Snapping at Transparency
An experimental approach to transparency and consumer behavior
Master Thesis
Master Business Administration – track Marketing
Maud J. Slingenberg (10616888)Supervisor: Mr. J. Demmers MSc
University of Amsterdam (UvA) Faculty of Economics and Business
MSc. in Business Administration – Marketing Track Date 01-06-2015
“The power of a brand resides in the minds of customers”
TABLE OF CONTENTS
LIST OF FIGURES ... vi
LIST OF TABLES ... vii
ABSTRACT ... viii
INTRODUCTION ... 1
THEORETICAL FRAMEWORK... 5
Transparency ... 5
The Mental Construal ... 6
Attitude ... 13
Product Choice ... 15
PRE-TEST (STUDY 1) METHOD ... 21
Participants & design ... 21
Measures ... 22
Procedure... 22
RESULTS PRE-TEST ... 22
FIELD EXPERIMENT (STUDY 2) METHOD ... 24
Participants ... 24 Design ... 24 Measures ... 25 Procedure... 27 Hypotheses Testing ... 28 DISCUSSION ... 38
Implications & future research ... 45
Limitations ... 45
REFERENCES ... 47
APPENDICES ... 56
Appendix A: Survey Supermarket... 56
Appendix B: Conditions ... 58
LIST OF FIGURES
Figures Page
1. Level of Construal Matrix 2x2……….. 12
2. Conceptual Model 1: Conditional Indirect Effects in a Moderated-mediation Model. Process Model = 16..………..
20
3. Mean numbers of Product & Brand Categorization in the different Conditions (n = 90)………
30
4. Hypotheses concerning Mental Construal………... 30 5. Mean numbers of Product & Brand Attitude in the different Conditions
(n = 90)…...………...
32
6. Hypotheses concerning Attitude ……….. 32
7. Product Choice in the different Conditions, divided by the Nature of the Information (n=150)………...
LIST OF TABLES
Tables Page
1. Transparency Statements………... 23
2. Description of the Variables and their Levels……… 25 3. Demographic Variables, (In)dependent variables: Descriptive Statistics
and Correlations………...
28
4a. Unstandardized OLS Regression Coefficients with Confidence Intervals of Positive Brand Disclosed Product Transparency on Product Choice through Attitude depended on Level of Construal………..
60
4b. Conditional Indirect Effects of Positive Brand Disclosed Product
Transparency on Product Choice at the Values of Construal (Values are the Mean and +/- one SD from Mean)……….
61
5a. Unstandardized OLS Regression Coefficients with Confidence Intervals of Negative Brand Disclosed Product Transparency on Product Choice through Attitude depended on Level of Construal………..
62
5b. Conditional Indirect Effects of Negative Brand Disclosed Product Transparency on Product Choice at the Values of Construal (Values are the Mean and +/- one SD from Mean)……….
ABSTRACT
With the rise of information technology and the subsequent increasing access to information, consumers have become more conscious about their environment and are demanding more transparency of companies and their products. Transparency has become an important factor in the decision making process of consumers. In this study we investigated the effect of product transparency on consumer decision-making through the mechanisms, grounded in the Construal Level theory (CLT) and attitude behavior relationship. A field experiment was conducted in a supermarket approaching consumers with an (transparent vs. non transparent) ad, measuring their mental construal, attitude and product choice. The results show that there was no effect of transparency on product choice. There was also no significant effect of transparency on attitude. Concerning the level of construal, positive information disclosure led to a high level construal, or abstract representations of the brand. Last, the attitude-behavior relationship was not dependent on the level of construal. Implications concerning marketing of transparency will be discussed.
“Honesty and transparency make you vulnerable. Be honest and transparent anyway.”
Mother Teresa (1910-1997);
Founder of The Missionaries Of Charity
Due to several organizational as well as environmental developments over the last decade, the markets’ focus shifted from transactional marketing to relationship marketing (Grönroos, 1996; Sheth & Parvatiyar, 1995). Because of the development of new information technology, marketers were able to gain more information about consumers, and built up consumer databases. Different digital tools, such as e-mail, mobile phone and internet, resulted in direct relationships between producers and consumers (Sheth & Parvatiyar, 1995). But not just the marketer assembled more knowledge about the consumer; the consumer also gained more access to information about products and companies. With the rise of information technology and the subsequent increasing access to information, the power has now shifted from the marketer to the consumer (Fournier & Avery, 2011; Labresque, Vor dem Esche, Mathwick, Novak, & Hofacker, 2013). The increased accessibility of information has led to more educated, sophisticated consumers (Labresque et al., 2013). Consumers have become more conscious about their environment and are shifting their preference towards sustainable products (Bhaduri & Ha-Brookshire, 2011). The growing need for information is also driven by the skepticism and distrust towards companies, which was enhanced by the financial crisis in 2007-8 (Kirby, 2012; Tapscott & Ticoll, 2003). Transparency, defined as the disclosure of relevant, accessible, comprehensive, qualitative, and reliable information (Vishwanath & Kaufmann, 2001), has become an important factor in overcoming distrust (Bhaduri & Ha-Brookshire, 2011; Jahansoozi, 2006; Kanagaretnam, Mestelman, Nainar, & Shehata, 2010; Kirby, 2012). Another consequence of the growing information technology is that organizations can no longer
disguise information (Christensen, 2002). These developments have led us to the age of transparency (Tapscott & Ticoll, 2003), in which, even though still outweighed by price and quality, transparency has become an important factor in the decision making process of consumers (Cohn & Wolfe, 2013).
Although transparency is quickly becoming more widely acknowledged among marketing practitioners, academic research has not yet established how transparency is cognitively processed and trough which mechanisms it may eventually induce trust and/or influence the decision making process. Before the consumer evaluates or makes a decision concerning a product or brand, a mental representation is retrieved, perceived or construed about that specific product or brand (Petty & Cacioppo, 1979). This touches the ground of the construal-level theory (CLT), which theorizes that the mental representations can be divided into two levels, namely low- and high-level. High-level construals can be viewed as abstract and superordinate mental representations. Low-level construals can be viewed as concrete and subordinate mental representations. Much research has been done concerning the relation between the degree of abstractness and the level of construal, as the higher the level, the higher the degree of abstractness (Kardes, Cronley, & Kim, 2006; Trope & Liberman, 2003). An important factor within these studies is the psychological distance. As the deviation of the direct experience increases, the degree of abstractness will be higher. But if levels change, the representations might change as well as the importance of certain context, attributes or features.
More specifically, one could question how, even though the consumer does not experience the product directly, transparency can influence the degree of abstractness, or in terms of the CLT, influence the level of construal? And how can change in level of construal effect consumers’ attitude and product choice? The current study is part of an
attempt to answer this question. Hence, the research question is formulated as follows:
How does transparency affect consumer decision-making through its effect on mental representations about products and brands?
The aims of this study are introducing the new aspect transparency to consumer behavior research, extending knowledge on the construal level theory as well as investigating new insight into the attitude-behavior relationship, and exploring the effect of transparency on cognition, attitude and choice.
The current study contributes to existing literature on CLT in enhancing the understanding of construal level theory in relationship to transparency. Subsequently the effect on the resulting attitudes about the product as well as the brand will be examined. Besides these effects, a moderation effect is modeled, implying that the attitude-behavior relation is dependent on the level of construal. It will also provide insight into the consequences of these effects on products/brands. We hypothesize that, dependent on the nature of information, transparent product descriptions lead to more concrete thoughts about that product, and more abstract thoughts about the brand. As a result, this may increase the credibility of the product or brand and perhaps the relationship between the company and the consumer. By understanding how different types of transparency (positive, neutral, negative) affect the representation of consumers, and consequently can lead to the executive control of attitude and action or choice, organizations will be able to change or enhance their transparency strategy accordingly. Transparency and the mental construal are central concepts for this study, and will be described within the theoretical framework. In order to answer the research question, supported by several hypotheses a survey and a field experiment were
conducted. The methods and procedures concerning this experiment are described in the method section, followed by the results. After discussing discrepancies and/or hiatuses, conclusions are drawn regarding the research question.
THEORETICAL FRAMEWORK
Transparency
Within the field of business, transparency is defined as “a situation in which business is done in an open way without secrets, so that people can trust that it is done fair and honest” (Cambridge Dictionaries, 2014). In the field of financial accounting, the definition for transparency is defined as the “availability of relevant, reliable information about the periodic performance, financial position, investment opportunities, governance, value, and risk of publicly traded firms (Bushman & Smith, 2001). Tapscott and Ticoll (2003, p. 22) define transparency as “access of information to stakeholders of institutions, regarding matters that affect their interest”. These definitions are formulated from a perspective of stakeholders or general public. What these definitions have in common is that they refer to important dimensions of information, namely availability and relevance. Thus, information need not only be relevant and accessible, but also comprehensive, qualitative and reliable. Hultman & Axelsson (2007) expand on the concept of transparency and define four relevant types of transparency within the Business-to-Business relationship (B2B), creating a transparency typology for marketing management research. The four types are technological, organizational, supply, and cost/price transparency. The authors also include the direction of transparency (sharing information or reciprocity) in their model, and the distribution of transparency (transparency in a network) (Hultman & Axelsson, 2007).
Several studies have examined the concept of transparency in relation to different outcomes. For example, Kanagaretnam et al. (2010) conducted a study on transparency (here conceptualized as ‘complete information’) and trust and found that higher transparency significantly increases trusting behavior in one-shot interactions. Furthermore, Bhaduri & Ha-Brookshire (2011) argue that transparency can change
distrusting consumers into trusting consumers. Still, the transparency of an organization needs to be proven by standardized authority or certificates (Bhaduri & Ha-Brookshire, 2011).
Organizations still encounter boundaries concerning the kind and the amount of information they opt to disclose. Just as consumers will have access to the disclosed information, so will the competition. As indicated by Christensen (2011, p. 167) “… “transparency” first and foremost is a question of establishing a consensual system of meaning between different actors in the corporate landscape.” In the article he argues that transparency is staging. Interestingly, research has shown that organizations may also benefit from disclosing unfavorable information. The literature on two-sided messages indicates that, under certain specific circumstances, disclosure of unfavorable information can positively influence consumer preferences (i.e., enhance source credibility, brand attitude, purchase intention and decrease negative cognitions) (Eisend, 2006).
Increased transparency may also have a positive impact on consumers purchase intentions and evaluations of products and brands (Bhaduri & Ha-Brookshire, 2011). For example, Carter & Curry (2010) focus on transparent pricing and found that when the price of a product is completely transparent (cf. distribution of transparency), consumers are willing to pay a premium price “to ensure that one or more supply-side agents receives an appropriate share of the retail proceeds” (p. 772). In this study it was shown that consumers are willing to pay more for transparency.
The Mental Construal
Within the cognitive science the mental construal is a “… basic concept of the Computational Theory of Mind” (Pitt, Fall 2013, p. 1). This theory postulates that
storage (in the mind/brain) of information-bearing structures (representations) of one kind or another” (Pitt, Fall 2013, p. 1). The construal level theory (Liberman, Sagristano, & Trope, 2002; Liberman, Trope, & Stephan, 2007; Trope & Liberman, 2010) theorizes that objects or event are mentally represented or construed. The theory implies that the mental construal can be divided into two levels, namely low- and high-level. Low-level construals can be viewed as concrete, and subordinate mental representations, such as a ‘German pointer’. High-level construals can be viewed as abstract, and superordinate mental representations, such as a mammal or animal.
The construal level is defined by Kardes, Cronley & Kim (2006) as the “degree of abstractness of mental representations”. Shifting from a concrete to an abstract representation involves disregarding contextualized features that appear secondary and incidental. But at the same time retaining certain features, which are primary, central and invariant. An example presented by Trope & Liberman (2010) is the shift from the low-level representation of an object as a cellular phone towards the high-level representation as a communication device (p. 441). During the shift information about size or form is omitted.
A primary factor that can change the level of mental construals is psychological distance, which is defined by Liberman, Trope, & Stephan (2007) as the extent of deviation from direct experience in time, space, as well as social, and hypothetical distance. Many studies have examined or discussed the effect of psychological distance on the mental construal (Trope & Liberman, 2010; Trope, Liberman, & Wakslak, 2007). For instance, the temporal distance can determine the level of abstraction (Kim, Zhang, & Li, 2008; Liberman et al., 2002; Trope & Liberman, 2003). Study 1 by Liberman & Trope (1998) focused on when an action was construed in a high- or low-level construal by changing the temporal perspective (tomorrow or next year). On the basis of the
participants’ descriptions of the actions, the results revealed that high-level descriptions were more common in the distant future condition. The same was true for low-level descriptions and the near future condition. Thus, there is a positive relation between temporal distance and the degree of abstractness concerning actions or objects. Changing the level (high/low) at which objects are represented can change the relative relevance of the different features of these objects (Maglio, Rabaglia, Feder, Krehm, & Trope, 2014). In line with this, changing the level can modify the way one thinks of, evaluates (Maglio et al., 2014; Trope & Liberman, 2010) and selects or chooses objects (Dhar & Kim, 2007; Fiedler, 2007; Lynch Jr & Zauberman, 2007; Trope et al., 2007). As Maglio et al (2014) states in her article: “this shift not only determines which types of features of the object are most cognitively salient but also configures downstream attitudes and preferences.” (p. 1084). As such, organizations can benefit from changing the relevance of certain features, as well as the consumers evaluation, by means of changing the level of construal accordingly.
Liberman, Trope & Stephan (2007) connects the level of construal to the lack of knowledge. They mention that as the subjective distance from direct experience of objects increases, we obtain less information about these objects. Intrinsically, the amount of information available for proximal and distal objects can lead to a concrete or abstract construal of these objects. They propose that because there is less information about distal objects compared to proximal ones, one will construe the distant object more abstract than the proximal one. This preposition links the psychological distance to level of construal, but it also refers to the amount of information available. An aspect omitted in their article is the quality of the information. Information can be divided in information quantity and quality. Studies have focused on the effect of increased information about a product on the decision making process, resulting in possible
information overload and influencing the information process capacity of the individual (Bawden, Holtam, & Courtney (1999); Schneider (1987) in Eppler and Mengis (2004). The information quality can modify information overload (Eppler & Mengis, 2004; Keller & Staelin, 1987). The information needs to be of high quality and comprehensible for the consumer. Thus, not only the amount of information but also the quality of information can have an impact on the level of construal. The providence of qualitative and relevant information is related to the disclosure of relevant, accessible, comprehensive, qualitative and reliable information, or transparency (Vishwanath & Kaufmann, 2001). People want more specific information to avoid choice uncertainty and increase trust (Christensen, 2002; Fournier & Avery, 2011). They need to understand the product, or rather transcend to the here and now of the direct experience of the product. This raises an important question of how transparency influences consumers’ thoughts, evaluation and choice of a product, brand or organization. The study by Lafferty & Goldsmith (1999) found that consumers are more favorable towards brands with transparent business practices. Thus in this case, the information about the sustainability or transparent practices of a business can have an impact on consumers’ evaluations of a product, brand and organization (Bhaduri & Ha-Brookshire, 2011). But before the evaluation or choice can be made, one will, in absent of direct experience, construe a mental representation of the product or brand. In this paper, we propose that transparency, as defined as disclosure of information, has an effect on the level of construal, such that higher transparency leads to lower level product construals, consequently leading to direct implications for behavior. Transparency refers to disclosure of information. More specifically, transparency entails the disclosure of concrete information, whereby the adjective concrete refers to the actual, specific, tangible thing and its features (Stevenson, 2010). Companies can be transparent about
their product or brand revealing information about business practices (Bhaduri & Ha-Brookshire, 2011; Hultman & Axelsson, 2007). Business practices reveal specific features, and processes that consumers can comprehend. As such, transparency can be defined as the disclosure of ‘concrete’ information. As mentioned above, the degree of abstractness lies within the amount of deviation from the direct experience of the object. The abstract representation is formed due to less information, or rather lack of knowledge. With the disclosure of concrete information, the lack of knowledge is reduced and less abstract representations are required. Thus, when knowledge of the consumer is enhanced by more concrete information about the object, the consumer has the ability the use this information to construe a more concrete mental representation of the object. Therefore, it is expected that the disclosure of information will lead to a low-level construal, resulting in the following hypothesis:
H1. Product transparency (non-transparency) leads to low (high) level of product representations.
As the quantity and quality of information are discussed, another aspect is the nature of the information. A consumer can either perceive the information as positive or as negative. Marketers often present information about a product or brand in a favorable light towards the consumer (Eisend, 2006; Eisend, 2007). They expect to influence consumers’ attitude by enhancing positive features of their product or brand (Crowley & Hoyer, 1994). By means of presenting positive features to influence preference, the information presented discloses more concrete information about the product. As such, it is expected that the disclosure of positive product information will lead to low-level representations of that product. An interesting research about attitude processing and CLT was conducted by Eyal, Liberman, Trope, and Walther (2004). By linking action
based considerations, in terms of pros and cons, with CLT (temporal distance) they found several results. One of the results was that pros are superordinate to cons, as “the subjective importance of cons depends on the existence of pros” (p. 781), whereas the importance of pros is not dependent or rather independent of the existence of cons. As such, they found an asymmetry in conditional importance whereby pros constituted a higher level of construal. Intrinsically, it is expected that disclosing positive information concerning brands, enhancing the pros, constitutes high-level brand construals. Thus, although the information itself makes the representation of the product more concrete, the positive nature of information will lead to abstract representations of the brand.
H2a. Positive product transparency (non-transparency) leads to low (high) level of product representations.
H2b. Positive product transparency (non-transparency) leads to high (low) level of brand representations.
Several studies have researched the impact of negative information and even “demonstrated that negative information has a stronger influence on overall impressions than does positive information of equal intensity” (Price, 1996, p. 53). This negativity effect occurs because people place more weight on negative information compared to positive information during the process of product evaluation (Ahluwalia, Burnkrant, & Unnava, 2000; Fiske, 1980). People show greater sensitivity towards negative information, which strongly influences their attitudinal and behavioral expressions (Ito, Larsen, Smith, & Cacioppo, 1998). One of the possible explanations for the negative effect is that, compared to positive information, negative information is more informative (Ahluwalia et al., 2000). If information is experienced as more informative, the negative information about a product provides more concrete
information about that product. As such, it is expected that the disclosure of negative product information will lead to low-level representations of that product. Besides the fact that the weight of information has a stronger impact on consumers, the value of the negative information is subordinate to positive information, as explained before. For instance during decision making, one will not consider only negative aspects, but will consider positive aspects first (Study 5 in Eyal et al., 2004). “The subjective importance of cons depends on the existence of pros” (p. 781). This hierarchical relationship implies that by disclosing negative information, enhancing the cons, consumers will construe subordinate, low-level representations. Intrinsically, it is expected that the disclosure of negative product information will constitute low-level representations of the product as well as the brand. The hypotheses have been summed in figure 1.
H3a. Negative product transparency (vs. non-transparency) leads to low (high) level of product representations.
H3c. Negative product transparency (vs. non-transparency) leads to low (high) level of brand representations.
Figure 1. Level of Construal Matrix 2x2
Nature of information Positive Negative
Sub
je
ct Product Low-level (-) Low-level (-) Brand High-level (+) Low-level (-)
Attitude
Attitude is generally defined in the literature “as evaluations of a product or brand” (Berger & Mitchell, 1989, p. 269), either favorable or unfavorable (Katz, 1960). Bohner and Dickel (2011) define attitude formation and change as “whenever people process information with the result of forming an evaluation of an object of thought (p. 397).” Disclosure of information can positively or negatively influence consumer preferences, source credibility, attitude, and purchase intention (Eisend, 2006). More specifically, it is expected that positive product transparency will positively influence product attitude. As mentioned earlier, people show greater sensitivity towards negative information, which strongly influences their attitudinal and behavioral expressions (Ito et al., 1998). As such, the following hypotheses have been formulated:
H4a. Positive product transparency positively influences product attitude. H4b. Negative product transparency negatively influences product attitude.
According to Gardner and Levy (1955), products are associated with brands, either clearly or vaguely. Favorable information about a product or its attributes can lead to more positive attitudes towards the product or increased product attribute beliefs (Olson & Mitchell, 2000). These product attributes can, besides product attitude, also establish brand associations (Keller & Lehmann, 2006). Vice versa, brands can help consumers choosing certain products, among others by implying a quality level or reducing risk. As Keller and Lehmann (2006, p. 740) indicate “Brands are built on the product itself…” or as they conclude “brands thus reflect the complete experience that customers have with products.” Intrinsically, it is expected that evaluations of the product impact brand evaluation. Thus, we hypothesize that as positive product
transparency will positively influence product attitude, the same will occur for brand attitude:
H4c. Positive product transparency positively influences brand attitude.
Following the reasoning above, negative or unfavorable information can lead to a negative brand attitude. Although marketers often aim for shedding favorable light on their products and brands, the disclaim of certain attributes of their products can enhance the positive claims (Eisend, 2007; Kuster & Eisend, 2012). This positive or persuasive effect of two-sided advertising has been discussed in the literature (Crowley & Hoyer, 1994; Eisend, 2006; Eisend, 2007; Kuster & Eisend, 2012). Under certain specific conditions, disclosure of unfavorable information can positively influence consumer preferences, by means of enhancing source credibility and truthfulness, brand attitude, and purchase intention (Eisend, 2006; Kuster & Eisend, 2012). In the age of transparency, companies have different strategies to disclose specific information to its customers, which may even be forced under certain circumstances. For example, it has been found that when companies disclose information voluntarily or before the announcement of another source (such as the competition or regulators), it has a stronger impact on source credibility (Crowley & Hoyer, 1994; Eisend, 2007). Nowadays, trying to hide information as opposed to being truthful may do more harm than good (Cohn & Wolfe, 2013). That does not mean that disclosure of negative information is without risk. For instance, even though negative information can increase source credibility, it may have a contrary effect on brand attitude. This leads to a trade-off, whereby “the amount of negative information particularly strengthens attitudes by increasing source credibility, and at the same time weakens them through the negative information presented in the message.” (Eisend, 2006, p. 189). The disclosing of negative
information is a strategic and risky move, but consumers appear to want more transparent information. Consumer research indicated that consumers are highly skeptical towards acts of transparency by companies, as consumers believe that companies disclose information involuntary or stage the information (Cohn & Wolfe, 2013). Disclosing negative information seems to be a solution for reducing this skepticism of the consumer. By disclosing negative product information, the message can be received as honest or truthful influencing its attributes, beliefs and attitudes. Thus, we postulate that negative information may positively influence or strengthen the evaluation of the brand.
H4d. Negative product transparency positively influences brand attitude.
Product Choice
Consumers have to make choices on a regular basis about different services, products, or actions. An overload of information provided in commercials, adds, billboards, internet, packaging, and worth of mouth (WOM) can create pre-choice confusion, (post-) choice uncertainty and therefore may lead to information overload (Scheibehenne, Greifeneder, & Todd, 2010). Within the labyrinth of information, much information is lost to the consumer, hidden away in small letters, or even omitted. During the decision-making process the choice of consumers can be affected by many external factors, such as price, options (Iyengar & Lepper, 2000; Scheibehenne et al., 2010), attributes (Richard & Allaway, 1993), and time (Reutskaja, Nagel, Camerer, & Rangel, 2011; Wright, 1974). But also internal factors, such as benefits, experience, knowledge (Bettman & Park, 1980), and implicit and explicit attitudes (Karpinski & Hilton, 2001, Study 2) can influence consumers’ choice, all of which can interact. Taking all factors into consideration is beyond the scope of this research. In this study, as more product or brand transparency
and organizations’ credibility are desired by consumers during their decision–making process (Goldsmith, Lafferty, & Newell, 2000; Lafferty & Goldsmith, 1999), we will explore the effect of product transparency on product choice.
During everyday decision-making, information is being processed purposively, which can affect thought and behavior. This purposiveness of thought and behavior is the foundation of integration information theory (Anderson, 2014). Anderson indicates that the purposiveness of behavior implies that “thought and action are conceptualized in terms of their functions in goal-directed behavior”, such as choosing a product (Anderson, 2014, p. 3). These thoughts and actions are based on values. Anderson (2014) theorizes a successively process starting with valuation (V), extracting information from the stimuli which lead to value representations. This is followed by an integration of these representations (I) leading to an implicit response, which again leads ultimately to the observable, explicit response (A). The process valuation can be positive or negative, and referring to approach-avoidance behavior, they are associated with goals. Hence, according to the theory, (more) positive value representations obtained from the information will lead to positive implicit and eventually positive responses. Concerning the current study, if indeed the information is positive, leading to positive value representations associated with choosing the product, the observable response would be buying the product. The opposite action would occur with negative information. In view of this theory, the relation between disclosing product information and product choice is expected to be positive:
H5. There is a positive relation between information valence and product choice.
Concerning consumer behavior, a great body of research has focused on the process from implicit to explicit response, or rather the attitude-behavior relationship (Ajzen,
1985; Berger, Ratchford, & Haines, 1994). The traditional view of this relationship indicates that attitudes lead to behaviors (Bentler & Speckart, 1981; Kahle, Klingel, & Kulka, 1981). This view is based on rationality, as it is often “considered to be logic or consistent for a person who holds a favorable attitude towards some object to perform favorable behaviors, and not to perform unfavorable behaviors, with respect to the object” (Ajzen & Fishbein, 1977, p. 889). The same can be said about unfavorable attitudes, as they lead to unfavorable behavior. The object of evaluation is often in general addressed as “the object” or “action. In marketing, as defined earlier, the object stimuli regularly referrers to products or brands. Thus, in view of previous theory and results, favorable product and brand attitudes are expected to increase product choice. This positive relationship has been hypothesized as follow:
H6. There is a positive relation between product attitudes and product choice. H7. There is a positive relation between brand attitudes and product choice.
Although research found supportive results concerning this traditional view (e.g. Bentler & Speckart, 1981), it did not always apply as contrary evidence was found by, among others, LaPiere's (LaPiere's (1934) & Corey (1937) in Armitage & Christian, 2003). Based on generally low correlation between the variables, Wicker (1969, p. 65) indicated that “taken as a whole, these studies suggest that it is considerably more likely that attitudes will be unrelated or only very slightly related to overt behaviors than that attitudes will be closely related to actions”. With these different positions regarding the relation between attitudes and behaviors, research started to focus on variables that could influence this relationship. The moderator attitude strength is one example, which can be divided in dimensions and measured in many different ways (Krosnick, Boninger, Chuang, Berent, & Carnot, 1993). The strength of attitudes on behavior might also be
moderator by the level of construal in which the product attitude is formed. As the time horizon of choosing a product, often in a supermarket, is small, consumers might change their strategy of processing information (Bettman, Johnson, & Payne, 1991). Known from the dual-process model, there are two types of information processing (Petty & Cacioppo, 1986), namely systematic vs. heuristic processing. As defined by Fujita, Eyal, Chaiken, Trope, and Liberman (2008) systematic processing involves “careful, elaborative, and reflective thinking and reasoning about relevant stimuli (such as the arguments presented, the source of and causes for the arguments)” (p. 2). Heuristic processing is described as “the application of well-learned decision rules” that will help reduce extensive information processing and complexity in choosing a product (Fujita et al., 2008, p. 2). For example attribute based heuristics, which can focus on the comparison of a single attribute, which is cognitive comprehensible (Bettman et al., 1991), as well as allowing evaluations to be formed quickly. Concerning CLT, high and low-level construals focus on and enhance different features of objects (abstract vs. concrete, desirability vs. feasibility). The characteristics of low-level construals, among which concrete, vivid and easy to visualize, align with the effects of time horizon (Kardes et al., 2006) or heuristic processing (Fujita et al., 2008). Additionally, low-level construals have more immediate and direct implications for behavior (Kardes et al., 2006; Vallacher & Wegner, 1989). Therefore, concrete representations (low-level construals) would have a higher influence on behavior, which are related to consumers’ attitudes (Kardes et al., 2006). Hence, if positive product attitudes, or positive attribute beliefs, are formed within low-level construal, positive product attitude should have a stronger effect on choosing that product. Therefore, we predict that positive attitude construed on a low-level vs. high-level will have greater impact on product choice. As such, product construal moderates the relation between product attitude and choice
behavior, such that based on a low-level product construal, product attitudes of consumers have the higher ability to predict choices.
H8. The positive relation between product attitudes and product choice is moderated by product construal level, such that the effect is stronger when product construal level is low.
Besides forming attitudes towards products, attitudes can also be formed towards brands. Defined by the American Marketing Association a brand is a “name, term, sign, symbol, or design, or a combination of them, intended to identify the goods and services of one seller or group of sellers and to differentiate them from those of competitors” (in Kotler & Keller, 2009). The differentiation of a brand can lie within the products and services a brand provides, but they may also lie in what the brand itself represents. The concept brand can differentiate on a more abstract level, being more symbolic, emotional or intangible. During construing brand attitudes it is not always easy to visualize or contextualize the brand as the distance between direct experience and brand is larger than the target product and its functionality. As such, in low-level construals concrete and subordinate features are enhanced, whereby brand attitudes would have a lesser impact in the decision making. In high-level construals the focus is on abstract, primary, superordinate features, which do not have to be contextualized, such as a brand. If positive brand attitudes are formed/construed in a high level construal, the abstract concept of the brand would be enhanced and have a greater impact on behavior. Therefore, we predict that positive brand attitude construed on a high-level vs. low-level will have greater impact on product choice. Hence, brand construal moderates the relation between brand attitude and product choice, such that
consumers’ brand attitudes have the higher ability to predict choices when consumers hold high-level construals (figure 2).
H7 The positive relation between brand attitudes and product choice is moderated by brand construal level, such that the effect is stronger when construal level is high.
Figure 2. Conceptual model: Conditional Indirect Effects in a Moderated-Mediation Model.
PRE-TEST (STUDY 1) METHOD
The aim of this pre-test was to select two product statements to be used in Study 2 that were considered equal transparent by respondents, yet opposite in nature of the information given (positive vs. negative).
The product chosen for this research was spiced gingerbread by Royal Peijnenburg. Nowadays a lot of focus is on biological and healthy food products and claims. People have become more aware of certain products, its ingredients and impact on health. Besides the end product, focus has also shifted towards supply chain transparency as well as production process transparency. Even though these are interesting topics, the focus in this study will be the transparency of the product itself as well as the brand through its product. In order to test this effect of transparency the product had to be familiar (M = 1.01, SD = .11).
Participants & design
A total of 50 participants entered the study. Due to the length several participants did not finish the survey. As the survey was divided into two parts, the first part concerning the nature of the information contained less missing data (n=45), compared to the second part regarding the definition of transparency (n=33). Thus, the final sample consisted of 45 participants concerning the nature of information of the statements and 33 participants concerning the transparency of the statements. Participants were recruited through Qualtrics1. The research used a within-subjects design, with claim
transparency and nature of information as the dependent variables.
1 Qualtrics is an online survey software tool. This software has been used and cited in professional and
Measures
Statements. In order to measure statements on their nature of information and
transparency, the statements were formulated based on the information presented by the products website, commercials, reports (Consumentenbond, 2009) and the label of the product itself. A total of 27 statements were presented to the sample.
Procedure
Participants were asked to rate the statements on a 7-point Likert scale. In the first part the statements were rated by the participants based on the nature of the information, ranging from negative (1) to positive (7). In the second part, participants were asked to rate the same statements based on the informational criteria’s used by Vishwanath et al. (2001) to define transparency. These criteria’s were informational relevancy, reliability, comprehensiveness, and quality. In order to test the hypothesis concerning transparency, participants were also asked to rate the statements on “concreteness”. The rating ranged from strongly disagree (1) to strongly agree (7).
RESULTS PRE-TEST
By comparing means different statements were found positive (5<M>7) and negative (1<M>3). Next, the degree of transparency of these statements was measured by computing a mean of the mentioned criteria per statement. Statements were considered transparent if their average was higher than 5 (M>5). By comparing the results, two statements held opposite means of favorability but both high means of transparency scores.
To examine whether favorability (positive/negative) differed between statements, a repeated measures ANOVA was conducted on average favorability scores with the type of statement (statement 1 vs. statement 2) as the within-subjects variable.
In line with our hypothesis, the analysis revealed a main effect of statement type, F(1,43) = 68,29, p < .001, 𝜂𝑝2 = .614, such that participants rated the favorability of statement 1
(M = 5.18, SE = .18) higher than that of statement 2 (M = 2.73, SE = .26). Thus, participants favored statement 1 more than statement 2.
Next, we examined whether the statements differed in transparency. To this end, a repeated measures ANOVA was conducted on mean transparency scores with statement type (statement 1 vs. statement 2) as the within-subjects variable. A main effect of statement type was absent, F(1,32) = .45, p = .506, 𝜂𝑝2 = .014, thus supporting
our hypothesis that transparency did not differ between statements.
Thus in line with the expectations, statement 1 scored higher on favorability than statement 2 (p < .001), while not differing in perceived transparency (p = .506). Thus, these two statements qualify to be used in the follow up experiment. The results are presented in Table 1.
Table 1. Transparency Statements
Condition Statement M Favorability Σ M Transparency σ Transparent + positive information Statement 1:
Bevat weinig vetten (1,5g/100g). 5.18 1.17 5.09 1.16
Transparent
+negative information
Statement 2:
Bevat meer dan 2 suikerklontjes per plakje.
2.73 1.74 5.26 1.17
FIELD EXPERIMENT (STUDY 2) METHOD
We designed a field study to investigate the hypothesized effects outlined in the theoretical model.
Participants
A total of 173 participants entered the study. Due to missing data and invalid responses, data from 23 participants were removed from the sample. Thus, the final sample consisted of 150 participants (59 men and 91 women), ranging in age from 14 to 81 years (M = 35.08 years, SD = 14.88 years). The educational background of the sample consisted of 73 (48,7%) University graduated, 43 (28,7%) HVE graduated and 21 (14%) MVE graduated. The remaining 13 (8,7%) participants had a lower educational level (Secondary school). Participants were recruited at a supermarket.
Design
The research used a 3 (non-transparency, positive, negative) x 2 (high, low) between-subjects design. The independent variable was product transparency, divided into non-, positive, and negative transparency. Several dependent variables were investigated in this study, namely the mental construal of the product and brand, product attitude, brand attitude and product choice (Table 2). Age, gender, educational level and source disclosure were the control variables.
Table 2. Description of the Variables and their Levels.
Variable Type Scale
Transparency Independent variable Nominal (0= non, 1=positive, 2=negative)
Product Construal Dependent variable / Moderator Interval (Numerical data)
Brand Construal Dependent variable / Moderator Interval (Numerical data)
Product Attitude Dependent variable / Mediator Interval (Numerical data)
Brand Attitude Dependent variable / Mediator Interval (Numerical data)
Product choice Dependent variable Dichotomous
Measures
In order to measure each element, various instruments were used and modified to be suitable for this research. All items are found and used in previous research.
Transparency. Transparency is defined as the disclosure of concrete
information, which is relevant, reliable, comprehensive, and qualitative (Vishwanath & Kaufmann, 2001). In order to develop the stimuli for this study a pre-test was conducted, whereby items (statements) were identified to be transparent, as well as containing negative or positive information. These two claims formed the two product transparency conditions vs. the control condition non-transparency.
Mental Construal. A measurement used to conceptualize the level of
abstractness was the level of inclusiveness or categorization breadth. The term inclusiveness indicates “the degree of similarity among objects in a category” (Gutman, 1982, p. 63). Abstract categories (e.g. food) have a higher level of inclusiveness than concrete, subordinate categories (e.g. snacks) (Liberman et al., 2002). In the original study 1, individuals had to read about 4 scenarios, either in the near or distant future. For each of the 4 scenario’s, the participants had to classify a list of 38 related objects into as many mutually exclusive groups as they considered appropriate (Liberman et al., 2002; Trope & Liberman, 2003) The authors of the study (Liberman et al., 2002)
expected the individuals to use fewer, broader categories to group the objects related to distant future situations than to near future situations. In the current study the measure of categorization breadth was used to measure the level of construal. This measurement was modified marginally to operationalize it for this research. First of all, the participants were not presented with 4 scenarios or actions, but they were shown only one advertisement obtaining information. As such this research did not make use of the temporal distance effect pertained in the scenarios. In order to measure product construal every participants had to classify an identical list of 31 related objects into as many mutually exclusive and exhaustive groups as they considered appropriate. The instructions for the participants were adapted from the study by Liberman et al. (2002, p. 526), minimally modified in Dutch due to translation:
“Take a look at the following items and place them into groups by writing the items that belong together next to each other on the right, and then circling the items that belong in the same group. Please make sure to include every item, even if you would not use it in reality. Additionally, please do not overlap, that is, place each item in only one group.’’ The same process was conducted to measure the brand construal. Only this time the participants had to classify an identical list of related objects to the brand instead of the product. As noted earlier, participants who categorize objects into fewer, broader groups (high inclusiveness) attain an abstract construal about either the product or brand.
Product/brand attitude. At the end of the questionnaire the participants were
asked about their current attitude towards both product and brand. Product (brand) attitude refers to the participants overall evaluation of the object (brand) (Lee, Park, & Han, 2008). A semantic differential approach was used (Andrich & Styles, 1998;
the participants on three bipolar adjective pairs on a 5-point rating scale (bad–good, unsatisfactory–satisfactory, and unfavorable–favorable).
Product Choice. The participants were asked if they wanted to buy the product
after finishing the survey and seeing the advertisement. Instead of proving a discount coupon for participants for payment at the cash registers (Study 1 in Iyengar & Lepper, 2000), the product was directly distributed for a price below the current price in the supermarket. As such, actual consumer purchase behavior was measured.
Control variable. In order to truly test the effect of product transparency on
product choice transparency the disclosure source needed to be controlled. Therefor two conditions were added with a different disclosure source.
Procedure
The data was gathered trough a cross-sectional survey suggested in Lewis, Thornhill & Saunders (2007). The questionnaire was distributed randomly at a supermarket (Appendix A) on the same repeatable days. All participants were appointed randomly to one of the five conditions. After filling in three basic questions (age, gender and education) the participants were shown an ad that was different for each condition. In condition 1, 2, and 3 the text within the ad was distributed by the brand itself as the source. In condition 4 ad 5 the text in the ad was communicated by a different source (Appendix B). After the observation of the ad, the participants were asked to categorize several objects related to the product, as well as several objects concerning the brand. The follow up two questions were concerned with the product and brand attitude. To control for brand familiarity, the last question of the survey asked if the participants were familiar with the brand. After finishing the questionnaire the participants were asked by the researcher if they would like to buy the advertised product for €1,-.
RESULTS FIELD EXPERIMENT
First, in order to give an overview of the variables used in the upcoming analyses a correlation matrix was conducted with the variables’ means, standard deviations and reliability values (Table 3). Second, the relations between product transparency and the dependent variables mental construal and attitude were investigated. Third, the possible positive relation between attitude and product choice was examined. Finally, the relationship between product transparency and product choice combined with the predicted underlying mechanism were investigated.
*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed). Listwise N=150
Hypotheses Testing
The effect of product transparency on product construal (H1, H2a & H3a)
The objects concerning the product as well as the brand were classified into different categories by the participants. The number of these categories was counted, with a higher count meaning that participants applied a more concrete categorization, whereas a lower count corresponds with a broader, more abstract classification. To examine whether product transparency increases the amount of categories mentioned by Table 3. Demographic Variables, (In)dependent variables: Descriptive Statistics and Correlations
Variables M SD 1 2 3 4 5 6 7 1. Age 35.08 14.88 - 2. Education 5.09 1.14 -.155 - 3. Product Construal 6.15 2.52 .128 .096 - 4. Brand Construal 6.43 2.79 .010 .124 .657** - 5. Product Attitude 3.63 .91 .080 -.215** .008 .014 (.86) 6. Brand Attitude 3.87 .72 .099 -.275** -.037 -.006 .712** (.84) 7. Brand Familiarity 1.01 .08 -.061 -.079 -.005 .105 .033 .053 -
participants (indicative of a low level product construal), a one-way ANOVA was conducted on product categorization with condition (positive brand disclosed product transparency vs. negative brand disclosed product transparency vs. control) as the between subjects variable. Contrary towards the hypotheses, a main effect of condition was absent, F(2,87) = .534, p = .588, n2 = .01, meaning that there was no difference
between conditions on product categorization.
The effect of product transparency on brand construal (H2b, H3b)
To examine whether product transparency decreases the amount of groups, indicating a high level brand construal, a one-way ANOVA was conducted on brand categorization with condition (positive brand disclosed product transparency vs. negative brand disclosed product transparency vs. control) as the between subjects variable. Contrary towards the hypotheses, a main effect of condition was absent, F(2,87) = 2.171, p = .120,
n2 = .05, meaning that there was no difference between conditions on brand
categorization. However, post hoc analyses revealed that, in line with H2b, participants formed marginally significant (p = .075) fewer groups when shown positive product transparency (M = 5.57, SD = 2.29) with a maximum of 12 groups compared to the control group (M = 6.97 SD = 3.09) with a maximum of 15 groups, meaning that participants tend to hold a higher level of construal towards the brand, when the information is positively presented by the brand. Even, participants formed marginally
significant (p = .075) fewer groups when shown positive product transparency (M = 5.57, SD = 2.29) with a maximum of 12 groups compared to the negative product
transparency (M = 6.97 SD = 3.51) with a maximum of 16 groups, meaning that participants tend to hold a higher level of construal towards the brand, when the information is positively presented by the brand compared to negative information.
The differences between product and brand construal within conditions.
A repeated measures ANOVA was used to compare the amount of groups of the two types of categorization. The ANOVA results show that there was a main effect of categorization, F(1,87) = 7.848, p = .006, 𝜂𝑝2 = .083, indicating that participants formed
more categories (high categorization breadth) about the brand (M = 6.50, SD = 3.04) than for the product (M = 5.94, SD = 2.52). Further analysis showed that there was a
significant interaction between the types of categorization and condition,
F(2,87) = 3.996, p = .022, 𝜂𝑝2 = .084, meaning that the conditions differed in amount of
product and brand categories.
Figure 3. Mean numbers of Product & Brand Categorization in the different Conditions (n=90).
H1 Product transparency (non-transparency) leads to low (high) level of
product representations.
Rejected
H2a Positive product transparency (non-transparency) leads to low (high)
level of product representations.
Rejected
H2b Positive product transparency (non-transparency) leads to high (low)
level of brand representations.
Marginal sig.
H3a Negative product transparency (non-transparency) leads to low (high)
level product representations.
Rejected
H3b Negative product transparency (non-transparency) leads to low (high)
level of brand representations.
Rejected
Figure 4. Hypotheses concerning Mental Construal.
5.77 5.73 6.33 6.97 5.57 6.97 0 1 2 3 4 5 6 7 8
Controle conditie:
Non-transparency C2: Positive transparency,brand disclosure C3: Negative transparency,brand disclosure
The role of product transparency on product and brand attitude (H4a,b,c,d)
The reliability statistics and Cronbach’s alpha were measured for the variables product attitude and brand attitude. Reliability of the scale for product attitude was good, (Cronbach’s α = .863). The same accounted for the items measuring brand attitude. The reliability of the items (Peijnenburg_Q1, Peijnenburg_Q2, Peijnenburg_Q3) measuring brand attitude was good (Cronbach’s α = .841). All items concerning the two variables have been computed in mean scales.
To examine whether positive product transparency increases product attitude a one-way ANOVA was conducted on product attitude with condition (positive brand disclosed product transparency vs. negative brand disclosed product transparency vs. control) as the between subjects variable. Contrary towards the hypotheses, a main effect of condition was absent, F(2,87) = .450, p = .639, n2= .01, meaning that there was
no difference between conditions on product attitude.
Next, we ran the same one-way ANOVA on brand attitude. The analyses revealed no main effect of condition on brand attitude F(2,87) = .205, p = .815, n2 = .005, meaning
that our hypothesis that product transparency increased brand attitude was not confirmed (figure 5).
The differences between product and brand attitude within conditions.
Again, a repeated measures ANOVA was used to compare attitudes between the two types of attitude. There was a main effect of attitude, F(1,87) = 18,86, p < .001, 𝜂𝑝2 = .178,
indicating that participants held a more positive attitude towards the brand (M = 3.99,
SD = .63) compared to the product (M = 3,67, SD = .96). However, the interaction
Figure 5. Mean numbers of Product & Brand Attitude in the different Conditions (n=90)
H4a Positive transparency regarding a product positively influences product attitude. Rejected
H4b Negative transparency regarding a product negatively influences product attitude. Rejected
H4c Positive transparency regarding a product positively influences brand attitude. Rejected
H4d Negative transparency regarding a product positively influences brand attitude. Rejected
Figure 6. Hypotheses concerning Attitude.
The role of product transparency and brand disclosure on product choice (H5)
The relationship between transparency and product choice was tested by asking the participants if they wanted to buy the product for €1,-, after finishing the survey and seeing the advertisement. It was expected that product transparency would increase product choice, for both positive and negative information. The Pearson Chi-Square test indicates there is no evidence for an effect of product transparency and brand disclosure on product choice as p > .05 (χ2 = 3.904, df = 4, p = .419). This indicates that product
transparency and brand disclosure did not increase the likelihood that participants will buy the product (figure 7).
3.8 3.57 3.66 4.03 4.01 3.93 3 3.5 4 4.5 5
Controle conditie:
Non-transparency C2: Positive transparency,brand disclosure C3: Negative transparency,brand disclosure
Figure 7. Product Choice in the different Conditions, divided by the Nature of the Information (n=150) 26,67 % 36,67 % 23,33 % 0 10 20 30 40 50 60 70 80 90 100 Control condition:
Non-transparantie disclosureC3: Brand C5: Non-branddisclosure
Negative information
26,67 % 26,67 % 43,33 % 0 10 20 30 40 50 60 70 80 90 100 Control condition:Non-transparency disclosureC2: Brand C4: Non-branddisclosure
Positive information
The effect of product and brand attitude on product choice (H6 & H7).
To examine whether product and brand attitude positively affects product choice, a binary logistic regression was conducted. The logistic regression indicated that the model predicting the effect of product and brand attitude on product choice was not significant (χ2 = 2.745, df = 2, p = .253). As the null model predicted an overall accuracy
of 70%, the new model predicted in classification accuracy the same percentage. Adding product and brand attitude did not increased or decreased the likelihood of a correct prediction of buying the product. The Hosmer and Lemeshow test indicates that our model did fit the data as p > .05 (χ2= 2.596, df = 7, p = .920), only this test was sensitive
to sample size because of the formulation of small subjects clusters. Nagelkerke’s Pseudo R2 was .043 again indicating a poor fit as only 4,3% of the variance in the
dependent variable was explained by the predictor variables. Even as individual
predictors, both variables were not significant (product attitude p = .118, brand attitude p = .156). Henceforth, product and brand attitude were weak predictors
of product choice, not supporting hypotheses 6 and 7.
The conditional indirect effect of brand disclosed product transparency on product choice through attitude (H8 & H9)
The variable attitude (product attitude (M1) & brand attitude (M2)) was modeled as
mediator between the independent variable brand disclosed product transparency (X) and binary dependent variable product choice (Y). The model included all possible indirect effects between X and Y and also allows the effects of M1 and M2 on Y be linearly
moderated by the level of product (V) and brand (Q) construal. In order to test these hypotheses a second stage moderation model was used described by Edwards and Lambert (2007). More specifically, a second stage moderated parallel multiple
mediation was used as there were two mediators, which operated parallel according to the Process macro (Hayes, 2012). Operating parallel means that the mediators product and brand attitude were not modeled to affect one another. Even though the model excluded this relationship, there was a strong and positive correlation between the two variables, r(148) = .712, p < .001.
To examine the relation of positive brand disclosed product transparency on product choice, the PROCESS macro of Hayes (2012) was used. The PROCESS command with model = 16, specifying a moderator-mediator model, generated the model of the direct effect, conditional indirect effects, and interactions. As the assumption of normality of the sampling distribution of the indirect effects was questionable because the sample was small, the indirect effects of attitude were bootstrapped (5.000 resamples) with 95% confidence intervals (CIs). The direct effect of positive brand disclosed product transparency on product choice was not significant, as evidence by the 95% bootstrap CIs included zero (c’ = .3031, CI95% [-1.0737, 1.6799], p = .666). This
finding was consistent with previous results of the direct effect. The unconditioned effects of product transparency on product (a1) and brand attitude (a1a) were also not
significant (a1 = -.2333, CI95% [-.7297, .2631], p = .351 and a1a = -.0222, CI95% [-.3513,
.3069], p = .893). Furthermore, the interaction product attitude and product construal was not significant (p = .900), meaning there was no moderating effect of product construal on product attitude and product choice (b4 = .0418, CI95% [-.6111, .6948]). The
interaction brand attitude and brand construal was also not significant (p = .959), meaning there was no moderating effect of brand construal on brand attitude and product choice (b5 = -.0208, CI95% [-.8145, .7730]). Equally, the conditional indirect
effects of product and brand attitude were not significant at any level of the moderators, as evidence by the 95% bootstrap CIs included zero. Appendix C, Table 4b shows the
output of the conditional indirect effects at different values of the moderator construal (mean and +/- 1 SD). Thus, the indirect effect through attitude did not depend on the level of construal.
To examine the relation of negative product transparency on product choice, again the PROCESS macro of Hayes (2012) was used. The PROCESS command with model = 16, specifying a moderator-mediator model, generated the model of the direct effect, conditional indirect effects, and interactions. Again the indirect effects were bootstrapped (5.000 resamples) with 95% CIs. The direct effect of negative brand disclosed product transparency on product choice was not significant, as evidence by the 95% bootstrap CIs included zero (c’ = .6198, CI95% [-.1209, 1.3605], p = .101). This
finding was consistent with previous results of the direct effect. The unconditioned effects of product transparency on product (a1) and brand attitude (a1a) were also not
significant (a1 = -.0722, CI95% [-.3117, .1672], p = .584 and a1a = -.0500, CI95% [-.2081,
.1081], p = .529). Furthermore, the interaction product attitude and product construal was not significant (p = .111), meaning there was no moderating effect of product construal on product attitude and product choice (b4 = .9222, CI95% [-.2131, 2.0576]).
Contrary, the interaction brand attitude and brand construal was significant (p = .0497), meaning there was a moderating effect of brand construal on brand attitude and product choice (b5 = 1.2375, CI95% [.0018, 2.4731]). The conditional indirect effects of
product and brand attitude were not significant at any level of the moderators, as evidence by the 95% bootstrap CIs included zero. Appendix C, Table 5b shows the output of the conditional indirect effects at three values of the moderator construal (mean and +/- 1 SD). Thus, although the conditional indirect effects through attitude did not depend on the level of construal, there was an interaction effect of brand attitude and brand construal on product choice, meaning that when the amount of
