Master Thesis for the Master of Business Administration (M.Sc.)
“Negative publicity: The shielding effect of customer-company commitment and source credibility on customer attitudes and behavior intentions”
Student: Luka Lipovšćak Student number: 10622268
Study: Master of Business Studies (M.Sc.)
Supervisor: Dr. Karin Venetis
Faculty: Faculty of Economics and Business
Marketing Department, M.2.25 Date: 25.09.2014
Table of Contents1. Introduction ... 1 1.1 Research question ... 4 2. Theoretical background... 6 2.1 Negative publicity ... 6 2.2 Customer commitment ... 9 2.3 Customer-company identification ... 11
2.4 Customer commitment versus customer identification ... 12
2.5 Source credibility ... 13 3. Conceptual model ... 16 4. Pre-test ... 20 4.1. Methods... 20 4.1.1 Design ... 20 4.1.2 Sample ... 20 4.1.3 Materials ... 20 4.1.4 Procedure ... 22 4.2 Results ... 23 5. Main Study ... 24 5.1 Methods... 24 5.1.1 Design ... 24 5.1.2 Materials ... 24 5.1.3 Procedure ... 28 5.1.4 Sample ... 30
5.1.5 Missing values & reliability ... 31
5.1.6 Normality of distribution ... 31
5.1.7 Correlation of variables ... 32
5.1.8 Groups descriptives ... 34
5.1.9 Hypotheses testing ... 36
6. Discussion ... 44
6.1 Theoretical and managerial implications ... 44
6.2 Future research ... 46
7. Conclusion ... 48
8. Literature ... 50
9. Appendix ... 57
9.1 Explanation of group names ... 57
This study investigates from a new perspective the impact of negativity on customer attitudes and behavior intentions. In the recent years negative publicity has become something companies have to deal with more than ever as the media keeps favoring reports about crises. Hence
companies had to look for new ways to deal with negative publicity. Some studies have
researched the so called “shielding” effect of customer commitment and identification and found that it protects customers’ attitudes and behavior intentions. Firstly, research will focus on the impact of customer-company commitment and its “shielding” effect on the customer attitudes and behavior intentions towards the company when the company dealt with moderate and extremely negative publicity. Secondly, a new moderator is introduced that creates a
multiplicative moderation in order to further test the turning point up to which high commitment customers counter argue negative news. The research is based on the three main theoretical constructs: the customer commitment and its protective effect on customer attitudes and behavior proposed by Ahluwalia (2000), the customer-company identification and its protective effect on negative news and customers attitudes and behavior developed by Einwiller, et.al (2006) and the source credibility construct and its impact on customer attitudes and behavior intentions
researched by Griffin, et. al (1991). A quantitative research was conducted on a sample (N=174) exploring the proposed concept.
Corporate reputation has a major impact on customer purchase intentions. Not only today but also for the last fifty years has research placed a major focus on this (Boddy, 2012). The
relevance of corporate branding lies in the fact that companies can differentiate themselves from competition. Corporate reputation is considered as a highly valued company asset that can hardly be imitated and is imperfectly mobile as it is deeply integrated in the company, therefore serving as a source of competitive advantage. Among the many dimensions of corporate reputation financial performance, value added in terms of a comparative advantage, corporate conduct, leadership, customers, corporate governance and crisis management are the most relevant ones (Reddiar et al., 2011).
Although being often mistaken and mingled with other terms, corporate reputation plays a different role in reputation development than corporate identity, brand, and image. It is the estimation in which the organization is held: a company can be seen as good or bad, it can be admired and respected as well as it can be seen as better or worse than its rivals (Dowling & Moran; 2012). Nowadays, the corporate reputation of a company has a great influence on its business performance, either positively or negatively.
Over the years there have been many corporate brand crises, unexpected, non-routine events that create uncertainty and threaten an organization’s priority goals (Seeger, Sellnow, & Ulmer 1998.) There can be different causes of a crisis such as accounting scandal or product harm with usual results such as creating negative publicity that threatens the corporate image. To prevent these consequences a good corporate reputation should be created (Dean, 2004).
2 Negative publicity situations are becoming increasingly common, what is shown by the
approximate increase of 15% in negative news coverage in the years 2000 to 2004 compared to the 1990s (Institute for Crisis Management, 2004). The most famous examples of those
reputation crises can be seen when looking at companies such as Exxon (i.e. the Valdez oil spill incident), Union Carbide (i.e. the Bhopal explosion), and Arthur Andersen (i.e. accounting scandals; Greyser, 2009).
One example most worth looking at is the one of the Arthur Andersen and Enron audit failure. One of the main reasons why audit companies can charge a premium price is their reputation. When that reputation is lost as in the Andersen and Enron audit case, the company lost its market value. Arthur Andersen was considered to be one of the top auditing companies and as such, companies were willing to pay a premium due to perception of safety and expertise. This changed immediately when Arthur Andersen was proven to be guilty for shredding documents. In this moment not only has the audit company lost its reputation and credibility, but so have its clients (Linthicum et.al, 2010.) Corporate reputation is an important determinant that institutional and individual investors take into consideration before investing in a company (Dowling, 1986.). This example proves how delicate, fragile and important corporate reputation is. Therefore it is important to know how and when companies can protect themselves against negative publicity.
We can conclude that negative news has an impact on customers’ attitudes and behavior.
Nevertheless, the question remains whether the news always have an impact or whether there are situations in which customers are protected from negative news. The impact of negative publicity and moderators has been researched by Ahluwalia (2000) study that focused on customer
commitment and its moderating role between consumers and negative publicity. Results have shown that consumers whose brand commitment is high are less likely to go through attitude
3 change towards the company and are expected to counter-argue when faced with negative
information, while on the other hand consumer who have a lower commitment to the company are also less inclined to counter-argue this negative information. Hence, higher commitment customers are less likely to go through attitude changes. This research was further expanded by Einwiller, et.al (2006) who focused on consumer company identification and its ability to prevent negative associations within customers, when they see the company exposed to negative publicity. In detail, this study tested how likely the customers would be to invest in the company if they had money to invest while dealing with different levels of crisis caused by negative publicity. It concluded that consumers with strong company identification have less negative associations when being exposed to negative publicity, unless it exceeds a certain extremity level. Results have implications for many companies as they show the specific ways in which customers who identify with the company react to negative news that they are exposed to. The research upon negative publicity and customers’ response to it has been looked upon from another perspective as well (Griffin et.al, 1991; Sherrel et.al 1985). Griffin et.al (1991) had determined three main dimensions that influenced the relation between negative news and customers’ attitudes and behavior. The dimension that is important for our research is source credibility. In their research it was determined that when less credible sources are reporting negative news, the recipients are more likely to keep more positive attitudes and purchase intentions towards the company.
4 1.1 Research question
This master thesis focused on providing a deeper insight into the “shielding” effect of two negative publicity moderators that are source credibility of a news source and customer commitment to the company, two important moderators of the effects of negative publicity on customer behavior and attitudes towards the company. In an experiment a crisis situation was created where the company was dealing with negative publicity. This thesis argues that an important moderator on the effects of negative publicity is not only the customer commitment or identification with the company but also the credibility of the source the news report is coming from. Results further expand the research of Einwiller et al. (2006.) and Ahluwalia (2000) by giving us a new insight into the turning point in which customers truly change their attitudes and behavior towards a company.
The way that the existing gap will be filled is by implementing an important factor in the research that is credibility of the source that is reporting the news. This will at the same time expand the research of Griffin et.al of (1991) on the impact of different credibility sources and customers purchase intentions but this time not only of products of the company but also the company stocks. Griffin et.al (1991) focused on the effects of source credibility for negative news but not on the degree of the negativity what will prove to be a highly important factor. Therefore in this thesis will be tested if the “shielding” effect of low source credibility appears also for extremely negative news and if so, it will be further looked into what type of customers does it occurs for, the low or high commitment ones? By doing this research the relation between negative publicity from different credibility sources and customers will be further explored.
5 Concluding, this master thesis analyzed in how far source credibility of negative information about a company affects customer attitudes and their behavior towards the company, basing its analysis on the assumption that this relationship is mediated by level of customer commitment towards the company reported. Based on these analyses, results have shown the point at which customers change their attitudes and behaviors towards the company. In summary, this study focuses on the question up to what turning point do source credibility and customer commitment protect customers from negative news; when does customer attitudes and behavior (i.e. purchase intentions) towards the company change or the “shielding” effect persists?
The following sub-questions should be answered before answering the research question of this thesis:
How do different levels of negative publicity affect customer attitudes and behavior intentions?
What is the effect of customer commitment on the relation between negative publicity and customers attitudes and behavior intentions?
How does source credibility moderate the relation between negative publicity and customers attitudes and behavior intentions?
2. Theoretical background
2.1 Negative publicity
Negative publicity has been defined by Sherrel and Reidenbach (1983) as: “Noncompensated dissemination of potentially damaging information by providing disparaging news about a product, service, business unit, or person via print, broadcast media or by word of mouth.”
Nowadays companies in order to attract and keep clients must more than ever take care not just of their product quality but of their reputation as well. One of the main issues companies must deal with is negative publicity that has had, as mentioned earlier, a steady growth in news coverage over the past years. As such negative publicity represents possible threats to
companies, a marketer’s worst nightmare. A better insight in how consumers react to negative publicity could give the marketers the knowledge of how to best deal with such situations as they occur or how to best prepare for them in advance. Dean (2004) focused on three factors that affect consumers’ response to negative publicity that are: companies’ response to the event, company reputation for social responsibility prior to the event and companies’ responsibility for the event.
Companies and marketing departments spend billions of dollars in the United States, as they are trying to influence consumer attitudes (Griffin et.al, 1991). By doing so they are potentially trying to prevent possible damage that might occur when dealing with a crisis situation that could give a devastating blow to the company. Media is rather reporting negative news than positive (Dennis & Miller, 1996.), hence companies are more likely to receive negative press than positive.
7 There are many articles that have focused on the effects of negative publicity and consumer responses to them (Griffin et.al, 1991; Sherrel et.al 1985., Ahluwalia, 2000.). A significant finding has been reported on that relation in the research of Mizerski (1982). The researcher focused his research on positive and negative types of information and concluded that unfavorable ratings compared to positive ratings on the same attributes, trigger stronger attributions to product performance, belief, strength and affect towards the product. These findings were a direct expansion to the research of Richey et.al (1975) whose results shown that a single negative behavior neutralized five positive behaviors.
Another important aspect of negative publicity is that people consider harder negative
information then they do positive when they are forming an overall evaluation of a product or a company. A great example in support of this information is the research done by Maheswaran and Meyers-Lewy (1990) who have found that when the processing of information is focused on message content, negative framing was more effective then positive framing. Another reason for this aspect of the negativity effect they have concluded was due to the fact that people consider negative information to be diagnostic and informative then positive information (Maheswaran and Meyers-Lewy, 1990), hence it is weighted more than positive is in the evaluation of people, objects and ideas (Mizerski, 1982). Different countries have different cultures. In the Netherlands a dominant source of information is newspaper journals what has been recognized by Renkema and Hoeken (1998). The focus of their study was on the
seriousness of the damage caused by the negative publicity as well as the duration of the negative effects. Their findings implied that not just the negativity of the news could affect the customers’ attitudes and behavior but the source of information as well.
8 How consumers interpret the negative news has a critical impact upon their associations towards the company. Corporate associations include all the beliefs and information a person holds about a certain company (Brown & Dacin, 1997). As mentioned earlier negative information has a greater weight than positive, therefore it is likely to have a strong negative impact upon forming of corporate associations. Hence it is likely that negative information about a product or a company will result in negative associations. As such the associations are likely to negatively influence behavior of customers towards the company and therefore the sales of products, and at the same time profits of the company and its stock price.
The question that imposes itself is whether negative information always has to influence
customers’ attitudes and behavior. A great example that helped to answer that question was when the Mercedes-Benz A-Class rolled over in 1997, during the elk test (which is designed to assess the ability of the car to evade objects in case of emergency), the brand’s reputation was not lastingly damaged, as it seemed to have been protected (Einwiller et.al 2006). This was explained by the research of Ahluwalia et.al (2000) that had shown that when costumers are highly committed to a brand, they are likely to dismiss the negative information in a biased manner. Afterwards Einwiller et.al (2006) continued the research on the shielding effect of customer commitment by connecting it to a very similar and connected construct of customer identification. Their research has found that customers, who highly identify themselves to the company, will keep their attitudes and behavior intentions positive and unchanged towards the company as long as the negativity of the news presented is not extreme.
The research upon the “shielding” effect of different moderators was continued by Norton et.al. (2011.) who challenged the fact that negative publicity will always have an impact on
9 the relation between consumer attitude changes and impact of negative publicity. In their study was found that negative publicity is less likely to affect high powered individuals unless the negative information is highly credible or the consumers’ confidence is high.
2.2 Customer commitment
Research in consumer behavior has shown that consumers are becoming more and more attached to different brands and are forming attitudes towards them ( Fournier, 1998.), what at the same time has a positive impact on brand equity (Keller, 1993). The attitudes created are diverse in strength from each other, so Petty & Krosnick (1995) have done a research and concluded that people with stronger attitudes will exhibit greater resistance to negative information.
Commitment has been identified as one of the major dimensions of attitude strength (Krosnick et.al, 1993.).
Over the years there has been a lot of research upon commitment and identification. Because of that there have been a lot of variations on what customer commitment is (see Hrebiniak& Alluto, 1972; Kanter, 1968; Sheldon, 1971; Brown, 1969) A lot of the mentioned research focuses on commitment related behavior, in a way that peoples’ behavior exceeded normative expectations. Based on the already mentioned research commitment can also be defined in terms of attitudes, attitudinal or organizational commitment. Meyer & Allen(1984, 1987) diversified among the following: commitment from necessity (continuance commitment), commitment from obligation (normative commitment) and the most researched one, most similar to organizational
10 Attitudinal or organizational commitment occurs when people connect their identity to the organization (Sheldon, 1971). Hence we can conclude that attitudinal commitment stands for a state in which a person identifies with a particular organization, its goals, and wants to stay within the organization in order to facilitate the goals. Crosby& Taylor (1983) stated that commitment is an emotional or psychological attachment that evokes within people a tendency to keep the same preferences towards an object, when presented with a conflicting experience or information. Shiv et.al (1997) found that commitment also affects the number of favorable or unfavorable thoughts that resulted from an exposure to news.
Mowday et.al (1979.) defined organizational commitment as:
“the relative strength of an individual’s identification with and involvement in a particular organization. It can be characterized by at least three related factors: 1) a strong belief in and acceptance of the organization’s goals and values; 2) a willingness to exert considerable effort on behalf of the organization; and 3) a strong desire to maintain membership in the organization.”
In the customer relationships literature, commitment was described as a lasting desire to continue an affiliation (Morgan & Hunt, 1994.) and resist conflicting information. By resisting conflicting information the authors Crosby& Taylor (1983.) meant that because of it, customers ignored negative information or blocked positive about a competing company. Funk & Pritchard (2005.) have continued the research upon the moderating effects of commitment. Their research was based on the impact that positive and negative print communication had on attitudes towards professional sport teams. Readers’ beliefs and feeling were trying to be manipulated by columnists. The results shown that commitment moderated the message effects, showing that
11 less committed readers were recalling more negative facts from the articles, while the higher committed fans tried to counter argue the negative information with more favorable thoughts.
2.3 Customer-company identification
The effect of organizational identification on the organizational members and organization has long been recognized (Brown, 1969). In order to understand the customer company
identification, we must first understand what identification is. Customer company identification is a form of organizational identification, based on the social identity theory (Hogg & Abramms, 1988). Burke et.al (2006) defined social identity theory as:
„a social psychological analysis of the role of self-conception in group membership, group process, and intergroup relations.“
The theory was used to address phenomena such as prejudice, organizational behavior,
leadership, group deviance, group cohesiveness and other. Identification is a cognitive construct, so in order to identify, customers must see themselves as psychologically intertwined with the future of the group (Ashforth and Mael, 1989). It was proposed that organization identification is a perception of unity or belongingness to an organization, as members and individuals people defined themselves in the terms of the organization. Hence the conclusion could be made that organizational identification is a perceived unity with the organization, in which people experience the organization’s successes and failures as their own (Ashforth and Mael, 1992).
12 More and more companies are trying to build long lasting relationships with their customers, but very few of them are truly successful in doing so. Bhattacharya & Sen (2003) have determined the consumer company relation as the main reason for the high committed, and meaningful relationship that marketers are constantly trying to build with the customers. The main focus of their research was determining under what conditions consumers were entering into the long lasting and committed relations with certain companies. One of the findings was that the strongest consumer company relationships were based on consumers’ identification with the companies that helped them satisfy one or more of their key self-definitional needs.
Even thou identification increases and grows over time, it is still possible that a person identifies with a company that is yet unknown to him or her. Such identification is possible when
consumer and the organization share the same values (Einwiller et.al 2006) or better to say the perceived identity of the company (Dutton et.al 1994). The companies’ identity is shaped by the organization’s mission, structure, processes and climate. When strongly identified consumers’ who have positive beliefs about a company are exposed with negative information about the company they are likely to try and preserve those beliefs, however the protective effect of identification should reach a limit when negativity increases (Bhattcharya & Sen, 2003).
2.4 Customer commitment versus customer identification
Van Knippenberg and Sleebos(2006.) differentiated between customer identification and customer commitment. As both identification and commitment reflected psychological connections between people and companies, the differences that were found were that identification reflected self-conception, whereas commitment was more attitudes oriented in
13 nature and had its roots in social exchange processes, employees’ attitude towards their jobs and the company.
From the mentioned theory upon commitment and identification we can conclude that they are extremely interconnected. Hence we shall for the brevity of this paper use the term commitment.
2.5 Source credibility
As mentioned earlier over the past years there had been an increasing number of negative publicity reports. Hence a better insight into negative publicity literature may give the marketers a better understanding of consumers’ reactions and as such marketers may more efficiently deal with such situations. To better understand and answer the question:” How does negative publicity influence consumers’ attitudes and behavior?” researchers have looked upon on relation from many perspectives. Firstly researchers explored the persuasiveness of the message what was done by McGuire (1978) who focused on identifying the components of persuasive conversation and found them to be the following: source, channel, message, receiver and destination variables. The source variable had three main aspects, which were: credibility, attractiveness and power. Source credibility was determined as a good moderator and answer to the proposed question (Griffin, et.al 1991; McGinnies and Ward, 1980; Berlo,et.al., 1969; Petty, et.al., 2002, 2004).
Griffin et.al (1991) had focused on the impact of negative news upon customers’ attitudes and intentions. While doing so they have determined three main dimensions that moderated the relationship. These dimensions were locus of responsibility, source credibility and performance history. The researchers determined that when less credible sources are reporting negative news,
14 the recipients are more likely to keep more positive attitudes and purchase intentions towards the company.
When looking into the source credibility literature, it is important to mention that over the years different dimensions of it have been discovered. The most influential ones according to Berlo et.al (1969) are safety, qualification and dynamism. McGinnies and Ward (1980) had continued the research upon source credibility and its two main components that are trustworthiness and expertise. Their research concluded that both trustworthiness and expertise contribute
significantly to higher credibility of the source, but also that trustworthiness has a greater impact then expertise. Trustworthiness was defined as the degree to which an audience perceived the messages valid made by a communicator, while expertise referred to the level to which the message recipients found the communicator to be believable. Other dimensions of source credibility were proposed in many studies. For example Whitehead (1968) proposed four dimensions: trustworthiness, competence, dynamism and objectivity, while McCroskey (1966) proposed only two: authoritativeness and character.
Petty et.al (2002) had looked at source credibility from a different perspective. They have based their research on examining of what happens when people resist persuasion. The conclusion was that participants in their experiment actually became more certain of their initial attitudes in situations when they resisted persuasion. Petty et.al (2004) continued the same research trying to define in what factors had to be satisfied in order for the participants to increase their initial attitudes. The conclusion was that participants’ initial attitudes increased when they resisted persuasion coming from high expertise sources, therefore sources with high credibility.
15 The research on source credibility and its effect on persuasiveness had been further explored by Pornpitakpan (2004), whose goal was to determine whether it was more beneficial for companies to use high versus low credibility sources and vice versa. It was found that a high credible source is commonly considered to persuade more towards the advocacy then a low credible one.
Marketers have long proposed that more credible sources are more influential then low credible ones (Griffin et.al 1991; Pornpitakpan, 2004). The conceptual background for this assumption had been supported by traditional attitude theory and attribution theory, that suggested that people changed their beliefs as they were exposed to persuasive messages. Source credibility was a crucial ingredient when creating such messages (Griffin, et.al. 1991).
The research upon source credibility had been contradictory over the years. McGinnies (1973) concluded a highly credible source was more persuasive then a low credible one when the initial opinion of the message recipient was very negative, but no source credibility effects were found when less negative initial attitudes were held. These finding were contrary to those of Dholakia and Sternthal (1977) that found no source credibility effects even thou the high credible source was found to be much more expert and trustworthy in comparison with the low credible one. The same contrast in results was found in the research of Weiner& Mowen (1986) who found that consumer discounted the message in situations where the source was expected to be biased. The opposite findings were found by Wegner, et.al (1981) that manipulated the media source
credibility(New York Times versus National Enquirer) to influence the presented information. The results shown that regardless of the source, the presented information was found equally effective.
3. Conceptual model
As mentioned in the previous chapter there have been studies (Griffin et.al, 1991; Sherrel et.al 1985., Ahluwalia, 2000; Mizerski, 1982) that focused on the effect of negative publicity upon customers’ attitudes and behaviors, meaning that the more extreme the negativity of the news reported, customers’ attitudes and purchase intentions towards the company will be lower. The main effect of our study is the difference in attitude change between moderate and extremely negative news as was done in previous research of Einwiller, et.al (2006). Hence for the first hypothesis of our study we assumed that depending on the extremeness of the news negativity, customers’ attitudes and behavior intentions would significantly differentiate.
H1: The more negative the news the more negative (lower) the attitudes and behavior intentions would be from consumers towards company.
As mentioned in the previous chapter Ahluwalia (2000.) researched and proved that customer commitment moderates the relation between negative publicity and customers attitudes. This thesis will further expand the existing knowledge by implementing an important factor that is the level of negativity of the news in a way that Einwiller et.al. (2006.) tested the relation of different levels of negative publicity and the moderating effects of customer identification with the
company. This was done in order to prove that customers who identify themselves with a company were able to justify negative publicity until the moment it became extreme. In the following hypothesis, we assumed that customers with high commitment to the company would
17 have more positive attitudes and behavior intentions than low committed customers, due to the “shielding” effect of customer commitment. As customers are high committed they are more likely to counter argue negative news.
H2: Customer company commitment moderates the relationship between negative news and customers attitudes and behavior intentions. It moderates the relationship as it “shields” high committed customers’ attitudes and behavior intentions from negative news by keeping them more positive than those of low commitment customers. This moderation effect occurs up to the moment when the negativity of the news becomes extreme. When given extremely negative information, attitudes and behavioral intentions will not vary between weak and strong identifiers
In hypothesis one the impact of the negative information on customers’ attitudes and behavior was discussed. In hypothesis two the moderation effect of customer commitment to the company on the relation was tested. In the final hypothesis three, a gap in the research needs to be tested by adding the source credibility as a second moderator to the model. Marketers have been interested for a long time and researched source credibility. As one of the main characteristics, it has long been assumed that high credible source is more influential than a low credible one (Sternthal et. al, 1977) Source credibility was chosen as one of the main dimensions affecting customers attitudes and behavior while researching the impact of negative public publicity (Griffin et.al , 1991). It was done in the way that the same information was presented to be coming from two different sources and was determined that the lower the source credibility the more positive the attitudes and intentions of customers were. In our research we assumed that the
18 source credibility will moderate the impact of customer commitment on the negativity, attitudes and behavior intentions relation. By doing so customers, who are high committed to the company would “shield” themselves from extremely negative news, when it reported from a low credible source. We are expecting that the extremity of the news would be counter argued due to the low credibility of the source and the high commitment of the customer to the company. This would provide a new perspective on the turning point in which customers change their attitudes and behavior towards the company.
H3: When exposed to extremely negative news, both low and high commitment customers are influenced by the news. Source credibility creates a moderated moderation upon customer commitment in such a way that even when exposed to extreme negative news high commitment customers will counter argue the negativity in case the news is reported from a low credible source. If the source credibility is high and negativity extreme, the “shielding” effect will not occur. The low commitment participants will be influenced by the negativity regardless of the source credibility.
19 Figure 1: Graphical illustration of the conceptual model
Negative publicity (moderate, extreme)
Customer behavior intentions Customer attitudes towards the company
Credibility of the source reporting
negative news(low,high) Customer commitment(low, high)
with the company Hypo 3 H yp o 2 Hypo 1
A pre-test was designed in order to ensure that the manipulations worked as predicted, meaning that the texts that would be used for the main experiment had recognizable characteristics needed for specific treatments. The characteristics that needed to be ensured were, that the participants differentiated between the negativity of the information presented in the texts and the source credibility that the information was coming from.
The pretest consisted of 40 students (male: 17, female: 23; =24.05, =1.319, age
range: 22-27) from Western countries.
The study was designed using the online survey tool Qualtrics. Questions to assess text
negativity and manipulation texts were based on the study by Einwiller, et al. (2006) on a 7 point Liker scale where the lower the number presented more extreme negativity (“I found the
performance of the company Conscience Funds in the above mentioned text was excellent”, and “I found that the performance of the company Conscience Funds was very favorable”), while the two questions for source credibility, were also measured on a 7 point Likert scale where also the lower the value chosen the lower source credibility was. The source credibility scales were based
21 on a study by Dholakia (1987) (“I find the above statements to be very trustworthy”, and “I find that the above statement has a highly expertise source”).
The pre-test consisted of four different conditions. Depending on the condition they were put in, participants were presented with one of the four possible texts that were used for the research of Einwiller et.al (2006) :
a) moderate negative news from a highly credible source
A well known business newspaper The Wall Street Journal focuses its research in investment companies. This year it has created a list of 163 funds that it analyzed on the basis of their financial performance, meaning the lower the rating the worse the financial performance of the company was in this year. In this list, Conscience Funds was listed on the 117th place out of 163, due to the poor earnings and overall bad financial performance of the company. Furthermore because of the low ranking, analysts have advised to keep a close look on the near future performance of The Conscience Funds and if the bad financial performance continues to consider selling your shares if you have any within The Conscience Funds.
b) moderate negative news from a low credible source
A celebrity gossip magazine "Stars Magazine" usually contains one page about investment companies and stocks. This year it has created a list of 163 funds that it analyzed on the basis of their financial performance, meaning the lower the rating the worse the financial performance of the company was in this year. In this list, Conscience Funds was listed on the 117th place out of 163, due to the poor earnings and overall bad financial performance of the company.
22 future performance of The Conscience Funds and if the bad financial performance continues to consider selling your shares if you have any within The Conscience Funds.
c) extremely negative news from a low credible source
A celebrity gossip magazine "Stars Magazine" usually contains one page about investment companies and stocks. This year it has created a list of 163 funds that it analyzed on the basis of their financial performance, meaning the lower the rating the worse the financial performance of the company in this year was. In this list Conscience Funds was listed in last place at 163rd. Due to the financial losses the company made, analysts have advised to sell your shares as fast as you can before they lose all their value.
d) extremely negative news from a high credible source
A well known business newspaper The Wall Street Journal focuses its research in investment companies. This year it has created a list of 163 funds that it analyzed on the basis of their financial performance, meaning the lower the rating the worse the financial performance of the company in this year was. In this list Conscience Funds was listed in last place at 163rd. Due to the financial losses the company made, analysts have advised to sell your shares as fast as you can before they lose all their value.
The study was distributed among participants via email and social media. Participants who decided to participate in the study, read the study information and agreed to the informed consent. Afterwards they were randomly assigned to one of the four conditions and presented
23 with a text as was done by Einwiller et.al (2006). After reading the text they were asked to answer two questions measuring negativity of the news and two questions measuring credibility of the source. After answering the questions they were thanked for participation and that was the end of the pre-test.
The pretest was analyzed in SPSS (Statistical Package for Social Sciences). An independent sample t-test was conducted to test for the difference between the source credibility groups. The test shown that in the low source credibility group the perception of the news was M=1.90 with an SD=0.552, while in the high source credibility group the mean level was M=6.02 with SD= 0.850. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s F test, F (38)=3.118, p=0.085. The independent samples t-test was associated with a statistically significant effect, t (38)= 18.192, p=0.00. Thus, the high source credibility group was associated with a statistically significantly larger mean then the low source credibility group participants. The difference in the means of the high and low source credibility was clear with a great statistical difference in value therefore we concluded that our source manipulation worked.
Another independent t-test was conducted in order to test difference in negativity between groups. The test shown that the high negativity group had a perception of news negativity of M=1.82 with an SD=0.654, while the low negativity group had a mean of M=2.72 with
SD=0.715. Additionally, the assumption of homogeneity of variances was tested and satisfied via Levene’s F test, F(38)=0.275, p=0.603. The independent samples t-test was associated with a statistically significant effect, t(38)=-4.15, p=0.00. Thus, the extreme negativity group was associated with a statistically significantly lower (more negative) mean then the moderate
24 negativity group of participants. For our study we needed texts that would be perceived as
moderately and extremely negative with a significant difference. Our texts have satisfied these criteria but we must emphasize that even though they have satisfied the criteria both values of the texts were considered quite negative with mean values of 1.82 and 2.72. Therefore in our study we needed to achieve quite strong customer company commitment in order to counter argue these negativity levels.
Based on the results of our analysis, we concluded that our manipulation was successful and there was a significant difference both for the negativity and source credibility of the texts created, hence we could use the texts for our main study experiment.
5. Main Study
The main study used a 2 (commitment: high vs. low) x 2 (source credibility: high vs. low) x 2 (negativity of news presented: high vs. low) between subjects design. The independent variable was negativity of the news presented (high vs. low); the moderators were customer commitment (high vs. low) and source credibility of the news source (high vs. low); and the dependent variables were customers’ attitudes and behavior intentions towards the brand presented in the news.
The main study made use of the information texts that were tested in the pre-test (see Materials section of the pilot study). Moreover, the following scales were used to evaluate participant indications of the variables.
25 Negativity of the news
The independent variable negativity of the news was measured with 2 items on a 7 point Liker scale (1 = Strongly Disagree; 4= neither agree nor disagree; 7 = Strongly Agree; example “In my opinion performance of Conscience Funds was excellent”). The scale was selected from the research of Einwiller et.al (2006). Negativity of the news was measured in order to ensure the differentiation in thoughts valence among the participants upon the negativity of the texts presented. The lower the value of negativity of the news variable meant that the more negative the participants found the information in the texts.
An 8 item measure was used to measure customer commitment that was selected from the research of Einwiller et.al (2006). The first five items were measured on a 7 point Likert scale (1 = Strongly Disagree; 4= neither agree nor disagree; 7 = Strongly Agree) to determine the strength of their identification with the company. These are the examples of items that have served to measure the connection with the company ( “ I am somewhat associated with Conscience Funds” , “I have a sense of connection with the Conscience Funds” , “ I consider myself to be a part of the group of people who are in favor of Conscience Funds” , “ Customers of Conscience Funds are most likely similar to me” , “ Employees of Conscience Funds are probably similar to me” ). The following three items on a 7 point Likert scale (1 = Strongly Disagree; 4= neither agree nor disagree; 7 = Strongly Agree) measured the perceived overlap in beliefs with the investment fund and to what extent this will be self-referential for participants (“Conscience Funds shares my values”, “Being a customer of Conscience Funds is a part of
26 sense of who I am “and “Purchasing Conscience Funds mutual funds would help me express my identity” ). Customer commitment was measured as it was a moderator affecting the impact of negative news on customers’ attitudes and behavioral intentions towards the company in the experiment. The higher the value of the customer-company commitment variable, ensured the higher commitment to the company. All the participants whose commitment was measured above 4 on the 7 point Likert scale were considered to be high committed to the company.
Social Responsibility Values
Participants social responsibility values were measured as their commitment to the Conscience Funds should be highly correlated to these values as the introductory text in the experiment was about a company with high corporate social responsibility to which the participants were
supposed to relate and commit. Social responsibility values were measured using a question on a 7 point Likert scale (1 = Strongly Disagree; 4= neither agree nor disagree; 7 = Strongly Agree; “It is extremely important for me that companies behave responsibly when it comes to social and environmental matters”) that was selected from the research of Einwiller et.al(2006). The higher the value of the variable meant that participants held social responsibility more important, with the values considered to be positive above 4 on a 7 point Likert scale.
As one of the dependent variables people’s behavior intentions were measured with a 7 point Likert scale ( 1 = Strongly Disagree; 4= neither agree nor disagree; 7 = Strongly Agree)on the
27 basis of agreement with the following three statements (“Would you keep your 50 000 €
inheritance invested in the Conscience Funds “, “How likely would you be to invest more money in the Conscience Funds, if you had more money to invest?”, “ If a friend asked you for advice about investment, how likely would it be that you recommended Conscience Funds to him or her?”). The higher the value of the variable meant that participants had more positive intentions towards the company in the experiment, with truly positive attitudes being above the value of 4 on a 7 point Likert scale. The scale selected to measure the participants’ behavioral intentions was taken from the research of Einwiller et.al (2006).
As a dependent variable participants attitudes towards the company were measured using a 7 point Likert scale (1 = Strongly Disagree; 4= neither agree nor disagree; 7 = Strongly Agree) on the basis of agreement with the following two statements (“After reading the text, I have very positive emotions towards the Conscience funds company” and “After reading the text, I have very favorable feelings for the Conscience funds company”). The higher the values of customers’ attitudes meant that the more positive they were towards the company. The scale used to measure participants’ attitudes in the experiment was selected from the research of Einwiler et.al (2006).
Investment knowledge variable
At the end of our experiment participants’ knowledge was assessed. Firstly they were asked if they have money invested in stocks, bonds or mutual funds in order to have a better
understanding of our sample. Afterwards they were asked how knowledgeable they are about mutual fund investing where they could choose on a 3 point Liker scale between (below average,
28 average and above average knowledge). The investment knowledge questions were selected from the research of Einwiller et.al (2006).
On the beginning of the experiment, participants were told they are participating in an online survey to see how people evaluate investment options. They were guaranteed anonymity and were told that the data would be used for the sole purpose of this paper. All the participants were put into a situation where they have inherited 50 000€ from a family relative and the money is invested into mutual funds managed by the fictitious company Conscience Funds. This scenario was used due to increased chances of identification and therefore higher commitment to the company. After the introduction participants were asked to read a one page flier that was used to indicate the company had strongly held corporate values in corporate responsibility. The flier was also emphasizing on the company creating mutual funds only with other companies that satisfy the screening requirements upon corporate social responsibility. The social responsibility was emphasized upon, in order to determine the low identifying from the high identifying
customers or better to say the high commitment from low commitment customers. This was done because the customers for whom the corporate social responsibility is important were more likely to commit to a company that is doing a lot of corporate social responsible work. The following flier was selected from the research of Einwiller et.al (2006):
Conscience Funds enables investors to align their financial goals with their personal values through a selection of professionally managed socially responsible mutual funds. By screening companies not only by their investment potential but also by standards of social responsibility, Conscience Funds challenges companies to reach for a higher "bottom line" and offers investors the opportunity to do good while doing well.
29 After reading the flier participants answered an 8 item measure that measured customer company commitment. Participants social responsibility values were measured too as their commitment to the Conscience Funds should be highly correlated to these values.
After participants have completed the part with social values and commitment level, depending if their commitment was high or low they were assigned to one of the 8 conditions. As being assigned to a specific condition they were presented with one of the texts that were checked in the pretest about the investment fund their money was invested in. This was done in order to manipulate source credibility and negativity of the news. Depending on the condition they were put in, they could have been presented with one of the four possible texts for both low and high commitment participants. The texts stated that the news came from either the well known business magazine “The Wall Street Journal” what was used as a high credibility source or the celebrity gossip magazine “Stars magazine” that was used as a low credible source. Depending on assigned condition, the text contained either moderate or extremely negative news. The story in the text contained information about the annual rankings of the mutual fund earnings. The financial performance of the company was chosen to be manipulated as it mentioned as one of the central dimensions of corporate reputation (Fombrun et. al. 2000.) and because it was likely not to have an impact on the social responsibility. In the moderate negativity scenario,
participants learned that the Conscience Funds earnings this year were poor, that the company
was ranked on the 117th place out of 163 possible and that the analysts recommended considering
selling shares. In the extremely negative scenario participants learned that the Conscience funds ended the year on the 163rd last place, that they were losing investors’ money and that analysts recommended immediately selling shares(see Methods section of the pre-test for full texts).
30 Following exposure to one of the texts the participants were asked to answer several questions that were used to measure their behavior and attitudes as described in the materials section of the main analysis.
Participants were then asked how they perceived the company to be portrayed in the article in order to ensure the difference in thoughts valence about the negativity of news. Finally
participants were asked if they have money invested in stocks, bonds or mutual funds and how knowledgeable they are about mutual fund investing, as well for their age and country.
Afterwards they were thanked for their participation, and that was the end of the study.
338 participants took part in our main study. 56 participants were excluded due to incomplete study and another 14 participants were excluded as they could not be fitted into a high or low commitment group due to giving all the answers as a 4 (neither agree nor disagree) on all the 7 point Likert scales that were used to measure commitment. Out of the 268 participants remaining 174(65%) considered themselves to have an average or above average knowledge of investing and 94(35%) participants considered themselves to have below average knowledge of investing. After excluding the participants with below average knowledge of investing from our analysis, we analyzed data of 174(avg. age=25.8 years) participants. From the 174 participants remaining 60(35%) had money invested into stocks or bonds while 112(65%) had not money invested into stocks or bonds. 2 participants did not answer whether they had invested money or not.
5.1.5 Missing values & reliability
In order to test our hypotheses firstly the data set had to be checked for missing values. We have found five missing values in the items measuring commitment, one missing value in items measuring negativity, attitudes and behavior. The missing values were replaced by the mean value of the missing variables. After replacing missing values variables that measured negativity, commitment, attitudes, behavior and sense of connectivity with the company had to be checked for reliability. All of the variables had a Cronbach’s alpha value above 0.7 what clearly showed they were reliable. The following table shows the results of reliability analysis:
Table 1: Reliability values
VARIABLE N of items Cronbach’s α
Commitment 8 α = 0,958
Negativity 2 α = 0,939
Behavior int. 3 α = 0,900
Attitudes 2 α =0,943
5.1.6 Normality of distribution
We have computed new variables and thereafter normality of the data distribution had to be checked. In order to do so skewness and kurtosis had to be analyzed. For a normal distribution skewness and kurtosis must be must be in range from -1 to +1 (Tabachnick and Fidell, 2001.). As the values of skewness and kurtosis were out of that range for the variables negativity,
32 commitment, behavior and attitudes, we have computed new variables in order to have normal distribution. As data were not normally distributed, we logarithmically or exponentially transformed the data on the variables used so that they satisfy the normality assumption standards placed by ANOVA measurements. The values of kurtosis and skewness of the new variables Behaviornorm, Attitudesnorm, Commitmentnorm and Negativitynorm were as presented in the table:
Table 2: Skewness and kurtosis
VARIABLE N Skewness Std.error Kurtosis Std.eror
Commitmentnorm 174 -0,386 0,184 -0,715 0,366
Attitudesnorm 174 0,092 0,184 -0,933 0,366
Behaviornorm 174 -0,245 0,184 -0,727 0,366
Negativitynorm 174 -0,394 0,184 -0,509 0,366
5.1.7 Correlation of variables
When looking at the correlations between all the five variables, the table 3 shows that they were all significantly correlated on the 0.01 level. As expected, participants commitment to
Conscience Funds was significantly and strongly correlated to the extent to which they held their social values (r=0.787; p=0.00). This was a very important correlation as in our introductory text on the basis of which the participants had to commit and identify themselves with the company was appealing on their social values and was talking about corporate social responsibility of the company. Another important significant and strong correlation was between negativity and participants attitudes(r=0,771, p=0.00) and behavior intentions(r=0,775, p=0.00) what confirmed
33 that we could further test hypothesis one. The relation was positive due to the fact that we
measured higher negativity on lower values of the 7 point Likert scales that we used. Hence the higher (lower value on the scale) perceived negativity meant lower attitudes and behavior intentions value(for detailed information check the materials section of the main study). Therefore the relationship was positively correlated.
Participants commitment to the company was significantly and moderately correlated to their attitudes(r=0,483, p=0.00) and behavior intentions (r=0,461, p=0.00) what confirmed that we could further test our hypothesis two. The positive correlation meant that the higher the commitment values the higher their attitude and behavioral intentions would be as well. The participants social values were also as to be expected moderately correlated to their
attitudes(r=0.444) and their behavior intentions(r=0.421) on the significance level of p<0.01. As the correlation strength was moderate and the relationship itself positive, it meant that the higher the social values participants shown the higher the values of their attitudes and behavior
intentions would be.
5.1.8 Groups descriptives
By answering questions in the study participants were categorized as explained in the variables section of the paper as either high (N=73) with a customer commitment average group mean of 5.1197 with SD=0.636, or low commitment (N=101) with a customer commitment mean of 2.41 with SD=0.674. According to their commitment level they were assigned to one of the 4
conditions for each type of participant, hence into 8 groups. When looking at an overall commitment level value we can conclude that the high commitment group mean average (M=5.1197) was not very high as the scale measured on a 7 point Likert scale while the high Table 3: Correlations table
Variable 1 2 3 4 5 1 Commitment - 2 Social values 0,787** - 3 Attitudes 0,483** 0,444** - 4 Behavior int. 0,461** 0,421** 0,781** - 5 Negativity 0,271** 0,248** 0,771** 0,775** -
** Correlation is significant at the 0.01 level
35 commitment participants were considered those above the value of 4. We can also conclude that the low commitment participants truly did not feel committed as their average group mean was a mere 2.41. The relatively weaker strong commitment to the company was connected to their measured social values to which it was highly correlated (table 3). Next we looked at the average mean value (M=5.77) of social values measured of the high commitment participants groups, hence we could conclude that the participants in the sample were not very oriented towards the social values on the basis of which they were supposed to commit and identify themselves with the fictional company used in the experiment. The negativity of the texts participants found quite negative both for extreme negative news and for the moderate negative news texts (table 4). As last we looked at the participants expressed attitudes and behavior intentions that were all mostly quite negative (table 3). Further exploration of the variables was done in the hypothesis testing section.
36 The following table presents the 8 groups(names explained in Appendix) that the participants were divided into:
Table 5: Condition, number of participants per group, and means of variables as shown
Conditions N Commitment Soc.values Negativity Attitudes Beh.intentions
LHH 25 2,34 2,96 1,56 1,46 1,48 LLH 25 2,46 3,08 2,30 2,58 2,4 LLL 29 2,43 3,17 2,53 2,87 2,54 LHL 23 2,36 3,39 2,06 2,36 2,31 HHH 20 5,1 5,75 2,47 3,65 3,00 HLH 18 4,86 5,67 2,72 3,44 3,66 HLL 16 5,11 5,94 2,37 3,37 3,22 HHL 17 5,33 5,72 3,16 3,86 3,35 TOTAL 174 - 5.1.9 Hypotheses testing
In hypothesis one, we assumed that less negative news will result in more positive customers’ attitudes and behavior intentions towards the company. From the conceptual model figure 1 (see conceptual model section) we tested the impact of negativity upon participants’ attitudes and behavior intentions.
We used an independent t-test to check for the impact of negativity of the news upon customers’ attitudes. Firstly, before using a t-test we checked the assumption of homogeneity of variances that was not satisfied, via Levene’s F test, F (160.118)= 11.289, p=0.001. An independent
37 sample t-test conducted, showed that extremely negative texts had a significantly stronger effect on customer attitudes than moderately negative texts, t (160.118)= -2.593, p=0.010. Therefore, when comparing the two means of participants attitudes connected to the extreme
news(M=0.3829, for more information check tables 12 and 13 in Appendix) and strength of attitudes of the moderate negative news(M=0.4737), we could conclude that more negative flier texts led to lower attitudes that participants had towards the company. Cohen’s d was estimated at d=-0.409 what was a medium sized effect.
To fully test hypothesis one, we conducted another independent samples t-test for customers’ behavior intentions. Firstly, before using a t-test we checked the assumption of homogeneity of variances that was not satisfied via Levene’s F test, F(152.614)=20.867, p=0.00. An independent sample t-test conducted, showed that extremely negative texts had a significantly stronger effect on customers’ behavior intentions than moderately negative texts, t (152.614)= -2.579, p=0.011. Therefore, when comparing the two means of participants behavioral intentions (M=0.3505, for more information look at tables 12 and 13 in Appendix) caused by exposure to the extreme negative news with the mean of behavioral intentions (M=0.4405) caused by the exposure to the moderate negative news, we concluded that more negative flier texts led to more negative behavior intentions that participants had towards the company. Cohen’s d was estimated at d=-0.417 what was a medium sized effect.
As there was a significant impact with a medium sized effect of negativity both on customers’ attitudes and behavior intentions, we accepted hypothesis one.
Hypothesis two assumed that the customers’ attitudes and behavior intentions are not only influenced by the negativity of the news they are exposed to, but that customers’ commitment to
38 the company that moderates the relationship. According to this assumption, the effect of
exposure to negative news on customers’ attitudes and behavior intentions would be smaller, hence customers who have a high commitment to the company would show more positive attitudes and behavior intentions until the moment the negativity of the news became extreme, due to the commitments’ “shielding” effect.
Firstly we have checked for the main effects of customer commitment on the dependent variables customer attitudes and behavior intentions by conducting two separate analysis of variance (ANOVA).
Before conducting the analysis of variance for customer attitudes we checked the assumption of homogeneity of variances that was satisfied via Levene’s F test, F (1, 172)= 1.736, p=0.189. Afterwards we have performed an analysis of variance (ANOVA) that yielded a significant effect, F (1, 172)= 36.370, p=0.00. Therefore, when comparing the two means of participants attitudes connected to their high commitment (M=0.545, check tables 8 and 9 for more
information in Appendix) and attitudes connected to low commitment participants (M=0.3464), we could conclude that the higher commitment the more positive the attitudes would be, what confirmed the positive correlation between the variables. The effect size was estimated by eta
square with estimated value of 2 0.175 what made it a medium effect size (Field, 2013). An
important fact was found while looking at the descriptive table of the attitudes variable before we exponentially and logarithmically transformed it in order to have normal distribution. We could conclude that even though high commitment participants had more positive attitudes (M=3.58) then low commitment participants (M=2.35), the overall attitudes for both groups were low and therefore quite negative (appendix table 8).
39 Next we checked the effects of commitment on customer behavior intentions. Before conducting the analysis of variance we checked the assumption of homogeneity of variances that was
satisfied via Levene’s F test, F (1, 172)= 0.184, p=0.669. Afterwards we have performed an analysis of variance (ANOVA) that yielded a significant effect, F (1, 172)= 28.384, p=0.00. Therefore, when comparing the two means of participants behavior intentions connected to their high commitment (M=0.501, check tables 10 and 11 in Appendix) and behavior intentions connected to low commitment participants (M=0.3219), we could conclude that the higher commitment the more positive the behavior intentions would be, what confirmed the positive correlation between the variables (table 3). The effect size was estimated by eta square with
estimated value of 2 0.142 what made it a medium effect size (Field, 2013). An important
fact was found when looking at the descriptive table of the behavior intentions variable before we exponentially and logarithmically transformed it in order to have normal distribution. We could conclude that even the high commitment participants had more positive behavior
intentions (M=3.29) then low commitment participants (M=2.21), the overall behavior intentions for both groups were low and therefore quite negative (appendix table 10).
Moderating effect customer commitment
In order to fully test hypothesis two we had to check for the moderating effect of customer commitment on the relation between negative news and customer’s attitudes. Therefore the moderator variable (commitment) and the independent variable (negativity of the news) have been centered and multiplied in order to calculate the interaction effect. Next a multiple regression has been conducted (table 6) to examine the significance of this effect. The
40 Coefficients table indicated that the multicollinearity assumption was not violated
(Tolerance=0.964). The normal probability plot indicates no major deviations from normality since the point were lying on a reasonably straight line (Appendix figure 2) The Scatterplot (Appendix: figure 3) was concentrated in the middle and no points were outside the ±3,3 (no outliers) (Tabachnick and Fidell, 2001). The model was evaluated with the R square =0.589, hence 58.9% of the variance of the dependent variable customers’ attitudes could be explained by the variables commitment and negativity of news. The adjusted R square had a value of 0.582 which was very close to the observed value of R square (0.589) indicating that the cross-validity of the model was very good. The significance of the model was appointed with the p value of p=0.113 what made the model insignificant for customer attitudes as a dependent variable.
Furthermore negativity of the news made the strongest unique contribution to explaining the dependent variable with standardized β=0,641 in comparison to customer company commitment with β=0,338. The unstandardized β value for negativity of the news was higher with β=0,611 than customer commitment β=0,412. These values were used in the regression equation,
indicating that a decrease(higher value on the measurement scale) in negativity by 1 results in an increase in customer attitudes by 0,611. This finding further supported hypothesis one. If
commitment increased by 1 that would have resulted in an increase in attitudes by 0.412. This shown the possibility for moderation effects of commitment. In our study we did not find a significant effect of moderation of customer commitment (p=0.113) upon their attitudes, hence we could reject hypothesis two for customer atittudes.
41 Table 6: Regression analysis attitudes
Variable unstand. B Std. error Stand. B Sig Tolerance
(constant) 0.431 0.012 0.000 Commitment 0.412 0.061 0.338 0.000 0.964 Negativity 0.611 0.049 0.641 0.000 0.964 Interaction com.*neg -0.395 0.249 -0.081 0.113 R 0.767 R square 0.589 Adj. R square 0.582 F 81.220 N 174
Next to further test hypothesis two we had to check for the moderating effect of customer commitment on the relation between negative news and customer’s behavior intentions. A multiple regression has been conducted (table 7) to examine the significance of this effect. The Coefficients table indicated that the multicollinearity assumption was not violated
(Tolerance=0.964). The normal probability plot indicates no major deviations from normality since the point were lying on a reasonably straight line (Appendix figure 4) The Scatterplot (Appendix: figure 5) was concentrated in the middle and no points were outside the ±3,3 (no outliers) (Tabachnick and Fidell, 2001). The model was evaluated with the R square =0.607, hence the model explained 60.7% of the variance of the dependent variable customers behavior
42 intentions. The adjusted R square had a value of 0.600 which was very close to the observed value of R square (0.607) indicating that the cross-validity of the model was very good. The significance of the model was appointed with the p= 0.254 what made the model insignificant for customer attitudes as a dependent variable.
Furthermore negativity of the news made the strongest unique contribution to explaining the dependent variable with standardized β=0,681 in comparison to customer company commitment with β=0,290. The unstandardized β value for negativity of the news was higher with β=0,647 than customer commitment β=0,352. These values were used in the regression equation,
indicating that a decrease in negativity by 1 results in an increase in behavior intentions by 0,647. This finding further supported hypothesis one. If commitment increased by 1 that would have resulted in an increase in behavior intentions by 0.352. This shown the possibility for moderation effects of commitment. In our study we did not find a significant effect of moderation of
customer commitment (p=0.254) upon their behavior intentions, hence we could reject hypothesis two for customer behavior intentions.
43 Table 7: Regression analysis behavior intentions
Variable unstand. B Std. error Stand. B Sig Tolerance
(constant) 0.398 0.011 0.000 Commitment 0.352 0.060 0.290 0.000 0.964 Negativity 0.647 0.048 0.681 0.000 0.964 Interaction com.*neg -0.278 0.243 -0.057 0.254 R 0.779 R square 0.607 Adj. R square 0.600 F 87.426 N 174
To conclude we did not find a significant moderation effect of customer company commitment upon attitudes and behavior intentions and therefore we had rejected hypothesis two.
Hypothesis three was supposed to explore a moderated (multiplicative) moderation of source credibility upon the tested moderating effect of commitment upon the relation between negativity of the news and customers attitudes and behavior intentions. As the moderation of customer commitment to the company was not found in our research further exploration of hypothesis three was not possible. Hence we had to reject hypothesis three. In the discussion section we