Master’s Thesis
The interaction between evaluative conditioning and attitudinal ambivalence on implicit attitudes towards fruit juice
Name: Yasmine Nadine Yamani Student ID number: 11806524 Graduate School of Communication
Persuasive Communication
Name of the supervisor: dr. Gert-Jan de Bruijn 1st February 2019
Abstract
The purpose of the following research paper is to identify whether an interaction between evaluative conditioning and level of attitudinal ambivalence on implicit attitudes towards fruit juice exists. The present study aims to shed light on whether individuals high in ambivalence are more affected by evaluative conditioning than individuals low in ambivalence. The study used a 2 (evaluative conditioning: positive vs. negative) x 2 (level of attitudinal ambivalence: high vs. low) between-subjects design. Additional analyses were conducted using the median value of an Implicit Association Test (IAT) to predict product choices. In total 300 participants took part in the study. They were partly recruited through Amazon Mechanical Turk and were paid for their participation (5ct). Stimuli for the evaluative conditioning procedures were selected based on a pre-test. Ambivalence was measured with an existing questionnaire. The results revealed no statistically significant interaction effect. The findings of the present study suggest that more research needs to be conducted to identify how implicit attitudes towards a complex and ambivalent product can be altered.
Introduction
Despite numerous efforts, emphasizing the advantages of healthy eating over the last decades, the average body mass index (BMI) has increased in Western countries (Palermo et al., 2016). While in many parts of the world hunger and poverty prevail, individuals in Western societies are faced with a great availability of food. Food is available in almost every
environment. This forces individuals to continuously make decisions about “what to eat, where to eat” and how often to eat (Rothman, Sheeran, & Wood, 2009, p. 4). In order to create and execute strategies to encourage healthy eating choices, a better understanding of these decisions is
important (Rothman et al., 2009).
To date, the majority of research has focused on measuring explicit attitudes through questionnaires or interviews. Participants were simply asked to express their attitudes towards a product or a brand (Hollands & Marteau, 2016). However, throughout the day, many people behave impulsively and in ways that might not correspond to their declared health goals (Rothman et al., 2009; Papies & Hamstra, 2010).
According to a growing body of research, people have two different attitudes: one that is explicit and one that is implicit. Explicit attitudes are thought to predict deliberate, controlled behavior, while implicit attitudes are believed to predict spontaneous behavior (Glashouwer, Bennik, De Jong, & Spruyt, 2018). In contrast to explicit attitudes, implicit attitudes are more automatic and based on associations in memory (Gawronski & Bodenhausen, 2006).
Implicit attitudes can be altered using the method of evaluative conditioning (EC), which is the pairing of a neutral stimulus (CS) with a positive or negative valenced stimulus (US) (De Houwer, Thomas, & Baeyens, 2001). The Implicit Association Test (IAT), a measure of implicit attitudes, is a computerized model that indirectly assesses the relative strength of automatic
associations between pairs of concepts via a classification task (Greenwald, McGhee, & Schwartz, 1998).
Previous research has successfully shown that evaluative conditioning changed implicit attitudes in different contexts, such as towards healthy food choices (Hollands, Prestwich, & Marteau, 2011) and snack foods (Lebens et al., 2011). However, no study has investigated the interaction between evaluative conditioning and ambivalence on implicit attitudes. This is problematic because prior research has demonstrated that the effect of evaluative conditioning depends on people’s existing attitudes. A study by Gibson (2008) has shown that people’s existing cognitions influence their susceptibility to evaluative conditioning. With regard to the concept of ambivalence, past research has indicated that people high in ambivalence are more open to persuasion compared to people low in ambivalence (Hodson, Maio, & Esses, 2001).
Attitudes are considered to be high ambivalent when an object is evaluated both positively and negatively at the same time. Low ambivalent attitude is defined as having either strong positive evaluation or strong negative evaluation with an object (Thompson, Zanna, & Griffin, 1995). Based on this, it is expected that people high in ambivalence will be more affected by evaluative conditioning than people low in ambivalence. Furthermore, because implicit attitudes are particularly effective in predicting automatic behavior (Glashouwer, 2018), the effect is expected to be stronger for people high in ambivalence.
Many attitudinal objects have various (negative and positive) features (Glaser, Woud, Iskander, Schmalenstroth, & Vo, 2018). Positive features of fruit juices, for instance, are that they are healthy, contain no fat and are cheaper than fresh fruits (Dennison, 1996). However, they also lose fresh juice flavors and experience color degradation (Cassano, Jiao, & Drioli, 2004). In addition they lack fiber and are usually high in sugar (Bazzano, Li, Joshipura, & Hu, 2008). Due
to the complex positioning of fruit juices, they are selected as the type of product for this research.
Analyzing whether an interaction between evaluative conditioning and attitudinal
ambivalence on the IAT exists will extend current EC research. Furthermore, it will shed light on whether individuals high in ambivalence are more affected by evaluative conditioning than individuals low in ambivalence. In terms of practical relevance, this study gives marketers insights into the effectiveness of commercial campaigns. It will explain why some people remain indifferent when they see a commercial that pairs a product with a stimulus that possesses
affective value (Schachtman, Walker, & Fowler, 2011). The following research questions are proposed:
RQ1: Is there an interaction between evaluative conditioning and attitudinal ambivalence on implicit attitudes towards fruit juice?
RQ2: Is there a three-way interaction between evaluative conditioning, level of attitudinal ambivalence and level of implicit attitudes on product choices?
Theoretical background
Until now, analyses of deliberative factors have dominated research on health-related behavior (Conner, Perugini, O'Gorman, Ayres, & Prestwich, 2007). This is reflected in various attitude-behavior theories, for instance the Theory of Planned Behavior (Ajzen, 1991), the Theory of Reasoned Action (Fishbein, 2008) or the Social Cognitive Theory (Bandura, 1999). Even though these models focus on different determinants of eating behavior, they agree on the assumption that deliberation about the effect of performing the behavior (attitudes), beliefs, views and behavior of other people (social norms), one’s capacity to perform a behavior (self-efficacy) and the desire to act (intention) are crucial determinants of behavior change (Rothman et al., 2009). However, a meta-analysis on techniques changing behavior, showed that changing these
determinants only had little results on dietary outcomes (Michie, Abraham, Whittington, McAteer, & Gupta, 2009). This suggests that behavior is not always affected by reflective/cognitive processes, but that there is an impulsive part to behavior that many models do not acknowledge (Rothman et al., 2009).
Reflective-Impulsive Model (RIM)
According to the Reflective-Impulsive Model, there are two different memory systems in the human mind: the automatic system and the reflective system. While the automatic system is always active and learns through experiences, the reflective system learns through language and logic. When cognitive capacity is available, the reflective system is activated. Depending on how much cognitive capacity is available, any given behavior will either be driven by the automatic or reflective system (Strack & Deutsch, 2004).
Implicit attitudes
Implicit attitudes are of interest in this context. They measure the relative strength of associations between a concept and its evaluation. Implicit attitudes are learned through repeated experience, are automatic and unreflective. They are assessed through indirect methods such as reaction-time based tasks (Rothman et al., 2009). Results have shown that implicit attitudes not only predict behavior (Poehlman, Uhlmann, Greenwald, & Banaji, 2005), but that they also manage to do so better than explicit attitudes (Conner, Perugini, O’Gorman, & Ayres, 2007). The most commonly used method to assess implicit attitudes is the IAT. It was introduced in scientific literature in 1998 by Greenwald, McGhee and Schwartz. The IAT measures implicit attitudes through an underlying automatic evaluation (Greenwald et al., 1998).
Evaluative conditioning
One method to change implicit attitudes is evaluative conditioning, meaning the repeated pairing of a neutral stimulus (CS) - e.g. high calorie snacks - with a positive or negative valenced
stimulus (US) - positive or negative images/words - in order to create the desired implicit attitude in a person (Rothman et al., 2009). EC-based interventions have been studied extensively in the past and have shown to be effective in changing implicit attitudes. A study by Hollands, Prestwich and Marteau (2011) has found that pairing pictures of snack foods with pictures of potential adverse health consequences changed implicit attitudes negatively towards energy-dense snacks. De Bruijn, Keer, Conner and Rhodes (2011) used an IAT to analyze the interaction effect of habit strength, implicit attitudes and fruit consumption. Results have indicated that implicit attitudes moderated the relationship between habit strength and fruit consumption; the relationship was stronger when implicit attitudes were more positive.
In EC research, interesting moderators have been examined intensively throughout the years. A study by Vogel, Hütter and Gebauer (2017) has found that individuals high in neuroticism and agreeableness were more affected by evaluative conditioning than individuals who were low in these traits. One of the most discussed and most debated moderators in evaluative conditioning research is contingency awareness; the awareness of CS–US association (Blask, Walther, Halbeisen, & Weil, 2012). While some studies have suggested that the awareness of CS-US associations is important in order to create such conditioning (Pleyers, Corneille, Luminet, & Yzerbyt, 2007), other studies have reported that the effect of evaluative conditioning only occurs in the absence of awareness (Baeyens, Eelen, & Bergh, 1990; Fulcher & Hammerl, 2001).
Attitudinal ambivalence
While a fair deal of controversy exists regarding contingency awareness, no research has been done on the effects of evaluative conditioning where attitudinal ambivalence is considered. Traditionally, attitudes have been analyzed in dichotomous terms. That is to say, attitudes are considered as being either in favor or against something (Thurstone & Chave, 1929). However,
an individual may simultaneously hold both negative and positive attitudes at the same time (Cacioppo, Gardner, & Berntson, 1997). In this context, particular attention has been given to the concept of attitudinal ambivalence (Thompson, Zanna, & Griffin, 1995). An individual high in ambivalence may, for instance, think that a chocolate cake is really delightful because it is tasty and soft (positive) but may also think that it is high in calories and sugar, which is detrimental to one’s health (negative). While studies have been done in the field of attitudinal ambivalence, there is a lack of consensus about the best way to measure ambivalence (Brecker, 1994; Priester & Petty, 1996).
According to Kaplan (1972), attitudinal ambivalence can either be measured directly or indirectly. Direct measures of attitudinal ambivalence are based on the subjective impression of one’s own attitude (Conner et al., 2002). Tourangeau, Rasinski, Bradburn, and D’Andrade (1989) have used this measure, which required individuals to show if they were against, in favor or had mixed feelings about a topic. However, it has been said that these measures tend to be more susceptible to extraneous factors, which can threaten their validity (Bassili, 1996). Indirect measures collect separate measures of the positive and negative feelings, thoughts and beliefs towards an object or a person. They are favored by researchers because they do not force participants to combine both positive and negative evaluations and are therefore less prejudiced than direct measures (Conner et al., 2002). Indirect measures have been demonstrated by Kaplan’s (1972) semantic differential measure, which tests separate positive and negative evaluations of an object. The measure requires calculating the mean of the separate positive and negative evaluations and then subtracting from them the absolute difference between the two
(Conner et al., 2002). The most widely used equation is exemplified by Thomson et al. (1995):
P(ositive) and N(egative) are measured on unipolar scales in two separate questions. The equation is also known as the Griffin calculation and is used in this study (Conner et al., 2002).
Attitudinal ambivalence has previously been investigated in evaluative conditioning. A study by Glaser, Woud, Labib, Schmalenstroth, and Vo (2018) paired positive, negative and ambivalent USs with neutral CSs. Findings have shown that EC effects were also found for the ambivalent USs. Jonas, Brömer and Diehl (1997) found stronger attitude-behavior intentions for people who are high in ambivalence compared to people who are low in ambivalence. To date, the interaction effect between attitudinal ambivalence and evaluative conditioning has not been investigated.
Firstly, for evaluative conditioning to alter implicit attitudes, it is important that both negative and positive associations exist in the memory in order for them to be activated.
Secondly, evaluative conditioning has an effect on implicit attitudes only to the extent to which a person does not already have strong positive or negative associations towards a product or a brand (Rothman et al., 2009). If a person already holds strong positive or negative associations, it will be difficult to alter these associations through a brief conditioning procedure. A study by Gibson (2008) has found that neutral people are most affected by evaluative conditioning: people who neither had a strong preference for Pepsi nor for Coca Cola reported strong EC effects. In order to categorize the neutral group, participants were asked to complete an explicit measure of preference for soft drinks. Questions were based on a 7-point differential scale (good-bad, high quality-poor quality etc.) and participants who expressed an equal preference for both products were categorized in the neutral group. The scale required participants to place their attitude towards an object on a unidimensional continuum, a scale ranging from positive to negative with neutral in the middle. However, having a neutral attitude score does not indicate whether a person is considered as high or low involved. Ambivalence is therefore a better indicator of involvement
than an attitude score. Based on literature and earlier findings the following hypothesis can be formulated:
H1: The effect of evaluative conditioning (positive and negative) on implicit attitudes towards fruit juice is stronger for people high in ambivalence compared to people low in ambivalence.
One of the key elements of the IAT is its capacity to predict consumer choice (Brunel, Tietje, & Greenwald, 2004). A study by Maison, Greenwald and Bruin (2004) for instance has measured explicit and implicit preferences for consumer goods. The results have indicated that the IAT increase the prediction of behavior, compared to using explicit measures alone. In a study by Karpinski and Hilton (2001), findings have revealed that the explicit measure predicts the choice between a snack and a fruit whereas the IAT did not. Although many studies have analyzed the relationship between the IAT and behaviors, only a few studies have examined under which circumstances individuals are more influenced by their implicit or explicit attitudes. As previously mentioned, implicit measures are particularly effective in predicting spontaneous behavior whereas explicit attitudes are assumed to predict reflective, controlled and conscious decisions (Macy, Chassin, & Presson, 2012).
The MODE Model
According to the MODE model (Motivation and Opportunity as Determinants) whether a person engages in deliberate processing or not depends on one’s motivation and opportunity to act or to think. If both motivation and opportunity are present, people are more likely to base their judgments on deliberate processing instead of automatically triggered attitudes (Fazio, 1990). However, if either the motivation or the opportunity for deliberate processing is low, people are more likely to rely on their implicit attitudes (Hahn & Gawronski, 2018).
The assumption of the MODE Model is supported by several studies. A study by Friese, Wänke and Plessner (2006) has investigated implicit consumer preferences and its direct impact on product choices. Findings have revealed that participants rely more on their implicitly preferred product than on their explicitly preferred one, when they are pressured to make a quick choice. Time constraints increase the value of implicit versus explicit preferences because it restricts the possibility of great information processing. A study by Gibson (2008) has found that implicit brand attitudes predict product choices, but only under cognitive load. These findings are in line with a study by Echabe (2013). His findings have indicated that when cognitive capacities are directed elsewhere or when time constraints forces participants to make a quick decision, implicit attitudes are better predictors of behaviors (Gibson, 2008). Based on earlier findings, a three-way interaction between evaluative conditioning, level of attitudinal ambivalence and implicit attitudes is expected. This interaction will only hold for people high in ambivalence. This leads to the following hypotheses:
H2a: There will be a three-way interaction effect between evaluative condition, level of attitudinal ambivalence and implicit attitudes on product choices (fruit juice vs. fresh fruits), such that the effect will be stronger for people high in ambivalence compared to people low in ambivalence.
H2b: There will be a three-way interaction between evaluative conditioning, level of attitudinal ambivalence and implicit attitudes on product choices (fruit juice vs. snacks), such that the effect will be stronger for people high in ambivalence compared to people low in ambivalence.
Method
Research Design
The true experiment used a 2 (evaluative conditioning: positive vs. negative) x 2 (level of attitudinal ambivalence: positive vs. negative) between-subjects design. The dependent variable for the first hypothesis was the results from an IAT. Additional analyses were conducted using the median value of the IAT (positive vs. negative implicit attitudes) to predict product choices. Procedure
The online experiment used a convenience sampling method. Participants were recruited through my social networking websites (e.g. Facebook, Instagram) and Amazon Mechanical Turk. Each participant recruited through Amazon Mechanical Turk received 5 cents for taking part in the study. Individuals recruited through social networking websites participated
voluntarily. The survey and experiment were conducted using Qualtrics. They were able to access the experiment through clicking on a link that was sent to them. After participants had declared that they had been informed in a clear manner about the nature and method of the research, they filled in some demographic questions and were asked to answer several questions regarding their knowledge about fruit juice and their attitudes towards it. While filling out the survey,
participants were either exposed to a sunny weather picture with a fruit juice (experimental condition 1) or a stormy weather picture with a fruit juice (experimental condition 2). Participants were randomly assigned to experimental conditions 1 or 2.
In order to assess implicit attitudes, participants completed an IAT. Specifically, implicit attitudes were measured with the Shiny web app offered by Iatgen (2018). Upon completing the Implicit Association Test, participants were asked to make a choice between a fruit juice and
fresh fruits (4x) and between a fruit juice and a snack (4x) in 5 seconds. The amount of time given was based on previous research on implicit/explicit attitudes and food choices (Friese, Wänke, & Plessner, 2006). After completion of the survey, participants were debriefed and had the opportunity to drop out of the study. They were thanked for their participation.
Participants
The study contained 345 participants. Of the 345 participants, 33 were removed from the final dataset due to missing values. Of the remaining 312 participants, 12 participants asked at the end of the experiment to be excluded from the analyses and were therefore removed. Analyses were thus conducted over a sample of 300 participants in total. The age range was 19-68 years old (M = 32.58, SD = 9.89). 151 of the participants were female (50.7%) and 148 were male (49%)1. On average, most participants had a Bachelor degree (54.3%).
Material
Prior to the main experiment, a pre-test was run to select the most appropriate pictorial stimuli for the two evaluative conditioning procedures. In total 8 participants, recruited through Facebook, rated the smileys and other images used on a 7-point Likert scale that ranged from 1 = 'extremely positive' to 7 = 'extremely negative'. Results of the pretest showed that participants rated the sunny weather picture as most positive and the stormy weather as most negative. Thus, those pictures were selected for the evaluative conditioning procedures. Participants of the pretest were not part of the actual experiment. Fruit juice was used as the stimulus in the experiment because of its ambivalent position. It simultaneously possesses positive features such as its nutritional value (i.e. fruit juice contains a lot of minerals, vitamins and antioxidants), and negative features such as its high sugar and calorie levels, which can have many negative health effects (Bazzano et al., 2008).
Measures
Fruit juice knowledge. Participants were provided with four statements on their attitude towards fruit juice. The statements were “Fruit juice are a healthy replacement for fresh fruits”, “Drinking fruit juice will reduce the risk of illnesses such as heart disease and high blood pressure”, “There are more fiber in fruits than in fruit juices” and “Drinking every day fruit juices will have more benefits than eating fresh fruits 2-3 per week”. In the negative conditioning, each answer scale was based on a 7-point Likert scale ranging from 1 = ‘strongly agree’ to 7 = ‘strongly disagree’. In the positive conditioning, the answer scale was based on a 7-point Likert scale ranging from 1 = ‘strongly disagree’ to 7 = ‘strongly agree’. To score the scales in the same direction so that a higher score indicated a more positive attitude towards fruit juice, opposing items were reverse scored. A principal component analysis (PCA) showed that 3 items formed a single unidimensional scale: only one component had an eigenvalue above 1 (eigenvalue 2.29). All items, except item 3 (factor loading was -.21), correlated positively with the factor. The first statement ("Fruit juice are a healthy replacement for fresh fruits”) had the strongest association (factor loading was .83). Reliability of the scale was low (Cronbach's alpha = .56). Item 3 was deleted, which increased Cronbach’s alpha to .82. Next, a new variable was computed to create the mean score of each participant regarding fruit juice knowledge (M = 4.31, SD = 1.53).
Level of attitudinal ambivalence. In order to assess positive and negative evaluations of fruit juice, unipolar scales were used. Three measures were utilized. For the first measure, two questions were taken and adapted from Thompson et al. (1995): “When you consider only the positive aspects of fruit juice and completely ignore the negative ones, how positive are these positive aspects of fruit juice for you?” (7-point Likert scale: 1 = ‘not at all positive’ to 7 = ‘extremely positive’) and “When you consider only the negative aspects of fruit juice and
completely ignore the positive ones, how negative are these negative aspects of fruit juice for you?” (7-point Likert scale: 1 = ‘not at all negative’ to 7 = ‘extremely negative’). The items for the second measure were taken from Sparks et al. (2001): “To what extent do you have negative feelings/thoughts about drinking fruit juice?” and “To what extent do you have positive
feelings/thoughts about drinking fruit juice?” (for both: 1 = ‘not at all’ and 7 = ‘to an extremely great extent’). The items for the third measure were also taken from Sparks et al. (2001): “To what extent are there advantages for you in drinking fruit juice?” and ‘To what extent are there disadvantages for you in drinking fruit juice?’ (for both: 1 = ‘not at all’ and 7 = ‘to an extremely great extent’). For each measure ambivalence was assessed using the Griffin calculation
described in the previous section. The inter-item reliability for these measures was low (Cronbach's alpha = 0.68). The first measure was deleted to increase the inter-item reliability (Cronbach's alpha = 0.73). Higher and lower ambivalence groups were constructed via a median split (Mdn = 3.50) on the ambivalence scores. Participants with a score under the median value were categorized as ‘low ambivalent’ and participants with a score above the median value were categorized as ‘high ambivalent’.
Implicit measures. Participants completed a survey-based IAT in Qualtrics comparing ‘fruit juice’ and ‘tools’ targets on ‘pleasant’ vs. ‘unpleasant’ categories. The IAT took about five minutes to complete and consisted of a series of practice blocks next to the compatible and incompatible blocks. The survey based IAT consisted of 7 survey questions. Four permutations were created in order to counterbalance left/right starting positions of targets and categories (Iatgen, 2018). Participants were randomly assigned to one of the four permutations: (1) target fruit juice started on the right with positive words, (2) target fruit juice started on the right with negative words, (3) target fruit juice started on the left with positive words and (4) target fruit juice started on the left with negative words. Block 1 (20 trials) consisted of a practice trial of
targets (fruit juice vs. tools). Block 2 (20 trials) was a practice trial of categories (pleasant vs. unpleasant). Then the first combined block started using both targets and categories, either incompatible or compatible, depending on the left/right assignments at the beginning (e.g.
‘compatible’ block: fruit juice + pleasant; tools + unpleasant). The combined block was divided into 20 practice trials (Block 3) and 40 critical trials (Block 4). After that, another practice block (Block 5; 40 trials) followed which consisted only of the categories with the sides reversed
(unpleasant vs. pleasant). Finally, participants repeated the combined block with the categories in their reversed position (e.g. ‘incompatible’ block: fruit juice + unpleasant; tools + pleasant). As before, this was split into 20 practice trials (Block 6) and 40 critical trials (Block 7) (Carpenter et al., 2018). Stimuli were photos of fruit juices and tools, taken from the website Pexels (Pexels, 2018). Error feedback was provided by displaying an “X” for 300 milliseconds (Greenwald et al., 1998). The target words used in the IAT were pleasant words (good, lovely, happiness, joy, friendly) and unpleasant words (disgust, awful, hate, sad, horrible) taken from previous research on implicit attitudes (Wiers, Van de Luitgaarden, Van den Wildenberg, & Smulders, 2005; de Liver, Van der Pligt, & Wigboldus, 2007; Greenwald et al., 1998). A D-score was then calculated for each participant based on the recommendations by Greenwald (1998). The assumption of the IAT is that if two concepts are strongly associated in memory (e.g., fruit juice + pleasant), the pairing will be easier and faster. When the category pairings are switched (e.g., fruit juice + unpleasant), the task will be slower. The extent to which an individual is faster in one section (e.g. ‘compatible’ Block) or the other (e.g. ‘incompatible’ Block) is a measure of one’s implicit attitude (Iatgen, 2018). Higher scores reflected a more positive implicit attitude towards fruit juice whereas negative scores reflected a more negative implicit attitude towards fruit juice. The internal consistency of the IAT was Cronbach's alpha = .78 and the error rate 0.07 (proportion of
trials in which erroneous response occurred). Pictures of the fruit juices and tools used in the experiment can be found in the appendix.
Product choices - fruit juice vs. fresh fruits. In the first four questions, participants were asked to choose between a fruit juice and fresh fruits. Of the four questions, two had fruit juice as the first option. First, each question was recoded into a new variable, where 0 = fresh fruits and 1 = fruit juice. Second, variables were added together to compute a new variable. The new variable had scores that ranged from 0 to 4. Higher scores implied that participants chose the fruit juice more often than the fresh fruits.
Product choices - fruit juice vs. snack. Next, participants were asked to choose between a fruit juice and a snack. Of the four questions, two had fruit juice as the first option. First each question was recoded into a new variable, where 0 = snack and 1 = fruit juice. Variables were then added together to compute a new variable. The new variable had scores that ranged from 0 to 4. Higher scores implied that participants chose the fruit juice more often than the snacks.
D-scores - groups. The sample was divided based on the median-split (Mdn = .65) in order to create two groups. Any value below the median split was categorized as having a negative
implicit attitude towards fruit juice and any value above the median split was categorized as having a positive implicit attitude towards fruit juice.
Results
Manipulation check
A cross-tabulation was conducted in order to analyze whether participants were able to recognize the right condition they were exposed to. Of the 143 participants who were in the positive condition, 68.5% people (n = 98) stated the right answer (sunny weather picture), 7.7% participants (n = 11) stated the wrong answer and 23.8% (n = 34) did not remember what image
they saw. Of the 157 participants who were in the negative condition, 52.2% (n = 82) stated the correct answer (stormy weather picture), 27.4 % (n = 43) stated the wrong answer and 20.4 % (n = 32) did not remember what image they saw. The results reveal that there was a significant association between the conditions (positive and negative) and the pictures participants were exposed to X2 (2, n = 300) = 75.30, p < .001. Hence, the manipulation was successful. Randomization checks
Gender. A cross-tabulation was conducted in order to check whether males and females were evenly distributed over the four conditions. Conditions served as the independent variable and gender as the dependent variable. The results revealed that the conditions (positive, low ambivalent; positive, high ambivalent; negative, low ambivalent; negative, high ambivalent) did not significantly differ on gender X2 (3, N = 2992) = .73, p = .866. Hence, the randomization of participants across conditions was successful for gender.
Age. The second randomization check entailed conditions as the independent variable and age as the dependent variable to see whether age was comparable over the four conditions. A one-way ANOVA showed that the respondents’ mean age in the positive, low ambivalent condition (M = 32.55, SD = 10.15) was not significantly different from the mean age in the positive, high ambivalent condition (M = 31.26, SD = 8.78) or from the mean age in the negative, low ambivalent condition (M = 33.77, SD = 10.75) or from the mean age in the negative, high ambivalent condition (M = 32.49, SD = 9.65), F (3, 294) = .81, p = .489. Hence, the
randomization of participants across conditions was successful for age.
Fruit juice knowledge. The third randomization check entailed Conditions as the independent variable and fruit juice knowledge as the dependent variable to see whether fruit
juice knowledge was comparable over the four conditions. A one-way ANOVA revealed that fruit juice knowledge
did not differ across conditions, F (3, 296) = .68, p = .564. Hence, the randomization check of participants across conditions was successful for fruit juice knowledge.
Correlation analyses
Fruit juice knowledge and IAT. A correlation analysis was conducted to assess the relationship between fruit juice knowledge and the IAT. Results showed that there was no significant correlation between fruit juice knowledge and the IAT: r = -.08, p = .193.
Fruit juice knowledge and product choices (fruit juice vs. fresh fruits). A correlation analysis was conducted to assess the relationship between fruit juice knowledge and the first choice task. Results indicated that there was a significant, moderately strong positive correlation between fruit juice knowledge and product choices: the more positive the attitude was towards fruit juice, the more often people chose fruit juice over fresh fruits, r = .36, p < .001.
Fruit juice knowledge and product choices (fruit juice vs. snacks). A correlation analysis was conducted to assess the relationship between fruit juice knowledge and the second choice task. Results showed that there was a significant, weak positive correlation between fruit juice knowledge and product choices: the more positive the attitude was towards fruit juice, the more often people chose fruit juice over snacks, r = .13, p = .027. Because significant results were found, fruit juice knowledge was included as a control variable in the analyses.
Interaction effects
Hypothesis H1. A two-way analysis of covariance (ANCOVA) was conducted to test the interaction between evaluative conditioning and level of attitudinal ambivalence on implicit attitudes while controlling for fruit juice knowledge. The ANCOVA contained evaluative
conditioning (positive vs. negative) and level of attitudinal ambivalence (high vs. low) as the independent variables, fruit juice knowledge as the control variable and implicit attitudes (d-scores) as the dependent variable. The assumptions for ANCOVA were met. The ANCOVA revealed no statistically main effect of evaluative conditioning, F (1, 295) = 1.13, p = .289, or attitudinal ambivalence, F (1, 295) = 1.91, p = .168 and no interaction between evaluative conditioning and level of attitudinal ambivalence on implicit attitudes F (1, 295) = .89, p = .346. H1 is therefore not supported (see Table 1 and Table 2 for an overview).
Table 1 - Implicit attitudes (d-scores).
EC Ambivalence N Mean Standard
Deviation
positive low ambivalent 74 .65 .38
high ambivalent 69 .63 .37
Total 143 .64 .37
negative low ambivalent 83 .64 .37
high ambivalent 74 .54 .43
Total 157 .59 .40
Total low ambivalent 157 .65 .37
high ambivalent 143 .58 .41
Total 300 .62 .39
Table 2 - Results of a two-way analysis of covariance (N = 300).
Sum of Squares df Mean Square F p n2 Fruit Juice Knowledge .26 1 .26 1.71 .192 .006 EC .17 1 .17 1.13 .289 .004 Ambivalence .29 1 .29 1.91 .168 .006 EC*Ambivalence .14 1 .14 .89 .346 .003 Error 44.59 295 .15 Total 159.58 300
Hypothesis H2a. A three-way ANCOVA was conducted to test the interaction effect between evaluative conditioning, level of attitudinal ambivalence and implicit attitudes on product choices (fruit juice vs. fresh fruits) while controlling for fruit juice knowledge. The three-way ANCOVA contained evaluative conditioning (positive vs. negative), level of attitudinal
ambivalence (high vs. low), implicit attitudes (positive vs. negative) as the independent variables, fruit juice knowledge as the control variable and product choices (fresh fruits vs. fruit juice) as the dependent variable. The assumptions for ANCOVA were met. The ANCOVA revealed no statistically main effect of evaluative conditioning, F (1, 291) = .66, p = .419, or attitudinal ambivalence, F (1, 291) = .05, p = .825 or implicit attitudes, F (1, 291) = .05, p = .825 on product choices. Furthermore, no interaction effect was found, F (1, 291) = .03, p = .858. H2a is therefore not supported (see Table 3 and Table 3 for an overview).
Table 3 - Product choices - fruit juice vs. fresh fruits
EC Ambivalence Implicit attitudes
N Mean Standard
Deviation
positive low ambivalent negative 36 2.03 1.11
positive 38 2.24 1.02
Total 74 2.14 1.06
high ambivalent negative 33 2.39 1.20
positive 36 2.22 1.22
Total 69 2.30 1.20
Total negative 69 2.20 1.16
positive 74 2.23 1.12
Total 143 2.22 1.13
negative low ambivalent negative 39 2.10 1.19
positive 44 2.25 1.24
Total 83 2.18 1.21
high ambivalent negative 42 2.21 1.02
positive 32 1.97 1.40
Total 74 2.11 1.20
Total negative 81 2.16 1.10
positive 76 2.13 1.31
Total 157 2.15 1.20
Total low ambivalent negative 75 2.07 1.14
positive 82 2.24 1.14
Total 157 2.16 1.14
high ambivalent negative 75 2.29 1.10
positive 68 2.10 1.31
Total 143 2.20 1.20
Total negative 150 2.18 1.12
positive 150 2.18 1.22
Table 4 - Results of a three-way analysis of covariance (N = 300). Sum of Squares df Mean Square F p n2
Fruit juice knowledge 48.94 1 48.94 40.13 .000 .121
EC .80 1 .80 .66 .419 .002 Ambivalence .06 1 .06 .05 .825 .000 D_scores .06 1 .06 .05 .825 .000 EC*Ambivalence .29 1 .29 .24 .627 .001 EC*D_scores .02 1 .02 .01 .912 .000 Ambivalence*D_scores .90 1 .90 .74 .391 .003 EC*Ambivalence* D_scores .04 1 .04 .03 .858 .000 Error 354.886 291 1.22 Total 1834.00 300
Hypothesis H2b. A three-way ANCOVA was conducted to test the interaction effect between evaluative conditioning, level of attitudinal ambivalence and implicit attitudes on product choices (fruit juice vs. snacks) while controlling for fruit juice knowledge. The three-way ANCOVA contained evaluative conditioning (positive vs. negative), level of attitudinal ambivalence (high vs. low), implicit attitudes (positive vs. negative) as the independent variables, fruit juice knowledge as the control variable and product choices (fruit juice vs. snacks) as the dependent variable. The assumptions for the ANCOVA were met. The ANCOVA revealed no statistically main effect of evaluative conditioning, F (1, 291) = .80, p = .371, or attitudinal ambivalence, F (1, 291) = .57, p = .450 or implicit attitudes, F (1, 291) = .00, p = .970 on product choices. Furthermore, no interaction effect was found, F (1, 291) = .03, p = .872. H2b is therefore not supported (see Table 5 and Table 6 for an overview).
Table 5 - Product choices - fruit juice vs. snacks
EC Ambivalence Implicit attitudes
N Mean Standard
Deviation
positive low ambivalent negative 36 2.97 1.08
positive 38 2.87 1.04
Total 74 2.92 1.06
high ambivalent negative 33 2.70 1.24
positive 36 2.70 1.24
Total 69 2.70 1.23
Total negative 69 2.84 1.16
positive 74 2.78 1.14
Total 143 2.81 1.14
negative low ambivalent negative 39 2.72 1.12
positive 44 2.64 1.24
Total 83 2.67 1.18
high ambivalent negative 42 2.64 1.21
positive 32 2.78 1.13
Total 74 2.70 1.17
Total negative 81 2.68 1.16
positive 76 2.70 1.19
Total 157 2.69 1.17
Total low ambivalent negative 75 2.84 1.10
positive 82 2.74 1.15
Total 157 2.79 1.13
high ambivalent negative 75 2.67 1.21
positive 68 2.74 1.18
Total 143 2.70 1.19
Total negative 150 2.75 1.16
positive 150 2.74 1.16
Total 300 2.75 1.16
Table 6 - Results of a three-way analysis of covariance (N = 300).
Sum of Squares
df Mean
Square
F p n2
Fruit juice knowledge 7.80 1 7.80 5.83 .016 .020
EC 1.08 1 1.08 .80 .371 .003 Ambivalence .77 1 .77 .57 .450 .002 D_scores .00 1 .00 .00 .970 .000 EC*Ambivalence 1.81 1 1.81 1.35 .245 .005 EC*D_scores .27 1 .27 .20 .656 .001 Ambivalence*D_scores .94 1 .94 .71 .402 .002 EC*Ambivalence* D_scores .04 1 .04 .03 .872 .000
Error 389.310 291 1.338
Total 2664.000 300
Discussion & Conclusion
The present study aimed to investigate whether an interaction between evaluative conditioning and level of attitudinal ambivalence on implicit attitudes towards fruit juice exists. In addition, this work investigated whether there is a three-way interaction between evaluative conditioning, level of attitudinal ambivalence and implicit attitudes on product choices. The results revealed no statistically significant main or interaction effect.
However, it is questionable whether final conclusions can be drawn based on the present study. Previous research has discussed under which conditions an evaluative conditioning procedure is most effective. One of the requirements is the repeated pairing of an object with a positively or negatively valenced stimulus in order to create the desired implicit attitude. In the present study, participants were conditioned through continuous exposure to either a positive or a negative stimuli while filling in a survey. The picture was placed above the questions on the right side of the screen. When participants were asked to select the picture they saw during the
conditioning procedure, almost half of the participants either did not remember what they saw or picked the wrong picture. The strength of an experimental research lies in its internal validity; i.e. the ability to establish causality by combining cause and effect through treatment manipulation while controlling for extraneous variables (Johnson & Christensen, 2008). Considering the high amount of people who failed the manipulation check, it is questionable whether internal validity in this experiment was guaranteed.
The fact that many people were not conditioned as intended, also explains the results on the IAT. It was expected that participants who were exposed to negative conditioning, would
have a negative D score compared to participants who were exposed to positive conditioning, and that the effect would be stronger for people high in ambivalence. As previously mentioned, a positive D score indicated a positive implicit attitudes towards fruit juice whereas a negative D score indicated a negative implicit attitudes towards fruit juice. A zero D score indicated no bias. Results showed that more than 80% of participants, including people high in ambivalence in the negative experimental condition, had a positive D-score on the IAT. As a result, the groups (negative vs. positive implicit attitudes) were not created ‘naturally’ but were split based on the median value.
The type of product chosen for this study can also explain the results. To date, evaluative conditioning research has mostly concentrated on using unambiguously positive or negative stimuli (Glaser et al., 2018). This means studies have used attitude objects that were clearly positive, such as fruits (e.g. Lebens et al., 2011) or clearly negative such as snacks (e.g. Hollands et al., 2011). In this experiment, the only target object used was fruit juice, a complex product which can trigger mixed feelings regarding its health benefits. The non-significant results suggest that further investigation is required to determine how to change implicit attitudes towards
complex products especially of people high in ambivalence. Thus, the study is low on external validity because the findings are not generalizable to other products. Different conclusions could have been drawn if other stimuli were used, such as fruits or snacks.
Furthermore, the study has shown that fruit juice knowledge was strongly correlated with product choices. Perhaps other control variables such as hunger/thirst or BMI could also have played a role in the experiment. For instance, if someone was very thirsty in the experiment, it is difficult to be certain that their implicit preference for fruit juice resulted from the pairing of fruit juice with a positive stimulus. Although it is possible to implement appropriate control variables
in a laboratory environment, outside of the controlled setting these controls are often hard to measure.
Evaluative conditioning is an associative learning; the impulsive system learns through repeated exposure (Baeyens, De Houwer, Vansteenwegen, & Eelen, 1998). It is possible that the conditioning procedure in this study was not long enough to create the desired associations between fruit juice and the positively or negatively valenced stimuli. A suggestion for further research in this context could be to expose participants longer to the evaluative conditioning procedure through a game that is fun and exciting. Previous research has shown that positive feelings that are triggered through a game can help shape a child’s brand associations even though he or she is not aware of it (Cohen, Pham, & Andrade, 2007). The same can be
experienced by an adult and, by playing a game, participants may be less likely to be aware of the conditioning procedure (contingency awareness). Furthermore, since this study only investigated positive and negative conditioning procedures, it would be interesting to include a control group that would not be exposed to any manipulation, which can help rule out alternative explanations driving the results.
The non-significant results can also be explained by the MODE model. The MODE model indicates that behavior is widely influenced by controlled processes, if a person has the necessary resources such as time and cognitive capacity and is sufficiently motivated to engage in deliberate reasoning. If either motivation and/or opportunity are missing, the model predicts that behavior will be influenced by attitudes that are automatically activated (Fazio, 1990). In this study, participants who were recruited through Amazon mechanical Turk, were motivated to complete the study because they were paid for their participation. The other participants were mostly personal acquaintances recruited through social media. This perspective suggests that their behavior was less likely to be driven by their implicit attitudes but rather by their deliberate
reasoning. Moreover, while intended as a constraint, it might be possible that 5 seconds provided a sufficient amount of time to make a thoughtful and controlled decision. The fact that both motivation and opportunities were present, explains why the behavior was influenced by controlled processes. In terms of the Reflective-Impulsive model, the fact that participants may have had enough time to elaborate on the questions, provides an explanation why behavior was driven by the reflective system and not by the automatic system.
In terms of practical contributions, the data from this study imply that marketers need to find stronger methods to manipulate people high in ambivalence. More investigation is needed in order to know how to implicitly change their attitudes towards complex products. Evaluative conditioning is a popular method used in commercials. Understanding for which target population the procedure is most effective is crucial. In terms of theoretical contributions, the study did not confirm that evaluative conditioning and level of attitudinal ambivalence are correlated. However, this result should be considered in relation to this study’s limitations.
The present study aimed to address a research gap. The results did not reveal an
interaction effect as expected. The study indicates that using pictures as evaluative conditioning, as presented in this online study, is not the right method to create the desired implicit attitude.
References
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179-211. doi:10.1016/0749-5978(91)90020-t
Baeyens, F., Eelen, P., & Bergh, O. V. (1990). Contingency awareness in evaluative
conditioning: A case for unaware affective-evaluative learning. Cognition & Emotion, 4(1), 3-18. doi:10.1080/02699939008406760
Baeyens, F., De Houwer, J., Vansteenwegen, D., & Eelen, P. (1998). Evaluative Conditioning Is a Form of Associative Learning: On the Artifactual Nature of Field and Davey's (1997) Artifactual Account of Evaluative Learning. Learning and Motivation, 29(4), 461-474. doi:10.1006/lmot.1998.1007
Bandura, A. (1999). Social Cognitive Theory: An Agentic Perspective. Asian Journal of Social Psychology, 2(1), 21-41. doi:10.1111/1467-839x.00024
Bassili, J. N. (1996). Meta-judgmental versus operative indexes of psychological attributes: The case of measures of attitude strength. Journal of Personality and Social
Psychology, 71(4), 637-653. doi:10.1037//0022-3514.71.4.637
Bazzano, L. A., Li, T. Y., Joshipura, K. J., & Hu, F. B. (2008). Intake of Fruit, Vegetables, and Fruit Juices and Risk of Diabetes in Women. Diabetes Care, 31(7), 1311-1317. doi:10.2337/dc08-0080
Blask, K., Walther, E., Halbeisen, G., & Weil, R. (2012). At the crossroads: Attention, contingency awareness, and evaluative conditioning. Learning and Motivation, 43(3), 99-106. doi:10.1016/j.lmot.2012.03.004
Breckler, S. J. (1994). A Comparison of Numerical Indexes for Measuring Attitude Ambivalence. Educational and Psychological Measurement, 54(2), 350-365. doi:10.1177/0013164494054002009
Brunel, F. F., Collins, C. M., Greenwald, A. G., & Tietje, B. C. (1999). Making the public private, accessing the inaccessible: Marketing applications of the implicit association test. Polish Psychological Bulletin, 32,1-9.
Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1997). Beyond Bipolar
Conceptualizations and Measures: The Case of Attitudes and Evaluative Space. Personality and Social Psychology Review, 1(1), 3-25.
doi:10.1207/s15327957pspr0101_2
Carpenter, T., Pogacar, R., Pullig, C., Kouril, M., Aguilar, S. J., LaBouff, J. P., …
Chakroff, A. (2018). Survey-Based Implicit Association Tests: A Methodological and Empirical Analysis. doi:10.31234/osf.io/hgy3z
Cassano, A., Jiao, B., & Drioli, E. (2004). Production of concentrated kiwifruit juice by integrated membrane process. Food Research International, 37(2), 139-148. doi:10.1016/j.foodres.2003.08.009
Cohen, J. B., Pham, M. T., & Andrade, E. B. (n.d.). The Nature and Role of Affect in Consumer Behavior. Handbook of Consumer Psychology.
Conner, M., Sparks, P., Povey, R., James, R., Shepherd, R., & Armitage, C. J. (2002). Moderator effects of attitudinal ambivalence on attitude-behaviour relationships. European Journal of Social Psychology, 32(5), 705-718. doi:10.1002/ejsp.117 Conner, M. T., Perugini, M., O'Gorman, R., Ayres, K., & Prestwich, A. (2007). Relations
Between Implicit and Explicit Measures of Attitudes and Measures of Behavior: Evidence of Moderation by Individual Difference Variables. Personality and Social Psychology Bulletin, 33(12), 1727-1740. doi:10.1177/0146167207309194
De Bruijn, G., Keer, M., Conner, M., & Rhodes, R. E. (2011). Using implicit associations towards fruit consumption to understand fruit consumption behaviour and habit strength relationships. Journal of Health Psychology, 17(4), 479-489.
doi:10.1177/1359105311421049
De Houwer, J., Thomas, S., & Baeyens, F. (2001). Association learning of likes and dislikes: A review of 25 years of research on human evaluative conditioning. Psychological Bulletin, 127(6), 853-869. doi:10.1037//0033-2909.127.6.853
De Liver, Y., Van der Pligt, J., & Wigboldus, D. (2007). Positive and negative associations underlying ambivalent attitudes. Journal of Experimental Social Psychology, 43(2), 319-326. doi:10.1016/j.jesp.2006.02.012
Dennison, B. A. (1996). Fruit juice consumption by infants and children: a review. Journal of the American College of Nutrition, 15(sup5), 4S-11S.
doi:10.1080/07315724.1996.10720475
Echabe, A. E. (2013). Relationship Between Implicit and Explicit Measures of Attitudes: The Impact of Application Conditions. PsycEXTRA Dataset. doi:10.1037/e562792013-005
Fazio, R. H. (1990). Multiple Processes by which Attitudes Guide Behavior: The Mode Model as an Integrative Framework. Advances in Experimental Social Psychology, 75-109. doi:10.1016/s0065-2601(08)60318-4
Fishbein, M. (2008). Reasoned Action, Theory of. The International Encyclopedia of Communication. doi:10.1002/9781405186407.wbiecr017
Friese, M., Wänke, M., & Plessner, H. (2006). Implicit consumer preferences and their influence on product choice. Psychology and Marketing, 23(9), 727-740. doi:10.1002/mar.20126
Fulcher, E. P., & Hammerl, M. (2001). When All Is Considered: Evaluative Learning Does Not Require Contingency Awareness. Consciousness and Cognition, 10(4), 567-573. doi:10.1006/ccog.2001.053
Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional processes in evaluation: An integrative review of implicit and explicit attitude change.
Psychological Bulletin, 132(5), 692-731. doi:10.1037/0033-2909.132.5.692
Gibson, B. (2008). Can Evaluative Conditioning Change Attitudes toward Mature Brands? New Evidence from the Implicit Association Test. Journal of Consumer Research, 35(1), 178-188. doi:10.1086/527341
Glaser, T., Woud, M. L., Labib Iskander, M., Schmalenstroth, V., & Vo, T. M. (2018). Positive, negative, or all relative? Evaluative conditioning of ambivalence. Acta Psychologica, 185, 155-165. doi:10.1016/j.actpsy.2018.02.006
Glashouwer, K. A., Bennik, E. C., De Jong, P. J., & Spruyt, A. (2018). Implicit Measures of Actual Versus Ideal Body Image: Relations with Self-Reported Body Dissatisfaction and Dieting Behaviors. Cognitive Therapy and Research, 42(5), 622-635.
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102(1), 4-27. doi:10.1037//0033-295x.102.1.4 Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual
differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74(6), 1464-1480. doi:10.1037//0022-3514.74.6.1464
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Implicit Association Test. PsycTESTS Dataset. doi:10.1037/t03782-000
Hahn, A., & Gawronski, B. (2018). Implicit Social Cognition. Stevens' Handbook of Experimental Psychology and Cognitive Neuroscience, 1-33.
doi:10.1002/9781119170174.epcn412
Hodson, G., Maio, G. R., & Esses, V. M. (2001). The Role of Attitudinal Ambivalence in Susceptibility to Consensus Information. Basic and Applied Social Psychology, 23(3), 197-205. doi:10.1207/s15324834basp2303_6
Hollands, G. J., & Marteau, T. M. (2016). Pairing images of unhealthy and healthy foods with images of negative and positive health consequences: Impact on attitudes and food choice. Health Psychology, 35(8), 847-851. doi:10.1037/hea0000293
Hollands, G. J., Prestwich, A., & Marteau, T. M. (2011). Using aversive images to enhance healthy food choices and implicit attitudes: An experimental test of evaluative conditioning. Health Psychology, 30(2), 195-203. doi:10.1037/a0022261 Houben, K., Havermans, R. C., & Wiers, R. W. (2010). Learning to dislike alcohol:
conditioning negative implicit attitudes toward alcohol and its effect on drinking behavior. Psychopharmacology, 211(1), 79-86. doi:10.1007/s00213-010-1872-1 iatgen. (2018). Retrieved from https://iatgen.wordpress.com/
Jiao, B., Cassano, A., & Drioli, E. (2004). Recent advances on membrane processes for the concentration of fruit juices: a review. Journal of Food Engineering, 63(3), 303-324. doi:10.1016/j.jfoodeng.2003.08.003
Johnson, B., & Christensen, L. (2008). Educational Research: Quantitative, Qualitative, and Mixed Approaches. Thousand Oaks, CA: SAGE.
Jonas, K., Diehl, M., & Brömer, P. (1997). Effects of Attitudinal Ambivalence on
Information Processing and Attitude-Intention Consistency. Journal of Experimental Social Psychology, 33(2), 190-210. doi:10.1006/jesp.1996.1317
Kaplan, K. J. (1972). On the ambivalence-indifference problem in attitude theory and measurement: A suggested modification of the semantic differential technique. Psychological Bulletin, 77(5), 361-372. doi:10.1037/h0032590
Karpinski, A., & Hilton, J. L. (2001). Attitudes and the Implicit Association Test. Journal of Personality and Social Psychology, 81(5), 774-788. doi:10.1037//0022-3514.81.5.774 Lebens, H., Roefs, A., Martijn, C., Houben, K., Nederkoorn, C., & Jansen, A. (2011). Making
implicit measures of associations with snack foods more negative through evaluative conditioning. Eating Behaviors, 12(4), 249-253. doi:10.1016/j.eatbeh.2011.07.001 Macy, J. T., Chassin, L., & Presson, C. C. (2012). The Association Between Implicit and
Explicit Attitudes Toward Smoking and Support for Tobacco Control Measures. Nicotine & Tobacco Research, 15(1), 291-296. doi:10.1093/ntr/nts117
Maison, D., Greenwald, A. G., & Bruin, R. H. (2004). Predictive Validity of the Implicit Association Test in Studies of Brands, Consumer Attitudes, and Behavior. Journal of Consumer Psychology, 14(4), 405-415. doi:10.1207/s15327663jcp1404_9
March, D. S., Olson, M. A., & Fazio, R. H. (2018). The Implicit Misattribution Model of Evaluative Conditioning. Social Psychological Bulletin, 13(3).
doi:10.5964/spb.v13i3.27574
Michie, S., Abraham, C., Whittington, C., McAteer, J., & Gupta, S. (2009). Effective techniques in healthy eating and physical activity interventions: A meta-regression. Health Psychology, 28(6), 690-701. doi:10.1037/a0016136
Olson, M. A., & Fazio, R. H. (2009). Implicit and explicit measures of attitudes: The perspective of the MODE model. In R. E. Petty, R. H. Fazio, & P. Briñol (Eds.), Attitudes: Insights from the new implicit measures (pp. 19–63). New York, NY: Psychology Press.
Papies, E. K., & Hamstra, P. (2010). Goal priming and eating behavior: enhancing self-regulation by environmental cues. Health Psychology, 29(4), 384.
Pexels. Free stock photos. (2018). Retrieved from http://pexels.com/
Pleyers, G., Corneille, O., Luminet, O., & Yzerbyt, V. (2007). Aware and (Dis)Liking: Item-Based Analyses Reveal That Valence Acquisition via Evaluative Conditioning Emerges Only When There Is Contingency Awareness. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(1), 130-144. doi:10.1037/0278-7393.33.1.130
Poehlman, T. A., Uhlmann, E., Greenwald, A., & Banaji, M. (2005). Understanding and using the implicit association test: III: A meta-analysis of predictive validity. PsycEXTRA Dataset. doi:10.1037/e633942013-708
Priester, J. R., & Petty, R. E. (1996). The gradual threshold model of ambivalence: Relating the positive and negative bases of attitudes to subjective ambivalence. Journal of Personality and Social Psychology, 71(3), 431-449. doi:10.1037//0022-3514.71.3.431
Rothman, A. J., Sheeran, P., & Wood, W. (2009). Reflective and Automatic Processes in the Initiation and Maintenance of Dietary Change. Annals of Behavioral Medicine, 38(S1), 4-17. doi:10.1007/s12160-009-9118-3
Schachtman, T. R., Walker, J., & Fowler, S. (2011). Effects of Conditioning in Advertising. Associative Learning and Conditioning Theory, 481-506.
doi:10.1093/acprof:oso/9780199735969.003.0157
Sparks, P., Conner, M., James, R., Shepherd, R., & Povey, R. (2001). Ambivalence about health-related behaviours: An exploration in the domain of food choice. British Journal of Health Psychology, 6(1), 53-68. doi:10.1348/135910701169052
Strack, F., & Deutsch, R. (2004). Reflective and Impulsive Determinants of Social Behavior. Personality and Social Psychology Review, 8(3), 220-247.
doi:10.1207/s15327957pspr0803_1
Thompson, M. M., Zanna, M. P., & Griffin, D. W. (1995). Let's not be indifferent about (attitudinal) ambivalence. In R. E. Petty & J. A. Krosnick (Eds.), Ohio State
University series on attitudes and persuasion, Vol. 4. Attitude strength: Antecedents and consequences (pp. 361-386). Hillsdale, NJ, US: Lawrence Erlbaum Associates, Inc.
Thurstone, L. L., & Chave, E. J. (1929). Attitude Toward the Church Scale. PsycTESTS Dataset. doi:10.1037/t04689-000
Tourangeau, R., Rasinski, K. A., Bradburn, N., & D'Andrade, R. (1989). Belief accessibility and context effects in attitude measurement. Journal of Experimental Social
Vogel, T., Hütter, M., & Gebauer, J. E. (2017). Is Evaluative Conditioning Moderated by Big Five Personality Traits? Social Psychological and Personality Science, 10(1), 94-102. doi:10.1177/1948550617740193
Palermo, A., Tuccinardi, D., Defeudis, G., Watanabe, M., D’Onofrio, L., Lauria Pantano, A., … Manfrini, S. (2016). BMI and BMD: The Potential Interplay between Obesity and Bone Fragility. International Journal of Environmental Research and Public Health, 13(6), 544. doi:10.3390/ijerph13060544
Wang, Y., Wang, G., Zhang, D., Wang, L., Cui, X., Zhu, J., & Fang, Y. (2017). Learning to Dislike Chocolate: Conditioning Negative Attitudes toward Chocolate and Its Effect on Chocolate Consumption. Frontiers in Psychology, 8.
doi:10.3389/fpsyg.2017.01468
Wiers, R. W., Van de Luitgaarden, J., Van den Wildenberg, E., & Smulders, F. T. (2005). Challenging implicit and explicit alcohol-related cognitions in young heavy drinkers. Addiction, 100(6), 806-819. doi:10.1111/j.1360-0443.2005.01064.x
Appendix A Pictures in the IAT
Tools
Appendix B
Product choices - fruit juice vs. fresh fruits
Fruit juice
Appendix C
Product choices - fruit juice vs. snacks
Fruit juice
