Nation-wide economic inequality and its
effect on individual switching behavior
Nicole D’Amore
Master Thesis MSc. Marketing Management
Nation-wide economic inequality and its effect on
individual switching behavior
By: Nicole D’Amore
S3158322
University of Groningen
Faculty of Economics and Business
Department of Marketing
Supervisors
First Supervisor: Dr. Moeini-Jazani, M. Second Supervisor: Dr. Leliveld, M. C.
Nicole D’Amore Eerste Baan 1B
8011 - Zwolle Ph: +31 649237099
Abstract
Research on brand switching behavior and its antecedents generates important insights for managerial implications. Taking a broad scope on potential societal indicators of varying brand switching behavior, this paper explores the effects of economic (in)equality on individual switching behavior. In this study we have additionally analyzed the innerworkings of this
relationship and therewith we have found a significant mediation effect of sense of control on the relationship between the two conditions: equal society, and unequal society, and the variance in the brand switching intention as a result thereof. This mediator, was established along with the significant mediating effect of perceived equality. Which was measured as a result of the manipulation performed to replicate an equal/unequal society with respect to individual American states. These measures were useful in providing additional understanding for the direct, positive relationship between equal societies and brand switching behavior, which was equally significant in our analysis. This paper further discusses the theory related to this relationship, supported with a literature review, and a final evaluation of its limitations and implications for further research.
Preface
As the final project in for my Master in Marketing Management I present to you my master thesis supervised by Dr. Mehrad Moeini-Jazani. I would like to take this opportunity to express my gratitude to the people that have supported me in the process of writing this paper and that have encouraged me to be persistent in doing the necessary to present the final results in my paper. Firstly, I would like to thank my supervisor Dr. Mehrad Moeini-Jazani for the time he has spent on guiding me and my thesis group, and his patience and flexibility therewith. Mainly due to his hands-on, proactive working style, he has been an invaluable factor for me in completing this thesis program. Additionally, I would like to thank my second supervisor Dr. Marijke C. Leliveld for her time and attention allocated to reviewing my thesis.
Secondly, I would like to show my appreciation to my sister who has been available to me day and night, to enlighten me in the process of performing my data analysis where I was lacking, and her encouragement when I was feeling less confident about my ability to complete this project. Without the motivational support of both my sister and my mother I would not have been able to deliver the result I am presenting to you today.
Table of contents
Introduction………....5
Theoretical framework Perceived Inequality………..……..6
System justification theory………..……...8
Sense of control………...……..…12
Brand Switching behavior……….……13
Conceptual model………..15
Methodology Plan of analysis………...…16
Power and sample size………...….17
Manipulation of inequality………...……...17
Mediator measures………...19
Measurement of Switching behavior………...…...20
Mood check and distractions……….………..……21
Introduction
“You have brains in your head, You have feet in your shoes, You can steer yourself in any direction you choose.” Dr. Seuss (Theodor Seuss Geisel)
Personal autonomy and emancipation have been a common theme in societal development for the past few decades. Whether it concerns our race, sex, religion or social class, we are taught that we all matter equally. Our inherit need for free choice, and our belief in the exertion of control in attaining desired outcomes are fundamental for our well-being (Leotti et al., 2010). However, how do we behave when the systems in our environment are out of balance and our prospects of the future are less certain? Will our belief in our ability to exert control diminish, and if so, what influence does this have on the choices we make?
The literature on economic inequality and financial constraints provides support for two main consequences of our challenged perception of control in uncertain environmental conditions. One consequence is that we are driven by our need to restore control, and another is that we seek stability and become resistant to change in attempt to correct for the uncomfortable feeling of “not knowing what will happen”. With that said, the need for order, structure and routine that is frequently linked to economic inequality, widely focused on the macro consequences of the matter. Inequality is often measured on a national, macro-level, exploring its relation to common societal affairs, in attempt to discover the roots of our often-flawed social systems, and other nation-wide concerns (Jencks, 1972; Boudon, 1974; Smeeding, 2005; Alesina et al., 2004; Trump, 2014).
Theoretical framework
This chapter covers the most relevant scientific literature regarding 4 interrelated theories; perceived inequality, system justification theory, sense of control, and brand switching tendency. Consequentially, two potential mediators are explored from which 3 hypotheses will be derived. Eventually, one concluding, main hypothesis will be proposed.
Perceived inequality
The distribution of wealth and its relation to our societal system, has been a prominent area of research amongst economists, psychologists, sociologists and researchers in other widespread disciplines. Wealth distribution can be assessed in many different terms and forms, but most often it is evaluated on a global scale with its inherent effect on macro mechanisms in our society. Globally, the mere part of individuals live in environments where wealth distribution is remarkably unequal. It has been well-established over the past 70 years that national inequality has significant consequences for the preferences that individuals develop for both economic and social policies, and the overall attitudes and behaviors they portray as members of social groups (Kuznets, 1955; Persson and Tabellini, 1994; Torgler et al., 2008; Wilkinson and Pickett, 2009). In terms of economic inequality, it is common to have the affiliation with the parts of the world, of which we have always been taught were poor and underdeveloped (i.e. countries in Africa, South-East Asia and Latin-America).
As poverty and economic inequality are commonly used interchangeably in assessing causality and relational construct, misperceptions exist on the degree to which our Western society is negatively affected by those particular antecedents (Coulter et al. 1992; Houser and Norton, 2017). Nevertheless, academics are in agreement on the fact that the level of inequality in the United States has reached a new high since last century, even higher than before the Great Depression (Keister, 2000; Wolff, 2002; Davies et al. 2009). Hence, although it remains unknown whether it is poverty or perceived economic inequality at the individual-level that is more likely to predict individual’s diverging preferences for social and economic policies, empirical research suggests that perceived inequality may be more relevant in this regard.
to the results from a large survey on perceived inequality across 23 countries in the International Social Survey Program (ISSP, 2011), it was found that across all countries, the citizens significantly underestimated their society’s inequality level. This effect was additionally found multiple times in experiments amongst American citizens related to the wealth and income distribution in their respective society (Osberg and Smeeding, 2006; Nieheus, 2014), where Norton and Ariely (2011) established that they did not only dramatically underestimate the level of inequality, but they also indicated their ideal levels to be even more equal. This effect was established when measuring perceived inequality in the broadest sense, which included: wealth, income and social mobility and moreover, their perception of their own stance on the so-called “societal ladder” (Hauser and Norton, 2017). According to Gimpelson and Treisman (2016), the underestimation was not only attributed to general inequality levels, but additionally tothe change in inequality level in their countries over the years and to the mobility level in their country. Davidaj and Gilovich (2015) and Kraus and Tan (2015) particularly demonstrated that Americans estimated their chances to move upward far higher than to move downward the “societal ladder”.
Overall, these results indicated a general optimistic perception of equality, which is painfully inaccurate from reality. When researchers attempted to learn about the drivers of these inaccurate beliefs, it was often found that the sources were threefold. Firstly, people tend to use their direct environment as an important cue for their estimations of national inequality levels. Secondly, the intensity of media coverage on social inequality is directly related to the estimations that people make, however, this effect was found to only have a short-term influence on people’s perceptions. Lastly, it is argued that a strong determinant of people’s perceptions of equality is a person’s level of endorsement of hierarchical societies and their beliefs in merit versus luck. Individuals who belief in personal choice as to predicting the quality of outcomes, are less likely to see inequality between groups of people than people who believe in luck as the determinant of outcomes, with perceived injustice as a result (Kteily et al. 2017). The latter driving force has not only been shown to be a determinant of the (mis)perceptions that individuals hold about the inequality levels in their societies, but more importantly of the attitudes, beliefs, and behavior they develop as a result. And therefore, this also holds for an individual’s approval of, and support for the maintenance of wealth inequality as the status quo (Mollerstrom and Seim, 2014; Savani and Rattan, 2012).
the degree to which they decide to support governmental redistribution of wealth therewith (Cruces et al., 2013). Nonetheless, the locus of control with regards to people’s belief in merit or luck as the source of their economic outcomes, has been shown to be a more evident explanation for one’s preference for the redistribution of wealth than income is (Fong, 2001). Preferences for wealth redistribution in the aforementioned studies is generally represented by one’s support of political systems with social policies e.g. social welfare programs, government subsidies and progressive taxation (Alesina et al., 2004). With that said, extensive research on human motive for justification of controversial societal systems, initially performed by Jost and Banaji (1994) elicited our inherent tendency to justify and legitimize the systems we live by. More interestingly, although it was previously argued that our motivation for supporting, or opposing political and societal policies was mainly driven by our self-interest depending on whether we were advantaged or disadvantaged by the system. Congruently, Jost and Banaji (1994) found that we have an equal tendency to support the status quo in our societies disregarding the way we are influenced by the implications of the system.
Hence, it is argued that rather than acting according to our self-interest with regards to how the existing systems affect us, either personally or as members of an in-group, we may more likely be
programmed to defend “the way it is”, no matter whether we are profiting or suffering therefrom
(Jost and Banaji 1994; Jost et al., 2002; Jost and Pelham 2002; Jost et al., 2004).
System justification theory
2013; Luttig, 2013). Moreover, cross-national research revealed that people in nations with more inequality were not more in favor of income redistribution neither were they more likely to think that the inequality in their countries was too high (Alesina and Glaeser 2004, Kenworthy and McCall 2008, Schroeder 2014, Trump 2014).
While the system justification theory endorses the generalization of this tendency as embedded in each of us, it does emphasize the fact that the degree to which each of us legitimizes the existing status quo, significantly varies between individuals. As the theory concerns a somewhat contradictory mechanism, one of the hypotheses by Trump (2014) for explaining this relationship is the so-called inequality-induced motivation hypothesis. Early on, Jost and Banaji (1994), in their attempts to clarifying this variance, had proposed the strong probability that the higher the degree of perceived “unfairness” and seriousness of a problematic social arrangement, the higher the motivation for legitimization and justification would consequently be (Jost and Banaji 1994, p.16). This hypothesis strongly leans on the framework of the ascertained cognitive dissonance theory, which was first officially developed in 1957 by Festinger.
preference for processing information that we believe is most consistent with our goals and behavior (Walster, 1964; Frey, 1986).
With that said, while the system justification theory draws on the cognitive dissonance theory, what truly distinguishes cognitive dissonance from social justification theory is that system legitimization is believed to be group-based and entails the collective process of support of the status quo i.e. its classification as “group justification” (Jost & Banaji, 1994; Jost et al., 2004). With that said, social arrangements are imbedded in our society and as such, a big homogamous and centralized system simply cannot conform to each individual’s idiosyncratic characteristics. With other words, we are all negatively affected by it in one way or another, depending on our belief systems. Nevertheless, it may be precisely that collectivistic nature of the system as a whole, with regards to the status quo that we legitimize, as opposed to the justification of our individual actions and behavior linked to cognitive dissonance, that have a differential effect on the antecedents and consequences of justification mechanisms.
As Havel (1978, 1985, 1991) argued repeatedly in his extensive study on societal powerlessness the following; “everyone . . . is both a victim and a supporter of the system” (p. 144). In additional system justification argumentation, Lerner (1980) explained that people need to believe to be living in a just world, just so that they “can go about their daily lives with a sense of trust, hope, and confidence in their future’’ (p. 14). The fundamental and common driver of system justification, which is likewise believed to be a consequence of perceived inequality, is the unwarranted sense of powerlessness and lack of control. The mechanism is characterized by the cognitive acceptation of “the way it is” in order to minimize the uncertainty of social change and consequently, to maintain the familiar social structure (Jost et al., 2003). As a result, it is argued that by rejecting the uncertainty of social change in the process of system justification, the overall motivation to induce change and the support for redistribution of wealth is reduced. This is the reason why some have stated that indirectly, inequality itself fuels inequality, which is essentially a matter of lack of motivation to strive for social change (Alesina and Glaeser 2004, Kenworthy and McCall 2008, Schroeder 2014, Trump 2014).
correlation of system justification tendency to an individual’s openness to experience, therewith a reduced affinity for situations involving novelty, diversity and change.
As system justification may be elicited by perceived inequality as a coping mechanism for the feelings of dissonance resulting therefrom, and as system justification is negatively related to one’s need for novelty and change, it could be argued that this effect is similarly reflected in one’s decisions regarding consumer behavior. The need for structure and order, as well as the lack of affinity involving change, are all motivational cues for the decisions we make in our daily lives. According to Ellis et al. (2009), predictability is a fundamental determinant in all facets of our environment. For those who perceive themselves to suffer from financial constraints, studies have shown that they experience more chaos in day-to-day live, which makes it less predictable what the future will hold for them (Evans, 2004; Mittal & Griskevicius, 2014). This in turn creates a sense of environmental uncertainty, and thereby those who suffer from financial constraints are left experiencing a decreased sense of control as compared to those who have more access to financial resources that buffer their sense of control (Brunner 1997; De Witte 1999).
Hypothesis 1: Perceived equality is positively correlated with sense of control.
Notwithstanding, as stated earlier on, we all use system justification as a mechanism for coping with feelings of discomfort relating to the societal status quo e.g. lack of control. However, the degree to which we employ system justification individually, differs significantly. Therefore, it could be argued that individuals who more strongly engage in system justification, will have more reduced consequent feelings of discomfort and therewith feel less that they lack control than the people who engage less strongly in system justification. In other words, the uncertainty stemming from high perceived inequality produces a general feeling of lack of control in individuals. But the degree to which we apply system justification determines the degree to which we feel we lack control as a result.
Sense of control
One’s belief in their ability to exert control in their environment is argued to not only be favorable to our well-being, but additionally to be a necessity for us, both psychologically and biologically (Leotti et al., 2010). Our estimation of the degree to which we are able to achieve desirable outcome, as well as our need for autonomy in itself, are important drivers for us in the decisions we make. In fact, our ability to exert choice is so important to us, that experiments with both humans and animals have shown that when we are presented with two options, of which the expected outcome value is equal, we prefer the option that presents us with an additional choice, to the one that does not. This preference is therefore not driven by the extraneous outcome benefits, but solely by the choice over non-choice equation (Catania and Sagvolden, 1980; Suzuki, 1997, Bown, et al., 2003). This finding is interesting, as the preference for free choice therefore outweighs the incremental effort and energy that is required in the additional decision-making process.
Alike animals, humans demonstrate several negative affects resulting from the restriction or removal of choice. As a result, humans react in several ways, to which including an increased effort for restoring control (Crombez et al., 2008). Our aversive responses to the loss of control and tendency to preserve autonomy, is already noticeable at an early stage, as early as when we learn our very first skills after birth. A good example would be that of an infant that has just learned to feed themselves. When they are aware of their capability of exerting control in the successful achievement of their goal (being fed), they become resistant to their parents trying to take this control from them. Even though the rewards from the action will be the same when they are fed by their parents, or by themselves, they still prefer it being a product of their own action where they were in control (Sullivan and Lewis, 2003). The fact that this preference is already there at a very early stage in the human lifecycle, shows that our need for control is most likely biologically inherited.
Even though our preference for choice-tasks, considering the effort and energy involved in self-exertion of control are not economically justifiable, the idea of exercising control as a reward in itself is. This way we can rationalize the choice preference as a mechanism for maximization of utility (Leotti et al., 2010).
of control, is argued to reduce our affinity for novelty and change (Jost and Hunyady, 2005). It is furthermore argued that lack of control makes us resistant to change, and increases our preference for the familiar, order and structure in attempt to avoid the uncertainty of the unknown (Jost et al., 2003). There are many imaginable ways and domains in which this preference could influence our behavior, ranging from our social interactions, political engagement, career choices etc. Likewise, it could very likely influence our consumer behavior. The choices that we make from awareness to purchase, in the entire path-to-purchase, may very well be affected by our motivation to avoid uncertainty. One of the many decisions we make in our customer journey’s, is that between brands. More specifically, when we are accustomed to a product through repeat purchases, most often in relation to commodity goods, do we stick to the same brand, or do we switch to another brand with a similar offering when we are faced with the choice again? Based on the aforementioned natural preference for autonomy, and for choice over non-choice when we perceive we are in control, it is likely that brand switching tendency is higher for individuals that have a high sense of control.
Hypothesis 3: There is a positive relationship between sense of control and brand switching behavior.
Brand Switching behavior
Brand switching can be defined as the decision to switch to a new brand, either new or existing on the market, instead of continuing to purchase the brand we are used to buying (Lam et al., 2010). It is important to note that this paper treats brand switching behavior as a behavioral response to correctional cognitive-processes as opposed to variety-seeking tendencies.
their respective brand choice equation. According to Hamilton et al. (2018), financial constraints and economic limitations pose as strong determinants in brand choice strategies, as they drive a consumer in the attentive resource allocation considerations alike. Literature on economic limitations in the broadest sense, include the causal process in which a consumer’s focus shifts to money when this resource is perceived to be scarce (Mullainathan & Shafir, 2013).
Additionally, brand switching is considered to involve certain possible switching costs related to both reduced satisfaction, which reduces the utility obtained from the resources allocated, and increases the effort and time spend in finding an alternative brand Burnham, Frels and Mahajan (2003). And being that economic constraints shifts individuals’ focus to money, and increases the motivation to maximizing utility, the probability that people in reduced equality conditions are less inclined to taking this risk related to brand switching than those in more equal conditions, is high. In other words, the more equal the distribution of wealth in a society, the more an individual perceives its society to be equal, and the more they are willing to take the risks involved in brand switching as opposed to those in unequal societies.
Conceptual Model
The model below represents an overview of the variables derived from the literature review, and that were included in the subsequent experiment. The model consists of a direct and an indirect effect, from of which the design concerns a one factor design with condition: 2 levels
(equal/unequal).
IV: Condition (equal/unequal)
Med: Perceived equality (continuous) Med: Sense of control (continuous) DV: Brand switching (continuous)
Methodology
Plan of analysis
The goal of this study is to explore the effects of living in an equal versus living in an unequal society, on individuals’ brand switching behavior. Additionally, building on existing literature on inequality and its relation to consumer behavior, we ran an experiment in attempt to learn more about the potential mediators of the hypothesized main effect i.e. perceived inequality and sense of control. The main effect design concerns a one factor, design with inequality: 2 levels, and its differential effect on the continuous outcome variable: brand switching. The design included both between-subject and within-subject elements as the participants were randomly allocated to one of two conditions, in which they were exposed to one out of two types of wealth-distribution data, after which the entire group of participants was exposed to the same following tasks.
In order to gather our data set, three hundred sixty-five participants were eventually included after data cleaning through Amazon Mechanical Turk (M-Turk). The study that we launched on M-Turk was linked to our survey created on Qualtrics.
After having provided a short introduction, a captcha task, and a consent statement, the participants had the option to either continue or opt-out of the survey. Thereafter, the participants were asked to provide information on their demographics including; age, ethnicity, household income, state of residence, political orientation and educational background. Next, they were randomly assigned to two conditions i.e. equal society condition, or unequal society condition. As the participants were now equally divided in two levels of a factor in which a manipulation was performed, the entire set of participants were asked about their sense of control at that moment in time, in the form of a slider from low to high. In order to maximize the manipulation effects on the outcome variable; brand switching tendency, the participants were directly asked to perform the task about the probability of switching between brands. Only after having completed this task followed by an attention check, they were instructed to perform the tasks involving their agreement with the items that would measure the potential mediators to the precedingly measured main effect. Finally, the participants were subjected to a mood check and a declaration with regards to the quietness of their environment during their participation in the survey.
United States of America. This prerequisite was additionally validated with a skip-logic that was inserted in the demographics sections for people that selected: “I do not reside in the US” at the time that they were asked to select their state of residence, and in retrospect, according to their actual location displayed with their IP-addresses. The aforementioned validations were crucial for the success of the manipulation for the conditions in our experiment.
Power and sample size
For the determination of the sample size, an a-priori power analysis was performed with
G*Power. The sample size was pre-determined, expecting a small interaction effect size (f2= .02)
with the power of 80 rendered a sample size of at least 395. While the initial sample size was 489, this number was reduced to 411 after data cleaning according to the following criteria: incomplete survey, incorrect response to the attention checks, and disbelief about the pie-chart data in the manipulation, as expressed in the writing task. However, while thoroughly studying the responses, we discovered that some of the respondents were not actually residing in the United States in contrast with the selection they had made in the demographics section.
Therefore, considering that for this particular study it was an important criterion for the success of the manipulation that the respondents resided in the US, those participants were also excluded from the dataset at a later stage. This left us with the final sample size of 365.
Manipulation of inequality.
After having extensively studied the literature on inequality, a commonly used manipulation was chosen to be replicated in this study. Herewith, Coté et al. (2016) performed a manipulation in which the computer appeared to be retrieving the wealth-distribution data related to the state the respective participant had selected they resided in. After a few seconds, depending on the condition the participant had been randomly assigned to, a pie-chart representing the total wealth in their state was displayed. The pie-chart was then either equally or unequally divided into 5 income classes.
Figure 1: Equal wealth distribution pie-chart
As visible in the charts above, the distribution in the equal condition was fairly equal, while still relatively realistic (11%, 15%, 18%, 21%, 35%). Contrastingly, the distribution in the chart for the unequal condition, consisted of a far more unequal wealth distribution (2%, 6%, 9%, 37%, 46%).
In relation to the experiment by Coté et al. (2016), the difference between the studies lies in the geographic scope of the study. The study performed by Coté et al. (2016) consisted of a manipulation of inequality on a national level in comparison to a far more equal society i.e. Sweden (see appendix A). However, even though the study in this paper focused on a national inequality scope, we have purposely chosen to exclusively expose the participants to the supposed wealth distribution of the states they reside in, as this may increase the success of the manipulation, due to increased relatability (Kteily et al. 2017). Furthermore, after the exposure to the supposed graphical representation of the wealth distribution in their state, the respondents were asked to fill out a few comprehension tasks related to the percentages in the pie-chart that they had seen, in order to test whether they had paid attention to the figures in the chart. Then, in an attempt to increase the effect of the manipulation in the inequality condition, two tasks followed. In the first task they were asked how accurate they had predicted the level of economic equality in their state, with the notion that research on inequality found that Americans generally tend to significantly underestimate the level of inequality in their environments. And finally, the last task concluded a short explanation of the consequences of the either equal/unequal wealth distribution in their states relating it to (depending on the condition they were assigned to) either poor/good access to healthcare, education and jobs. Thereafter they performed a writing task, describing how they felt living in a society as such.
Mediator measures
compromised. The selection of the items included in the sense of control measure, was done following the exact 2 personal mastery and 4 perceived constraints items selected by Lachman and Weaver (1998) in their reduced item scale (α=.825). Furthermore, social justification tendency was measured according the commonly used scale, originally created by Jost and Banaji (2004), consisting of 8 items measured on a 7-point likert scale (α=.863).
Measurement of Switching behavior
Ultimately, after having studied the literature on switching behavior and the methods that are most frequently used for measuring switching behavior, we included 2 tasks that we believed could generate the most accurate output for this variable in the context of our experiment. The first task consisted of a short introduction explaining a situation, in which they needed to buy a new laptop because they had given their current laptop to their brother. But when they were in the store to buy the laptop, they saw that apart from the laptop they used to have (brand A), they also offered a laptop from another brand (brand B), which had the same price and functionalities. Then they were asked to report the probability that they would either repurchase the laptop from brand A, or switch to brand B, on a 10-point slider (see Appendix B for survey item).
Additionally, the participants were asked to perform a task in which they reported their preference for an ice-cream flavor from a selection of four flavors, for each of the four
consecutive nights on which they were told they would be receiving 1 free cup of ice-cream (see the image in Appendix C for the visual representation of this survey item). Even though this task has most often been used to measure variety seeking tendency, the literature related to this
method also provided support for its ability in measuring switching behavior. Therefore, this task was fundamentally included to possibly measure switching behavior on a broader scope, relating it to both utilitarian and hedonic product types. Nevertheless, even though the concepts variety seeking, and switching behavior are commonly used interchangeably, the fundamental
all functionalities and price between the two product choices were the same, leaving the only difference being the brand. Whereas the variance in switching between different ice-cream flavors, could possibly be related to personal preference for certain flavors and therefore it was a less reliable measure. As a result, we used only included the task with the choice between laptop brands as a measure for brand switching in our analysis.
Mood check and distractions
Results
Representativeness of sample
The final sample consisted of 365 participants, of which all were currently residing in the United States. The age of the participants in the sample ranged between 19 to 72 years old with M= 37. Furthermore, the descriptives of the sample revealed that the age and income level of the participants were relatively normally distributed (age; skewness = 0.962, kurtosis = 0.124 and income; skewness = 1.385, kurtosis = 2.303. Which according to George and Mallery (2010), means that skewness and kurtosis values are within the normality range (-2 and 2), except for the kurtosis of income, which was >2. This indicated that the peak of the distribution of income is slightly higher than in normal distribution centered around the mean (M= 5.76). Income levels were measured with a multiple choice item of 20 choices, each representing an incremental income bracket (see Apendix A for an image of the income brackets used in the survey). Therefore, since there were no big abnormalities, we could argue that our sample was fairly representative of the American society for an exploratory study as the one we performed.
Variable Descriptives
According to Table 1 below, both outputs for skewness and kurtosis of the main variables, which express the distributional shape, are within the range of -2 and 2 for all variables included. Therefore, in accordance with George and Mallery (2010), the normality of distribution of the variables can be considered accepted.
Table 1 shows the descriptive statistics of the four continuous variables in the model, namely: Perceived Equality, Sense of Control, and Brand Switching Behavior.
Table 1
Perceived Equality, System Justification Tendency, Sense of Control, and Brand Switching Behavior
Location Dispersion Distributional Shape
mean sd m
in
max skew kurtosis
Perceived Equality 4.38 2.63 0 10 0.13 -0.90
Sense of Control 6.40 2.63 0 10 -0.59 -0.46
Manipulation check
The aim is to understand whether the inequality manipulation had a significant effect on the equality perceptions of the participants. Following the manipulation, participants in the equality condition perceived equality in their state higher, by reporting a higher value on the 10-point scale from low inequality to high inequality (M = 5.72, SD = 2.09) than those randomly assigned to the inequality condition (M = 3.25, SD = 2.52) reported according to the outcome of the Levene’s test for equality of variances by which the assumption of equal variance was rejected (p = .034) (mean difference: 2.472, 95% CI [1.991; 2.953], t[362,938] = 10.261, p < .000). As the mean difference of perceived equality was significantly different between the two conditions, the manipulation can be considered successful (see figure 3 below).
Main analysis
Following our manipulation, our objective was to find a possible mediation effect for the main relationship between our predictor and response variable. To do this, we followed the causal steps approach by Barron and Kenny (1986), in which they suggest that in order to detect a mediating relationship, the main effect should be significant. Nonetheless, it is repeatedly argued that in order to proceed with the mediation analysis, the main effect between the predictor and the outcome variable should be established first. Nonetheless, Hayes (2009) argues in his evaluation of mediating relationships, that even when the main relationship is not significant, the indirect effect might still be present. In his argument, he specified that in order to detect the indirect effect, the mediator variables should be significantly correlated with both the independent and dependent variables of the model. Therefore, we started out by performing a correlation check between all variables in the model.
In this check all continuous variables were included i.e.: Perceived Equality, Sense of Control, and Brand Switching Behavior. The correlations in table 2 below revealed that all variables were significantly associated with each other at the 0.01 level.
Table 2 shows the correlations between all the continuous variables
Next, we tested whether we could find enough evidence to reject the null hypotheses of no group differences (condition: equal versus unequal), for each of the continuous variables above, with an independent sample T-test. The comparison of the means of the variables as a function of condition showed that all means were lower for the unequal condition, namely: the mediator variables, Perceived Equality, (mean difference: 2.472, 95% CI [1.991; 2.953], t[363] = 10.103, p < .000), Sense of Control (mean difference: 0.901, 95% CI [0.366; 1.437], t[363] = 3.312, p < .001), and for the dependent variable Brand Switching (mean difference: 0.385, 95% CI [0.027; 0.743], t[363] = 2.113, p = .035). Hence, condition (equal versus unequal) significantly and positively predicts
Perceived Equality
Sense of control Brand Switching
Perceived Equality 1 .296* .203*
Sense of Control .296* 1 .178*
Brand switching .203* .178* 1
group differences for each of the three mediator variables as well as for the dependent variable. Therewith, we have established the presence of a significant direct effect between Condition and Brand switching behavior.
Hypothesis testing
In order to adhere to the proposed mediation model by Baron and Kenny (1986), three conditions should be met. The first condition was that the independent 2-level variable Condition should be associated with both the mediator variables i.e.; Perceived Equality and Sense of Control. The support for having met this first criterium, was presented in the t-test for measuring group differences as a function of Condition earlier on. The second condition, defines the need for association between the mediator variables and the outcome variable. Therewith, as reported by the correlation table in the main analysis part above, the mediator variables Perceived Equality and Sense of Control were indeed associated with the outcome variable Brand Switching.
Nevertheless, the step-wise inclusion of mediation variables in the model might influence the associations that were described. With regards to this limitation, and conjointly in order to meet the third condition of the mediation guidelines by Baron and Kenny (1986), three multiple linear regression analyses were performed, and in each model, the relevant control variables were selected (i.e., in total we tested three models). To adhere to the third condition of this mediation, the effect of the independent variable, here Condition (equal/unequal), and the outcome variable Brand Switching, should decrease when both the mediators are added in the model.
the effect of Condition became non-significant (B = .067, 95% CI [-.333; .467], p = .740). In terms of explained variance, the R2 significantly increased with 2.9% (F(1,362) = 11.111, p < .001), resulting in a total R2 – proportion of the total variance explained – of 4.2%. As a third step, the second mediator variable – Sense of Control – was added to the model and showed a significant effect on Brand Switching (B = .085, 95% CI [.015; .155], p = .017). Resultantly, the effect of Perceived Equality slightly decreased yet remained significant (B = .105, 95% CI [.027; .183], p = .008). Additionally, the effect of Condition also slightly decreased and remained non-significant (B = .049, 95% CI [-.349; .466], p = .810). In this third and final model, the R2 again showed a significant increase of 1.5% (F(1,361) = 5.761, p = .017), resulting in a final proportion explained variance of 5.7% (R2 = .057, F(1,363) = 4.467, p = .035).
In figure 4 below, the final mediation model is shown. Hence, the results provide support for the complete mediation model, with Perceived Equality and Sense of Control mediating the effect of Condition (equal versus unequal) on Brand Switching.
Figure 4: Final mediation model (model 3) with zero-order correlations and partial correlations in brackets. ** means p < .01, * means p < .05.
Discussion
From our final mediation model (figure 4), we can conclude that the level of economic equality in a society, indeed influences consumers’ brand switching behavior. Additionally, with our mediation analysis we have shown that there is substantial reason to believe that this effect is mediated by perceived equality and sense of control. Relating these findings to our theoretical framework, the results of our study were consistent with our initial hypotheses.
Firstly, our hypothesis that perceived equality is positively correlated with sense of control, was confirmed (r = 0,296, p =.01). This effect could be explained by the holistic System justification theory that we covered in our literature review, where we used academic sources to connect perceived equality and sense of control as conjointly representing antecedents and consequences of the cognitive mechanism that we call system justification. And considering that sense of control is a more situational whereas system justification is more dispositional, we though that system justification was less susceptible for manipulation and therefore sense of control would fit well in our model as a measure of system justification tendency. Secondly, the inclusion of a condition to our study to influence perceived equality, led us to accept our
Conclusion and recommendations
This study provides a contribution to the literature on economic inequality in the context of consumer behavior. More specifically, as to how living in an environment where the uncertainty stemming from financial constraints is high, eventually influences the decisions that consumers make with regards to switching between brands. Even though the literature on both phenomena is extensive, economic inequality and switching behavior are different measures with regards to their scope, as one is a nation-wide, macro measure, and the other an individual, behavioral measure. This ‘funneled’ approach is less common in this research area, and therefore its uniqueness contributes to the existing literature and could aid in managerial marketing implications.
In conclusion, we found that switching behavior is significantly higher in society with higher economic equality, and that this is partially caused by the increased sense of control that people feel when they perceive that they live in an equal society. This finding could be relevant in further research on control-restoring marketing techniques, to be used especially in societies where sense of control is relatively low i.e. unequal societies. This in turn, could contribute to managerial implications with regards to marketing mix allocation and improvements of point-of-purchase conditions for brand management.
Limitations
While the results of this study could aid in implications for further research, we essentially disregarded what the exact role of system justification mechanism is in the relationship between economic equality and switching behavior. We currently measured the effect that sense of control has on the main relationship, interpreting it as consequence of the level of perceived equality.
However, literature on system justification provides strong support for the corrective role that system justification has when feelings of dissonance, such as lack of control, emerge, and therefore it would be useful to explore whether system justification tendency moderates, mediates, initiates or follows sense of control in our model. Nevertheless, considering the limitations of our resources, our experiment design, and the dispositional nature of system justification tendency, we suggest that further research should be performed with a larger sample, and in combination with qualitative research methods. As the study was performed online, using M-Turk, the drawbacks include the inability of acquiring more in-depth answers from the participants, as the length of the survey is crucial in the stimulation of engagement (Evans and Mathur 2005; Buhrmester, Kwang, and Gosling 2011).
Another limitation of this study, is the lack of variety in product types for which brand switching is explored. With that said, it is very likely that the act of switching between brands as a consequence of economic equality, varies between product categories e.g. electronics, foods, clothing etc., hedonic/functional consumption.
Appendices
A.
Equal condition (Sweden)
B.
References
1. Alesina, A.F., and G laeser, E. (2004). Fighting poverty in the U.S. and Europe: A world of difference. Oxford, UK: Oxford University Press
2. Alvaredo, F., Atkinson, A.B., Piketty, T., Saez, E. (2013) The top 1 percent in international and historical perspective. J Econ Perspect , 27,3-20.
3. Aronson, E., (1992). The return of the repressed: D issonance theory makes a comeback. Psychological inquiry, 3(4), pp.303-311.
4. Baharom, A. H., and Habibullah, M. S. (2009). Crime and income inequalit y: The case of Malaysia. J. Pol. & L., 2, 55.
5. Buhrmester, M., Kwang, T., and Gosling, S. D. (2011). Amazon's Mechanical Turk: A new source of inexpensive, yet high-quality, data?. Perspectives on psychological science, 6(1), 3-5.
6. Boudon, R. (1974). Education, opportunity, and social inequality: Changing prospects in western society.
7. Botti, S., Orfali, K., and Iyengar, S. S. (2009). Tragic choices: Autonomy and emotional responses to medical decisions. Journal of Consumer
Research, 36(3), 337-352.
8. Brehm, J.W., (1956). Postdecision changes in the desirabilit y of alternatives. The Journal of Abnormal and Social Psychology, 52(3), p.384.
9. Bow n, N., Read, D., and Summers, B. (2003). The lure of choice. Journal Of Behavioral Decision Making, 16(4), 297-308.
10. Cannon, C., G oldsmith, K., and Roux, C. (2018). A Self‐Regulatory Model of Resource Scarcity. Journal of Consumer Psychology.
11. Catania, A.C., and Sagvolden, T. (1980) Preference for free choice overforced choice in pigeons. J. Exp. Anal. Behav. 34, 77 –86
12. Coulter, F. A., Cowell, F. A., and Jenkins, S. P. (1992). Equivalence scale relativities and the extent of inequality and poverty. The Economic Journal, 1067-1082.
13. Crombez, G., Eccleston, C., De V lieger, P., Van Damme, S., and De Clercq, A. (2008). Is it better to have controlled and lost than never to have
controlled at all? An experimental investigation of control over pain. Pain, 137(3), 631-639.
15. Davidai, S., and G ilovich, T. (2015). Building a more mobile America —O ne income quintile at a time. Perspectives on Psychological Science, 10(1), 60-71.
16. Davies, J.B., Sandström, S., Shorrocks, A., and Wolff, E. (2009). The global pattern of household wealth. J Int Dev, 21:1111 -1124. 4.
17. Ellis, B. J., Figueredo, A. J., Brumbach, B. H., and Schlomer, G. L. (2009). Fundamental dimensions of environmental risk. Human Nature, 20(2), 204-268.
18. Evans, G.W. (2004). The environme nt of childhood poverty. American Psychologist, 59(2), 77
19. Faber, J., Fonseca, L.M. (2014) How sample size influences research outcomes. Dental Press J Orthod. 19(4):27 -9.
20. Festinger, L. and James M.C. (1959). "Cognitive consequences of forced compliance." The journal of abnormal and social psychology 58.2203.
21. Festinger, L. (1957). A Theory of Cognitive Dissonance (Stanford University Press: Stanford, California).
22. Fong, C. (2001). Social preferences, self-interest, and the demand for redistribution. J Public Econ 82: 225 –246.
23. Frey, D. (1986). "Recent research on selective exposure to information." Advances in experimental social psychology . Vol. 19. Academic Press, 41 -80. 24. Gimpelson, V., and Treisman, D. (2016) Misperceiving Inequality, Working
Paper. 1-56. 21.
25. George, D., and Paul, M. (2010)."SPSS for Windows step by step. A simple study guide and reference
26. Havel, V., and Wilson,P., (1985). The Power of the Powerless. International Journal of Politics. 15. 23-96.
27. Hauser, O., and N orton, M. (2017). (Mis)perceptions of inequality. Current Opinion In Psychology, 18, 21-25.
28. ISSP Research Group (2011). International Social Survey Progra mme (ISSP): Social Inequality. Cologne: GESIS Data Archive .
29. Jencks, C. (1972). Inequality: A reassessment of the effect of family and schooling in America.
31. Jost, J. T., and Banaji, M. R. (1994). The role of stereotyping in system -justification and the production of false consciousness. British Journal of Social Psychology, 33, 1–27.
32. Jost, J. T., Pelham, B. W., and Carvallo, M. (2002). Non -conscious forms of system justification: Cognitive, affective, and behavioral preferences for higher status groups. Journal of Experimental Social Psychology, 38, 586 -602.
33. Jost, J. T., G laser, J., Kruglanski, A. W., and Sulloway, F. J. (2003). Political conservatism as motivated social cognition. Psychological bulletin, 129(3), 339.
34. Jost, J., Pelham, B., Sheldon, O., and Ni Sullivan, B. (2003). Social inequality and the reduction of ideological dissonance on behalf of the system: evidence of enhanced system justification among the
disadvantaged. European Journal Of Social Psychology, 33(1), 13-36. 35. Jost, J.T., Banaji, M.R., and Nosek, B.A. (2004). A decade of system
justification theory: Accumulated evidence of conscious and unconscious bolstering of the status quo. Political Psychology, 25, 881 –919.
36. Keister, L.A. (2000). Wealth in America: Trends in Wealth Inequality . 37. Kenw orthy, L., and McCall, L. (2008). Inequality, public opinion and
redistribution. Socio-Economic Review, 6 (1), 35–68.
38. Kraus, M. W., Côté, S., and Keltner, D. (2010). Social class, contextualism, and empathic accuracy. Psychological science , 21(11), 1716-1723.
39. Kraus, M. W., & Tan, J. J. (2015). Americans overestimate social class mobilit y. Journal of Experimental Social Psychology , 58, 101-111. 40. Kteily NS, Sheehy-Skeffington J, Ho AK: H ierarchy in the eye of the
beholder: (anti)egalitarianism shapes perceived levels of social inequality. J Pers Soc Psychol 2017, 112:136 -159.
41. Kuznets, S. (1955). “Economic Growth and Income Inequality.” American Economic Review 45(1): 1–28.
42. Lachman, M. E., and Weaver, S. L. (1998). The sense of control as a moderator of social class differences in health and well -being. Journal of personality and social psychology, 74(3), 763.
43. Leotti, L., Iyengar, S., and Ochsner, K. (2010). Born to choose: the origins and value of the need for control. Trends In Cognitive Sciences , 14(10), 457-463.
45. Luttig, M. (2013). The structure of inequality and Americans’ attitudes toward redistribution. Public Opinion Quarterly, 77 (3), 811-821. 46. McCall, L. (2013). The undeserving rich: Beliefs about inequality,
opportunity, and redistribution in American society . Cambridge, UK : Cambridge U niversity Press.
47. McFadden, Daniel, (1986), The Choice Theory Approach to Market Research, Marketing Science, 5, issue 4, p. 275-297.
48. Mittal, C., and Griskevicius, V. (2014). Sense of control under uncertainty depends on people’s childhood environment: A life history theory approach. Journal of Personality and Social Psychology, 107 (4), 621–637.
49. Mollerstrom, J., Seim, D. (2014) Cognitive ability and the demand for redistribution. PLOS ONE
50. Niehues, J., (2014). Subjective Perceptions of Inequality and Redistributive Preferences: An International Comparison, Working Paper. 2014:1 - 23. 51. Norton, M.I., and Ariely, D., (2011). Building a better America — one wealth
quintile at a time. Perspect Psychol Sci 2011, 6:9 -12.
52. Osberg, L., Smeeding, T. (2006) “Fair” inequality? Attitudes toward pay differentials: the U nited States in comparative perspective. Am Sociol Rev 2006, 71:450-473.
53. Persson, T. and Tabellini, G.(1994), Is Inequality Harmful for Growth?, American Economic Review, 84, issue 3, p. 600-621.
54. Rao, H., Davis, G., and Ward, A. (2000). Embeddedness, Social Identity and Mobility: Why Firms Leave the NASDAQ and Join the New York Stock Exchange. Administrative Science Quarterly,45 (2), 268-292.
55. Savani, K., and Rattan, A. (2012). A choice mind-set increases the
acceptance and maintenance of wealth inequality. Psychol Sci , 23:796-804 56. Schroeder, M. (2014). “How income inequality is related to tolerance for
inequality” Working Paper. Marburg, Germany: Philipps-Universität Marburg.
57. Seuss, D. (1990) Oh, the Places Y ou’ll Go! Random House
58. Smeeding, T. M. (2005). Public policy, economic inequality, and poverty: The United States in comparative perspective. Social Science Quarterly, 86, 955-983.
59. Sullivan, M.W. and Lew is, M. (2003) Contextu al determinants of anger and other negative expressions in young infants. Dev. Psychol. 39, 693 –705 60. Suzuki, S. (1997) Effects of number of alternatives on choice in humans.
61. Suzuki, S. (1999) Selection of forced - and free-choice by monkeys (Macaca fascicularis). Percept. Motor Skills 88, 242 –250
62. Tajfel, H., and Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin, & S. Worchel (Eds.), The social psychology of intergroup relations, 33-37
63. Torgler, B., Schaffner, M., Frey, B. S., Schmidt, S. L., and Dulleck, U. (2008). Inequality Aversion and Performance in and on the Field (N o. 2008-18). Center for Research in Economics, Management and the Arts (CREMA). 64. Trump, K-S. (2014). Public opinion and economic inequality: Acceptance or
ignorance?” Paper Presented at the Annual Meeting of the American Political Science Association, Washington D.C.
65. Walster, E. (1964). "The temporal sequence of post-decision processes." Conflict, decision, and dissonance, 112-128.
66. Wicklund, R.A., and Frey, D. (1981). "Cognitive consistency: Motivational vs. nonmotivational perspectives." Social cognition: Perspectives on
everyday understanding, 141-163.
67. Wilkinson, R., and Pickett, K. ( 2010). The spirit level. Why equality is better for.
