 20 respondents who took more than one hour’s time. The remaining sample includes 637 observations. Analyses with the full sample are presented in the Supplementary Materials.

Table 2 shows basic socioeconomic attributes of the full and the screened sample, along with measures at the population-level from official statistics. The age and gender are close to the population measures due to stratified sampling. The lowest and medium educational level are somewhat underrepresented, and the highest level is somewhat overrepresented in the sample. Chi-square tests showed no significant differences among the six treatment subsamples.

3.5. Statistical Analysis

The examined dependent variables include the measures of objective knowledge, perception of the ads, beliefs, and policy opinions. Frequency distributions of categorical variables in the subsamples were tested using chi-square tests. Treatment effects and associations with covariables were tested in ordinary least squares regressions, ordered logit and binary logit models, using White’s robust standard errors. In binary logit models, we computed marginal effects (at means) and predicted values using the delta method as implemented in Stata 13. Ordinal explanatory variables for age, education, frequency of meat consumption, and beliefs were included as linear variables. For transparency, we always included the same basic explanatory variables in the regression analyses.

4. Results

4.1. Descriptive Statistics of Variables

Table 3. Description of variables.

Variable Name	Description	Mean	SD	N
Sociodemographic characteristics
Age	Age category (18–30 = 1; 31–45 = 2; 46–60 = 3; and 61–75 = 4), included as linear variable	2.65	1.05	637
Female	Gender (female = 1; male/other = 0)	0.50	0.50	637
Education	Highest education (categories see Table 2), included as linear variable	2.99	1.26	636
Urban	Dummy for urban residence (reference is periurban)	0.29	0.46	637
Rural	Dummy for rural residence (reference is periurban)	0.43	0.50	337
Relationship	Dummy for personal relationship with agriculture (family, work, and earlier life)	0.37	0.48	637
Meat	Frequency of meat consumption: 1 = vegetarian or (almost) never; 2 = few times in a year; 3 = once or twice a month; 4 = about once per week; 5 = several times a week; and 6 = (almost) daily	4.46	1.45	637
Knowledge
Know report	Self-reported knowledge about agriculture and meat production in Switzerland (1 = I am not familiar with the subject; (…); 5 = I am very familiar with the subject)	2.68	1.45	637
Know error	Mean deviation of eight responses from the true values in the knowledge test (see Table 4 for details)	32.1	9.22	637
Know level	Knowledge groups based on interquartile range of Know error (1 = lowest; 4 = highest)	2.48	1.12	637
Beliefs (before Intervention 1) a
Belief environ. I	Agreement with ‘More meat production in Switzerland is good for the environment’	2.63	1.28	620
Belief welfare I	Agreement with ‘[…] animal welfare’	2.73	1.30	616
Beliefs (after Intervention 1)  a
Belief environ. II	Agreement with ‘Meat production in Switzerland is more sustainable than abroad’	3.82	0.88	637
Belief welfare II	Agreement with ‘In Switzerland, the requirements for animal welfare are higher than abroad’	4.11	0.81	637
Policy opinions (before intervention 1) b
Subsidy meat	‘Should meat production be supported more or less in Switzerland?’	2.81	1.05	619
Subsidy agriculture	‘[…] agriculture in general […]?’	3.51	1.06	626
Policy opinions (after intervention 2) b
Subsidy meat ads	‘Should advertising for meat be subsidized more or less?	2.91	1.14	608
Subsidy ag. ads	‘[…] agricultural products in general […]?’	3.44	1.07	615

Notes: ^a Scale: 1: ‘do not agree at all’; (…); 5: ‘fully agree’. ^b Scale: 1: ‘much less’; 5: ‘much more’.

presents descriptive statistics of the variables for the examined socio-demographic characteristics, topic knowledge, beliefs, and opinions.

4.2. Objective Knowledge

Figure 2 shows the responses to the first seven knowledge questions about animal feeding in Switzerland. Figure 3 shows the responses to the last knowledge question. The figures illustrate the large variation in the responses and large differences in estimates from the true values which are indicated by red lines. For instance, the average estimate of the percentage of beef farms feeding their animals exclusively with the farm’s own feed was 42 percent, while the true value was 7 percent (see Figure 2a, horizontal red line). In contrast, the percentage of cattle feed that is imported to Switzerland was greatly overestimated (horizontal axis and vertical red line).

Table 4. Knowledge test: true values, estimates, and correlation with subjective knowledge.

Question	True Value	N	Do Not Know (%)	Mean Estimate	SD	Mean Abs. Difference from True Value	SD	Corr. with Self-Reported Knowledge (Pearson r)
Percentage of farms feeding their animals exclusively with the farm’s own feed
Beef farms	7 a	544	17	41.8	21.5	34.9	21.3	0.028
Dairy farms	2 a	555	15	50.8	21.9	48.8	21.9	−0.019
Pig farms	1 b	542	18	40.1	23.5	39.1	23.5	−0.064
Poultry farms	1 b	543	17	39.4	25.0	38.4	24.9	−0.054
Percentage of animal feed that is imported to Switzerland
Cattle feed	10 c	532	20	49.3	19.5	39.4	19.4	0.064
Pig feed	44 c	534	19	51.4	21.3	18.5	12.9	0.067
Poultry feed	61 c	528	21	54.9	21.5	17.9	13.3	0.062
Percentage of cropland in Switzerland that is used for producing animal feed
Cropland	60 d	563	13	45.6	19.3	19.8	13.7	−0.104

Notes: Data sources: ^a Agroscope (Farm Accountancy Data Network, farm management sample, data 2017–2020, extracted 22 November 2021); ^b no official data available (value below 1); ^c Agristat ([74], data 2020); and ^d FOAG [56] (data 2020).

summarizes the responses to the knowledge questions, their mean difference from the true values, and the correlation between the differences and self-reported knowledge. The overall result is that the level of knowledge about local vs. non-local sources of animal feed in Swiss meat production is very low.

We also examined the determinants of knowledge for each of the eight knowledge variables, measured as differences between the estimated and true values (OLS regression estimates are available in the Supplementary Materials). Education was the most important predictor and had the expected negative effect on the difference from the true values. Self-reported knowledge was not significantly associated with knowledge. In some cases, meat consumption and rural residence were positively associated with larger differences between the estimates and true values. For a summary knowledge variable that measures the mean difference between estimated and true values (after replacing ‘don’t know’ with the value ‘50’ to include all respondents), only education was (highly) significant. The models do not explain much of the variation, however (with R² between 0.04 and 0.08). Regarding our hypothesis H1.1, the results suggest that many consumers had very limited knowledge about those aspects of production that are relevant for placing the Showcase Ad for ‘Swiss meat’ in perspective.

4.3. Perception of Ads

Table 5. Perception of ads: percentage of respondents who (rather or fully) agreed with statements about ads.

Item	Statement	Showcase Ad	Retailer Ad	Control	N
‘credible’	The information in the ad[s] is credible	83.1	48.3	43.8	619
‘realistic’	The ad[s] convey[s] a realistic image of Swiss meat production	52.4	41.8	35.3	610
‘as it is’	The ad[s] show[s] the production of Swiss beef as it is	40.5	30.9	36.2	595
‘what we can buy’	The ad[s] show[s] what we can buy in the supermarket	31.7	63.9	47.5	612
‘how sustainable’	The ad[s] show [s] clearly how sustainably the meet is produced	66.0	28.6	30.9	616

provides the statements about the advertisements used in the questions and the percentage of respondents who (rather or fully) agreed under each treatment condition—Showcase Ad, Retailer Ad, and Control. The Showcase Ad received more positive ratings than the Retailer Ad or the control treatment for all items except ‘the ad[s] show[s] what we can buy in the supermarket’. But even here, about one third of the respondents agreed.

Figure 4 focuses on the perceptions of the ‘realism’ of the ads. For instance, 20 percent of the respondents to the Showcase Ad (which presented the showcase farm) ‘fully agreed’ on the statement about the commercial’s ‘realism’. This is about twice as many as in the Retailer Ad.

Table 6. Perception of ads: ordered logit estimates.

	‘Credible’		‘Realistic’		‘As It Is’		‘What We Can Buy’		‘How Sustainable’
Variable	Coeff.	z	Coeff.	z	Coeff.	z	Coeff.	z	Coeff.	z
Showcase Ad	1.970 ***	9.78	0.996 ***	5.16	0.531 **	2.82	−0.687 ***	−3.61	1.669 ***	8.64
Retailer Ad	0.325 (*)	1.77	0.239	1.29	−0.299	−1.48	0.585 **	3.19	−0.438 *	−2.34
Info	0.064	0.42	−0.025	−0.16	0.023	0.15	−0.010	−0.07	−0.115	−0.76
Age	0.062	0.85	0.065	0.91	0.189 **	2.61	−0.107	−1.6	0.177 *	2.43
Female	0.011	0.07	0.057	0.36	0.111	0.7	−0.270 (*)	−1.71	0.037	0.24
Education	−0.068	−1.04	−0.082	−1.36	−0.060	−0.97	−0.001	−0.02	−0.153 *	−2.38
Know report	−0.008	−0.08	0.158	1.64	0.068	0.72	0.165 (*)	1.81	0.122	1.39
Meat	0.362 ***	5.69	0.407 ***	6.36	0.356 ***	5.7	0.190 **	3.12	0.279 ***	5.47
Rural	0.028	0.15	−0.076	−0.41	0.013	0.07	0.250	1.39	−0.133	−0.7
Urban	0.350 (*)	1.75	−0.087	−0.46	−0.002	−0.01	0.355 (*)	1.83	0.149	0.73
Relationship	0.176	1.09	0.187	1.14	0.173	1.07	−0.017	−0.11	0.110	0.68
Know error	0.044 ***	5.29	0.072 ***	7.33	0.070 ***	7.07	0.047 ***	5.34	0.063 ***	6.88
Ps.-Logl.	−794		−848		−823		−859		−855
N	618		609		594		611		615
Pseudo-R2	0.10		0.08		0.08		0.06		0.11

Notes: Significance levels: (*): p < 0.1; *: p < 0.05; **: p < 0.01; and ***: p < 0.001.

shows the models used to test the differences in the perception of the ads and the associations of the perceptions with the co-variables. The differences between the perception of the Showcase Ad (showcase farm) and the control (‘meat ads you recently encountered’) were highly significant for all questions. The Showcase Ad was perceived as particularly credible and realistic, suggesting that the consumers did not, as hypothesized (H1.2), perceive an issue with the credibility or realism of the ad. To the contrary, the ad was perceived as more credible and realistic than the average meat ad. The same cannot be said of the Retailer Ad where the perceptions were different only for ‘The ad[s] show[s] what we can buy in the supermarket’ (more agreement) and ‘The ad[s] show[s] clearly how sustainably the meat is produced’ (less agreement). Turning to the covariables, the positive effects of meat consumption and negative effects of knowledge are highly significant for all examined items.

Regarding our hypothesis 1.2, the results provide evidence that perceptions of ad credibility and realism are maintained even when a showcase beef farm is used to promote a much larger product category: ‘Swiss meat’.

4.4. Beliefs About Policy Outcomes

Figure 5 shows the distribution of beliefs about the consequences of ‘more meat production in Switzerland’ for the environment (Figure 5a) and for animal welfare (Figure 5b) as assessed before the video-commercial intervention. The beliefs regarding animal welfare and the environment were similar. More than half of the respondents (54.2 and 51.1 percent, respectively) ‘partly’, ‘rather’, or ‘fully’ agreed with the statements. Similar questions—about the consequences of meat production in Switzerland compared with other countries—were asked after the intervention (see

Table 7. Ordered logit estimates of beliefs about policy outcomes.

	Beliefs (I) Before Video-Commercial Intervention Outcomes of More Meat Production in Switzerland				Beliefs (II) After Video Commercial Intervention Outcomes of Meat Production in Switzerland Relative to Other Countries
	Environment (I)		Animal Welf. (I)		Environment (II)		Animal Welf. (II)
	Coeff.	z	Coeff.	z	Coeff.	z	Coeff.	z
Showcase Ad	0.188	1.00	0.311 (*)	1.73	0.191	1.00	0.091	0.49
Retailer Ad	−0.187	−1.06	−0.108	−0.61	−0.236	−1.31	−0.104	−0.56
Info	0.175	1.19	0.046	0.31	−0.114	−0.73	0.033	0.21
Age	−0.107	−1.50	−0.107	−1.50	−0.173 *	−2.38	−0.100	−1.42
Female	−0.297 (*)	−1.94	−0.440 **	−2.90	−0.179	−1.13	−0.078	−0.49
Education	−0.16 **	−2.68	−0.027	−0.45	0.051	0.80	0.032	0.49
Know report	0.050	0.56	0.068	0.75	0.232 *	2.38	0.317 **	3.23
Meat	0.367 ***	5.04	0.407 ***	5.64	0.266 ***	4.02	0.360 ***	5.64
Rural	−0.025	−0.14	−0.137	−0.79	0.136	0.72	0.105	0.57
Urban	−0.001	0.00	−0.238	−1.15	0.123	0.64	0.230	1.13
Relationship	0.097	0.60	0.163	1.03	0.315 (*)	1.88	0.516 **	3.12
Know error	0.040 ***	4.76	0.047 ***	4.91	0.034 ***	3.74	0.025 **	2.86
Ps.-Log-l.	−922		−917		−756		−684
Pseudo-R2	0.04		0.05		0.04		0.05
N	619		615		636		636

Note: Significance levels: (*): p < 0.1; *: p < 0.05; **: p < 0.01; and ***: p < 0.001.

).

As mentioned in Section 3.1, the environmental outcome of Swiss meat production has not been shown to be better than elsewhere. Regarding hypothesis H1.3, the responses indicate that, with the information available to them, many Swiss consumers are not able to form correct beliefs about the environmental benefits of Swiss meat production relative to other origins.

Table 7 examines the responses in more detail. For comparability, we also included the treatment factors in the model for beliefs before the intervention. The patterns of responses before and after the treatment intervention are similar overall. The video commercials did not significantly affect the beliefs (Table 7, right-hand side). Some of the co-variables were significantly associated with the beliefs: self-reported knowledge (Know report), frequent meat consumption (Meat), personal relationship with agriculture (Relationship), and low knowledge (Know error) were all positively associated with positive beliefs. The relationship with frequent meat consumption was highly significant in all four models.

The hypothesis (H2.1) that the Showcase Ad positively affects beliefs about environmental outcomes is not supported. The highly significant association of (overly) positive beliefs with the frequency of meat consumption and with low knowledge indicate that motivations may be an important driver of beliefs.

4.5. Policy Opinions

Figure 6 shows the distribution of responses to the central questions about policy opinions concerning government subsidies for agriculture. The opinions about government support for the advertisements of Swiss meat were about symmetrically distributed, with the middle category ‘same as today’ chosen by 37 percent of the respondents, and 34 percent preferring less, and 30 percent preferring more support (Figure 6a). Opinions concerning support for the advertisement of Swiss agricultural products in general were skewed, with 34 percent choosing ‘same as today’, while 17 percent preferred less, and 50 percent preferred more support.

Figure 7 shows the opinions about meat subsidies for each combination of the experimental treatments (video-commercial and information treatments). The y-axis shows the predicted probability of support for ‘more’ or ‘much more’ subsidies. The co-variables in the model are the same as in

Table 8. Logistic regression estimates of policy opinions.

	Support of Subsidies for Meat Production (Before Video-Commercial Intervention)		Support of Subsidies for Meat Advertisement (After Video-Commercial Intervention)
	Ordered Logit		Ordered Logit		Ordered Logit		Binary Logit
	Coeff.	z	Coeff.	z	Coeff.	z	ME	z
Showcase Ad	0.169	0.87	0.460 (*)	1.68	0.555 **	2.96	0.136 **	2.90
Info	−0.016	−0.08	−0.473 (*)	−1.74	−0.383 (*)	−1.92	−0.051	−1.07
Sc. Ad × Info			0.184	0.48
Age	−0.032	−0.38	−0.120	−1.2	−0.120	−1.21	−0.030	−1.30
Female	−0.049	−0.23	0.926 ***	4.38	0.926 ***	4.38	0.169 **	3.31
Education	−0.087	−1.09	−0.233 **	−2.67	−0.233 **	−2.66	−0.044 *	−2.15
Know report	0.043	0.34	0.190	1.55	0.193	1.59	0.071 *	2.56
Meat	0.486 ***	5.47	0.531 ***	5.58	0.531 ***	5.53	0.073 **	3.25
Rural	0.064	0.27	−0.074	−0.31	−0.071	−0.29	−0.032	−0.53
Urban	−0.136	−0.49	−0.008	−0.03	−0.010	−0.04	0.017	0.27
Relationship	0.186	0.85	0.242	1.11	0.241	1.10	0.054	1.07
Know error	0.034 **	2.78	0.034 **	2.68	0.034 **	2.68	0.005 *	2.23
Belief env. I	0.804 ***	9.06	0.385 ***	4.55	0.386 ***	4.55	0.058 **	3.19
Ps.-Logl.	−490		−526		−526		−209
Pseudo-R2	0.15		0.11		0.11		0.14
N	409		400		400		400

Notes: Significance levels: (*): p < 0.1; *: p < 0.05; **: p < 0.01; ***: p < 0.001. ME: Marginal effect.

. The mean probability of approval is higher under the Showcase Ad treatment than the Retailer Ad or the control (with no video commercial shown). The Info treatment, which presented the correct answers to the preceding knowledge questions, slightly decreased the probability of approval under the treatments: Showcase Ad and Control.

Figure 8 shows the support for each of the video-commercial treatments at different levels of the co-variable indicated on the x-axis and in the legend (the values shown are from models without additional covariables). Support for more subsidies tended to be higher among women (Figure 8a), lower among respondents with a higher education (Figure 8b), and higher among respondents with a more frequent meat consumption (Figure 8c). Higher levels of knowledge about meat production were associated with less support (Figure 8d), while the belief that ‘more meat production in Switzerland is good for the environment’ was associated with higher support (Figure 8e). In the following, we test if these treatment effects and associations with covariables are statistically significant.

Table 8 shows the statistical models testing the effects of the experimental treatments on policy opinions and the associations with co-variables. Ordered logit models were used to examine the effects on the 5-point Likert-type opinion variable. The presented models include the treatments: Showcase Ad and Control. The first model is shown for reference and completeness and examines the opinions about subsidies for meat production elicited before the video-commercial intervention. The treatment variables: Showcase Ad and Info are included for a transparent comparison with the following models.

The second model examines the main hypotheses, which concern the determinants of the policy opinions elicited after the experimental interventions. The Showcase Ad had a (weakly significant) positive effect on the support of meat ad subsidies (relative to the control). The hypothesis (H2.2) that the Showcase Ad positively affected policy opinions about meat ad subsidies is therefore supported by the experimental test.

Turning to the moderators of greenwashing effects (hypotheses H3), the Info treatment tests the effect of exposure to contextual information about animal feeding. The effect is weakly significant and negative. The interaction of the Showcase Ad and Info tests if the exposure to factual information decreased the effect of the Showcase Ad, which is not the case. The hypothesis (H3.1) that the contextual information eliminates (or at least decreases) the effect of the ad is not supported. When the non-significant interaction term is omitted (in the third model), the effect of the Showcase Ad is highly significant and the Info effect again weakly significant.

To examine potential individual-level moderators (hypothesis H3.2), we also computed this model for subsamples by education levels. The coefficient of the Showcase Ad was positive in all the subsamples and significant in one case. The sample size and p-values for education levels in ascending order were N = 20/p = 0.096, N = 181/p = 0.18, N = 60/p = 0.12, N = 64/p = 0.017, and N = 84/p = 0.78. In models by the knowledge levels (interquartile groups), the Showcase Ad was significant in the highest-knowledge group (N = 104/p = 0.015) but not in the other groups (N = 108/p = 0.16, N = 100/p = 0.36, and N = 97/p = 0.50, in descending order). The interactions of the Showcase Ad with the education and knowledge type were not significant. Accordingly, the experiment does not provide clear evidence supporting hypothesis H3.2 (cf. Section 5).

The fourth model in Table 8 shows the marginal effects based on the binary logistic model. The dependent variable is coded 1 for respondents supporting an increase in subsidies (responses ‘more’ or ‘much more’). The video-commercial treatment increased the percentage of these respondents by 13.6%. The information treatment decreased support by 5%, but this latter coefficient is not statistically significant. The covariables: female, education, frequency of meat consumption, knowledge, and beliefs about environmental outcomes were all significant. The marginal effects show the effects on the probability of approval associated with a one-category increase in the explanatory variable. Self-reported knowledge, rural or urban residence, and a personal relationship with agriculture did not have any effect.

Analogous models with the treatments: Retailer Ad and Control did not show any treatment effects on policy opinions, while the associations with the co-variables were similar. Policy opinions about government support for the advertisement of ‘agricultural products in general’ were not affected by the experimental treatments. Hence, this additional test of hypothesis H2.2 is not supported. The regression estimates are provided in the Supplementary Materials.

5. Discussion

Our main empirical findings in this study which experimented with materials from actual ad campaigns to promote agricultural products and policies are threefold. First, the respondents lacked the knowledge to properly contextualize an untypical showcase product presented in a misleading ad. In line with this finding, many respondents perceived the showcase production as ‘realistic’ and believed that Swiss products are a preferable choice for environmental reasons—the vague and unsubstantiated claim of the misleading ad. Second, the misleading ad changed policy opinions towards increased support of government subsidies for meat advertisement. Third, information that contextualized the showcase farm affected policy opinions in the opposite direction but did not decrease the effect of the misleading ad. Several aspects of these findings deserve discussion. We begin with consumer understanding and greenwashing effects before turning to implications for regulation and limitations of this study.

Regarding consumer understanding, the present study goes beyond previous work by combining the experimental approach with a careful analysis of relevant objective product knowledge. Although not a formal hypothesis test, the detailed documentation and descriptive analysis of objective knowledge indicates that consumers did not have the background to usefully contextualize the unsubstantiated and misleading claim of the Showcase Ad. Low knowledge about production systems, perceptions that a showcase farm that ‘feeds only the farm’s own feed’ presented ‘a realistic image of Swiss meat production’, and beliefs about a superior environmental performance of Swiss products confirm that many consumers are largely defenseless against influences by misleading green claims (cf. [21]).

The relatively weak effects of the experimental treatments could be explained with pre-existing perceptions and beliefs, which may be the long-term effect of decades of agricultural industry advertisements of which the presented video commercial is only one example. This larger context and background could explain why about half of the respondents found that the Showcase Ad featuring a farmer who ‘feeds only the farm’s own feed’—as emphasized multiple times in the 40 s commercial—presented ‘a realistic image of Swiss meat production’. The finding that beliefs are more stable than policy opinions or preferences is also consistent with findings of other recent research that examined beliefs and policy opinions [75].

The failure to decrease the effect of the Showcase Ad by providing contextual information (about sources of animal feed) parallels the finding in other studies. Parguel et al. [5] found that mere numbers (about gas mileage of cars)—unlike traffic-light ratings—did not help respondents revise their judgements. Similarly, Kolcava [8] found only a weakly significant effect of a counter-framing on policy opinions (about mandatory reporting). In a study examining halo effects, Lanero et al. [46] report that providing accurate information did not neutralize biased heuristic thinking. A possible interpretation in our study is that the Showcase Ad affected opinions (mainly) for other reasons than the technical information on animal feeding, such as, for instance, the pleasing nature-evoking images. The information treatment in our study may have served as a generic warning message rather than through the specific numerical information.

With the commercially produced ads and actual context information, the survey experiment examined the challenges to consumers more realistically than previous work. Respondents who received the context information on animal feeding had to make the connection between this information and the ad’s (vague) claim that Swiss meat comes from beautiful farms that feed only their own feed. Using such context information is clearly more challenging than reacting to overt warning messages [13,50] or traffic-light symbols [5]. Moreover, the information was not available in the ad itself but in a subsequent step of the survey, which may have added to the challenge. More experimentation is needed to clarify which forms of mandatory information disclosure are most effective in preventing deception through showcase designs and other greenwashing strategies and, at the same time, increasing consumer information and competence.

Regarding topic knowledge and education as mediators of effects, we did not find a consistent pattern. The Showcase Ad affected opinions among the most knowledgeable respondents and among the lower educational groups. A tentative explanation is that only the knowledgeable respondents fully understood (and reacted positively to) the ad’s message content, while higher education may have decreased a more general positive effect of the farm story and imagery. In contrast, the direct associations of knowledge and education with perceptions, beliefs, and opinions were consistent. Higher levels of education and knowledge were linked with less favorable perceptions of the ads, less positive beliefs, and less favorable opinions about subsidies. Among the other individual characteristics, the far strongest association with beliefs was found for the frequency of meat consumption. We do not have an empirical basis to determine whether ‘diet follows beliefs’ or ‘beliefs follow diet’. However, since dietary habits are difficult to change and eating meat is deeply rooted in Western culture [76], the main direction of causality is likely from dietary habits to beliefs. The same causality would also be expected from theories of motivated information processing [38] (cf. Section 2.3) or self-serving bias [77].

Regarding practical implications, the findings indicate that current beliefs and opinions about Swiss meat are influenced by a combination of low consumer knowledge, weak (self-)regulation of the advertising industry, and government support for misleading industry advertisement. Efforts to change this situation would be most effective if they addressed each of the causes. Serious regulation of green claims along the lines of the FTC guidelines (cf. Section 2.1) may prevent misleading general claims and comparisons involving Swiss meat. Given that Swiss consumers pay higher prices than in the neighboring countries, and taxpayers contribute substantially to agriculture and food (cf. Section 3.1), the government could also consider more actively informing consumers and citizens about the production systems and environmental measures rather than contributing to misleading industry communications.

The finding of incorrect beliefs about production systems and environmental outcomes and their association with low knowledge suggests a possible approach for regulation. Advertisers could be required to provide relevant contextual information about the entire product category for which they advertise. This would be effective in multiple ways. The regulation would make the use of misleading showcase designs less attractive to advertisers, provide information to interested consumers, and, in the long run, contribute to consumer knowledge. From an institutional perspective, of course, such a regulation may be difficult to implement when the interests of the regulators and the regulated industry appear to be largely aligned (cf. Section 3.1). Far from supporting norms of transparency and disclosure [78], the current financial support for greenwashing ads and lax regulation of advertising (cf. Section 3.1) could even enhance the perception among the agricultural and communication industries that the greenwashing of Swiss products is an appropriate behavior and in line with the public interest.

A few limitations should be mentioned. First, it may be tempting to interpret the experiment as a study of misleading vs. acceptable communications. However, as mentioned (Section 3.2), the misleading Showcase Ad and the acceptable Retailer Ad differed in many other dimensions, including the source (a German retailer vs. the Swiss meat industry organization), the audiences (consumers vs. consumers and citizens), and the message content (environment vs. animal welfare). Hence, the study approach does not allow us to isolate the effect of misleading vs. acceptable communications. The relevant comparison for both the Showcase Ad and the Retailer Ad is with the control treatment which referred to ‘meat ads you recently encountered’. Related to this limitation, we cannot say which elements of the Showcase Ad were decisive for the observed effect on policy opinions, and we cannot make any predictions for effects of different ads. To isolate the effects of specific elements of actual ads, one would need to apply the experimental treatments within the ad campaign itself (cf. Section 6). Third, some respondents may have encountered the presented video commercials before seeing them in the survey. This may have changed their responses relative to a situation in which the commercial is new to them. Finally, the sequence of the sets of questions has some disadvantages. In particular, the knowledge test was conducted after the first intervention (the video-commercial treatment), which may have influenced the outcome. The decision about the sequence is not an easy one, however, as conducting the knowledge test earlier in the survey could likewise influence subsequent responses.

6. Conclusions

Misleading green claims are widely recognized to be harmful for consumers, for producers of more environmentally friendly products, and for society at large. This study used an online experiment with an age- and gender-representative sample of respondents and commercially produced video materials to examine consumer responses to an ad that presented a showcase farm with a very untypical production system to suggest that ‘Swiss meat’ is preferable for environmental reasons. We examined relevant consumer knowledge, perceptions of the ad, and beliefs about environmental outcomes. Furthermore, we experimentally examined how the ad affected product beliefs and opinions about subsidies relative to a control treatment, and whether an intervention providing relevant context information could eliminate any effects of the ad. Parallel analyses were conducted for a second video commercial, which made a true and verifiable statement about a retailer’s entire beef offering. We found that the consumers did not have the necessary knowledge to put the message of the misleading ad in perspective. While the ad did not change beliefs about the environmental and animal welfare outcomes of Swiss meat production, we found evidence that it changed political opinions in favor of agricultural subsidies. Contextual information to counter the misleading information affected opinions in the opposite way but did not decrease the effect of the misleading ad. While the observational evidence on knowledge and perceptions suggests that the consumers were misled, the experimental test could not confirm that this was achieved specifically through the ‘showcase’ design.

The detailed documentation of consumer knowledge, perceptions, and beliefs related to the video commercial that had been classified as ‘non-misleading’ by the self-regulatory organization of the Swiss communication industry makes this study relevant for debates about the regulation of greenwashing in Switzerland. Furthermore, the empirical results underline the problem of government support for communications that are designed to mislead, using a range of mechanisms well known in marketing and organization theory. The findings suggest that the examined commercial undermined markets for more environmentally friendly meat and misled public opinion on agricultural policy. Beyond the Swiss context, the results show that more research is needed to determine how the mandatory disclosure of information about larger product categories could discourage misleading showcase designs and help improve consumer competence. This research would be particularly useful if it combined experimental approaches with the realism of commercial ad campaigns and representative samples of respondents with pre-existing knowledge and beliefs about the advertised products. An interesting avenue to combine experimental control and external validity would be to conduct field experiments in which the experimental treatment conditions are applied within the commercial ad campaign, and a survey approach is used only to measure the consumers’ responses.