Social media petitions matter.
Social media petitions, shareholder proposals and firm
responsiveness
Amsterdam Business School
Final version
Date: 22nd of January, 2018 By: Daphne Sweers
Student number: 10460675
MSc Thesis Business Administration International Management Track
Supervisor: Michelle Westermann-Behaylo Second reader: Lori DiVito
Statement of Originality
This document is written by Student Daphne Sweers who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is 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. Abstract………...5
2. Introduction………...6
3. Literature review………...8
3.1 Stakeholder activism………...9
3.1.1 Social shareholder proposals……….9
3.1.2 Social media petitions………11
3.1.3 Comparison social shareholder proposals & social media petitions…..13
3.2 Firm responsiveness………...14
3.3 Firm reputation………...17
4. Conceptual model………....18
4.1 Shareholder proposals and firm responsiveness……….18
4.2 Social media petitions and firm responsiveness……….20
4.3 Firm reputation as moderating effect……….21
5. Methodology………23
5.1 Research design………...23
5.2 Research measures………..25
5.2.1 Dataset shareholder proposal………..25
5.2.2 Dataset social media petitions……….27
5.2.3 Control variables………28
5.3 Method………29
6. Results………..30
6.1 Descriptive statistics………..31
6.1.1 Descriptive statistics shareholder proposal dataset………..31
6.2 Ordinal regression output………34
6.2.1 Shareholder proposal dataset………...34
6.2.2 Social media petition dataset………36
7. Discussion……….39
7.1 Discussion descriptive statistics………39
7.2 Discussion of hypotheses………..40
7.3 Managerial implications………43
7.4 Implications for other parties………44
7.5 Limitations……….45
7.6 Future research………..46
8. Conclusion………...47
9. References………...50
Appendix A: examples shareholder proposals………....55
Appendix B: examples social media petitions………..55
1. Abstract
This study examines whether degree of pressure positively impacts firm responsiveness in the context of social activism. The traditional form, shareholder proposals, is compared to the emerging modern form, social media petitions, by analysing the two datasets. The moderating effect of firm reputation is also taken into account, expecting to negatively influence the relationships mentioned above because of a buffer effect. Data is mostly gathered manually from the action website Change.org and SEC filings. Degree of pressure was conceptualized by ratio of stock ownership for shareholder proposal and number of petition supporters for social media petitions. Firm responsiveness was coded ordinally in line with Ashforth and Gibbs (1990). Ordinal regression analysis looked into shareholder proposals and social media petitions received by 25 of the largest firms in varying industries in a longitudinal setting. The predicted relationship between degree of pressure of social media petitions and firm
responsiveness proved to be significant. The other predicted relationships did not have a significant outcome. The contribution of this research thus lies in the significant result that when the degree of pressure of social media petitions increases, the level of firm
responsiveness also increases. It is, as far as is known, the first empirical investigation of this emerging concept and its significant result has important implications of executives and other parties.
Keywords: shareholder proposal, social media petition, firm responsiveness, degree of pressure, social activism
2. Introduction
In 2014, Victoria’s Secret launched a marketing campaign with the slogan ‘The Perfect Body”, featuring 10 models with the same body type. This marketing campaign led to such indignation that an online petition was launched to protest against their unhealthy and unrealistic standards of ‘The Perfect Body’. People were asked to sign the social media petition and tweet #iamperfect to Victoria Secret to spread the word and show their support in this matter. Within weeks, the petition had over 30,000 supporters and created social media turmoil. As response, Victoria’s Secret eventually changed the slogan to ‘A Body for Every Body’. This is just one example where the impact of a social issue was magnified through the power of a social media petition.
A social issue is not a new phenomenon in the business environment and thus has been researched extensively in previous literature (Groysberg, Lin, Serafeim & Abrahams, 2016; Tybout & Roehm, 2009; Bundy, Buchholtz and Shropshire 2013; McDonnell & King, 2013; David, Bloom and Hillman, 2007). Amongst other topics, scholars have examined how firms can respond to such issues (David et al., 2007; Bundy et al., 2013), and what the effects are on firm reputation (Tybout & Roehm, 2009; Perrault & Clark, 2016). Despite this coverage there are emerging factors that pressures firms’ responsiveness to a social issue, which require more attention (McDonnell, 2015; McDonnell & King, 2013; Waldron et al., 2013). One emerging factor is the social media petition, which sheds a new light on the established literature of social activism. Do social issues that arise through social media petitions pressure a different level of response from firms, opposed to the more conventional way of shareholder
proposals? Interestingly, the impact of these forms has not been compared yet while it can infer more about the relative power of the two concepts. Moreover, the concept of social media petitions has not been investigated empirically to date (Shirky, 2011). Recent research involved social media up to 2007, but not social media petitions specifically (McDonnell,
2015).There is thus a dire need for empirical research to examine the impact of social
activism in the recent decade with a new perspective. This research tries to bridge this gap by empirically testing the impact of social media petitions on the degree of firm responsiveness to social issues, compared to the more conventional way of shareholder proposals and their impact. Moreover, it examines the moderating effect of firm reputation. This will be investigated through the following research questions:
RQ1: To what extent do shareholder proposals have an impact on firm responsiveness, moderated by firm reputation?
RQ2: To what extent do social media petitions have an impact on firm responsiveness, moderated by firm reputation?
sTo answer these questions, the degree of pressure of social media petition and degree of pressure of shareholder proposals will be examined in relation to the firm responsiveness from S&P 500 companies. Degree of pressure of shareholder proposals and the firm’s responsiveness are extracted from the database WRDS as well as coded manually. Degree of pressure of social media petitions and the corresponding firm’s responsiveness are extracted and coded from the social change platform Change.org. Moreover, the moderating effect of firm reputation will be examined, in line with research such as McDonnell & King (2013). Firm reputation will be conceptualized using the Most Admired Firm ranking from Fortune.
This thesis contributes to current literature in several ways. First, the impact of social media and firm responsiveness will empirically be investigated, building upon the limitation noted by Shirky (2011) and Bundy et al. (2013). Moreover, this is the first thesis to assess the difference in impact between shareholder proposals and social media petitions. Whereas previous research looked into the process of shareholder proposal and the outcome (Tkac, 2007; Perrault & Clark, 2016), it has never been put in perspective to the new form of social activism: social media petitions. The results show a significant relationship between the
degree of pressure of social media petitions and firm responsiveness, which provides a new insight for the relative power of shareholders. While adding to the academic literature by confirming the power of social media petitions in relation to firm responsiveness, this outcome is also relevant for managers. When faced with social activism in the form of a petition, executives will be aware how and to what degree they should respond. Especially, this study shows that they should acknowledge the power that stems from social media petitions and form their response accordingly.
This thesis continues with a review of the relevant, contemporary literature about the constructs of the research questions.Next, the conceptual framework will present the links of the constructs as well as hypotheses of this study. After that, the methodology will explain the research design, the conceptualization of the variables and the procedure of analysis. The results section will lay out the outcome of this research. In the discussion, the results will be examined as well as focusing on the implications and limitations of the study. Suggestions for future research are also included. Finally, a conclusion will shortly summarize this thesis and its findings.
3. Literature review
Having introduced the topic of this research, the literature review will focus on current literature to extend knowledge on this field. In the first section, stakeholder activism is discussed. Here, the division is made between the more traditional manner of stakeholder activism, social shareholder proposals, and the more modern form: social media petitions. The latter concept will first address the relatively new phenomenon of social media before
continuing with the relevant literature concerning social media petitions. Next, the definition of ‘firm responsiveness’ will be provided, as well as how a firm can respond and what internal and external factors influence such response. Last, firm reputation will be discussed.
3.1 Stakeholder activism
Stakeholder activism will be regarded in a broad type of definition, namely “a player that seeks to change the practices of the firm” (Baron, 2001, p. 12). The goal of such action is twofold: to influence the perception of the stakeholders and to compel the target firm to alter a specific aspect of their strategy or policy (McDonnell & King, 2013). Previous research shows that the use of activism provides great challenges for strategy and practices of organizations (Eesley & Lenox, 2006; McDonnell & King, 2013; McDonnell & Werner, 2015). When faced with activism, firms need to think about strategies to defend themselves (McDonnell & King, 2013) and consider devices that help them manage social issues
(McDonnell & Werner, 2015). Eesley and Lenox (2006) point out that a social issue raised by activists also can be a severe economic burden for the target firm. This thesis examines two types of stakeholder activism: social shareholder proposals and social media petitions.
3.1.1 Social shareholder proposals
In this thesis, the focus is specifically on proposals that concern a social issue. A social shareholder proposal is seen as a tool to achieve social change by altering business practices (Logsdon & Van Buren, 2009). To set the process in motion, the shareholder activists first detect a social issue that needs to be addressed and select a strategy to achieve that. Then, the party either informs the firm about their concerns or immediately files a shareholder proposal. Once the firm received such proposal, it is bound by laws from the Securities and Exchange Commission (SEC) to omit, withdraw or vote upon the proposal (Logsdon & Van Buren, 2009; Gillian & Starks, 2007). As the name indicates, only a vote results in shareholders having their say about the proposal during the annual meeting. Omission of the proposals means that firms are allowed to exclude it, whereas withdrawal signals a type of action from the firm which resulted in filers to withdraw their proposal (Tkac, 2006). Already active in
the 1940s, the number of social shareholder proposals increased considerably over the years thanks to the formalization of the process and the founding of social responsibility centres (Logsdon & Van Buren, 2009; Goranova & Ryan, 2014).
Despite an increase in popularity, the true impact of a social shareholder proposal remains a challenge to pinpoint. First of all, success is only achieved when the action or response from the firm is what the shareholder demanded. It is, however, not always clear from the proposal what the shareholder really wants to achieve. Perhaps attributable to the challenge of interpretation, the outcome of a social shareholder proposal shows conflicting results in the literature. On one hand, stakeholders experience beneficial effects from social activism (Reid & Toffel, 2009), and it can be linked to an increase in profits and positive publicity (Tkac, 2006). In that light, a social shareholder proposal is an effective avenue for realising change. On the other hand, David et al., (2007) argue that the impact can be discounted because a social shareholder proposal serves as window dressing to reflect their commitment to social change or merely functions as a political play called ‘signalling’, without really acting to it. Logsdon and Van Buren (2009) agree by concluding that the act of a shareholder proposal is understated, as they very rarely win votes during annual meetings. Actual achievement occurs more through dialogue, which takes place in a non-public setting and thus is difficult to analyse (Logsdon & Van Buren, 2009; Gillian & Starks, 2007).
With whom, however, does the dialogue take place? Merely looking at the target firm does not cover the full spectrum of a shareholder proposal. It is also important to examine who pressures the firm to make adjustments (Goranova & Ryan, 2013).The shareholders that are behind those proposals can be divided into different groups. First, there are individuals who may pursue social change in the firm that they deem essential on an individual level (Tkac, 2006). On the other hand, the majority of social shareholder proposal filers consist of organized groups. Pension funds, labor unions, religious organizations, social organizations
such as PETA and socially responsible mutual funds are all active in social activism (Tkac, 2006; Monks, Miller & Cook, 2004; O’Rourke, 2003). Each of them has different motives behind their actions. Labor unions want to ensure that their members will benefit from a social shareholder proposal (Tkac, 2006) whereas religious organizations tend to focus more on social and economic justice (O’Rourke, 2003). Nonetheless, they all share the drive to generate social change within a firm. The socially responsible mutual funds are most strategic in nature as they invest in stocks of a firm while submitting proposals to change the behaviour of the firm (Tkac, 2006).
3.1.2 Social media petitions
Before diving into the relevant research concerning social media petitions, the concept of social media will be addressed first. This step is necessary to establish the context in which social media petitions come to existence and what the impact of such petition can be.
Social media
As stated in the introduction, social media is a relatively new phenomenon in the field of research. This thesis will address the term social media as defined by Aula (2010, p. 43): “Social media is characterized by interactivity- participants freely send, receive and process content for use by others. Social media services include social networking, content producing, the distribution of services and websites that are collectively constructed by users.”. Scholars have only recently recognized the magnitude of its role. As mentioned by Aula (2011), the immense popularity of this medium makes it a factor that cannot be ignored in the academic field. The power of social media can be found in a variety of disciplines; the impact on political change (Shirky, 2011), the effect on an individual (Drobocky, 2013) and of course the consequences for the business environment (Aula, 2010). Despite of the evidence
provided by the scholars mentioned above, research is still lacking empirical investigation (Shirky, 2011; Wiederhold, 2012). Nevertheless, some implications of social media are already uncovered in the academic field.
An important implication of social media is the impact it has on corporate strategy and practices (Aula, 2010; Hennig-Thurau, Hofacker & Bloching, 2013). Because the
environment is virtually entirely open, with low boundaries for the audience to engage and participate, firms can no longer control the discussions of which they are the objective (Aula, 2010; Hennig-Thurau et al., 2013). Thus, an increase in active participation and networked interconnectedness has drastically increased the power of the consumer (Hennig-Thurau et al., 2013; Wiederhold, 2012). With loss of control and power, firms need to reconsider their current organizational practices and routines to be able to provide an adequate response when faced with a social media petition (Hennig-Thurau et al., 2013).
Social media petitions
Whereas the more traditional setting of activism is offline, this thesis focuses also on the impact of online activism. Online activism does not differ much from offline, it merely moves the classic tactics from an offline to an online setting (Earl, 2006). As such, the goal of the tactics does not change. In general, there are four tactics that can occur online: boycotts, letter-writing campaigns, e-mail campaigns and petitions (Earl, 2006). This thesis will specifically investigate online petitions, here addressed as social media petitions.
A social media petition can be defined as “A statement published online that individuals can sign as a show of support” (Earl, 2006, p. 362). This is a relatively unexplored tactic in the current field of social movement theory (Earl, 2006). Of course, scholars have dedicated attention to the concept of a petition before. Caren, Ghoshal & Ribas, (2011), for instance, show a stark increase in petition signing over time, in contrast to protests.
With the emergence of Internet, however, various scholars have built upon that by looking into the relationship of social media and a petition (Atengble, 2014; Earl & Elliott, 2016; Valenzuela, 2013). Atengble (2014) illustrates that social media was used to ensure reach of the Ghana’s 2012 Election Petition, but this still describes merely the online promotion of an offline petition, rather than a petition that was created online. Moreover, online activism is more likely to take place when an individual learned about an occurrence via a social networking site (Earl & Elliott, 2016). Work of Valenzuela (2013) suggests that might be because social media used for communication of opinion and activism, works as mediator in protest behaviour. Nonetheless, research about the impact of social media petition, thus as an independent variable, remains unexplored.
As social media petitions occur through online channels, the forces behind social media petitions generally consists of individuals. Earl (2006) describes the act of signing a social media petition as individual participation. This description, however, indicates that whereas the signing is a one person action, it does contribute to a larger group with many more participants. Also large organizations can be behind a social media petition (Earl, 2006), yet each signature online still comes from a single person. As such, the force behind a social media petition can be both individual and group effort.
3.1.3 Comparison shareholder proposals and social media petitions
As mentioned in the introduction, this thesis will address shareholder proposals and social media petitions as independent variables. The two concepts were chosen for one resemblance; they are both linked to stakeholder activism as explained by Baron (2001). Both serve to generate change in the practices of the firm, in this context by resolving a social issue (Baron, 2001). Nevertheless, there are also differences described in the literature between them.
terms. Shareholder proposals stem from shareholders, who are investors of the firm. As such, they hold a direct interest in the firm, as well as a certain level of control over the practices of the firm (Gillian & Starks, 2007; Vasi & King, 2012). In contrast, any stakeholder can launch social media petitions with an indirect interest in the firm (Vasi & King, 2012) and a lower degree of control. Vasi and King (2012), classify the latter as ‘secondary stakeholders’ and the former as ‘primary stakeholder’.
The second difference stems from the previous one. Because investors are involved in the firm, any proposal that they file is bound by rules and regulations (Gillian & Starks, 2007). As described above, the process that shareholders have to go through is very formal, even closely monitored by the SEC in the U.S (Logsdon & Van Buren, 2009). Social media petitions, on the other hand, can be launched by anyone from everywhere. As pointed out by Earl (2006), online petitions are hosted by several providers that are run by both individuals and large organizations, allowing millions of signatures for thousands petitions. The entire process is quite informal, with core goal to encourage public debate (Vasi & King, 2012). The last distinction is quite unmistakable, yet for the integrality important to mention. A
shareholder proposal is a traditional, offline process, where as social media petition only occurs online and is regarded as a very novel phenomenon.
3.2 Firm responsiveness
As elaborated upon above, a firm can receive a social shareholder proposal and/or a social media petition. Once this has occurred, the firm can, or is in the case of a proposal even obligated by law to, respond. The responsiveness of the firm, however, can differ. To define firm responsiveness, the definition provided by Bundy et al. (2013) will be used: “The degree to which a firm is willing to provide a thoughtful response to stakeholder concerns and commit to continued work on the issue” (p. 352). The explanation of this term indicates the scale of such response, as there numerous ways possible. Here, the focus will solely be on the
responsiveness of a firm faced with a social issue. A social issue is an umbrella term for a wide range of issues: “environmental, moral or religious, animal rights, civil rights or
diversity and human or labor right issues” (McDonnell, 2015, p. 60). It is important to make a distinction between a governance issue and a social issue as the content differs greatly and is thus likely to trigger different reactions.
Several authors have examined how a firm responds to a social issue. One focus point can be the action from a firm to assess three level of responsiveness when facing an activist shareholder proposal (David et al., 2007; Logsdon & Van Buren, 2009). Their research defines three categories from least responsive to most responsive where they would challenge, oppose or settle the proposal (David et al., 2007; Logsdon & Van Buren, 2009). The latter two approaches can be linked to the research of Ashforth and Gibbs (1990). Ashforth and Gibbs note that a firm can respond in a symbolic or in a substantive way to realize legitimacy (1990). When responding in a symbolic manner to a social issue, there is no actual output. Instead, the firm achieves in appearing to comply with the requests of the stakeholders. Examples of such behavior are denial and concealment, redefining means and ends and ceremonial conformity (Ashforth & Gibbs, 1990). David et al., link therefore symbolic response to the act of ‘window dressing’ to satisfy activists (2007). In contrast, a substantive response involves real change in goals, structures and social practices (Ashforth & Gibbs, 1990). As such, the firm meets the requests of the stakeholders and acts to it. Examples of substantive actions are termed coercive isomorphism, adjusting resource dependencies and conforming to socially institutionalized practices (Ashforth & Gibbs, 1990).
Bundy et al. (2013), also look into the manner of reply from Ashforth & Gibbs 1990). Their starting point for addressing the firm’s responsiveness, however, is issue salience. Issue salience is “the degree to which a stakeholder issue resonates with and is prioritized by management” (Bundy et al., 2013, p. 352). They argue that the relationship of the issue with
both the organizational identity and strategic framework can be unrelated, conflicting or consistent, from which several courses of action can follow. Those courses of action can either be substantive or symbolic (Ashforth & Gibbs, 1990; Bundy et al., 2013).
When examining factors that play an external role in the responsiveness from the firm, the focus shifts from the target firm, to the group that pressures such reaction. Here,
stakeholder salience theory aids in explaining the reaction that the action group elicits (Mitchell, Agle & Wood, 1997; Eesley & Lenox, 2006; David et al., 2007). Mitchell et al. (1997), proposes a framework whereby the salient, the important, stakeholder has three attributes: power to pressure the firm, urgency of the issue and legitimacy of relationship with the firm. Eesley and Lenox (2006), and David et al. (2007) build upon that framework by empirically investigating whether the three attributes have an impact on the executive’s
perception of importance of the issue, in turn influencing the level of responsiveness. Whereas David et al., (2007) find evidence that the responsiveness of the firm is more likely to be satisfying when the stakeholder group is salient (thus possessing the three attributes), Eesley and Lenox (2006) only offer strong support for the attributes of power and legitimacy.
Last, there are also internal factors that can impact the responsiveness from the firm. Important to consider are those who decide what the response will be: executives from the firm. According to Hambrick and Mason, organizations act as they do because they are
“reflections of the values and cognitive bases of powerful actors in the organization” (1984, p. 193). Guided by their cognition and values, managers will respond to issues according to their interpretation (Hahn, Preuss, Pinkse & Figge, 2014). When managers have a more business case oriented framework, opposed to a more paradoxical framework, they sense a high level of control over the issue and interpret it as either clearly positive or clearly negative (Hahn et al., 2014). On the contrary, corporate sustainability can also provide a cognitive context for executives (Grewatsch & Kleindienst, 2015). Managers with a focus on corporate
sustainability in their cognitive framework prioritize social issues and view them as opportunities instead of threats (Grewatsch & Kleindienst, 2015).
3.3 Firm reputation
The reputation of the firm can be defined as the way firms are perceived among a broad range of stakeholders (Fombrun & Van Riel, 2004). It is essential to maintain a good reputation, as it is a key success factor that results in a high level of support from stakeholders, and works as a differentiating mechanism (Fombrun & Van Riel, 2004). Therefore, a high level of firm reputation is related to many positive economic outcomes (Lange, Lee & Dai, 2011; Decker 2011). Moreover, a good reputation can also function as a buffer when novel, negative
information about the firm arises (McDonnell & King, 2013; Lange, Lee & Dai, 2011; Decker 2012). Reputations are, however, a challenge to manage, because they consist of subjective aspects such as credibility, reliability and competence (Helm, Liehr-Gobbers & Storck, 2011). Any reputational threat can cast a shadow on the carefully constructed brand image of its stakeholders with drastic consequences (McDonnell & King, 2013). Wei, Ouyang & Chen (2016) for instance show the influence of reputation on loss of firm value.
As mentioned, a firm reputation can have a buffering effect. This is specifically the context in which reputation will be regarded in this study. Various scholars show that a positive firm image can reduce negative impact in several occasions (Lange et al., 2011; Decker, 2011; McDonnell & King 2013). When novel, negative information about the firm arises, a good reputation can dilute the loss of reputation (Lange et al., 2011). Moreover, when firms are targeted by boycotts, they are least likely to engage in responses that reduce the threat when they have a high reputational ranking (McDonnell & King, 2013). The
strength of the effect is reflected in research by Decker (2011), who found partial support that with a good reputation, the best degree of response to allegations would be complete denial.
On the other hand, a good reputation can result in certain expectations that the firm cannot live up to (Lange et al., 2011). Grunwald & Hempelmann (2010), for instance show that the buffer effect of a good reputation failed during a product crisis, negatively affecting the consumers’ perceptions of problem severity. Thus, research shows a reputation can function in general as a positive buffer but in some cases can also backlash on the firm.
The provided definition above does not reflect the recent evolvement attributable to the rise of Internet, which is important to note. As stakeholders are increasingly present online, this also has several consequences for perception of the brand. Kaul, … Carroll et al. (2015) for instance state that the perception of a company’s brand is shaped by the
information gathered through different channels: media, press releases, website, online resources and/or word of mouth. Such focus on online channels implies an additional difficulty for firms to control their reputation (Ott & Theunissen, 2015). Users expect open and honest communication from firms but also have the tools to uncover news that firms would not prefer to air (Greyser, 2009).
4. Conceptual model
The previous section reflected the relevant literature concerning the constructs of the research questions: firm responsiveness, shareholder proposals, social media petitions and firm
reputation. This section will build upon that by establishing the links between the constructs, as well as the expectations about these relationships.
4.1 Shareholder proposals and firm responsiveness
As stated in the literature review, shareholder proposals are one of the most traditional and formal tools to achieve social change (Logsdon & Van Buren, 2009). With the rise of the Internet and consequently the emergence of social media petitions, the impact of shareholder proposals has a new point of comparison. The process to file a shareholder proposal, as
described above, is one of several steps and thus can take a considerable amount of time. Linking this to the three attributes of the salient stakeholder as mentioned before, legitimacy, urgency and power (Mitchell et al., 1997), the latter two aspects might be more feasible to reach through social media than the traditional form of filing a proposal because of a shorter time span of the total process (Mitchell et al., 1997; Perrault & Clark, 2016). Momentum can thus be generated at relative high speed for social media petitions, but not for social
shareholder proposals. Moreover, actual achievement is more likely to be reached through the form of dialogue as noted by Logsdon & Van Buren (2009). The relatively closed, offline environment of filing shareholder proposals can have a detrimental effect on achieving said proposals in comparison to social media petitions.
Nevertheless, research did confirm the effectiveness of shareholder proposals and its beneficial effects (Gillian & Starks, 2007; Reid & Toffel, 2009; Tkac, 2006). This might be attributable to two characteristics of shareholder activism. First, the traditional, formal process ties a certain level of earnesty to the request of the stakeholders (Logsdon & Van Buren, 2009). Second, the power of the shareholders is rather high due to direct control (Gillian & Starks, 2007; Vasi & King, 2012). As they are in possession of stocks of the firm, they have the ability to steer the course of action of the firm to a certain degree through the voting system. The fact that they own a small part of the firm, in general indicates that they also hold a direct interest in the firm and thus also want to steer it to a certain goal (Gillian & Starks, 2007; Vasi & King, 2012). Combining the aspects above reflects a substantial degree of pressure a shareholder can have in the response of the firm. Building on that, we expect the degree of pressure of shareholder proposal to be positively linked to the responsiveness of the firm. As described in the literature review, an increase in degree of pressure will correspond to an increase in responsiveness to encompass a symbolic approach whereby the firm appears to comply with the proposal as identified by Ashforth and Gibbs (1990). The responsiveness
of the firm could also be of the highest degree: the substantive manner. In that case, the firm will reflect real change in goals, structures and or social practices (Ashforth & Gibbs, 1990). This leads to the following hypothesis:
Hypothesis 1: The degree of pressure of shareholder proposals, conceptualized by relative stock in possession of the proposal filer, will have a positive impact on the degree of firm responsiveness
4.2 Social media petitions and firm responsiveness
A petition to realize social change within a firm is not a new custom and has been discussed in the literature (Caren et al., 2011), but the rise of internet has marked a new variation of the concept: the social media petition. Even though the goal of a social media petitions does not differ from an offline petition, this change in setting has important implications. One
implication stems from the fact that a social media petition is characterised by a virtual open environment, with low boundaries to engage (Aula, 2010; Hennig-Thurau et al., 2013). When a social issue occurs, people will be more compelled to voice their opinion, simply because it is easier to do. Anyone who has an interest in the firm, direct or indirect, can show their support in the manner (Vasi & King, 2012), which can create a very large action group. As the individuals can now easily group together and stand strong together, the power of the consumer drastically increases in an online setting (Hennig-Thurau et al., 2013). In turn, the target firm becomes inferior when faced with a social media petition, losing some level of control.
Another important aspect of the virtual setting is the time span to create attention for the issue. A shareholder proposal is a formal process, which requires time to be processed (Logsdon & Van Buren, 2009). On the other hand, a social media petition can be launched by anyone at any time with a time span of a few clicks. The process is informal which allows support and discussion to be shown in a few clicks too (Vasi & King, 2012), from everywhere
over the world. This allows a social issue to create much more momentum in a very short time span. It may even go viral within a few hours. The ability to create attention at a rapid pace is also reflected in the example that introduced the topic, where the campaign about Victoria’s Secret had more than 30.000 supporters within weeks.
These aspects that encompass a social media petition have implications on the degree of responsiveness of the firm. Because of the high degree of power of the large action group of individuals that is usually behind a social media petition, as well as its ability to create a high degree of attentiveness in a short time span, the degree of pressure can be interpret as relatively high (Vasi & King, 2012; Henning-Thurau et al., 2013; Aula, 2010). It is thus expected that a firm feels pressured to respond in a very adequate way. After all, if they react in the least responsive degree, which is to provide no response and neglect the social media petition, the attention to the issue will not cease and people will not be satisfied about the firm, negatively affecting the firm reputation. As such, it is likely that firms will respond in a more satisfying manner, which is either symbolic or substantive: the two degrees of
responsiveness (Ashforth & Gibbs, 1990). With a substantive response, action will then mount to reflect real change in goals, structures and or social practices as noted by Ashforth and Gibbs (1990). Therefore, the following hypothesis is constructed:
Hypothesis 2: The degree of pressure of social media petitions, conceptualized by the number of petition supporters, will have a positive impact on the degree of firm responsiveness
4.3 Firm reputation as moderating effect
Last, firm reputation is also expected to have a moderating effect on the shareholder proposals – firm response and social media petitions – firm response relationship. The brand image that people have about the target firm can shape how people regard and react to social issues. If people are very favourable towards a certain brand, they can either be shocked when they
discover any issue with their brand or their natural, positive bias towards the brand might diminish the impact (Decker, 2011). With the rise of Internet, people expect and pressure firms into a response that they deem adequate in the shortest time span possible (Ott &
Theunissen, 2015). Also, as indicated by Kaul et al. (2015), online channels play an important role in shaping the perception of a brand. Here, it does not matter whether the form of
activism is through shareholder proposals or social media petitions, as both tactics can gain a lot of attention through online channels and create a reputational threat (Aula, 2010).
We expect firms with a good reputation to be shielded to a certain degree from shareholder proposals and social media petition, extending previous research. The negative relation that McDonnell & King (2013) show between firm reputation and boycotts is likely to be found in this setting too, as boycotts is also a form of activism. Research by Lange et al. (2011) states that negative information harms firms with a good reputation less. We expect a similar effect to occur for a social issue, as it in general does not put a positive light on the firm (as the Victoria’s Secret example displays). A firm with a high reputational ranking opposed to a firm with a lower reputational ranking can react to a social issue, to a certain extent, less satisfying because of the shield a strong reputation forms. Because of that buffer, firms with a good reputation that are targeted by (online) activism will feel less pressured to respond in the substantive manner Ashforth and Gibbs identified (1990). On the other hand, if a firm with a bad reputation is targeted with a social issue, the pressure to respond in an adequate manner increases, because they have nothing to fall back on. Thus, we expect to find that reputation functions as a buffer and influences the way a firm feels pressured to respond. This expectation results in the following two hypotheses:
Hypothesis 3a: A high level of firm reputation will have an increasingly
negative effect on the shareholder proposal – firm response relationship, compared to a low level of firm reputation.
Hypothesis 3b: An increase in the level of firm reputation will have an increasingly negative effect on the social media petitions – firm response relationship, compared to a low level of firm reputation.
The three hypotheses combined together form the following conceptual model:
Figure 1: Conceptual model
5. Methodology
The previous sections elaborated on examining current literature and eventually built up to the hypotheses that will be tested. In this section, the focus will be on the conducted research. First, the research design will be explained, followed by an elaborate explanation of the measures used. This section will close by shortly stating the method and procedure of analysis.
5.1 Research design
This research will investigate the impact on firm responsiveness from both shareholder
2006) yet no relevant empirical research investigating the difference of effect between shareholder proposals and the relatively new concept of social media petitions was found. As reflected in the conceptual model, the research design is constructed by employing two different datasets. One is focused on shareholder proposals, the firm’s degree of
responsiveness and the moderating effect of firm reputation. The other dataset is accordingly focused on social media petitions, the firm’s degree of responsiveness and the moderating effect of firm reputation. The sample consists of the 25 largest firms from the S&P 500, who have been targeted by social activism through a social media petition and a shareholder proposal in a longitudinal setting. S&P 500 firms serve as a suitable sample as previous research highlighted that social activism is most likely to occur with the largest, high status organizations within an industry (McDonnell, 2015; Rehbein et al., 2004).Therefore, each firm in the sample is likely to face the same pressure and influence when targeted by social activism, which increases the likelihood of triggering the same level of response. Moreover, the datasets specifically only include firms that received both a shareholder proposal and a social media petitions to accurately assess the difference in effect between the two
independent variables.
Both datasets will be longitudinal but with a different time span. Because of data collection constraints, shareholder proposals are examined from the time span 2007 – 2013. The data of social media petitions was collected in the time period 2007 – 2017 as the dataset showed it is increasingly more popular to post petitions online. A longitudinal study will accurately assess the impact of shareholder proposals and social media petitions on the firm’s degree of responsiveness as it allows mitigating effects that certain conditions, e.g. an
5.2 Research measures
For discussing the conceptualization of the different variables, this section is divided into three. First, the dataset of shareholder proposals will be discussed, then the dataset of social media petitions. Last, the control variables will be elaborated upon, as they are the same for both datasets.
5.2.1 Dataset shareholder proposal
Degree of pressure of shareholder proposals will be examined in line with research by David et al. (2007) and Eesley and Lenox (2006) by examining the relative power of ownership stakes. First, data from WRDS was extracted concerning S&P500 firms in the time span 2007 – 2013. These were filtered to only include social issues. Next, the list of firms was compared to the list of firms who received a social media petition. Firms that were not present in both files were also excluded from the sample to increase accuracy of comparison. Then, a list of the 25 largest firms was composed by examining their market capitalization after 2007. Reason is that larger firms have a greater likelihood to be targeted by social activism
(McDonnell, 2015; Rehbein et al., 2004). Appendix C displays a full list of the companies in the sample. Those 25 firms resulted in the final sample size, respectively 373 shareholder proposals across a variety of industries.
To conceptualize the degree of pressure, we did an intensive online search for each individual proposal to manually code the number of shares of the firm that were in possession of the proposal filer. This number was either displayed in the proxy statement of the firm or in the SEC fillings, as proof of ownership is required. For missing cases, a proxy of 1% of the total number of common shares outstanding was taken because this is one of the requirements a shareowner needs to fulfil to be able to submit a proposal (Ertimur et al., 2010). To
# shares in possession of proposal filer for firm x, year y # common shares outstanding of firm x, year y
A frequency analysis of three percentiles resulting in the following categories: Category 1 (Low pressure): 0 – 0.000166425 of stock in ownership
Category 2 (Medium pressure): 0.00017 – 0.001 of stock in ownership Category 3: (High pressure): 0.0011 – 0.399 of stock in ownership
Next, we looked into the degree of firm responsiveness to a shareholder proposal. As stated by Tkac (2006), it is a challenge to observe the response of the firm after a shareholder proposal. Because shareholder proposals are non-binding, the firm can decide for itself whether to act, or not (Tkac, 2006). This means that a firm can implement a shareholder request even though it was omitted, or that the firm can decide to neglect a shareholder
request even though it received a majority of votes in favour. Therefore, rather than taking the shareholder proposal outcome provided by Compustat, this thesis examines the response of the firm in line with the research conducted by Tkac (2006): using Internet searches,
proponent and company Web sites and annual reports to find information about the proposals. Thus, the 373 shareholder proposals that were gathered for the variable shareholder proposal were again individually investigated online to examine the actual response of the firm. This was mostly noted in the SEC filings but sometimes mentioned on the website of the proposal filer. Then, the response was assessed to place it in one the different categories of firm response noted by Ashforth and Gibbs, (1990): none, symbolic or substantive to adequately measure the firm response.
Last, ‘firm reputation’ is measured as reputational ranking; it’s ranking in the US Fortune’s “Most Admired Companies”. This is in line with previous research such as
McDonnell and King (2013) and Carter and Ruefli (2006), taking the overall score of the each firm present in the sample size. The scores come to existent by interviewing approximately
8,000 people (Carter & Ruefli, 2006). Lange et al. (2011) noted that firm reputation consists of three aspects: firm can be known in general, for a specific aspect or there may be favorable feelings in general towards the firm. However, as all three aspects can influence firm
reputation, we will regard firm reputation as a whole by examining the overall score of the reputational ranking. The reputational ranking for each firm was noted down for each year in which the petition or proposal occurred, to determine if there would indeed be any ‘buffer’ effect, as discussed above. As the reputational rankings are ordinal in nature (Carter & Ruefli, 2006), the variable was divided into three categories according to a frequency analysis, to accurately display either a low, medium or high level of firm reputation.
5.2.2 Dataset social media petitions
As mentioned, degree of pressure of social media petitions is conceptualized using data from the composed dataset of Change.org petitions. Social media petitions of firms from the Global Fortune 500 were examined to include only social petitions. At the start of the process, a group of six students searched for and coded social media petitions for a total of
approximately 630 firms. This exhaustive search resulted in 2467 petitions. Once the file was complete, the list was filtered to also include only the top 25 S&P 500 firms that received both a social media petition and a shareholder proposal. Correcting for missing values resulted in the final sample of 698 social media petitions in the time span of 2010-2017. Degree of pressure is conceptualized by noting the number of supporters each petition received. A frequency analysis divided all the cases into three percentiles, resulting in the following categories:
Category 1 (Low pressure): 0 – 34 supporters Category 2 (Medium pressure): 35 – 465 supporters Category 3: (High pressure): 466 – 648449 supporters
responsiveness to social media petitions was examined online at the Change.org website what the response of the firm was, if there was any. This could include an action whereby e.g. the product was removed from the stores or an actual statement issued by the firm to, for
example, adopt certain policies. Thus, all the 694 social media petitions were placed into one of the different categories of firm response from Ashforth and Gibbs (1990): none, symbolic or substantive. Last, firm reputation was conceptualized in the same manner as described above, for the dataset of shareholder proposals.
5.2.3 Control variables
Both datasets include the same control variables. The perceived level of social responsibility of a firm can impact the likelihood of being targeted by social activism. If a firm is known to be committed to social responsibility, it is likely that shareholders are more eager to file a petition or proposal as the likelihood is also higher that such requests will be implemented. To control for any of such effects, the membership of the firm in the UN Global Compact is taken as a proxy. At this largest international corporate social responsibility (CSR) event, firms voluntarily sign principles of sustainability and corporate citizenship and promise to report annually on their social activity (McDonnell, 2015). Even though it is difficult to determine the true impact of such initiative, it does signal a clear commitment of the firm to CSR (McDonnell, 2015).
Moreover, the success of a petition or proposal may depend on the level of effort that corresponds to the type of request (Eesley and Lenox, 2006). The adjustment of
organizational practices or a different strategy e.g. require a substantial amount of resources and time to implement. The request to adopt a principle, or to label a product, is less
consuming on the other hand. Thus, to control for the level of effort that the type of request may have, a dummy variable will be included that indicates the degree of effort involved with the request. The five coded categories were split, whereby adopting principles/pledges or
labelling products or processes is seen as a low level of requested effort, whereas a request for report on operations or adjustment of operational changes (with or without cost) is coded as a high level of effort (Eesley and Lenox, 2006).
We do not control for time because the collected data is longitudinal, covering a time span in the final sample of eight years and seven years. As mentioned, this should already flat out any short term effects such as an economic crisis. Also, we refrain from controlling for industry because we already pre-filtered to only include the largest firms who belong to different industries. Scholars noted that firm size was a major selection criterion for activist groups, whereas industry reflected conflicting results in the literature (Tkac, 2006; Logsdon & Van Buren, 2009). Therefore, the final sample size already controls sufficiently for industry. Focusing specifically on the 25 largest S&P 500 firms in terms of market capitalization in each industry, we immediately control for performance in the final sample too. As McDonnell (2015) noted, performance of a firm may influence the engagement in social activism, where low performance is generally linked to lower social change. Because the S&P 500 index is based on market capitalization and our research is longitudinal we already take the long term performance into account in our sample size.
5.3 Method
As mentioned before, firm responsiveness is measured by examining the categories ‘none’, ‘symbolic’ or ‘substantive’ response. Because of this ordinal ranking, the hypotheses will be tested through ordinal regressions (Harrell, 2015). Models that allow analysis of ordinal data are increasingly more popular (Taylor, West & Aiken, 2006). Advantageous of such
regression is that the model’s assumptions are more likely to be met compared to OLS regression when symmetry in the dataset cannot be guaranteed, as is the case here (Taylor et al., 2006). This increases the likelihood to run the model for analysis of the hypotheses.
Another advantage is that the model does not assume spacing between the different levels Y, only examining the rank ordering between the values (Harrell, 2015).
The two datasets will follow the same line of analysis to ensure proper comparison. After assessing the descriptive statistics, through correlation tables, we will turn to the ordinal regressions. For this thesis, we focus on the cumulative odds ordinal regression with
proportional odds model, including an interaction effect.First, the assumptions of ordinal regression model will be checked (Harrell, 2015). There are four assumptions that must be met. First, the dependent variable must be measured at the ordinal level and the independent variables must be continuous or categorical. Next, there should not be multicollinearity present. This will be tested by running a linear regression of independent and dependent variables, temporarily transforming the categorical variables in the model. The last assumption is that there are proportional odds present, so an independent variable has the same effect on the dependent variable as any other independent variable. This assumption is tested by performing a test of parallel lines. If the assumptions are met, we will continue running the ordinal regression for both of the datasets. To test for the moderation effect of firm reputation on the relationship between degree of pressure and firm responsiveness, we include the interaction effect of degree of pressure * firm reputation. Last, the parameter estimate of significant variables will be turned into the exponential to allow better judgement of the variable’s relative effect.
6. Results
After describing the method of analysis, this section will continue with noting the results of that analysis. First, some descriptive statistics will be provided for the dataset of shareholder proposals and the dataset of social media petitions. Next, the results of the ordinal regressions to test for the three hypotheses will be shown.
6.1.1 Descriptive statistics shareholder proposal dataset
A first look at the data through descriptive statistics allows some deduction of the examined variables. Firm responsiveness is slightly unbalanced, with 32.4% of the 373 cases falling into the response category of ‘none’, 48.3% in the symbolic category and 19.3% in the
substantive category. This is also reflected in a mean of 1.87 (SD: 0.708). Approximately the same distribution can be found in the degree of pressure of shareholder proposal, with
respectively 32.2%, 44.2% and 23.6% of the cases in the ‘low, ‘medium and ‘high’ pressure category. This resulted in a mean of 1.91 (SD: 0.743), or looking at the ratio, an average degree of pressure of 0.0144164. In this dataset, Exxon Mobil received the highest number of social shareholder proposals: 59 proposals. They also received the proposal that displayed the highest degree of pressure, 0.3944, which concerned hydraulic fracturing. Appendix A shows some random examples of social shareholder proposals that were included in the final sample.
Firm reputation is quite balanced, which is not remarkable as it was categorized by examining a frequency analysis. This resulted in a categorized mean of 1.96 (SD: 0.827), or a mean score of reputational ranking of 7.24. The highest reputational ranking in the dataset was awarded to Apple in 2012, receiving 8.42. The lowest reputational ranking of 6.06 occurred for Johnson and Johnson in 2012. Perceived social responsibility is unbalanced, where 90.6% is not perceived as socially responsible opposed to 9.4%, with a mean of 0.09 (SD: 0.292). From the descriptive statistics, we can also tell that the majority of the proposals require a rather high level of effort, with 17.7% of the cases only falling into the category of a low level of effort. This is also reflected in the mean of 0.18 (SD: 0.328).
The Spearman’s correlation table allows us to explore the association between variables before running regression analysis (see table 1). There are not many significant relations between the variables present. A negative, statistically significant medium correlation between perceived social responsibility and firm reputation can be observed. It
thus appears that when firm reputation increases, perceived social responsibility decreases and vice versa. There is a statistically significant small effect between firm reputation and firm response, which suggests that the firm reputation is mildly positively related to a change in firm responsiveness. Last, there is a small significant relation between firm reputation and degree of pressure, reflecting that a higher firm reputation can have a positive effect on the degree of pressure of SP and vice versa. In general, the poor presence of significant
correlations indicates weak relationships between the variables in the model.
Table 1: Means, standard deviations and correlations shareholder proposal dataset
Variables Mean SD 1 2 3 4 5 1. Firm response 1.87 0.708 . 2. Degree of pressure 1.91 0.743 -0.016 . 3. Firm reputation 1.96 0.827 .171** .144** .
4. Perceived social responsibility 0.09 0.292
-0.008 -0.085 -.297** .
5. Level of effort of request type 0.18 0.328 0.005 0.006 0.021 -0.053 .
**. Correlation is significant at the 0.01 level (2-tailed)
6.1.2 Descriptive statistics social media petitions dataset
The average response of a firm is 1.23, falling in the symbolic response category, with a standard deviation of 0.606. Unfortunately, the dataset is unbalanced with 87% of the cases falling in the first firm response category of ‘none’, 3% belonging to the firm response of ‘symbolic’ and 10% with a substantive response. Not unexpectedly, the degree of pressure of social media petitions is balanced, with a categorized mean of 2.00 (SD of 0.823) and an average degree of pressure of 12988 supporters. In this dataset Amazon received the highest number of petitions, 123 cases. The highest number of supporters a petition received was 645229, petitioning Walmart to “Stop this American Horror Story. Tell Walmart to Stop Torturing Pigs”. See appendix B for some random examples of social media petitions in the final sample size. Firm reputation was also expectedly balanced, with an average of 1.96 (SD of 0.802) which resulted in a mean reputational ranking of 7.05. The highest reputational
ranking of 8.6 was, again, attributed to Apple in 2016, whereas the lowest reputational ranking of 5.2 belonged to Walmart in 2016. On the other hand, perceived social
responsibility and level of effort of the type of request are unbalanced again. Approximately 70% is being perceived as not socially responsible and also the type of request that requires a high level of effort is approximately 70%. This is reflected with an average of 0.26 (SD of 0.437) and 0.28 (SD of 0.45) respectively. Compared to the dataset of shareholder proposal, this dataset is slightly less unbalanced but still does not display perfect balance.
Examining the Spearman’s correlation table (see table 2 below) shows us that the relation between firm response and the degree of pressure of social media petition is significant at the 0.01 level with a medium effect of 0.310. Thus, as the degree of pressure increases, we can expect firm responsiveness to increase too which suggests the relationship as was hypothesized exists. Another significant, medium effect of 0.3 occurs between the perceived social responsibility and the firm reputation, reflecting that an increase in perceived social responsibility is positively associated with an increase in firm reputation. Other small but significant effects can be found between the level of effort of the type of request and firm reputation, between level of effort of the type of request and perceived social responsibility and between firm reputation and firm response. Interestingly, a small but significant negative effect is found between perceived social responsibility and firm response, suggesting that as the degree to which a firm is seen as socially responsible decreases, the firm responsiveness will also decrease. In comparison to the dataset of shareholder proposal, this dataset reflects more significant relationships between the variables.
6.2.1 Ordinal regression shareholder proposal dataset
Before running the ordinal regression, the assumptions of the model were checked. The first two assumptions mentioned above are met. Testing for multicollinearity reflected no
constraints in the data. As all the VIF values of the variables are around 1 respectively, there appears to be no multicollinearity present and thus the third assumption is met. The
assumption of proportional odds is tested by performing a test of parallel lines. This generates a non-significant effect (p=0.292) so we did not violate this assumption. Having met all the assumptions, an ordinal regression can be run.
Ordinal regression with interaction effect was performed in SPSS to investigate the ability of the degree of pressure of a shareholder proposal and firm reputation to predict the firm responsiveness, controlling for perceived social responsibility and effort level of the type of request. The output of SPSS shows a warning that there are 18 cells with zero frequencies, yet this is only a small percentage of the total sample size of 373 cases. The case processing summary confirms that the firm response is rather unbalanced in this dataset, as already reviewed above. On the other hand, the model fitting information confirms that the models fitted well with the data (p=0.023). The goodness-of fit confirms that the data is consistent with the fitted model, as both the Pearson and Deviance test reflect statistical values greater than 0.05 (p=0.276 and p=0.121). The Nagelkerke Pseudo R-Square, however, displays the Table 2: Means, standard deviations, correlations of social media petitions dataset
Variables M SD 1 2 3 4 5
1. Firm response 1.23 0.606 .
2. Degree of pressure social media petition 2.00 0.823 .310
** . 3. Firm reputation 1.97 0.802 .093 * 0.056 .
4. Perceived social responsibility 0.26 0.437 -.080
*
-0.047 .300** .
5. Level of effort of request type 0.28 0.45 0.034 0.071 .127
**
.100** . **. Correlation is significant at the 0.01 level (2-tailed)
value of 6.2%, which is a fairly low percentage. Thus, only 6.2% of the variance of firm responsiveness is explained by degree of pressure of shareholder proposals, firm reputation, perceived social responsibility and effort level of the type of request, which does not really confirm the value of our model.
Table 3 below displays the result of the ordinal regression to provide more information about the hypotheses. Unfortunately, the different levels of degree of pressure of shareholder proposals do not statistically differ from each other (p=0.665 and p=0. 573). This provides no evidence that as the degree of pressure increases, the responsiveness of the firm increases. There is thus not enough statistical evidence to support H1. The lowest level of firm
reputation is statistically significant to the highest level with p=0.028. A medium level of firm reputation, on the other hand, is not statistically different to the highest level (p=0.639). Taking the exponent of the significant low level of firm reputation results in exp(1.171) = -3.225. The odds ratio (OR) thus suggests that the odds of a low firm reputation entering a higher level of firm responsiveness than a higher firm reputation is 3.2 times more likely. Looking at the moderation effect of degree of pressure * firm reputation reflects no significant result for any of the levels. Based on that result we can infer that any levels of firm reputation do not appear to have any effect on the relation between degree of pressure of shareholder proposals and degree of responsiveness, not supporting H3a. Both the control variables show non-significant results, contrary to previous research (McDonnell, 2015). Perceived social responsibility does not seem to influence the relation between degree of pressure and firm responsiveness (p=0.370). The same, non-existing effect, can be noted for the effort level of type of request (p=0.806).
Table 3: Ordinal regression model shareholder proposal dataset*
Variable Description Parameter
Estimate Standard Error Wald Chi-Square Sig.
[FirmRespCat = 1] Firm response none -1.496 0.541 7.656 0.006
[FirmRespCat = 2] Firm response symbolic 0.763 0.536 2.022 0.155
[DegPresCat=1] Low degree of pressure 0.211 0.487 0.188 0.665
[DegPresCat=2] Medium degree of pressure -0.232 0.412 0.317 0.573
[DegPresCat=3] High degree of pressure 0a
[FirmRepCat=1] Low firm reputation -1.171 0.532 4.843 0.028
[FirmRepCat=2] Medium firm reputation -0.219 0.467 0.220 0.639
[FirmRepCat=3] High firm reputation 0a
[PerceivedSocialResponsibility=0] PSR not present
-0.319 0.356 0.803 0.370
[PerceivedSocialResponsibility=1] PSR present
0a
[LowEffTypReq=0] High effort level type of request -0.064 0.260 0.060 0.806
[LowEffTypReq=1] Low effort level type of request 0
a
[DegPresCat=1]*[FirmRep=1] Moderation effect
DegPres1*FirmRep1
0.410 0.691 0.352 0.553
[DegPresCat=1]*[FirmRep=2] Moderation effect
DegPres1*FirmRep2
-0.372 0.659 0.319 0.572
[DegPresCat=1]*[FirmRep=3] Moderation effect
DegPres1*FirmRep3
0a
[DegPresCat=2]*[FirmRep=1] Moderation effect
DegPres2*FirmRep1
0.313 0.633 0.244 0.621
[DegPresCat=2]*[FirmRep=2] Moderation effect
DegPres2*FirmRep2
0.694 0.601 1.334 0.248
[DegPresCat=2]*[FirmRep=3] Moderation effect
DegPres2*FirmRep3
0a
a. This parameter is set to zero because it is redundant. * Displayed as per Pérez and Pla-Barber, 2005
6.2.2 Ordinal regression social media petitions dataset
Before running the ordinal regression, the assumptions of the model were again checked. Similar to the previous dataset, the first two assumptions are met. Checking for
multicollinearity resulted in VIF values between 1 and 5 respectively, indicating that there are no problems with multicollinearity. Last, the test of parallel lines generated a non-significant effect (p=0.636) so we can assume the presence of proportional odds.
ability of the degree of pressure of social media petition to predict the firm responsiveness, moderated by firm reputation and controlled for perceived social responsibility and effort level of the type of request. Running the ordinal regression resulted in 40 cells with zero frequencies. However, this is only a small number in comparison to the actual sample size of 698 cases. Also, both the model fitting information and goodness-of fit confirm the fit and consistency of the model and the data with significant values on all tests (model fitting information: p<0.000, Pearson test: p=0.112, Deviance test: p=0.324). Reviewing
Nagelkerke’s Pseudo R. Square shows that around 19.9% of variance is explained by the explanatory variables, which confirms the value of our model.
Table 4 below reflects the outcome of the ordinal regression with interaction effect. The model shows that both the first and second level of degree of pressure for social media petitions are significant with values of p=0.000 and p=0.002. This indicates that the two levels significantly differ from the highest level of pressure. Thus different levels of degree of pressure generate a statistically significant response from the firm, supporting H2. Parameter estimates can provide more information about the explanatory variables individually in this context. Taking the exponent from the parameter estimate of the first degree of pressure, so 2.340) results in 0.096. The exponential of the medium degree of pressure is exp(-1.656), thus 0.190. So, the OR of the estimates reflects that the odd that a low degree of pressure results in a higher level of firm response is approximately 90% less likely than a higher degree of pressure. For a medium degree of pressure, the odd that it results in a higher firm responsiveness is approximately 80% less likely compared to the reference group of high degree of pressure. For firm reputation, only the lowest level is significantly different from the highest level (p=0.009), whereas a medium level of reputation is not significantly different (p=0.785). Taking the exponent of this significant variable, exp(-0.954) results in the OR of 0.385. Therefore, the odd that a low level of firm reputation generates a high level of firm
responsiveness is 61.5% less likely than a higher level of firm reputation. Looking at the moderating effect of firm reputation on the relationship, table 4 shows that none of the levels have a significant value. This result therefore provides no support for H3b, as an increase in the level of firm reputation appears to have no increasingly negative effect on the degree of pressure of social media petitions – firm response relationship. Perceived social responsibility also reflects a significant result (p=0.0450), which complements previous findings
(McDonnell, 2015). Thus, the odds that perceived social responsibility is likely to generate a higher degree of firm responsiveness is 1.9 times higher than no perceived social
responsibility, as the exponent of the estimate, exp(0.656) is 1.927. Last, a low effort level of the type of request appears to have no statistical difference with a high level of effort with a p-value of 0.920.
Table 4: Ordinal regression model social media petition dataset*
Variable Description Parameter
Estimate Standard Error Wald Chi-Square Sig.
[FirmRespCat = 1] Firm response none 1.165 0.362 10.336 0.001
[FirmRespCat = 2] Firm response symbolic 1.540 0.367 17.638 0.000
[DegPresCat=1] Low degree of pressure -2.340 0.644 13.219 0.000
[DegPresCat=2] Medium degree of pressure -1.656 0.530 9.765 0.002
[DegPresCat=3] High degree of pressure 0a
[FirmRepCat=1] Low firm reputation -0.954 0.416 5.268 0.022
[FirmRepCat=2] Medium firm reputation -0.156 0.343 0.206 0.650
[FirmRepCat=3] High firm reputation 0a
[PerceivedSocialResponsibility=0] PSR not present 0.656 0.322 4.148 0.042
[PerceivedSocialResponsibility=1] PSR present 0
a
[LowEffTypReq=0] High effort level type of request -0.039 0.264 0.022 0.883
[LowEffTypReq=1] Low effort level type of request 0a
[DegPresCat=1]*[FirmRep=1] Moderation effect DegPres1*FirmRep1 -0.692 1.239 0.312 0.576
[DegPresCat=1]*[FirmRep=2] Moderation effect DegPres1*FirmRep2 -1.010 1.221 0.684 0.408
[DegPresCat=1]*[FirmRep=3] Moderation effect DegPres1*FirmRep3 0a
[DegPresCat=2]*[FirmRep=1] Moderation effect DegPres2*FirmRep1 0.471 0.807 0.340 0.560
[DegPresCat=2]*[FirmRep=2] Moderation effect DegPres2*FirmRep2 0.570 0.651 0.765 0.382
[DegPresCat=2]*[FirmRep=3] Moderation effect DegPres2*FirmRep3 0a
a. This parameter is set to zero because it is redundant. * Displayed as per Pérez and Pla-Barber, 2005
