University of Groningen

(1)

University of Groningen

Can Consumers Make Inferior Decisions When Trusting

in Their Feelings?

Preference for Consistency Strengthening the Relationship between

Trust in Feelings and Confirmation Bias.

Master Thesis

Jasper Bos

(2)

Can Consumers Make Inferior Decisions When Trusting

in Their Feelings?

Preference for Consistency Strengthening the Relationship between

Trust in Feelings and Confirmation Bias.

Jasper Bos

University of Groningen, Department of Marketing

Master Thesis

16 June, 2019

Address: Hesselingen 30, Meppel

Phone: +316 83840520

E-mail:

j.bos.33@student.rug.nl

Student Number: S2978296

(3)

Management Summary

It is widely known in the literature that human reasoning could lead to inferential error. One reason for making inferior decisions is the fact that there are some problematic aspects which reduces the human’s ability to reason properly. One of the most known problematic aspect of human reasoning is the confirmation bias. In our paper, we defined confirmation bias as ‘’the human tendency to see what we expect, need, and/or want to see from our environment’’. Therefore, people tend to selectively seek and interpret information in a way that it confirms their existing belief. We investigated trust in feelings as a possible factor that could lead to an individual engaging in confirmatory reasoning. We defined trust in feelings as ‘’the degree to which a person believes that his or her feelings point toward the right direction in judgements and decisions’’. We hypothesized that individuals high on trust in feelings are more likely to engage in confirmation bias compared to individuals that do not trust their feelings. Furthermore, we also investigated whether preference for consistency could influence such an effect of high trust in feelings on confirmation bias. In our research, preference for consistency is treated as a personality trait. We hypothesized that high

preference for consistency strengthens the effect of trust in feelings on confirmation bias.

We proposed a 2 (trust in feelings: high vs low, manipulated) x 2 (cognitive

consistency: high vs low, measured) design. We conducted a survey over 197 participants on Amazon Mechanical Turk, who were randomly assigned to either the high trust in feeling HTIF condition or the low trust in feelings LTIF condition. To manipulate the participants’ trust in feelings, we started the experiment with the manipulation task derived from Pham et al. (2012). Subsequently, we put in the stepwise evolution of preference method of Chaxel and Han (2018) in order to look whether participants engaged in the confirmation bias. We put the brief preference for consistency scale of Cialdini and Trost (1995) at the end of the survey in order to measure participants’ preference for consistency.

Results indicated that the manipulation task used in the survey did not work, meaning that people in the HTIF condition did not actually trust their feelings more in decision

(4)

interaction effect of preference for consistency on the effect of high trust in feelings on confirmation bias. We can thus say that marketing appeals that encourage individuals to rely on their feelings should not increase confirmation bias.

Additional analyses pointed out that there was a significant effect of preference for consistency on the use of feelings, indicating that individuals with high preference for consistency use their feelings more in decision making compared to people with low

(5)

Preface

The basis of this thesis stems from my interest in consumer behaviour I have developed from the Consumer Psychology course. My interests are in the limited cognitive capacity consumers have which results in the use of heuristics, biases and habits. As marketer, it is crucial to know why consumers buy certain products and how decision-making processes work.

(6)

Introduction

Decision making has become more difficult over the last decades with the increase of offerings and marketing all around us. Take for example the supermarkets, offering tons of alternatives for each product category which makes it hard to choose for consumers. Besides, people get more and more exposed to the marketing of different products. It takes a lot of effort to process all the available information of all available offerings, hence making purchase decisions. As a result, when there is too much information to sort through, people might engage in conformation reasoning. Rehak et al. (2010) define confirmation bias as ‘’the human tendency to see what we expect, need, and/or want to see from our environment’’. Therefore, people tend to seek information that is in line with their existing beliefs, rather than seek information to disprove their beliefs (Chia, 2005). People may take inferior decisions when they engage in confirmation reasoning. For example, Park et al. (2010) found investors use message boards to seek information that is in line with their beliefs about their

investment. Hence, when investors process information from message boards, they exhibit confirmation bias. As a result, this confirmation bias leads to overconfidence which results in higher trade frequency, but lower realized returns. Another example concerns policy

rationalization. Tuchman (1984) argues that governments becomes focused on justifying a policy once it is adopted and implemented. An example is the policy that drew the United States into the Vietnam war and kept the military engaged for 16 years, despite all the evidence that indicated it was a lost cause from the beginning.

(8)

2

In this paper, we are investigating trust in feelings as an important additional factor that contributes to the confirmation bias. This factor can be considered important because previous research has shown that people generally rely on feelings to make a variety of judgements and decisions (Pham 2004; Schwarz and Clore, 1996). For example, consumers often tend to rely on their feelings towards a product or service when making a purchase decision (Pham, 1998). One of the conditions that promotes a greater influence of feelings in judgement and decision making is the perceived diagnosticity (Informativeness) of the feelings (Greifeneder et al., 2011). Besides representativeness and relevance as determinants of the informativeness of feelings, Avnet et al. (2012) investigated trust in feelings as an important determinant for perceived informativeness of feelings in judgement and decision-making. In their study, Avnet et al. (2012) defined trust in feelings as ‘’the degree to which a person believes that his or her feelings point toward the right direction in judgements and decisions.’’ Although Avnet et al. (2012) identified trust in feelings as third determinant of informativeness of feelings, there is little research about decision outcomes when trusting on your feelings. Pham et al. (2012) found in eight studies that individuals with higher trust in feelings are better in predicting outcomes of future events compared to individuals with lower trust in feelings. High trust in feelings (compared with low trust in feelings) improved the prediction accuracy. Yet, the improvement in prediction accuracy only occurred when the individual possessed sufficient background knowledge about the domain. However, we expect that trust in feelings could also promote biases such as the confirmation bias. We expect that people who truly believe that their feelings point toward the right direction, will tend to seek information confirming their existing beliefs. In other words, feelings can serve as a shortcut, but at the same time, the reliance on our feelings can also promote confirmation bias.

(9)

3

expect that preference for consistency strengthens the effect of trust in feelings on confirmation bias.

In this paper, we investigate the effect of trust in feelings on the likelihood of confirmation reasoning. Furthermore, we investigate if the effect of trust in feelings on confirmatory reasoning is moderated by individuals’ general preference for consistency (see conceptual framework in figure 1). It is expected that when an individual believes that his or her feelings point towards the right direction in judgement and decisions, he is not paying attention to subsequent information and therefore engages in confirmation reasoning. We therefore expect that individuals high on trust in feelings engages in confirmatory reasoning, whereas low trust in feelings has no effect on the confirmation bias. Besides, we expect that this effect is even stronger for people with high preference for consistency compared to individuals that score low on preference for consistency.

(10)

4

Literature review

Feeling as information

There used to be the assumption that human beings have the ability to make good and rational decisions all the time. However, our cognitive capacities are limited which entails that we cannot make rational decisions all the time. Historically, the consumer decision-making literature represented decision making as a process of assessing information about different attributes. Traditionally, behavioral decision theorists and consumer psychologists defined the notion of information in decision-making as perceptions, beliefs, and declarative knowledge structures (Pham, 2004). In 1983, Schwarz and Clore expanded the notion of information by demonstrating that affective feelings are also used as input in judgements and decision-making. This expansion significantly changed how researchers looked at judgement and decision-making in general. A growing body of research suggests that people also tend to rely on their feelings when making a decision (Pham 2004; Schwarz and Clore, 1996). For

example, people often rely on their feeling when making a purchase decision (Pham, 1998), or assess risk based on the feeling of anxiety and fear (Loewenstein et al., 2001). These findings had for example a significant influence on marketing. Knowing that people also rely on their feelings changed how marketing is done within businesses.

As a result of the expansion of the notion of information, it is nowadays well known in the psychology literature that people use two different systems in decisions-making, referred to as system 1 and system 2. System 1 is referred to as the ‘’Automatic mode’’, in which decisions are made fast, frequent and emotional, whereas system 2 is called the ‘’Reflective mode’’ in which decisions are made rationally (Szmigin & Piacentini, 2018). Because conscious thinking takes a lot of effort, most decisions are delegated to system 1. System 1 is also known as the peripheral route in the Elaboration Likelihood Model (Szmigin &

Piacentini, 2018). People use heuristics when they make decisions via the peripheral route. Heuristic are simple rule of thumbs that facilitates the decision-making process. Heuristics can be efficient and effective, but also lead us astray. One of the heuristic is the affect

(11)

5

serve as a shortcut in judgement and decision-making. But in what conditions do feelings influence the judgemental and decision-making process?

Prior studies categorized two conditions in which influence of feelings on judgement and decision-making increases. The first condition pertains the accessibility of the feelings during judgement or decision-making, compared to available inputs. Relative accessibility of feelings, compared to other inputs, increases when people pay attention to their feelings (White and McFarland, 2009). Similarly, people are also more influenced by their feelings when their ability to process information is reduced, whether by cognitive load (Shiv and Fedorikhin, 1999), time pressure (Pham et al., 2001), or distraction (Albarracin and Kumkale, 2003). The second condition pertains the perceived diagnosticity (informativeness) of the feelings. The review of Greifeneder et al. (2011) shows that feelings have greater influence in judgement and decision-making when these feelings are perceived to be informative for the judgement or decision at hand. Besides representativeness and relevance as determinants of the informativeness of feelings, Avnet et al. (2012) investigated trust in feelings as an important determinant for perceived informativeness of feelings in judgement and decision-making. When an individual trust in his/her feelings, he/she will likely use their feelings more in decision-making compared to an individual who does not trust his/her feelings.

Trust in feelings

Avnet et al. (2012) defined trust in feelings as ‘’the degree to which a person believes that his or her feelings point toward the right direction in judgements and decisions.’’ Avnet, Pham and Stevens (2012) performed six different studies in which they found evidence for the direct role of trust in feelings as a distinct determinant of informativeness of feelings in

judgement. Besides, they also found that trust in feelings interacts with other determinants of perceived informativeness of feelings in judgement. Hence, trust in feelings is an indicator for whether the feelings someone has can be used in decision-making (Avnet et al., 2012). Unlike representativeness, which is target specific, and relevance, which is goal and judgement specific, trust in feelings is person specific (Avnet et al., 2012). Avnet et al. (2012) were the first that believed and showed that person-specific factors beyond the task itself contribute to the perceived informativeness of feelings. As mentioned in the previous section, people tend to rely on their feelings when they perceive their feelings as informative for the decision at hand. The reliance on feelings depends on whether individuals trust what they feel, which differs among people (Pham, 2017). It is thus about an individual perception about the

(12)

6

point in the right direction regarding in decision-making and judgement, hence perceive their feelings as trustworthy, whereas others may perceive their feelings as untrustworthy, since they think their feelings usually point in the wrong direction in decision-making and

(13)

7

interpersonal consistency compared to individuals using reason-based assessments. Another study showed that individual who highly relied on the emotional automatic mode during decision making generated a higher preference for consistency (Lee et al., 2009). The studies above show positive outcomes from trusting in your feelings. However, we also believe that trusting in your feelings could also lead to inferior decisions. In the next session, we will discuss a possible negative outcome from trusting in your feelings and using these feelings for making decisions.

Confirmation bias

It is widely known in the literature that human reasoning could lead to inferential error. One reason for making inferior decisions is the fact that there are some problematic aspects which reduces the human’s ability to reason properly. One of the most known problematic aspect of human reasoning is the confirmation bias. Rehak et al. (2010) define confirmation bias as ‘’the human tendency to see what we expect, need, and/or want to see from our environment’’. Therefore, people tend to seek information that is in line with their existing beliefs, rather than seek information to disprove their beliefs (Chia, 2005). Nickerson (1998) mentioned two ways how people could evaluate and search information/evidence. The first way is to seek information on all sides of a question, evaluate every piece of information as objective as one can, and subsequently come to an unbiased, objective conclusion. The other way is to first draw a conclusion, then selectively gather, and give weight to information to justify the conclusion you have already drawn. In the latter way, one may seek evidence that supports his/her already drawn conclusion while neglecting counterarguments and give more weight to supporting information compared to counterarguments. Hence, in the second way, one engages in the confirmation bias. The confirmation bias does not only impact how people gather information, but also how people interpret and recall information (Nickerson, 1998). In other words, the confirmation bias refers to selectivity in the acquisition and use of information (Nickerson, 1998), preventing us from looking at situations objectively.

(14)

8

when searching for and process additional information in order to protect their existing belief. In other words, confirmation bias pertains to the biased search and interpretation of new information to favour emerging beliefs (Russo et al., 1998). Another concept used for this is information distortion, which is a form of confirmation bias.

To put this again in a consumption perspective, the following example is given.

Imagine, for example, that you need to make a decision between two hotels (hotel A and hotel B) where you can stay when going on a holiday with your family. You search information regarding the facilities they offer at both hotels and you develop a slight preference for Hotel B after comparing the facilities of both hotels. However, you decide to search for additional information about the hotels and you come across reviews about both hotels. Normally, you might evaluate the additional information (reviews) as neutral when processing it without the knowledge about the facilities they offer. However, since you already develop an initial preference you might shift your evaluation of information to favour hotel B. An explanation why people engage in confirmatory information processing is because people have a

preference for consistency (Chaxel and Han, 2018).

(15)

9

rationally is reduced. Subsequently, people may use their feelings to come to a biased decision or judgement.

As mentioned above, prior research mentioned relative accessibility of the feelings, similarly the reduced ability to process information, as a condition that increases the influence of feelings in decision making. Dror & Fraser-Mackenzie (2008) found that when an

individual’s ability to process information is reduced, confirmation biases are even more likely to occur. In other words, when people’s ability to process information decreases, the influence of feelings in decision making increases, which makes confirmation reasoning more likely to occur. A possible explanation could be that when someone believes that his or her feelings point toward the right direction in judgements and decisions, they selectively acquire and put weight on information that support their beliefs. Besides, someone with high trust in their feelings is also likely to neglect and put less weight on information that counters his or her beliefs. Therefore, we come up with the following hypothesis:

H1: High trust in feelings contributes to the confirmation bias

However, it is also interesting to look at preference for consistency as a potential influencer on the effect between trust in feelings and confirmation bias, hence show

differences among people. As mentioned earlier, Lee et al. (2009) stated that individuals who highly rely on their emotional responses during decision-making generate a higher preference for consistency. Besides, other study showed that preference for consistency is a significant driver of confirmatory information processing (Chaxel et al., 2018). In the following section, we will look at preference for consistency, and try to establish links to the effect of trust in feelings on confirmation bias.

Preference for consistency

People strive for consistency in order to get a coherent understanding of their surroundings. This has always been the case. For centuries back, it was essential to seek consistency in the surroundings for people who had to make survival-oriented predictions (Chaxel and Han, 2018). Early theorists of social psychology contributed to the development of research and theories regarding the need for consistency among humans (Festinger, 1957; Berkowitz & Devine, 1989; Lecky, 1945). Lecky (1945) described the need for

self-consistency as a fundamental human motive, whereas Festinger (1957) even viewed

(16)

10

their behaviors and attitudes and that people have a need to appear consistent to others. In his book about the cognitive dissonance theory, Festinger (1957) mentioned that people have a need for a positive intrapersonal state, which means that they have the need to maintain consistency between their attitudes and behaviors. On the other hand, an individual will generate a negative intrapersonal state (dissonance) when he or she perceives an inconsistency among his/her cognitions, which motivates the individual to seek and implement a strategy to alleviate this state. Therefore, when people are aware of inconsistencies, they will take steps to reduce either the inconsistency itself or their awareness of it since it is not the desired state of people’s cognitions. The motivation to reduce cognitive dissonance results from the negative psychological consequences of experiencing cognitive dissonance. Zajonc (1960) even described this negative intrapersonal state (cognitive dissonance) as ‘’a painful or at least psychologically uncomfortable state.’’ Brehm and Cohen (1962) reframed the cognitive dissonance theory of Festinger and hypothesized cognitive dissonance as a state of arousal. Later on, experiments of Elkin and Leippe (1986) provided the first definitive evidence supporting the hypothesis that there is a psychological arousal component to the state of cognitive dissonance. In short, the studies above indicate that people want to maintain

consistency between their attitudes and behaviour, and thus avoid cognitive dissonance, since it puts them in a psychologically uncomfortable state.

Over the past decades, There has been a focus on why people are motivated to reduce cognitive dissonance. Some psychologists consider consistency as an indicator of good mental health and successful adaptation (Rogers, 1959; Funder, 1995). Cross, Gore and Morris (2003) state that a person has one ‘’real’’ self, whereas consistent expression of stable attitudes, behaviour and other personal characteristics form the foundation for validating and defining the real self. Cross, Gore and Morris (2003) stated that individuals who believe to know their real self are able to resist influence of others and therefore behave autonomously. Inconsistency can lead to the sense of having a divided self, or self-concept confusion. Consequently, individual consistency is indicative of self-integrity, unity, and maturity, and therefore associated with a positive well-being (Cervone & Shoda, 1999). Other research also showed that individuals who have consistent self-concepts report higher levels of well-being compared to individuals who have more inconsistent self-concepts (Donahue et al, 1993; Sheldon et al., 1997).

(17)

11

particular way and help to (2) predict human behavior. Assuming that others want to maintain a consistent personality allows people to explain but especially predict human behaviour, but also facilitate smooth social interactions. Cialdini (1993) provided the following example: ‘’ If an individual is characterized as “honest,” he or she is assumed to be honest across

situations, and other people predicate their interactions with this person on the belief that he or she will behave honestly in the future.’’ Swann et al. (1992) even indicated with their self-verification theory that individuals select relationship partners based on the consistency between their perception of them and their own self-perception since this leads to smooth interpersonal interactions.

Cialdini (1993) suggested that consistency serves as a heuristic for decision making. The consistency rule provides an efficient and quick shortcut for decision making once a person has taken a stand on an issue for the very first time. Acting the same way as previously in a given situation is a result from this consistency shortcut. For marketers, the notion of preference consistency helps to understand, predict, and influence consumer behaviour. Most marketing activities assume that consumers behave in a consistent way. However, there are differences between individuals regarding decision making and it is expected that not every person has equal preference for consistency. In other words, preference for consistency is person-specific. As mentioned earlier, prior research established that people use two different systems when making judgements or decisions. There is a distinction between the automatic emotional mode and the reflective cognitive mode, which has a substantial value in explaining behavior. Individuals use the automatic mode to solve evolutionarily recurrent situations and the carry out fast and accurate decision making and judgemental evaluations (Lee et al., 2009). The set of programs that are used when making a decision via the emotional system have the effect of activating, mobilizing, and coordinating a pool of resources, mental

(18)

12

significant driver of confirmation bias.

Therefore, we expect that preference for consistency plays a role in the effect of trust in feelings on confirmation bias. Consequently, individual differences in preference for consistency (Cialdini & Trost, 1995) may moderate the effect of trust in feelings on

confirmation bias. Specifically, we expect that the effect trust in feelings on the confirmation bias will be strengthened by preference for consistency. Specifically, we expect that

individuals high on trust in feelings (compared with individuals low on trust in feelings) will be more likely to engage in confirmatory reasoning and this effect should be even stronger for individuals high on trust in feelings who generally exhibit a high preference for consistency. We do not have an hypothesis for individuals low on preference for consistency.

H2: High preference for consistency strengthens the effect of trust in feelings on confirmation bias.

Method

Experimental Design

We propose a 2 (trust in feelings: high vs. low, manipulated) x 2 (cognitive

consistency: high vs. low, measured) design. Participants are randomly assigned to the high or low trust in feelings condition. The study uses trust in feelings as independent explanatory variable which consist of two different conditions, namely, the high trust in feelings condition and the low trust in feelings condition. As mentioned above, We expect that high trust in feelings has a positive effect on the confirmation bias, strengthened by preference for

consistency. The independent variable, trust in feelings, will be manipulated using the method Pham et al. (2012) used in their study. The dependent variable, confirmation bias, will be measured using the exact same experimental setup and stimuli as in the fifth experiment of Chaxel and Han (2018). Cognitive consistency will be measured using the brief preference for consistency scale developed by Cialdini and Trost (1995). We will elaborate further on the different measurements and tasks in the experimental procedure section.

Subjects and design

(19)

13

accessibility for every worker account on MTurk, which resulted in 200 responses within an hour. When analysing the data in Qualtrics, it became clear that 79 people failed one or both attention check(s) that were in the survey. Consequently, we rejected those responses that failed one or both attention check(s). Besides, we also rejected 10 other participants because they filled in irrational answers which indicated that the survey had been filled in by a bot. Those participants did also not receive the reimbursement. Because we rejected and deleted 89 responses in total, we had to set up a second batch on MTurk. For the second batch of 89 respondents, we included some specific qualifications that workers had to meet in order to fill in our survey. The first qualification was location based, workers had to come from United States to ensure that they were native English speakers and to ensure that the incentive was attractive for them since this was in dollars. The second qualification that workers had to meet in order to work on our survey was that they needed to have a HIT Approval Rate (%) for all Requesters’ HITs equal or higher than 90%. We included the second qualification in order to ensure that bots got excluded (because they get rejected more often) and therefore increase the possibility that participants filled in the survey the best they can. Most workers on MTurk that have a high approval rate want to keep it high so that they can participate in studies that have a high HIT approval rate as qualification criteria. There were still 18 respondents rejected and not rewarded from the second batch that we published because they either failed one or both attention check(s), or filled in answer that were not logical which indicates that it has been filled in by a bot. The 18 participants who got rejected were not deleted from the dataset. For the third batch of 18 responses, we changed the second qualification criteria to an approval rate equal or greater than 95%. Eventually we kept 218 surveys within our dataset which we exported to SPSS.

Experiment Procedure

The experiment consisted of four different phases, namely introduction, manipulation, confirmation task and measurement. At first, participant got on a screen with the disclosure form for research participation. After the disclosure form, participants were exposed to instructions of the experiment, which stated that participant will participate in two unrelated small studies for which they will be rewarded. Furthermore, the instructions drew

(20)

14

Subsequently, participants started with the manipulation task. As mentioned above, participants were randomly assigned to one of the two conditions, either the high trust in feelings conditions or the low trust in feelings condition. The manipulation task is derived from the study of Pham et al (2012) and capitalizes on the well-known ease-of-retrieval phenomenon. We manipulated participants’ trust in feelings by asking them to retrieve a number of situations in which they successfully relied on their feelings in judgements and decision making. The task started with a brief description explaining the task, asking participant to describe different situations in which they trusted their feelings to make a judgement of decisions in and it was the right thing to do. Participants in the low trust in feelings (LTIF) condition were asked to describe 10 situations in which they trusted their feelings in making a judgement or decisions and it was the right thing to do, whereas

participant in the high trust in feelings (HTIF) condition were asked to describe 2 situations. All the participants were given 4 minutes to come up with as many situations as possible. After the 4 minutes, they completed the ease-of-retrieval task. The manipulation will create a perception of success in past reliance on feelings (HTIF) or lack of success in past reliance on feelings (LTIF), which should increase participants’ trust in feelings (HTIF) or decrease participants’ trust in feelings (LTIF).

After finishing the manipulation task, all participants (both conditions) could continue with the second study. For the second study, participants had to choose between two

restaurants (K and J). This study derived from the paper of Chaxel and Han (2018). First, participants were briefly informed about study two to ensure they knew what was expected from them. For the task, they had to read and evaluate information about two different restaurants (restaurant J and restaurant K) and subsequently make a choice between the two restaurants. Participants were subsequently exposed to different kind of information about both restaurant (appearance, dessert, hours of operation, etc.). After each piece of information, they had to fill in whether the piece of information they just received favoured restaurant K or J on a nine point-scale rating (1= favors restaurant K, 5= favors neither option, 9= favors restaurant J), which restaurant they would choose to eat (based on all information received so far), and how confident they were that their current choice would be their final choice (rating from 1 to 100). Finally, after having received all information about the restaurant, participants were asked to make the final choice and how certain they were about their final choice. After the task, participants were asked to what extent they used their feelings in making the

(21)

15

Scale (Cialdini, 1995) to measure to what extent participants had a preference for consistency. The brief PFCS consists of nine different questions which indicates to what extent a

participant has a preference for consistency. In our sample, we could compute the items into one variable (α = .884)(see Appendix A). We expect that preference for consistency is a fixed personality trait, therefore, the previous tasks should not be of influence for the results on the preference for consistency scale. We will provide evidence that preference for consistency is not influenced by the manipulation task in the result section.

Preliminary Analysis

After exporting the data from Qualtrics to SPSS, we cleared the data to make it ready for analysis. First, we checked whether participants passed or failed the attention checks. We had two attention checks hidden in our survey. From the 218 participants in our dataset, 197 passed both attention checks, whereas 21 participants failed one or both attention check(s). Subsequently, we excluded those participants who failed one or both attention check(s) from our analysis. Secondly, we created the independent variable which we named TIF_CON, which is an abbreviation of trust in feelings conditions. Besides, we labelled TIF_CON as conditions which is more obvious when doing analysis. TIF_CON holds two values, namely 0 or 1. The value 0 indicates that a participant was assigned to the low trust in feelings

condition, whereas the value of 1 indicates that a participant was assigned to the high trust in feelings condition. Next, we checked whether every measurement item that wielded a Likert scale had the right values in SPSS, which was not the case.

(22)

16

0.6). Subsequently we decided to compute all the 9 variables of the brief preference for consistency scale into one variable which we labelled Preference_for_consistency.

Next, we looked at the manipulation measurement variables, which measures whether participants used their feelings when they decided which restaurant to choose. We included the two following questions in the survey: ‘’How much did you choose the restaurant on your gut feeling?’’ and ‘’To what extent did you base your choice on how right the restaurant felt?’’ The items statistics (see Appendix B) of both items show that participants scored high on the first item (M = 5.22 , SD = 1.67) and the second item (M = 5.69 , SD = 1.27), indicating that participants generally used their feelings while making decisions. Subsequently, we carried out a reliability analysis to check whether we could compute both variables into one variable. The reliability analysis (see Appendix B) indicates that both variables could be factored into one (α = .604 , α > 0.6). Since the reliability analysis only included two variables, it was logical that Cronbach’s alpha was not higher than 0.75. Therefore, we computed both variables into one variable as a measurement for use of feelings which we labelled Use_of_feelings.

We analysed information distortion (confirmation bias) following the guidelines of Russo et al. (1998), from which Chaxel and Han (2018) also derived their analysis. First, we computed the absolute difference between the participants’ rating of the diagnosticity of each attribute and the neutral value of the scale (5), which we called abs2, abs3, abs4, abs5. It is important to note that we did not calculate information distortion after the first attribute because there were no preferences formed yet. Hence, we started to calculate from the second attribute. For example, if a participant choose brand K as the brand they preferred after reading about the appearance (first attribute), and subsequently rated the diagnosticity of the information about the second attribute (desserts) as 1 (favouring K), the information distortion was + abs(1-5) = 4. On the other hand, if a participant preferred restaurant J after reading about the appearance, and subsequently rated diagnosticity as 1 (favouring K), the

information distortion was –abs(1-5)= -4. Subsequently, we computed the average distortion score among all attributes which we labelled Confirmation_Bias. The variable

(23)

17

Results and discussion

Manipulation check. First, we checked whether the manipulation worked in our

experiment. As mentioned earlier, we excluded 21 participants who failed the attention check(s) in further analysis. The descriptive statistics (see Appendix C) shows that in general, all participants scored high on the manipulation check measure Use_of_feelings (M = 5.45 ,

SD = 1.25). From the 197 participants, 89 were in the low trust in feelings condition (LTIF)

and 108 in the high trust in feelings condition (HTIF). We carried out an independent sample t-test analysis (see Appendix C) to measure the difference of the use of feelings between the high trust in feelings conditions and the low trust in feelings condition. Therefore, we used TIF_CON as our independent variable and Use_of_feelings as our dependent variable. The independent sample t-test shows there was not a significant difference between the low trust in feelings (M = 5.53 , SD = 1.30) and the high trust in feelings (M = 5.39 , SD = 1.22) conditions; t(195)= .776, p = .439 (figure 2). These results indicate that the manipulation derived from Pham et al. (2012) did not work in our survey. It is even the case that the LTIF condition scored higher on use of feelings compared to the high trust in feelings, which is contradicting to what we expected. However, as mentioned above, it is important to note that both conditions scored high on the manipulation check measurement, indicating that

participants in both conditions relied on their feeling while deciding which restaurant to choose. All in all, the insignificant results of independent samples t-test thus indicates that the manipulation did not work in our experiment.

Independent Samples Test

Levene's Test for Equality of

Variances t-test for Equality of Means

F Sig. t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper Use_of_feeling s

Equal variances assumed ,033 ,856 ,776 195 ,439 ,13920 ,17944 -,21469 ,49309 Equal variances not

assumed

,771 182,728 ,442 ,13920 ,18057 -,21708 ,49548

Figure 2: Independent Samples Test; Insignificant differences regarding trust in feelings between both conditions

(24)

18

feelings could have also resulted in, for example, increased disappointment, self-threat or something similar. Another possible explanation for the failed manipulation is the fact that the survey was conducted online via Amazon Mechanical Turk (MTurk). Because we conducted the survey online, we could not control the experimental conditions. Therefore, it is possible that participants were distracted by their surroundings while filling in the survey, resulting in the manipulation not having effect on them.

Although the manipulation did not work, we continued with the analysis. Because there is no difference between the HTIF and LTIF conditions, we carried out further analysis twice. First, we used the TIF_CON variable as independent variable. Subsequently, we carried out analysis using the Use_of_feelings measurement variable as independent variable. Note that this analysis is only exploratory and cannot serve to provide causal evidence for our hypothesized effects.

Manipulation task on preference for consistency. As mentioned earlier, we expect that

preference for consistency is a personality trait and therefore should not be influenced by the manipulation task. We conducted an independent samples t-test (see Appendix D) to compare difference on preference for consistency between both conditions. As expected, there was not a significant difference of scores on preference for consistency between the low trust in feelings (M = 6.30 , SD = 1.48) and high trust in feelings (M = 6.39 , SD = 1.31) conditions;

t(195)= -.453 , p = .651. The insignificant results indicate that preference for consistency is

indeed not influenced by the manipulation task. It therefore provides evidence for our claim that preference for consistency is a personality trait.

Main effect of trust in feelings on confirmation bias. Next, we analysed the main effect

of trust on feelings (manipulated binary IV) on confirmation bias (information distortion). As mentioned earlier, we followed the guidelines of Russo et al. (1998) to calculate confirmation bias. We computed the absolute difference between participants’ rating of diagnosticity of each attribute which we called abs2, abs3, abs4, abs5. Subsequently, we calculated

(25)

19

confirmation bias between the low trust in feelings (M = 0.71 , SD = 1.90) and the high trust in feelings (M = .83 , SD = 1.93) conditions; t(195)= -.411 , p = .682 (figure 3).

F Sig. t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper Confirmation_Bia s Equal variances assumed ,199 ,656 -,411 195 ,682 -,11291 ,27492 -,65511 ,42930

Equal variances not assumed

-,411 189,005 ,681 -,11291 ,27448 -,65434 ,42853

Figure 3: Independent Samples Test; Insignificant difference regarding confirmation bias between both conditions

The insignificant results indicate that high trust in feelings does not result in information distortion as we hypothesized. We expected that people who highly trust that their feelings point toward the right direction would selectively acquire and put weight on information that support their beliefs/preferences, whereas people low in trust in feelings would not engage in the confirmation bias. However, the results of the test shows that there is no significant differences between high trust on feelings and low trust in feelings. Since there is no difference between the participants in HTIF and LTIF conditions regarding the use of feelings, we did an additional analysis for the main effect. We carried out a simple linear regression analysis to predict confirmation bias based on use of feelings. An insignificant regression equation was found (F(1,195) = 0.651 , p > .05) with an R2 _{of .003 (see Appendix}

F). Results thus indicate that confirmation cannot be predicted by use of feelings. When looking at the coefficients, it even indicates that Confirmation_Bias slightly decreases when Use_of_feelings increases (β = -0,058). The negative β is opposite from what we

hypothesized based on the literate we found. However, the independent t-test did indicate that people in the HTIF scored slightly higher on information distortion compared to the LTIF condition. Since both analyses showed and insignificant effect/correlation, we cannot draw any conclusion on the coefficients and correlations between both variables. Based on the results of the analyses mentioned above, we reject H1.

Interaction effect preference for consistency. Next, we performed a moderation

(26)

20

confirmation bias. First, we used TIF_CON as independent variable X, the composite measure of confirmation bias: Confirmation_Bias as dependent variable Y, and PFC_ALL as

moderator M. The results of the moderation analysis (see Appendix G) indicate there is no interaction effect of preference for consistency on the effect of trust on feelings on

confirmation bias; β = -.0832, SE = .1963, t(193) = -.4240, p = .67, 95% CL [-.4705 , .3040]. Based on these results of the first moderation analysis, we can conclude that preference for consistency does not have any effect on the relationship between trust in feelings and confirmation bias.

Next, just as we did with the analysis of the main effect, we conducted an additional analysis for the interaction effect, using Use_of_feelings as independent variable instead of TIF_CON. Again, this is done because the manipulation did not work at the beginning of the survey. The results (see Appendix H) indicate there is no interaction effect of preference for consistency on the effect of trust on feelings on confirmation bias; β = -.0584, SE = .0638, t(193) = -.9164,

p = .36, 95% CL [-.1842 , .0673]. The results suggest there is no interaction effect. Based on

our moderation analyses, we reject H2.

Additional Analyses

Apart from analysing the data regarding our hypothesis, we also carried out additional analyses. Chaxel and Han (2018) found that preference for consistency is a significant driver of confirmation bias (information distortion). We wanted to check whether we could predict confirmation bias based on preference for consistency. Therefore, we carried out a simple linear regression analysis (see Appendix I). A significant regression equation was found (F(1,195) = 6.448 , p < .012) with a R2 of .032 (figure 4). Coefficientsa Model Unstandardized Coefficients Standardized Coefficients t Sig.

95,0% Confidence Interval for B

B Std. Error Beta Lower Bound Upper Bound 1 (Constant) 2,346 ,633 3,706 ,000 1,098 3,594

preference_for_consisten cy

-,247 ,097 -,179 -2,539 ,012 -,439 -,055

Figure 4: Coefficients table linear regression analysis; X = Preference for consistency, Y = confirmation bias

Results thus indicate that confirmation bias can be predicted by preference for consistency. Participants predicted engagement in confirmation bias is equal to 2.346 – 0.179 (preference for consistency) score (which could vary between -4 and 4) whereas preference for

(27)

21

confirmation bias decreased -.179 for each point they score higher on the brief preference for consistency scale. The analysis indicate that we can predict confirmation bias based on the equation mentioned above. However, the equation suggests that people that score higher on preference for consistency engage less in confirmation bias, which is contradicting to the study of Chaxel and Han (2018). It is therefore interesting for future research to investigate the link between preference for consistency and confirmation bias (information distortion).

Besides, we were also interested whether individuals with high preference for

consistency were more likely to use their feelings in making decisions. Therefore, we carried out a simple linear regression analysis with preference_for_consistency as predictor and Use_of_feelings as dependent variable (see Appendix J). A significant regression equation was found (F(1,195) = 31.496 , p < .000) with an R2

of .139. Results thus indicate that the use of feelings can be predicted by preference for consistency. Participants predicted use of feelings is equal to 3.313 + 0.373 (preference for consistency) score when preference for consistency is measured on a likert scale of 1 to 9. Participant’s use of feelings increases .373 for each point they score higher on the brief preference for consistency scale. The results thus suggest that the higher an individual’s preference for consistency, the higher the use of feelings in decision making. We reason differently compared to the theory of Lee et al. (2009), since they do not treat preference for consistency as a personality trait. Our results suggest that preference for consistency can predict the use of feelings, whereas Lee et al. (2009) found that the reliance on emotional responses increases preference for consistency.

As mentioned in the result part, there was no effect between trust in feelings on confirmation bias. However, we still wanted to look whether the first choice of a participant influenced their final choice. According to the confirmation bias literature, this should be the case. A Chi-square Test of Independence (see Appendix K) was performed to examine the relationship between a participant’s first choice of restaurant and final choice of restaurant. The relationship between these variables is significant Χ2 (1, 197) = 31.861 , p = .000. This test suggests that there is a statistically significant relation between the first and the final choice.

(28)

22

(Restaurant K) is a significant predictor of choosing restaurant J; ExpB = .177, p = .000, CL [.095 , 0.33]. The results show that a participant who first chose restaurant K, has a

probability of 0,177 of choosing restaurant J as final choice, suggesting that participants who had restaurant J as first choice are more likely to choose J as final choice. All in all, the Chi-square Test of Independence and logistic regression show that the final choice of a participant is influenced by their first choice. The logistic regression analysis also made a prediction variable which shows the probability of choosing restaurant J as final choice. This predicted probability variable shows that P(choosing restaurant J) = 68,4% if a participant’s choice_1 was restaurant J, whereas P(Final_choice = J) = 27,7% if a participant’s choice_1 was restaurant K. Hence, P(Final_choice = K) = (1- 0.277) = 72,3 % if a participant’s choice_1 was restaurant K, whereas P(Final_choice = K) = (1 - 0.684) = 31,6 % if participant’s choice_1 was Restaurant J.

Variables in the Equation

B S.E. Wald df Sig. Exp(B)

95% C.I.for EXP(B) Lower Upper Step 1a_{Choice_1(1)} _-1,732 _,317 _29,778 ₁ _,000 _,177 _,095 _,330

Constant ,773 ,201 14,725 1 ,000 2,167

Figure 5: Logistic regression Choice_1 and Final_Choice; variables in the equation

Since the Chi-square Test of Independence and logistic regression show a statistically significant relation between the first choice and final choice of participants, we decided to do another additional analyses to look whether use of feelings or preference for consistency moderates the relationship between Choice_1 and Final_Choice. We conducted a moderation analysis in order to check whether use of feelings moderates the influence of first choice on the final choice. Results show a significant effect of Choice_1 on Final_Choice, however, results also show that there is no significant interaction effect; p = .57 (see Appendix M) We conducted an additional moderation analysis in order to check whether preference for consistency moderates the effect of first choice on the final choice. Results show there is no significant interaction effect; p = .9294 (see Appendix N).

General Discussion

(29)

23

confirmation bias. We expected that people who highly trust that their feelings point toward the right direction would selectively acquire and put weight on information that support their beliefs/preferences. Besides, people with higher trust on feelings would also be more likely to neglect and put less weight on information countering their beliefs/preferences. Hence, high trust in feelings contributes to the confirmation bias. Although the manipulation task derived from Pham et al. (2012) did not work in our experiment, we decided to analyse the first hypothesis using TIF_CON as independent variable and later on change the independent variable with the manipulation measurement variable Use_of_feelings. Our study show no significant result, indicating that trust in feelings does not have any effect on the confirmation bias. In our additional analysis, we looked at the influence of the first choice on the final choice. According to the confirmation bias literature, the final choice should be influenced by the first choice. A Chi-square Test of Independence and logistic regression indicated that the first choice influences the final choice of participant. Participants who chose restaurant J after the first piece of information were also likely to pick restaurant J as their final choice and vice versa.

In this study, we also looked at a possible moderator for the effect of trust in feelings on confirmation bias, namely, preference for consistency. Hypothesizing that preference for consistency strengthens the effect of trust in feelings on the confirmation bias. However, our analysis showed no significant interaction effect of preference for consistency, which is logical since there was also no significant main effect. Figure 6 shows the results regarding our hypotheses.

Figure 6: Results regarding hypotheses; Red means rejected

(30)

24

preference for consistency increases, Chaxel and Han (2018) found the opposite, stating that preference for consistency is a significant driver of information distortion. Not only is this contrary with the findings of Chaxel and Han (2018), but it also does not make sense when thinking rationally about both concepts. Is it logical that someone with preference for consistency engages in confirmation bias in order to keep a consistency mindset. Possible explanations are the use of MTurk for conducting our survey, hence different conditions, and the limited number of participants. However, we found a significant negative correlation between both concepts so the link between both concepts is interesting for further analysis to establish the exact correlation.

The correlation between preference for consistency and use of feelings was positive,

indicating that the higher an individual’s preference for consistency, the more they use their feelings in decision-making. Although we reason differently compared to Lee et al (2009), we both found positive correlations between the use of feelings in decision making and

preference for consistency. It is interesting for further analysis to focus on whether preference for consistency is an individual-difference measure (personality trait) or not. If yes, we can state, according to our results, that people that have a high preference for consistency will rely more on their feelings while making decisions compared to people having low preference for consistency.

Practical implications

Based on our results, we cannot draw similar conclusions as Chaxel and Han (2018) and Russo et al. Whereas Chaxel and Han (2018) found that people higher on preference for consistency are more likely to engage in information distortion (confirmation bias), we found that the higher an individual’s preference for consistency, the less likely he/she is to engage in the confirmation bias. Therefore, further research is needed to establish under which

(31)

25

customers and target them in your marketing message (with appeals that encourage reliance on feelings). Our results also proved that the final choice of people can be predicted by the first choice. Therefore, it is important for marketers to appeal customers when they first come in contact with their products. When they develop an initial preference for the product, it is expected that they will also buy the product.

Limitation

One of the main limitations of this study was the lack of time. Due to time restrictions, we were not able to pre-test whether the manipulation derived from Pham et al (2012) would actually work in our setting. As mentioned above, our manipulation did not work which has influence in further analysis.

Another limitation is that the data is gathered online via Amazon Mechanical Turk. Because the experiment was put online, we had no control in how people actually filled in the survey. There could have been distractions while filling in the survey, which influences the procedure of the experiment. Because we conducted the survey online, we could not control the experimental setting which could be of influence in answering the questions. Besides, we used no requirements for the first batch of 200 participants, which resulted in participant who did not speak English as a native language. Therefore, it is questionable whether those

participants understood all the tasks and question included in the survey.

The third limitation of the study is a budget limitation. We had a budget limitation of €300,-, with which we were able to recruit 197 participants. Because of our budget limitation, we had limited number of participants in our analysis. As known, a small sample size

(32)

26

References

Albarracín, D., & Kumkale, G. T. (2003). Affect as information in persuasion: A model of affect identification and discounting. Journal of Personality and Social Psychology, 84(3), 453.

Avnet, T., Pham, M. T., & Stephen, A. T. (2012). Consumers’ trust in feelings as information. Journal of

Consumer Research, 39(4), 720-735.

Berkowitz, L., & Devine, P. G. (1989). Research traditions, analysis, and synthesis in social psychological theories: The case of dissonance theory. Personality and Social Psychology Bulletin, 15(4), 493-507. Brehm, J. W., & Cohen, A. R. (1962). Explorations in cognitive dissonance.

Cervone, D., & Shoda, Y. (Eds.). (1999). The coherence of personality: Social-cognitive bases of consistency,

variability, and organization. Guilford Press.

Chaxel, A. S., & Han, Y. (2018). Benefiting From Disagreement: Counterarguing Reduces Prechoice Bias in Information Evaluation. Journal of Consumer Psychology, 28(1), 115-129.

Chia, C. W. (2005). Countering positive confirmation biases in command teams: An experiment with different

interventions. SINGAPORE ARMED FORCES CTR FOR MILITARY EXPERIMENTATION (SINGAPORE)

FUTURE SYSTEMS DIRECTORATE.

Cialdini, R. B. (1993). Influence: The psychology of persuasion (Rev. ed.). New York: Morrow.

Cross, S. E., Gore, J. S., & Morris, M. L. (2003). The relational-interdependent self-construal, self-concept consistency, and well-being. Journal of personality and social psychology, 85(5), 933.

Donahue, E. M., Robins, R. W., Roberts, B. W., & John, O. P. (1993). The divided self: Concurrent and longitudinal effects of psychological adjustment and social roles on self-concept differentiation. Journal of

personality and social psychology, 64(5), 834.

Dror, I. E., & Fraser-Mackenzie, P. A. (2008). Cognitive biases in human perception, judgment, and decision making: Bridging theory and the real world. In Criminal investigative failures (pp. 79-94). CRC Press. Elkin, R. A., & Leippe, M. R. (1986). Physiological arousal, dissonance, and attitude change: Evidence for a dissonance-arousal link and a" Don't remind me" effect. Journal of personality and social psychology, 51(1), 55. Epstein, S. (2003). Cognitive‐experiential self‐theory of personality. Handbook of psychology, 159-184.

Festinger, L. (1957). A theory of cognitive dissonance Row. Peterson and Company.

Funder, D. C. (1995). On the accuracy of personality judgment: a realistic approach. Psychological

review, 102(4), 652.

Greifeneder, R., Bless, H., & Pham, M. T. (2011). When do people rely on affective and cognitive feelings in judgment? A review. Personality and Social Psychology Review, 15(2), 107-141.

(33)

27

Lecky, P. (1945). Self-consistency; a theory of personality.

Lee, L., Amir, O., & Ariely, D. (2009). In search of homo economicus: Cognitive noise and the role of emotion in preference consistency. Journal of consumer research, 36(2), 173-187.

Loewenstein, G. F., Weber, E. U., Hsee, C. K., & Welch, N. (2001). Risk as feelings. Psychological

bulletin, 127(2), 267.

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

psychology, 2(2), 175-220.

Park, J., Konana, P., Gu, B., Kumar, A., & Raghunathan, R. (2010). Confirmation bias, overconfidence, and investment performance: Evidence from stock message boards.

Pham, M. T. (1998). Representativeness, relevance, and the use of feelings in decision making. Journal of

consumer research, 25(2), 144-159.

Pham, M. T. (2004). The logic of feeling. Journal of Consumer Psychology, 14(4), 360-369.

Pham, M. T. (2014). Feels Right… Go Ahead? When to Trust Your Feelings in Judgments and Decisions. GfK

Marketing Intelligence Review, 6(1), 22-27.

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.

Rehak, L. A., Adams, B., & Belanger, M. (2010, September). Mapping biases to the components of rationalistic and naturalistic decision making. In Proceedings of the Human Factors and Ergonomics Society Annual

Meeting (Vol. 54, No. 4, pp. 324-328). Sage CA: Los Angeles, CA: Sage Publications.

Rogers, C. R. (1959). A theory of therapy, personality, and interpersonal relationships: As developed in the

client-centered framework (Vol. 3, pp. 184-256). New York: McGraw-Hill.

Russo, J. E., Meloy, M. G., & Medvec, V. H. (1998). Predecisional distortion of product information. Journal of

Marketing Research, 35(4), 438-452.

Russo, J. E., Meloy, M. G., & Wilks, T. J. (2000). Predecisional distortion of information by auditors and salespersons. Management Science, 46(1), 13-27.

Russo, J. E., & Yong, K. (2011). The distortion of information to support an emerging evaluation of risk. Journal of Econometrics, 162(1), 132-139.

(34)

28

Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judgments of well-being: informative and directive functions of affective states. Journal of personality and social psychology, 45(3), 513.

Schwarz, N., & Clore, G. L. (1996). Feelings and phenomenal experiences. Social psychology: Handbook of

basic principles, 2, 385-407.

Sheldon, K. M., Ryan, R. M., Rawsthorne, L. J., & Ilardi, B. (1997). Trait self and true self: Cross-role variation in the Big-Five personality traits and its relations with psychological authenticity and subjective

well-being. Journal of personality and social psychology, 73(6), 1380.

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.

Stephen, A. T., & Pham, M. T. (2008). On feelings as a heuristic for making offers in ultimatum negotiations. Psychological Science, 19(10), 1051-1058.

Swann, W. B., Stein-Seroussi, A., & Giesler, R. B. (1992). Why people self-verify. Journal of personality and

social psychology, 62(3), 392.

Szmigin, I., & Piacentini, M. (2018). Consumer behaviour. Oxford University Press.

Thurstone, L. L. (1924). THE NATURE OF GENERAL INTELLIGENCE AND ABILITY (III) 1. British

Journal of Psychology. General Section, 14(3), 243-247.

Tuchman, B. W. (2011). The march of folly: From Troy to Vietnam. Random House.

White, K., & McFarland, C. (2009). When do moods influence consumers’ product evaluations? The role of mood–acknowledgment and perceived appropriateness of using moods. Journal of Consumer Psychology, 19(3), 526-36.

(35)

29

Appendices

Appendix A: Reliability Analysis Preference for Consistency

Case Processing Summary

N % Cases Valid 197 100,0

Excludeda ₀ _,0

Total 197 100,0

a. Listwise deletion based on all variables in the procedure. Reliability Statistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items ,884 ,896 9 Item Statistics Mean Std. Deviation N PfC_1 6,20 2,130 197 PfC_2 6,68 1,740 197 PfC_3 6,81 1,895 197 PfC_4 6,61 1,777 197 PfC_5 6,51 1,777 197 PfC_6 6,54 1,872 197 PfC_7 6,87 1,884 197 PfC_8 6,78 1,933 197 PfC_9 4,15 2,272 197

Summary Item Statistics

Mean Minimum Maximum Range

Maximum /

(36)

30 Item-Total Statistics Scale Mean if Item Deleted Scale Variance if Item Deleted Corrected Item-Total Correlation Squared Multiple Correlation Cronbach's Alpha if Item Deleted PfC_1 50,95 117,901 ,719 ,638 ,863 PfC_2 50,48 123,077 ,767 ,663 ,861 PfC_3 50,34 118,358 ,818 ,701 ,855 PfC_4 50,54 122,995 ,750 ,633 ,862 PfC_5 50,64 123,273 ,742 ,592 ,862 PfC_6 50,61 119,412 ,801 ,713 ,857 PfC_7 50,28 119,998 ,779 ,635 ,858 PfC_8 50,38 119,328 ,773 ,654 ,859 PfC_9 53,00 158,990 -,147 ,148 ,940

Appendix B: Reliability Analysis Manipulation Measurement

Case Processing Summary

N % Cases Valid 197 100,0

Excludeda ₀ _,0

Total 197 100,0

(37)

31 Item Statistics

Mean Std. Deviation N How much did you choose the

restaurant on your gut feeling? - Not at all:A lot

5,22 1,665 197

To what extent did you base your choice on how right the restaurant felt? - Not at all:A lot

5,69 1,267 197

Summary Item Statistics

Mean Minimum Maximum Range

Maximum /

Minimum Variance N of Items Item Means 5,452 5,218 5,685 ,467 1,089 ,109 2 Item Variances 2,189 1,605 2,774 1,169 1,729 ,683 2

Appendix C: Independent-Samples T-Test (X = TIF_CON , Y = Use_of_feelings)

Group Statistics

TIF_CON N Mean Std. Deviation Std. Error Mean Use_of_feelings Low TIF condition 89 5,5281 1,29764 ,13755

High TIF condition 108 5,3889 1,21581 ,11699

F Sig. t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper Use_of_feeling s

Equal variances assumed ,033 ,856 ,776 195 ,439 ,13920 ,17944 -,21469 ,49309 Equal variances not

assumed

University of Groningen

University of Groningen

Can Consumers Make Inferior Decisions When Trusting

in Their Feelings?

Preference for Consistency Strengthening the Relationship between

Trust in Feelings and Confirmation Bias.

Master Thesis

Jasper Bos

Can Consumers Make Inferior Decisions When Trusting

in Their Feelings?

Preference for Consistency Strengthening the Relationship between

Trust in Feelings and Confirmation Bias.

Jasper Bos

University of Groningen, Department of Marketing

Master Thesis

16 June, 2019

Address: Hesselingen 30, Meppel

Phone: +316 83840520

E-mail:

j.bos.33@student.rug.nl

Student Number: S2978296

Table of contents

Introduction

Literature review

Method

Results and discussion

General Discussion

Practical implications

Limitation

References

Appendices

Summary Item Statistics

Summary Item Statistics