Trust in Feelings & Confirmation Bias: A relationship mediated by processing fluency

(1)

Trust in Feelings & Confirmation Bias:

A relationship mediated by processing fluency

University of Groningen Faculty of Economics and Business Master Thesis for MSc Marketing Management

Groningen, The Netherlands 16-06-2019 Rianne Lukkien S2768968 b.r.lukkien@student.rug.nl +31655544895 Westerhavenstraat 3-1 9718 AJ Groningen The Netherlands

(2)

ABSTRACT

In this research, we study how the concept of trust in feelings influences judgment. Specifically, it was studied how the degree to which a person trusts their feelings leads to biased judgment of incoming information; confirmation bias. Feelings are often used as information – they signal value about a subject; “what feels good, must be good”. Judgment is then based on feelings rather than on thorough rational evaluation. Additionally, perception of truth is found to be influenced by the “truth effect”; something that is seen frequently is thought to be familiar and familiar things are more frequently judged as true (Bacon, 1979). This familiarity-based fluency can lead to biased judgments of information. In this thesis, we propose that trust in feelings influences confirmation bias in such a way that high trust in feelings is associated with an increased confirmation bias compared to low trust in feelings. We expect that this is because the relationship is mediated by fluency; high trust in feelings increases receptiveness to fluency, which in turn increases confirmation bias. The results of a logistic regression show that trust in feelings increases the confirmation bias when trust in feelings is high. However, results of a comparison of means showed no significant difference in information distortion scores between the two conditions for high or low trust in feelings. Thus, we find partial evidence to support our first hypothesis: high trust in feelings leads to an increased confirmation bias. The results of a mediation analysis for the effect of fluency in the relationship between trust in feelings and confirmation bias showed no significant interaction effect. Thus, we find no evidence to support the hypothesis that fluency effects mediate the relationship between trust in feelings and confirmation bias. Additionally, testing the direct effects between trust in feelings and fluency and the direct effect of fluency on confirmation bias, we also find no significant results. Thus, we did not find evidence to argue for direct effects between these concepts. A possible explanation for the insignificant findings of this study is that the manipulation of trust in feelings, based on Avnet et al. (2012), was not successful.

(4)

INTRODUCTION

“Next to sound judgment, diamonds and pearls are the rarest things in the world.” – Jean de la Bruyere, 1713

In our day-to-day life, we are continuously confronted with new experiences, situations and information. All of which we filter for importance, relevance and truth in order to construct a comprehensive understanding of the world around us. By this, we form beliefs about what we consider as our reality. However, this judgment of truth is often biased by our pre-existing beliefs. People who hold strong beliefs are likely to examine new information in a biased manner; supporting evidence is more likely to be accepted whereas disconfirming evidence will sooner be rejected (Lord et al., 1979). This concept of distorted information processing is known as confirmation bias (Nickerson, 1998).

Confirmatory reasoning is found to occur in consumer decision-making; people preferred an inferior alternative over a more favourable alternative when the inferior alternative was more in line with their emerging preferences (Russo et al., 2006). The

(5)

However, using feelings in judgment can sometimes also be misleading –for instance, when the current feelings are misattributed in the affect evaluation. This concept, known as the mood congruence judgment effect (Mayer et al., 1992), shows that individuals let their mood influence their judgment; a positive overall mood leads to a more positive judgment where a more negative mood leads to a more negative judgment of a target. Thus, the current experienced feelings are not necessarily related to the subject of evaluation, yet the evaluation is influenced by these feelings.

In this research, we consider the potentially misleading effect feelings can have on the way we process incoming information. Specifically, we focus on individuals’ belief that their feelings generally point toward the “right” direction in judgments and decisions; the concept of trust in feelings (Avnet et al., 2012). Individuals differ in their beliefs that feelings should be relied upon in judgments and decision-making. This originates from two sources; the personal history of success or failure people have with trusting their feelings and secondly the individuals’ cultural and social environment (Avnet et al., 2012). Thus, people can either have high or low trust in their feelings based on these two sources.

We will investigate how individuals’ trust in their feelings influences their likelihood to engage in confirmatory reasoning. Specifically, we expect that high trust in feelings, compared to low trust in feelings, is associated with an increased level of confirmation bias. This is because thought processes are accompanied by metacognitive experiences such as the ease or difficulty that information comes to mind — the fluency of information processing. These experiences of processing ease are informative on their own and people rely on them when forming judgments (Schwarz, 2004). These fluency effects are found to influence truth judgments (Koch & Forgas, 2012; Scholl et al., 2014). Specifically, Schwarz (2004) states that fluency is related to perceived familiarity; familiar information is easier to process. In turn, when encountering new information, it is seen as familiar when it is easy to process. Thus, information that is easy to process might feel familiar and therefore feel right, resulting in it being accepted as truth, leading to a confirmation bias in judgments.

The way we process information is found to influence how receptive we are to these effects of fluency, in such a way that more intuitive processing is associated with increased receptiveness to fluency effects (Dijkstra et al., 2014). Thus, in this research, we suggest that individuals who score high on trust in feelings will be more susceptible to fluency effects, which in turn increases confirmatory reasoning.

(6)

feelings as it leads to preference consistency, we argue that this consistency can lead to biased final judgment and thus is not always positive. Additionally, this study will contribute to existing literature by testing for the mediating role of fluency in this relationship. Earlier, fluency was mainly assessed in relation to several forms of judgment and decision making (Schwarz, 2004); brand evaluation (Lee & Labroo, 2006), affective judgments (Reber et al., 1998), judgment of risk (Song & Schwarz, 2004). Yet, no framework for the relationship between trust in feelings, confirmation bias and the role of fluency was previously researched. Researching this relationship may provide insights for marketers and brand managers in how feelings of consumers influence information processing, evaluation and thus the choice making process. This knowledge can then be used to design and develop marketing strategies to optimize business performances. Therefore, in this study, an experimental design will be used to test how high versus low trust in feelings influences the level of confirmation bias and we test for the mediating effect of fluency in this process.

THEORETICAL FRAMEWORK

Confirmation bias

(7)

There are several factors that facilitate confirmation bias. The level of confidence people have in their attitudes shows to influence how they perceive new information; people who feel very confident about their attitudes were more receptive for attitude-challenging information, whereas people who had more doubts about their attitudes and thus were low in confidence about their attitudes preferred attitude-congruent information (Albarracin & Mitchell, 2004). Another factor that facilitates confirmation bias is mood; a good mood leads to increased pre-decisional distortion; in an experiment, participants were given a bag of candy to induce a good mood which led them to evaluate new information more positively than participants who had not been given candy (Meloy, 2000). Koch and Forgas (2011) also demonstrated that the way information is processed is influenced by mood, in such a way that positive mood makes people more receptive to information that is easy to process – fluent information. Additionally, reliance on feelings in judgment increases preference consistency and decreases cognitive noise (Lee et al., 2009). Thus, judgment of new information is influenced by many factors, of which some even lead to biased judgment as people tend to judge information so it fits their previous beliefs.

Trust in feelings

Research by Schwarz and Clore (1988) set the base for addressing feelings as a source for judgment as they showed that feelings were seen as a valuable source of information. However, the extent to which individuals rely on their feelings instead of relying on reason based judgment is dependent on several factors. So is found that an independent self-construal leads to an increased reliance on feelings (Hong & Chang, 2015), social exclusion also increases reliance on affect in information processing (Lu & Sinha, 2017) and even when consumers try to suppress feelings in product choices, this suppression of feelings leads to an increase in reliance of feelings in judgment, a paradoxical effect (Qiu et al., 2009). Also, people tend to rely more on their feelings in extraordinary situations that are uncommon and unfamiliar to them, than in more common, ordinary situations (Xuhong, 2016) and under circumstances of limited processing resources available, people tend to base their decision-making more on affect rather than on cognition (Shiv & Fedorikhin, 1999).

(8)

trust in feelings originates from a person's history of success with reliance on feelings in judgment and whether trust in feelings in judgment is supported or rejected by the person’s cultural and social environment. High trust in feelings showed to lead to an increased reliance on feelings in judgment (Avnet et al., 2012).

Trusting feelings in judgments affects the information processing and judgment system in multiple ways. People who highly trusted their feelings were more likely to choose a more affective, less statistically favourable option over a more statistically, less affective favourable option (Avnet et al., 2012). However, research by Pham et al. (2012) showed that individuals who have high trust in their feelings were more successful in predicting future events; a concept known as the oracle effect. The authors propose that this is due to the fact that people have gathered information over time regarding the subject and have stored this unconsciously in the back of their mind. Addressing feelings then provides a “privileged window” into tacit knowledge stored by the individual (Pham et al., 2012). According to Pham et al. (2001) the use of feelings as a base for judgment shows to be faster, more consistent stable and consistent across individuals in comparison to reason based judgment. These affective feelings are used for a wide range of inferences and judgments as affective feelings can signal value to the individual; what feels good must be good (Pham, 2004), a concept known as fluency.

Trust in Feelings & Fluency

Research by Aydin (2018) showed that thinking style in judgment affects one’s receptiveness to fluency effects; intuitive judgment showed to induce a more fluent evaluation process compared to rule-base judgment. Processing fluency is the subjective experience of ease with which people process information (Alter & Oppenheimer, 2009). The concept of fluency generally can be divided in perceptual fluency and conceptual fluency (Reber et al., 2004). Perceptual fluency is the ease of identifying physical identity of the stimulus, which is influenced by perceptual priming, clarification, presentation duration, repetition or figure-ground contrast (Reber et al., 2004). Conceptual fluency is the ease with which the target comes to consumer’s mind and pertains to the processing of meanings (Lee & Labroo, 2004). In this research we use the concept of processing as a whole as both might be at play here. From now on, we will refer to processing fluency simply as “fluency” in this paper.

(9)

their gut feelings, rather than using the analytical format, which requires more elaboration. This is in line with Dijkstra et al. (2014), who found that the processing style of information influences people's reliance on fluency; more global, intuitive processing showed an increased reliance on fluency in contrast to local processing (more analytical thinking).

However, Tsai and Thomas (2011) found that abstract thinking, in contrast to concrete thinking, reduces the effects of fluency. Interestingly, they found that this was only the case when feelings were considered as unimportant. When feelings were considered important, fluency increased liking in the abstract thinking condition (Tsai & Thomas, 2011). Additionally, Sundar et al. (2015) showed that when people were encouraged to trust their feelings, they were more receptive to fluency effects. For the present study, these finding imply that that participants who score high on trust in feelings (compared to low) and thus, are more likely to rely on their feelings during judgment, should be more susceptible to fluency effects. Consequently, we propose that people are more likely to interpret new information in a biased way to affirm pre-existing beliefs and preferences. This is because information that confirms pre-existing beliefs appears more fluent due to perceived familiarity (Schwarz, 2004).

In comparing disfluency and fluency, it was found that disfluently processed stimuli are perceived as riskier than fluently processed stimuli, given that they are more unfamiliar and thus more risky (Song & Schwarz, 2009). Fluent processed information however is found to result in positive affect in the study by Winkielman and Cacioppo (2001) where fluency effects on smiling and frowning muscles were tested. Fluent processing was associated with higher activity in the facial muscles that cause smiling, indicating that processing fluency shows to increase liking. Additionally, processing fluency is found to influence people’s judgment; highly fluent processed subjects are preferred over disfluent objects (Reber et al., 2004) and high levels of fluency lead to an increase in affect, which in turn leads to more positive brand evaluations (Lee and Labroo, 2004).

(10)

Fluency & Confirmation Bias

According to Schwarz (2004) fluency-based familiarity has important implications for judgment of truth; when little other information is available to determine truth, people rely on cues and associate fluency with truth and disfluency with something not being true. A related concept is the so called “truth effect”. The illusion of the truth effect is when repetition of a statement makes that the statement is perceived as more true (Bacon, 1979); repetition of statement leads to greater experienced processing fluency of the information (Unkelbach, 2007) and this fluency induces greater feelings of familiarity (Schwarz, 2004). This is confirmed by (Alter & Opperheimer, 2009) who state that fluency is a mechanism that influences truth judgments in such a way that fluently processed information is perceived truer that less fluent information. Thus, fluency is found to have several effects on judgment; it induces feelings of familiarity, increases liking and preferences for fluently processed information and by that influences judgments of truth.

Research by Bacon (1979) showed that changed information (differing from original) was especially likely to be rated as false. He argues that this is because truth-preserving information is found to be more credible and truth-altering information is perceived less credible. Especially, familiar and easy to retrieve tendencies may make congenial information (information that is in line with previous beliefs) easier to process than ungenial confirmation and then also more attractive (Winkielman & Cacioppo, 2001). According to Silva et al. (2016) the process of accepting a fact as true seems to be cognitively fluent and therefore effortless, where categorization of information as false is more demanding. Given that changing current beliefs takes time and effort, people often prefer to not regard alternative perspectives that question their existing beliefs as this is easier than adapting your beliefs (Hernandez & Lee Preston, 2013). Also, Hernandez and Lee Preston (2013) found that disfluency of processing disrupts the confirmation bias because this disfluency prompted a more analytical and critical mind-set, which made people more receptive to uncongenial information. Thus, fluency shows to affect judgment in several ways and, specifically, it influences truth judgments and then in turn contributes to a confirmation bias in judgment. Current research

(11)

that participants were more susceptible to the fluency effect and the truth effect when they were encouraged to trust their feelings (Sundar et al., 2015). Thus, we propose that high trust in feelings (compared with low trust in feelings) is associated with increased receptiveness to fluency.

Perceived fluency has several effects on judgment; leading to increased liking and preference (Schwarz, 2004; Winkielmann & Cacioppo, 2001; Reber et al., 2004) and leading to more positive brand evaluations (Lee & Labroo, 2004). Also, fluency affects judgment of truth in such a way that high processing fluency leads to an increased perception of truth of information (Alter & Opperheimer, 2009). Given that people prefer to accept information that is line with their existing beliefs (Hart et al., 2009), highly fluent information, which is more likely to be perceived as true, might enhance people’s confirmation bias. Thus, we propose that individuals high on trust in feelings display an increased confirmation bias compared to individuals low on trust in feelings. We argue that this is because individuals high on trust in feelings are more susceptible to fluency effects and therefore, are more likely to interpret information as true that is in line with their emerging/pre-existing preferences. Thus, we propose the following hypotheses:

H1: Trust in feelings affects the confirmation bias in such a way that individuals high on trust in feelings show an increased confirmation bias compared to individuals low on trust in feelings.

H2: The relationship between trust in feelings and confirmation is mediated by fluency effects.

(12)

METHODOLOGY

In the previous section, the existing literature on the concepts was discussed as well as the proposed relationship between the concepts. In the following section, we describe the experimental design of the study, the procedure for data collection and measures of the variables will be explained, followed by the description of the participants. Finally, we describe how the collected data will be analysed.

Study Design

This study uses an experimental design to test the mediating effect of fluency in the relationship between trust in feelings (TIF) and the confirmation bias. In this experiment, we manipulate trust in feelings (IV) based on Avnet et al. (2012). Participants are randomly assigned to either the high trust in feelings or the low trust in feelings condition. We then test how high (vs. low) trust in feelings influences confirmation bias (DV). Then, we test for the mediating effect of fluency in the relationship between trust in feelings and confirmation bias. We thus use a 2 (trust in feelings: high vs. low) between-subjects design with trust in feelings manipulated. To make sure that no biases arise from differences in participants due to selective assigning, participants are randomly assigned to the conditions.

Measures

In the following section we describe the measures used for the independent and dependent variables. Also, we describe how trust in feelings is manipulated.

Demographics. To gain insight in the nature of our participant sample, the participants will be questioned on their demographic information. This will include gender, age and nationality. Also, participants will be asked to rate their English language proficiency to foresee any problems with the questionnaire due to misunderstanding of the content.

(13)

situations, but more difficult to retrieve 10 situations, two different conditions will be created. The participants who get the task to only retrieve two situations find this relatively easy to do and thus experience successfully trusting their feelings as more common, whereas the participants who are asked to come up with 10 situations find this more difficult and thus assume that these situations, of successfully reliance on feelings, are less common (Avnet et al., 2012). This manipulation then leads to the creation of two conditions amongst the participants; high trust in feelings (HTF) and low trust in feelings (LTF).

Confirmation Bias. For measuring the dependent variable, confirmation bias, we use measurements composed by Russo et al. (1998) and Chaxel and Han (2018). Predecisional distortion of product information was measured by Russo et al. (1998) by initially providing a task in which people had to choose between two products. These two products were presented with several attributes, assuring that both products could be rated equally; one product would not be preferred over the other product. Then, secondly, additional information would be provided, aside from attribute information, that would enhance the preference for one of the two products. Then, as a later task, people were provided with information statements about the brand in which they have to indicate which brand was favored, based on this information. Consequently, participants were asked which of the brands was the leading brand in their choice. By this, it was measured to what extent people engaged in confirmation bias; the biased evaluation of incoming information to favor emerging beliefs (Russo et al., 1998).

(14)

right choice. By using this method, information distortion can be measured because the participants have no prior brand or product knowledge and thus no preference when starting the choice process. Also, the information on the attributes was designed to neither favour one of the two options; the equivocality of information (Russo et al., 1998). Therefore, with this method, we measure to what extent people evaluate information in favour of their previous choices and how this influences their final choice for a restaurant, thus to what extent they engage in the confirmation bias.

Fluency. To measure the perceived fluency of information we use the measures as tested by Lee and Aacker (2004). They measured processing fluency by asking participants to rate information on ease of processing using a seven-point scale (1 = difficult to process, 7 = easy to process) and to rate the comprehensibility of the information (1 = difficult to understand, 7 = easy to understand). These items were averaged to form a fluency index (r = .88). In our study, we implement the fluency measure in the study for confirmation bias; participants will be asked to rate the information they are given according to the measurements as discussed. Thus, the participants will be presented with the information on the attributes of Restaurant K and Restaurant J and then, after each information statement about an attribute, will be asked to rate the ease of processing and comprehensibility of that informational statement. Thus, fluency is measured multiple times as participants are asked to rate each information statement they receive. By this, we measure the perceived processing fluency of the participants for the information they are given.

Manipulation check. To be able to see if the induced manipulation of trust in feelings worked as we intended, we included a manipulation check in the survey. Following Avnet et al. (2012) the participants were asked to what extent they chose the restaurant based on their gut feeling, and to what extent did they base their choice on how the restaurant felt. Both these questions were measured using a 7-point Likert scale (1 = not at all, 7 = a lot). These questions were asked at the end of the survey to measure how participants perceive they have trusted their feeling.

Survey & data collection

(15)

presented the disclosure form. In this disclosure form, participants were informed that their participation was completely voluntary, there are no risks involved in taking part in the study, no costs would be associated with participation and that the participant was free to leave the survey at any time. Importantly, by ensuring that data would be handled anonymously, we ensured total confidentiality. As an incentive for participation, participants could enter their name and e-mail address in order to participate in winning one of the two Bol.com gift cards of a value of 20 euros. As this might endanger their anonymity, it was ensured that this data would not be linked to their responses in the survey. Lastly, contact details of the researcher were provided in case of questions, concerns or complaints about the study. In case of consent with the participation, the participant then could continue to the instructions. Here, brief instructions for the survey were given which included an overview of the studies, the note that attention checks were included, the request to not talk about the content of the study with other parties and a reminder for the price winning.

As the participant then continues to the first part of the study, the software randomly assigns the participant to one of the two conditions for trust in feelings. After finishing this part, all participants would be provided with the same questions for the remaining part of the survey; the confirmation bias measurement and the demographics.

At the end of the second part of the survey (measure of confirmation bias) an attention check was included to see if participants were still paying attention, in order to be able to filter out people who do not put effort in doing the survey and therefore could possibly decrease validity of the study (Oppenheimer et al., 2009) This was followed by the manipulation check as used by Avnet et al. (2012). Next, the respondents were asked for their demographic information and personal details in case they would like to participate for the prices. This was optional. Lastly, the participants were thanked for their time and participation in this study. In total, the survey was available for 30 days, from 01.04.2019 to 30.04.2019.

Plan of analysis

(16)

To see if the measurements scales are valid to use for analysis, a reliability analysis will be conducted for the measures of fluency and for the measures of the manipulation check.

To start, we check whether the manipulation in this research has been successful. Then, we test for the main effect of our research design; the effect of trust in feelings on confirmation bias. This is tested by using an independent samples t-test. Subsequently, we test for the mediation role of fluency by conducting a mediation analysis in which we test for the interaction effect. We use Hayes’ (2013) PROCESS Marco (model 4) to do this. The results of the analyses then will be interpreted and discussed in the conclusion and discussion section of this thesis.

RESULTS

Preliminary data analysis

Participants

First, we deleted incomplete responses from the dataset. In total 145 participants completed the survey. Seven participants failed the attention check and were thus excluded, leading to a final sample size of 138 respondents for analysis. Amongst these respondents there were 97 females, 41 males (Mage = 24.5, SD = 5.98). Of the participants, 65.2% is of Dutch nationality, 7.2% German, 7% is of Belgian nationality and 26.8% of other nationalities.

Data preparation

(17)

scale. In this case, with the 9-point scale, choosing a restaurant would result in absolute value compared to the neutral point (5). For example, fully favouring restaurant K (left of the scale) and thus selecting 1 on the 9-point scale, would lead to an absolute value of -4 (1-5). For restaurant J the other way around. With this, both the favour of and choice of the participants towards the leading and trailing restaurant was measured. As participants do not have a leading option when starting the survey, we calculate absolute values from the second attribute onwards. For each participant, an information distortion score was then calculated. Note that the level of information distortion is a form of confirmation bias (Chaxel & Han, 2018). Participants’ information distortion score was used as dependent variable in our analyses.

Manipulation check

With an independent samples t-test, we tested whether the manipulation of trust in feelings has indeed created two conditions; one group high on trust in feelings, one group low on trust in feelings. Following the measures of Avnet et al. (2012), respondents are asked to what extent they relied on their gut feeling in answering the questions and to what extent they rated the information on how good it felt (α = .73). The results of the independent samples t-test (see Appendix 2.1) show a significant difference between the two groups, with the mean score on the manipulation for low trust in feelings higher (MTIFlow = 5.34, SD = 0.97) than for the high trust in feelings group (MTIFhigh = 5.28, SD = 1.23), with a 95% confidence interval (t(136) = .36, p = .04, CI [-.31, .44]). This would indicate that the low trust in feelings group relied more on their feelings than the high trust in feelings group, concluding that our manipulation of the variable trust in feelings did not work. Despite these results, it is still interesting to test for the effect of trust in feelings on fluency and the relation to confirmation bias.

Main analysis

Trust in feelings and confirmation bias

(18)

for a restaurant and the manipulated variable for trust in feelings. The logistic regression was done using PROCESS Macro by Andrew F. Hayes (Model 1) with a confidence interval of 95% and 10.000 bootstrapping samples. The results show that there is a significant interaction effect (b = 1.93, χ2 (1,1) = 2.43, p < .05). Looking at the conditional effects, we find no significant effect for condition 0 (low trust in feelings) at b = .70, χ2 (1,1) = 1.38 and p = .17. For condition 1, high trust in feelings, we do find a significant effect (b = 2.63, χ2 (1,1) = 4.33, p < .05). The total analysis can be found in Appendix 2.3. As we do find a significant interaction, this provides partial support for our first hypothesis; confirmation bias increases under high trust in feelings conditions. We speak of partial support because we find mixed evidence; we find support with the logistic regression, however, we do not find support based on the independent samples t-test looking at the difference in information distortion scores.

Trust in feelings and fluency

(19)

difference in fluency per condition. The full output of the analysis can be found in Appendix 2.6.

Fluency and confirmation bias

We tested whether there was a correlation between fluency measured and the confirmation bias the participants showed. For this, we use a linear regression analysis with the average measure of fluency and the calculated information distortion scores to represent confirmation bias. The results (see Appendix 2.7) show that there is no significant correlation between the fluency measured and the confirmation bias (F(1,132) =.00, p = .10), with an R2

of .00.

Mediation analysis

(20)

analysis using PROCESS Macro (model 4) with a 95% confidence interval and 10.000 bootstrapping samples. The results (Appendix 2.9) show that there is no significant indirect effect (β = -.10, SE = .08, 95% CI [-.29, .03]). Looking at the direct effect between trust in feelings and the difference measure of fluency, we find no significant effect (path a)(β = -.28, SE = .20, 95% CI [-.67, .11]). Then, we assess the effect of fluency on confirmation bias and find a significant effect (path b)(β = .35, SE = .12, 95% CI [.11, .59]). Looking at the direct effect of trust in feelings on confirmation bias, we find no significance (path c)(β = -.10, SE = .28, 95% CI [-.65, .45]). Thus, also for the difference measure of fluency, we do not find a significant mediation effect for the relationship of trust in feelings and confirmation bias.

Alternative explanations

Because we find little significant effects, we conducted an extra analysis to look for an alternative explanation. As it is often claimed in Western society that women are more emotional and men are more rational (Fischer, 1993), we ran the mediation analysis a second time and included gender as a covariate, as gender then might explain a difference in trust in feelings (Appendix 2.10). We ran the mediation analysis by using PROCESS by Andrew F. Hayes (2013) (Model 4) with a 95% confidence interval and 10.000 bootstrapping samples. The results show no significance here as well (β = .24, SE = .31, 95% CI [-.38, .85].

DISCUSSION

Conclusions

(21)

place. Therefore, to discover further possible insights, we continued the analysis. The results showed that there was no significant difference in the confirmation bias between the two conditions of trust in feelings when looking at the information distortion scores. However, the results of the logistic regression showed a significant interaction effect for trust in feelings on the effect of the first choice of the restaurant on the final choice for a restaurant. Specifically, we found that this was only significant for the condition of high trust in feelings. This indicates that the first choice for a restaurant influenced people’s final choice if they had high trust in feelings. Considering both these analysis for the effect of trust in feelings on confirmation bias, we conclude that we find partial support for our hypothesis that states that trust in feelings influences confirmation bias in such a way that people high on trust in feelings (compared to people low on trust in feelings) show an increased confirmation bias.

In assessing the effect of trust in feelings on fluency, we found no significant difference between the two conditions of trust in feelings, which contradicts our expectations but is not surprising since the manipulation was not successful. Then, in assessing the effect of fluency on confirmation bias, we found no significant correlation between the two variables; indicating that increased fluency in this study is not associated with increased confirmation bias. What we did find is that the perception of fluency changed over time; the first time fluency was measured, it was significantly lower that the last time fluency was measured. In addition, we tested if this change in fluency differed per condition of trust in feelings. However, results showed that there was no significant difference for the two conditions of trust in feelings when it comes to change in fluency perception during the survey. Then, lastly, we conducted a mediation analysis to test for the indirect effect of trust in feelings on confirmation bias through fluency. The results showed that there was no significant direct effect between trust in feelings and confirmation bias and also the indirect effect, trough fluency, was not significant. From this, we can conclude that we did not find evidence to support our second hypothesis that fluency effects mediate the relationship between trust in feelings and confirmation bias. As alternative explanation, gender was included as covariate, however, this was not significant and can we conclude that gender does not serve as covariate in this relationship.

(22)

effects of this study not being significant, this study contributes to the existing literature with an overview of these possibly related concepts and might serve as base for further testing in future research.

Limitations

The main limitation of this study is that the manipulation did not work as was intended. We were unable to successfully create two conditions; one high on trust in feelings, one low on trust in feelings. This makes that is was not possible to assess the difference between the two groups and their effect on the fluency perception and on the confirmation bias. A possible cause for the manipulation being unsuccessful can be the occurrence of the reactance effect. Reactance is behaviour that is the opposite of the behaviour or attitude that is being asked or being pressured to adopt (Stieger et al., 2007). This even can have a boomerang effect; the behaviour respondents show is the opposite behaviour as expected from literature. Especially, Stieger et al. (2007) found that forced-response design invoked a reactance effect. In our case, as we also saw an opposite measure of trust in feelings compared to what we expected, this might be because of a reactance effect. Another limitation is that we did not consider cultural differences. As the respondents origin from different countries, different cultures are also at play as people rely on cultural systems to help them interpret meaning of encountered events or objects (Hanges et al., 2000). Therefore, there might be differences in the degree to which people trust and rely on their feelings per culture, possibly biasing this study. A third limitation is that despite the intended manipulation, individual differences should also be taken into account. Some individuals might rely more on their feelings than other individuals, as part of their personality. This can then bias the manipulation as for example; individuals that are naturally high on trust in feelings now are assigned to the low trust in feelings condition. Even though the software will assign conditions to the participants completely random, there might be a pre-existing difference in degree to which people rely on their feelings within and between the conditions. Implications

(23)

literature by assessing difficulties with this manipulation method and can serve as a base for exploring room for improvement, so that, in future research, the manipulation of trust in feelings can be more successful. Additionally, the results of the analyses for the mediation effect were not significant, implying that there is no mediation by fluency for the relationship between trust in feelings and confirmation bias. This contributes to the existing literature in such a way that it contradicts expectations and therefore needs further research to explore whether and how these concepts are related. Even though we did not find evidence to support our hypothesis for the proposed mediation, we did find that for people who where high on trust in feelings, their initial choice more influences their final choice, compared to people low on trust in feelings. As for the field or marketing, this provides an interesting and even promising insight. Knowing that people high on trust in feelings are more influenced by their initial choice when it comes to the final decision-making, enables marketers to optimize their strategies. It might be interesting to specifically target consumers who highly trust their feelings or manipulate this condition in order to influence their evaluation of a product, service or a brand. For example, for people high on trust in feeling, the initial contact with a brand can then be of great influence on how information is processed during the further relationship with the brand and thus also influence the final evaluation, eventually determining customer loyalty to the brand. However, this study only provides as slight insight in the relationship between these concepts. For marketers to be able to effectively assess trust in feelings in their marketing strategy, there first needs to be more research on these relationships.

Future research

(24)

Specifically, different forms of fluency can be tested as this study used existing stimuli to measure information distortion, into which the fluency measures where incorporated; participants were asked for their perceived fluency. Future research can explore the relationship between trust in feelings and confirmation bias with different fluency stimuli and test other forms of fluency; think of figure-ground contrast, conceptual fluency or compositional fluency. Concluding, as this mediated relationship between trust in feelings and confirmation bias was not studied before, this current study provides a starting point which can inspire researchers to further study this relationship and other possibly related factors.

REFERENCES

Albarracín, D., & Mitchell, A. L. (2004). The role of defensive confidence in preference for proattitudinal information: How believing that one is strong can sometimes be a defensive weakness. Personality and Social Psychology Bulletin, 30(12), 1565-1584.

Alter, A. L., & Oppenheimer, D. M. (2009). Uniting the tribes of fluency to form a metacognitive nation. Personality and social psychology review, 13(3), 219-235.

Avnet, T., Pham, M. T., & Stephen, A. T. (2012). Consumers’ trust in feelings as information. Journal of Consumer Research, 39(4), 720-735.

Aydin, A. E. (2018). Processing fluency: examining its relationship with thinking style and purchase intention. Journal of Marketing Communications, 24(6), 588-598.

Bacon, F. T. (1979). Credibility of repeated statements: Memory for trivia. Journal of Experimental Psychology: Human Learning and Memory, 5(3), 241.

Carlson, K. A., & Russo, J. E. (2001). Biased interpretation of evidence by mock jurors. Journal of Experimental Psychology: Applied, 7(2), 91.

Chaxel, A. S., & Han, Y. (2018). Benefiting From Disagreement: Counterarguing Reduces Prechoice Bias in Information Evaluation. Journal of Consumer Psychology, 28(1), 115-129. de la Bruyère, J. (1713) A defence of M. de La Bruyere. Volume 1 of The Works of Monsieur de La Bruyere. California, E. Curll and J. Pemberton.

(25)

Festinger, L. (1962). Cognitive dissonance. Scientific American, 207(4), 93-106.

Fischer, A. H. (1993). Sex differences in emotionality: Fact or stereotype?. Feminism &

Psychology, 3(3), 303-318.

Hanges, P., Lord, R., & Dickson, M. (2000). An Information‐processing perspective on leadership and culture: A case for connectionist architecture. Applied psychology, 49(1), 133-161.

Hart, W., Albarracín, D., Eagly, A. H., Brechan, I., Lindberg, M. J., & Merrill, L. (2009). Feeling validated versus being correct: a meta-analysis of selective exposure to information. Psychological bulletin, 135(4), 555.

Hauser, D. J., Ellsworth, P. C., & Gonzalez, R. (2018). Are manipulation checks necessary?. Frontiers in psychology, 9.

Hayes, A. F. (2013). Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach, New York: Guilford

Hernandez, I., & Preston, J. L. (2013). Disfluency disrupts the confirmation bias. Journal of Experimental Social Psychology, 49(1), 178-182.

Hong, J., & Chang, H. H. (2015). “I” follow my heart and “We” rely on reasons: The impact of self-construal on reliance on feelings versus reasons in decision making. Journal of Consumer Research, 41(6), 1392-1411.

Koch, A. S., & Forgas, J. P. (2012). Feeling good and feeling truth: The interactive effects of mood and processing fluency on truth judgments. Journal of Experimental Social Psychology, 48(2), 481-485.

Kostopoulou, O., Russo, J. E., Keenan, G., Delaney, B. C., & Douiri, A. (2012). Information distortion in physicians’ diagnostic judgments. Medical Decision Making, 32(6), 831-839. Lee, A. Y., & Aaker, J. L. (2004). Bringing the frame into focus: the influence of regulatory fit on processing fluency and persuasion. Journal of personality and social psychology, 86(2), 205.

(26)

Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased assimilation and attitude polarization: The effects of prior theories on subsequently considered evidence. Journal of personality and social psychology, 37(11), 2098.

Lu, F. C., & Sinha, J. (2017). Speaking to the heart: Social exclusion and reliance on feelings versus reasons in persuasion. Journal of consumer psychology, 27(4), 409-421.

Mayer, J. D., Gaschke, Y. N., Braverman, D. L., & Evans, T. W. (1992). Mood-congruent judgment is a general effect. Journal of personality and social psychology, 63(1), 119.

Meloy, M. G. (2000). Mood-driven distortion of product information. Journal of Consumer Research, 27(3), 345-359.

Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of general psychology, 2(2), 175-220.

Oppenheimer, D. M., Meyvis, T., & Davidenko, N. (2009). Instructional manipulation checks: Detecting satisficing to increase statistical power. Journal of experimental social psychology, 45(4), 867-872.

Pham, M.T. (2004). The Logic of Feeling. Journal of Consumer Psychology, 14(4), 360-369 Pham, M. T., Cohen, J. B., Pracejus, J. W., & Hughes, G. D. (2001). Affect monitoring and the primacy of feelings in judgment. Journal of consumer research, 28(2), 167-188.

Pham, M. T., Lee, L., & Stephen, A. T. (2012). Feeling the future: The emotional oracle effect. Journal of Consumer Research, 39(3), 461-477.

Qiu, C., Lee, Y. H., & Yeung, C. W. (2009). Suppressing feelings: A double-edged sword to consumer judgment and choice. Journal of Consumer Psychology, 19(3), 427-439.

Reber, R., Schwarz, N., & Winkielman, P. (2004). Processing fluency and aesthetic pleasure: Is beauty in the perceiver's processing experience?. Personality and social psychology review, 8(4), 364-382.

Russo, J. E., Carlson, K. A., & Meloy, M. G. (2006). Choosing an inferior alternative. Psychological Science, 17(10), 899-904.

Russo, J. E., Meloy, M. G., & Medvec, V. H. (1998). Predecisional distortion of product information. Journal of Marketing Research, 35(4), 438-452.

(27)

Schwarz, N. (2004). Metacognitive experiences in consumer judgment and decision making. Journal of Consumer Psychology, 14(4), 332-348.

Schwarz, N., & Clore, G. L. (1988). How do I feel about it? The information function of affective states. In K. Fiedler & J. P. Forgas (Eds.), Affect, cognition and social behavior: New evidence and integrative attempts (pp. 44-63). Toronto: C.J. Hogrefe.

Shiv, B., & Fedorikhin, A. (1999). Heart and mind in conflict: The interplay of affect and cognition in consumer decision making. Journal of consumer Research, 26(3), 278-292. Silva, R. R., Garcia-Marques, T., & Mello, J. (2016). The differential effects of fluency due to repetition and fluency due to color contrast on judgments of truth. Psychological research, 80(5), 821-837.

Song, H., & Schwarz, N. (2009). If it's difficult to pronounce, it must be risky: Fluency, familiarity, and risk perception. Psychological Science, 20(2), 135-138.

Stieger, S., Reips, U. D., & Voracek, M. (2007). Forced‐response in online surveys: Bias from reactance and an increase in sex‐specific dropout. Journal of the American society for

information science and technology, 58(11), 1653-1660.

Sundar, A., Kardes, F. R., & Wright, S. A. (2015). The influence of repetitive health messages and sensitivity to fluency on the truth effect in advertising. Journal of Advertising, 44(4), 375-387.

Tsai, C. I., & Thomas, M. (2011). When does feeling of fluency matter? How abstract and concrete thinking influence fluency effects. Psychological Science, 22(3), 348-354.

Unkelbach, C. (2007). Reversing the truth effect: Learning the interpretation of processing fluency in judgments of truth. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(1), 219.

Winkielman, P., & Cacioppo, J. T. (2001). Mind at ease puts a smile on the face: psychophysiological evidence that processing facilitation elicits positive affect. Journal of personality and social psychology, 81(6), 989.

Xuhong, Z. (2016). The Effect of the Extraordinary: Reliance on Feeling Over Reason in Decision Making. ACR North American Advances.

(28)

APPENDIX 1: Survey

Start of Block: Default Question Block

Q1 DISCLOSURE FORM

This is a disclosure form for research participation. It is standard and contains important information about this study and what to expect if you decide to participate. Please consider the information carefully.

You are being asked to take part in a research study. The survey will consist of two parts about how consumers make decisions. Your participation is voluntary. You may refuse to participate in this study. If you begin the study, you may leave the study at any time. There are no known risks involved. There is no direct benefit but you may learn about consumer decision making. We anticipate no costs apart from the time you spend completing the research.

Efforts will be made to keep your study-related information confidential. If you agree to participate in this study, you will not give up any personal legal rights you may have as a participant in this study.

By finishing the participation in this study, you have a chance on winning one of the Bol.Com giftcards worth of 20 euros! If you would like to participate for the prices, do not forget to fill in your name at the end of the survey.

For questions, concerns, or complaints about the study you may contact me anytime via email: b.r.lukkien@student.rug.nl

Thank you very much for your consideration of our study.

IF YOU GIVE YOUR CONSENT TO PARTICIPATE IN THIS STUDY PLEASE CONTINUE. IF YOU DO NOT GIVE YOUR CONSENT, PLEASE CLOSE YOUR BROWSER NOW.

(29)

Q2 Dear Participant,

In the following you will participate in two small studies. The first study will be a study for your trust in feelings, which will take 4 minutes. Then, the second study will be a decision making study in which you have to choose between two restaurants. Please read the following instructions to ensure a smooth process.

1. Note that this survey will include attention checks. Therefore, please always read the question first before answering.

2. Please do not talk about the content of the studies to other people.

3. Do not forget to fill in your name and email adress in the end to have a chance on the price; the Bol.com Giftcard of 20 euros!

Again, thank you for your participation! Please click on the arrow below to start.

End of Block: Default Question Block Start of Block: TIF manipulation_high

Q39 For this first study, we would like you to briefly describe different situations in which you trusted your feelings to make a judgment or a decision and it was the right thing to do.

In other words, describe specific situations that really happened to you in which you felt confident in making a judgment or a decision based on how you felt, and later found out that you were right in using your feelings for this judgment or decision. The situations you list could involve any type of decision, provided that they were real situations in which you felt pretty certain that you should go by what your feelings were telling you to do, and it turned out that your feelings were right.

For example, one participant in a previous session of this study mentioned the following:

“I had a very good feeling toward my new roommate from the moment I met him. Although objectively we were quite different, my feelings told me that we would be good friends. My feelings were correct! We became best friends after only a week. Trusting how I felt about my new roommate was definitely the right think to do.

(30)

Start listing these TWO (2) situations in which you were right in trusting my feelings on the next page…

PLEASE CLICK ON THE ARROW BELOW TO START THE TASK.

Page Break

Q43

START OF THE TASK.

You will only be able to proceed after 4 minutes. If you don't know a situation anymore, please write "I don't remember"

Q45 1ST situation in which I was right in trusting my feelings:

(31)

Q47 2ND situation in which I was right in trusting my feelings: ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ Q49 THANK YOU

PLEASE WAIT UNTIL THE END OF THE 4 MINUTES. THEN AN ARROW WILL APPEAR ON THE BOTTOM RIGHT CORNER. CLICK THIS ARROW TO

CONTINUE

End of Block: TIF manipulation_high Start of Block: TIF manipulation_low

Q51 For this first study, we would like you to briefly describe different situations in which you trusted your feelings to make a judgment or a decision and it was the right thing to do.

In other words, describe specific situations that really happened to you in which you felt confident in making a judgment or a decision based on how you felt, and later found out that you were right in using your feelings for this judgment or decision. The situations you list could involve any type of decision, provided that they were real situations in which you felt pretty certain that you should go by what your feelings were telling you to do, and it turned out that your feelings were right.

For example, one participant in a previous session of this study mentioned the following:

“I had a very good feeling toward my new roommate from the moment I met him. Although objectively we were quite different, my feelings told me that we would be good friends. My feelings were correct! We became best friends after only a week. Trusting how I felt about my new roommate was definitely the right think to do.

(32)

your decision (three or four sentences per situation will be enough).

Start listing these TEN (10) situations in which you were right in trusting my feelings on the next page…

PLEASE CLICK ON THE ARROW BELOW TO START THE TASK.

Page Break

Q55

START OF THE TASK.

You will only be able to proceed after 4 minutes. If you don't know a situation anymore, please write "I don't remember"

Q57 1ST situation in which I was right in trusting my feelings:

(33)

Q59 2ND situation in which I was right in trusting my feelings: ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________

Q61 3RD situation in which I was right in trusting my feelings:

________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________

Q63 4TH situation in which I was right in trusting my feelings:

(34)

Q65 5TH situation in which I was right in trusting my feelings: ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________

________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________

(35)

Q71 8TH situation in which I was right in trusting my feelings: ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________

________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________ ________________________________________________________________

(36)

Q77 THANK YOU

PLEASE WAIT UNTIL THE END OF THE 4 MINUTES. THEN AN ARROW WILL APPEAR ON THE BOTTOM RIGHT CORNER. CLICK THIS ARROW TO

CONTINUE

End of Block: TIF manipulation_low

Start of Block: DV confirmation bias measure - on chaxel 2018

Q145

YOU HAVE COMPLETED THE FIRST STUDY.

We will now move on to the second study for today. Click on the arrow to proceed.

(37)

Q149 STUDY 2- Restaurant Choice Study

On the pages that follow, you will be asked to read and evaluate information that will help you make a decision between two restaurants. Imagine that you are on holiday in the US and you are planning a dinner with your partner. Your budget is not limited. Please try to become the decision maker and really think about which of the 2 restaurants you would pick. Page Break Q153 APPEARANCE Restaurant J:

Restaurant J has a main dining area with a mixture of large and small tables. In the smaller adjacent dining area, a fireplace dominates the room. Soft lighting comes from dimmable, electrified candles located on the wall in the main dining room. In the smaller dining area, the fireplace, when lit, provides a rosy glow.

Restaurant K:

Restaurant K is composed of multiple dining areas, with tables suited to the decor of the room. A dining nook (= small private dining area) is available and has small tables with wing-back chairs. Candles decorate each table. At the larger tables, multiple candles illuminate the table. In the largest dining are, a chandelier hangs from the ceiling.

(38)

Q157 Please consider only the information that you have just received. Rate it on the scale below according to your personal judgment.

o

1 (Information strongly favors Restaurant K) (1)

o

2 (2)

o

3 (3)

o

4 (4)

o

5 (Information favors neither option) (5)

o

6 (6)

o

7 (7)

o

8 (8)

o

9 (Information strongly favors Restaurant J) (9)

Page Break

Q161 Considering all the information you have received so far, which restaurant are you leaning toward choosing at this time? Remember that more information is coming, so this is only your current, not final opinion.

o

Restaurant K (1)

o

Restaurant J (2)

(39)

Q163 How confident are you that your current choice will also be your final choice? Absolutely uncertain 0 Absolutely certain 100 Page Break

Q188 Please consider only the information that you have just received. How would you rate the information in terms of how difficult it was to process for you?

(40)

Q189 How would you rate the information in terms of how easy it was to understand?

o

1= difficult to understand (1)

o

2 (2)

o

3 (3)

o

4 (4)

o

5 (5)

o

6 (6)

o

7= easy to understand (7) Page Break Q167 DESSERTS

Restaurant J: Restaurant J has a wide assortment of standard desserts, and daily specials always feature one type of cheese-cake, a fruit sorbet, and a chocolate dessert. In addition, a professional dessert chef is on staff and will accommodate requests.

Restaurant K: Restaurant K has a variety of desserts including cakes, pies, and ice cream. In addition, specialty desserts are prepared daily. Dessert and coffee are included in the price of the meal.

(41)

o

2 (2)

o

3 (3)

o

4 (4)

o

6 (6)

o

7 (7)

o

8 (8)

o

Page Break

o

Restaurant K (1)

o

Restaurant J (2)

(42)

Q193 Please consider only the information that you have just received. How would you rate the information in terms of how difficult it was to process for you?

(43)

o

2 (2)

o

3 (3)

o

4 (4)

o

5 (5)

o

6 (6)

o

7= easy to understand (7) Page Break Q181 HOURS OF OPERATION Restaurant J:

Restaurant J serves lunch between 11 A.M. and 1:30 P.M. Dinner is served between 5 P.M. and 9:30 P.M.

Restaurant K

Restaurant K serves lunch between 11:30 A.M. and 2 P.M. Dinner hours are between 4:30 P.M. and 9 P.M.

(44)

o

2 (2)

o

3 (3)

o

4 (4)

o

6 (6)

o

7 (7)

o

8 (8)

o

Page Break

Q189 Considering all the information you have received so far, which club are you leaning toward choosing at this time? Remember that more information is coming, so this is only your current, not final opinion.

o

Restaurant K (1)

o

Restaurant J (2)

(45)

Q195 Please consider the information that you have just received. How would you rate the information in terms of how difficult it was to process for you?

(46)

o

2 (2)

o

3 (3)

o

4 (4)

o

5 (5)

o

6 (6)

o

7= easy to understand (7) Page Break

Q195 DINING REVIEWS FROM Trip Advisor

Restaurant J:

"Loved the atmosphere, the food and the service. The waiter and staff was polite and friendly. The food was pricey but delicious with large portions. I got to indulge in the macaroni and cheese, the deep fried pickles, and the fried onion brick. It was all sooo good! Then my girlfriends and I spoiled ourselves with the milkshakes and alcohol shakes. By the time our meal came, we could barely attack our plate as we went a bit overboard. Next time, I return I plan to hit up the dessert menu and drinks! A wonderful place for drinks with a friend or a full meal get together!"

Restaurant K:

(47)

Page Break

o

2 (2)

o

3 (3)

o

4 (4)

o

6 (6)

o

7 (7)

o

8 (8)

o

Page Break

o

Restaurant K (1)

o

Restaurant J (2)

(48)

Q197 Please consider the information that you have just received. How would you rate the information in terms of how difficult it was to process for you?

Trust in Feelings & Confirmation Bias: A relationship mediated by processing fluency