The influence of voting systems on satisfaction with the outcome and attitude

(1)

5-2-2019

Co-creation

contests

The influence of type of voting

system on satisfaction with the

outcome the vote and attitude

towards a company

Niek Stortelder

Studentnumber: 4348354

Supervisor: Dr. P. H. Driessen

Second examiner: Dr. H.W.M. Jooste

Master thesis Business Administration,

Marketing

(2)

1

Co-creation contests

The influence of type of voting system on satisfaction with the

outcome the vote and attitude towards a company

Niek Stortelder (S4348354 Athlonestraat 4, 6524 BH Nijmegen +31627572668 n.stortelder@student.ru.nl Business Administration Master’s thesis in Marketing

Dr. P. H. Driessen

Dr. H.W.M. Joosten 5 February 2019

(3)

2

Preface

This thesis was written as final part of my Master Business Administration, specializing in Marketing at the Radboud University Nijmegen. A year from the start this thesis is finally finished. It took many hours of studying literature, setting up experiments, collecting and analyzing data, interpreting results, writing conclusions and finally tons of re-writing. I know proudly present you my thesis with the title: Co-creation contests: the influence of type of voting system on satisfaction with the outcome of a vote and attitude towards a company. During this process I got to sharpen my research skills, learned more about marketing, but also learned more about myself.

This thesis would not have been possible without great support from a number of people, which I would like to especially thank. First of all I would like to thank Dr. P.H. Driessen for his supervision of this thesis. Without his support, feedback and guidance this thesis might have never seen the light of day. Next I would like to thank my second examiner, Dr. H.W.M. Joosten, for the time he took reading and examining this thesis. Furthermore, I would also like to thank all the participants that took time out of their day to partake in this experiment. Without them this thesis could not have been finished. Lastly, I would like to thank my parents and brother for their unconditional support during the process of writing this thesis.

I hope you enjoy reading my master thesis.

Niek Stortelder

(4)

3

Abstract

Co-creation is common practice these days. In order to be able to keep creating value for their consumers, companies are more and more engaging in co-creation with their consumers. They try to do this in numerous ways, one of which is by organizing co-creation contests. With these contests they provide their consumers with the power to vote for what new product they would like to see produced.

This research tries to examine in what way such a contest should best be designed in order to achieve a more positive attitude towards the company that organized the contest after the results of the vote. This was done by comparing a plurality voting system to a preferential voting system and how they influence voter’s satisfaction with the outcome of the vote. Next to that, the influence on preference match on satisfaction with the outcome of the vote was researched, as well as the influence of satisfaction with the outcome of the vote on attitude towards the company that organized the contest.

If was expected that voters would show higher satisfaction with outcome of the vote when they voted by means of a preferential voting system and experienced a preference match. Also, when voter’s satisfaction with the outcome of the vote wash high, it was expected that voters would also have a more positive attitude towards that company.

Results, however, showed that voting by means of a plurality voting system induces higher feeling of satisfaction with the outcome of the vote compared to a preferential voting system. Furthermore, when the voter experienced a preference match, that voter also experienced higher satisfaction with outcome of the vote. Lastly, when voters where satisfied with the outcome of the vote, they also had a more positive attitude towards the company that organized that contest.

(5)

4

1. Introduction

1.1 Background

In recent years it has become increasingly important for companies to include consumers in new product development (NPD)(Prahalad and Ramaswamy, 2004; Hoyer et al., 2010). Since the rise of Web 2.0, allowing social media to thrive, more options have become available for companies to reach out to its consumers more easily (Constantinides and Fountain, 2008). Companies are now able to create and sustain online platforms on which customers can share their ideas for new products. They are able to listen and integrate these ideas into their NPD processes (Fuchs et al., 2011). This method of including consumers in NPD processes is gaining popularity amongst companies (Füller et al. 2009). In order to include consumers in the NPD process, companies often make use of contests, consisting of voting systems, to provide consumers with tools to influence what product is to be produced (Fuchs et al, 2011).

These kind of co-creation contests are however mostly used by companies that produce products, compared to service providing companies, and that want to provide their consumers with influence over what product they want to see on the market. Companies like Lays, BIC and M&M all have used its consumers to decide what new products they are going to produce. Lays for instance allowed its consumers to come up with flavors of chips that they would like to see produced, selecting three of the best options and then let their consumers vote for their favorite flavor. BIC called for its consumers to create new designs for their newest lighter collection and come up with a slogan for the campaign, the winner would receive 2000 dollars and see the design being produced (Djelassi and Decoopman, 2013). These contests usually make use of a voting system.

Companies need to think carefully about how they want to design their contest in order to receive an outcome that is pleasing to the company, while also satisfying the consumers that took part in the contest. As every contest knows winners and losers, it might seem difficult to have an outcome that pleases all consumers.

What kind of voting system to use in a contest is however interesting, as with contests there are always winners and losers. Making use of a voting system that might provide the losers with a higher degree of satisfaction, could maybe provide the company with more favorable intentions towards its contests outcome and perhaps ultimately towards its company as a whole. Although these contests have been gaining popularity, what kind of voting systems they make use of has been mostly overlooked in the field of marketing. This is a shortcoming in the literature, as these voting systems are a common presence in today’s marketing landscape.

(9)

8 Companies use them to enable their consumers to engage in co-creation and have an influence on the products that are to be produced. Consumers want to have an influence on the NPD of companies, companies can no longer create valuable products without including their consumers (Prahalad & Ramaswamy, 2004). In order to keep extracting economic value out of consumers, companies need to co-create. One could argue that companies keep their consumers satisfied by allowing them to engage in co-creation.

The influences of the types of voting system that is used in a contest or election have been studied extensively in political science (Farrel & McAllister, 2006; Blais & Gélineau, 2007; Grönlund & Setälä, 2007). Many different types of voting systems are used in political systems. However, plurality and preferential voting systems are most widely used (Reynolds and Reilly, 1997). Plurality voting systems are systems in which the voter can only select one option of their choice. Preferential voting systems come in different ways but are mostly used as ranked systems. These systems allow the voter to select multiple options and rank order them in their way of preference. (Reynolds and Reilly, 1997)

This thesis aims to compare these two types of voting systems and their influence on consumer satisfaction with a company. The goal is to research whether consumer satisfaction with a company increases, after taking part in a voting system on which product is to be launched, depending on what type of voting system is used. Two types of voting will be compared: plurality voting and preferential voting. Meaning that one voting system only provides the option to select one product, whereas the other voting system asks consumers to list the products in preferred order. Afterwards satisfaction with the company will be measured, as well as the outcome of the vote, and compared between the two voting systems. Participants will take part in an experiment in which they are presented with a fictional company and fictional products.

1.2 Research question

This research aims to contribute to the insights in the most favorable way of how to design a voting contest in order to achieve a higher level of positive attitude of the consumer towards the company that organizes the voting contest. This leads to the following research question: What influence does the type of voting system and outcome of the vote in NPD co-creation contests have on consumer satisfaction with the outcome of the vote and overall consumer attitude towards the company that organizes the contest?

(10)

9 1.3 Theoretical contribution

That companies nowadays need to co-create with their consumers has been established by research (Prahalad and Ramaswamy, 2004). Fuchs et al. (2011) for instance have researched what type of NPD consumer co-creation results in the highest favorable corporate associations amongst consumers that do not take part in the co-creation efforts. They found that when the company provides full empowerment to consumers it results in more positive and in stronger behavioral intentions attitudes towards the company (Fuchs et al, 2011).

However, they have not researched how that contest should be designed in order to achieve the highest possible favorable associations. Marketing research has not yet concerned itself with what type of voting system can best be used in such co-creation contests. What influence a type of voting system can have is however extensively researched in political research. In politics a multitude of voting systems are used and it was found that different systems can have a differing influence on the satisfaction with an outcome, preferential voting systems tend to induce a higher satisfaction with the outcome of the vote (Farrel and McAllister, 2006). This can also be interesting in the field of marketing research as these types of voting systems are also more and more used in company co-creation contests, providing insight in how these voting systems influence consumers can add to the overall co-creation literature. Even for political science it can prove interesting, adding insights in not only the influence on the outcome of a vote but also on overall attitude towards the political system.

This thesis tries to add to this literature by researching what type of voting system is most likely to induce higher feelings of satisfaction with the outcome of the vote. Next to that, this thesis also tries to provide insight into how satisfaction with outcome of a vote carries over to attitude towards a company.

1.4 Practical contribution

There is also a practical component to the knowledge this research will try to provide. Companies are aware that they need to co-create, given the amount of co-creation contests that have been issued. Contests always have winners and losers, winners might have more positive attitudes towards a company and losers might have less positive attitudes towards a company. Organizing a co-creation contest can prove to be quite a risk for companies as the outcome of the contest cannot be influenced by the company. How many of their consumers will identify as a winner or a loser is uncertain when organizing such a contest. Having the ability to mitigate this risk when organizing such a contest might prove useful to companies and could make them

(11)

10 more likely to engage more easily in such types of co-creation. This research will try to provide insights in what type of voting system can best be used in such contests to mitigate this risk. By using a voting system that induces the feeling to consumers that they partially won, by their vote coming in second place for instance, feelings of dissatisfaction towards the company might be less.

Knowledge on how to best design such a contest in order to make sure that the co-creation efforts have a positive influence on most consumers is beneficial to researchers as well as marketers. To marketing managers this research is useful when they have to make decisions about how to apply consumer co-creation contests in a way that is most beneficial to their company.

1.5 Structure

This research will be structured in the following manner: First an extensive literature review will be conducted, in which the conceptual model and all its components will be explained and defined. Next, the method by which the research will be executed will be discussed. Following, the results will be presented and will afterwards be discussed. Finally, a conclusion will be presented, along with future research possibilities and managerial implications.

2. Conceptual framework

2.1 Conceptual model

The conceptual model shown in the figure below will be used to guide this research. The conceptual model consists of four parts: the type of voting system, preference match, satisfaction with outcome of the voting system and attitude towards the company. In this model it is expected that the type of voting system used in a contest organized by a company has an influence on consumers satisfaction with the outcome of that vote. It is hypothesized that a plurality voting system has neither a positive nor a negative effect on satisfaction with outcome of the vote. On the other hand, it is expected that a preferential voting system does have a positive outcome on satisfaction with the outcome of the vote.

Whether or not the consumer experiences a preference match with the outcome of the vote can also have an influence on their satisfaction with the outcome of that vote. It seems

(12)

11 probable that when consumers experience a preference match with the outcome, they are more satisfied with the outcome.

Lastly it is expected that the satisfaction with the outcome of the vote has a positive influence on overall satisfaction with the company. These expectations are further elaborated upon in this chapter.

Figure 1. Conceptual model

2.2 Voting systems

2.2.1 plurality voting systems

Voting systems are widely used in political systems. Many different types of voting systems exist. One of the most commonly used systems is that of plurality voting (Bormann & Golder, 2013). This system, which is sometimes called the “first past the post” method, is the simplest voting system. Within this system every voter is enabled to cast one vote. The winner of the vote is the candidate that receives the highest percentage of votes (Blais & Massicotte, 1996). Although this system is most widely used, it has its shortcomings. Arguably the largest shortcoming is that plurality voting allows for the possibility of a very small majority to win the vote (Reilly, 2004). One example of this is the 1988 prime minister election in Canada. The main point of this election was free trade with the United states, to which 60% of Canadians were against. Three candidates ran in the election, two of them being opposed to free trade and the third candidate being pro free trade. Contrary to what might be expected, the pro free trade candidate won the election, because voters that were against free trade split their votes between the two anti-free trade candidates (Van Newenhizen, 1990).

It is, amongst lesser shortcomings, for this reason that a multitude of voting systems have been designed, most often in the design of preferential voting systems.

(13)

12 2.2.2 Preferential voting systems

Preferential voting systems are different from the plurality voting system, as voters are able to choose from multiple options and put them in order of their preference. There is a multitude of voting systems that are accumulated under the term of preferential voting, but it is mostly defined as “electoral systems in which voters can rank-order candidates on the ballot paper in order of their choice” (Reynolds and Reilly, 1997). Within this type of preferential voting system, different terms are used to describe preferential voting systems. Alternative voting, single-transferable voting and ranking methods are all names given to preferential voting systems and most only slightly differ from each other (Toplak, 2017). Table 1 contains an overview of different types of preferential voting systems. Some examples of how these voting systems are put into practice are presented in appendix A.

(14)

13

Name of system Way of voting Pros and Cons

Single transferable vote (STV) and Instant runoff voting (IRV)

Voters order their vote by most preferred candidates. If a candidate reaches a first-choice majority, he is elected. If not, the candidate with lowest first-choice votes is eliminated. His votes, along with extra votes from elected candidates, are reallocated to the next candidate on the preference list of the voter. This continues until a candidate reaches a majority or all positions are filled. (Farrel & McAllister, 2004)

Pros:

Voters that vote for a smaller party can still influence results with their other preference votes.

Cons:

Complex side effects might occur (Van Erp, 2002) Allows for strategic voting, bigger parties will want the preference votes of smaller parties.

(Sharman, Sayers & Miragliotta, 2002)

Single

nontransferable vote (SNTV)

In this system voters vote for a candidate within a party. It is very similar to list proportional representation. Each party presents a list of candidates that voters can vote for. The candidates that are elected to fill the seats is based on the total number of votes each candidate received. (Bartholdi & Orlin, 1991)

Pros:

Small parties have an advantage over larger parties. (Cox & Niou,

1994)

Cons:

Provides problems for large parties, as they might run too many or too few candidates in a district.

(Cox & Niou, 1994)

Borda-count Voters rank all the optionable candidates from most preferred to least preferred. Every position is awarded points. The mean rank of each candidate is then computed and the classes are reranked. This produces a rank order of which the winner has the highest mean over all voters. (Van Erp, Vuurpijl and Schomaker

, 2002)

Pros:

Not susceptible to manipulation

Most likely to pass the Condorcet principle (Saari, 1990a)

Cons:

Voters might put their preferred candidates largest rival in last place in order to influence results. When not forced to list every candidate, voters might not list certain candidates to not allow them points (Reilly, 2002)

Alternative vote Voters rank their choices by putting a number in front of the candidate. First preference gets a “1”, second a “2” and so on. If a candidate has a majority in number 1 votes, he is automatically elected. If not, candidate with lowest number of preferences is eliminated. His votes are then transferred to the number 2 from that voter. This

Pros:

Supporters of smaller parties can still have an influence over the winning party.

Cons:

Smaller candidates can be discriminated against. Allows for strategic voting, bigger parties will want the preference votes of smaller parties.

(15)

14 Table 1. Overview of preferential voting systems

As can be seen in the table, all types of preferential voting systems have their pros and cons. Of these methods the Borda-count voting system is the oldest and easiest to implement (Reilly, 2002). As opposed to the other systems it only contains one stage and does not deal with eliminating and re-assigning votes. This might be beneficial to the comprehension of this system amongst voters. Connecting this to the organizing of contests, it might seem like the most favorable preferential voting method to use. For companies it is not complicated to implement and to explain to their customers how the voting system works, which for instance would be harder if a single transferable vote system was to be used.

What is however also important to consider, is if the election method used is deemed as being fair. One method to decide if a voting method is a fair method is that of the Condorcet principle. The idea of the Condorcet principle is one of fairness. It entails the principle that: “the method must select from more than two competitors the one supported, as against any of the other competitors taken singly, by more than half of the voters who have expressed a preference between the two, if such a measure or candidate there be” (Hoag and Hallet, 1926). To explain this, it means that in a Condorcet voting system, the winner of the vote does not only need to have the most votes, but should also win from every other candidate when they are solely pitted against one another. For example: in a Condorcet election with candidates A, B,C and D, candidate A has won the election. To check if the Condorcet principle has been met, every candidate will be pitted against the others. If A is to be deemed winner, it should become clear that more voters have picked candidate A over B, picked A over C and have picked A over D. If this is the case, the Condorcet principle has been met and the election is to be deemed fair.

continues until a candidate has a majority. This term is mostly used in Australia. (Reilly, 2004)

(Sharman, Sayers & Miragliotta, 2002)

(Closed) List proportional representation (PR)

European version of preferential voting. Voters vote for their favorite candidate, automatically voting for the party that candidate is in. In this system candidates of a party that received the most votes within that party will be elected. They can fill the seats that were assigned to that party, determined by the total number of votes the party received. (Toplak, 2017)

Pros:

No wasted votes.

Allows for a simple preferential vote. (Marsh, 1985)

Cons:

Favors the candidate at the top of a parties list. A candidate needs many favor votes to move up on the parties list.

(16)

15 The preferential voting system that is most likely to pass the Condorcet principle is the Borda-count voting system (Van Newenhizen, 1992). The Borda-count does not always satisfy the Condorcet principle, but is the method that is most likely to pass the principle.

This method provides voters with the option to rank their preferences, awarding points to each rank. The means of each option are then calculated and subsequently re-ranked in order of their highest means (Van Erp, Vuurpijl and Schonmaker, 2002).

This means that, with the Borda-count system, when results are conducted, the candidate with the most top-rank votes does not necessarily win the vote. This might seem unfair at first glance, however it allows for the majority of voters to see their choices reflected in the outcome. The Borda-count allows for finding the middle way when voters are divided over a number of options. For instance, half the voters put candidate A as their most favorable pick and Candidate C as their least favorite pick. The other voters prefer Candidate C and see Candidate A as least favorable. However, all voters agree that Candidate B is their second pick. In this case Candidate B would be the winning candidate, allowing for most voters to be presumably satisfied. In a plurality voting system half the voters would be presumably disappointed, as the winning option was only their third pick. This might also be applicable for co-creation contests designed by companies, this voting system seems likely to be able to decrease the gap between winners and losers.

It is because of this reasoning (least complicated, easy to implement, least difficult to comprehend and the Borda-count being most likely to pass the Condorcet principle), that the Borda-count will be used in this research as a means of a preferential voting system. For the remainder of this research the Borda-count will be referred to as “preferential voting system”.

2.3 Satisfaction and voting systems

The type of voting system used might have an influence on the satisfaction with the outcome of the vote, therefore with the overall satisfaction of the company, according to the conceptual model. In the literature two possible explanations for this relationship were found.

2.3.1 Voting systems in political science

The first explanation is based in political science. In political research it was found that preferential voting can make a difference in voters’ satisfaction with democracy. Voters who experienced a higher degree of fairness about the election’s outcomes, had an increased level

(17)

16 of satisfaction with their democracy (Farrel & McAllister, 2006). Other research found that winning or losing a vote has a significant effect on satisfaction with democracy, even when satisfaction with democracy beforehand was controlled for (Blais and Gélineau, 2007).

It was also found in research that voters who ended up voting for the minority are more likely to be less satisfied with their democracy compared to those voters who voted for the majority. Losers that voted for the minority, but have a voting system that is more consensual tend to have higher satisfaction, compared to losers in a system with a more majoritarian outcome. (Anderson and Guillory, 1997)

Further suggested is that winners tend to experience higher satisfaction compared to losers, the reason for this being that the outcome of the vote has their support. This is the other way around for losers (Anderson & Guillory, 1997). This negative effect might be stronger in plurality voting systems compared to systems that also allow for the candidate in second place to be elected (Henderson, 2008).

It was also suggested in research that preferential voting “reduces the rewards for winners and reduces punishment for losers.” The system was thought of as “accounting for limited rewards of winners and absence of dissatisfaction among losers” (Henderson, 2008). This implies that a preferential voting system does not induce feelings of dissatisfaction when a voter’s choice does not win. Also, winners did not experience as much of a winning feeling after their choice won the vote (Henderson, 2008).

These findings suggest that an outcome of the vote that reflects better the way the votes were distributed, induces a higher level of overall satisfaction with the outcome of that vote. This might also apply to this research; in a preferential system the chances that the outcome reflects the opinions of the voters is higher compared to that of a plurality voting system.

Other research found that people who experienced a higher degree of controllability over the outcome of something they achieved, had more satisfactory feelings towards that outcome, compared to when they felt like they had less control over the outcome. (Forsyth & McMillan, 1981) This might apply to providing voters with more control over their vote, by having them rank order their preferences, instead of only picking one option to vote for. This might then suggest that preferential voting systems can induce a higher degree of satisfaction with the outcome, because of voters having more control over how they voted.

This theory might also be applicable to company contests. When using a preferential voting system in their contests, the literature suggests that voters experience less of a winning attitude, but also less of a losing attitude. Overall this might result in higher satisfaction with

(18)

17 the outcome of the vote. This might also be the case when voters see their preferred product win, they experience not as high a winner mentality. Losers, who do not see their preferred product win, however experience less of a loser mentality. Because of this absence of a gap between winner and loser mentality, overall satisfaction with the voting system might result to be higher. Combined with the higher feeling of satisfaction when control over the outcome seems higher, preferential voting systems might induce a higher degree of satisfaction with the outcome compared to plurality voting systems that do not offer these benefits.

2.3.2 Voting systems and cognitive dissonance

The second explanation for the relationship between voter satisfaction with the outcome of a vote that was found in the literature is that of cognitive dissonance. Cognitive dissonance is defined as: “a negative drive state which occurs whenever an individual simultaneously holds two cognitions (ideas, beliefs, opinions) which are psychologically inconsistent” (Aronson, 1969). This theory suggests that when people do not see their beliefs or ideals met, they try to reduce this dissonance. This increases as the degree of dissonance increases, also the strength of these beliefs has an influence (Festinger, 1957). It means that when people do not see their beliefs met, they try to convince themselves that they were still right in some way or another. In an example: when someone has voted for option A and beliefs option A won, but the outcome shows that option B won and A is second, they will try to convince themselves that they actually wanted to vote for option B. Thus reducing the dissonance between the two beliefs and creating more consonance. This theory might be of influence in the relationship between voting system and satisfaction with the outcome of the vote. Voters might experience a degree of cognitive dissonance when they do not see their vote win. To try and reduce this degree of cognitive dissonance voters might look to the options that did not win. In a preferential voting system voters might look to their second preferred winner, if that option did indeed come in at second place, the voter’s cognitive dissonance might be reduced and their satisfaction with the outcome increased. In a plurality vote this theory might not be of such influence as the voter did not vote for more than one option, so there is not second preferred winner to look at.

These explanations provide a basis for the assumption that outcome of a voting system has an influence on the voter’s satisfaction of the subject. To determine if satisfaction of outcome with a voting system differs between type of voting systems used, two types of voting systems will

(19)

18 be compared in this study. First a plurality voting system will be examined and second a preferential voting system, in the form of the Borda-count, will be examined

This leads to the following hypothesis:

H1: Use of a preferential voting system in an NPD-contest will induce higher feelings of satisfaction with the outcome of the vote, compared to a plurality voting system.

2.3.3 Additional concepts

In order to be able to better understand the possible underlying relationship between the type of voting system and satisfaction with outcome of the vote, two more concepts will be included as variables in this research. The two concepts that will be included are “perceived fairness of the voting system” and “consumer comprehension of the voting system”. These concepts will be referred to as “fairness” and “comprehension” for the remainder of this thesis. If there proves to be a relationship between type of voting system used in a contest and the satisfaction with the outcome of that contest, it might be insightful to be able to analyze what characteristics of that particular voting system are the cause of this relationship. For this research the variable fairness will be defined as: the degree to which a consumer believes that the voting system is able to determine the winner of a contest that reflects the votes of consumers best. Comprehension will be defined as: the ability of a consumer to understand how the voting system works and how a winner of a contest is calculated.

Because these concepts are only added to provide further insights where possible, they are not included in the conceptual model.

2.4 Preference match

Voters might also experience a higher feeling of satisfaction because of the idea of preference match. It is assumable that consumers might be more satisfied when they see the outcome of the vote is in accordance with their own voted preferences. It might therefor be the case that consumers are not so much more satisfied with the voting system per se, as that they are satisfied with their preference match. This also comes with the risk that in a preferential voting system consumers get more insight in their preference match compared to a plurality system. As in a preferential voting system the voter puts all options in his most favorable order, however the outcome might be very different from their own preferences. The consumer has thought about their preference about every option and might therefor feel more strongly about the eventual position of each option in the outcome. While in a plurality system consumers only have a

(20)

19 connection to one option, as they did not have to put every option in order of their preference. Preference match might have a less strong effect in this type voting of system.

On the other side however, consumers that see their preferences reflected in the outcome might feel more satisfied with this outcome for the same reason. They have stronger feelings for every option and when they see their preferences shared by the majority they might feel empowered and more satisfied.

These assumptions are backed by literature. It was found that satisfaction is influenced by ideological proximity, how far the standpoints of the winning party are removed from the voter’s political beliefs. It was found that when a voter did not see his preferred candidate win, but the voter’s political beliefs were relatively close to the winner, the voter was relatively more satisfied (Henderson, 2008). This finding can be extrapolated to this research in the form of preference match. When a voter in a preferential system does not see his preferred product win, but does see that it came in second, the voter might be more satisfied with the outcome. In a plurality system this does not apply, as the voter will not have thought about his second preferred product. It might therefor be the case that preferential voting systems have a higher chance of inducing higher feelings of satisfaction with the outcome of the vote.

Another theory that might explain why preference match has an influence on satisfaction with outcome of the vote is that of belongingness and cognitive dissonance theory.

Belongingness suggests that it is “the act of self-identification or identification by others, in a stable, contested or transient way (Yuval-Davis, 2006). The reason that people want to affiliate with one another is that humans have a drive to have positive, affective and stable interpersonal bonds (Yuval-Davis, 2006). Baumeister and Leary (1995) refer to this as “the need to belong”. They claim that people need to feel that they belong to a certain group in order to have their need to belong satisfied. This theory has claimed to apply under many different circumstances and widely ranging conditions. When the need to belong is not satisfied, people may experience negative effects, like anxiety, distress and even health problems (Van Prooijen & Van den Bos, 2004). This translates to voting systems as people that experience a preference match might feel that they belong more. When their preferred products are in the winning group, they might feel that they belong more, inducing feelings of satisfaction. When they do not experience a preference match they might feel that they do not belong, possibly experiencing the negative effects mentioned above. This would possibly decrease their satisfaction with the outcome of the vote.

(21)

20 Next to the theory of belongingness, the theory of cognitive dissonance might also have an influence on preference match. This theory was already explained above, regarding satisfaction with outcome of the vote. This might also be applicable to voting contests and preference match. When voters see that their option is the winner they will not experience cognitive dissonance, possibly resulting in higher satisfaction with the outcome. When they do not experience a preference match, cognitive dissonance might occur, possibly resulting in lower satisfaction with the outcome of the vote. Taken together, these theories lead to the following hypothesis:

H2: voters that experience a preference match will have a higher degree of satisfaction with the outcome of the vote, compared to voters that do not experience a preference match.

2.5 Attitude towards the company

The final concept in the conceptual model is attitude towards the company. The conceptual model expects that the voter’s satisfaction of the vote will have a positive influence on the voter’s attitude towards the company that organized the contest. Before this relationship can be explained, the concept of attitude towards a company has to be defined.

The term attitude is rather loosely but also commonly used in literature. The fundamental assumption regarding attitude however, is commonly accepted as “attitudes in some way, guide, influence, direct, shape or predict actual behavior (Kraus, 1995).” This is interesting as it seems to insinuate that a certain attitude might lead a person to do a certain thing. With regard to a company, this could be that a person’s positive attitude towards a company might urge them to show behavior that is in favor of that company.

On the matter of a definition of attitude, in most marketing related studies the concept of attitude has been used as a predictor of the consumers beliefs about a brand (Mitchell & Olson, 1981). The dictionary of Cambridge defines attitude as: the way you feel about something or someone, or a particular feeling or opinion. In this research attitude will be defines as: The beliefs, or feelings, a consumer has towards something or someone.

A consumer’s attitude towards something is made up of that person’s salient beliefs regarding that something. In the case of this research the consumers salient beliefs about the company that organized the contest. In order to change a consumer’s attitude towards something, that person salient beliefs about that something need to be changed (Fishbein & Ajzen, 1975). This can be done in a number of ways: by changing the strength of a salient belief, changing the evaluation of a belief, or by changing a salient belief in such a manner that it is no

(22)

21 longer salient (Lutz, 1975). In the case of this research the consumers salient beliefs about a company are tried to be positively altered using the consumers satisfaction with the outcome of a vote organized by a company.

The relationship between the two mentioned concepts might not be directly clear and will therefore be clarified in this section. The broad idea is that a consumer who sees his choice in a vote win, will be more satisfied with the outcome of the vote, compared to consumers who do not see their choice win. The consumers who are more satisfied with the outcome of the vote, might as a result also feel more positive towards the company that organized the contest. This type of relationship is supported by the halo effect. The halo effect is a popular theory that is mostly defined as “the influence of a global evaluation on evaluations of individual attributes of a person” (Nisbett & Wilson, 1977). Meaning that one is unable to discriminate between different attributes of a person, crediting his score on one attribute also to other attributes. For instance the idea that a beautiful person will more often be described as a nice person, without this having to be the case.

It was found that halo effects occur between brands and retailers with regard to repurchase intentions (Zboja & Voorhees, 2006). In the case of this research voters transfer the satisfaction with the winning product, to overall satisfaction with that company. For instance if a consumer takes parts in a contest about what new flavor of chips is to be produced and he votes for paprika flavor, if that flavor is also the winning flavor that consumer will be satisfied with that outcome. But, because of the halo effect the consumers will also be more positive about the company that organized that contest.

This expectation is supported by findings of Andreassen and Lindestad (1998). They found that consumers that are satisfied with services that were deployed by a company, have higher positive attitudes towards that company. In this research, the contest organized by the company, can be seen as such a service. Therefor it might be able to create higher degrees of satisfaction with the company as a whole.

In the case of this research consumers might feel more positive towards a company when the outcome of the vote was in line with their preferences. The voters will experience a certain amount of satisfaction with the outcome, which will change their salient beliefs about that company (due to the halo effect) and therefor change their attitude towards that company. It is expected that a higher satisfaction with the outcome of the vote will change the voter’s

(23)

22 salient beliefs about the company in a positive way, so that they have a more positive attitude towards that company. This leads to the third hypotheses:

H3: Satisfaction with the outcome of a voting contest leads to a more positive attitude towards a company.

3. Method

3.1 Method introduction

This research will be conducted by means of an experiment. The main goal of and experimental design is to identify a possible causal relation between independent and dependent variables (Kirk, 1982). This means that an environment will be created in which the researcher can control what happens and under which conditions this happens. Another advantage of an experimental design is that with relatively low effort and costs, a large amount of data can be gathered (Kirk, 1982). This approach suits this research well, as this research aims at comparing the relation of two independent variables (the voting system and preference match) on one dependent variable (satisfaction with outcome of the vote). In a second analysis the relationship between satisfaction with the outcome of the vote and attitude towards the company will be analyzed. This design makes an experiment very suitable as a means of data gathering.

3.2 Design

Because of the relative intricateness of the setup of this experiment, it will first be briefly explained before advancing further into this chapter.

The idea of this experiment is to create four groups, differentiating between the two voting methods and experiencing or not experiencing a preference match. The experiment is divided into two parts: in the first part participants only vote, in the second part the participants are presented the outcome of the vote and are asked multiple questions about the voting procedure, the outcome and about the company.

During the first part of experiment participants are presented four different t-shirts. They can then either vote for one t-shirt or rank order the t-shirts in order of their preference, depending on which one of the four groups the participant got put in. After they have voted, participants are asked for their e-mail address, which is needed to contact them personally to take part in the second part of the experiment.

Three days after taking part in the first part of the experiment participants were sent an e-mail, inviting them to partake in the second part of the experiment. Here they were shown the

(24)

23 outcome of the vote, which was manipulated to suit the group that the participants were put in and are asked a multitude of questions about the voting process.

The design of this research will therefore be a full factorial between-subjects design. This design allows for the comparison between a multitude of groups that are all only subject to one stimulus. Two groups of participants will both take part in the experiment, but each group will be influenced by a different factor. In this case on group will only get to vote using the plurality voting system voting, whereas the other group will get to vote using the preferential voting system. This research design is shown in the table below.

Preference match Non-preference match

Plurality voting Group 1 Group 2

Preferential voting Group 3 Group 4

Table 2. Research design 3.3 Material

First of all the type of contest needs to be determined. Here the choice has to be made between creating a fictional company or choosing an already existing one. Both have their advantages and disadvantages, which will be discussed below.

The problem with choosing an already existing company is that consumers might already have strong attitudes towards the selected company. It is then difficult to measure the degree of satisfaction that is induced due to the voting contest. Their preexisting feelings toward the company might not be altered enough by an experiment to contribute that change to the experiment. The advantage however is that with an existing company most consumers will be familiar with it, it therefor does not need as much of an introduction.

With a fictional company the problem occurs that consumers are not yet familiar with the company at all. It is then of importance to design a fictional company that is believable enough for all kinds of consumers. Otherwise a bias might be created just by how the company is presented. This is something that could, and should, be tested in a pilot experiment.

Having considered both options, this research will make use of a fictional company in its experiment. The problem that consumers preexisting satisfaction with an existing company will not be able to be changed by the experiment is an issue that cannot be ignored. To make sure that this is not possible to happen, a fictional company will be created.

The next problem is what type of company this will be and what type of product will be used in the experiment. It is favorable to select a product that is of low-involvement, as these types of products are most likely to be used for co-creation (Lays chips, Bic lighters, M&M

(25)

24 flavors i.e.). The product that will be featured in this experiment needs to be of a sort that actually benefits from customer involvement. Products like furniture or t-shirts might spring to mind. A company that heavily relies on customer co-creation is a company called Threadless. They enable the consumer to design a variety of clothing types and then organize a vote amongst consumers to decide on the design that is going to be sold (Brabham, 2010). This prime example shows that contests, with t-shirts as products, do work and are a viable option to be used for this experiment.

This experiment will therefor make use of t-shirts as its contest product. Participants will be introduced to a fictional company, that sells a variety of t-shirt designs.

The next step is then to pick several designs that participants will be able to vote for. It is important that these designs are all evenly desirable, to avoid creating a bias to one of the designs. Having such a bias could influence the results of the experiment, rendering them unusable. Also important is to choose designs that are not yet in existence, or at least not widely known. Participants might already have experience with these existing designs and might therefor be biased towards that design. The designs were mostly chosen at random, but it was taken into account that there had to be some variety amongst the t-shirts. Also the choice to only select t-shirts for men was a deliberate one. Men tend to have less of a clear opinion about woman’s clothing compared to the opinion’s women have of men’s clothing (Kwon, 1997). These eight designs will first be used in a pilot experiment, to determine if the designs are evenly liked. The four designs that are ranked closest to each other will then be used in the actual experiment. Only four designs will be used to not overload participants with options and to keep the experiment representing of reality.

What is also important is that participants will see the experiment as an actual contest that could also take place in reality. This is important because if the contest does not resemble a believable experience the results of the experiment will not be representable of reality. To make sure participants consider the experiment to be a possible real contest, a pilot experiment will be done.

The pilot experiment now consists of testing the fictional company, the eight selected designs that will be checked for desirability and the fairness and believability of the contest will be tested.

Next to the setup of the experiment, the stimuli will also have to be manipulated. This study consists of two conditions that need to be manipulated. First the type of voting systems needs to be looked at. Two of the four groups will be presented with a plurality type

(26)

25 voting system. In this system the participant only gets the option to vote for one of the four presented t-shirts. The other two groups will be presented with the preferential voting system. In this system the participants need to rank the four t-shirts according to their preference.

The other stimulus that needs to be manipulated is the preference match. Two groups will be presented with a preference match. This means that the results of the vote will be in line with their own vote. For the plurality system this means that the participants favorite t-shirt will also be the winner of the vote. For the preferential system it means that the t-t-shirt the participant ranked as their number one preference will be the winner of the vote. The other two groups will not experience a preference match. For the plurality system this means that the t-shirt the participant voted for will not be the winner of the vote. With the preferential system it means that the winning t-shirt is not the t-shirt the participant ranked as their number one preference.

3.3 Procedure

3.3.1 Pilot study

First a pilot study has been conducted to make sure that the experiment corresponds with the aim of the research and that there could be no mistakes in the layout of the experiment. This pilot study was distributed amongst a small group of participants. The participants were somewhat directly selected out of the immediate social circle of the researcher. The pilot was distributed via an internet link and was conducted using the program Qualtrics. In total 17 participants took part in the pilot study, of which four participants took the pilot survey by hand with the researcher present. This was done to get a better feel of how the participants reacted to the pilot survey and to allow them to ask questions directly if any clarification was needed. In the pilot study the company TwoDesign was introduced, this company was presented as one that specializes in the sale of t-shirts with a striking design. Participants were asked a number of questions about this company, to check if TwoDesign would be passable as an actual company. The items for this variable can be found in table 4.

Results showed that this variable scored a mean 3.10 out of 5, which means that TwoDesign was treated rather neutral. This is considered sufficient as only a negative outcome would indicate that TwoDesign was not believed to be an actual company that could exist. Next the participants were shown eight t-shirts, which they were asked to rank to their preference. These t-shirts are all featured on the website of Threadless.com. To ensure that the original designers of these t-shirts get the credit they deserve, they are all linked in appendix…

(27)

26 together with the link to the specific page the t-shirt is featured on. The eight t-shirts are shown below:

Figure 2. T-shirts in the pilot experiment

Of these eight t-shirts, four would be selected to be presented in the experiment. This choice was made so that during the experiment participants would not be overwhelmed with a large number of t-shirts, as to much choice can lead stress regarding the voter (Schwartz, 2004). In the pilot study eight designs were presented to ensure that the four designs that would be presented in the experiment were all equally favorable. The four t-shirts that were grouped closest together regarding favorableness would then be selected. The t-shirts that were selected are presented below:

Figure 3. Four selected t-shirts for experiment

This selection of t-shirts was based on the results in the table below. The table shows all the means of in what place the t-shirts were ranked. As can be seen the t-shirts 3,6,7 and 8 group

(28)

27 very close together. This means that participants in the pilot study regarded these t-shirts as equally favorable. That is why these t-shirts were selected to the options participants could vote for during the experiment. This ensured that participants in the experiment were equally likely to vote for either of the four t-shirts.

T-shirt Mean 1 3,08 2 5,50 3 3,80 4 2,35 5 6,00 6 4,00 7 4,08 8 4,08

Table 3. Means of t-shirt ranks

One of the most important reasons to conduct the pilot study was to make sure that participants understood both the plurality voting system and the preferential voting system. This was examined by providing an example of both voting methods and have the participants select which option would be the winner in each voting contest. This was done to make sure that participants understood the voting system, but also so they could experience the difference between the two systems.

This was researched by presenting the participants with examples of both voting systems and asking for each example which option would be the winner of the vote. After participants were asked questions about the fairness and comprehension of both the voting systems.

In order to provide better insight in the underlying relations between the before mentioned concepts, some more variables will be used in the analysis. To provide better insight in the relationship between the type of voting system and satisfaction with outcome of the vote, the variables “fairness” and “comprehension” will be included in the analysis.

The variable “attitude towards company contests” will be used as a control variable to make sure that preexisting attitudes of participants towards company organized contests will not be of influence on results.

(29)

28 The examples and complete pilot study can be found in appendixes B and C. Lastly the participants were also asked about their general attitude towards contests organized by companies.

The items for these variables are presented in the table below:

Variable Author Items Translated item

Credibility of TwoDesign Chadhuri, et al. (2001) 1. This brand is dependable

2. This brand is reliable 3. This brand is trustworthy 1. Op TwoDesign kun je bouwen 2. TwoDesign is betrouwbaar 3. TwoDesign is te vertrouwen Comprehension of voting system Kamoen et al. (2007) De tekst is: 1. Eenvoudig te begrijpen 2. Ingewikkeld 3. Overzichtelijk 4. Logisch opgebouwd 5. Omslachtig

Deze stemmethode is: 1. Eenvoudig te begrijpen 2. Ingewikkeld 3. Overzichtelijk 4. Logisch opgebouwd 5. Omslachtig Fairness of the voting system Kwak et al. (2015) 1. Fair 2. Just 3. Reasonable 4. Acceptable

Deze stemmethode is: 1. Eerlijk 2. Rechtvaardig 3. Redelijk 4. acceptabel Attitude towards company organized contests Belezza et al. (2014) 1. I like them 2. I react favorably to them

3. I feel positive about them 1. Ik vind ze leuk 2. Ik reageer er gunstig op 3. Ik ben er positief over Table 4. Pilot study measurements

(30)

29 The means for each of these variables were calculated using the program SPSS. Results are shown in the table below and the Cronbach’s Alpha for each variable can be found in appendix D.

Table 5. Results of pilot study

By consulting the mean scores of all the variables it became apparent that all variables had a neutral or positive mean (i.e. a mean score of 3 or higher). Both attitude towards TwoDesign and Attitude towards company contests showed relatively low means, almost reaching neutral. This was positive, given the fact that the attitude towards TwoDesign variable was only included to make sure that the fictional company of TwoDesign would not make a difference in participants answers and that the company itself could actually exist. Attitude towards company contests was included to test the average attitude participants had toward these types of contests. A strong preexisting opinion on this matter might have influenced results. The pilot showing that this variable is almost neutral is a positive, as results will likely not be influenced by the participants attitude towards these types of contests. In the experiment the company TwoDesign was thus used.

Fairness and comprehension both showed high means for the plurality voting system as well as for the preferential voting system. This means that participants were able to understand both voting systems and also regarded them as fair. The plurality voting system as well as the preferential voting system, in the form of the Borda-count, will thus be used in the experiment.

What is however interesting to take away from this pilot study is that the fairness of the preferential voting system is regarded higher compared to the plurality voting system. This might already proof to be some insight in hypothesis 1, the experiment however will provide actual insight in these relationships.

Variable Mean

Attitude towards TwoDesign 3.10

Comprehension plurality voting 4.74

Comprehension preferential voting 4.2

Fairness plurality voting 3.9

Fairness preferential voting 4.5

(31)

30 3.4 The experiment

3.4.1 Setup of experiment

Participants were put in groups so that they either got to vote by means of the plurality voting system, in which they got to vote for only one t-shirt. Or by preferential voting, in which they could rank order the t-shirts according to their preference. The other part of the group was created by either showing the participants a preference match or a non-preference match in the results of the voting contest.

In the first part participants were randomly put into one of these four groups. In this part they only got to vote for the t-shirts. After having voted, either by plurality or preferential system, they were asked for their e-mail address. This was necessary so that the participants could be personally contacted to take part in the second part of the experiment.

To ensure that the participants remained in the same group as they were in in the first part of the experiment, a number of sixteen different experiments were created. Each of these experiments consisted of a specific part and a general part. The specific part was tailored to the group the participant was in in the first part of the experiment. This worked so that a participant who voted by means of the plurality voting system with a preference match, received a personal link to an experiment in which the participant got to see the four t-shirts he could vote for, the t-shirt he voted for and the t-shirt that has won the contest. Because the participant was in a preference match group, the winning t-shirt was also the t-shirt the participant voted for.

This is done for each possible outcome of the contest in the experiment. In group 1 participants could vote for four t-shirts, so four experiments were created. Each one showed another t-shirt that was voted for and showing the corresponding winner.

This worked the same for the preferential voting group with a preference match group. Here participants had to rank order the t-shirts in order of their preference. However, because of how preference match is defined in this research, participants only got to see the t-shirt they ranked on top together with only the winning t-shirt. Which in the case of the preference match group is the same t-shirt as the one they ranked on top.

The non-preference match groups work the same. However, instead of receiving a link to an experiment in which the shirt they voted for/ranked first is the winning shirt, the winning t-shirt is one that they did not vote for/that they not rank on top.

The participants are contacted for the second part of the experiment approximately three days after they took part in the first part of the experiment. This is purposefully done to give the participants a feeling that they are actually taking part in a real voting contest. This way it

(32)

31 would seem that votes were actually being collected and that the outcome was not already predetermined. The participants have been contacted by e-mail. Containing the personal link to continue in the experiment. This research aims to achieve at least twenty participants per group, which allows for a viable analysis (Field, 2013). Because of an expectation of a high mortality rate (i.e. participants that finished part 1 of the experiment, but will not finish part 2), the aim for the first part is to reach at least 200 participants.

The general part of the experiment consists of a series of items about TwoDesign, fairness of the voting system, comprehension of the voting system, satisfaction with outcome of the vote and attitude towards TwoDesign. Preference match will also be checked for, as well as general attitude towards company organized contests.

3.4.2 Measurements

The table below shows what measurements were used to measure the different variables in the experiment. Because the variable voting system is the independent and will be manipulated in the experiment it is not included in the table. The independent variable preference match is included, as a means a manipulation check. For this variable only one item was used.

First the variables comprehension and fairness are presented in the table. Both variables use the same scales as used in the pilot. Both scales are borrowed, their respective Cronbach’s alpha’s being .81 and .96. The scale for comprehension was already in Dutch and needed no translation or other. The scale was originally used to measure how well people understood a text they had to read. The scale translates well to the purpose of this experiment. The scale for fairness was originally used to identify price fairness, but is general enough to be purposed to this research. This scale had to be translated into Dutch, which was done as literal as possible. Both variables are measured by a 5-point Likert scale.

For the variable satisfaction with the outcome of the vote, two sets of measurements were used, both consisting of 3 items. The first set was a scale developed by Allen et al. (2015) with a Cronbach’s alpha of .97. The scale was originally used to asses peoples satisfaction with a business. The scale was translated to Dutch for the purpose of this experiment and because all three items were open ended, it made them easily adjustable to suit this experiment. This variable is measured by a 7-point Likert scale.

The second set of items was constructed by Haumann et al. (2015) with a Cronbach’s alpha of .94. It was originally used to measure satisfaction with a co-production process. For this experiment the scale was translated to Dutch and used to measure how satisfying the outcome

(33)

32 of the vote was. No adaptions to the scale had to be made, only the preceding question was altered in comparison to the original scale. This variable was measured by seven-point bi-polar adjectives.

The variable attitude towards a company was measured by a scale created by Park and John (2014) and has a Cronbach’s alpha of .90. The scale was originally used to measure attitude towards an article, but is also used to measure attitude towards an advertisement and credibility (Bruner, 2017, p.70). For this experiment the items were translated to Dutch as literally as possible without further alterations. The variable was measured using a 7-point semantic differentials.

The final variable, attitude towards company contests, was measured by the same set of items that was used in the pilot. This scale is created by Belezza and Keinan (2014) and has a Cronbach’s alpha of .94. The scale was originally used to measure attitudes towards non-core students. For this experiment the items were translated to Dutch and measured on a 5-point Likert scale.

The items for measuring the demographics of the participants in the experiment are also presented in the table below. Participants will be asked to provide information about their gender, age and education.

(34)

33

Variable Definition Author of scale

Original Scale Adapted and translated scale

Comprehension The ability

of a consumer to understand how the voting system works and how a winner of a contest is calculated Kamoen et al. (2007) 1. Eenvouding te begrijpen 2. Ingewikkeld 3. Overzichtelijk 4. Logisch opgebouwd De stemmethode was: 1. Eenvoudig te begrijpen 2. Ingewikkeld 3. Overzichtelijk 4. Logisch opgebouwd

Fairness The degree

to which a consumer believes that the voting system is able to determine the winner of a contest that reflects the votes of consumers best Kwak et al (2015) 1. Fair 2. Just 3. Reasonable 4. Acceptable De stemmethode was: 1. Eerlijk 2. Rechtvaardig 3. Redelijk 4. Acceptabel Satisfaction with outcome of the vote

Degree to which the consumer is content with the winner of the contest Allen et al (2015)

1. I am satisfied with this…. 2. I think that I did the right

thing when I selected this…

3. I am happy with this…

1. Ik ben tevreden met het winnende T-shirt

2. Ik ben van mening dat het winnende T-shirt de juiste was 3. Ik ben blij met het winnende

T-shirt

Haumann et al. (2015)

The overall co-production process was: 1. Dissatisfying/Satisfying 2. Displeasing/Pleasing 3. Terrible/Delighting De uitkomst is: 1. Onbevredigend/Bevredigend 2. Onaangenaam/Aangenaam 3. Onprettig/Prettig Preference Match A consumer vote for an option is also the

- The winning shirt is also the T-shirt of my highest preference

Het winnende T-shirt is ook het T-shirt van mijn hoogste voorkeur.

The influence of voting systems on satisfaction with the outcome and attitude

5-2-2019

Co-creation

contests

The influence of type of voting

system on satisfaction with the

outcome the vote and attitude

towards a company

Niek Stortelder

Studentnumber: 4348354

Supervisor: Dr. P. H. Driessen

Second examiner: Dr. H.W.M. Jooste

Master thesis Business Administration,

Marketing

Co-creation contests

The influence of type of voting system on satisfaction with the

outcome the vote and attitude towards a company

Preface

Abstract

Contents

1. Introduction

2. Conceptual framework

3. Method