Think or sleep on it? The advantages of unconscious thought

(1)

Think or sleep on it?

The advantages of unconscious thought

Author: Maureen Smeenk Email: m.j.e.smeenk@student.rug.nl

Phone number: +316 20806575 Student number: s2960982

Department: Faculty of Economics and Business Master: MSc Marketing Management

Supervisor: Dr. J. Berger Second supervisor: Dr. J. Hoekstra

(2)

2

Abstract

Consumers experience a lot of online and offline information about products and their attributes. In recent years distracting features, such as pop-ups and banners, are also increasingly employed by (online) retailers. Distracting features together with the variety of information can increase the difficulty for consumers to choose the maximum product choice. When consumers have to choose between several choice alternatives, they can either conscious deliberate about the alternatives or perform a distraction task. During the distraction task consumers will unconsciously process product information. There are mixed findings about the performance of unconscious thought. This research contributes to the ongoing debate and aims to examine the effect of unconscious thought on maximum choice and the moderating effect of decision complexity and presentation mode, on the relationship between unconscious thought and maximum choice. We find with the experimental design, there is a positive relationship between unconscious thought and maximum choice in decision making, which is amplified when the decision is complex and not amplified by experiential information. These results are further validated by controlling for circadian asynchrony. The findings and implications of this research show that consumers who take time away from a decision or ‘sleeping on it’ before they make a final choice, are more likely to choose the maximum choice option. This is in line with the Unconscious Thought Theory.

(3)

3

Preface

Although it has been a turbulent time, with online thesis meetings and without the small talk conversations at the coffee corner at the University Library, I have been writing my thesis for the MSc Marketing Management at the University of Groningen. I would like to thank those involved in my graduation period. Firstly, Dr. J. Berger, who provided me with useful feedback and support during the process. Moreover, I would like to thank my second supervisor Dr. J. Hoekstra for taking the time to read and evaluate my thesis. I also want to thank all the participants; without their cooperation I would not have been able to conduct the research analyses.

Furthermore, I want to thank my colleagues at Nestlé for their support and insights. Through them, this interesting topic has been brought to my mind. They have inspired me with their new in store experiential food plates and I was interested how experiential information impacts consumer decision making.

(4)

4

INTRODUCTION

Consumers can counter a challenge when trying to make a purchase decision. Through a lot of information about products, their attributes, evaluations and opinions of other consumers it becomes difficult for consumers to choose the maximum choice option (Gao, Zhang, Wang & Ba, 2012). Distracting website features, such as pop-ups and banners are increasingly employed by online retailers, this can further increase the difficulty for consumers to make their maximum choice (Dang-Longani, 2018). A product that is not perceived as the maximum choice can result in consumer regret and return of the product (Tsiros & Mittal, 2000). Therefore, maximizing a purchase decision is important, as it increases the possibility that consumers' expectations of the product are met. This will lead to greater consumer satisfaction (Wang & Benbasat, 2008). Therefore, in marketing and information systems literature it has been a focal topic to study how to help consumers to make their maximum choice (Li & Sun, 2019).

Some studies are primarily focused on how to promote the learning ability of consumers in conscious ways (Dawes & Corrigan, 1974; Keeney & Raiffa, 1976; Simon, 1982; Baron, 2005). They have examined that conscious thought and deliberation is needed prior to decision making. On the other hand, other studies have examined the benefits of unconscious thought in decision making. Unconscious thought, referred to as “sleeping on it”, can occur during an unrelated (distraction) task (Dijksterhuis, 2004).

(7)

7 Nowadays there are a substantial number of published replications and extensions of the UTT (e.g., Ham, Van den Bos, & van Doorn, 2009; Lerouge, 2009; Usher, Russo, Weyers, Brauner, & Zakay, 2011; Abadie, Waroquier, & Terrier, 2017; Hu, Yu, Chu, Zhao, Jude, & Jiang, 2018; Hasford, Hardesty, & Kidwell, 2019; Shen, Sun, Heng & Chan, 2020). However, there are also studies published indicating that the Unconscious Thought Theory does not (or not always) lead to maximum choices (e.g., Acker, 2008; Calvillo & Penaloza, 2009; Lassiter, Lindberg, Gonzalez-Vallejo, Bellezza, & Phillips, 2009; Newell Wong, Cheung, & Rakow, 2009; Rey, Goldstein, & Perruchet, 2009; Thorsteinson & Withrow, 2009; Waroquier, Marchiori, Klein, & Cleeremans, 2009).

As a result, there is an ongoing debate about the advantages of unconscious thought in choice environments. For example, Bos, Dijksterhuis, & Van Baaren, (2011) demonstrated how conscious thought is more likely to lead to the choice of a car with a higher total number of favorable attributes, while unconscious thought is more likely to lead consumers to choose a car with favorable attribute ratings across a smaller, but more important set of attributes. Messner and Wänke (2011) also compared conscious and unconscious thought, showing that a period of unconscious thought can overcome choice overload effects to increase product satisfaction after consumption. Payne, Samper, Bettman, & Luce, (2008), on the other hand, showed that in a lottery task, where participants were presented with precise numerical values, conscious thought resulted in more maximum choices than unconscious thought.

According to Strick, Dijksterhuis, Bos, Sjoerdsma, Van Baaren & Nordgren, (2011) the Unconscious Thought Theory is observed only when certain moderators are met with regards to decision tasks. These moderators can pertain to the nature of the choices or to the way information acquisition and modes of thought are implemented. The present research examines two potential moderators that may influence the UTT. The first moderator regarding the nature of the choice and the second moderator regarding the information acquisition.

(8)

8 did not propose a criterion stating when a choice is complex. Therefore, it is interesting to examine how decision complexity functions for unconscious thought. The second moderator is presentation mode. Studies that have examined the UTT often hold all product attribute information constant across participants to isolate the impact of modes of thought on maximum choice (Abadie, Villejoubert, Waroquier, & Vallee-Tourangeau, 2013). For example, all attributes are presented in an identical format using verbal or numerical attributes to characterize choice alternatives (e.g., Dijksterhuis, et al., 2004; Payne, et al., 2008; Shen, et al., 2020). However, it could of interest to examine how the presentation mode function for unconscious thought, especially the presentation of experiential information. Experiential information is never used in the UTT while individuals gain increasingly more experiential information online and offline relating to feelings and emotions (Hasford, et al., 2019).

Finally, this research is controlling for circadian asynchrony. Most people have an optimal time of day, in which they are most alert and able to perform at their best. The ‘time of day’ preference is defined as circadian preference (Folkard, 1982). In general, morning type individuals perform better in the morning and evening type individuals perform better in the evening. There is a circadian synchrony when circadian preference matches with the time at which a decision is made. When circadian preference does not match with the decision time, there is circadian asynchrony. This means that people are working outside their preferred time of day (Shen, et al., 2020). Shen, et al., (2020) emphasize that consumers' circadian asynchrony should be considered when studying the effects of unconscious thought because it can impact cognitive function and consequently the decision task. Besides, it is critical to model time-related factors since many e-commerce sites and mobile technologies allow transactions at any point in time and thus not necessarily at times coinciding with their circadian preference (Bittman, Wajcman and Brown, 2009).

The discrepancies in findings about the advantages of unconscious thought suggest that the Unconscious Thought Theory deserves further investigation and there might be potential moderators of the UTT. This results in the following research question:

(9)

9 Accounting for decision complexity and presentation mode may help to further scrutinize the advantages of unconscious thought. This research contributes to marketing and information system research because it helps to clarify the mixed |findings in prior research on unconscious thought. Moreover, it will help marketeers to decide whether to use distracting (website) features (e.g. music, banners, pop-up windows). This way, consumers will be stimulated to make maximum purchase decisions which results in more customer satisfaction and less product return.

(10)

10

LITERATURE REVIEW

In this chapter, the theoretical background of the research will be covered. First, the following concepts will be defined: mode of thought, decision complexity, presentation mode. Second, relationships between the concepts will be described and hypothesized. Finally, a conceptual model will visualize the relationships.

Mode of thought

All studies that examined the UTT distinguished two modes of thought, namely, conscious and unconscious thought. Hence, it is important to compare these modes against each other when examining the advantages of unconscious thought. Conscious and unconscious thought function in different ways in the brain (Dijksterhuis, 2004). Conscious thought refers to the cognitive processes a person is consciously aware of while attending a task. For instance, one may compare two holiday destinations and consciously think, “The French coast is beautiful, but I do not want to go there because it is way too crowded.” Therefore, conscious attention is directed at the task at hand. Unconscious thought, on the other hand, refers to cognitive processes that take place outside conscious awareness. A person may compare two holiday destinations and does not know which one to choose. Subsequently, the person does not consciously attend to the problem for a few days, and suddenly it pops into the persons mind “It’s going to be Paris!”. The thought itself is conscious, but the decision transitioning to a preference a few days later is the result of unconscious thought and happens often with the use of a distraction task. Thus, one critical point of distinction between conscious and unconscious thought lies in whether the attention is directed to the primary task or other irrelevant non-primary (distraction) tasks (Dijksterhuis & Nordgren, 2006).

Decision complexity

(11)

11 Nieuwenstein, Wierenga, Morey, Wicherts, Blom, Wagenmakers, & van Rijn, (2015) defined complex decisions as decisions with at least 30 attributes and simple decisions as decisions that involve only four attributes. Strick, et al., (2011) argued that it is difficult to pin down the exact amount of decision information required, as complexity of the problem depends on various factors not solely on the number of attributes (e.g., expertise of the participants, the amount of filler information). However, they suggested that utilizing the number of attributes across choice options seems to work well for distinguishing between simple and complex decisions.

Presentation mode

Information can be presented in many ways and studies on the UTT have used different ways to present attribute information (presentation mode). It can be presented with quantitative attribute information or qualitative attribute information (Abadie, et al., 2013). Quantitative attribute information can be characterized by numerical attributes, which are easy to quantify and more homogeneous (Payne, et al., 2008). On the other hand, qualitative attribute information is characterized by attributes that are vague and difficult to quantify or to combine in chunks (Abadie, et al., 2013). There are two different qualitative ways to present attribute information. First, qualitative information can be provided with verbal attributes. These are attributes with purely descriptive words (Brakus, Schmitt & Zhang, 2014). Second, qualitative information can be provided with experiential attributes. This refers to sensory and affective attributes presented in a nonverbal way (Brakus, et al., 2014). Experiential attributes can appear on products and packages, in logos, as part of ads, in shopping environments, or as backgrounds on web sites (Henderson, Cote, Leong, & Schmitt, 2003). Experiential attributes are frequently adopted in marketing communications, for example, by presenting a color or shape rather than naming the color; or by presenting an emoticon of a “smiley face” rather than the word “smile” (Brakus, et al., 2014).

Mode of thought and maximum choice

(12)

12 deliberation system (Epstein, Pacini, Denes-Raj & Heier, 1996; Evans, 2003; Kahneman 2003; Keren & Schul, 2009)

Traditionally, decision-making is a cognitive activity as it benefits most from rational deliberation. Accordingly, the attributes of choice alternatives need to be considered consciously through careful analysis (Dawes & Corrigan, 1974; Keeney & Raiffa, 1976). Subsequently, Wilson & Schooler (1991) indicated the need to re-evaluate the assessment of conscious deliberation. According to their research, conscious deliberation can have a disruptive effect on attitudes (Wilson & Schooler, 1991). Moreover, neuropsychological studies of decision-making criticized the traditional deliberation view. They argued that the power of the intuitive/affective mode is underestimated, as the affective process can play a substantial role in value integration when a decision eventually needs to be made (Bechara, et al., 1994; Damasio, 1994; Lieberman, 2000).

Besides the intuitive/affective mode, Smith & Blankenship (1989), introduced the idea that a period of distraction allows problem solvers to return to a problem with “fresh eyes” and forget inappropriate strategies. They called this phenomenon ‘incubation’ which refers to superior performance after a distraction period rather than working continuously on the problem. Subsequently, researchers demonstrated the ability to process information in the absence of awareness during the execution of a distracting task (Lewicki, Hill, & Czyzewska, 1992; Reber, 1992; Betsch, Plessner, Schwieren, & Gütig, 2001). The researchers revealed that distraction is not detrimental for product evaluation. For example, Betsch, et al., (2001) demonstrated that participants, who were simultaneously engaged in a distracting task were able to differentiate between good and bad shares, especially when participants were using an affective cue. In addition, the role of distraction in decision-making is further explored in different literature (Dijksterhuis, 2004; Dijksterhuis & Nordgren, 2006). These researches argue that the unconscious process during distraction is responsible for enhanced performance in decision-making. An important explanation for enhanced performance is the underlying bottom-up mechanism of unconscious thought.

(13)

13 expectancies, whereas the bottom-up mechanism is based on the available information. Hence, there is no bias prior to examining the information and it can integrate every piece of information in a more polarized and organized way (Gao, et al., 2012). Recently, Li, Li, Zhang, Shi and He (2020) elaborate on the bottom-up mechanism of unconscious thought and argue that unconscious thinkers can detect relational similarities because of the bottom-up process. This means that they can detect higher-order relations, which are important in problem solving and decision making. To conclude, information can be organized, weighted and integrated in an optimal fashion during unconscious information processing, which will increase the likelihood of choosing the maximum choice option. This results in the following hypothesis: H1: Unconscious thought will positively affect maximum choice

The moderating effect of decision complexity

When consumers make a purchase decision, they proceed through three decision-making steps. First, the intelligence phase, concerning recognizing and gathering information. Second, the design phase, concerned with integrating and structuring the information and constituting criteria of the information. Finally, consumers choose the best alternative that meets their criteria during the choice phase (Simon, 1995). Consumers effectively process the information in order to integrate their needs with perceived information in the design and choice phase. However, Simon, (1995) examined that consumers can have a capacity limitation when they effectively process information in decision-making. Therefore, effective information processing depends on the amount of information a choice contains.

(14)

14 correlates positively with the amount of information up to a certain point (Eppler & Mengis, 2008).

According to Dijksterhuis (2004), unconscious thought, due to its unconscious processes, has a virtually infinite processing capacity. The virtually infinite capacity allows consumers to process large amount of attribute information. Hence, the advantage of unconscious thought is argued to be strongest when a decision is complex because unconscious thought does not suffer from capacity limitations when large amount of information needs to be processed. Thereby, consumers will still be able to choose the maximum choice. This results in the following hypothesis:

H2: The positive relation between unconscious thought and maximum choice is moderated by decision complexity and will be amplified by complex decision making.

The moderating effect of presentation mode

(15)

15 can have the strongest impact on unconscious thought because pictorial information usually leads to more holistic processing than solely verbal information. However, they do not mention whether solely pictorial information leads to holistically processing and thus an unconscious thought advantage.

As mentioned, unconscious thought consists of bottom-up mechanisms during the distraction task. Besides these mechanisms, other specific information processing methods exist for unconscious thought, such as holistic processing (Usher, et al., 2011). According to Lerouge, (2009) and Usher, et al., (2011) the advantage of unconscious thought can only emerge when consumers holistically process the information during the distraction task. Holistic processing occurs when consumers process attribute information into coherent mental (holistic) representations. Consequently, they form global impressions of each choice alternative (Li, Wang, Shen & Fan, 2017). Furthermore, Dijksterhuis & Strick (2016) stated that unconscious thought operates on holistic representations of options rather than on recollection of individual attributes.

Unconscious thought has a holistic and associative nature. Therefore, it performs better when the information induces holistic processing (Usher, et al., 2011). According to Strick, et al., (2011) pictorial information can induce holistic processing. Experiential information consists of affective and sensory pictorial information (Brakus, et al, 2014). Therefore, it is likely that experiential information can impact unconscious thought and increases the likelihood of choosing the maximum choice option. This results in the following hypothesis:

H3: The positive relation between unconscious thought and maximum choice is moderated by presentation mode and will be amplified by experiential information.

The control variable circadian asynchrony

(16)

16 memory recall when there is circadian synchrony. Circadian asynchrony, on the other hand, leads to a constraint in an individuals’ cognitive efficiency. The arousal level is lower, which results in more dependence on the capacity of working memory. Eventually, this can affect the decision task (Shen, et al., 2020). Based on recent literature of Shen, et al., (2020), the present research wants to control for the possible effect of circadian asynchrony.

Conceptual model

A conceptual model has been developed in alignment with literature that has been provided in the previous paragraphs (figure 1). The three independent variables, dependent variable and control variable are visualized in the model. Besides the variables, the model visualizes the relationship between unconscious thought and maximum choice and the moderating role of decision complexity and presentation mode.

(17)

17

METHODOLOGY

In this chapter, the research design and methods will be explained. First, the experimental methods will be outlined. This includes the experimental material, the procedure, and the manipulations of the variables. After that, the analysis plan will be explained, followed by test assumptions.

Data collection

The experimental data for this research is collected amongst 173 respondents to the questionnaire between the 18th_{of November until the 2}nd_{of December. The questionnaire,}

created with Qualtrics, was shared on various social media platforms, such as WhatsApp, LinkedIn and Facebook (appendix A). Participants were asked to share the questionnaire with their contacts to increase the sample size. This is a form of convenience sampling, which allows to select all respondents who are at the right place at the right time (Malhotra, 2010). So, everyone could participate in the questionnaire including relatives, friends and colleagues. This was most appropriate because of willingness of the participants to participate.

Research design

(18)

18 Table 1: Overview of the conditions

Condition Decision

complexity Mode of thought Presentation mode

1 Complex Conscious Experiential

2 Simple Conscious Experiential

3 Complex Conscious Verbal

4 Simple Conscious Verbal

5 Complex Unconscious Experiential

6 Simple Unconscious Experiential

7 Complex Unconscious Verbal

8 Simple Unconscious Verbal

Material

The experiment contains four different hotels described on four different website pages of Vakantieveilingen.nl. Hotels are service providers and therefore characterized by high involvement (Hochgraefe, Faulk & Vieregge, 2012). High involved customers engage in extensive pre-purchase information search before they make a purchase decision (Schiffman & Kanuk 2004; Mathwick & Rigdon 2004). Hence, it is more likely that participants will take effort to scrutinize the hotel attribute information. The experiment incorporated hypothetical hotels, labeled A, B, C and D, to control for possible confounding effects of brand preference. Pre-existing brand preference can bias participants, which influences the attitude formation towards the brand (Mau, Silberer & Constien, 2008).

Each hotel alternative shows certain attributes. The attributes either possesses positive or negative values. Each hotel has the same set of attributes, but different attribute values. Also, the sequence of the attributes is randomized for each hotel to avoid possible order effects (Eisenberg & Barry, 1988).

(19)

19 Procedure

The sequence of the questions in the questionnaire is based on theory of Krosnick (2018). He describes the ideal sequence of questions and argues that first questions should be related to the research subject while demographic questions should be at the end. For this reason, the questions at the beginning of the questionnaire are designed to influence the willingness of the participants to respond to the questionnaire (Krosnick, 2018).

The participants are informed about the procedure and the general subject of the study. After the short introduction, participants are instructed that during the experiment, they would be asked to evaluate and compare four hotel alternatives in Amsterdam. Subsequently, participants are randomly assigned to one of the eight experimental conditions. Participants get 12 seconds to evaluate the positive and negative hotel attributes of each alternative. After 12 seconds participants are automatically directed to the next hotel alternative and cannot proceed to the previous alternative. Next, the dependent variable is measured. Participants are asked to choose one of the four hotels for their family or friend. After completing the choice task, participants answer questions about which attributes they base their choice upon. This is followed by questions regarding circadian preference. In the last questions, participants are asked to fill in their demographic details. The questionnaire ends with a short debrief, and the participants are thanked for their participation.

Manipulation mode of thought

This research manipulates mode of thought by shifting participants attention between the primary task and non-primary task (Dijksterhuis & Nordgren, 2006). Therefore, the n-back distraction task is used to shift participants’ attention. It aims to limit conscious thought and has been proven effective in evoking unconscious thought in decision making literature. It triggers unconscious thought because the n-back task is irrelevant for the hotel choice and thereby distracts the participants (Jonides, Schumacher, Smith, Lauber, Awh, Minoshima & Koeppe, 1997).

(20)

20 matches the one presented 3 steps earlier in the sequence (Jonides, et al., 1997). After the 3-back task, participants are redirected to the Vakantieveilingen.nl website and are asked to indicate their maximum choice. Participants in the conscious thought condition do not perform the 3-back task. Instead, they are instructed to carefully think about the hotel attribute information as well as their decision for 2 min, without any distraction or access to the hotel information again. Participants can advance before the time expired. Conscious thought is self-paced to avoid forcing participants to think longer than they want (Payne, et al., 2008).

Manipulation decision complexity

In order to manipulate decision complexity, the number of attributes varies between choice tasks. Simple tasks involve 4 alternatives with each 4 attributes. Complex tasks involve 4 alternatives with each 12 attributes. The manipulation is based on the theory of Miller (1956) who argues that information processing capacity is limited to a temporary “store” of seven items. Also, other studies that tried to define decision complexity are considered in this manipulation (Dijksterhuis & Nordgren 2006; Strick, et al., 2012; Nieuwenstein, et al., 2015).

Manipulation presentation mode

The presentation mode is manipulated by the material of the attribute information. According to Brakus, et al., (2014) there are different experiential attribute types and dimensions. They highlight sensory and affective types, which can be characterized by vivid pictures (Brakus, et al., 2014). Therefore, in the experiential information condition, the attributes are presented with vivid pictures (appendix B). Each attribute is described with an appropriate picture for that specific attribute and marked with a check mark when the attribute is positive and with a x mark when the attribute is negative. According to Yoon & Vargas (2018) people associate a check mark with good and an x mark with bad. In the verbal information condition, the hotel attributes are described with words. The experiential and verbal information conditions both adopt the same set of positive and negative attributes to describe each alternative. Also, the sequence, in which the attributes are presented, is equal for both conditions.

Maximum choice measure

(21)

21 when it is characterized by more positive attributes and fewer negative attributes (Abadie, et al., 2013). This is evaluated from a normative perspective (Dijksterhuis, 2004). It is corresponding with maximizing the overall probability of winning something in decision-making literature (Payne, et al., 2008). When a participant chooses the hotel with the most positive attributes the choice is coded as “1” and when a participant chooses one of the three other hotels the choice is coded as “0”.

In addition, participants are asked to choose the hotel as a gift for their family or friend. This is critical to mention because it justifies the decision-making process based on an objective standard rather than individuals own preferences or whims and fancies. Thus, it helps to overcome the potential influence of personal preferences (Shen, et al., 2020).

As a result, a hotel is identified as the maximum choice in simple decision making, when it is possessed with 3 positive attributes (75%). Two alternatives are identified as mediocre options, with each 2 positive attributes (50%). One hotel is marked as the poor option, with only 1 positive attribute (25%). Hotel C possesses the most positive attributes in all cases (table 2).

Table 2: Overview of the hotel attributes in simple decision making

A B C D

Breakfast included Yes (+) No (-) No (-) No (-)

Parking No (-) No (-) Yes (+) Yes (+)

Bar No (-) Yes (+) Yes (+) No (-)

Pool Yes (+) No (-) Yes (+) Yes (+)

Number of positive attributes

2 1 3 2

(22)

22 Table 3: Overview of the hotel attributes in complex decision making

A B C D

Breakfast included Yes (+) No (-) No (-) No (-)

Rooftop Yes (+) Yes (+) No (-) No (-)

Parking No (-) No (-) Yes (+) Yes (+)

Room service No (-) No (-) Yes (+) Yes (+)

Sauna No (-) No (-) Yes (+) Yes (+)

Bar No (-) Yes (+) Yes (+) No (-)

Tea/coffee facility No (-) Yes (+) Yes (+) No (-)

Gym Yes (+) No (-) No (-) Yes (+)

Pool Yes (+) No (-) Yes (+) Yes (+)

Jacuzzi Yes (+) No (-) Yes (+) No (-)

Garden No (-) No (-) Yes (+) Yes (+)

Patio Yes (+) Yes (+) No (-) No (-)

Number of positive attributes

6 4 8 6

Circadian preference classification

To control for circadian asynchrony, participants answer questions from Horne and Ostberg’s Morningness-Eveningness Questionnaire (MEQ) (Horne and Östberg, 1977). Horne and Ostberg have developed a questionnaire consisting of 19 items to assess the degree to which participants are active and alert at certain times of day. When participants answer all the questions in the MEQ, they obtain a total score and can be classified within each of the scale’s five diagnostic categories. The experiment uses a reduced scale of 5 items to classify participants in one of the five categories. Each category identifies a preference group (table 4). Adan and Almirall (1991) have proven that this reduced scale is sufficiently sensitive with respect to the results of Horne and Ostberg’s (1976) MEQ as to classify participants into the five preference groups (Adan and Almirall, 1991).

Table 4: Preference groups based on the range of the direct total score

(23)

23 In addition, each participant has to fill in the time on their screen, which refers to the experiment time. Morning time ranges from 6am-12pm and evening time ranges from 17pm-22pm (Meurisch, Hussain, Gogel, Schmidt, Schweizer & Mühlhäuser, 2015). There is an asynchrony effect when a type participant completes the experiment in the evening and a morning-type participant completes the experiment in the evening. These participants are coded as “1”. All the other participants do not have an asynchrony effect and are therefore coded as “0”.

Attention, manipulation and classification check

First, an attention check assesses whether participants carefully read the study instructions and base their choice on the positive and negative attributes that are presented for each hotel. Participants are asked if they have based their choice on the hotel with the most positive attributes (no = 0, yes = 1).

Another check question checks whether the experiential information manipulation induces holistic processing. After participants make their decision, they are asked to indicate whether their choice is based on a global impression or on only one or two specific attributes (global = 0, specific =1).

Finally, to ensure that participants are correctly classified as circadian asynchrony, a classification check question is included (Shen, et al, 2020). Participants are asked at the end of the study if they have completed the questionnaire at their favorite part of the day (no = 0, yes =1).

Plan of analysis

(24)

24 Cross table chi-square tests (𝑋!_{) are conducted because the dependent variable is measured at} a nominal level. Moreover, group differences can be analyzed with a cross table chi-square test. The calculations required to compute the chi-square provide considerable information about how each of the groups perform in the experiment. The chi-square is a significance statistic and should be followed with a strength statistic (McHugh, 2013). Therefore, the Cramer’s V is used to test the data when a significant chi-square result is obtained. A limitation of the Cramer’s V is that it cannot quantify the relationship between variables, even for highly significant results.

A binary logistic regression model can quantify the relationship between independent variables and the dependent variable by predicting certain variables. Also, a binary logistic regression model, allows for additional moderator variables, by including interaction effects (Allison, 2012). The dependent variable in a binary logistic model takes the value “0” or “1”. Parameters of the binary regression model are estimated by means of Maximum Likelihood Estimation methods, which focuses on probabilities instead of observed values. Instead of linear regression and general linear models, which are based on ordinary least squares algorithms, a binary logistic regression model is based on cumulative logistic distribution. This means that the variables need to be exponentiated and do not require to be linear related. The cumulative logistic distribution transforms the model such that the probabilities follow an s-shape curve so that the independent variables are linearly related to log odds (Allison, 2012).

Non- parametric tests assumptions

The chi-square test and binary logistic regression model are both non- parametric tests. This means that the distribution of the dependent variable is robust and does not require normal distribution. Therefore, both tests use a distribution free statistic and do not require equality of variances among the experiment groups or homoscedasticity in the data. To be able to

perform chi-square tests and a logistic regression model, several assumptions have to be met (McHugh, 2013; Schreiber-Gregory, 2018).

(25)

25 - The chi-square test and logistic regression model require observations to be

independent of each other. In other words, the observations should not come from repeated measurements or matched data. This assumption is met because participants are randomly assigned to one of the eight conditions.

- The chi-square test and logistic regression model require no (or little) multicollinearity among the independent variables. This means that the independent variables should not be too highly correlated with each other. When there is high correlation between independent variables, it becomes difficult to verify which independent variables explain the dependent variable (Alin, 2010). The variance inflation factor (VIF) can determine whether multicollinearity is present. The VIF test estimates the extent to which the variance of one independent variable is inflated with another independent variable (O’brien, 2007). The results of the VIF test are presented in the next chapter. - For the chi-square test, the expected observations should be 5 or more in at least 80%

(26)

26

RESULTS

This chapter will start with descriptive statistics of the sample, followed by the results of the preliminary checks. Furthermore, the results of the logistic regression and the corresponding chi-square test will be given. Finally, each individual hypothesis test result is presented.

Descriptive statistics of the sample

Based on the data, participants who finished the total questionnaire, the sample size is 173. The attention check checked whether participants carefully read the experiment instructions and whether their choice was based on the attribute information that was provided. From this check, it can be concluded that 50 participants did not follow the experiment instructions as intended. This threatened the internal validity of the research results, which means that the confidence of the cause – and - effect will be lower (Wilson, Aronson and Carlsmith, 2010). Therefore, these participants were excluded from the sample, resulting in a sample size of 123 used in the analysis. Table 5 presents for each condition the number of participants (n). The assumption about the sample size is met, and therefore the logistic regressions can be done.

Table 5: Number of participants (n) in each condition Condition Decision complexity Mode of thought Presentation mode n

1 Complex Conscious Experiential 15

2 Simple Conscious Experiential 23

3 Complex Conscious Verbal 18

4 Simple Conscious Verbal 17

5 Complex Unconscious Experiential 10

6 Simple Unconscious Experiential 18

7 Complex Unconscious Verbal 10

8 Simple Unconscious Verbal 12

Of the participants, 35.8% (n=44) were men and 64.2% (n=79) were women. The age of the participants was between 19 to 74 years old (M=34.22 SD=14.68). The majority of the participants, 86.2%, had advanced educational levels: HBO-degree or higher. Thereby, it can be concluded that the sample consisted of relatively young and well-educated people,

(27)

27 Preliminary checks

A cross table chi-square test checked if participants in the experiential unconscious conditions rely more on global impressions. In the experiential unconscious conditions, 75% of the participants indicated they made a global judgment. This percentage was lower for participants in the verbal unconscious conditions (68.9%). However, the percentages are not significantly different, 𝑋!_{(1, 𝑁 = 50) = 0.268 𝑝 = 0.594. These findings are important when interpreting} the effects of experiential information.

Finally, the control variable circadian asynchrony was computed. First, the total preference group score was obtained for each participant. The scores showed that a substantial number of participants end up in the ‘neither’ preference group (table 6). Consequently, this results in only a few participants who could be classified as circadian asynchrony. Therefore, participants with a group score of 12 and a group score of 17 were considered to be classified into a morning/evening group. This changed the number of participants in each preference group considerably to further compute the circadian asynchrony variable (table 6). Finally, 24 participants were classified as circadian asynchrony.

Table 6: Number of participants (n) classified in each preference group

No split in neither type Split in neither type

Type Score n Type Score n

Definitely Morning 22-25 0 Definitely Morning 22-25 0

Moderate Morning 18-21 15 Moderate Morning 17-21 23

Neither 12-17 91 Neither 13-16 72

Moderately Evening 8-11 17 Moderately Evening 8-12 28

Definitely Evening 4-7 0 Definitely Evening 4-7 0

(28)

28 Multicollinearity

One of the assumptions of non-parametric tests is that there should be little or no multicollinearity among the independent variables. According to O’Brien (2007) a VIF score of 4 is the threshold value for multicollinearity. Table 7 shows that all the VIF scores do not exceed the value of 4, therefore multicollinearity is not a problem.

Table 7: VIF scores of the independent variable

Variable VIF

Mode of thought simple 1,000 Circadian preference simple 1,000 Presentation mode simple 1,000 Mode of thought complex 1,001 Circadian preference complex 1,001 Presentation mode complex 1,001

Logistic regression

The binary logistic regression was conducted to define the relation between the independent and dependent variables. The structure of the binary logistic regression model is showed with the equation, which ensures that the value of the predictor is between 0 and 1 (Allison, 2012):

𝑃𝑃" = _$_!"##

𝑃𝑃_" = the probability that a participant chooses the maximum choice

𝑍_" = combination of the three independent variables and their interaction effects

𝑍" = 𝛽𝑜 + 𝛽1𝑀𝑇" + 𝛽2𝐷𝐶" + 𝛽3𝑃𝑀" + 𝛽4(𝑀𝑇 ∗ 𝐷𝐶)" + 𝛽5 (𝑀𝑇 ∗ 𝑃𝑀)"+ 𝛽 control + 𝜀

𝑀𝑇_" = Count variable for mode of thought in decision i; 𝐷𝐶_"= Count variable for decision complexity in decision i; 𝑃𝑀"= Count variable for presentation mode in decision i;

(29)

29 (𝑀𝑇 ∗ 𝑃𝑀)_" = The interaction variable for mode of thought and presentation mode in decision i;

Control = Control variable circadian asynchrony ε = The error term

A logistic regression model is non-linear and follows a cumulative distribution function. The coefficients in the model only show if there is a positive or negative relationship between the independent and the depend variable. A positive (negative) and significant coefficient leads to an increase (decrease) in the probability of a consumer chooses the maximum choice option. Table 8 shows all the independent variables in the model are nominal and represent a group that is either ‘0’ or ‘1’. This means that the positive or negative value for each group in the independent variable, can also be interpreted from the model. The following interpretations can be drawn when the independent variable represents two groups:

- A positive and significant coefficient leads to an increase in the probability of choosing the maximum choice option for each ‘1’ group of the independent variable.

- A positive and significant coefficient leads to a decrease in the probability of choosing the maximum choice option for each ‘0’ group of the independent variable.

- A negative and significant coefficient leads to an increase in the probability of choosing the maximum choice option for each ‘0’group of the independent variable.

- A negative and significant coefficient leads to a decrease in the probability of choosing the maximum choice option for each ‘1’group of the independent variable.

Table 8: The groups of the independent variables

Independent variable Group 0 Group 1

Mode of thought Conscious Unconscious

Decision complexity Simple Complex

Presentation mode Verbal Experiential

(30)

30 coefficient will lead to an odd ratio lower than 1. Table 9 shows the log odds and coefficients of each independent variable.

Table 9: results logistic regression

Variable Coefficient St. error P-value Log odds

Constant 0.144 0.215 0.504 1.155 Mode of thought 0.517 0.196 0.008** 1.677 Decision complexity -0.084 0.198 0.671 0.919 Presentation mode -0.166 0.195 0.393 0.847 Mode of thought x decision complexity 0.473 0.204 0.020* 1.605 Mode of thought x presentation mode -0.076 0.200 0.706 0.927 Control: circadian asynchrony -0.285 0.493 0.504 0.752 Significant codes: 0.01**, 0.05*

Based on the results in table 9, the following conclusions can be drawn:

- Mode of thought is significant (β = 0.517; p < 0.01). The odds ratio shows that for one unit increase in unconscious thought, the probability that a consumer chooses the maximum choice option, increases by 1.677.

- Decision complexity is not significant (p = 0.671). The variable does not affect the probability of choosing the maximum choice option.

- Presentation mode is not significant (p = 0.393). The variable does not affect the probability of choosing the maximum choice option.

- The interaction of mode of thought and decision complexity is significant (β = 0.473; p < 0.05). However, interaction effects in nonlinear models are hard to interpret because the odd ratio of the interaction effect should be interpreted as the ratio by which the odds ratio changes (Buis, et al., 2010). This means that when a consumer is thinking unconsciously after a complex decision, the odds-ratio of unconscious thought (1.677) increases by a factor of 1.605. Hence, decision complexity does positively affect the effectiveness of mode of thought.

(31)

31 - The control variable circadian asynchrony is not significant (p = 0.504). The variable

does not affect the probability of choosing the maximum choice option.

Cross table chi-square tests

To further analyze the differences between the conditions, chi-square tests were conducted. Firstly, differences between the conscious (condition: 1, 2, 3 & 4) and the unconscious (condition: 5, 6, 7 & 8) conditions were compared. As illustrated in Figure 2, participants in the unconscious condition were more likely to choose the maximum choice option than participants in the conscious condition. The analysis revealed that the difference was significant (χ!_{(1, N = 123) = 6.585, p < 0.05, φ = 0.231).}

Figure 2: Percentage of participants with the maximum choice

Secondly, the difference between the complex unconscious thought (condition: 5 & 7) and the simple unconscious thought (condition: 6 & 8) conditions was compared. As illustrated in Figure 3, participants in the complex unconscious thought condition were more likely to choose the maximum choice option than participants in the simple unconscious thought condition, as a significantly higher proportion of them chose the maximum choice option (χ!_{(1, N = 50) =} 11.448 p < 0.05, φ = 0.465). 0% 10% 20% 30% 40% 50% 60% 70%

Conscious thought Unconscious thought

(32)

32 Figure 3: Percentage of unconscious thought participants with the maximum choice

Finally, the difference between the experiential unconscious thought (condition: 5 & 6) and the verbal unconscious thought (condition: 7 & 8) conditions was compared. As illustrated in Figure 4, participants in the experiential unconscious thought condition did not choose the maximum choice option more often compared to the verbal unconscious thought condition. The analysis confirmed that the difference was not significant (χ!_{(1, N = 50) = 0.792 p = 0.373).}

Figure 4: Percentage of unconscious thought participants with the maximum choice Support for the hypotheses

Based on the logistic regression and the chi-square tests, hypotheses 1 & 2 are accepted and hypotheses 3 is rejected. Table 10 provides an overview of the hypotheses and whether these

0% 10% 20% 30% 40% 50% 60% 70% 80% Simple Complex

Percentage of participants with the maximum choice

0% 10% 20% 30% 40% 50% 60% 70% 80% Verbal Experiential

(33)

33 are rejected or accepted. First, H1 is accepted because the logistic regression shows a positive and significant relationship between unconscious thought and maximum choice. This was confirmed with the chi-square test which also showed a significant difference between conscious and unconscious thought.

Additionally, H2 is accepted. The logistic regression shows a significant interaction of mode of thought and decision complexity on maximum choice. The chi-square test confirmed that this effect was due to the complex decision conditions. Therefore, complex decisions amplify the relationship between unconscious thought and maximum choice.

Lastly, H3 is rejected. The logistic regression did not show a significant interaction of mode of thought and presentation mode on maximum choice. The chi-square test confirmed that there was no difference between conscious and unconscious thought in the experiential information conditions. Therefore, experiential information does not amplify the relationship between unconscious thought and maximum choice.

Table 10: overview of the hypotheses

Hypothesis Findings

H1: Unconscious thought will positively affect maximum choice accepted H2: The positive relation between unconscious thought and maximum choice is

moderated by decision complexity and will be amplified by complex decision making.

accepted

H3: The positive relation between unconscious thought and maximum choice is moderated by presentation mode and will be amplified by experiential information.

(34)

34

DISCUSSION AND CONCLUSION

First, the conclusion of the results in the previous chapter will be presented. Second, the results will be discussed and compared to the literature mentioned in the introduction and literature review. Subsequently, theoretical and practical implications of the results will be presented. Finally, limitations and suggestions for future research will be discussed.

Conclusion

This research examined the effect of unconscious thought on choosing the maximum choice option, with the moderating effect of decision complexity and presentation mode. Based on the literature about unconscious thought, the expectations were that unconscious thought positively affects maximum choice, especially when the decision is complex and provided with experiential information. The expectations were based on the bottom-up mechanisms, the capacity principle and the holistic processing theory. Furthermore, the control variable circadian asynchrony was taken into account, because recent literature suggests that this could have an impact on cognitive function when choosing the maximum choice option. The findings did support two of the three hypotheses. Furthermore, controlling for circadian asynchrony did not have a significant impact on choosing the maximum choice option. Next, the results will be discussed more in depth.

General discussion

(35)

35 in an optimal way. As a result, this increased the likelihood of choosing the maximum choice option.

Hypothesis 2 indicates that the positive relation between unconscious thought and maximum choice is moderated by decision complexity and will be amplified by complex decision making. The research findings support hypothesis 2. This means that unconscious thought is conducive in complex decision making. Decision complexity was defined in terms of the number of attributes for each choice alternative. According to the capacity theory of Miller (1956), 12 attributes are more difficult to process than 4 attributes. However, the present research shows that this theory does not hold for unconscious thought. This is in line with the unconscious thought capacity principle, which states that unconscious thought has a virtually infinite processing capacity (Dijksterhuis & Nordgren, 2006). Therefore, the advantage of unconscious thought becomes more salient when the decision contains 12 attributes instead of 4 attributes.

(36)

36 processing the information holistically and could not focus on forming an overall global impression of the alternatives.

Theoretical and practical implications

This research provides theoretical and practical implications. Theoretically, this research contributes to the ongoing debate regarding the merits of unconscious thought. For example, Strick, et al., (2011) identified a reliable unconscious thought effect across a variety of domains, while Nieuwenstein, et al., (2015) attribute the unconscious thought effect to various methodological problems within the unconscious thought experimental paradigm. The present research replicates the Unconscious Thought Theory and strengthens the core assumptions of the theory. Furthermore, it identifies critical new factors underlying the effectiveness of unconscious thought in decision making.

Practically, this research provides implications for managers regarding the choice environment of consumers. According to Tsiros & Mittal (2000), choosing the maximum choice option leads to consumer satisfaction and less product return. So, it is important for managers to ensure that consumers choose the maximum choice option. Managers are able to provide product information containing many attributes while utilizing distracting (website) features, such as pop-ups and banners in an online retail environment and displays and music in an offline retail environment. The distraction induces unconscious thought thereby increasing the likelihood of choosing the maximum choice option.

(37)

37 Limitations and future research

There are several limitations to this research. Firstly, the maximum choice was defined normatively as having the greatest number of positive attributes. Newell, et al., (2009) pointed out, defining maximum choice normatively entails the assumption that participants assign a subjective utility value of 1 to all positive attributes and a subjective utility of 0 to all negative attributes. This assumption is questionable in the present research as 50 participants indicated they did not choose the hotel with the most positive attributes. The importance of certain attributes differs per person. This is because people have idiosyncratic preferences. To overcome this problem, the present research adopts the procedure from Shen, et al., (2020) by asking participants to choose the hotel as a gift for their family or friend. However, the attention check showed that people still based their choice on their idiosyncratic preferences and not on the greatest number of positive attributes. This could have been avoided by a pre-test or the weighting principle. For example, a pre-test could have aimed at assessing which attributes people find most important. Consequently, only attributes that participants indicated as moderately important could have been employed. Another option could have been to let participants rank the set of attributes in the pre-test. Thereafter, the alternatives could have been composed by a computer program, which ensures that one of the alternatives was the maximum option for each participant. Lastly, an attribute importance task could have been used at the end of the experiment in which participants should have allocated 100 points to the attributes, such that more allocated points to a particular attribute, represented greater importance of that attribute with respect to their choice. However, a critical point of this task is whether importance allocation reflects participants’ intrinsic preferences rather than a product of their earlier choices (Calvillo & Penaloza, 2009). Future research can take the present research to a ‘higher level’ by testing the effect of these idiosyncratic preferences methods.

(38)

38 bootstrapping tests to ensure statistical power. Furthermore, the present research included a nonrepresentative sample in comparison with the Dutch population, which signifies that the present results cannot necessarily be generalized to the Dutch population. According to Lassiter, et al., (2009) the unconscious thought is subject to individual differences. Therefore, future research should utilize a better mix between educational levels and age.

This research is also limited by ecological validity. Researchers noted that the decisions made through unconscious thought are more acceptable when the tasks are more ecologically valid (Strick, et al., 2011; Dijksterhuis & Strick, 2016). The present research utilizes a relatively simple decision task by asking participants to select the best possible hotel based on the attributes provided. The Vakantieveilingen.nl environment tried to make the decision task more ecological valid, but the results can still be an artifact of setting up participants to make deliberations (Lassiter, et al., 2009). However, no systematic examination of the ecological validity of the decision tasks is conducted by previous studies (Li, et al., 2020). It is recommended that future research selects decision tasks that are realistic and can stimulate participants' intrinsic motivation to further reveal the influence of unconscious thought in decision making.

(39)

39

References

Abadie, M., Villejoubert, G., Waroquier, L., & Vallee-Tourangeau, F. (2013). The interplay between presentation material and decision mode for complex choice. preferences. Journal of Cognitive Psychology, 25(6), 682-691.

Abadie, M., Waroquier, L., & Terrier, P. (2017). The role of gist and verbatim memory in complex decision making: Explaining the unconscious-thought effect. Journal of Experimental Psychology: Learning, Memory and Cognition, 43, 694-705

Adan, A., & Almirall, H. (1991). Horne & Östberg morningness-eveningness questionnaire: A reduced scale. Personality and Individual differences, 12(3), 241-253.

Alin, A. (2010). Multicollinearity. Wiley Interdisciplinary Reviews: Computational Statistics, 2(3), 370-374.

Allison, P. D. (2012). Logistic regression using SAS: Theory and application. SAS Institute.

Baron, J. (2005). Rationality and intelligence. Cambridge University Press.

Bechara, A., Damasio, A. R., Damasio, H., & Anderson, S. W. (1994). Insensitivity to future consequences following damage to human prefrontal cortex. Cognition 50, 7–15.

Betsch, T., Plessner, H., Schwieren, C., & Gütig, R. (2001). I like it but I don’t know why: a value-account approach to implicit attitude formation. Pers. Soc. Psychol. Bull. 27, 242–253.

Bittman, M., Brown, J. E., & Wajcman, J. (2009). The mobile phone, perpetual contact and time pressure. Work, employment and society, 23(4), 673-691.

(40)

40 Brakus, J. J., Schmitt, B. H., & Zhang, S. (2014). Experiential product attributes and preferences for new products: The role of processing fluency. Journal of Business Research, 67(11), 2291-2298

Buis, M. (2010). Interpretation of interactions in nonlinear models. Stata Journal, 10(2), 305– 308.

Chen, Y. C., Shang, R. A., & Kao, C. Y. (2009). The effects of information overload on consumers’ subjective state towards buying decision in the internet shopping environment. Electronic Commerce Research and Applications, 8(1), 48-58.

Damasio, A. R. (1994). Descartes’ Error: Emotion, Reason, and the Human Brain. New York: Putnam.

Dang-Longani, K. (2018). Impact and effectiveness of advertising through pop ups on internet on young working women’s buying behavior-a scale validation study in Pune city. International Journal, 1(5), 53-60

Dawes, R. M., & Corrigan, B. (1974). Linear models in decision making. Psychol. Bull. 81, 95–106.

Dijksterhuis, D. (2004). Think different: The merits of unconscious thought in preference development and decision making. Journal of Personality and Social Psychology, 87 (5) (2004), pp. 586-598

Dijksterhuis, A., & Nordgren, L. F. (2006). A theory of unconscious thought. Perspectives on Psychological science, 1(2), 95-109.

(41)

41 Epstein, S., Pacini, R., Denes-Raj, V, & Heier, H. (1996). Individual differences in intuitiveexperiential and analytical-rational thinking styles. Journal of Personality and Social Psychology, 71(2), 390-405.

Eppler, M. J., & Mengis, J. (2008). The concept of information overload-a review of literature from organization science, accounting, marketing, mis, and related disciplines (2004). In Kommunikationsmanagement im Wandel (pp. 271-305). Gabler.

Evans, J. S. B. T. (2003). In two minds: dual process accounts of reasoning. Trends Cogn. Sci. 7, 454–459.

Evans, J. S. B. (2008). Dual-processing accounts of reasoning, judgment, and social cognition. Annu. Rev. Psychol., 59, 255-278.

Eysenck, M. W. (1976). Arousal, learning, and memory. Psychological bulletin, 83(3), 389.

Folkard, S. (1982). Circadian rhythms and human memory. Rhythmic aspects of behavior, 241-272.

Gao, J., Zhang, C., Wang, K., & Ba, S. (2012). Understanding online purchase decision making: The effects of unconscious thought, information quality, and information quantity. Decision Support Systems, 53(4), 772-781.

Ham, J., van den Bos, K., & Van Doorn, E. A. (2009). Lady Justice thinks unconsciously: Unconscious thought can lead to more accurate justice judgments. Social Cognition, 27(4), 509-521.

(42)

42 Henderson, P. W., Cote, J. A., Leong, S. M., & Schmitt, B. (2003). Building strong brands in Asia: Selecting the visual components of image to maximize brand strength. International Journal of Research in Marketing, 20(4), 297-313.

Hochgraefe, C., Faulk, S., & Vieregge, M. (2012). Links between Swiss hotel guests' product involvement and brand loyalty. Journal of Hospitality Marketing & Management, 21(1), 20-39.

Horne, J. A., & Östberg, O. (1977). Individual differences in human circadian rhythms. Biological psychology, 5(3), 179-190.

Hu, F., Yu, X., Chu, H., Zhao, L., Jude, U., & Jiang, T. (2018). Similar effects for resting state and unconscious thought: both solve multi-attribute choices better than conscious thought. Frontiers in psychology, 9, 1360.

Jonides, J., Schumacher, E. H., Smith, E. E., Lauber, E. J., Awh, E., Minoshima, S., & Koeppe, R. A. (1997). Verbal working memory load affects regional brain activation as measured by PET. Journal of cognitive neuroscience, 9(4), 462-475.

Kahneman, D. (2003). A perspective on judgment and choice: mapping bounded rationality. Am. Psychol. 58, 697–720.

Keren, G., & Schul, Y. (2009). Two is not always better than one: a critical evaluation of two-system theories. Perspect. Psychol. Sci. 4, 533–550.

Keeney, R. L., & Raiffa, H. (1976). Decisions with Multiple Objectives: Preference and Value Tradeoffs. New York: Wiley.

Kirk, R. E. (2012). Experimental design. Handbook of Psychology, Second Edition, 2.

(43)

43 Lewicki, P., Hill, T., & Czyzewska, M. (1992). Nonconscious acquisition of information. American psychologist, 47(6), 796.

Li, J., Li, X., Zhang, X., Shi, K., & He, Y. (2020). Can unconscious thought detect relational similarities?. International Journal of Psychology, 55(1), 60-66.

Li, J., Wang, F., Shen, M., & Fan, G. (2017). The tendency of unconscious thought toward global processing style. Consciousness and Cognition, 53, 14-22.

Li, S., & Sun, H. (2019). Herd Booster: Examining the Impacts of Recommendation and Product Type on Herd Behaviors.

Lieberman, M. D. (2000). Intuition: a social cognitive neuroscience approach. Psychol. Bull. 126, 109–137.

Malhotra, N.K. (2010). Marketing Research: an applied orientation (6th edition). Prentice Hall: Pearson Education Inc.

Malhotra, N. K. (1982). Information load and consumer decision making. Journal of consumer research, 8(4), 419-430.

Mathwick, C., & Rigdon, E. (2004). Play, flow, and the online search experience. Journal of consumer research, 31(2), 324-332.

Mau, G., Silberer, G., & Constien, C. (2008). Communicating brands playfully: Effects of in-game advertising for familiar and unfamiliar brands. International Journal of Advertising, 27(5), 827-851.

(44)

44 Messner, C., & Wänke, M. (2011). Unconscious information processing reduces information overload and increases product satisfaction. Journal of Consumer Psychology, 21(1), 9-13.

Meurisch, C., Hussain, T., Gogel, A., Schmidt, B., Schweizer, I., & Mühlhäuser, M. (2015). A Spatiotemporal Approach for Social Situation Recognition. In International Conference on Mobile Computing, Applications, and Services (pp. 309-316). Springer, Cham.

Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological review, 63(2), 81.

Monk, T. H., & Leng, V. C. (1982). Time of day effects in simple repetitive tasks: some possible mechanisms. Acta Psychologica, 51(3), 207-221.

Newell, B. R., Wong, K. Y., Cheung, J. C., & Rakow, T. (2009). Think, blink or sleep on it? The impact of modes of thought on complex decision making. Quarterly Journal of Experimental Psychology, 62(4), 707-732.

O’brien, R. M. (2007). A caution regarding rules of thumb for variance inflation factors. Quality & quantity, 41(5), 673-690.

Perfetti, C. A., & Lesgold, A. M. (1979). Coding and comprehension in skilled reading and implications for reading instruction. Theory and practice of early reading, 1, 57-84.

Reber, A. S. (1992). The cognitive unconscious: An evolutionary perspective. Consciousness and cognition, 1(2), 93-133.

Schiffman, L. G., & Kanuk, L. L. (1991). Consumer Behavior, 4-th Edition.

(45)

45 Shen, Y., Sun, H., Heng, C. S., & Chan, H. C. (2020). Facilitating Complex Product Choices on E-commerce Sites: An Unconscious Thought and Circadian Preference Perspective. Decision Support Systems, 137, 113365.

Simon, H. A. (1997). Models of bounded rationality: Empirically grounded economic reason (Vol. 3). MIT press.

Simon, H. A. (1955). A behavioral model of rational choice. The quarterly journal of economics, 69(1), 99-118.

Smith, S. M., & Blankenship, S. E. (1989). Incubation effects. Bulletin of the Psychonomic Society, 27(4), 311-314.

StatLine (2020, November 13). Bevolking; onderwijsniveau; geslacht, leeftijd en

migratieachtergrond. Retrieved from:

https://opendata.cbs.nl/statline/#/CBS/nl/dataset/82275NED/table?fromstatweb

Strick, M., Dijksterhuis, A., Bos, M. W., Sjoerdsma, A., Van Baaren, R. B., & Nordgren, L. F. (2011). A meta-analysis on unconscious thought effects. Social Cognition, 29(6), 738-762.

Strick, M., Dijksterhuis, A., van Baaren, R. B., & Nordgren, L. F. (2010). Unconscious-thought effects take place off-line, not on-line. Psychol. Sci.

Tan, B., Yi, C., & Chan, H. C. (2015). Research note—deliberation without attention: the latent benefits of distracting website features for online purchase decisions. Information Systems Research, 26(2), 437-455.

(46)

46 Wang, W., & Benbasat, I. (2008). Attributions of trust in decision support technologies: A study of recommendation agents for e-commerce. Journal of Management Information Systems, 24(4), 249-273.

Waroquier, L., Marchiori, D., Klein, O., & Cleeremans, A. (2009). Methodological pitfalls of the unconscious thought paradigm. Judgment and Decision Making, 4(7), 601-610.

Wilson, T. D., Aronson, E., & Carlsmith, K. (2010). The art of laboratory experimentation. InS.T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (p. 51–81).

Wilson, T. D., & Schooler, J. W. (1991). Thinking too much: introspection can reduce the quality of preferences and decisions. J. Pers. Soc. Psychol. 60, 181–192

Think or sleep on it? The advantages of unconscious thought