How does Facebook impact other social media companies?

53  Download (0)

Full text


How does Facebook impact other social media


Spillover effect of brand scandals from Facebook to subsidiaries and competitors

Author: Ivo Coutinho Student number: 11321105 Date of submission: 30-06-21

Program: BSc. Business Administration Track: Management in the Digital Age

Thesis supervisor: dr. R. de Bliek ABSTRACT

Facebook has been plagued by multiple scandals throughout its history, one of which is the Cambridge Analytica scandal. This thesis looks at the relationship between Facebook, their subsidiaries (WhatsApp and Instagram) and their competitors in the social media industry (Snapchat and TikTok). To this end this thesis examines to what extent there is a spillover effect on either subsidiaries or competitors after a brand scandal. 38 participants were recruited and divided into two different groups: either presented with a filler article or an article about the Cambridge Analytica scandal. The results show no evidence for a spillover effect, however this could be due to customers already having a low baseline trust in Facebook (and other social media companies).

Keywords: brand scandals, spillover effect, brand trust, brand loyalty


Statement of Originality

This document is written by Ivo Coutinho who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.


Table of contents

1. Introduction 4-7

2. Theoretical Framework

2.1 Brand Scandals 8-11 - What is a brand scandal?

- Product Harm Crisis - Effects of a brand scandal - Minimizing the impact of a brand scandal

- Types of brand scandals

2.2 Brand Scandal Spillover 12-14 2.3 Brand Trust 14-15 2.4 Brand Authenticity 16 2.5 Brand Loyalty & Willingness to use 17-18

3. Methods and Data

3.1 Research Design 19-20 3.2 Sample 21

3.3 Variables 21-22 3.4 Analyses 22-23 4. Results

4.1 Correlation 24 4.2 Hypothesis Tests 25-29 5. Discussion

5.1 Implications 30-31 5.2 Limitations 31-32 6. References 33-39


1. Introduction

It all started with a simple idea from a Palantir (US-based Big Data Analytics firm) employee: create an app to get access to valuable data about Facebook users and their network of friends on the social media platform. Cambridge Analytica later went on to implement this idea. (Confessore & Rosenberg, 2018a). They used it to obtain the data of over 87 million people, of which the vast majority did not explicitly give consent to do so. (Revell, 2018). The data was collected through the guise of a personality test and with the help of Michal Kosinski, a researcher at Cambridge University. This data was then primarily used to create psychological profiles of users to influence the US elections of 2016. (Confessore & Rosenberg, 2018a). While Cambridge Analytica filed for

bankruptcy in 2018 (Confessore & Rosenberg, 2018b), Palantir is still in business and has even been listed on the New York Stock Exchange since September 30, 2020. (Levy, 2020). Peter Thiel, one of the co-founders of Palantir, also serves on the board of directors at Facebook. (Millard, 2018).

Research by Hinds et al. (2020)indicated that contrary to expectations, users of Facebook did not take much action with regards to their Facebook account after the Cambridge Analytica scandal broke out (actions for users include for example deleting their Facebook account or changing their privacy settings). The article further points out that the reasons for this are sometimes even contradictory, for example some users feel as though they “have had enough of Facebook, but can’t leave”. (Hinds et al., 2020, p. 8).

A situation arises where users mention they feel ‘trapped’, or somehow forced to stay on Facebook to keep in contact with their friends on the platform. These findings also appear


to be contradictory with existing literature on perceptions of digital (information) security. According to research by Huang et al. (2011) a low level of perceived security in IT systems can result in users not using these IT systems. After a widespread scandal such as the Cambridge Analytica case, it appears that users’ intention to keep using the platform (in this case Facebook) to keep in touch with their friends overrides their security concerns. (Hinds et al., 2020). A study involving in-depth interviews with college students (N=10) by Brown (2020) showed that in the aftermath of the Cambridge Analytica scandal none of the respondents deleted their Facebook account permanently.

While multiple studies have been done on the specifics of this scandal for Facebook users, there have been no studies to date on the spillover effect of brand scandals on users perception of other social media platforms. In this light it is also important to note that Facebook owns two subsidiaries which are also active in the social media industry or closely related to it (Instagram and WhatsApp).

Even if Facebook users do not quit the platform as a result of these scandals, it can still result in decreased trust in the brand, which could in turn spillover to their subsidiaries and perhaps also to competitors of the brand (social media platforms not owned by Facebook). Decreased trust might practically mean that a hypothetical future scandal could be the catalyst that will deter users from using the platform permanently. In the case of a highly negative spillover effect this has the potential to snowball to other social media platforms (both Facebook and non-Facebook owned). Therefore conducting research in this domain will have significant implications for management. Today many consumer-oriented companies apply the practice of umbrella branding for their portfolio


of products (examples include Unilever, Nestlé, Coca-Cola). (Lei et al., 2008). Facebook itself has also recently started with this practice, a prominent example of this can be seen when launching the Instagram app on a smartphone (see Appendix 1).

A study by Trump and Newman (2016) found a spillover effect from Volkswagen to similar car manufacturers in the aftermath of the Dieselgate emission scandal.

Specifically they found that the unethical perception of Volkswagen spilled over to BMW, a competing German car manufacturer. On the other hand the study found no evidence that the Dieselgate scandal had either a negative or positive spillover effect on the geographically dissimilar competitor Ford (American car manufacturer).

High brand authenticity can offset the negative effects of a brand scandal according to research done by Guèvremont and Grohmann (2017). However their study also indicated that even firms with a higher level of brand authenticity cannot fully absorb all negative effects of a brand scandal (such as willingness to pay, perceived responsibility for the scandal, perceived hypocrisy). Even so, offsetting even a small percentage of the negative effects of a brand scandal can prove to be invaluable for large companies such as

Facebook, marketing and public relations implications of this finding are therefore not to be underestimated.

Based on the above this research aims to investigate the following research question: To what extent is there a spillover effect of brand scandals from Facebook to their

subsidiaries and to competing social media platforms?


To answer this research question data from participants will be collected through means of a between-subjects experimental design. The control group (scandal: absent) will be presented with an article not related to the purpose of the study, while the experimental group (scandal: present) will be presented with a summary article about the Cambridge Analytica scandal and the data harvesting from Facebook users. The group composition will be decided randomly. Afterwards both groups will be presented with the same survey, which will feature questions on a 7-point Likert scale about participants

perceptions on various metrics such as brand trust, brand authenticity and willingness to use the brand’s service.

The brands featured in these questions will be Facebook, as the parent company, WhatsApp and Instagram as the subsidiaries, and Snapchat and TikTok as the competitors. My hypothesis is that there will be a statistically significant negative

spillover effect for Instagram (positive correlation with Facebook) but not for WhatsApp.

For the competing platforms I predict there will be a statistically significant positive spillover effect (negative correlation with Facebook). I predict the negative spillover for Instagram will be caused due to the platform being the most similar to Facebook.


2. Theoretical Framework

2.1. Brand Scandals What is a brand scandal?

A brand scandal, used interchangeably by researchers with terms such as brand

misconduct, brand crisis and brand failure (Li & Wei, 2016), has been defined by Dawar and Lei(2009)as a situation in which a key brand proposition turns out to be false. This situation can arise from factual (damaging) information about the company, or in some cases even from rumors about the company which may or may not be true. If key brand propositions such as authenticity, safety, climate-friendly etc. turn out to be false this can lead to customers losing their trust in the brand. In addition, there are also cases where the stock price of affected companies drop substantially as a result of a brand scandal.

(Dawar & Lei, 2009).

The reputation of a brand will also be hurt by scandals, and customers will be less willing to purchase products from the implicated company. While product-related (also referred to as performance-related) scandals have been studied adequately, there has been less academic research on value-related (i.e. ethical) scandals. (Guèvremont and Grohmann, 2017). The Cambridge Analytica scandal is a clear example of a value-related scandal and this study therefore aims to fill the research gap on this topic. Additionally, brand equity can also be negatively impacted by brand scandals. This has important

implications for brand management, since high brand equity is one of the primary drivers that add value to the brand name of a product. It is the difference between the Coca-Cola brand and a supermarkets white-label Cola product. Damage to brand equity can destroy the premium that strong brands can charge their customers. (Hegner et. al., 2014).


Product-harm crisis

Product-harm crisis is a term closely related to brand scandals, and researchers tend to use the terms interchangeably. However there are some key differences between the two concepts. A product-harm crisis is present when a product fails to deliver its expectations.

This could be caused by issues such as a defective product, dangerous product etc.

Common actions taken after a product-harm crisis would be the recalling of the failed product, compensating affected customers and taking responsibility for potential liabilities resulting from the product failure. (Vassilikopoulou et al., 2009). A product- harm crisis, if dealt with inadequately can turn into a full-scale brand scandal or brand crisis. That is, the crisis with one product trickles over to the rest of the brand. Research by Siomkos and Kurzbard (1994) identified three major factors which can amplify the effect of a product-harm crisis. These are internally: the company’s existing reputation and the company’s response to the crisis and externally: (negative) media coverage. A Product-harm crisis affects large, well-established companies in a different manner than smaller, lesser known companies. Large corporations usually own a wide portfolio of different products sold under the name of one underlying ‘umbrella’ or parent brand (Unilever, Nestle, P&G etc.). Therefore if customers are not aware that brand X, which suffers a product-scandal, is owned by company Z it would likely not dilute the brand equity of company Z. However the brand value of brand X will likely be effected by this hypothetical product scandal. (Siomkos and Kurzbard, 1994).


Effects of a brand scandal

Brand scandals or corporate scandals have the ability to strongly change customers behavior towards the offending company. Of course, customers can choose to set aside any ethical or other concerns they have with the company and continue purchasing their products or using their services. However, it is also possible that a line will be crossed for customers, resulting in them not being willing to engage with the company in the future.

(Guckian et al., 2017). Research by Guckian et al. (2017) further indicates that future brand engagement is highly dependent on customers perception of the corporate culture in offending companies. In the case of Volkswagen’s Diesel Emission scandal, VW customers who believed that only a few people within Volkswagen were responsible for the scandal were more willing to continue engaging with the company in addition to feeling less anger about the situation. On the other hand, customers directly affected by the scandal (in VW’s case, owners of diesel cars containing the so-called ‘defeat devices’

which manipulated the emission values.) were found to be less likely to continue engaging with VW and expressed more feelings of anger towards the situation, in line with the researchers expectations. (Guckian et al., 2017). Van Heerde et al. (2007) argue that brand crises diminish the effectiveness of marketing practices. Further they suggest that affected brands will be more prone to having their market share captured by


Minimizing the impact of brand scandals

It is evident that it will be in a corporation’s best interest to prevent brand scandals, and in the case a scandal is unavoidable, to minimize the impact of it. One way to minimize


the effect of a brand scandal is the ‘buffering’ effect. (Trump, 2014). Prior research by Ahluwalia et al. (2000) has shown that the buffering effect is able to compensate the effect of a brand scandal if the customer is highly connected with the brand. In this situation negative information about the brand can be resisted and argued against by customers. In contrast, the study by Trump, (2014) shows that this buffering effect does not occur when brand transgressions are self-relevant for the customer (closely impacted by the brand scandal, due to for example a failure to deliver a service by the brand). In fact, the research indicates that the negative effects are actually amplified if a customer is highly connected to a brand, because they can experience a feeling of betrayal due to the actions of the transgressing company. Notably, the buffering effect also did not occur if the brand scandal was ethical in nature.

Types of Brand Scandals

Brand scandals can be broadly divided into two categories, value-related scandals and performance-related scandals. Value-related scandals can be ethical or social in nature (Dutta & Pullig, 2011), examples of these are the bad working conditions in Amazon’s warehouses and the huge pressure on their delivery drivers. Performance-related scandals are present when the company cannot adequately satisfy consumers needs due to faulty products or related issues. (Dutta & Pullig, 2011). In Amazon’s case, this can be seen in their recent heavy push to allow more third-parties to sell on their platform, there is little control on the products these third parties sell which in some cases can lead to unsafe or even dangerous products being sold through the marketplace Amazon provides. (Berzon et al., 2019).


2.2 Brand Scandal Spillover

To put it simply, brand scandal spillover refers to a situation in which negative

information about one brand transfers over to other brands. This spillover effect can be seen in competing brands in the same industry. A scandal can potentially benefit

competing brands if the scandal is deemed to be unique to the situation of the scandalized company. If a scandal is not isolated then it could cause harm to competitors who might be considered ‘guilty by association’ (even if they actually have nothing to do with the scandal in question). (Roehm & Tybout, 2006). While the spillover effect in general has been discussed and studied in multiple papers, until now there has been limited research concerning the potential presence of a brand spillover effect in subsidiaries of the company affected by a brand scandal. (Lei et al., 2008).

Research on this subject by Lei et al. (2008) discusses negative spillover in brand

portfolios. However, it is important to note that ‘positive’ spillover can also occur, that is, a situation in which sub brands benefit from positive news or other promising

developments in the parent brand. Positive spillover can also be the result of dedicated marketing efforts, with the aim to transfer over brand equity In fact, umbrella-branding is often pursued with this goal in mind. While umbrella-branding can thus amplify

marketing effectiveness, the downsides of adverse events are also amplified due to the strong linkages between the sub brands and parent brand. The study by Lei et. al (2008) further found that there is an asymmetric relationship concerning spillover in brand portfolios. A brand crisis might have a greater spillover effect when the scandal

originates in sub brand A and transfer over to sub brand B than vice versa. This finding has important implications for management, since this indicates that different brands in a


portfolio can have asymmetric vulnerability to adverse events. Given this, different brands might need to take differing measures to combat potential scandals. Actions that could be taken include reducing associations with the parent for more at risk sub brands, but this needs to be carefully balanced with the resulting lessened marketing


What other measures can be taken to minimize the impact of brand scandal spillover?

Roehm & Tybout (2006) argue that competing firms would benefit from issuing a denial of the scandal behavior, but only in case the spillover has already occurred. If spillover has not yet occurred but a denial is issued, it can cause customers to think that the company has something to hide and will therefore be considered guilty of wrongdoing (the so-called boomerang effect).

Another important question to consider is why the spillover effect occurs. The

accessibility-diagnosticity theory proposed by Feldman and Lynch (1988) provides an explanation for this. This theory states that if brand A is informative about brand B then customers perceptions of aspects of brand A such as product quality will be used to infer the quality of brand B. For this effect to occur, the information needs to be easily

accessible to the customer, in other words easy to remember. This effect can also be strengthened if the two brands are perceived to be similar to each other, for example German car manufactures being more similar to other German car manufacturers than to Japanese or American car manufacturers. (Borah & Tellis, 2016).

This study takes a novel approach by simultaneously studying the spillover effects for subsidiaries of Facebook (WhatsApp, Instagram) and competitors of Facebook (TikTok,


Snapchat), by doing so the aim is to see whether the spillover effect of a brand scandal is, relatively speaking, stronger for subsidiaries or for competitors.

2.3 Brand Trust

Quoting research by Delgado‐Ballester and Luis Munuera‐Alemán (2005, pp. 2) Brand trust is defined as “(…) the consumer’s belief that the brand has specific qualities that make it consistent, competent, honest, responsible and so on”. This finding indicates that customers have high expectations of brands, they expect the brand to fulfill their

promises and bring them positive experiences. Brands (or more accurately the

corporations behind them) are not perfect and scandals affecting them cannot always be prevented, therefore it becomes inevitable that customers trust in a brand will fluctuate over time. In this light, it can be expected that Brand trust is strongly correlated to brand loyalty. Munuera-Aleman et al.(2003) confirmed this view earlier by showing that brand trust had a significant influence on brand loyalty. Brand trust can be seen as consisting of two major components: trustworthiness and expertise. The trustworthiness dimension is about the recurring theme of the customers ‘belief’ in a brand to deliver quality and to act in a sincere way. (Sung & Kim, 2010). Expertise is self-explanatory, but there can be major differences here between ‘young’ brands and brands with a rich history.

Brand trust also has an influence on the extent to and manner in which brand scandal spillover occurs. Research by Gao et al. (2015). found moderate support for the hypothesis that distrust in ‘contaminated’ brands leads to a corresponding increase of trust for differentiated brands (in this study the differentiating factor was the country of


origin of the brand). This finding shows a situation in which positive spillover occurred, competing brands benefiting from the scandal at another company. However negative spillover did occur for the other categories, being 1: hybrid brand (domestic production, foreign origin) spillover to noncontaminated brands and 2: contaminated domestic brand spillover to noncontaminated domestic brands. (Gao et al., 2015).

In the context of the social media industry, country of origin can also be considered a differentiating factor: Facebook, WhatsApp, Instagram and Snapchat are all American companies while TikTok is a Chinese company. Another category distinction can be made between subsidiary (WhatsApp, Instagram) and non-subsidiary competitors in the industry (Snapchat, TikTok). Since brand trust is one of the factors determining whether a spillover effect will occur, the following hypotheses can be proposed to test the

influence it might have on the potential creation of such a spillover effect.

H1: After a brand scandal, there will be a spillover effect negatively (positively) affecting brand trust for the subsidiary (non-subsidiary) firm

H2: After a brand scandal, there will be a spillover effect positively affecting brand trust for the foreign firm (TikTok)


2.4 Brand Authenticity

Brand authenticity has been defined as ‘a subjective evaluation of genuineness ascribed to a brand by consumers’. (Napoli et al., 2014, pp. 2). Since it is a subjective evaluation, different customers can ascribe different factors as contributing to higher brand

authenticity. Several attributes commonly associated with brand authenticity are

sincerity, nostalgia, quality commitment, design consistency and craftsmanship. (Napoli et al., 2014). From a marketing perspective it has also been described as the essence of

‘what customers want’ For this to be the case, the products the brand offers must be considered real and genuine by them, as opposed to imitated or copied from competitors.

For this reason products with a long history are often considered to be more authentic (think of centuries old beer brewers). (Hernandez-Fernandez & Lewis, 2019).

According to Guèvremont and Grohmann(2017), brand authenticity can moderate customer response to a brand scandal. When customers believe that brands are authentic and thus acting in a honest manner, situations can arise where brand scandals will have lessened impact on them. Most notably, perceived responsibility for the scandal was found to be lower for firms considered more authentic. (Guèvremont & Grohmann, 2017). Extrapolating this finding to the spillover effect, the assumption can be made that a negative spillover effect is less likely to occur in more (vs less) authentic firms. As such, the following hypothesis will be proposed:

H3: Firms with higher (vs lower) brand authenticity are less likely to suffer a negative spillover effect after a brand scandal


2.5 Brand Loyalty & Willingness to use

Brand loyalty is said to occur when customers continue to purchase the products of a brand over time. Loyal customers may also have a preference and be willing to pay more for the products of the brand they are loyal to. (Kapoor & Banerjee, 2020). However, prior brand loyalty does not guarantee that customers will stay loyal even in the event of a brand scandal. (Elbedweihy et al., 2016). Another definition of brand loyalty is stated as follows by Han et al. (2018, pp. 4) : “the biased (non-random) behavioral response (purchase) expressed over time by some decision-making unit with respect to one or more alternative brands out of a set of brands (…)”. The key takeaway from this is that

customers make a conscious decision to choose one brand over another.

Willingness to pay is a concept closely linked to brand loyalty (Chaudhuri & Holbrook, 2001), however since Facebook and other social media services are primarily free

services, willingness to pay is not the correct term to use here. Instead, willingness to use can be applied in the context of the social media industry. The core idea being that willingness to pay is strongly related to the extent of a customer’s loyalty to a brand.

Customers may be more willing to pay (a premium) for products of the brand they are loyal to, this can be due to perceived uniqueness of the brand offerings. (Chaudhuri &

Holbrook, 2001). Social media services are highly differentiated from each other, not only based on the look and feel of the app/website and the features of the platform, but perhaps most importantly due to the different contacts (Facebook friends, Instagram followers, WhatsApp mobile contacts etc.) people have on different platforms. As such instead of loyalty there might be more of a necessity to staying on social media platforms,


to keep in contact with friends or to prevent fear of missing out. A paradoxical event might occur, in the sense that customers might not necessary ‘like’ the social media platform but instead feel pressure to stay on it and keep using it due to the reasons mentioned above. In research by Hinds et al. (2020) this was also shown by participants feeling conflicted. They were ‘fed-up’ with Facebook, but not enough to take the action to delete or deactivate account. Therefore it remains unclear if brand loyalty is strong enough in the social media industry to counteract the effect of a brand scandal (thus preventing negative spillover). Thus, taking into account the willingness to use and brand loyalty of customers the following hypotheses can be proposed to determine whether a spillover effect could occur:

H4: Firms with higher (vs lower) brand loyalty are less likely to suffer a negative spillover effect after a brand scandal

H5: Firms with higher (vs lower) willingness to use are less likely to suffer a negative spillover effect after a brand scandal


3. Methods and Data

3.1 Research Design

To answer the research question “To what extent is there a spillover effect of brand scandals from Facebook to their subsidiaries and to competing social media platforms?”

several hypotheses were proposed in the theoretical framework. Since these hypotheses depend on the result of several correlation and regression tests a quantitative approach needed to be taken to test them. As such a between-subjects experimental survey was designed for this purpose. A between-subjects design was chosen to create a control group for the experiment condition: brand scandal. By doing this the potential impact of a brand scandal is expected to be gauged more accurately. The between-subjects design was accomplished by dividing the participants in two groups randomly using the survey software (Qualtrics). The control group was first presented with an irrelevant article, a filler task, and afterwards the actual survey questions. In contrast, the experiment group was first presented with an abridged article about the Cambridge Analytica scandal, the same filler task as the control group, and afterwards the same survey questions as the control group. The survey was designed in Qualtrics and provided to participants using an anonymous link, all respondents provided consent to use their data for the purpose of this research prior to filling in their responses.

Regarding the firms questioned in the survey, Facebook was chosen since it was the subject of the Cambridge Analytica scandal and since it is currently the largest social media platform in terms of users. (H. Tankovska., 2021). WhatsApp and Instagram are both subsidiaries of Facebook, but differ in relevant ways. WhatsApp is mainly used to


stay in contact with close friends, either through 1-on-1 chats or group chats with friends, while Instagram can be more easily used to share your content with not only (close) friends but everyone who follows your account (private profile) or even everyone who uses Instagram (public profile). The same distinction can be made for Snapchat and TikTok, two main competitors of Facebook. Snapchat is more focused towards sharing content with close friends while on TikTok you can post content publicly so that everyone who follows your account (private profile) or everyone who uses TikTok (public profile) can see your posts.

The survey contained 37 questions in total. 2 of which were demographic questions (gender, age) to be used as control variables. The filler task was comprised of 5 questions, each asking respondents to rate a different brand on a scale from 1-10. (The brands were Spotify, Apple, Google, Uber and Amazon). The remaining 30 questions were divided in three sections (Facebook, subsidiaries, competitors) containing 6, 12 and 12 questions respectively. These 30 questions consisted of 6 main questions substituted with the names of the 5 different brands involved in the study. (Ex. I consider

Facebook/WhatsApp/Instagram etc. to be an ethical company). The questions tested for hypotheses concerning the variables brand trust, brand authenticity, brand loyalty and willingness to use. the Respondents could fill in the questions based on a 7-point Likert- scale (1= strongly disagree, 7 = strongly agree) as commonly used in marketing research.


3.2 Sample

Data was collected for 14 days from June 1 until June 14, 2021. The participants for the survey were recruited through convenience sampling, mostly utilizing various social media platforms. Data was collected from a total of 54 participants, however the data of 16 participants had to be discarded due to partially incomplete surveys. Therefore the data of 38 participants in total was analyzed. This indicates a survey completion rate of 70.37%. Of these 38 participants 18 were presented with the ‘scandal’ condition while 20 participants were part of the control group. Ideally this division would be exactly 50%, however due to the discarded responses a slight variance occurred here. The control variable gender was distributed as follows: 20 males, 16 females with the remaining 2 being non-binary/third gender. The other control variable was age and it was distributed as follows: (M= 36,21, SD = 15,70). 4 values were missing due to respondents

(unintentionally) being able to fill in a blank space in the corresponding question.

Therefore the total valid cases was 34 instead of 38.

3.3 Variables

Brand Trust: To measure brand trust a 2-item scale was constructed, since there were 5

brands in the study respondents answered a total of 10 questions about brand trust.

Testing the scale for each brand separately gives the following values for Cronbach’s Alpha (Facebook: α = 0.63, Whatsapp: α = 0.76, Instagram: α = 0.85, Snapchat: α = 0.73, TikTok: α = 0.96, average: α = 0.79). Given these high values the scale can be assumed to be sufficiently reliable.


Brand Authenticity: To measure brand authenticity a 2-item scale was constructed, the

total questions are therefore 10 split among the different brands. Testing the scale for each brand separately gives the following values for Cronbach’s Alpha (Facebook: α = 0.77, Whatsapp: α = 0.96, Instagram: α = 0.88, Snapchat: α = 0.89, TikTok: α = 0.996, average: α = 0.89). The values are high enough to assume the scale to be reliable,

however the value for TikTok could be considered too high which could indicate using a 2-item scale here might be redundant.

Brand Loyalty / Willingness to Use: Brand loyalty and willingness to use were both

measured using a 1-item scale, however since Cronbach’s Alpha cannot be calculated with only one item both variables were combined here. Testing the scale for each brand separately produces the following values for Cronbach’s Alpha (Facebook: α = 0.65, Whatsapp: α = 0.54, Instagram: α = 0.70, Snapchat: α = 0.65, TikTok: α = 0.75, average:

α = 0.66). While these values are a bit lower than for the other variables, they can still be considered sufficiently high to use the scale.

3.4 Analyses

Hypothesis 1 and 2 were tested using the independent samples t-test. For Hypothesis 1 two sub hypotheses needed to be tested: H1a and H1b. For Hypothesis 1a the

(continuous) dependent variable was subsidiary brand trust, which was calculated as the mean of WhatsApp and Instagram’s brand trust (measured using the corresponding 2- item scale). The independent variable was the experiment condition: the categorical variable ‘scandal present’ (coded as 0 = no scandal, 1 = scandal). Hypothesis 1b used the


(continuous) dependent variable of competitor brand trust, which was calculated as the mean of Snapchat and TikTok’s brand trust. The independent (categorical) variable (was the same: ‘scandal present’. Hypothesis 2 featured TikTok’s brand trust as the dependent variable, and the same independent variable: ‘scandal present’.

Hypothesis 3, 4 and 5 were all analyzed using the PROCESS macro (Hayes, 2017) in SPSS. Specifically, model 1 of the PROCESS macro was used to perform moderation analysis. The dependent variable was brand trust for all three hypotheses. Specifically, the average brand trust values for all 5 firms combined. The independent variable was again ‘scandal present’ the same categorical variable used for Hypothesis 1 and 2. The moderator variables were: brand authenticity (H3), brand loyalty (H4) and willingness to use (H5).


4. Results

4.1 Correlation

Variable M SD 1 2 3

1. Facebook* 1.76 1.03

2.Subsidiary* 2.15 1.19 ,815**

3. Competitor* 2.21 1.16 ,587** ,612*


** p < 0.01 (2-tailed); N = 38

* Brand Trust

The correlation matrix above shows the correlations between Facebook, their subsidiaries and their competitors. All correlations are significant at the 99% confidence interval (p <

0.01) and all correlations are positive. The strongest correlation is found between Facebook and their subsidiaries at a value of r(38) = 0.82, p < .0.1, indicating a very strong positive relationship. Interestingly, the weakest correlation is between Facebook and their competitors at r(38) = 0.59, p < .0.1. Recalling H1: “After a brand scandal, there will be a spillover effect negatively (positively) affecting brand trust for the subsidiary (non-subsidiary) firm” a negative correlation between Facebook and their competitors would ideally be expected for this hypothesis to be accepted. That is, a decrease in brand trust for Facebook (after a scandal) would lead to a increase in brand trust for competitors. This is not the case here since the correlation is still strongly positive. Mean brand trust is highest for competitors and lowest for Facebook, however the values are still quite close to each other.


4.2 Hypothesis Tests

H1a: After a brand scandal, there will be a spillover effect negatively affecting brand trust for the subsidiary firm

For the independent sample t-test the data need to be approximately normally distributed.

The Shapiro-Wilk test gave values of W(20) = 0.82, p < .01 for the control group (Scandal0) and W(18) = 0.92, p = 0.14 for the experiment group (Scandal1). This indicates that the control group might violate the normality assumption. However as an additional check, skewness and kurtosis values for Scandal0 indicate an approximately normal distribution. (Skewness(20) = 0.95, Kurtosis(20) = -0.45). Levene’s test for equality of variances showed a significant value of F(36) = 4.35, p = 0.04, indicating that equal variances cannot be assumed.

The t-test showed a higher mean for the control group (MNoScandal = 2.23, SD = 1.41) then for the experiment group (MScandal = 2.07, SD = 0.91), indicating that there might be some effect attributed to the scandal condition. The t-test revealed this effect as t(33) = 0.41, p

= 0.69. This finding is non-significant and therefore H1a cannot be supported based on these results. In other words, there was no evidence found for a statistically significant difference between the brand trust for subsidiaries in either a control or experiment condition. The presence of a spillover effect cannot be proven based on these data and therefore H1a needs to be rejected.

H1b: After a brand scandal, there will be a spillover effect positively affectingbrand trust for the competitor firm

For the competitor brand trust data, there was severe skewness in the control group data:


Skewness(20) = 1.111, to solve this, one outlier was removed from the data. After removing the outlier the skewness value was 0.88, bringing it closer to a normal

distribution. Levene’s test for equality of variances showed a significant value of F(35) = 6.18, p = 0.02, rejecting the hypothesis that equal variances can be assumed.

Contrary to the results for subsidiary firms, the mean brand trust rating for the control group (MNoScandal = 2.01, SD = 0.93) was lower than for the experiment group (MScandal = 2.26, SD = 1.23). This finding corresponded with a value of t(32) = -0.70, p = 0.49.

While this result is non-significant due to the high p-value, it is important to note that a inverse relationship was found. After a brand scandal, brand trust for the competitors increased, however this increase was not statistically significant. Based on these findings H1b cannot be supported.

H2: After a brand scandal, there will be a spillover effect positively affecting brand trust for the foreign firm (TikToK)

The data for TikTok was found to be severely positively skewed for the control condition: Skewness(20) = 1.47. As such, a log transformation was performed on the data, resulting in a skewness value for the control condition of 0.99, showing a

considerable improvement. The mean value for this transformation cannot be interpreted properly as a value between 1-7, so in the results below the mean and standard deviation for the original data will be presented. For both the original data and the transformation the Levene’s test for equality of variances showed a non-significant value, indicating that equal variances can be assumed. (Original: F(36) = 0.21, p = 0.65). (Log transformation:

F(36) = 0.02, p = 0.87).


Brand trust was higher for participants in the Scandal group (MScandal = 2.17, SD = 1.26) compared to the control group (MNoScandal = 1.95, SD = 1.49). The log transformation produced a value of t(36) = -0.77, p = 0.49, indicating a non-significant result. For reference the original data was also non-significant with t(36) = -0.48, p = 0.63. Again, while there is a difference between means in the two groups in the right direction

(expected increase in brand trust for the scandal group), this difference is not statistically significant and therefore H2 needs to be rejected.

H3: Firms with higher (vs lower) brand authenticity are less likely to suffer a negative spillover effect after a brand scandal

The Durbin-Watson test produced a value of 2.33 which means the condition of

independence of observations has been met. A scatterplot between brand trust and brand authenticity showed a linear relationship. There is no severe multicollinearity present considering the very low VIF value of 1.624 for brand authenticity and values below 10 for the interaction (brand authenticity * scandal 0/1) and the scandal binary variable. (VIF

= 6.59, VIF = 5.80). The Q-Q Plots appear to show that the residuals are approximately normally distributed.

The PROCESS output shows that the model is overall significant with R2 = 0.84, F(5,28)

= 29,97, p < .01. The model explains 84% of the variation for the dependent variable brand trust. However the interaction between brand authenticity and the scandal variable is not significant with corresponding values of: R2 change = 0.005, F(1,28) = 0.88, p = 0.35. The model overall is still a significant predictor of brand trust due to significant


results for brand authenticity: t = 9.03, p < .01 and the control variable age: t = 2.71, p = 0.01. Since there was no significant interaction effect present we cannot conclude that a spillover effect is the cause of the change in brand trust. Therefore H3 cannot be

supported based on the available data.

H4: Firms with higher (vs lower) brand loyalty are less likely to suffer a negative spillover effect after a brand scandal

The Durbin-Watson test with a value of 2.34 confirms the independence of observations.

The scatterplot between brand trust and brand loyalty shows a linear relationship,

although the distribution is a bit uneven. The VIF values for the interaction (VIF = 5.64), brand loyalty (VIF = 1.82) and the scandal variable (VIF = 4.55) show no serious

multicollinearity. The Q-Q plots show that the residuals are close to a normal distribution.

The overall model is significant with R2 = 0.78, F(5,28) = 20,02, p < .01. Therefore the model as a whole explains 78% of the variation in the brand trust ratings. The interaction between brand loyalty and the scandal variable is not significant (R2 change = 0.02, F(1,28) = 2.75, p = 0.11, however this value is quite close to being statistically significant compared to the other observations. There is again significance for age: t = 2.53, p = 0.18 and there is also significance for brand loyalty: t = 7.35, p < .01. Based on these results brand loyalty, and to a lesser extent age, have a significant impact on the brand trust variable. However since there is no interaction between brand loyalty and the scandal variable, H4 cannot be supported. There is not sufficient evidence to show the presence


of a spillover effect.

H5: Firms with higher (vs lower) willingness to use are less likely to suffer a negative spillover effect after a brand scandal

Independence of the observations can be assumed due to a Durbin-Watson score of 2.32.

A scatter plot with willingness to use and brand trust shows a moderate linear

relationship. Multicollinearity is also not an issue for the variables. The VIF values are 4.51 for the interaction, 1.80 for willingness to use and 3.7 for the scandal binary variable. The Q-Q plot between willingness to use and brand trust are close to the linear line and therefore it can be assumed that the residuals are normally distributed.

The overall model is significant (R2 = 0.40, F(5,28) = 3.75, p = 0.01. This means that the model explains 40% of the variation for the brand trust ratings, this is considerably lower than the results for H3 and H4. Willingness to use (t = 2.54, p = 0.02) and age (t=3.22, p

< .01. are both significant predictors of brand trust. The interaction between willingness to use and the scandal variable is not significant (R2 change = 0.01, F(1,28) = 0.39, p = 0.54. Since the interaction is not significant there is no sufficient evidence to conclude that H5 is true, and therefore the hypothesis needs to be rejected.


5. Discussion

5.1 Implications

In this thesis the findings from the survey were analyzed in order to answer the research question: To what extent is there a spillover effect of brand scandals from Facebook to their subsidiaries and to competing social media platforms?. The various hypotheses proposed to answer this question could not be supported by the data. In other words, the there was insufficient statistical evidence to conclude there was a meaningful spillover effect of brand scandals at the parent company to subsidiaries and competing platforms.

However, while there was no statistical significance to the data, the results generally did fall in line with the expectations set out in the hypotheses. For hypothesis 1a, subsidiaries did have a lower brand trust score in the brand scandal scenario compared to the control group. Hypothesis 1b and 2 showed that competitors in fact experienced increased average brand ratings in the scandal condition. Again, this difference was simply not sufficiently large to conclude that it was caused by a spillover effect. Hypothesis 3, 4 and 5 all found no interaction effect between the moderator (brand authenticity, brand loyalty and willingness to use) and the independent variable, the scandal condition. However, the results showed age and the moderators individually to be a significant predictor of brand trust. While it is not very surprising that a variable such as brand loyalty can have a large effect on a related measure such as brand trust, age was not expected to be a predictor.

This could be caused by the relatively small sample size, or due to different age groups having wildly different experiences with social media and the internet in general. The younger age group was perhaps in high school or university during the rise of social media, and therefore experienced it fully, while the older age group is already well into


the workforce and thus might pay little attention to it (and the dangers of privacy

breaches). These are just predictions however, future research could further look into the role of age concerning attitudes to social media companies.

These results were not in line with other research on the potential presence of a spillover effect at brands. Research by Trump and Newman (2016) for example did find evidence for a spillover effect based on unethical perceptions of a brand. This study succeeded in more successfully forming a difference between the control group and the experiment group, therefore the differences between means were statistically different here. However this study was about the car industry, a completely different industry than the social media industry. Furthermore factors such as the safety of cars can potentially be a life or death situation and therefore it is expected that customers will place a higher importance on this factor and related ones. This difference in industry is further highlighted by Hinds et al. (2020), where participants to the study generally did not show much concern after the Cambridge Analytica scandal. Hence this study can actually be considered consistent with the results of this thesis, willingness to keep using the service was not significantly impacted by a brand scandal.

5.2 Limitations

This study did not find evidence for a spillover effect, this could have been caused by multiple factors. First, the sample size was 38, which could normally be considered sufficient, however this sample was split in two, leading to two groups of 18 and 20 participants. This caused some issues with the data, because the data sometimes showed high levels of skewness. Notably the mean for the scale results usually was around 2, this


can be considered quite low, for a Likert-scale consisting of 7 levels. Since the mean ratings on attributes of the brand were already low, the scandal condition perhaps did not shock participants enough to cause significantly lower ratings. Participants in the control group might also already have been familiar with the scandal, and therefore rated the social media companies lower then if they did not know about the scandal at all. The use of convenience sampling could also have introduced bias to the results.



Ahluwalia, R., Burnkrant, R. E., & Unnava, H. R. (2000). Consumer Response to Negative Publicity: The Moderating Role of Commitment. Journal of Marketing Research, 37(2), 203–214.

Berzon, A., Shifflett, S., & Scheck, J. (2019, August 23). Amazon Has Ceded Control of Its Site. The Result: Thousands of Banned, Unsafe or Mislabeled Products. WSJ. thousands-of-banned-unsafe-or-mislabeled-products-11566564990

Borah, A., & Tellis, G. J. (2016). Halo (Spillover) Effects in Social Media: Do Product Recalls of One Brand Hurt or Help Rival Brands? Journal of Marketing Research, 53(2), 143–160.

Brown, A. J. (2020). “Should I Stay or Should I Leave?”: Exploring (Dis)continued Facebook Use After the Cambridge Analytica Scandal. Social Media + Society, 6(1),

Chaudhuri, A., & Holbrook, M. B. (2001). The Chain of Effects from Brand Trust and Brand Affect to Brand Performance: The Role of Brand Loyalty. Journal of Marketing, 65(2), 81–93.

Confessore, N., & Rosenberg, M. (2018a, March 27). Spy Contractor’s Idea Helped Cambridge Analytica Harvest Facebook Data. The New York Times. Retrieved from:


Confessore, N., & Rosenberg, M. (2018b, May 2). Cambridge Analytica to File for Bankruptcy After Misuse of Facebook Data. The New York Times. Retrieved from: down.html

Dang, S. (2021, April 20). Apple goes to battle with Spotify in premium podcast push.

Reuters. premium-podcast-push-2021-04-20/

Dawar, N., & Lei, J. (2009). Brand crises: The roles of brand familiarity and crisis relevance in determining the impact on brand evaluations. Journal of Business Research, 62(4), 509–516.

Delgado‐Ballester, E., & Luis Munuera‐Alemán, J. (2005). Does brand trust matter to brand equity? Journal of Product & Brand Management, 14(3), 187–196.

Dutta, S., & Pullig, C. (2011). Effectiveness of corporate responses to brand crises: The role of crisis type and response strategies. Journal of Business Research, 64(12), 1281–1287.

Elbedweihy, A. M., Jayawardhena, C., Elsharnouby, M. H., & Elsharnouby, T. H. (2016).

Customer relationship building: The role of brand attractiveness and consumer–

brand identification. Journal of Business Research, 69(8), 2901–2910.


Feldman, J. M., & Lynch, J. G. (1988). Self-generated validity and other effects of measurement on belief, attitude, intention, and behavior. Journal of Applied Psychology, 73(3), 421–435.

Gao, H., Zhang, H., Zhang, X., & Knight, J. G. (2015). Spillover of Distrust from Domestic to Imported Brands in a Crisis-Sensitized Market. Journal of International Marketing, 23(1), 91–112.

Granville, K. (2018, May 14). Facebook and Cambridge Analytica: What You Need to Know as Fallout Widens. The New York Times. explained.html

Guckian, M. L., Chapman, D. A., Lickel, B., & Markowitz, E. M. (2017). “A few bad apples” or “rotten to the core”: Perceptions of corporate culture drive brand engagement after corporate scandal. Journal of Consumer Behaviour, 17(1), e29–e41.

Guèvremont, A., & Grohmann, B. (2017). Does brand authenticity alleviate the effect of brand scandals? Journal of Brand Management, 25(4), 322–336.

H. Tankovska (2021). Global social networks ranked by number of users 2021. Statista.

Retrieved from networks-ranked-by-number-of-users/


Han, H., Nguyen, H. N., Song, H., Chua, B. L., Lee, S., & Kim, W. (2018). Drivers of brand loyalty in the chain coffee shop industry. International Journal of

Hospitality Management, 72, 86–97.

Hayes, A. F. (2017). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. Guilford publications.

Hegner, S. M., Beldad, A. D., & Kamphuis op Heghuis, S. (2014). How company responses and trusting relationships protect brand equity in times of crises.

Journal of Brand Management, 21(5), 429–445.

Hernandez-Fernandez, A., & Lewis, M. C. (2019). Brand authenticity leads to perceived value and brand trust. European Journal of Management and Business

Economics, 28(3), 222–238.

Hinds, J., Williams, E. J., & Joinson, A. N. (2020). “It wouldn’t happen to me”: Privacy concerns and perspectives following the Cambridge Analytica scandal.

International Journal of Human-Computer Studies, 143

Huang, D.-L., Patrick Rau, P.-L., Salvendy, G., Gao, F., & Zhou, J. (2011). Factors affecting perception of information security and their impacts on IT adoption and security practices. International Journal of Human-Computer Studies, 69(12), 870-883


Kapoor, S., & Banerjee, S. (2020). On the relationship between brand scandal and consumer attitudes: A literature review and research agenda. International Journal of Consumer Studies. Published.

Lei, J., Dawar, N., & Lemmink, J. (2008). Negative Spillover in Brand Portfolios:

Exploring the Antecedents of Asymmetric Effects. Journal of Marketing, 72(3), 111–123.

Levy, A. (2020, September 30). Palantir closes below first trade after NYSE debut.

CNBC. Retrieved from: nyse.html

Li, M., & Wei, H. (2016). How to Save Brand after Crises? A Literature Review on Brand Crisis Management. American Journal of Industrial and Business Management, 06(02), 89–96.

Millard, D. (2018, March 30). Cambridge Analytica is bad, but Palantir is fucking terrifying. The Outline. Retrieved from:

Munuera-Aleman, J. L., Delgado-Ballester, E., & Yague-Guillen, M. J. (2003).

Development and Validation of a Brand Trust Scale. International Journal of Market Research, 45(1), 1–18.


Napoli, J., Dickinson, S. J., Beverland, M. B., & Farrelly, F. (2014). Measuring

consumer-based brand authenticity. Journal of Business Research, 67(6), 1090–


Revell, T. (2018, April 16). How Facebook let a friend pass my data to Cambridge Analytica. New Scientist. Retrieved from: to-cambridge-analytica/

Roehm, M. L., & Tybout, A. M. (2006). When Will a Brand Scandal Spill Over, and how Should Competitors Respond? Journal of Marketing Research, 43(3), 366–373.

Siomkos, G. J., & Kurzbard, G. (1994). The Hidden Crisis in Product‐harm Crisis Management. European Journal of Marketing, 28(2), 30–41.

Sung, Y., & Kim, J. (2010). Effects of brand personality on brand trust and brand affect.

Psychology and Marketing, 27(7), 639–661.

Trump, R. K. (2014). Connected consumers’ responses to negative brand actions: The roles of transgression self-relevance and domain. Journal of Business Research, 67(9), 1824–1830.


Trump, R. K., & Newman, K. P. (2016). When do unethical brand perceptions spill over to competitors? Marketing Letters, 28(2), 219–230.

Van Heerde, H., Helsen, K., & Dekimpe, M. G. (2007). The Impact of a Product-Harm Crisis on Marketing Effectiveness. Marketing Science, 26(2), 230–245.

Vassilikopoulou, A., Siomkos, G., Chatzipanagiotou, K., & Pantouvakis, A. (2009).

Product-harm crisis management: Time heals all wounds? Journal of Retailing and Consumer Services, 16(3), 174–180.



Appendix 1

Umbrella Branding in the Instagram App Launch Screen


Appendix 2

Survey Background text: Control Group

Apple Inc (AAPL.O) is fighting to retain control of the fast-growing podcasting market it popularized years ago but did not monetize, analysts and industry experts told Reuters.

Nearly 16 years after Apple added the ability to find podcasts -- a portmanteau of "iPod"

and "broadcasting" coined by a Guardian journalist -- to its iTunes software, the iPhone maker now seeks to court podcast creators with new subscription and creator tools, and fend off competition from streaming audio company Spotify (SPOT.N).

Apple announced on Tuesday it will launch Apple Podcast subscriptions, which will let users pay to unlock new content and additional benefits like ad-free listening, said Apple Chief Executive Tim Cook during the presentation. Pricing for each subscription will be set by the creator and billed monthly, Apple noted in a press release. It also introduced a new Apple Podcasters Program that will cost $19.99 per month, and will provide creators the tools they need to offer podcast subscriptions.

Sourced from: premium-podcast-push-2021-04-20/


Survey Backround text: Experiment Group

Cambridge Analytica, a political data firm hired by President Trump’s 2016 election campaign, gained access to private information on more than 50 million Facebook users.

The firm offered tools that could identify the personalities of American voters and influence their behavior. The data, a portion of which was viewed by The New York Times, included details on users’ identities, friend networks and “likes.” The idea was to map personality traits based on what people had liked on Facebook, and then use that information to target audiences with digital ads.

Researchers in 2014 asked users to take a personality survey and download an app, which scraped some private information from their profiles and those of their friends, activity that Facebook permitted at the time and has since banned.

Facebook said no passwords or “sensitive pieces of information” had been taken, though information about a user’s location was available to Cambridge.

Facebook in recent days has insisted that what Cambridge did was not a data breach, because it routinely allows researchers to have access to user data for academic purposes

— and users consent to this access when they create a Facebook account.

But Facebook prohibits this kind of data to be sold or transferred “to any ad network, data broker or other advertising or monetization-related service.”

Sourced from: analytica-explained.html


Appendix 3: Survey Intro Text Welcome!

In this survey, we are interested in your attitude towards and perception of certain brands. In answering the questions, please note that there is no right or wrong answer. The best answer is the one that is the closest to your experience or feeling. This survey takes approximately 5-10

minutes to finish.

Your participation is voluntary and anonymous. Your response will only be used for the purpose of my thesis in Business Administration at the University of Amsterdam. If you have any further questions concerning this study please feel free to contact me: Ivo Coutinho at

Appendix 4: Survey Questions

Segment 1: Demographic Questions + Filler Task 1. What is your age?

2. What is your gender?

Filler Task: Please rate the following brands on a scale from 1-

10 according to your personal preferences 1. Spotify

2. Apple 3. Google 4. Uber 5. Amazon

Segment 2: Facebook Brand Trust

1. I consider Facebook to be an ethical company

2. I trust Facebook to handle my personal data with care


Brand Authenticity

1. I believe that Facebook will not betray me 2. I believe that Facebook cares about their users Brand Loyalty

1. I feel more loyal to Facebook than to other brands Willingness to use

1. I would still be willing to use Facebook even if my personal data were given to third-parties without my consent

Segment 3: WhatsApp and Instagram Brand Trust

1. I consider WhatsApp to be an ethical company

2. I trust WhatsApp to handle my personal data with care Brand Authenticity

1. I believe that WhatsApp will not betray me 2. I believe that WhatsApp cares about their users Brand Loyalty

1. I feel more loyal to WhatsApp than to other brands Willingness to use

1. I would still be willing to use WhatsApp even if my personal data were given to third-parties without my consent

Segment 3: WhatsApp and Instagram Brand Trust

1. I consider Instagram to be an ethical company

2. I trust Instagram to handle my personal data with care


Brand Authenticity

1. I believe that Instagram will not betray me 2. I believe that Instagram cares about their users Brand Loyalty

1. I feel more loyal to Instagram than to other brands Willingness to use

1. I would still be willing to use Instagram even if my personal data were given to third-parties without my consent

Segment 4: Snapchat and TikTok Brand Trust

1. I consider Snapchat to be an ethical company

2. I trust Snapchat to handle my personal data with care Brand Authenticity

1. I believe that Snapchat will not betray me 2. I believe that Snapchat cares about their users Brand Loyalty

1. I feel more loyal to Snapchat than to other brands Willingness to use

1. I would still be willing to use Snapchat even if my personal data were given to third-parties without my consent

Segment 4: Snapchat and TikTok Brand Trust

1. I consider TikTok to be an ethical company

2. I trust TikTok to handle my personal data with care


Brand Authenticity

1. I believe that TikTok will not betray me 2. I believe that Tiktok cares about their users Brand Loyalty

1. I feel more loyal to TikTok than to other brands Willingness to use

1. I would still be willing to use TikTok even if my personal data were given to third-parties without my consent

Appendix 5: SPSS Output H1a







Process Output




Process Output




Process Output




Related subjects :