The relation between multihoming and consumers innovation
adoption intentions: an analysis in the video game industry.
Eleonora Toso Student ID: 11838000
eleonora.toso@student.uva.nl 22nd June 2018
MSc. in Business Administration - Entrepreneurship & Innovation University of Amsterdam – Amsterdam Business School
Statement of originality
This document is written by Student Eleonora Toso 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.
Statement of originality 1 Abstract 3 Introduction 4 Research question 8 Literature review 9 Multihoming 9
The video game industry 12
Innovation adoption behaviour 16
Model development 19
Research design 21
Method 21
Measures 22
Sample 23
Preliminary analysis of the sample 25
Preliminary analysis 29
Hypothesis testing 33
Discussion 36
Conclusion 37
Limitations of the study & Future research 38
Abstract
This thesis is an exploratory study into the relationship between multihoming and adoption intention of a certain product with an innovative feature. In particular, it focuses on the market of video games consoles, a context where people are likely to own more than one device. The topic of multihoming has been investigated in literature, highlighting especially the firms’ perspective and the benefit that video game companies could gain from it. However, the studies that analyse multihoming under users’ perspective are still scarce. This research was built on literature review about adoption intentions and it focused on two characteristics of an innovation: relative advantage and compatibility. These characteristics have been found to have a strong effect on adoption intention by past research. The purpose of this study was to investigate if this relationship was somehow affected by the ability of multihoming. On a sample of 267 people, results showed that multihoming does not affect relative advantage in the adoption intentions, but it does affect compatibility. However, the difference in adoption intention between people who do not multihome and people who do is not as significant as expected. Interesting outcomes came from the analysis of genre and features in videogames: results showed that for some genres there is an influence in the adoption intention of a videogame with innovative features. This would be in favour of what was also stated in previous studies about video games: users are always looking for a balance between familiarity and innovation, therefore some genres would facilitate the user decision to adopt a videogame with innovative features.
Introduction
Today mobile applications have a central role in our daily life and where one application can be used for multiple purposes. WeChat allows people in China to purchase, book restaurants, transfer money, get in touch with friends and so on, all within the same application. In Western countries Facebook1 tried to integrate a large amount of services into its own application. However, this process of integration of services under the same platform (or in the example above, the same application) is not valid in the case of platforms. When it comes to platforms, differentiation is still a reality and people have to use different ones to run different software and therefore they are forced to multihome. The first time that multihoming became prevalent was at the beginning of the Nineteenth century, when the invention of the telephone started to spread among citizens. Each household had a different phone line and, without compatibility, a separate phone line was required to reach another phone. Even if some hundreds of years have passed, in some network markets, such as payment systems and game platforms, compatibility has not seriously been taken into account (Doganoglu & Wright 2006). As a consequence, we still have more than one credit card in our wallet, we use both Windows and Linux, some people might use both whatsapp and telegram to message and some might own both Xbox and Playstation, even if they all serve the same purpose. This practice of consumers using more than one platform has been also seen as a way to gain maximal network advantages (Choi 2007). Thus, consumers are not just forced to use more than one platform because of the absence of compatibility, but they might want to do so.
1See for example the attempt from Facebook to connect directly instagram stories - a feature which allows users
to share videos and pictures for 24 hours- with facebook stories.I.Lunden (2018).Instagram tests letting users post Stories directly to WhatsApp. Posted Jan 3, 2018 retrieved on https://techcrunch.com/2018/01/03/instaapp/
In fact, some authors have pointed out how compatibility might not be pursued by firms when users have more than one platform at the same time (in other words when they multihome), affecting the incentives for compatibility. If the consumer buys two platforms, each firm will increase its total sales. This gives firms a reason to remain incompatible, going against the tide, even when compatibility is socially desirable. In addition, the fact that there are consumers that multihome reduces incentives to be compatible, due to the fact that consumers do not mind buying twice even if the prices are high. (Doganoglu & Wright 2006).
Starting from these assumptions, the aim of this research is to further investigate the reason why a consumer who engages in multihoming chooses to buy an application for either one or the other platform he owns. Given that users multihome, what are the factors that drive their budget allocation in buying applications?
Some literature has emphasized how complementary applications will attract users and increase a platform installed customers base (the number of customers that are using the platform) (Cennamo & Santalo 2013). However, this kind of analysis has been carried out mostly under the firm’s point of view rather than the consumers’ one. If a user multihomes, why would he buy an application for one platform rather than another? Do innovative features play a role in this decision?
A platform is defined as an interface that mediate between different groups of users (Cennamo & Santalo 2013). In other words, platform is the hardware that runs the software (eg. computer, smartphones, video game consoles) (Marchand & Henning-Thurau 2013). Many topics have been studied in relation to platforms. The most investigated ones are network effects, two sided markets and multihoming. As mentioned above, in this thesis focuses especially on the latter and the relation it has with consumer innovation adoption. Given that a certain group of users can multihome, how does a consumer’s innovation adoption behaviour change?
Research consumers’ attitudes in relation to innovation has been carried out. Part of the literature has also investigated how at different stages of innovation adoption, situations in which the innovative product might be used or bought can be seen differently every time, leading to different adoption decisions (Arts et al. 2011). For example, when the innovation of the videophone was introduced for people to test it, many stated that they would use or buy such an innovation; however, when the product was actually launched different factors such as the perceived cost prevented people from actually buy and adopt the innovation (Arts et al. 2011). If consumers intention of innovation adoption is related to multihoming, the question could be how multihoming can change a consumer adoption intention and behaviour. In fact, knowing that the users have the possibility to choose an application for different platforms they own, how would their acceptance of an innovative application change? Would they choose an application because of its innovative features or would they keep making decisions based on factors such as price consciousness and brand familiarity? (Arts et al. 2011). More in general the problem addressed here is part of the research concerning the acceptance of an innovative feature, and if the innovative feature is considered to be intriguing enough for consumers when they have the possibility to choose whether to buy an innovative application or not. Since for platforms firms it is difficult to predict users’ satisfaction when it comes to complementary applications (Cennamo & Santalo 2013), knowing consumers attitude toward innovative features when they can multihome, could help reducing the asymmetric information problems that firms face when dealing with multihoming.
This research on the relation between multihoming and innovation adoption is going to focus on the video game industry. There are two main reasons behind this choice. The first reason is that this industry is often used as an example of industry where multihome takes place (Doganoglu & Wright 2006, Landsman & Stremersch 2011). In addition it is a classic example of “platform-based systems” and after the launch of the latest console by Sony (PS4)
and Microsoft (Xbox one) it has become a popular topic for academics and researcher in marketing (Steiner et al. 2016). As a matter of fact, since more recent generations of console have become more targeted, it is more likely that the same family purchases different consoles (Dubè et al. 2010). A famous example is the one of Nintendo Wii: not only it had the innovative motion controller, but for its launch it targeted mostly families and children with games that were suitable for all ages and for multiplayer games. The second reason is that this industry, as part of the creative industries, seek continuous market advantage through novelty and innovation, given that it has a hit-oriented nature and a short product life cycle. However, even if consumers of video games expect novelty in products, at the same time they wish it to be familiar and accessible (Tschang 2007), which is the dichotomy this research wants to investigate while adding the variable of multihoming. In the video game industry innovation has been measured studying different factors. Some of them have been identified in the form of gameplay, the visual style and the story or background and genre. In addition, games that are similar in gameplay, visual styles and story can be labelled as innovation if they extend a genre (Tschang 2007). There are many studies related to genre in the video game industry. Although videogames are considered an aesthetic medium as movies or theatre, the genre division made by literature differs. Video Games genres could be divided according to their aesthetic or mechanic elements. Some video games might have storytelling elements that can be easily identified, some others don’t. In fact, some genres are defined mostly by the controls mechanisms that the player can use in the gameplay rather than the story itself, whereas some others are defined more by the aesthetic elements, in a more similar way to the movie genres. (Karhulahti 2011). For example, a genre based on aesthetic elements could be sci-fi, or mystery or far-west, while the ones based on mechanics could be adventure games, puzzles etc. This is why the incremental innovation in video games could concern the extension of the genre, meaning innovating either the mechanic
element (e.g. gameplay) or the aesthetic one (e.g. context). Therefore, a question that might arise is which element a video game user pays attention to. Would he stick to a certain genre or would he be drawn by an innovative feature? Finally, this thesis aims at giving a further insight in the games field, which has grown in the last decade but still does not get as much attention from scholars, as other entertainment industries (Marchand & Henning-Thurau 2013). The intention of this thesis is to further investigate consumers innovation acceptance in the videogame industry, adding the variable of multihoming, helping the research into consumers behaviour that could turn out to be fruitful for firms that work in the game and entertainment industry.
This thesis is organized as follows. The first section will concern the literature review that is going to discuss what is known about multihoming, video game industry and consumers innovation acceptance. Then it will present the conceptual model, that summarize what introduced above, and the main hypothesis. The core part of this thesis will be the quantitative analysis where a sample of video game users will be investigated to test the main hypothesis. After an analysis of the main characteristics of the sample and some required statistical procedures, a regression analysis will be performed in order to check the relationship between the main variables that this thesis wants to investigate. This thesis work will end with the discussion of the results and conclusion will be drawn.
Research question
Given what stated in the introduction the research question for this thesis is the following:
“When users are considering buying and adopt a new application, how much innovative characteristics influence the decision? How this influence is modified if the user can multihome?”
Literature review
Multihoming
The first definition of multihome appeared in the IT sector in relation to internet connection. Multihoming takes place when the so-called host has different and multiple connections to the network via one or more Internet Service Providers (ISP), making connectivity stable even if one connection fails (Crémer et al. 2000). If taken out of the IT sector, multihoming takes place when the same user purchases and generally uses two competing products, that serve the exact same purpose, in order to gain a higher advantage thanks to the maximization of network benefits (Doganoglu & Wright 2006). Another definition of multihoming has been offered by Caillaud and Jullien (2003) where multihoming is addressed as the behaviour users have, when they request a service to several intermediaries at the same time. On the same page is the definition by Rochet & Tirole (2003) who states that a user multihomes when he connects to several platforms. The opposite of multihoming is single-homing, where an agent decides to use just one platform (Landsman & Stremersch 2011). Multihoming behaviour started diffusing with the introduction of the telephone, in the 19th century. Before telephone networks were merged and developed as we know them today, each user would need one telephone line per person to make a phone call. Even though telephones have followed the path of compatibility, many other sectors are still relying on multihoming. The most common example are payment systems and game platforms. How many different credit cards do you have in your wallet? How many platforms do you play games with? It is likely that without even realizing it, you have been multihoming for a long time.
Multihoming can be seen from different points of view. The first one is called seller-level multihoming and takes place when an application offered on one platform by a seller is made available for different platforms. This happens when instead of subscribing exclusive
contracts with independent software developers, the application developed by a company is available on different competing platforms (e.g. when a game published on a platform is also available for other platforms). The second, platform-level multihoming, occurs when an application for one specific platform is available for buyers of a competing platform. In this second situation the application is developed internally by the company that develop also the platform; however instead of making the application available only for the platform it produces, it sells it also to competing platforms (e.g. Microsoft Office, developed by Microsoft, is not available only for Microsoft computers but also for Apple ones; Nintendo usually does quite the opposite with its games, i.e. Super Mario is developed by Nintendo and sold just on Nintendo platforms). The last level of multihoming is buyer-level and occurs when users own multiple platforms and use them at the same time or uses them equally (e.g. the same user owns different video console platforms and usually plays with them both) (Landsman & Stremersh 2011). This is the level that the research is going to focus on. If there is little empirical research on how multihoming impacts on sales and competition, that is even less when it comes to buyer level multihoming and the influence it has on purchase decisions.
Most of the literature has investigated multihoming on a platform level and its relationship with platform competition, since multihoming decisions are important to firms in many different contexts such as platform sales (Landsman & Stremersh 2011).
However, buyer-level multihoming is still an unexplored path that, if analysed, could bring a new perspective to firms when it comes to deciding if and which kind of application to develop.
Given what said above, the purpose of this research is to address these gaps in the literature, as pointed out by Landsmann & Stremersch (2011) and analyse buyer level multihoming and in particular the relationships between users’ buying behaviour toward applications.
Secondly, literature on platform adoption, even when focused on consumers expectation, has investigated mainly the relation with variables such as network effects (Steiner et. al 2016). Literature has also focused on hardware quality in relation to network effects (Gretz 2010). However, as Steiner et al. (2016) point out, most of the previous literature that investigates platforms adoption has treated consumers as homogeneous, taking into consideration the market average. Even when consumer expectations were considered in a network context, it was done using strong assumptions (Steiner et al. 2016). An example of these assumptions is a study conducted by Zhu & Iansiti (2012), on platform-based markets. In fact, when analysing the role of consumers expectations on the success of entrants in those markets, Zhu & Iansiti’s research consider the same consumers population with the same type of expectation, therefore it does not consider casual game players that might value different aspects of a console. Given the fact that now, more than in video games’ early years, the segmentations have become more relevant especially because of the different genres available, the different platforms and the different targets, it is not possible to generally assume that there is a unique kind of player. Moreover, the study carried on by Zhu & Iansiti explicitly excludes multihoming, associating each group of consumers with a certain platform, without taking into account the fact that each gamer could be playing with more than one platform. In addition, when selecting consumer preferences, it aggregates platform quality and application quality in the same measure, assuming that platform quality is more important. Another study about the increase in market share due to indirect network effects, by Dubè et al. (2010), gives consumer expectations a central role. As in Zhus & Iansiti’s study, consumers are assumed to single-home. Consumer’s expectations are based on the assumption that adoption of a certain software is related to future hardware prices and software availability. Dubè et al. do not take into account consumer heterogeneity either, just as they do not take into account the game content per se when talking about console adoption.
Therefore, some characteristics such as video game genres or features are not taken into account even though some literature concerning the video game industry have been found to be relevant when talking about video game innovation and design. However, having analysed some literature about video game designs (Tschang 2007, Pinelle et al. 2008), it is undeniable that a study about videogames cannot disregard a genre analysis.
The video game industry
As mentioned in the introduction, among the many industries where multihoming is a reality, we are going to focus on the video game console market. This market can be considered as an established example since it has already been studied in relation to multihoming, given that seller-level multihoming decision significantly affect a firm performance and can influence many decisions in different contexts (Landsman & Stremersh 2011). The home video game systems consist of two parts: the console or hardware and the software, which are the games that can run in that specific hardware. The 1980s has seen the dominance of Nintendo’s NES, but in the following generations of video game platforms, the market has seen the launch of other consoles; currently the market share is owned by three competitors: the Japanese Nintendo and Sony and the U.S. Microsoft. In addition, also personal computers have increased their performance and are often used as a gaming platform. Home video games have been and still are incompatible, which means that the same video game (software) cannot run on a different platform rather than the one it has been developed for. This incompatibility not only applies to games for consoles of different producers, but also to consoles of different generations.
The software can be developed by the console manufacturer itself (these videogames are also known as “in-house” titles) or the manufacturer can sign a contract with independent software publishers (eg. EA, Ubisoft) that develops the game for the platform. It is common
that the firm that develops the console has the rights to approve games before publishing them. Moreover, the console manufacturer can hold the rights of the developed game, so that it cannot be published for any other platform. However, even if at the very beginning video games were developed specifically for one platform, in the last 20 years the number of game titles developed and published for more than one platform at the same time has increased up to 40% (Corts and Lederman 2009).
Another interesting aspects that should be taken into account is related to the software development. The performance of the software (that in our scenario corresponds to a video game title) depends on the hardware (here a console) on which the game is run. Console’s firms have increased their market share with the release of portable systems (called handheld systems) giving gamers the possibility to differentiate. Examples that can be found in the market are Nintendo Ds and Playstation Portable.
Many studies about the effect of exclusivity for manufacturers, have been carried out by researchers. However, there is still very little literature on users’ buying behaviour when multihoming takes place.
Furthermore, as mentioned in the introduction, the video game industry - as part of the creative industries - has a hits-oriented nature, a short product cycle and, most relevant for this research, difficulties in predicting product acceptance (Tschang 2007). The fact that in creative industries, and so the video game industry, there is this need to pursue innovation, makes it a good ground to investigate consumers’ attitude toward innovation. In addition, it has been studied that this particular industry faces the struggle of balancing technical knowledge and creativity in order to seek the constant innovation that is required to maintain a competitive advantage. The struggle that firms face comes when the consumer expects not only innovation and novelty but also familiarity (Lampel et al. 2000). As a consequence, in the video game industry competition is harsh both on firms that do not innovate and keep up
with times and at the same time on firms that experiment too much failing to meet player’s expectations (Tschang 2007). Some authors have stressed the fact that users know if they like a game only after they have played it. This means that when purchasing a new game, users are undertaking a risk that could influence their decision (Marchand and Henning-Thurau 2013). Even though they can easily gather information about prices, the one about quality are more difficult to gather. Due to that, consumers face the problem of having to evaluate every time the options they have before buying the product itself. Some authors, such as Nelson, have distinguished the process of gathering information on prices and the one that aim to evaluate the product. Some consumers, in fact, would directly buy the product, in order to gather information. Nelson calls this “information process experience” (Nelson 1970). Information about a certain product can also be obtained from friends or family members, specialized magazines or advertising. Consumers then decide if they want to take into account the information they acquired or if they want to continue their research. When basing the decision on experience, consumers should have bought and tried more than one brand or product in order to compare them and decide which is the best one. Different consumers experiment different brands based on the input received by friends and family, therefore there is a relationship between information obtained from research and the one obtained from experience. In our context, it must be also taken into account the ability of the consumer to multihome. Based on Nelson’s conclusions, the assumption is that users who multihome not only take into account experience of a brand or friends and family suggestions, but also the fact that they possess more than one platform. This could mean that before taking a decision user implicitly compare the two brands or console. Due to multihoming, it is assumed that when experimenting, users just consider applications for the platform they possess, limiting the cost of research or experiments they have to face.
Another aspect that past literature has investigated and should be considered in this research is the so called “experiential aspect of consumption” (Holbrook & Hirschman 1982). As previously underlined, consumers have to make decisions when purchasing a product, and to do so they use logical thinking. However, the purchase decision is not based just on comparison of prices and advantages. Especially when consumption has to do with leisure activities, it involves feelings, fantasies and fun. All these factors are called “experiential view” by Holbrook & Hirschman.
In this experimental view the consequences of the consumers’ choice do not depend on the mere utility -if the good served the purpose- but on multiple factors such as the fun, the enjoyment and the pleasure it generates. Thus, the benefits are not tangible or do not carry what is usually called “utilitarian function” but a “symbolic meaning” (Holbrook & Hirschman 1982). This is particularly relevant in entertainment and arts industries such as the video game one, as we already mention the fact that in this industry, users are looking for enjoyment and in general for game variety before adopting a console rather than the utilitarianism strictu senso. Another important factor is given video game quality and the so called “superstar software” (Steiner et al. 2016). These popular and best seller games are not only an incentive to the adoption of a certain console, but also, when they become the user’s favourite title, an incentive to prefer to use that console or keep buying that title and sequel for the same console.
In this chapter we have learned that the video game industry is wide and complex, and consumers’ behaviour cannot be studied based only on price expectations. In fact, other factors such as the enjoyment and experience do influence the purchase decision. In addition, the continuous search for innovation and familiarity makes the challenge to understand users’ behaviour harder.
This is why the video game industry is a perfect field to look into to contribute to the research on consumers’ behaviour through innovation and its acceptance. In addition, the fact that it is common that gamers multihome, and therefore have a choice in buying an application because of its innovative features or stick to something they are familiar with, could help firms and academics shortening the gap, considering the fact that in the video game industry, consumers strongly influence firms in decisions concerning the development of a new videogame and the balance between innovation and consumer satisfaction is essential.
Innovation adoption behaviour
Innovation adoption has been defined as the decision that a consumer takes whether to use or not an innovation (Arts et al. 2011). Many are the models that have been used to study and explain consumer innovation adoption and among them, there is the Technology Acceptance Model. This model developed by Davis (1989) uses two main variables called “perceived ease of use” and “usefulness”. In particular when talking about innovation adoption in the video game industry, the ease of use mentioned in Davis’ model should be taken into account. In this case it can be seen as the perception the user has about the ease of adapting to the game style of the new video game. Usually studies on adoption behaviour analyse the perceptions and the characteristics that a consumer who had adopted an innovation has, in contrast with consumers who have not. Moreover, the characteristics of the consumers are considered to be the main drivers of the innovation adoption (Arts et al. 2011). In many studies those different characteristics have been used to divide the consumers who adopt an innovation in different groups (for example age, gender). Less used, but still relevant variables, are price consciousness and brand familiarity. When adopting an innovation, consumers usually take into account their perception on compatibility, complexity, advantage and risk of the innovation. This is because they have limited awareness of the new product or
innovation and all the purchasing decisions are made on preferences (Hoeffler 2003). This is well adjusted with what was said in the previous paragraph about videogames: users do not have a lot of information about a new game therefore they base their decision either on suggestion or on genre or feature preferences or on the platform they own. Another important aspect of innovation adoption have been acknowledged by some authors with the fact that the process is dynamic, therefore intention and behaviour might change during this process (Castagno 2008). As a matter of fact, when talking about intentions there have been some studies that have described the intention as an “immediate determinant of behaviour”. However, the same studies have also pointed out how the measure of an intention won’t always be an “accurate predictor of behaviour”. This discrepancy originates from two main reasons: the fact that the intention might not last over time, meaning that intention is not always constant, and the fact that the degree to which the measure of the intention and the behaviour correspond might change and be less accurate for each situation (Ajzen & Fishbein 1980). All in all, researchers argue that if the measure of the intention corresponds to the measure of behaviour, it is possible to make an accurate prediction. On the other hand, the longer the time interval that occur between the intention and the behaviour observation, the highest the chance that the intention has changed. Despite this, sometimes it might be neither possible nor feasible to measure intentions and behaviour in a close time frame. The fact that it is hard to do so does not mean that it is not possible. As a matter of fact, sometimes it is more relevant to predict or study behaviour at an aggregate level. At this level it is more likely that the intentions persist, since the events that could influence a change in the intention could be balanced out by the aggregation. Given this consideration it can be deduced that a good way to balance intention and behaviour is to analyse a significant sample. This research will consider those factors while investigating consumers’ behaviour through innovation. In particular, given what said by Ajzen & Fishbein, the analysis will be
carried out using the law of large numbers in order to reduce the weight of each individual on a macro level, trying to reach as much respondents as possible.
As mentioned earlier there are different effects of innovation characteristics. In particular the one that have been studied in (Arts et al. 2011) are the following:
Table 1
Feature DESCRIPTION
Relative Advantage when an innovation is perceived as being better than its prior idea or model
Compatibility when an innovation is seen as consistent with existing value, past experiences, life styles and needs
Complexity the degree to which an innovation is perceived to be difficult to to use
Trialability when an innovation can be experimented
Observability when the results of an innovation are visible to others
Uncertainty the degree to which social and financial consequences cannot be established
This research is going to consider the effects studied by previous literature with the purpose of seeing if the same expected effects occur when the variable multihoming is present. In fact, after checking in an empirical analysis if these innovation characteristics have the same effect on adoption intention in the video game industry, the variable multihoming will be included and the results studied.
Model development
As stated in the introduction this thesis wants to investigate how innovative characteristics influence consumers’ adoption intention in case of multihoming. Literature have identified different innovation characteristics that can affect the adoption intentions of an innovation (Arts et al. 2011). As highlighted in Art et al., in literature concerning innovation adoption there are six main features consumers use to evaluate an innovation, however it has also been pointed out and recognised that the ones that have a positive impact in consumer innovation adoption are four.
As a matter of fact, it is relative advantage, compatibility, trialability and observability that have a positive effect on the consumer’s evaluation of innovation. In addition, according to the study performed by Arts et al., the innovation characteristics that seem to have a strong positive relation with adoption intention are compatibility and the relative advantage.
For the purpose of this research I have decided to limit the analysis to these two characteristics in relation to adoption intention, when talking about the video game industry, because some authors have pointed out how video game users look for both novelty and familiarity when purchasing a new application, making game design no easy task (Tschang 2007). Consumers tend to buy games within genres and games that drift apart from consumers’ expectation can fail. This need of familiarity can be related to what Arts et al. call compatibility since this characteristic is related to how much an innovation is in line with past experience (i.e. previous video games) and styles (i.e. genres). Concerning adopter characteristics Arts et al. have found limited effects on adoption intentions, therefore in this thesis socio-demographics will be analysed but not run in the regression and in relation with the hypothesis studied. On the other hand, when talking about innovation adoption, adopter characteristics have also been divided in psychographics and the less used one are brand familiarity, price consciousness, self-confidence and dogmatism (Arts et al. 2011). Because
these have not been used that much in past literature, as Arts states, it was decided not to include them in the survey for this research.
This being said, in my analysis the independent variables are going to be compatibility and relative advantage, while the dependent variable is adoption intention. Multihoming will play a role as a moderator, to verify if the presence or absence of multihoming influence the direct relationship with innovation adoption. Regarding this interaction the hypothesis is that, if people multihome, the interaction is going to be stronger. In fact, the assumption is that a user that buys more than one console was driven by the conviction that owning a variety of platforms would better suit his demand for a customized experience and so would fit well his needs. Since also the independent variable “compatibility” expresses this demand for personalization in the experience – insofar as it represents the perception of familiarity with the gaming style – this research assumes that when circumstance “multihoming” occurs, the users’ adoption intention is more influenced by their perception of the consistency of the new features with their gaming style.
Moreover, it is presumable that the same exacting user purchased different consoles in order to satisfy the desire of playing the latest version of his games, no matter which platforms they are published for. Being the independent variable “Relative Advantage” an indicator of the perceived enhancement produced by the innovation – so that the video game appears to be a cutting-edge one –, this work supposes that in presence of “multihoming”, users’ adoption intention is more likely to be affected by the improvement provided by new innovative features in their games.
In addition, when running the analysis, games genres and characteristics will be introduced in the model as covariates to see if there is any influence in the main relationship.
Hypothesis
Given the main research question and the conceptual model, the hypothesis have been formulated as follow:
H1: The relative advantage of an innovation has a positive effect on the adoption intentions. H1a: The positive relationship between relative advantage and adoption intentions is moderated by the presence of buyer-level multihoming. The presence of multihoming makes the relationship stronger.
H2: The compatibility of an innovation has a positive effect on the adoption intentions.
H2a: The positive relationship between compatibility and adoption intention is moderated by the presence of buyer-level multihoming. The presence of multihoming makes the relationship stronger.
Research design
Method
My research method used a deductive approach and is of explanatory nature. It is based on a quantitative research. In addition, the research design is observational. Previous literature
Compatibility Relative Advantage Multihoming Adoption intentions H1a +
+
+
H2 + H1 + 2a +suggests that the most common research design method used when it comes to understanding consumers’ behaviour are surveys. This method has frequently been used in analysing games usage (Boyle et al. 2012). In addition, when talking about consumer expectations, survey was used as a data collection method (Frels et al 2003). When studying platform adoption in markets, one of the approaches used in literature is the so-called “mindset metrics” such as customer attitudes, expectations and intentions in the adoption process (Steiner et al. 2016). Since this research wants to link buyer-multihoming behaviours to customer’s attitude toward innovation the methodology used in past research about adoption innovation behaviour has been analysed and it has emerged that questionnaires and surveys are the most common methods to gather information (Claudy et al. 2015). Therefore, following this trend, data were collected through a survey.
Measures
Some literature states that there is more than one reason why, in the gaming industry, the internet is a good channel to carry out research. Among those reasons Wood et al. (2004) pointed out the fact that the internet is widely used by gamers on a daily basis, allows to reach a large sample in a quick and efficient way, reduces social desirability leading to a higher validity and simplify participant recruitment through websites and social networks. Wood et al. state also that gamers are usually interested in research and are often willing to take part in it. In addition, gamers usually are part of communities and often can recommend good places where to contact other gamers (Wood et al. 2004). In the same paper they affirm that surveys are the most useful applications for online research when this involves video game players. In fact, not only they can be launched quickly and efficiently but they also guarantee a high degree of anonymity, therefore participants might be keener in taking part. The questionnaire firstly asked questions about age (ratio variable), type of console owned
(nominal variable) and number of consoles owned (nominal variable). The other questions were based on a seven-point Likert type scale with a range that goes from “strongly disagree” =1 to “strongly agree” =7. In the questionnaire there were other questions about the preferred genre, the game characteristics considered to be important, the number of people living in the household and income. The survey was administered in English.
Sample
Stage 1: Before launching the final version of the survey, a trial round was made among friends. In this round in addition to close ended answers, open ended questions were asked in order to have a better understanding on the specific reasons that drive people to buy a new videogame and/or a new console and why, if that’s the case, they decided to own more than one console. Even though the number of respondents was not wide, some interesting answers were retrieved. Nine out of ten people stated that the decision to buy a new videogame for one of the consoles or the other they own is highly influenced by the genre. The factors that influence the decision to buy a videogame were confirmed to be reviews from friends or other gamers as stated by Nelson (1970). In addition, it is interesting for our research to report the following statement by a respondent: “For example if you compare Play Station and Nintendo, you see both offer different genres of games - that’s why another console is a good addition” where he highlights the fact that game genre could influence the decision. Because of this interesting insight it was confirmed that the question about which type of genre do people play was an interesting aspect to add in the final version of the survey. To narrow down the types of genres some literature research was made. It turned out that there are different ways to divide game genres and the number of different genres has increased recently and become more detailed and sophisticated. For this reason, it is difficult to find a universal classification, as video games genres classifications tend to change according to
popularity. In general, it can be said that genres have been identified and grouped together, when some similarities in the gaming style are found (Pinelle et al. 2008). In the survey for this research the division used was the one by Apperley (2006) where he distinguishes the following genres:
Table 2
Genre Description
Simulation Videogames that simulate real life activates
such as sports, flying, driving or different dynamics of communities
Strategy Usually divided in real time strategy and turn
based strategy.
Action With the difference between first person and
third person games.
Role playing/adventure Similar and strongly related to the literature
fantasy genre.
Stage 2: The final version of the questionnaire was made of multiple choice answers and a 7 Likert scale when measuring relative advantage and adoption intention. In total 371 surveys were filled in by users. The sampling combined the snowball sampling (a non-probability technique were people share the survey with their acquaintances) and self-selection ones (where people choose freely to take part in a survey). In fact, respondents were reached through Facebook groups focused on video games (e.g. Video game fans, Nintendo switch Friends codes and Community), among student at the University of Amsterdam, among friends and some volunteers recruited through Amazon mechanical Turk. Especially in the survey on Facebook targeted groups, what was stated about Wood et al. (2004) was
confirmed. In fact, gamers showed an interest in the research and showed to be involved and helpful in commenting the questionnaire. Future research should consider the use forums also for the trial round, in order to gain a better picture of the main issues that gamers pay attention to.
Preliminary analysis of the sample
Firstly, since Mturk was also used to retrieve some respondents, the data were cleared from all the answers that corresponded to the ones that failed to answer correctly the check question. In particular, people who failed answering correctly to the statement “This is a check question please answer strongly agree”, were deleted due to the strong suspicion that those surveys were completed by a bot. The total number of answers deleted was 104, therefore the total number of respondents that have been analysed is 267. As a second step a frequency test was run, and no data entry errors were detected. In addition, a Normality check was done. A graphic test was run with histograms. To test how significant the deviation from normality were, a Kolmogorov-Smirnov test was made.
The test confirmed that the deviations from the normal are significant, with a confidence interval set at 95%. In fact for the relative advantage D(267)=.15,p<.05,for compatibility D(267)=.18, p<.05 and adoption intention D(267)=.18, p<.05.The Q-Q plot test confirmed this outcome. For the variable adoption intention (DV) the value of skew was checked. The results are that the distribution is a negatively skewed one, however, since the skewness is close to 0, with a value of -.40, the deviation from normality is considered to be acceptable for the purpose of this thesis. Z-scores were used to find outliers. No relevant outlier was detected therefore no answer was deleted.
Age
of analysis is already narrowed down by evidence in literature. In fact, users of video games are usually aged between teens and 40 years old (Marchand and Henning-Thurau 2013). In collecting the sample, it was asked to the respondent to identify themselves in certain age frames, the results are shown in the following table.
Table 3
It can be noticed that most of the video game users that completed the survey are between the age of 18 and 35 years old, which reflects what was noted by Marchand and Henning-Thurau (2013). It has to be noticed however that since the survey was administrated also at the University of Amsterdam among students, who are usually between 18 and 25, this could have influence the results, however there is no reason to think that this could have biased the analysis.
Household & Income
Other demographic Information were also gathered. In particular, the number of people living in the same household and the average income.
Out of 267 respondents, 181 (corresponding to 67,8%) live in a household with between two and four people, while 16,1 % live alone and 43 respondents have a household with more than four people. When asked the gross income per month it resulted that the majority of the people who answered is positioned in the range between 1.200 and 2.999 euro (25,8%) and 3.000 to 4.999 euro (28.8%).
Multihoming
The sample was also divided into users that do multihome and users that do not, due to the fact the research wants also to assess the role that multihoming have in adoption intentions. Of 267 valid responses 180 people in the sample do multihome and 87 don’t. In addition, among the people that multihome 29% own two consoles, 12% three and 26% more than three. In addition to understand better how people multihome and if they use mainly one console it was asked to specify how much time players that multihome spend on each console in %. It turned out that if people own two consoles 30,3 % of them spend half of the total playing time on each console, 20,2% spend between 50% and 69% of the total time on one console, 27% play with one console between 70% and 89% of the time while only 22,5% uses one console more than the 90% of the time they spend playing with videogames. To sum up it can be noticed that people who multihome with two different devices use the two consoles equally on average. This behaviour changes when people have more than two consoles. In fact, the time spent playing with the third console decreases. People that have three consoles claimed to play 50% of the time with one, 25% of the time each for the second console and the third. This shows that even when owning more than one device, there is a main used one. In addition, the fact that the number of consoles owned increase does not influence the amount of time that gamers spend on one console, presumably their favourite one. This preference should be taken into account in the following analysis.
In table 4 the reader can find the respondent profiles concerning household and income. Table 4
Entire sample Multihoming Non Multihoming
N % N % N %
Income less than 1.200 32 12 13 7,2 19 21,84
1200-2.999 69 25,8 40 22,2 29 33,33
3.000-4.999 77 28,8 56 31,1 21 24,14
5.000-8.999 31 11,6 23 12,8 8 9,20
Decline to answer 49 18,4 42 23,3 7 8,05
Household One 43 16,1 24 13,3 19 21,84
two-four 181 67,8 126 70 55 63,22
more than four 43 16,1 30 16,7 13 14,94
A correlation analysis was made to look at the relationship between multihoming and income. In fact, the assumption is that people or households with a higher income are more likely to possess more than one console and therefore multihome. The result confirms there is a relationship between multihoming and income with a r= .242, p<.01. On the other hand, no significant correlation was found between the number of people in the household and multihoming with a p=.121.
Genre
Literature has always found difficulty in defining video game genres, some authors have even been resistant to the notion (Apperley 2006). In addition, it has been noticed that users are not always satisfied with the genre distinction being repeated over and over. Thus, brings the necessity to innovate in order to make genres less static. At the beginning of this paragraph, during the trial test of the survey, it emerged that gamers take into account genres when they decide if to buy a new videogame or if they decide to diversify the type of console they own. In the survey submitted to people the genre preferences were investigated. Here are shown the results. The most common genre played by respondent is action, followed by role playing and adventure. The latest one turned out to be simulation and strategy. Most of the respondent chose at least two different genres when stating which one they usually play however there was no pattern in the association of genre played by people.
Video game characteristics
aspect is made of enjoyment of the experience and engagement. These characteristics are related to some aspects and features a videogame has (Laffan et al. 2006). In addition, other authors said how in videogame it is essential that users have a right balance between innovation and familiarity (Lampel et al. 2000). Therefore, in the survey these characteristics were identified in order to understand gamers’ preferences:
Presentation features Eg. Graphic and sound features
Social features Eg. In game voice and text chats, strategy
guides, high score list.
Narrative and Identity features Eg. avatar creation, storytelling features, theme and genre features
Engagement Eg. Immersion, flow, presence
The most important feature turned out to be presentation with 59,2%, followed by narrative and engagement. The least important appears to be the social feature.
Preliminary analysis
RecodingThe independent variable used for relative advantage was measured with three different items. One of these items needed to be recoded as it was a counter-indicative one, as explained at the end of the next section. Moreover, all the independent and dependent variable were standardized.
Reliability
As previously said, the variable relative advantage was measured with three different items with a 7 Likert scale. The tree items where worded as following:
“You currently have two video game console that you normally use. A new video game is going to be released next month on the market for the first time for both the consoles you own.
(eg. You own the ps4 and the nintendo switch. You have always have been a big fan of fifa and always played it on playstation, now fifa18 is coming out for the nintendo switch with some innovative features such as a new exclusive control scheme, new game play mechanics. Eg.you own the xbox one and the nintendo switch. You have always been a big fan of doom and always played it on the xbox, now doom is coming out for the nintendo switch with some innovative features such as motion aiming, multiplayer parties and updated game icon.)” Relad1 “The video game released on the other console would not offer me any new benefits
over the one for the console I usually play it on.”
Relad2 “The advantages of the new features in the video game for the new console outweight the disadvantages.”
Relad3 “This video game released on the new console will be more entertaining than the same video game on the usual console I play it on.”
In order to check the consistency of these three items, or whether some items should be deleted, a reliability test was conducted. The measure used to run this test was the Cronbach’s Alpha. For the three items of relative advantage the Cronbach’s alpha is .38, which is lower than what was expected according to the exploratory research. The three items were checked for errors; however, no error was detected. The results of the reliability analysis showed that if the item RelAd1 was deleted the overall Cronbach's Alpha would increase to .75 making the Δ>.10 if the Alpha was to be deleted. In addition, while the correlation in relation with the total score of the scale of the items RelAd2 and RelAd3 is good (above .30) the correlation of RelAd1 is <.30, meaning that it does not correlate well with the score of the scale (Fields 2013). Since deleting the first item would considerably increase the reliability, further investigations were made. Analysing the frequency of the rRelAd1 (not reversed), it seems that despite being phrased oppositely, the frequency of the answers was distributed in the same way of the two other items (meaning that people that answer e.g. strongly agree to that question also answered strongly agree to the items that were phrased oppositely). One hypothesis that seemed plausible is that the word “not” in the question was ignored. For this
reason, it was decided to delete the first item RelAd1 since this action would consistently improve the overall reliability.
Computing scale means
As a final part of the preliminary analysis a new variable was calculated. Total Relative advantage (recoded as TOTRelAd) was calculated as the mean of the items used in the questionnaire (except the deleted one).
Correlation
In table 1 are displayed the means, standard deviations and the correlations of the variables that are object of this study. Compatibility is significantly related with relative advantage with a Pearson’s r= .636, p<.01. Adoption intention is significantly related with relative advantage (r=.506, p<.01) and compatibility (r=.434, p<.01). Multihoming is rated lower on the outcome adoption intention (r=-.059) where the correlation is not as significant, since p >.01, however being it close to p<.05 it might be considered.
Another correlation test that was run is the one between the IV relative advantage and compatibility and the different genres, to see if there is a relation between these. Among the four different genres only the one called “Role play/adventure” had a significant correlation with relative advantage (r=-.216, p<.01), adoption intention (r=-.167, p<.01) and multihoming (r=.305, p<.01).
Because of the strong correlation between relative advantage and compatibility multicollinearity was investigated and specifically the variance inflation factor (VIF) was calculated (Field 2014). The VIF turned out to be 1.68 and a tolerance of 0.59. Because the VIF is below 10 and the tolerance is above 0.2 it can be concluded that there is no collinearity between compatibility and relative advantage. It can be noticed that 89% of the variance in
the regression coefficient of relative advantage and compatibility is associated with the smallest eigenvalue. This indicate dependency between the two variables, creating a potential bias of the model. It was not possible to use the principal component analysis (PCA) to reduce the collinearity, since the predictor involved are two.
Table 1: Means, Standard Deviations, Correlations Variables M SD 1 2 3 4 5 6 7 8 1. Relative Advantage 4,36 1,419 (0,75) 2. Compatibility 4,90 1,237 0,636** - 3. Adoption Intention 4,39 1,723 0,506** 0,434** - 4. Multihoming 2,32 1,182 -0,244** -0,085 -0,059 - 5.Strategy 0,41 0,492 0,103 0,077 0,062 0,099 -
6.Role playing / Adventure 0,52 0,501 -0,216** -0,106 -0,167** 0,305** -0,020 -
7.Action 0,60 0,490 -0,055 -0,050 -0,005 0,057 -0,027 0,150* -
8. Simulation 0,39 0,488 0,048 0,086 -0,061 -0,005 0,046 -0,004 -0,065 -
* Correlation is significant at the 0,05 level ** Correlation is significant at the 0,01 level
Note: Cronbach’s Alpha is displayed only for relative advantage since is the only variable with more than one item
The same correlation analysis was done for the variable characteristics with the following results: Social features have a significant correlation with relative advantage (r=.217, p<.01) and adoption intention (r=.205, p<.01). Engagement have a significant correlation with relative advantage (r=-.233, p<.01) and adoption intention (r=-.181, p<.01). None has a significant correlation with either compatibility nor multihoming.
Table 2: Means, Standard Deviations, Correlations Variables M SD 1 2 3 4 5 6 7 8 1. Relative Advantage 4,36 1,419 (0,75) 2. Compatibility 4,90 1,237 0,636** - 3. Adoption Intention 4,39 1,723 0,506** 0,434* * - 4. Multihoming 2,32 1,182 -0,244** -0,085 -0,059 - 5.Presentation 0,59 0,492 -0,030 -0,003 -0,031 -0,008 - 6.Social Features 0,31 0,465 0,217** 0,071 0,205** 0,002 -0,061 - 7.Narrative 0,55 0,498 -0,115 -0,015 -0,082 0,148* -0,138* -0,182** - 0,51 0,501
-* Correlation is significant at the 0,05 level ** Correlation is significant at the 0,01 level
Hypothesis testing
The four hypotheses were tested with a multiple regression analysis. Before running the regression, the numerical variables relative advantage, compatibility and adoption intention were standardized. Then the regression was done to test the capability of the independent variables to explain the different levels of adoption intention (See table 3).
As well as our DV and IV, the genres variable (strategy, role playing, action, simulation) and the characteristics (engagement, social features, presentation, narrative) were standardized. Firstly, the results from the regression confirmed that the model is statistically significant with an F (2, 264) =50,63, p<.001 and it explains the 27% of the variance in adoption intention. Both relative advantage and compatibility turned out to be statistically significant. The highest value of β is recorded by relative advantage (β = .387, p<.001) followed by compatibility with a β=.187, p<.01. This means that with an increase of one unit SD in relative advantage, adoption intention increases of .387 standard deviations and for one unit SD increase of compatibility, adoption intention increase of .187 standard deviations. The results of the regression confirmed both H1 and H2. This means that, as predicted, adoption intention has a positive relation with both relative advantage and compatibility.
Having analysed the relationship between the two dependent variables and the independent variable, as a second step, a moderation analysis was made to see whether these relationships interact with the moderator multihoming. Because of the presence of multicollinearity, the moderation was run separately for relative advantage and compatibility. The moderation role of multihoming in the relationship between relative advantage and adoption intention is not significant with a p=.69. The conclusion is that no moderation takes place. Therefore, hypothesis 1a should be rejected. On the other hand, for the variable compatibility, in this regression the interaction between compatibility and multihoming is significant with a p<.05 even if the variable multihoming itself it is not significant. This can help us conclude that the effect of compatibility on the adoption intention depends on multihoming. In addition, the model has an R-sq of 0.32 meaning that this solution explains the 32% of the variance of adoption intention and it is statistically significant with a p<.01. Due to this result, the interaction between relative advantage and multihioming was taken out and the regression was run once again. This made the interaction between multihoming and compatibility more significant. An additional test that was run on the variables genres and characteristics in games. Therefore, other regressions were run, in order to check whether they have an influence in the model proposed in this thesis. The variables regarding genre and characteristics were put in the regression as covariates. Since not all of the variable put in the regression were significant a variable selection was made. This process consists of running multiple time the regression and each time taking out the variable with the highest value above the significance of p<.05. The first variable taken out of the regression was “presentation” with a p=.95. The same procedure was repeated for all the covariates that each time resulted with a p<.05. At the end of the regressions run with the variable selection it turned out that a significant interaction exists with the variable genre “simulation” p<.05 and the variable characteristics “social features” (e.g. In game voice and text chats, strategy
guides, high score list) with a p<.05. The final regression confirmed the significance of the interaction. Taking a closer look to the relationship it can be said that this differs for people who don’t multihome -that have one console- (effect=.33, SE=.09, CI=.15 to .51) people who do multihome and answered to own two consoles (effect=.24, SE=.07, CI= .10 to .38) or more than three (effect=.07, SE=.09, CI=-.11 to .26). Moreover, looking at the plot showing the interaction, it can be said that since the slopes of the curve are positive in all three groups, the relationship between compatibility and adoption intention has the same trend in both people that multihome and people that do not. However, it seems that this positive effect is less strong for people who multihome than for those who own only one console.
Table 4 coefficient SE t p intercept i1 -0,1073 0,1228 -0,8735 0,3832 Compatibility c1 0,4248 0,1263 3,3637 0,0009 multihoming c2 0,0425 0,0484 -0,8784 0,3806 comp*multihoming c3 -0,0878 0,0444 -1,9787 0,0489 Graph 1
Discussion
The analysis conducted in the previous paragraph, albeit confirming what expected concerning the two main hypothesis about adoption intention (H1 and H2), gave us different results than what expected concerning the moderation role of multihoming.
The positive relationship, already investigated by previous authors (Arts et al. 2011, Davis 1989) of relative advantage and adoption intention, when talking about an innovation, was confirmed by this study. Video game users have stated that they take into account innovative features when thinking of buying a new videogame. In particular, in this research it turned out that the advantage perceived by users when thinking about adopting a new game that has an innovative feature, has a positive relationship with the adoption intention. This positive relationship is there, even when talking about the relationship between compatibility and adoption intention, despite being less strong.
The expectation was that if people use or own more than one console, the positive relationship between compatibility or relative advantage and adoption intention would be stronger. However, the results did not confirm that for both the main relationships. In fact, the regression showed that multihoming did not have a role in the relationship between relative advantage and adoption intention. This means that contrary to what had been supposed, if people have more than one console, the main relationship is not modified. No matter how many consoles people own, the effect of perceived advantage does not change in relation to adoption intention. On the other hand, when talking of compatibility multihoming has a significant role in the relationship. However, the presence of multihoming instead of making the relationship stronger, as hypothesised, makes the relationship weaker. This means that due to the fact that the user owns more consoles, he will be less demanding in relation of the consistency with gaming style when considering if adopting or not a videogame with innovative features. This could be explained with the fact that because of multihoming he is
keener in trying different things disregarding the level of compatibility with his game style. As a matter of fact, when the moderation becomes stronger, the influence of the compatibility on adoption intention becomes weaker. In addition, the higher the number of consoles owned the lower is the impact on the main relationship. This becomes clear in graph 1, where the slope of the line that represents the relationship between compatibility and adoption intention decrease when the number of consoles owned grows.
Herein it is relevant to mention the fact that some genres and characteristics also have an influence in the main relationship studied. In particular, among the four genres mentioned in the study (Simulation, Strategy, Action, Role playing/adventure) and characteristics (presentation, social feature, narrative, engagement), even though there was a significant correlation with relative advantage and compatibility of the genre role playing/adventure and the features called social features and engagement, in the end it turned out that only social features had a significant interaction. This means that when the main features considered by gamers are social ones, the effect of compatibility on adoption intentions increase. The reason could be that for videogame that have social features (such as online gaming, interaction etc.) the chance that is perceived as compatible with the game style are higher.
Conclusion
This study wanted to look deeper at the implications of multihoming, in particular when talking about adoption intention. The focus was on the videogame industry since several authors have used it as an example where multihoming occurs and it’s a canonical example of systems based on a platform (Steiner et al. 2016). However, while several authors have looked at it under a firm’s perspective (Landsman & Stremersh 2011), this research wanted to investigate the role it plays on users’ behaviour. The fact that in our research multihoming turned out not to have a role in the adoption intention is important for firms that would then
