How Self-confidence Moderates the Effect of Two-sided Messages on Purchase Intention

Abstract

This article seeks to clarify how two-sided messages work by explaining their causal effects on purchase intention moderated by the self-confidence of individuals. Three studies were performed. Study 1 measured individuals’ levels of self-confidence and their purchase intentions in front of one- or two-sided messages. Studies 2 and 3 manipulated individuals’ levels of self-confidence and analysed the strategies used to evaluate alternatives in cognitive decision-making. Both studies elaborated on the underlying mechanisms by which low (study 2) and high (study 3) self-confident individuals process two-sided messages, explaining the results of study 1. Self-confidence is a moderator variable for the effect of two-sided messages on purchase intention. When an individual’s self-confidence level is high (low), exposure to a two-sided message will increase (reduce) purchase intentions compared to exposure to a one-sided message. Low self-confident individuals apply a non-compensatory strategy to evaluate the offer, whereas high self-confident individuals apply a compensatory strategy to evaluate the offer. The findings suggest a communication strategy oriented to inform and increase consumers’ self-confidence. The findings contribute to a more detailed and refined knowledge of the underlying mechanisms by which two-sided messages are processed.

Keywords

Two-sided messages self-confidence cognitive decision-making compensatory strategy non-compensatory strategy

Introduction

The use of two-sided messages (2SMs), in which the source takes into account positive and negative sides of an issue but actually still favours one side, is an effective way of communicating persuasively (Hovland, 1954). In other words, the sender provides, intentionally or not, positive and negative information. Conversely, one-sided messages (1SMs) present only positive attributes (Crowley & Hoyer, 1994). For the last decade, consumers have no longer been limited to one-way, seller-to-buyer, communications (Moe & Trusov, 2011). Consumers are nowadays subjected to more information, from different sources like network online communities (Park & Cho, 2012), and different types of information, such as product (goods and services) experiential descriptions (Zhang et al., 2016) and reviews (Zhang et al., 2018). This means that consumers may be receiving 2SMs about products in many different forms. For example, individuals could read a publicity report about the pros and cons of using a product, watch a ‘YouTuber’ assessing its functions, and get involved in an online community related to the product. Those consumers would then be conscious of the positive and the negative attributes of the offer. In this scenario, companies need to understand the effects of 2SMs containing information about their products’ attributes, in order to reduce harmful effects and increase beneficial effects.

The assertion of Eisend (2006) is still valid, that is, knowledge of the effectiveness of 2SMs, and of the influences of potential moderating and mediating variables, are valuable to marketers in showing how communication strategies aimed at consumers can be improved. However, there is a lack of research into the persuasive impact of 2SMs (Cornelis et al., 2015). Therefore, this research seeks to clarify how 2SMs work by explaining their causal effects on purchase intention, moderated by the self-confidence of individuals, and to examine the underlying mechanisms by which 2SMs are processed.

The literature review identified two theoretical gaps in the study of the effect of 2SMs. The first one is the lack of clear explanations on the functioning of the effect of 2SMs on individuals’ responses, hampering their implementation by communication professionals. There have been several attempts to model and interpret the effect of 2SMs (Bohner et al., 2003; Chen, 2016; Cornelis et al., 2014, 2015; Eisend, 2007, 2008, 2010; Hastak & Park, 1990; Hunt et al., 1982; Kamins & Marks, 1987; Kao, 2011; Pizzutti et al., 2016; Trifts & Häubl, 2003). However, the diversity of the results, as highlighted by Eisend (2006), indicates the necessity of further research on moderating variables (Cornelis et al., 2015). Moreover, in the context of decision-making, moderators of negative information about the attributes of a product remain largely unexplored (Hair & Bond, 2018).

Research on persuasion is heavily fragmented, and it is difficult to generalise findings from the many diverse field studies (Hornik et al., 2016), as well as findings from individuals with different characteristics or situations. There is evidence that trivial attributes, which would ordinarily be ignored, may take on special meaning when individuals are in a state of high social risk, and that individuals who are particularly confident in their ability to fulfil their goals should be less likely to rely on trivial attributes (Xiao, 2016). In that research, the author concluded that self-confidence influences the evaluation of products’ attributes. Thus, the valence of primary and secondary attributes in 2SMs, considering the individuals’ level of self-confidence, could clarify the effect of 2SMs on individuals’ responses. In this scenario, the first theoretical contribution of this research is to offer better explanations of the functioning of the effect of 2SMs on individuals’ purchase intentions, by testing how self-confidence moderates that effect.

Research into consumer behaviour has shown that there are still gaps in our knowledge of how different individuals engage in the purchase decision-making process and how that process unfolds (Karimi et al., 2015). For this reason, a current interest in decision research is attribute elaboration strategy, understood as the process by which individuals interpret, organise and simplify attribute information to facilitate decision-making (Hair & Bond, 2018). Literature does not yet explain how individuals process 2SMs during the decision-making process, specifically when those individuals differ in their level of self-confidence, evidencing our second theoretical gap. Knowledge of how individuals with different levels of self-confidence process 2SMs should be of particular interest for communication management decisions.

Decision processes and product judgement theories suggest that consumers base their decision process patterns strategically according to context-specific constraints (Chang, 2005). Recent findings show that the importance of the product attributes and the valence of the information related to those attributes are critical factors in determining the extent to which attribute information will be processed (Hair & Bond, 2018). Those authors suggest that the attributes evaluation strategy may vary with the individuals’ familiarity with the object under evaluation, and highlight that investigation of this possibility would be an intriguing extension of their work. Following Hair and Bond’s (2018) suggestion, this research proposes that the attribute evaluation strategy will be different for individuals with different levels of self-confidence (a subjective evaluation), instead of the suggested construct familiarity (an objective evaluation). Self-confidence is defined as how much an individual feels capable or secure in relation to their purchase decisions (Bearden et al., 2001): it is an individual’s auto-evaluation of his/her self. Lesser comprehensive, product familiarity represents product-related experience and accumulated information (Lacey et al., 2010), forming the objective basis for consumers to evaluate a specific offer (Park & Kim, 2003). Because self-confidence is more of a self-perception, rather than how truly the individual is self-confident (Hellén & Saaksjarvi, 2011), it is possible to design interventions to de/increase individuals’ self-confidence (Atherton et al., 2016), which would be potentially useful for practitioners. Therefore, the second theoretical contribution of this research is to reveal the attribute evaluation strategy by which individuals’ self-confidence moderates the effect of 2SMs on purchase intention.

Theoretical Development of the Two-sided Message’s Effect

2SMs strengthen perceptions of source credibility (Bohner et al., 2003; Eisend, 2007, 2010; Hastak & Park, 1990; Hunt et al., 1982; Kamins & Marks, 1987; Trifts & Häubl, 2003). However, a meta-analysis that aggregated data on two-sided advertising showed that 2SMs also have a positive influence on consumers’ responses (Eisend, 2006). Specifically, that author concluded that such messages would enhance source credibility, reduce negative cognitive responses and impact positively on brand attitude and purchase intention. This is because consumers may think that the purpose of a 1SM is to sell a product by highlighting chosen features, whereas the perceived objective of a 2SM would be to describe the complete characteristics of a product with honest information (Eisend, 2007). Previous research suggests that, in order to work, 2SMs had to meet two conditions. First, the characteristics of the product described negatively had to relate to its secondary attributes (Crowley & Hoyer, 1994; Hoyer & Macinnis, 2008; Kao, 2011, 2012). Secondary attributes are trivial: those judged by experts to be of little importance to a product’s performance or quality (Carpenter et al., 1994). Primary attributes form the essential functionalities of the product (Kao, 2012). Second, that the amount of negative information could never exceed the amount of positive information (Crowley & Hoyer, 1994), and should ideally not exceed 40 per cent (Golden & Alpert, 1987).

The results of the meta-analysis presented by Eisend (2006) indicated that there should be continued investigation into the substantive moderating and mediating variables that influence the effect of 2SMs, such as involvement or need for cognition. Since then, several studies have tested moderators and mediators of the effect of 2SMs on consumers’ responses. Eisend (2007) tested four models of the 2SMs’ effect, by aggregating data from a meta-analytic procedure and performing two experimental studies. The conclusion was that the parsimonious model with the best fit was based on the ‘Attribution Theory’. In this model, the dependent variable is attitude towards the brand and the mediator variables are source credibility and attitude towards the advertiser. It explains that 2SMs would increase source credibility and positive attitudes towards the advertiser, which, in turn, would enhance attitudes towards the brand. The other model tested successfully was the ‘Integrative Model’. In this model, the dependent variable is purchase intention and the mediator variables are source credibility, cognitive responses, attitudes, attention and motivation to process the advertisement and perceived novelty. It explains that 2SMs would increase source credibility, leading to positive and negative cognitive responses and attitude towards the brand, this one having an effect on purchase intention. Source credibility would also enhance attention and motivation, resulting in positive effects on attitudes and cognitive responses, as well as on purchase intention. Finally, by following the source credibility and the attention and motivation pads, 2SMs would increase perceived novelty, and so have an effect on purchase intention.

Eisend (2010) confirmed that an individual’s cognitive load might moderate the impact of 2SMs that disclose important attributes negatively. In that research, the author concluded that when participants are under cognitive load, disclosure of negative attributes increases positive attitudes towards the brand, because negative information will enhance source credibility, so that individuals may fail to infer product uniqueness, increasing their attitudes towards the brand. In a different vein, Kao (2011) found that, for 2SMs, persuasion and favourable brand attitudes are more likely to occur when individuals are under moderate time pressure rather than under low or high time pressure. More importantly, in his research, Kao found that the need for cognition, defined as an individual’s propensity to engage in and enjoy cognitively demanding tasks (Cacioppo et al., 1984), moderates the effect of 2SMs on purchase intention. The greater the need for cognition, the greater is the effect of 2SMs on purchase intentions (in comparison to the effect of 1SMs). This is because the need for cognition affects individuals’ motivation to process information.

These findings reinforce the need to identify the factors that may increase the effectiveness of 2SMs. The effect of 2SMs was explained from a rational point of view by Florack et al. (2009), who stated that the stronger the promotion focus, the stronger the positive effect of two-sided ads (compared to one-sided ads). Because individuals in a promotion focus want the maximisation of positive results, a 2SM would make them feel more confident than a 1SM, increasing their positive responses. Similarly, it was found that it is possible to enhance trustworthiness through a 2SM, without mitigating the intentions of buying, by controlling the importance of the negative information provided (Pizzutti et al., 2016). These findings, confirming the importance of individuals’ confidence and trustworthiness, indicate that the individual’s self-confidence could have a moderating effect on the effectiveness of 2SMs. The reasoning is that both confidence and trustworthiness are related to the individual’s self-confidence.

Prior studies, summarised in Table 1, have shown that the effect of 2SMs is conditioned to situational factors, such as time pressure (Kao, 2011), disclosure uniqueness, cognitive load and involvement (Eisend, 2010). Negative attributes are also important (Pizzutti et al., 2016). Other evidence shows that the characteristics of the messages, such as regulatory congruence (Cornelis et al., 2014), issue ambivalence and message arguments (Cornelis et al., 2015) may moderate their effect. Similarly, the impact of 2SMs may be moderated by individuals’ characteristics, such as self-regulatory focus and confidence (Florack et al., 2009), cognitive effort (Eisend, 2010) and the need for cognition (Kao, 2011). The premise of this research is that, despite the higher credibility of 2SMs, their effect on purchase intentions will depend on the characteristics of each individual. Specifically, it is proposed that each individual’s self-confidence moderates the effect of 2SMs on their purchase intentions. The argument supporting this proposition is that the moderator effect of regulatory focus (Florack et al., 2009), trustworthiness (Pizzutti et al., 2016) and need for cognition (Kao, 2011) depends on how self-confident the individual is to make a purchase decision. Therefore, it is proposed in this research that the effect of 2SMs could be better explained by considering the individual’s level of self-confidence during the purchase decision process. Studying how consumers’ self-confidence influences the effect of 2SMs on their purchase intention is thus likely to aid our comprehension of the two-sided effect.

Table 1.

Experimental Studies on the Effect of Two-sided Messages on Observers’ Responses

Authors	Independent (I)/Moderator (M) Variables	Mediator Variables	Dependent Variables
Eisend (2007)		Source credibility, attitude towards the advertiser and attitude towards the ad	Attitude towards the brand
		Perceived novelty, source credibility, attention, motivation, attitudes and cognitive responses	Purchase intention
Florack et al. (2009)	Self-regulatory focus (M)	Credibility and positivity of thoughts	Product evaluation
Eisend (2008)	Proportion of negative information (I) and involvement (M)	Message, evaluation: central and peripheral routes of persuasion and attitude towards the ad	Attitude towards the brand
Eisend (2010)	Disclosure uniqueness (I) and cognitive effort (M)	Source credibility and product uniqueness	Source credibility, attitude towards the brand and product uniqueness
Kao (2011)	Time pressure and need for cognition		Ad attitudes, brand attitudes and purchase intention
Cornelis et al. (2014)	Regulatory congruence (I) and self-regulatory focus (M)	Central and peripheral routes of persuasion	Processing style influences regulatory Congruence effect
Cornelis et al. (2015)	Issue ambivalence: (non-refutational and refutational) (M) and message arguments: affective and cognitive (M)		Credibility
Pizzutti et al. (2016)	Attribute importance	Trustworthiness	Purchase intention

Source: The authors.

Self-confidence: Moderation and Hypothesis Development

Self-confidence reflects the subjective self-assessment of an individual’s ability to create positive experiences as a consumer (Adelmann, 1987). Therefore, it is a reflection of individuals’ perceptions of their ability to make effective decisions related to consumption, and to protect themselves from being cheated or treated unfairly (Bearden et al., 2001). Studies on self-confidence usually segregate consumers at high and low levels, because their behaviour could be different at these contrasting levels.

Consumers with high self-confidence (HSC) are more likely to engage in intense information research (Chaplin & John, 2007; Loibl et al., 2009), as they seek to know all about the product before deciding to purchase, diminishing uncertainty (Lutz & Reilly, 1974). Information search is part of the factor ‘need for cognition’, proposed by Kao (2011) as a moderator of the effect of 2SMs. The objective of the search for information by HSC individuals is to improve their knowledge of the product and the market, in order to make the best possible decision and to be satisfied with the purchase (Michaelidou & Dibb, 2006; Seiders et al., 2005; Smith & Sivakumar, 2004). Therefore, HSC individuals decide less compulsively and present a lower level of anxiety compared to low self-confidence (LSC) individuals (Locander & Hermann, 1979). A person’s mood can influence self-confidence at the time of purchase: individuals in a good mood tend to be more self-confident and make themselves heard when they are treated unfairly or have their rights violated (Hellén & Saaksjarvi, 2011). Thus, self-confidence is important for the well-being of the consumer and evoking good mood helps to support that feeling.

Consumers with LSC tend to make their decisions based on trivial attributes of the products to justify their decisions for themselves or for others, even though they have to create an additional goal to justify their choices based on these attributes (Carpenter et al., 1994). Because they based their decisions on secondary attributes, these additional objectives are beyond those related to the essential functions of the product, so they can be created at the time of product evaluation just in order to justify their decisions. They may arise from the need to reduce anxiety and uncertainty, and because the individual is unable to decide on what the purchase needs to do, so the decision is based on a secondary attribute (Schlenker & Leary, 1982). Consumers lacking in self-confidence are more likely to interpret trivial attributes in a biased way (Xiao, 2016). For this reason, a consumer with LSC will be biased towards the importance of secondary characteristics described negatively. Thus, an LSC individual will tend to decide based on secondary attributes (Carpenter et al., 1994). Because secondary attributes are described negatively in 2SMs (Crowley & Hoyer, 1994; Hoyer & Macinnis, 2008; Kao, 2011, 2012), it is expected that these individuals will have a lower intention to buy when exposed to a 2SM rather than to a 1SM.

The need of HSC individuals to search for information decreases when they are exposed to a 2SM. In such a situation, the individual receiving positive and negative information about the product will perceive it as enough. Negative information in the message will make consumers stop thinking about other negative characteristics that the product may have (Rucker et al., 2008), because HSC consumers ameliorate the need to rely on trivial attributes when deciding on a product (Xiao, 2016). This means that HSC consumers will ‘partial-out’ the negative information related to secondary attributes of the offer, and will base their responses on the important characteristics described in a positive way. Therefore, for HSC individuals, 2SMs will increase purchase intentions more than 1SMs. This explanation leads to the first two hypotheses of this research:

H1a: When exposed to a 2SM, individuals with HSC will have a higher purchase intention than if exposed to a 1SM.

H1b: When exposed to a 2SM, individuals with LSC will have a lower purchase intention than if exposed to a 1SM.

From these hypotheses, it is reasonable to propose that HSC and LSC individuals have different evaluation strategies in their decision processes. This proposal is supported by evidence showing that psychological determinants may serve as the underpinnings of the style of decision-making (Lysonski & Durvasula, 2013). Thus, this research investigates the attributes evaluation strategy followed in processing 2SMs by individuals with HSC and LSC.

Decision Process and Attributes Evaluation Strategies: Hypothesis Development

Considering the proposed moderator effect of self-confidence, it would be interesting to explain the cognitive process underlying an individual’s decision process when exposed to 2SMs. The literature indicates that general self-confidence plays an important role in decision-making, and thus has a great influence on decision processing (Chuang et al., 2013). The process of decision-making may be either cognitive or affective (Epstein, 1994). In the cognitive process, consumers use information about the brand and product attributes to make their decisions; in the affective process, they make their decisions based on their feelings and emotions. When decision-making is cognitive, consumers can use either compensatory or non-compensatory strategies to evaluate attributes of the alternatives (Blackwell et al., 2001). Individuals evaluate the quality of each attribute of the offer, assigning a level of importance to each of them (Hoyer & Macinnis, 2008). According to the compensatory rule, individuals are forced to weigh and sum all of the possible outcomes to assign a unidimensional value to each alternative, thereby exposing the individuals’ preferences. On the other hand, non-compensatory rules forgo weighing and summing. Instead, these rules eliminate any alternative that has a low value in one dimension, even if that alternative rates highly in the other dimension. In the non-compensatory strategy, individuals use only negative information to evaluate products and brands, consequently discarding them immediately from their consideration set (Klein & Bither, 1987). The negative attributes eliminate the possibility of purchase. The consumer does not take into account the positive aspects presented in the message, thus requiring a lesser cognitive effort. In this regard, previous research has found that processing 2SMs requires more cognitive resources for systematic scrutiny than 1SMs (Kao, 2011).

The less self-confident the individual is, the greater the likelihood of taking into account secondary information described negatively in 2SMs (Carpenter et al., 1994). Therefore, when individuals with LSC are exposed to a 2SM, their purchase intention would be less than if they were exposed to a 1SM. In this situation, it is expected that the individuals would use the non-compensatory strategy to evaluate the offer. Because their decision will be based on the secondary attributes (described in a negative way), they will discard the primary ones (described in a positive way). On the other hand, HSC individuals tend to search for more information to make better decisions (Chaplin & John, 2007; Loibl et al., 2009). Thus, it is expected that these individuals would use the compensatory strategy to evaluate the offer, processing a large amount of information about the product and assigning differentiated importance to each attribute. It is also expected that greater importance will be given to the primary attributes, because these are related to the essential functionalities of the product. In addition, the need to search for information tends to decrease when a 2SM is presented to HSC individuals, because it inhibits consumers from thinking about other negative characteristics that the product may present (Rucker et al., 2008). HSC individuals do not decide based on secondary attributes: for this reason, it is expected that when they are exposed to a 2SM, they would adopt a compensatory strategy to evaluate the attributes of the offer. This leads to the following hypotheses:

H2a: Individuals with LSC make use of the non-compensatory strategy to evaluate alternatives.

H2b: Individuals with HSC make use of the compensatory strategy to evaluate alternatives.

Methods and Materials

The formulated hypotheses were tested in three experimental studies. The experimental method is the most appropriate, because the objective was to test a cause and effect relationship.

The participants in all studies were undergraduates from a university in Southeast Brazil. No individual appeared in more than one study, nor in both the pre-test and the main experiment. The convenience sampling procedure considered a minimum of 30 participants per group of treatment, predetermined by research constraints. However, a sensitivity power analysis was conducted using G*Power3 (Faul et al., 2007) for each of the statistical tests in the experiments. Power analysis uses the observed sample effect size as the basis of the population effect and tells if the experiment had sufficient subjects to detect with inferential statistics the actual effect found (O’Keefe, 2007).

Study 1

The objective of this study was to test H1a and H1b. The independent variables were type of message, manipulated by two ads of a product (1SM vs. 2SM) and participants’ level of self-confidence (measured): the dependent variable was purchase intention (measured).

Preliminary Procedures

A schoolbag was the product chosen for the experiment because it was familiar to all of the participants. To prepare the advertisements, the primary and secondary attributes of a schoolbag were identified by three focus groups (six participants each). The focus groups discussed the attributes and benefits of a schoolbag in the participants’ daily life as students. Comfort, resistance and internal space were the primary attributes; waterproofing and colour were the secondary attributes.

Stimuli Pre-test

Two ads, a 1SM and a 2SM, were produced to serve as stimuli. In both ads, the title was ‘Just arrived the new Multi backpack!’ The subtitle of the ads manipulated the side of the message presenting the secondary attributes positively in the 1SM, or negatively in the 2SM. For the 1SM, the subtitle was ‘Besides being totally waterproof and having more than 15 colour options, its benefits are unsurpassed’. For the two-sided, the subtitle was ‘It is not fully waterproof and has only two colour options, but its benefits are unsurpassed’. Finally, the text of the message in both ads described three primary attributes: ‘100% anatomical, health for your back and shoulders’, ‘Great internal space with three compartments and 5 pockets’ and ‘Ultra-tough material, supporting up to 20 kg and fitting everything you need’. A group of 60 students were exposed individually and randomly to one of the ads for 20 seconds to pre-test the stimuli. They were then asked whether the advertisement of the schoolbag presented either: just positive, just negative, or positive and negative attributes. Of the 30 respondents submitted to the one-sided ad, 22 (73%) signalled that it had shown just positive attributes of the product, whereas of the 30 participants exposed to the two-sided ad, 26 (87%) signalled that it had shown positive and negative attributes (chi-square (c²) = 26.962, p = 0.000). As expected, the conclusion was that the ads were recognised, respectively, as one- and two-sided. The stimuli were considered to have represented satisfactorily both types of message.

Procedure and Measurements

Upon arrival, the 62 participants (who were not participants in the stimuli pre-test) were seated at an individual desktop and asked to fill out the survey. The research was introduced as a marketing survey for a product coming soon to the market. The first part of the experiment assessed the participants’ self-confidence. Five items were used on a seven-point scale, anchored by 1 = totally disagree and 7 = totally agree (adapted from Bearden et al., 2001): ‘I often have doubts about the purchase decisions I make’, ‘I frequently agonize over what to buy’, ‘I often wonder if I’ve made the right purchase decision’, ‘I never seem to buy the right thing for me’ and ‘Too often the things I buy are not satisfying’ (α = 0.88). The participants were then assigned randomly to conditions: 31 were submitted to the one-sided ad and 31 were submitted to the two-sided ad. After 20 seconds exposure, the participants were asked: ‘What was the product in the advertisement?’ The three individuals who answered incorrectly were discarded from the sample. Nothing was asked about the product’s features mentioned in the ad, in order to avoid calling attention to the positive and negative attributes of the product. Finally, the dependent variable ‘purchase intention’ was assessed. Three items were used on a seven-point scale, anchored by 1 = very low and 7 = very high (adapted from Dodds et al., 1991): ‘The likelihood of my purchasing this schoolbag is’, ‘If I were going to buy this product category, I would consider buying the advertised schoolbag’, and ‘My willingness to buy the schoolbag is’ (α = 0.90).

Results and Discussion

Model 1 of the SPSS Process application was used. It was confirmed that the difference in purchase intention at the levels of HSC or LSC was significant between 1SMs and 2SMs (R² = 0.1368, F(3.00) = 3.0635, p = 0.0351. The results indicated that type of message (β = 3.69, SD = 1.3164, t(58) = 2.802, p = 0.0069) and level of self-confidence (β = 0.5249, SD = 0.2185, t(58) = 2.402, p = 0.0195) were both associated with purchase intention. The interaction between type of message and level of self-confidence was also significant (β = −0.9182, SD = 0.3049, t(58) = −3.0113, p = 0.0038), suggesting that the effect of type of message depended on the level of self-confidence.

To support the interaction effect found with inferential statistics, a power analysis was conducted using G*Power3 (Faul et al., 2007). The power was calculated using the following parameters: effect size d = 0.1585 (from correlation coefficient R² = 0.1368), alpha α = 0.0351, n = 62. The result showed that the power of the test for the interaction effect of type of message and level of self-confidence was 0.67. This means that the probability of having found a true significance is 67 per cent. This proportion could be considered acceptable because, in behavioural sciences, a remarkable value for power is around 0.80 (Keppel & Wickens, 2004).

To support the hypothesis, simple slopes for the association between type of message and purchase intention were tested for low (−1 SD below the mean), moderate (mean) and high (+1 SD above the mean) levels of self-confidence (Figure 1). The effect of type of message on purchase intention was significant when self-confidence was one SD above the mean (MSelfConf + 1SD = 5.59, t = −2.15; p < 0.05), supporting H1a. Type of message was also related significantly to purchase intention when self-confidence was one SD below the mean (MSelfConf − 1SD = 2.46, t = 2.12; p < 0.05), supporting H1b. However, when self-confidence was at the mean (MSelfConf = 4.029, p = 0.9812), the effect of type of message on purchase intention was not significant. The Johnson–Neyman technique showed that the relationship between type of message and level of self-confidence was significant when the level of self-confidence was <2.63 and >5.39. This means that when the individual’s self-confidence is <2.63, the purchase intention for a 1SM is greater than for a 2SM and when the individual’s self-confidence is >5.39, the purchase intention for a 2SM is greater than for a 1SM (Figure 1).

Figure 1.

Source: The authors.

These findings are consistent with the theoretical explanation used to formulate H1. HSC consumers will exclude the negative information related to secondary attributes of the offer, and base their responses on the important characteristics described in a positive way. Conversely, LSC consumers will be biased towards the importance of the secondary characteristics described negatively, and will base their responses on those characteristics.

Study 2

The objective of this study was to test H2a. The independent variable was self-confidence, manipulated by a procedure of three instructions (LSC condition vs. HSC condition) and the dependent variables purchase intention (yes vs. no) and decision attribute (primary vs. secondary).

Stimuli Pre-test

Self-confidence was manipulated to high and low levels, following Atherton et al. (2016). However, emotions were not manipulated. Emotion is a valenced affective reaction to perceptions of situations (Ortony et al., 1988). In contrast to emotion, self-confidence is an individual’s subjective self-evaluation (Atherton et al., 2016). In order to manipulate self-confidence, 34 participants read a text, in which they were asked to follow three instructions: to remember, to think and to write about a particular purchase situation. For the HSC condition (n = 17), the first instruction was

Try to remember a time in your life when you made your best ever purchase decision. Try to remember this situation. Try to remember how you felt after realizing that you had made the right choice. Try to feel the sensations you felt when you were sure that you had made the right purchase.

After 30 seconds, the second instruction was ‘Think again about the feelings you felt, such as: accomplished, happy, joyful, satisfied, intelligent, skilful, smart, admired, recognized, powerful, assertive, respected, qualified’. After 20 seconds, the third instruction was ‘Using five of those feelings, write a text explaining why you felt in that way’. For the low level of self-confidence condition (n = 17), the first instruction was

Try to remember a time in your life when you made your worst ever purchase decision. Try to remember this situation. Try to remember how you felt after realising that you had made the worst choice. Try to feel the sensations you felt when you were sure that you had made the wrong purchase.

After 30 seconds, the second instruction was ‘Think again about feelings you felt, such as: frustrated, unhappy, sad, dissatisfied, not intelligent, not smart, eager, deceived, naive, shamed, despised, fragile, disrespected, unqualified’. After 20 seconds, the third instruction was ‘Using five of those feelings, write a text explaining why you felt in that way. In both cases there was no limit on time or words. After the manipulation, each respondent’s level of self-confidence was assessed as in study 1.

The Cronbach’s alpha coefficient (α = 0.90) showed the reliability of the self-confidence measurement. A post-test manipulation check verified that participants in the HSC condition (M(17) = 5.71) rated their self-confidence as significantly higher than those in the LSC condition (M(17) = 2.4; t(32) = −14.13, p < 0.01), t (32 ) = −14.13; p < 0.01. These results indicate successful manipulation of the level of self-confidence.

Procedure and Measurements

The experiment was introduced as a marketing survey about consumer behaviour. Initially, 62 participants (who were not participants in the stimuli pre-test) were submitted randomly to one of the self-confidence conditions: 32 were submitted to the HSC and 30 to the LSC. Then, participants were exposed to the two-sided ad used in study 1. After 20 seconds, the participants’ attention was controlled by asking ‘What was the product in the advertisement?’ All respondents answered correctly. Once again, the ad was showed to the participants for 20 seconds. Finally, the dependent variables were included. All respondents had to provide a categorical response for purchase intention (yes or no), and they were then asked about the reason for that decision. The question was ‘Why would you (would you not) buy this schoolbag? Think about the attributes of the schoolbag and write down which one of them was the most important in making your decision. Write only one attribute!’

Results and Discussion

As expected for a 2SM, individuals exposed to the HSC condition were more likely to purchase the product than were individuals submitted to the LSC condition. Of the 32 participants exposed to the HSC condition, 16 (50%) indicated that they would buy the bag and 16 (50%) would not buy it. Of the 30 participants exposed to the LSC condition, only 4 (13%) indicated that they would by the bag and 26 (87%) would not buy it (c² = 9.526, p = 0.002). Thus, it was concluded that 2SMs increase purchase intention when an individual’s self-confidence level is high.

The power for the c²-test was calculated from the following parameters: proportion in the HSC condition group = 50 per cent, proportion in the LSC condition group = 13 per cent, alpha α = 0.002, HSC condition n = 32 and LSC condition n = 30. The result showed that the power of the c²-test for the effect of the level of self-confidence on purchase decision was 0.53. This means that the probability of having found a true significance is 53 per cent.

To confirm H2a, the decision strategy used by participants was inferred from the type of attribute (primary or secondary) that they declared as the most important to make their decision. Of the 32 participants exposed to the HSC condition, 18 (56%) chose a primary attribute and 14 (44%) chose a secondary attribute. Of the 30 participants exposed to the LSC condition, 26 (87%) chose a secondary attribute and 4 (13%) a primary attribute of the product (c² = 12.458, p = 0.000). These results confirm H2a, because individuals with LSC make their purchase decision, positive or negative, based on some secondary attribute. That behaviour characterises the non-compensatory strategy to evaluate alternatives during the decision process.

The power for the c²-test was calculated from the following parameters: proportion in the HSC condition group = 44 per cent, proportion in the LSC condition group = 87 per cent, alpha α = 0.001, HSC condition n = 32 and LSC condition n = 30. The result showed that the power of the c²-test for the effect of the level of self-confidence on the type of decision attribute was 0.61. This means that the probability of having found a true significance is 61 per cent.

Study 3

The objective of this study was to test H2b. The independent variable was self-confidence, manipulated as in study 2 (LSC condition vs. HSC condition) and the dependent variables were purchase intention (measured) and importance of the attributes (measured).

Procedure and Measurements

There were 65 participants 34 were submitted to the HSC condition and 31 to the LSC condition. The procedure followed that of study 2, except that participants had to rate their purchase intention on a seven-point scale anchored by 1 = very low and 7 = very high (α = 0.86) (as used in study 1). They were not asked to choose between to buy or not to buy, as in study 2. Finally, to assess the attributes evaluation strategy, participants were asked to rank the importance of the schoolbag’s attributes presented in the ad. Based on theoretical definitions, it was stated that, in a compensatory strategy to evaluate the attributes of the offer, greater importance will be given to the primary attributes; and in a non-compensatory strategy, greater importance will be given to the secondary attributes. The participants had to indicate the order of importance of each attribute in making their decision on a scale from 1 = least important to 5 = most important. The attributes to be ranked were ‘Not fully waterproof’, ‘Only two colour options’, ‘100% anatomic’, ‘Great internal space, 3 compartments and 5 pockets’ and ‘Ultra-resistant material, supporting up to 20 kg’. The first and second attributes were secondary attributes, and the others were primary attributes.

Results and Discussion

As expected for a 2SM, individuals exposed to the HSC condition showed a higher purchase intention than individuals submitted to the LSC condition. A students’ t-test for independent samples was conducted to compare the effect of self-confidence conditions (HSC and LSC) on purchase intention. There was a significant effect of self-confidence on purchase intention for the two conditions. The purchase intention for the HSC condition (M = 5.56, SD = 1.90) was significantly different than the LSC condition (M = 4.30, SD = 2.41, t(65) = −2.358, p = 0.021). These results indicate that HSC has an effect on purchase intention. Specifically, the results suggest that, when an individual has HSC, he/she is more positively persuaded by a 2SM than when he/she has LSC.

The power for the students’ t-test was calculated from the following parameters: effect size d = 0.58 (from the means and standard deviations in the groups), alpha α = 0.021, HSC condition n = 34 and LSC condition n = 31. The result showed that the power of the t-test for the effect of the level of self-confidence on purchase intention was 0.49. This means that the probability of having found a true significance is 49 per cent.

To test H2b, the means of the importance level for primary and secondary attributes were calculated. It was expected that the importance level for primary attributes would be higher than for secondary attributes. A students’ t-test for independent samples was conducted to compare the effect of self-confidence conditions (HSC and LSC) on the levels of importance for primary and secondary attributes. There was a significant difference in the evaluation of the importance of primary and secondary attributes by the participants submitted to the HSC and LSC conditions. The level of importance for the primary attributes in the HSC condition (M = 3.39, SD = 0.63) was significantly different from the level of importance for the primary attributes in the LSC condition (M = 2.13, SD = 0.24), (t(65) = −9.432, p = 0.000).

The power for the students’ t-test was calculated from the following parameters: effect size d = 2.64 (from the means and standard deviations in the groups), alpha α = 0.0001, HSC condition n = 34 and LSC condition n = 31. The result showed that the power of the test for the effect of the level of self-confidence on purchase intention was 1.00. This means that the probability of having found a true significance is 100 per cent.

Also, the level of importance for the secondary attributes in the HSC condition (M = 2.41, SD = 0.90) was significantly different from the level of importance for the secondary attributes in the LSC condition (M = 4.22, SD = 0.65), (t(65) = 8.292, p = 0.000).

The power for the students’ t-test was calculated from the following parameters: effect size d = 2.31 (from the means and standard deviations in the groups), α = 0.001, HSC condition n = 34 and LSC condition n = 31. The result showed that the power of the test for the effect of the level of self-confidence on purchase intention achieved 1.00. This means that the probability of having found a true significance is 100 per cent.

These results indicate that the level of self-confidence has an effect on the evaluation of the attributes during the purchase process. Specifically, the results show that when an individual has HSC, primary attributes are more important than secondary attributes. The importance of the primary attributes increases as the individual’s self-confidence increases. These results could be considered evidence that individuals with HSC make use of the compensatory strategy to evaluate alternatives when making a decision.

Findings

The first finding of this research was the confirmation of self-confidence as a moderator variable for the effect of type of message (1SMs and 2SMs) on purchase intention. The results of study 1 confirmed that, for HSC individuals, the effect of 2SMs on purchase intention is greater than the effect of 1SMs. Conversely, for LSC individuals, the effect of 2SMs on purchase intention is lower than the effect of 1SMs.

The second finding of this research was the confirmation of the proposed underlying mechanisms by which self-confidence moderates the effect of 2SMs. In study 2, the results confirmed that 2SMs are more effective for individuals with HSC (than for individuals with LSC) in generating positive purchase decisions. It was also found that purchase decision, positive or negative, was based on primary attributes for individuals with HSC, and based on secondary attributes for individuals with LSC. In study 3, it was confirmed that 2SMs are more effective for individuals with HSC (than for individuals with LSC) in increasing purchase intention. It was also found that HSC individuals ascribe higher importance to primary attributes of the product for their purchase intention, and that LSC individuals ascribe higher importance to secondary attributes for their purchase intention. Studies 2 and 3 showed that, when exposed to a 2SM, the difference in purchase intention between individuals with HSC and LSC occurs because they have different strategies to evaluate alternatives.

Discussion

There are two theoretical contributions in this research: explanations regarding the moderator effect of self-confidence for the effect of type of message on purchase intention and explanations for the underlying mechanisms, specifically the attributes evaluation strategy, by which self-confidence moderates the effect of 2SMs.

Concerning the first theoretical contribution, the confirmation of the moderator effect of self-confidence for the effect of type of message on purchase intention corresponds with the explanation given by Kao (2011) for the effect of 2SMs: when exposed to a 2SM in conditions of moderate and low time pressure, individuals having a higher need for cognition tend to express a more favourable purchase intention than individuals with a lower need for cognition. Less anxiety and a high need for cognition are characteristics of HSC individuals. Confirmation of the moderator effect also amplifies the explanation given by Pizzutti et al. (2016) that it is possible to increase intentions of buying through a 2SM by controlling the importance of the negative information provided. HSC individuals will perceive negative information as of little importance, increasing purchase intention. The results of the current research confirm that 2SMs increase the purchase intention of HSC individuals, because they feel confident to make a purchase decision. Conversely, LSC individuals exposed to a 2SM will have lower levels of purchase intention compared to HSC individuals, because they do not feel confident enough to make a purchase decision.

The results infer that self-confidence is a condition related to other factors that have already been considered moderators of the effect of 2SMs, that is, self-regulatory focus (Florack et al., 2009) and regulatory congruence (Cornelis et al., 2014). In both studies, 2SMs had a greater effect on individuals with a high promotion focus. This is because promotion-focused individuals want the maximisation of positive results, and so a 2SM would make them feel more confident than a 1SM, increasing their positive responses. In this research, the effect of 2SMs was higher for HSC individuals. In other words, independently of the individual’s self-regulatory focus or the regulatory congruence of the message, an individual’s response to a 2SM would depend more on how confident the individual feels to make a decision.

Concerning the second theoretical contribution of this research, study 2 showed that individuals with LSC make use of the non-compensatory strategy to evaluate the offer during their cognitive decision process. LSC individuals are not motivated to process the information in the ad, applying the non-compensatory strategy to evaluate the offer. They rely on secondary attributes, described negatively, to make their decisions. Additionally, they tend to act on impulse, which makes their evaluation process faster when compared with the process adopted by individuals with HSC. Individuals with LSC decide not to buy based on secondary attributes of products, described negatively in a two-sided ad. These individuals look for mechanisms that facilitate their decision-making process, and nothing is easier for them than when there is secondary information about the products described in a negative way, because they adopt the non-compensatory decision model. This explanation confirms that LSC consumers are more likely to interpret trivial attributes in a biased way (Xiao, 2016). They do not take into account the information described in a positive way. For this reason, among individuals with LSC, purchase intention is less for 2SMs than for 1SMs.

On the other hand, HSC individuals, having a high need for cognition, are motivated to process the information in the ad and to apply the compensatory strategy to evaluate the offer (Eisend, 2010). Thus, in view of the fact that motivation and ability are determinant factors in a persuasive situation (Petty & Cacioppo, 1981), processing of the information in a 2SM would increase purchase intention. In contrast 1SMs, containing positive information only, would not be of interest to HSC individuals. These individuals will adopt the compensatory strategy to evaluate the offer because they tend to be less anxious and do not act on impulse. When the available cognitive resources are sufficient, individuals tend to scrutinise systematically either 1SMs or 2SMs. HSC individuals will analyse all the information of the product and will decide whether or not to buy only when they are sure that they have enough information to do so. They will give greater importance to primary attributes, which are described positively in both 1SMs and 2SMs. The negative information in 2SMs is likely to suppress counter-arguments and strengthen supporting arguments (Bohner et al., 2003; Kamins & Assael, 1987). This process of carefully analysing the information makes individuals give differentiated importance to each of the attributes, and so they will certainly attach greater importance to the primary attributes, because those are related to the essential functionalities of a product.

Because 2SMs generate more credibility, causing individuals not to think about other negative characteristics that the product may have, the need to search for information decreases, resulting in HSC individuals having more purchase intention when receiving a 2SM. This would happen because, when HSC individuals are exposed to a 2SM, their need for further information decreases, perceiving that the information in a 2SM as enough. This is because HSC consumers ameliorate the need to rely on trivial attributes when making a decision about a product. This finding is congruent with those showing the effect of 2SMs on purchase intention when the individual is under low or moderate time pressure (e.g., Kao, 2011), a situation in which the individual would have HSC.

The contributions in this research move theoretical development forward in two areas of knowledge. First, regarding the existing literature on 2SMs, our findings clarify how the effect of 2SMs occurs when taking into account the personality characteristics of target individuals, complementing studies of Florack et al. (2009) and Cornelis et al. (2014). Second, on the study of self-confidence, the majority of the literature considers this individuals’ personality characteristic as an antecedent of behavioural constructs, such as information seeking (Locander & Hermann, 1979), anxiety and post-purchase dissonance (Keng & Liao, 2013). Differently, our research examines self-confidence as a factor moderating the effect of a specific type of communication frequent nowadays in online environments (Zhang et al., 2018). In this theoretical development venue, the knowledge produced in this research is oriented to increase the expected communication effects in a creative and undirected manner.

Managerial Implications

Consumers’ exposure to various sources of information has augmented their contact with 2SMs. In this scenario, the findings of this research stress the value of knowledge about consumers’ self-confidence for management. HSC consumers would have more positive responses in front of 2SMs than those with LSC. Because of that, companies can affect individuals’ self-confidence in a positive way, by using bi-directional communication (e.g., as in online environments) and individualising messages oriented by this objective. It is possible to personalise marketing messages, managing interactive media (Kaptein, 2018), therefore raising the consumers’ self-confidence. A communication strategy oriented to inform and increase consumers’ self-confidence would be essential to access the potential benefits of 2SMs from external sources. On the other hand, companies using 2SMs in their marketing communication strategies should provide an environment conducive to self-confidence (e.g., selecting the right context information or enhancing good mood).

Limitations and Future Research

Theoretically, this research’s limitations provide paths for forthcoming research. In view of the decision process theory, this investigation explains the occurrence of the moderator effect of self-confidence on the effect of 2SMs on purchase intention. Specifically, the level of self-confidence was studied before exposure to a 2SM. However, the level of self-confidence after that exposure was inferred as augmented. Forthcoming research should analyse self-confidence as a mediator factor for the studied effect. In other words, upcoming studies must confirm the increase in self-confidence and the reduced need for information after a 2SM. Additionally, this investigation considered only one dimension of the self-confidence construct (i.e., personal results), and did not consider the level of knowledge about the advertised product. In following studies, the analysis of other dimensions of self-confidence and the previous knowledge of the product category could bring better clarity to the effect of 2SMs.

Still, from a theoretical perspective, the manipulation of self-confidence shows that it is possible to influence how individuals behave as consumers. An explanation for why and how that influence occurs is needed. It was proposed that the level of self-confidence (which was measured in study 1 and manipulated in studies 2 and 3) of consumers would influence their need for information. Given the low need for information by HSC consumers, the effect of 2SMs seems to be more effective. Therefore, forthcoming research should aim to explain which related constructs are mediators of the effect of 2SMs, considering the level of self-confidence, and the functioning of that mediation. Moreover, testing other characteristics of individuals as moderators for the effect of 2SMs would be of value. Such characteristics could be the level of knowledge of the product, the degree of connection with the brand and emotional state (e.g., distraction, anxiety and nervousness). Finally, an interesting topic of research could be the effect of 2SMs on a company’s other products and services.

Some weaknesses in the methodology of this research must be remedied in new studies. This research focused a product category oriented from a utilitarian perspective. Forthcoming research should include a wider variety of goods and services, which could bring different explanations. The convenience sampling procedure is another methodological limitation of this study. Nevertheless, it is obligation of the scientists to be cautious about the generalisability of their research results beyond the lab (Winer, 1999). In this concern, it is known that the more similar survey participants are to real purchasers, the more the research results may be generalised from the experiment to real decision situations. Consequently, to enlarge the external validity, future studies should enlist participants having a profile equivalent to that of the actual purchasers. By considering this feature in the sample, marketing professionals will be more confident in applying research results to their real decisions. The sample size determined by convenience before the experiments could be also considered a methodological limitation. It is recommended that, a priori, the sample size of experiments be computed to increase the power of the tests (Faul et al., 2007). In this research, a post hoc power analysis computed for the confirmed effects on the dependent variable purchase intention achieved the powers of 0.67 in study 1, 0.53 in study 2 and 0.49 in study 3. The effects confirmed on the dependent variable type of attribute (Primary v Secondary) in study 2 achieved a power of 0.61. Those levels of power are below 0.80, and are considered remarkable for researchers in the behavioural sciences (Keppel & Wickens, 2004). To increase the power, augmenting the probability of correctly rejecting the null hypothesis in statistical tests, future research must determine, a priori, the sample size using G*Power3 (Faul et al., 2007). There are also inadequacies regarding the stimuli used in the studies (ads) that may have had low credibility, because advertising professionals did not create them. In order to increase the generalisability of the explanations stated here, future studies must replicate the discoveries of this investigation in online environments, by working with different categories of products and services, other forms of manipulation and participants with diverse profiles.

Footnotes

Declaration of Conflicting Interests

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

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

References

Adelmann

P. K.

(1987). Occupational complexity, control, and personal income: Their relation to psychological wellbeing in men and women. The Journal of Applied Psychology, 72(2), 529–537.

Atherton

, Antley

, Evans

, Cernis

, Lister

, Dunn

, Slater

, & Freeman

(2016). Self-confidence and paranoia: An experimental study using an immersive virtual reality social situational. Journal of Behavioural and Cognitive Psychotherapy, 44(1), 56–64.

Bearden

W. O.

, Hardesty

D. M.

, & Rose

R. L.

(2001). Consumer self-confidence: Refinements in conceptualization and measurement. Journal of Consumer Research, 28(1), 121–134.

Blackwell

R. D.

, Miniard

P. W.

, & Engel

J. F.

(2001). Consumer behavior. Dryden Press, Harcourt College Publishers.

Bohner

, Einwiller

, Erb

H.-P.

, & Siebler

(2003). When small means comfortable: Relations between product attributes in two-sided advertising. Journal of Consumer Psychology, 13(4), 454–463.

Cacioppo

, Petty

, & Kao

C. F.

(1984). The efficient assessment of NFC. Journal of Personality Assessment, 48, 306–307.

Carpenter

G. S.

, Glazer

, & Nakamoto

(1994). Meaningful brands from meaningless differentiation: The dependence on irrelevant attributes. Journal of Marketing Research, 31(3), 339–350.

Chang

(2005). Ad and brand evaluations in a competitive processing context-the effects of number of attributes and repetition strategies. Advances in Consumer Research, 32(1), 548–553.

Chaplin

L. N.

, & John

D. R.

(2007). Growing up in a material world: Age differences in materialism in children and adolescents. Journal of Consumer Research, 34(4), 480–493.

10.

Chen

(2016). Can two-sided messages increase the helpfulness of online reviews? Online Information Review, 40(3), 316–332.

11.

Chuang

, Cheng

, Chang

, & Chiang

(2013). The impact of self-confidence on the compromise effect. International Journal of Psychology, 48(4), 660–675.

12.

Cornelis

, Cauberghe

, & De Pelesmacker

(2014). Regulatory congruence effects in two-sided advertising: The mediating role of processing fluency and processing depth. European Journal of Marketing, 48(7/8), 1451–1465.

13.

Cornelis

, Cauberghe

, & De Pelsmacker

(2015). The credibility of refutation in two-sided anti-drug messages. Journal of Social Marketing, 5(3), 241–257.

14.

Crowley

A. E.

, & Hoyer

W. D.

(1994). An integrative framework for understanding two-sided persuasion. Journal of Consumer Research, 20(4), 561–574.

15.

Dodds

W. B.

, Monroe

K. B.

, & Grewal

(1991). Effects of price, brand and store informationon buyers’product evaluations. Journal of Marketing Research, 25(3), 307–319.

16.

Eisend

(2006). Two-sided advertising: A meta-analysis. International Journal of Research in Marketing, 23, 187–198.

17.

Eisend

(2007). Understanding two-sided persuasion: An empirical assessment of theoretical approaches. Psychology & Marketing, 24(7), 615–640.

18.

Eisend

(2008). Dual-processing and message sidedness: The effects of involvement in two-sided advertising. In Brown

& Gurhan-Canli

(Eds.), Marketing theory and applications. AMA Winter Educators’ Conference Proceedings (Vol. 19, pp. 168–169). American Marketing Association.

19.

Eisend

(2010). Explaining the joint effect of source credibility and negativity of information in two-sided messages. Psychology & Marketing, 27(11), 1032–1049.

20.

Epstein

(1994). Integration of the cognitive and the psychodynamic unconscious. American Psychologist, 49(8), 709–724.

21.

Faul

, Erdfelder

, Lang

A.-G.

, & Buchner

(2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191.

22.

Florack

, Ineichen

, & Bieri

(2009). The impact of regulatory focus on the effects of two-sided advertising. Social Cognition, 27(1), 37–56.

23.

Golden

L. L.

, & Alpert

M. I.

(1987). Comparative analyses of the relative effectiveness of one-sided and two-sided communications for contrasting products. Journal of Advertising, 16(1), 18–28.

24.

Hair

, & Bond

S. D.

(2018). Attribute dismissal and valence effects in preferential decision processing. Journal of Behavioral Decision Making, 31, 164–178.

25.

Hastak

, & Park

J.-W.

(1990). Mediators of message-sidedness effects on cognitive structure for involved and uninvolved audiences. In Goldberg

M. E.

, Gorn

, & Pollay

R. W.

(Eds.), NA—Advances in consumer research (Vol. 17, pp. 329–336). Association for Consumer Research.

26.

Hellén

, & Saaksjarvi

(2011). Is consumer self-confidence a stable phenomenon? The effect of mood on self-confidence dimensions. Journal of Customer Behaviour, 10(3), 223–243.

27.

Hornik

, Ofir

, & Rachamim

(2016). Quantitative evaluation of persuasive appeals using comparative meta-analysis. The Communication Review, 19(3), 192–222.

28.

Hovland

(1954). Effects of the mass media of communications. In Lindzey

(Ed.), Handbook of social psychology (vol. 2; pp. 1062–1103). Addison-Wesley.

29.

Hoyer

W. D.

, & Macinnis

D. J.

(2008). Consumer behavior (5th ed.). Cengage Learning.

30.

Hunt

J. H.

, Domzal

T. J.

, & Kernan

J. B.

(1982). Causal attributions and persuasion: The case of disconfirmed expectancies. In Mitchell

(Ed.), NA—Advances in consumer research (Vol. 09, pp. 287–292). Association for Consumer Research.

31.

Kamins

M. A.

, & Marks

L. J.

(1987). Advertising puffery: The impact of using two-sided claims on product attitude and purchase intentions. Journal of Advertising, 16, 6–15.

32.

Kamins

M. A.

, & Assael

(1987). Two-sided versus one-sided appeals: A cognitive perspective on argumentation, source derogation, and the effect of disconfirming trial on belief change. Journal of Marketing Research, 24(1), 29–39.

33.

Kao

D. T.

(2011). Message sidedness in advertising: the moderating roles of need for cognition and time pressure in persuasion. Scandinavian Journal of Psychology, 52(4), 329–340.

34.

Kao

D. T.

(2012). Exploring the effect of regulatory focus on ad attitudes: The moderating roles of message sidedness and argument quality. International Journal of Psychology, 47(2), 142–153.

35.

Kaptein

(2018). Customizing persuasive messages; The value of operative measures. Journal of Consumer Marketing, 35(2), 208–217.

36.

Karimi

, Papamichail

, & Holland

C. P.

(2015). The effect of prior knowledge and decision-making style on the online purchase decision-making process: A typology of consumer shopping behavior. Decision Support Systems, 77, 137–147.

37.

Keng

C. J.

, & Liao

T. H.

(2013). Self-confidence, anxiety, and post-purchase dissonance: A panel study. Journal of Applied Social Psychology, 43(8), 1636–1647.

38.

Keppel

, & Wickens

T. D.

(2004). Design and analysis: A researcher's handbook. Upper Saddle River, N.J: Pearson Prentice Hall.

39.

Klein

, & Bither

S. W.

(1987). An investigation of utility-directed cutoff selection. Journal of Consumer Research, 14(2), 240–256.

40.

Lacey

, Close

A. G.

, & Finney

R. Z.

(2010). The pivotal roles of product knowledge and corporate social responsibility in event sponsorship effectiveness. Journal of Business Research, 63(11), 1222–1228.

41.

Locander

W. B.

, & Hermann

P. W.

(1979). The effect of self-confidence and anxiety on information seeking in consumer risk reduction. Journal of Marketing Research, 16(2), 268–274.

42.

Loibl

, Cho

S. H.

, Diekmann

, & Batte

M. T.

(2009). Consumer self-confidence in searching for information. Journal of Consumer Affairs, 43(1), 26–55.

43.

Lutz

R. J.

, & Reilly

P. J.

(1974). An exploration of the effects of perceived social and performance risk in consumer information acquisition. In Ward

& Wright

(Eds.), NA—Advances in consumer research (Vol. 01, pp. 393–405). Association for Consumer Research.

44.

Lysonski

, & Durvasula

(2013). Consumer decision making styles in retailing: Evolution of mindsets and psychological impacts. Journal of Consumer Marketing, 30(1), 75–87.

45.

Michaelidou

, & Dibb

(2006). Product involvement: An application in clothing. Journal of Consumer Behavior, 5(5), 442–453.

46.

Moe

, & Trusov

(2011). The value of social dynamics in online product ratings forums. Journal of Marketing Research, 48(3), 444–456.

47.

O’Keefe

D. J.

(2007). Brief report: Post hoc power, observed power, a priori power, retrospective power, prospective power, achieved power: Sorting out appropriate uses of statistical power analyses. Communication Methods and Measures, 1(4), 291–299.

48.

Ortony

, Clore

G. L.

, & Collins

(1988). The cognitive structure of emotions. Cambridge University Press.

49.

Park

, & Cho

(2012). Social network online communities: Information sources for apparel shopping. Journal of Consumer Marketing, 29(6), 400–411.

50.

Park

C. H.

, & Kim

Y. G.

(2003). Identifying key factors affecting consumer purchase behavior in an online shopping context. International Journal of Retail & Distribution Management, 31(1), 16–29.

51.

Petty

R. E.

, & Cacioppo

J. T.

(1981). Attitudes and persuasion: Classic and contemporary approaches. Wm. C. Brown.

52.

Pizzutti

, Basso

, & Albornoz

(2016). The effect of the discounted attribute importance in two-sided messages. European Journal of Marketing, 50(9/10), 1703–1725.

53.

Rucker

D. D.

, Petty

R. E.

, & Briñol

(2008). What’s in a frame anyway? A meta-cognitive analysis of the impact of one versus two sided message framing on attitude certainty. Journal of Consumer Psychology, 18(2), 137–149.

54.

Schlenker

B. R.

, & Leary

M. R.

(1982). Social anxiety and self-presentation: A conceptualization and model. Psychological Bulletin, 92(3), 64–69.

55.

Seiders

, Voss

G. B.

, Grewal

, & Godfrey

A. L.

(2005). Do satisfied customers buy more? Examining moderating influences in a retailing context. Journal of Marketing, 69(4), 26–43.

56.

Smith

D. N.

, & Sivakumar

(2004). Flow and internet shopping behavior: A conceptual model and research propositions. Journal of Business Research, 57(10), 1199–1208.

57.

Trifts

, & Häubl

(2003). Information availability and consumer preference: Can online retailers benefit from providing access to competitor price information? Journal of Consumer Psychology, 13(1–2), 149–159.

58.

Winer

R. S.

(1999). Experimentation in the 21st century: The importance of external validity. Journal of the Academy of Marketing Science, 27(3), 349–358. https://doi.org/10.1177/0092070399273005

59.

Xiao

(2016). How non-consumption goals (elicited by competitive setting or social risk) and self-confidence influence the importance of trivial attributes in product evaluation. European Journal of Marketing, 50(9/10), 1746–1766.

60.

Zhang

, Wu

, & Mattila

A. S.

(2016). The role of information load and peripheral factors. Journal of Travel Research, 55(5), 299–311.

61.

Zhang

, Zhao

, & Gupta

(2018). The role of online product recommendations on customer decision making and loyalty in social shopping communities. International Journal of Information Management, 38(1), 150–166.