This product reminds me of myself : how social identity packaging influences consumer behavior

This product reminds me of myself: how social identity

packaging influences consumer behavior

Maud Hink

10759751

Master Thesis MSc in Business Administration – Digital Business

University of Amsterdam | Amsterdam Business School

Supervisor: Javier Sese

Abstract

In light of the growing interest in product packaging as a communication tool, this study set out to identify how packaging displaying social in-group identities may positively influence consumer behavior. Purchase intentions, brand attitudes and willingness to pay are researched as the brand outcomes, as these variables are known to possess a high managerial relevance. Moreover, by manipulating the level of self-construal, it is tested whether individuals with an interdependent mindset enjoy a greater influence of the social in-group packaging compared to individuals with an independent self-construal. After conducting an experimental survey with a 2 x 2 factorial design, the results indicate that a moderation effect of self-construal is not present. Nevertheless, positive effects were found for the relationship between in-group identities and the dependent variables. This indicates how managers should pay attention to the inclusion of social identities while choosing a branding strategy for their products.

Keywords: social identity theory, self-construal, product packaging, product symbolism, consumer behavior

Tables of content

I.

INTRODUCTION ... 6

II.

LITERATURE REVIEW ... 8

2.1 Effects of product packaging on consumer behavior ... 9

2.2 Product symbolism... 10

2.3 Social identity theory ... 11

2.4 Self-construal ... 13

2.4.1 The concept ... 13

2.4.2 Cultural self-construal ... 14

2.5 Summary ... 15

III.

CONCEPTUAL FRAMEWORK ... 16

3.1 Conceptual model ... 16 3.1.1 Dependent variables ... 16 3.1.2 Moderator ... 17 3.2 Hypotheses ... 19 3.2.1 Main effects ... 19 3.2.2 Moderation ... 22

IV.

METHODOLOGY ... 23

4.3.2 Measurement and procedure ... 26

4.3.3 Results ... 27 4.4 Main study ... 28 4.4.1 Research design ... 28 4.4.2 Participants ... 29 4.5 Measures ... 30 4.5.1 Independent variables ... 30 4.5.2 Moderator ... 31 4.5.3 Manipulation checks ... 31 4.5.4 Dependent variables ... 33 4.6 Procedure ... 33

V.

RESULTS ... 34

5.1.1 Sample... 34 5.1.2 Randomization check ... 35 5.2 Reliability analyses ... 35 5.3 Manipulation checks ... 36 5.4 Descriptive statistics ... 38 5.4.1 Correlations ... 38 5.4.2 Data characteristics ... 39 5.5 Hypotheses testing ... 41 5.5.1 Assumptions ... 41 5.5.2 Regression analyses ... 42

VI.

DISCUSSION... 48

6.1 Interpretation of the results ... 48

6.2 Theoretical contributions ... 51

6.3 Practical contributions ... 53

6.4 Limitations & suggestions for future research ... 54

Conclusion ... 57

References ... 58

I. INTRODUCTION

In this era, packaging serves a function that goes beyond merely being a container for goods. When diving into the magical world of Youtube, one will find many hysterical teenagers opening all sorts of packaging in their ‘unboxing videos’. Across many industries, these videos tend to be generally popular. This is not surprising, given the fact that consumers are not solely interested in the product anymore. Consumers want the full experience and research shows that packaging may play a key role in fulfilling this desire (Joutsela, Latvala, & Roto, 2017). From a managerial perspective, the importance of packaging prevails from the fact that the share of product packaging in product satisfaction has risen from 18% in 2013 to 26% in 2015 (Westrock, 2015). Additionally, it has been found that 65% of the respondents tries a new product because of new packaging, whereas 35% completely changes brands because of new packaging (Westrock, 2017). Moreover, while managers are pressured by reduced traditional mass-advertising budgets and rapidly intensifying competition, research addresses that it is packaging, that may aid brands in cutting through the shelf clutter (Underwood, 2003).

Product packages may help brands with standing out from the crowd, as they often possess symbolic elements through which they communicate messages to its end users (Becker, Van Rompay, Schifferstein, & Galetzka, 2011). For example, for food products, the symbolic element ‘shape’ is known to communicate health messages (van Ooijen, Fransen, Verlegh et al., 2017). With the growing body of literature viewing packaging as a communication vehicle, it is also noted that the symbolic function of packaging allows individuals to reflect their identities in products (Underwood, 2003). The congruity between a product and one’s self-image has been shown to provide many benefits for brands, such as an increase in purchase intentions (Landon, 1974). So far, the literature has extensively investigated this relationship in terms of individuals’ personal identities (e.g. Sirgy, 1982). However, while consumers are

known to possess both personal and social identities, relatively little is known in the field of social self-image congruity within products. It is considered interesting putting effort into filling this gap, as Hogg, Terry and White (1995) describe how social identities are prone to guide behavior and influence evaluations. They add that the influence on behavior happens as individuals focus on the relationship between them and their social in- and out-groups. Accordingly, the boundaries of those groups characterize certain behaviors, thoughts and feelings (Brewer & Gardner, 1996). The relational focus on others is shared with one level of construal, another view on the concept. Where individuals with an independent self-construal view themselves as autonomous and unique, the interdependent self-self-construal focuses on relations and connections with others (Markus & Kitayama, 1991). Despite the relatedness of social identity theory and the interdependent level of self-construal, research has yet to examine these effects in a product related setting.

By creating a normal and social in-group related packaging, this paper will examine the effects of social in-group identities on purchase intentions, brand attitudes and willingness to pay. It is expected that consumers who feel part of their in-group after seeing the packaging, will report more positive outcomes for the dependent variables in contrast to individuals who are presented with the normal packaging. In addition, it will be tested whether self-construal is able to moderate these effects by priming either an independent or interdependent self-construal. This study proposes how individuals with higher levels of interdependency enjoy greater effects from the in-group related package. In order to fill the research gap in the literature, this paper will put effort into answering the following research question: “To what extent is social in-group related packaging able to influence brand outcomes and how is this effect moderated by self-construal?”

The present study builds on and extends the theoretical and managerial understanding of package design and buying behavior, especially in the context of social identity theory. First of all, this research adds to the growing body of literature that emphasizes the role of packaging as a communication vehicle (Underwood, 2003). Also, it provides additional support for the established relationship of self-image congruity in products, in particular for social self-images (Sirgy, 1982). On top of that, insights will be added to the within culture understanding of self-construal. This finding is particularly interesting from a managerial perspective, as it may aid managers with the introduction for a specific product across different cultural segments. As packaging may serve as an activator of particular social in-group identities, this research may also help with segmenting the product to the right audience.

In order to gain a better understanding of the different concepts, the following chapters will commence with synthesizing reported literature on product packaging, social identities and self-construal. Second, a detailed description of the conceptual framework will be provided, followed by the proposed hypotheses. Subsequently, the methodology will be discussed, after which the results will be presented. These results will be elaborately evaluated in the discussion section, where the theoretical and managerial contributions will make their appearance. Lastly, this paper will stimulate further exploration of the topics by addressing the limitations of this study and providing suggestions for further research.

II. LITERATURE REVIEW

This section will synthesize past literature on product packaging and its effects on consumer behavior, product symbolism, social consumer identities and self-construal. It will explicate major insights derived from these past studies and identify the research gaps this study is aiming to fill.

2.1 Effects of product packaging on consumer behavior

As of today, an increasing number of research highlights the importance of studying product packaging. Reasons include a need for standing out due to growing competition intensity (Baruk & Iwanicka, 2016), increased levels of market awareness of final buyers (Ahmad & Juhdi, 2010), rising expectations and needs of final buyers (Homburg, Wieseke, & Bornermann, 2009) and a need for protecting the natural environment (Rokka & Uusitalo, 2008) through purchasing sustainable products (Magnier, Schoormans & Mugge, 2016). Moreover, besides its original functions aimed at logistics and protecting the product against potential bacterial damage (Prendergast & Pitt, 1996; Stewart, 1995), a growing body of literature is respecting product packaging as a marketing communication tool (Underwood, 2003; Wells, Farley, & Armstrong, 2007; Rundh, 2009). From a managerial point of view, packaging is discussed as being the ‘silent salesmen’ for brands, as it has become the communicator of product qualities, features, benefits and brand image (Ampuero & Vila, 2006; Becker et al., 2011). Also, there is an increased recognition that packaging can create differentiation and identity, particularly for relatively homogenous goods (Ampuero & Vila, 2006).

The communicative function of packaging tends to hold several implications for consumer behavior. First of all, research demonstrates how packaging may increase consumers’ attention to the brand, which is an important factor in forming the consumers’ consideration set and making it easier for consumers to find products (Underwood, Klein, & Burke, 2001; Rundh, 2009). Moreover, packaging has the ability to effect consumers’ processing of product information, which in turn will have a positive effect on willingness to pay (Van Ooijen, Fransen, Verlegh, & Smit, 2016). Also, it is known that when packaging is more appealing, it will increase consumers’ valuation of products (Ghoshal, Boatwright, & Cagan, 2009). Where such valuations are used in the formation of attitudes, it is the effect of

these attitudes which may influence consumers’ perceived food product quality and brand preference (Wang, 2013; Magnier et al., 2016). However, despite the increased attention in product packaging design over the last few years, relatively little is known about this topic when compared to its relevance to buyer decision making, brand building, and overall marketing success (Bloch, 2011).

2.2 Product symbolism

The communicative power of packaging finds its origins within McCracken’s (1986) work on meaning transfer. Research shows that product packaging is an inherently multidimensional concept, consisting of various symbolic elements such as shape, color, logo, size, graphic design, material, texture, etc. (Underwood et al., 2001). McCracken (1986) proposes how symbolic meaning may transfer from consumer products to the individual consumers themselves. Such symbolism may portray communications such as convenience, environmental consciousness, health consciousness, family, and more (Underwood, 2003). Ease of use and convenience, are examples of symbolic messages which are known to positively influence purchase intentions (Silayoi & Speece, 2007). Color is argued to be an important symbolic cue for achieving immediate product recognition and consumer satisfaction (Piqueras-Fiszman & Spence, 2011). Shapes are shown to serve as a signal for product healthiness (Van Ooijen et al., 2017) and visual elements may impact perceived food product quality and brand preference (Wang, 2013). Hence, consumers tend to use symbolic cues for drawing inferences about the contents of products and this process aids them in their point-of-purchase decision making (Becker et al., 2011; Silayoi & Speece, 2007).

More importantly, it is demonstrated that symbolic meanings of product packaging may also present consumers with ways of expressing themselves (Elliott & Wattanasuwan, 1998). Consumers are prone to consume products in which they evaluate themselves and which reflect

parts of their identity (Claiborne & Sirgy, 1990). Their possessions may portray essences of their individuality or connection to others (Elliott & Wattanasuwan, 1998). Moreover, research shows a positive relationship between consumers’ personal self-image and the products they buy (Sirgy, 1982; Belk, 1988). The findings of congruity between the personal self and products have supported by many researches (e.g. Escalas & Bettman, 2005, Kang, Tang, Lee, 2015). In line with these findings, it is argued that symbolic consumption also helps consumers with categorizing themselves in society (Belk, 1988). Hence, Kleine, Kleine and Kernan (1993) note that products may serve as a facilitating artifact for both personal and social identities.

2.3 Social identity theory

The concept of social identities is well explained within the psychology field of research. Social identities are being described as the social view on the concept of the Self and are the core of Social Identity Theory (Tajfel & Turner, 1979). Social-identity theory entails that people fall in and belong to certain social categories, of which the boundaries of those categories provide a definition of who one is (Hogg et al, 1995). Hogg et al. (1995) describe that people possess a repertoire of category memberships and that the subjective membership into those categories are the purported social identities. A socio-cognitive process which underlies this phenomenon is self-categorization (Turner, 1985). The process of self-categorization emphasizes similarities between members or objects belonging to the same category, while accentuating differences between members or objects falling outside of that category (Hogg et al.,1995). Brewer and Gardner (1996) note that the categorization of the self and others in these in-groups and out-groups provides individuals with their frame of reference for self-evaluation. The authors add that this process happens as individuals are likely to ascribe characteristics and traits to themselves that they share with other members of the in-group, which makes them different from members of the out-group. In turn, the in- and out-group identities prescribe and describe

how one should feel, think, behave and evaluate according to that specific group (Hogg et al., 1995). That is, individuals are likely to respond to contexts in ways which are consistent with their social identity. The literature highlights the importance of in-groups by demonstrating how individuals may conform to their in-groups even at the expense of their own interests (Tajfel & Turner, 1986; Van Vugt & Hart, 2004). For example, even when consumers perceive a product as high quality, they might refuse to purchase this product if it does not sufficiently symbolize the group membership (Klein, Ettenson, & Morris, 1998). In line with these findings, research shows that consumers prefer brands which can serve as signals of memberships in relevant in-groups (White & Dahl, 2007).

The extent to which certain groups (identities) become activated is dependent on the context (Kleine et al., 1993; White, Argo, Sengupta, 2012). The likelihood that out of consumers’ repertoires of identities the most relevant one will be chosen is called salience

(Brewer, 1991). Broader identities such as gender are more likely to be activated than narrow ones such as profession (Oyserman, 2009). Nevertheless, research finds that identity salience may be increased by various contextual stimuli including visual images and words (Hong, Morris, Chiu, & Benet-Martínez, 2000), reference group symbols (Smith & Mackie, 1995) and out-group related symbols (Marques, Yzerbyt, & Rijsman, 1988). This implies how in a retail setting, packaging may serve as the contextual factor which activates a consumer’s in-group identity. However, so far, research has aimed little attention at examining how salient in-group related packaging plays a role in influencing consumer behavior. Despite this lack of research on social identity-oriented buying, a connection exists between belonging to a social group and a person’s attitude and buying behavior (Langner, Hennigs, & Wiedmann, 2013). It is therefore

2.4 Self-construal 2.4.1 The concept

A theory which proposes a different, yet similar view on the concept of the Self concerns self-construal. Alike social identity theory, self-construal entails a form of the self-concept which involves how individuals see themselves. Both theories may systematically influence the cognition, emotion and motivation of a person (Markus & Kitayama, 1991) and are therefore relevant to research in the context of consumer behavior.

Singelis (1994) defines self-construal as the extent to which a person views oneself as being separate from, or interconnected with, others. It is derived from a study conducted by Markus and Kitayama (1991), who distinguish two classifications, namely the independent self and the interdependent self. The authors conceptualize the independent self as autonomous and unique, whereby one’s opinions may be altered by individual thoughts and personal opinions.

Research shows that people with an independent self-construal prioritize autonomy and prefer consistency irrespective of social contexts (Suh, 2002). Monga and John (2008) assert that this is reflected in their thinking style, as independents tend to possess analytic thinking styles whereby the independence of individual objects is accentuated. The authors add that associations related to an object are mainly oriented toward the focal object itself. Research relates this finding to peoples’ goals, as independents generally have a focus on individual-level goals such as individual accomplishments and differentiation (Aaker & Schmitt, 2001; Zhang & Shrum, 2008).

On the contrary, the interdependent self is defined as highly interwoven with others, collectivist and communal, whereby one’s attitudes and behavior are based on the thoughts, feelings, and actions of others in the relationship (Markus & Kitayama, 1991). Researchers proclaim that people who have an interdependent self-construal value their social identities and maintaining strong bonds with their social groups more (Trafimow, Triandis & Goto, 1991;

Aaker & Schmitt, 2001). That is, interdependency is positively associated with the importance of group membership and the self (Sato & Cameron, 1999). Monga and John (2008) assert that interdependent individuals tend to adopt a holistic style of thinking, which emphasizes that the world is constituted by interrelated elements. They add that holistic thinkers focus on the relationship between objects and the field in which they are embedded. This competence of seeing relationships among objects and people, highly relates to the social self-view described in the previous section. However, existing literature shows a lack of research on the integration self-construal and social-identity theory.

The current study researches this concept, as the reported literature increasingly demonstrates the implications of self-construal in both behavioral and retail contexts. It is shown that individuals’ coping with stress is influenced by self-construal (Cross, 1995) and that self-construal may mediate the effect between ethnicity and this coping mechanism (Lam & Zane, 2004). Furthermore, research finds that self-construal effects nonconscious mimicry (Van Baaren, Maddux, Chartrand, et al., 2003) and that it plays a role in the relationship between consistent behavior and well-being (Cross, Gore, & Morris, 2003). Moreover, individuals with an independent self-construal are shown to engage in more impulsive buying behavior (Zhang & Shrum, 2008), while interdependents are better able to perceive a fit between a parent brand and a brand extension (Ahluwalia, 2008). Also, research shows interdependent people have a greater tendency to use price information to judge quality (Lalwani & Shavitt, 2013). Despite the practical relevance that self-construal has within behavior, it remains a topic of modest research activity.

2.4.2 Cultural self-construal

Existing research on construal initially attributes the differences in the two levels of self-construal to culture (Markus & Kittayama, 1991). The distinction of Western and non-Western

perceptions about the self. Accordingly, research finds that interdependency is associated with a collectivist culture while independent self-construals may be more likely to appear in individual cultures (Brewer & Gardner, 1996; Trafimow et al., 1991). Former research has mainly focused on a cross-culture and between-nation comparison of self-construal (Oyserman & Lee, 2008). However, there has been an incremental change in support for the notion how most individuals possess traits from both perspectives (Singelis, 1994). Similar to the social-identity view on the self, the self-construal traits may be triggered by situational cues (Agrawal & Maheswaran, 2005). However, little is known about how self-construal manifests itself within cultures and to what extent it may be effectively primed.

2.5 Summary

In sum, it has been widely addressed that consumers use symbolic product cues to construct their personal and social identities. Reed, Forehand, Puntoni et al. (2012) describe this as “identity driven effects”. As of today, examining these identity driven effects becomes even more relevant as brands need new ways to stand out from the crowd in hyper-competitive markets. Past research notes that this could be done by means of product packaging, especially for which symbolism is the key driver. However, apart from studies focusing on single symbolic elements, research is lacking on packaging conveying full social identities. Therefore, the present study will contribute to existing research by demonstrating how social in-groups embodied by product packaging are able to positively influence consumer behavior. Additionally, it is examined whether this effect may be moderated by self-construal, as the interdependent self-construal is highly related to key aspects of social-identity theory. Examining this possible moderation is meaningful, first because self-construals differ across cultures and second because they influence cognitive processes. Not only will this study add to academic research, as the concepts of social-identity theory, self-construal, product

packaging and identity driven effects are further explored. The findings will also be relevant from a managerial perspective, as these implications may be applied to actual retail settings. In order to investigate the effect of social in-group packaging and possible moderation of self-construal, this paper postulates the following research question:

To what extent is social in-group related packaging able to influence brand outcomes and how is this effect moderated by self-contrual?

III. CONCEPTUAL FRAMEWORK

3.1 Conceptual model

As described in the literature review, it has been shown that recognition of aspects of one’s identity in products has positive outcomes for brands. To extend onto the research of identity driven effect from product packaging, this research proposes that portrayed social in-groups will have stronger effects on brand outcomes in contrast to packaging not displaying such in-groups. In order to test this, the current research will analyze the effects on three dependent variables: purchase intentions, brand attitudes and willingness to pay.

3.1.1 Dependent variables

The first brand outcome studied within this paper concerns purchase intentions. Purchase intentions are described as personal action tendencies relating to the brand (Bagozzi, Tybout, Craig, & Sternthal, 1979). The marketing literature highlights that this construct is a reliable predictor of subsequent purchase and that it provides theoretical and practical insights regarding consumers’ buying decisions (Arts, Frambach, & Bijmolt, 2011). It is argued how this variable possesses strong correlational relationships with brand attitudes, the second

dependent variable of this research. Brand attitudes are defined as individuals’ internal evaluations of a brand (Mitchell & Olson, 1981). This construct has been demonstrated to serve as a driver for consumers’ purchase intent (Spear & Singh, 2004; Fishbein & Azjen, 1977). As brand attitudes are known to possess strong correlational relationships with behavior across multiple contexts, they are considered good predictors of consumer behavior (Mitchell & Olson, 1981; Ibrahim & Najjar, 2008; Michaelidou & Hassan, 2008). Although brand attitudes and purchase intentions are related, research conceptualizes them as separate dimensions which tend to complement each other (Spears & Singh, 2004). Including brand attitudes as a dependent variable is useful in the context of this social identity research, because its value-expressive function allows individuals to express their self-concept (Keller, 1993). Subsequently, the final dependent variable this paper focuses on concerns willingness to pay (WTP). It is defined as the maximum a consumer is willing to pay for a given quantity of a good (Wertenbroch & Skiera, 2002). Willingness to pay relates to the other variables as consumers utilize their WTP as an internal reference price that influences their purchase decisions (Shirai & Bettman, 2005). It complements the other variables by quantifying an estimation of the actual input consumers want to provide (Wertenbroch & Skiera, 2002). From a managerial perspective, WTP may aid managers in the construction of price response functions (Völckner, 2008), as well as providing opportunities for designing optimal pricing schemes such as price differentiation, value-based pricing, or price bundling (Wertenbroch & Skiera, 2002). It is important to marketers as reacting to an increase in WTP in the sale price of a product can lead to higher profits for the company (Dixit, 2007).

3.1.2 Moderator

In addition, this study proposes that the effects on purchase intentions, brand attitudes and willingness to pay may be moderated by either an independent or interdependent self-construal.

This variable is chosen as the moderator, because the literature study outlines that it is a valid self-concept related construct which is prone to influence consumer behavior (Markus & Kitayama, 1991). Moreover, the levels of self-construal are relevant to a wide range of judgements (Oyserman, 2009) and are likely to be chronically accessible during decision making (Markus & Kitayama, 1991). Another reason for investigating this variable resides in the fact that self-construal can easily be manipulated. Even those individuals with generally independent or interdependent self-construals can be induced to take the opposite perspective (Zhang & Shrum, 2008). Therefore, using this variable in an experimental research setting enables the opportunity of finding interesting results. Lastly, it is important to note that due to the dynamic nature of self-concepts, individual’s self-construal may be accessible along with other social identities at a particular moment (Oyserman, 2009). Therefore, when priming individuals with a certain self-construal category, an individual’s social in-group may still become salient. Hence, as these forms of the self may exist in symbiosis, self-construal is considered an appropriate choice of moderator.

Altogether, the following conceptual model is developed to show the relationships between Social Identity, Purchase Intentions, Brand Attitude, Willingness to pay and Self-construal, as seen in Figure 3.1. Subsequently, the following subsections will elaborate more on the dependent variables when forming the hypotheses.

Figure 3.1. Conceptual Framework.

3.2 Hypotheses 3.2.1 Main effects

From the literature study it is known that symbolic product packaging provides a value transmitting function to its end users (McCracken, 1986). This perceived value may be derived from consumers’ self-identification within products (Elliott & Wattanasuwan, 1998). Identification of the personal self in products has been widely researched and is shown to have positive effects for brands (Sirgy, 1982). More importantly, not only does this self-identification aid consumers in constructing their personal identities, it also helps them with categorizing themselves in society and building their social identities (Tajfal & Turner, 1986). Hence, individuals also tend to consume products to construct their social identities (Kleine et al., 1993). Therefore, it is expected that when products exhibit those social identities, consumers will be more favorable towards those products. This proposition is supported by research explicating that social in-groups guide evaluations and behavior (Hogg et al., 2015). This study views the above as the underlying foundation for the each of the three following

hypotheses. With these arguments in minds, this paper will proceed with a detailed explanation according to each of the three dependent variables.

For purchase intentions, it has been shown that recognition of the personal self within products may enhance the levels of this brand outcome (Landon, 1974). Research argues that when purchases elicit certain identities, the value perception consumers have is higher (Richins, 2004; Persaud & Schillo, 2017). Due to a strong and consistent relationship between perceived value and purchase intentions (Grewal et al., 1998), it is expected that products which elicit identities conforming to those of the consumer will increase reasons to purchase the product (Reed, 2002). Additionally, purchase intentions are affected by many factors such as packaging shape (Silayoy & Speece, 2004), visual images (Underwood et al., 2001) and perceived product quality (Mugge & Schoormans, 2012). It is therefore inferred that social identities may be another symbolic factor which is able to influence purchase intentions. Therefore, this paper proposes the following:

H1: Product packaging showing in-group social in-group identities will lead to higher purchase intentions compared to packaging not showing in-group identities.

In line with the effect of personal identities on purchase intentions (Landon, 1974), it is noted that social self-identification is a strong predictor of positive attitudes (Bartels & Hoogendam, 2011). Moreover, it is known that consumers are inclined to generate favorable perceptions towards the in-group (Hogg et al., 1995). As research relates such cognitions to the formation of brand attitudes (Mitchell & Olson, 1981), it is expected that the social in-group packaging may transform these favorable perceptions into positive brand attitudes. Furthermore, when consumers regard these attitudes as important, they are more likely to use them when processing information, forming intentions, and taking action (Boninger, Krosnick, & Berent,

1995). This provides additional reason for the expected positive impact of in-group packaging on purchase intentions. Accordingly, the following hypothesis is formulated:

H2: Product packaging showing in-group social identities will lead to more positive brand attitudes compared to packaging not showing in-group identities.

After reviewing the literature, it is appropriated that equity theory provides a sound theoretical background for proposing the effect of product packaging on willingness to pay. Equity theory describes how allocated justice among exchange partners is gained when the benefits of each are proportional to their investments (Adams, 1965). Hence, people involved in exchanges seek to adjust their input (e.g. money, effort) according to the expected outcome (i.e. the benefit of a product or service) (Huppertz, Arenson, & Evans, 1978). Thus, when consumers receive a greater benefit from the product they are more likely to provide higher monetary inputs. One such benefit may include the ability of products to contribute to one’s personal or social identity (Sirgy, 1982; Elliott & Wattanasuwan, 1998; Escalas & Bettman, 2005). Therefore, it is expected that the social self-image congruity has a positive effect on WTP:

H3: Product packaging showing in-group social identities will lead a higher willingness to pay compared to packaging not showing in-group identities.

All in all, for the purpose of constructing their social identity, it is expected that consumers are more likely to purchase products which reflect their in-group, have more favorable attitudes towards this group and are willing to pay a higher price.

3.2.2 Moderation

It is hypothesized that self-construal will moderate the effect of social identities on brand outcomes. From the literature study it is known that individuals with an interdependent construal tend to include others, such as in-group members, in the construction of their self-identities (Markus & Kitayama, 1991; Brewer & Gardner, 1996; Van Baaren et al., 2003). Compared to independents, interdependent individuals tend to be more flexible and adapt to different social contexts and roles (Singelis, 1994). Moreover, they often adopt holistic thinking styles whereby they are more likely to see relations between objects. Aguirre-Rodriguez, Bosnjak and Sirgy (2012) argue that the identification of the self in products is higher when individuals adopt such a holistic thinking style. From this, it is inferred that for interdependent individuals, in-group cues may be stronger drivers for behavior as compared to independent individuals. Hence, it is suggested that the in-group connection is more likely to occur when there is a greater focus on the relationship with others. Therefore, this paper proposes that an interdependent self-construal will enhance the effects of social in-groups on consumer behavior.

On the contrary, independent individuals place greater emphasis on the self and behave in a manner consistent with protecting their individual self-worth (Heine & Lehmann, 1995, 1997). As they are more autonomous, they are expected to be less likely affected by in-group symbols as compared to interdependents. Moreover, as such individuals often adopt analytic thinking styles and assess objects individually, it is proposed that they are less prone to see the relationship between their social groups and objects. Accordingly, it is hypothesized that when independency is the primed self-construal, this will have a weaker influence on the effects of social in-groups on consumer behavior. Accordingly, the following hypothesis is composed:

H4: The effect of social identities on brand outcomes (H1-3) is moderated by self-construal, such that the effect is stronger when people have interdependent mindsets as compared to independent mindsets.

IV. METHODOLOGY

This section outlines the methodological approach for this study. It will explain the chosen means for operationalizing the constructs and show how they are validated by the pre-test. Further, this section will explicate the used academic scales and how they are incorporated in the main survey.

4.1 Product choice

The majority of packaging studies aim their attention towards food products. Food-related products are considered low-involvement products, which are most often Fast-Moving Consumers Goods (FMCG) (Saunders & Guoqon, 1997; Assael, 1992; Silayoi & Speece, 2004). Researching these types of products is important, because research shows that symbolic product packaging has a bigger impact for low-involvement products as compared to high-involvement products (Brunner, Ullrich, Jungen, & Esch, 2016). Moreover, for low-involvement products, it has been shown that symbolic elements such as visual graphics (Silayoi & Speece, 2004), shape and color (Becker et al., 2011), text (Machiels & Karnal, 2016) and typeface (Celhay, Boysselle, & Cohen, 2015) have an effect on consumer decision making. Within the range of food products, it has been shown that especially supermarket food products are prone to influence behavior due to their stimulating packaging visuals (Underwood, 2003). Also, it has been found that products high on experiential benefits are positively affected by visual elements on packaging (Underwood et al., 2001). Following these reasons, this paper

will concentrate on a low involvement product which is high on experiential benefits and is sold within grocery shops and supermarkets: ice cream. Ice cream is a product with experiential benefits which provides sensory pleasure, variety and cognitive stimulation (Keller, 1993). In their own research, Underwood et al. (2001) use candy as a product high on experiential benefits. Based on this, ice cream is considered a decent similar choice.

4.2 Social in-group

From the literature study it is known that identities vary in the degree to which they are situationally activated and that broader identities may be activated more easily than narrow ones (Oyserman, 2009). For the scope of this thesis and reaching enough respondents per level of the social identity variable, it is decided to focus on a broad identity. This way, it is more feasible to collect a diverse sample whereby all the conditions are represented fairly. Moreover, this allows for not having to collect data from websites such as MTurk, which tend provide lower quality data.

Research shows that one example of a broad identity concerns gender (Bigler & Liben, 2006). Over the years, gender identity has been a focus point of many identity related researches (e.g. Dittmar, 1991; Frable, 1997). Maldonado, Tansuhaj and Muehling (2003) address how the marketing literature has mainly conceptualized gender as a personal identity. They add that looking at gender identity from a social point of view, provides a broader approach to understanding the influence of gender. This is in line with research done by Risman (1998), who postulates how gender should be looked at from a multilevel perspective, as it is more than solely the property of individuals. In the research on gender identity, it has been found that women are more likely to define the self as embedded in, contexts and relationships (Oyserman, 2009). It is shown that the feminine gender identity exhibits behaviors consistent with communal goals focused on members of an in-group (Winterich, Mittal, & Ross, 2009).

Such communal goals entail a focus on social relationships, interpersonal affiliation and harmony with others (Bakan, 1966). Accordingly, as women are generally more associated with social and relational contexts, it is expected that they will more easily identify with a gender related in-group as compared to men. It is therefore decided to choose the female gender as the social in-group.

What is worth noting, is that some researchers argue how gender may affect self-construal (Ahluwalia, 2008; Markus & Kitayama, 1991). This proposition holds because men and women may adopt contrasting approaches regarding cognition, self-related motivations and emotions (Cross & Madson, 1997). However, newer research finds no significant interactions between self-construal and gender and it is therefore assumed that gender will not impact the effects of self-construal (Winterich & Barone, 2011). Hence, it was expected that gender would not distort the proposed effect of gender in-groups on the brand outcomes.

4.3 Pre-test

4.3.1 Research design

The main experiment addressed the possible effects of social identities in product packaging on consumer behavior. In order to test this, two types of ice cream packaging were designed, one for each level of the social identity in- and out-group (Appendix A). Thus, one design resembled the female in-group, while the other portrayed the non-female outgroup. The packages were designed by an art student from the Willem de Kooning academy and resembled a fictive brand. The use of a fictive brand had the advantage that confounding variables such as brand image and pre-existing attitudes towards the brand were ruled out (Van Ooijen et al., 2016). Such pre-existing attitudes could have influenced the dependent variables, such as brand attitudes. As different colors, fonts and images were chosen for the packaging, all remaining package elements (shape, fair trade logo,size) were designed to be as homogenous as possible.

As it was of high importance that participants associated both types of packaging with the right social group, a pre-test was conducted. Moreover, the pre-test allowed for validating ice cream as the chosen product, amongst the conditions of low-involvement and being high on experiential benefits.

4.3.2 Measurement and procedure

This pre-test included two conditions, one for each of the levels of the social groups. Within this pre-test (N=21), respondents first had to rate the packages in terms of female associations. This was done on a three-item seven-point scale adopted from Brunner et al., (2016) containing the following items: rough/gentle, hard/soft and masculine/feminine (α = 0.843). Furthermore, to assess the extent to which participants thought of themselves in terms of their gender identity after exposure to product packaging, the twenty statements task (TST) was employed (Kuhn & McPartland, 1954). Within this task, respondents had to complete up to 20 self-descriptive statements following the sentence “I am___”. Statements that were completed with feminine nouns such as ‘female’ or ‘mother’ represented female associations, while other nouns such as ‘teacher’, ‘happy’, did not. Hence, for the female in-group packaging it was expected that the female associations would appear more frequently than for the normal packaging. This task was implemented as research has demonstrated that it is a valid measure of people’s momentary self-conceptions and because it was used reliably within self-construal studies (Gardner, Gabriel, & Lee, 1999; White et al., 2012).

Moreover, the pre-test included a measure to validate whether ice cream actually was considered a low-involvement product with experiential benefits. Both characteristics are important because the literature has shown that symbolic functions of product packaging have the most impact for high experiential and low-involvement, Fast Moving Consumer Goods (Underwood et al., 2003; Cacioppo and Petty, 1984). Whether ice cream was indeed high on

experiential benefits, was measured along a three-item, seven-point Likert-type scale. It was adopted from Laurent and Kapferer (1985) and was used reliable previously (Cooper-Martin, 1991). This scale consisted of the following questions “For me, ice cream is a real pleasure; Buying ice cream, it's a way to reward yourself; and I don't particularly like ice cream” (α=.882). Before analyzing the results, the last question was recoded in order to measure the same thing as the other two questions. To assess whether ice cream was perceived as a low-involvement product, a four-item, seven-point purchase-decision low-involvement scale was adopted from Mittal (1989) (α=.875). The four items of the scale had to be ranged on different dimensions (Appendix B). One example of a question includes: “In selecting from the many types and brands of this product available in the market, would you say that”, where the response could range from “I would not care at all as to which one I buy” to “I would care a great deal as to which one I buy”. This scale was chosen as it measured consumers’

involvement in a decision-making setting, rather than solely in a product-class setting (Mittal, 1989).

4.3.3 Results

For both the pre-test and the main study, all calculations were performed using the statistical software program SPSS. Firstly, the scale for measuring the female associations regarding the two packages was considered reliable, as it reported a Cronbach’s α of .843. By means of conducting a one way Analysis of Variance (ANOVA), the results indicated that that there is a significant difference in the perception between the two packages (F(1,20) = 26.453, p<.001). The in-group packaging was perceived as having more female associations (M=6.24) as compared to the out-group packaging (M=4.18). Additionally, the outcomes of the TST supported this finding (Appendix H). When comparing the second condition which included the normal packaging, respondents reported more female associations in the first condition

which displayed female packaging. In the first condition, 54.5% of the respondents completed their first response for the “I am__” statement with female associated nouns. This was 0% in the second condition with the normal packaging. Moreover, in condition one, one male responded how this packaging reminds him of females. It was therefore concluded that the female and normal packaging represented the in- and out-group well. Nevertheless, to further increase the distinctiveness between the packages, personal feedback towards optimizing the design of the female packaging was received. As a result, the female packaging was improved with the addition of more female elements, such as a white flower pattern. Lastly, the results showed that ice cream is somewhat considered a low-involvement product (M=4.31, SD= 1.05) and fully considered as being high on experiential benefits (M=5.22, SD=.79). Hence, it was concluded that the designed stimuli were appropriate measures and that they would be embedded in the main study.

4.4 Main study 4.4.1 Research design

The data of this study was collected by means of an experimental survey (Appendix I). As the effects of different social groups on brand outcomes had to be compared, a quantitative method would provide the most accurate measurements. The dependent variables purchase intentions, willingness to pay and brand attitude may be measured in many different ways. As the literature study has shown that these constructs can be significantly measured by existing scales, a survey was chosen for embodying the experiment. The survey was created by the online tool Qualtrics. To test the hypotheses, the experiment was set up with a 2 (product packaging: female in-group vs, non-female out-group) x 2 (self-construal: independent vs. interdependent) between-subjects design. In each of the conditions, the participants were presented the same

questions. Accordingly, the conditions differed in the stimuli materials for the independent variable and the moderator.

4.4.2 Participants

In order to avoid a bias in responses, individuals who participated in the pre-test were excluded from the main study. It is important to note that only female respondents were considered for running the statistical tests. When including males in these tests, the results would have been biased. The reason for this is that males were highly unlikely to associate with the female in-group, so their responses towards the female packaging would presumably be much more negative compared to females. Meaning that the female responses would have been evened out. However, even though the analyses were run only for female participants, this study aimed for reaching male respondents too. As the independent variable had the ‘normal’ packaging representing the out-group, no control condition was used. By incorporating men in the analyses, it could be tested whether differences existed between participants who could potentially associate with the in-group (females) and participants who could definitely not (males). For the sample, it was further preferred to have an equal mix of nationality, age and occupation category. For trying to achieve this, the survey was distributed across diverse social media channels, by e-mail or personal approach. The respondents were gathered by means of convenience sampling and were found in diverse sectors including universities, a hospital, an online marketing agency and a telecom company. Also, the respondents had to have experience with grocery shopping to enhance the visualization of a point-of-purchase decision. The survey could be taken by smartphone, tablet and computer, in order to enhance the likelihood of response.

4.5 Measures

4.5.1 Independent variables

Social in-group. The independent variable, social identity, was constructed by two levels. The first level represented the female in-group and the second level portrayed the out-group. As known from the pre-test, these two levels were embodied by two types of packaging which were shown to represent the correct social group. For comparing the means between the female and normal packaging, a dummy variable for each of the two levels was created. Hereby 0 represented the normal packaging and 1 stood for the female packaging. Additionally, a scale was used to measure whether females felt part of the in-group after seeing the packaging. This scale was also used as manipulation check and will be discussed in the following subsection. Furthermore, this scale represented the ‘in-group’ variable in the regression analyses. In this way, it could be measured how a sense of belonging to the in-group would influence the brand outcomes.

Control variables. For ruling out the chance that other factors served as possible explanations for a change in the dependent variables, control variables were added into the regression model. These included age, nationality, occupation type and the how often people purchase ice cream. As the regressions were run for females only, gender was not added as a control variable. The age question received responses by means of an open question and was implemented in the regression as a scale variable. The nationality variable was coded as a dummy variable whereby 0 = Other and 1 = Dutch. The occupation type was a nominal variable with 1 = Student, 2 = Student working part time, 3 = Employee and 4 = Other. Lastly, the purchase frequency of ice cream was added for checking whether the responses would be influenced by how often people purchase ice cream. This is done because it could be that people who buy ice cream very often have higher purchase intentions in general. Alike the age

variable, this variable consisted of ratio level data and was implemented in the regression as a scale variable.

4.5.2 Moderator

Self-construal. It is argued how most individuals possess traits from both self-construal perspectives, whereby these traits may be triggered by situational cues (Agrawal & Maheswaran, 2005). Therefore, this experiment will manipulate the level of self-construal, by priming respondents to fall in either one of the two categories. The respondents are primed according to a method employed by Trafimow and his colleagues (1991). Within this method, participants are instructed to take 5 minutes to think about and write down how they are similar to (interdependent) or different from (independent) their friends and family. In the interdependent condition, participants are suggested to think of themselves in terms of social roles and relationships with people, which will activate their dependence on others. Respondents in the independent condition will more likely think of themselves as being different and unique, which will make them feel more independent. For increasing the effectiveness of the priming material, pictures relevant to the primed situation were displayed above the question. For the interdependent item a group of friends was displayed, while respondents in the independent condition saw a highly distinguishable penguin standing in a large group of similar penguins (Appendix C). Participants were instructed to look at the picture before answering the open question.

4.5.3 Manipulation checks

For checking whether the designed package was correctly associated with the corresponding in- or out-group, a variant of the three-item scale used by Swaminathan et al. (2007) was adopted. The Likert-type scale consisted of the following items "This product makes me feel

connected with other females," "This product reminds me of females," and "This brand is a statement of how I am female" (α = .845). The Likert-type scale had seven points ranging from 1 = “strongly disagree” to 7 = “strongly agree”. It is worth noting that other researchers evaluating social identities have described the social identity construct as consisting of a cognitive, affective and evaluative aspect (Bagozzi & Dholakia, 2006). It was observed that these studies applied scales with questions representing these three aspects. When comparing such scales with the questions utilized within this research, it could be seen that both measurement tools were highly similar. Therefore, it is concluded that even though this research does not review social identity in terms of the three separate aspects, the scale used in this research represents these aspects well.

In order to test whether the priming of self-construal was effective, this research has adapted a manipulation check method used by White et al. (2012). For this method, the authors use one question of both the independent and interdependent self-construal scales constructed by Singelis (1994). The researchers Escalas and Bettman (2005) used other self-construal items, which were chosen based on having the highest corrected item-total correlations in Singelis’ scales. Based on the authors reporting a relatively low Cronbach’s α as compared to White et al. (2012), it is decided to combine both approaches. Additionally, the factorial loadings of the scales as reported by Singelis (1994) were considered. Combining the above led to the construction of two self-construal scales based on the following items: “It is important for me to maintain harmony within my group” and “I often have the feeling that my relationships with others are more important than my own accomplishments” (α = .820) for the interdependent scale and “My personal identity independent of others is very important to me” and “I enjoy being unique and different from others in many aspects” (α = .648) for the independent scale. Both were seven-point, Likert-type scales and had to be ranged from 1 = “strongly disagree” to 7 = “strongly agree”.

4.5.4 Dependent variables

Purchase intentions. To measure purchase intentions, a three-item scale was adopted from Grewal et al. (1998). This seven-point, Likert-type scale consisted of the questions "The probability that I would consider this product is", "The likelihood that I will purchase this product is", "When buying ice cream, the probability of buying this brand is". This scale showed high reliability, with a Cronbach’s α of .963 The corrected item-total correlations indicated that all the items had a good correlation with the total score of the scale. None of the three items would have substantially affected reliability if they were deleted.

Willingness to pay. WTP was measured by asking participants for the price they were willing to pay using the question ''When we ask you to buy this ice cream in the next three months, how much would you be willing to pay for it?". Participants were asked to indicate an amount between 0.00 and 7.00 Euros on a horizontal slider scale with intervals of one cent. This scale was used reliably before (Miller, Hofstetter, Krohmer, & Zhang, 2011).

Brand attitudes. Adjusted from the work of Swaminathan et al. (2007), brand attitudes was measured along a 7-point, semantic differential scale with 4 items (low quality/high quality, bad/good, unfavorable/favorable, negative/positive) (α = .940). These questions were embodied by a slider scale, with intervals of 1.

4.6 Procedure

The survey had to be taken only online. Participants were approached with a short message about the purpose of this survey and the question whether they would like to participate.

When opening the survey, all participants were randomly assigned to each of the conditions. Hereby, they were unaware of the other conditions nor saw the other conditions. Before the start of the survey, respondents were asked to give their consent to voluntarily participate in the survey. Subsequently, they were informed that the purpose of this study was

to help a startup company by evaluating a new supermarket product that was not available on the market yet. Throughout the survey, the respondents were forced to leave a response for all questions in order to avoid missing data. The participants were told that in light of learning about the startup’s potential consumers, the research begun with priming the respondents with either the independent or interdependent self-construal. This was followed up by a manipulation check to test whether the priming appeared successful. Along these lines, the manipulation check for the female in- and out-group was performed afterwards in order to test whether the findings of the pre-test would prevail within the main study. The survey continued with presenting participants the questions about the ice cream packaging, which measured the dependent variables. A picture of the package was presented above every question. As a validation check, the survey contained a question whether participants could guess the true purpose of this study. Lastly, the participants are thanked for their cooperation and informed of the opportunity of leaving their email address for receiving the results of the study.

V. RESULTS

5.1 Data screening 5.1.1 Sample

After distributing the sample across multiple channels including e-mail and social media, a number of 221 respondents opened the survey. From the 186 respondents who filled out the questions, 10 respondents were removed from the sample for not completing the full survey. This resulted in a completion rate of 79.6%. In the previous section it was described that the survey included a question for examining whether respondents could guess the true purpose of the study. The responses showed that all participants were naïve as to the purpose of the study and no additional person

had to be removed from the sample. Lastly, not one respondent reported never doing grocery shopping, meaning that all respondents were able to imagine the point-of-purchase decision situation implied in this survey. Within this final sample, the participants were aged between 20 and 58 (M=32.2, SD=10.5) of which 56.1% was female (M=1.56, SD=.48). The majority of the respondents was employee (64.8%). Importantly, the sample can be described as Western, as 89.1% of the respondents had a Dutch nationality.

5.1.2 Randomization check

In order to test whether the randomization of participants across the conditions was successful, it was tested whether the conditions significantly differed in gender, age and occupation type. Due to an effective set-up of the survey in Qualtrics, participants were already categorized by the variable ‘Condition’, which consisted of four levels (C1 = interdependent/female, C2 =

interdependent/normal, C3 = independent/female, C4 = independent/normal). An ANOVA test showed that age was normally distributed (F(3,172)=.497, p>.05), where homogeneity of variances was assumed. As gender, occupation category and the condition variable all were categorical variables, a Chi-Square goodness-of-fit test was employed for checking random attribution across the conditions. For gender, there was no statistically significant difference, χ2_{(3)=2.258, p>.05. Therefore, the null hypothesis was not rejected and an equal gender} distribution was assumed. Also, occupation type was shown to be equally distributed as no statistical difference was found, χ2_{(9)=8.039, p>.05.}

5.2 Reliability analyses

Before further analyzing the data, the used scales were tested for reliability. The manipulation check scale which tested whether the different packages represented a female social in-group had high reliability, with Cronbach’s α being .845. The corrected item-total correlations

indicate that all the items have a good correlation with the total score of the scale. None of the three items would have substantially affected reliability if they were deleted. The scales for testing the self-construal manipulation were adopted from Singelis (1994). The questions used to measure an individual’s independency were recoded as to measure the same construct as the interdependent scale did. Taken together, these four questions were shown to possess a Cronbach’s α of .762, concluding that this was a reliable scale that could be used for analyzing the data.

Also, the scales used for the dependent variables were shown to possess excellent reliability, as the purchase intention scale reported a Cronbach’s α of .963, while the brand attitude scale showed a value of .940. As WTP was measured by one question only, it was not possible to conduct a Cronbach’s α reliability test. However, as this measure was used reliably in previous research (Miller et al., 2011), it was assumed that it was valid to derive conclusions from this question.

5.3 Manipulation checks

This research manipulated both the two levels of the independent variable, as well as the two levels of self-construal. In order for this study to validly derive results from the different conditions, it was tested whether the manipulations were effective.

For the independent variable, two one-way Welch ANOVA tests were conducted. The first test examined the effectiveness of the stimulus material by checking whether females who saw the female packaging scored higher on the in-group scale as compared to females who saw the normal packaging. Equal variances assumed, the results revealed a significant difference across the four conditions (F(3,95)=23.392, p<.001), meaning that the designed packages were able to differentiate between the in-group and the out-group (Figure 5.1). For further evidence, it was tested whether there was a significant difference in belonging to the in-group between

females and respondents who were highly unlikely to associate with the in-group, which were men. An ANOVA test was run for males and females who saw the female in-group related packaging. These results revealed that there was a highly significant difference (F(1,80)=21.446, p<.001), where equal variances were assumed This indicated that the in-group related packaging specifically worked for the associated in-in-group and that it is likely to have low effects for people outside of that in-group. Hence, the designed stimuli materials were effective in manipulating the respondents for the purpose of this research.

Subsequently, the self-construal manipulation was validated. To perform this analysis, the total four-item interdependent scale described in the previous subsection was used. From these four questions, a mean variable for the complete interdependent scale was computed. Together with the ‘conditions variable’, this new variable was analyzed in a Welch ANOVA test where equal variances were not assumed. The results indicate that the self-construal manipulation worked, as there was a significant difference between the interdependent means across the conditions, F(3,172)=9.511, p<.001, (Figure 5.2). Accordingly, with both of the manipulations being statistically significant, it was concluded that the four conditions were successfully implemented and that the data could be analyzed according to these conditions.

Figure 5.2 Means of the interdependent scale for the four conditions

5.4 Descriptive statistics 5.4.1 Correlations

In Table 5.1 the means and standard deviations for the tested variables are presented for the female sample (Cronbach’s ɑ on the diagonal). Additionally, bivariate correlation tests were run for both the full and female sample. Firstly, this revealed that, as intended, gender correlates with the in-group packaging (r =.463, p<.01). This means that there was a positive relationship between females and feeling part of the female in-group after seeing the packaging. Self-construal was found not to be correlated with any of the other variables, which is especially interesting for the gender variable (r = .097, p>.05). This was in line with the findings of statistical models used by Winterich and Barone (2011), which showed that gender does not interact with self-construal. As expected, the in-group variable significantly associated with the dependent variable outcomes purchase intentions (r =.662, p<.01), willingness to pay (r =.632, p<.01) and brand attitude (r =.486, p<.01). All these relationships were found to be

moderate to strong. An unexpected negative significant relation between self-construal and willingness to pay was found. This indicates that individuals who are independent tend to have higher purchase intentions. Lastly, the correlation analysis showed that all of the dependent variables were related to each other. This preliminary finding was in line with the theoretical framework as previous research indicates the relatedness of the three dependent variables.

Table 5.1 Means, Standard Deviations, Cronbach's α

M SD 1 2 3 4 5 6 7 8 9 1. Age 31.54 10.97 - 2. Nationality (Dummy) .90 .30 -.159 - 3. Occupation 2.64 .80 .506** .015 - 4. Purchase frequency 2.48 .212 .044 -.193 -.045 - 5. In-group 3.33 1.82 -.063 .043 -.151 .184 - (.845) 6. Self-construal 5.04 .77 -.161 -.115 -.081 -.011 -.001 - (.762) 7. Purchase intentions 3.04 1.65 .100 -.020 -.088 -.022 .662** -.089 - (.963) 8. Brand attitudes 3.58 1.38 -.113 .086 -.185 -.010 .632** -.071 .763** - (.940) 9. WTP 2.49 1.30 -.059 .021 -.103 .032 .486** -.208* .611** .649** -

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

5.4.2 Data characteristics

Table 5.2 shows the means and standard deviations per condition for females. Before running the main statistical tests on the data, these results indicated some initial effects. Firstly, the data showed that participants generally did not perceive both packages as largely positive. For the normal packaging all the dependent variables were shown to have a mean far below the Likert-scale average of 4 across the two levels of self-construal. This result was also prevalent for the female packaging, where brand attitudes had the highest means for both the independent and interdependent condition.

Secondly, when comparing the means of the female packaging to the normal packaging, it could be seen that respondents generally had more positive purchase intentions, willingness to pay and brand attitudes for the female packaging. This was shown by higher means for every dependent variable across both of the two levels of self-construal for the female packaging. The responses of females who considered themselves as part of the social in-group were strong