How does pre-purchase media usage
influence online and offline spending?
The (moderating) effects of media involvement, product type,
and consumer segments.
Peter Sijtsema
s1924265How does pre-purchase media usage
influence online and offline spending?
The (moderating) effects of media involvement, product type,
and consumer segments.
Peter Sijtsema
s1924265
University of Groningen
Faculty of Economics and Business
Master Thesis MSc Business Administration
Marketing Management & Marketing Research
First supervisor: Prof. Dr. J.E. (Jaap) Wieringa Second supervisor: S.F.M. (Sander) Beckers, MSc.
Management Summary
In today’s constantly changing multimedia world, where consumers are more involved and more knowledgeable than ever, and are spending their money both online and offline, it is important to understand which media are used by today’s consumers in their pre-purchase process and which effects these media have on consumers’ online and offline spending. The goal of this research is to find out if the use of a certain type of medium during the pre-purchase process leads to a significantly different online or offline spending and under which type of product, which level of media involvement, and which type of consumer segment, online or offline spending is amplified or reduced.
Four regression models and a segmentation model are used to explore the effects of the type of medium used in the pre-purchase process on online and offline spending of consumers, and the (moderating) influences of media involvement, product type, and consumer segments.
For each increase in pre-purchase online media usage, a statistically significant increase in relative online spending is found. However, not enough evidence is found to conclude the same for offline spending, namely that a higher offline pre-purchase media usage leads to a significantly higher relative offline spending. Although positive associations between simultaneous media usage, pre-purchase online media usage, and the percentage of online spending are found, this cannot be statistically proven. Offline, the results show a lower relative offline spending for consumers with both a high pre-purchase offline media usage and a frequent simultaneous media usage. However, also in this case, not enough evidence is found to confirm or deny this finding.
For sensory products, a higher percentage of offline spending is expected, and for non-sensory products, the percentage of online spending is expected to be higher. This research statistically confirms these expectations, in that relative online spending for non-sensory products is indeed higher than for sensory products, and vice versa for relative offline spending. In terms of the moderating effect of product type on the relationship between pre-purchase media usage and spending, the expected effects are found, but the moderating effect on relative offline spending cannot be confirmed, and the effect on the percentage of online spending can only be confirmed with less certainty.
Preface
This report is my master thesis for the conclusion of my Master in Business Administration, specifically in Marketing Management and Marketing Research, at the Faculty of Economics and Business of the University of Groningen.
It was important to me to investigate a topic that was relevant today and to carry out a research that interests me. Because of my general interest in online and offline media and my pre-master thesis on the relationships between advertising medium, consumer purchase likelihood, and product type, I searched for a relevant topic within the field of consumer media usage. Although I started with a proposal to research the role of traditional and new media in each step of the consumer’s purchase process, this gradually became the report lying before you today.
Firstly, I would like to thank dr. Jenny van Doorn for her help in the early stages of my thesis. With her feedback, a specific research topic was selected and the basis for a strong theoretical foundation was formed. Secondly, many thanks to my supervisor at the University of Groningen, prof. dr. Jaap Wieringa, for his guidance, feedback, and advice throughout the writing process, and his suggestions with respect to the questionnaire and the analysis of the data. Also, thanks to all 212 respondents who took the time and effort to fill in the questionnaire.
Finally, I would like to thank my second supervisor Sander Beckers for his contribution in reading and revising the final version of my thesis.
Table of Contents
1 | Introduction ... 1
2 | Theory ... 5
2.1 | Consumer Purchase Behaviour ... 5
2.1.1 | Influencers of Purchasing Behaviour ... 5
2.2 | Online Media Usage ... 7
2.2.1 | Interpersonal Sources ... 7
2.2.2 | User-Generated Content ... 8
2.2.3 | Describing Online Purchasers ... 8
2.3 | Research Shopping ... 9
2.4 | Offline Media Usage ... 10
2.5 | Media Synergy ... 11
2.5.1 | Cross-Media Synergies ... 11
2.5.2 | Consumer Simultaneous Media Usage ... 12
2.6 | Media Involvement ... 13
2.7 | Product Type ... 14
2.7.1 | Sensory and Non-Sensory Products ... 14
2.7.2 | Search and Experience Goods ... 15
2.7.3 | Product Complexity... 15
2.8 | Consumer Heterogeneity ... 16
2.9 | Conceptual Framework ... 17
3 | Research Design ... 19
3.1 | Data Collection and Measurement ... 19
3.1.1 | Questionnaire ... 19
3.1.2 | Sample Characteristics ... 21
3.2 | Data Analysis: Regression ... 22
3.2.1 | Operationalization ... 24
3.2.2 | Specification ... 25
3.3 | Data Analysis: Segmentation ... 27
4 | Empirical Results ... 29
4.1 | Descriptive Statistics of Spending and Media Usage ... 29
4.2 | Regression Results ... 30
4.2.1 | Base Model ... 30
4.2.3 | Involvement Model ... 34
4.2.4 | Product Model ... 36
4.3 | Segmentation Results ... 38
4.3.1 | Segment Profiles ... 39
4.4 | Hypotheses Overview ... 41
5 | Conclusion and Discussion ... 43
5.1 | Managerial Implications ... 46
6 | Limitations and Future Research ... 47
References ... 49
| 1 |
1 | Introduction
Not that long ago, the entire media landscape consisted of television, radio, and printed media. In recent years however, the development of the internet led to the introduction of product review sites, social media and many more unique features which were not just impossible but unthinkable twenty five years ago. Combined with the traditional forms of media, these new media enable consumers to let the world and especially their friends and family know their opinions on products, companies, advertisements, and many more topics. Visiting shop after shop in order to research a specific product before actually making a purchase is not necessary anymore. Nowadays, consumers consult websites and social media, and are far better aware of the features, strong points and negative aspects of products. Instead of going to physical stores to make purchases, an increasing number of consumers are spending their money online.
In 2002, only two out of every ten Dutch internet users were frequent online shoppers (CBS, 2009). Eight years later, the number of online shoppers has increased (77 percent of all Dutch internet users from 12 to 74 years of age shopped online) as well as the number of frequent online shoppers (55 percent of online shoppers are frequent online shoppers) (CBS, 2011). In 2010, 86.2 percent of the Dutch population was online and spent in total almost 115 million hours on the internet over the entire year (STIR, 2011). In the same year, online consumer spending in the Netherlands increased with 11 percent over the previous year to 8.2 billion euros (Blauw Research, 2011), and is expected to increase in the coming years. Growth figures show that not only the number of online buyers, but also the number of orders, and the amount spent and the number of orders per consumer increased.
According to market research (Q&A Research & Consultancy, 2011), the majority of Dutch consumers do not have a preference for visiting (48 percent) or buying from (49 percent) online or offline shops. Slightly more consumers prefer offline shops for visiting (29 percent vs. 23 percent online) as well as buying (31 percent vs. 20 percent). The preferences are believed to still have roughly the same distribution in 2015. It is expected that consumers will buy offline as well as online.
| 2 | There are studies that have investigated consumers’ online information search with respect to
intended search and purchase channel (e.g., Burke, 2002; Verhoef, Neslin, and Vroomen, 2007), purchase intention (e.g., Rubinson, 2009), or the amount of money spent in physical or online stores (e.g., Sands, Ferraro, and Luxton, 2010; Pauwels, Leeflang, Teerling, and Huizingh, 2011). There have also been papers that studied consumers’ media usage in general (e.g., Heo and Cho, 2009) and those that identified segments of shoppers based on consumer attitudes towards various channels for search and purchase (e.g., Konuş, Verhoef, and Neslin, 2008).
However, in this changed world, where consumers are more involved and more knowledgeable than ever, and are spending their money both online and offline, the marketing department needs to understand which media are used by today’s consumers in their pre-purchase process, which effects these media have on consumers’ online and offline spending, and for which levels of media involvement, and which types of products and consumer segments, these effects are amplified or reduced. That is what this empirical research will try to contribute to existing academic knowledge by combining literature from the fields of consumer information search behaviour, consumer spending behaviour, and consumer media usage, and using the different theories and results to identify and explain the relationship between media usage and spending, and the effects of media involvement, product type, and consumer segments.
Marketing teams and managers can use the insights in consumer media usage to determine which media to use in their advertising campaigns, and which kind of information on which specific types of websites can benefit their product. Results for online and offline spending can help managers to decide for instance if opening an offline shop next to an online shop can increase revenues.
The goal of this research is twofold: first, to find out if the use of a certain type of medium during the pre-purchase process leads to a significantly different online or offline spending; second, to see under which level of media involvement, which type of product, and which type of consumer segment, online or offline spending is amplified or reduced.
The main research question is:
| 3 | Sub questions that will be answered by this research are:
1: How does a consumer’s media involvement affect the relationship between media usage and spending?
2: How does simultaneous media usage influence online and offline spending?
3: What are the differences between the effects of a sensory product and a non-sensory product on the relationship between media usage and spending?
4: Based on media usage across traditional and new media channels, how do the different segments of consumers look like?
5: What role do consumer segments play in the relationship between media usage and spending?
Four regression analyses and one segmentation analysis are performed in order to answer the questions posited above. The results show positive and significant effects of (1) online media usage on online spending, (2) online media involvement on online spending, (3) offline media involvement on offline spending, (4) product type on online and offline media, and (5) consumer segments on the relationship between media usage and spending. Positive and significant effects but with less certainty are found for (1) online media involvement on the relationship between online media usage and online spending, and for (2) product type on the relationship between online media usage and online spending.
| 5 |
2 | Theory
In this chapter, existing literature from the fields of consumer information search behaviour, consumer spending behaviour, and consumer media usage, among others, is discussed. This chapter starts with a section on consumer purchase behaviour and the factors that influence or explain this. Next, online media usage is investigated, followed by a discussion on research shopping. In the following section, literature on offline media usage is presented. Section 2.5 deals with synergy among media channels and the simultaneous media usage of consumers. The moderators are discussed in the next three sections, starting with media involvement (2.6). This is followed by product type (2.7), which presents different product type distinctions, and consumer segments (2.8). Throughout the entire chapter, hypotheses are presented, leading up to the last section, which presents the conceptual framework of this research.
2.1 | Consumer Purchase Behaviour
Consumers vary in terms of product knowledge, involvement and purchasing behaviour, but consumer research theories suggest that the steps consumers follow (unconsciously) in purchasing situations are similar. According to Kotler (1988), consumers pass through five stages to reach a purchase decision: need recognition, information search, evaluation of alternatives, purchase decision and post-purchase behaviour. However, consumers do not pass through all five stages with every purchase; they often skip some of the stages in more routine purchases. The five stages show all the considerations that arise when a consumer faces a new and complex purchase situation (Kotler, Armstrong, Saunders, and Wang, 1999). Because a purchase process starts long before the actual purchase and continues long after, marketing’s focus should lie on the entire purchase process rather than just the purchase decision. The pre-purchase process consists of the need recognition, information search, and evaluation of alternatives stages. Information search is an important part of the decision process for most consumers considering the purchase of a major durable product (Punj and Staelin, 1983).
2.1.1 | Influencers of Purchasing Behaviour
| 6 | Häubl and Trifts’ (2000) findings, which suggest that the use of interactive decision tools positively
influence consumers’ purchase decisions.
According to the results of a study by Riegner (2007), about one in ten purchases is influenced by online word-of-mouth. This is a relatively new and, with the recent increase in social media usage, an increasingly widespread phenomenon. Grant, Clarke, and Kyriazis (2007) found that consumer search behaviour is heavily affected by the manner in which consumers interact and use such (new) technologies.
In an early study on consumers’ search activity, Punj and Staelin (1983) link prior knowledge to information search. They say there are two major unique components of prior knowledge, namely specific product knowledge, which causes less external search, and general product-class knowledge, which causes more external search. Their results confirm the link between prior knowledge and external search: prior relevant knowledge is the most important factor in a consumer’s search activity, or in other words, the respondents who had the least need for information conducted the least amount of search.
Ahuja, Gupta, and Raman (2003) found that the use of the internet for research (the internet as a source of consumer information) is an important element of online purchasing behaviour, thereby confirming the findings by Bellman, Lohse, and Johnson (1999), who found that looking for product information on the internet is the most important predictor of online buying behaviour. Based on their findings, Bellman et al. (1999) state that the most important information for predicting both online and offline shopping habits are measures of past behaviour. Lohse et al. (2000) found that the factor that explained the most variation in whether a consumer would make an online purchase was the degree of internet usage to search for product information. Also, the longer the amount of time spent online, the greater the probability of making a purchase online.
| 7 | Although demographics show a slight influence on purchasing behaviour, this influence is barely
significant. This is again confirmed by Ahuja et al. (2003), who found evidence that demographics alone do not account for online shopping behaviour, and also by Kwak et al. (2002), who found age and education to be weak influencers of internet purchasing across all nine investigated product categories with the exception of computer products, although they did find significant effects of household income and gender.
A majority of consumers in Burke’s (2002) study expressed a preference for using multiple channels when shopping in four stages of the consumer’s purchase process: 82 percent preferred to use more than one channel to learn about new products, 77 percent to search for product information, 74 percent to compare and evaluate alternatives, and 63 percent to purchase and pay for products. It is therefore important to investigate both online and offline media.
2.2 | Online Media Usage
Searching for (product) information and spending money online has become increasingly common over the past few years. In the period 2005-2010, the number of internet users in the Netherlands that had bought a product online in the three months preceding the CBS survey, the so-called frequent online shoppers, rose from 36 to 55 percent of all online shoppers (CBS, 2011). In contrast, only two out of ten internet users were frequent online shoppers in 2002 (CBS, 2009). According to Kwak et al. (2002), consumers who frequently access the internet purchase more products online, and consumers who frequently search online for product information are more likely to purchase products via the internet. Likewise, Bellman et al. (1999) found that consumers who spend more time on the internet also spend more money online. Results by Verhoef et al. (2007) confirm that the internet is fairly popular for search (64 percent of the respondents), but also show that the internet is much less chosen as a purchase channel (13 percent).
2.2.1 | Interpersonal Sources
| 8 | 2.2.2 | User-Generated Content
From 2006 onwards, when Web 2.0 revolutionised how people make use of the internet, websites have reinvented themselves to transcend the format of a static publication by bringing a participatory quality to online activities (Riegner, 2007). Freely available user-generated content, collaborative tools, and networked communities created a dynamic and participatory internet culture, resulting in fundamental changes in the ability for people to communicate their ideas and opinions to many people (Riegner, 2007). User-generated content sources used by consumers are mostly consumer review and ratings sites, but also blogs, discussion boards, and social networking sites.
Approximately 33 percent of U.S. broadband internet users rate or review products online, 25 percent posts to forums, and 15 percent express their opinion online (Riegner, 2007). Zhu and Zhang (2010) suggest that the informational role of consumer reviews becomes more prominent in an environment in which alternative information is hard to find or not present. As such, less popular products may benefit from online consumer reviews. Moreover, online reviews are more influential when consumers have relatively more internet experience. However, in the case of online purchases, Cheema and Papatla (2010) found the higher a consumer’s internet experience, the lower the relative importance of online information. Nonetheless, Zhu and Zhang (2010) predict that, over time, online consumer reviews will become increasingly influential.
2.2.3 | Describing Online Purchasers
In a follow-up study on their earlier research, Lohse et al. (2000) reviewed U.S. panel data of 1997 and 1998, and found a higher income, more work hours, a longer time on the internet, and more hours per week spent online for consumers that had made an online purchase in both 1997 and 1998 compared to consumers who never made an online purchase or only in 1997 but not in 1998. Additionally, the average number of online purchases increased from 4.3 to 7.4 transactions per year, and the mean purchase increased with almost 75 percent from 49.53 to 86.31 dollars.
| 9 | 2.3 | Research Shopping
There is still a large share of consumers that search online for information regarding product features and prices but buy the product offline, the so-called ‘web-to-store’ shoppers or ‘research’ shoppers (Verhoef et al., 2007). Also, almost three quarters of the respondents did research shopping with the majority of consumer using multiple channels for their pre-purchase information search. Even in a time when the internet was not as developed as today, Wolfinbarger and Gilly (2001) noted that consumers were showing clear signs of what they called multi-modal shopping: investigating offline sources before purchasing online or looking for product or pricing information online before buying the product or service offline. They conclude that consumers value the unique attributes of each setting and that the two shopping methods complement each other.
Recent research by Q&A Research & Consultancy (2011) shows that in the coming years, the number of only-online buyers and only-offline buyers will decrease, and that the largest part of Dutch consumers (54.1 percent) will consist of cross-channel buyers. Likewise, only looking for product information online or offline and only comparing alternative products online or offline will also decrease in favour of multi-channel pre-purchase behaviour.
Sands et al. (2010) investigated the impact of internet search on consumers’ physical store purchase behaviour in terms of dollar spend. They found that online search plays a significant role in offline behaviour. Results show that the channel used in consumers’ pre-purchase search process has a significant effect on in-store purchase spend. Moreover, consumers who used the internet to search for information prior to purchase spent a significantly higher amount when purchasing in a physical store. This same effect is also found by Pauwels et al. (2011), who investigated the short-term and the long-term effects of the introduction of an informational site. They found that consumers who search more online spend more offline. Moreover, consumers with a high product need but also high price sensitivity buy more products and slightly more expensive products offline. In contrast, consumers who live close to a physical store spend less offline. Their internet visits are mostly for entertainment and serve as a (partial) substitution for taking trips to the store.
| 10 | H1: Consumers with a high pre-purchase online media usage will have a higher online
spending than consumers with a low pre-purchase online media usage.
2.4 | Offline Media Usage
Although the new media seem to provide the most opportunities for companies, Pfeiffer and Zinnbauer (2010) state that traditional advertising remains a necessity, especially in building brand strength or conveying a brand’s positioning relative to competitors toward a broad audience. The importance of offline advertising can clearly be noted through a survey by iProspect (2007) which shows that offline advertising channels clearly have an influence on online search users. A large share (67 percent) of the online search population is driven to search by offline channels (44 percent through television, 41 percent through word-of-mouth, 35 percent through newspaper and magazine advertisements, 23 percent through radio, and 13 percent through billboards). Of the online searchers who are influenced by offline channels, ultimately 39 percent make a purchase. Furthermore, general print media also help to increase buyers’ perception of durable online purchases since the media has been traditionally considered to be an effective medium for high-involvement consumers (Kwak et al., 2002). Cheema and Papatla (2010) found that offline information becomes relatively more important for consumers with high levels of internet experience.
Rubinson (2009) conducted a form of meta-analysis across seven different databases, including results from advertising-weight tests, marketing-mix modelling, copy testing, return-on-marketing analysis from quasi-experimental design, and media-planning tools, in order to assess the effectiveness of traditional media in today’s world. The results surprisingly show that television advertising has not lost any of its effectiveness over the years. For fast moving consumer goods, the effectiveness of television impressions at generating sales lift has not decreased over time. The analysis also shows that television is more effective than online and print media in generating awareness. Although television is also effective in generating familiarity and equally as effective as online media in generating purchase intent, print media appears to be the most effective at generating both familiarity and purchase intent. According to Verhoef et al. (2007), the physical store is preferred by consumers in both the search and the purchase stages with 81 percent intending to use the physical store for information search, and 84 percent for purchase.
| 11 | greater retailer and media search is associated with a higher purchase involvement. Besides,
purchase involvement, attitude towards shopping, product class knowledge and time availability are all positively associated with retailer search and media search, as well as total search effort across product categories.
A greater offline media search leads to a higher offline purchase involvement (Beatty and Smith, 1987). Also, television is more effective in generating awareness and equally as effective in generating purchase intent as online media (Rubinson, 2009). However, in the case of online purchases, offline information is more important for experienced internet users (Cheema and Papatla, 2010), which would suggest that these consumers with a high online media usage also have a high offline media usage but with a high online spending. Despite this finding, the expectations regarding offline media usage are:
H2: Consumers with a high pre-purchase offline media usage will have a higher offline spending than consumers with a low pre-purchase offline media usage.
2.5 | Media Synergy
Many researchers have long suggested that messages from multiple sources are more easily processed by and will motivate more consumers than repetitive messages (e.g., Harkins and Petty, 1987; Chang and Thorson, 2004). Presenting information in different contexts enhances the retrieval ability of consumers, which increases consumer awareness. Before the rise of the internet, media planning focused on individual media. Since then, the focus has shifted to the interaction between media, in particular between online and offline media. In his meta-analysis on cross-media synergy, Assael (2011) states that a key requirement in identifying cross-media synergies is the need to determine interactions between media at the individual consumer level. However, important synergistic effects such as the distinction between sequential and simultaneous media exposure have not been explored (Assael, 2011).
2.5.1 | Cross-Media Synergies
| 12 | Although internet advertising has provided many unique features, it has not displaced most media as
advertising sources. Tsao and Sibley (2004) measured to what extent consumers use the internet to displace or reinforce the use of other media as sources of advertising information and suggest that internet advertising serves a complementary rather than competitive role. In other words, many consumers found that internet advertising was an alternative medium complementary to their favourable attitude or frequent use of other media advertising. Similarly, Saeed et al. (2003) found evidence about the complementary effect of website value and offline advertising.
The study by Wakolbinger, Denk, and Oberecker (2009) on the effectiveness of combining online and print advertisements indicates potential advantages of cross-media advertising over single-medium advertising. This was applied to multiple media by Pergelova, Prior, and Rialp (2010), who found that firms that invested more in internet advertising had a greater relative share of broadcast advertising and achieved higher efficiency. Also between different online channels, there can be synergistic effects. In the context of social networks, word of mouth has a major impact on driving traffic to the website (Pfeiffer and Zinnbauer, 2010).
According to Tsao and Sibley (2004), there seems to be a particular group of consumers who have favourable attitudes, at least at the attitudinal level, to all advertising media. These favourable attitudes might lead to the reinforcement effects between online and offline channels. This finding was consistent with the study by Donthu and Garcia (1999), who found that internet shoppers had more positive attitudes toward advertising than non-shoppers.
2.5.2 | Consumer Simultaneous Media Usage
Zigmond and Stipp (2010) investigated the impact of television commercials on internet search queries through a series of case studies. It was found that television advertisements can stimulate consumers to take an immediate step to obtain more product information – a possible antecedent to an actual sale. Their data implies that consumers who are using television and the internet simultaneously continue to be reached by television advertisements, rejecting the speculation that simultaneous media usage could lead consumers to ignore commercial messages.
| 13 | more time with television and 65 percent more time with the internet than the average television or
internet user (Enoch and Johnson, 2010). However, although simultaneous usage is widespread, it is very limited in duration.
Since most simultaneous media users are also heavy online and offline media users (Enoch and Johnson, 2010), they continue to be reached by television advertisements (Zigmond and Stipp, 2010). These media users are exposed to and influenced by both online and offline media, and spend more time offline (television) and much more time online compared to non-simultaneous media users (Enoch and Johnson, 2010). Therefore, the hypotheses for simultaneous media usage are:
H3a: Simultaneous media usage will positively affect online spending for consumers with a high pre-purchase online media usage
H3b: Simultaneous media usage will positively affect offline spending for consumers with a high pre-purchase offline media usage.
2.6 | Media Involvement
Depending on their level of involvement, individual consumers differ in the extent of their decision process and their search for information (Laurent and Kapferer, 1985). Since different levels of involvement result in different consequences on consumer behaviour, media usage should differ when the level of involvement in media differs (Heo and Cho, 2009), which will have an effect on both online and offline spending. Kwak et al. (2002) found that internet involvement seems to have a large impact on online spending. Overall internet buying is positively affected by consumers’ involvement with the internet, and internet involvement is found to have significant positive effects on online purchasing.
It is expected that a high involvement with a certain type of medium is associated with a high usage, which will lead to a high spending via that same type of medium, leading to the following hypotheses regarding media involvement:
H4a: Consumers with a high online media involvement will have a higher online spending than consumers with a low online media involvement.
H4b: Consumers with a high offline media involvement will have a higher offline spending than consumers with a low media involvement.
| 14 | H4d: Consumers with a high pre-purchase offline media usage and a high offline media
involvement will have a higher offline spending than other consumers.
2.7 | Product Type
According to Weathers, Sharma, and Wood (2007), the intangible nature of e-commerce may increase consumer uncertainty about whether products shown online will fit their needs, i.e. if the internet is able to convey all information before purchase. It can therefore be said that a distinction between products suitable and products less suitable for online buying should be made. The existing literature makes a distinction between sensory products and non-sensory products (e.g., Pauwels et al., 2011) or search goods and experience goods (e.g., Huang et al., 2009). It is important to notice that the classification of sensory and non-sensory products is not identical to the classification of search and experience goods (Pauwels et al., 2011).
2.7.1 | Sensory and Non-Sensory Products
Pauwels et al. (2011) made a distinction between sensory products (e.g., clothes or cosmetics), and non-sensory products (e.g., electronics or DVDs) in their research. They indeed found a difference between the two categories in a higher offline consumer spending for sensory products. Riegner’s (2007) research shows the more expensive and valuable an item is, the more it is researched and the more likely it is to be influenced by user-generated content. Such higher priced and more complex products (e.g., consumer electronics) are more likely to be influenced by user-generated content because consumers devote more time researching them and look for opinions of people who already purchased the product before purchasing it themselves.
On the other hand, products that consumers need to see, feel, or try on are less likely to be influenced by user-generated content and are also more likely to be purchased offline. In general, products that are purchased online are almost twice as likely to be influenced by user-generated content as products purchased offline (Riegner, 2007). As with Pauwels et al. (2011), also here a distinction between sensory products, which are not likely to be influenced by user-generated content, and non-sensory products, which are likely to be influenced by user-generated content, can be made.
| 15 | such as with fashion products. For functional products such as computers on the other hand, brand
names are expected to be less important online because online stores can give detailed product information, as well as comparative information. In short, the importance of brand names depends on the extent of relevant information available for making choices.
2.7.2 | Search and Experience Goods
According to Huang et al. (2009), there are significant differences between search and experience goods in consumers’ perceived ability to evaluate product quality before purchase. Whereas experience goods are difficult to evaluate before purchase, search goods are relatively easy to compare. In other words, search products are characterised by attributes that can be assessed without the need for the user to experience the product directly. Experience products, on the other hand, are characterised by attributes that do need to be experienced prior to purchase (Xiao and Benbasat, 2007).
The differences between the two product categories can also be noted in the relationship between consumers’ purchase behaviour and the presence of product reviews and interactive multimedia, as mentioned earlier. Although these differences exist in traditional retail environments, they are blurred in online environments (Huang et al., 2009). Nelson (1970) predicted that recommendations of others will be used more for purchases of experience goods than search goods. A study by Chaing and Dholakia (2003) shows that product type influences consumers’ shopping intentions in such a way that the intention to buy search goods online is higher than the intention to buy experience goods.
2.7.3 | Product Complexity
In marketing research, the complexity of a product is usually defined in terms of the number of attributes used to describe the product (e.g., Keller and Staelin, 1987; Swaminathan, 2003) or the number of alternatives for the product category (e.g., Keller and Staelin, 1987; Swaminathan, 2003). The higher the number of attributes of a product and the higher the number of alternatives for the product category, the greater the product complexity. Keller and Staelin (1987) have examined the impact of product complexity on consumer decision quality and search behaviour, and developed an analytical model, which shows that a higher number of product attributes and alternatives results in a decrease in decision effectiveness.
| 16 | impact of the recommendation agent is greater in less complex product categories with a limited
number of alternatives or attributes. In other words, the interaction between product complexity and recommendation agent was significant, suggesting that the relationship between recommendation agent and amount of search is moderated by product complexity. Similarly, Xiao and Benbasat (2007) found that product complexity moderates the effect of included product attributes on users’ choice. Moreover, the inclusion effect is stronger for products with negative inter-attribute correlations, which are the more complex products, than for products with positive inter-attribute correlations, the less complex products. However, they did not find any empirical evidence to support that product type and product complexity affect the effectiveness of recommendation agents.
In summary, non-sensory products (e.g., consumer electronics) can be compared to search goods, are more likely to be influenced by user-generated content, easier to evaluate before purchase, and more likely bought online. In contrast, sensory products (e.g., clothing) can be compared to experience goods, and are less likely to be influenced by user-generated content, difficult to observe and compare before purchase, and more likely bought offline. This leads to the following hypotheses:
H5a: For non-sensory products, the online spending of consumers will be higher than the offline spending of consumers.
H5b: For sensory products, the offline spending of consumers will be higher than the online spending of consumers.
H5c: The online spending for non-sensory products will be higher for consumers with a high pre-purchase online media usage than for consumers with a low pre-purchase online media usage.
H5d: The offline spending for sensory products will be higher for consumers with a high pre-purchase offline media usage than for consumers with a low pre-purchase offline media usage.
2.8 | Consumer Heterogeneity
| 17 | those loyal to it. Chu, Chintagunta, and Cebollada (2008) found that an identical price reduction
across channels does not result in a bigger online response in purchase quantity than offline, but they do consider it to be possible. Furthermore, households exhibit lower price sensitivities when they shop online than when they shop offline.
When segmenting the respondents on the basis of their channel orientation in the information search and purchase phases of their shopping process, Konuş et al. (2008) found that there are clear differences between the different segments (uninvolved shoppers, multichannel enthusiasts, and store-focused consumers) both in general and across product categories (mortgage, insurance, computers, home electronics, clothing, holidays, and books).
In short, there are differences between consumer segments but this depends on the basis of the segmentation. A segmentation based on demographics will probably not yield the same results as a segmentation based on channel orientation. It is believed that media usage will provide a strong enough basis to reveal significant differences between the segments, which leads to the following hypothesis:
H6: There will be significant differences in online and offline spending between the different media usage segments.
2.9 | Conceptual Framework
This study investigates the use of offline and online media in the pre-purchase stages of the consumer purchase process and its relationship with online spending and offline spending among Dutch consumers. For the selection of online media channels, the approach by Chen, Clifford, and Wells (2002), who do not view the internet as a single medium but rather as a collection of media, is followed. As such, the internet will be divided into four separate areas, namely (1) social media, (2) discussion boards, forums, and blogs, (3) product review and comparison websites, and (4) company websites. Offline media in the research will include (1) television, (2) radio, (3) newspapers, and (4) magazines. According to literature, consumers’ media involvement, the type of product, and the consumer segments also have an effect on spending and might amplify or reduce the described relationships. Therefore, these three variables are included as moderators. Figure 1 gives an overview of the research.
| 18 | Although not visible in figure 1, two of the moderators are also divided into categories. Product type
will be divided into sensory product categories and non-sensory product categories, in line with Pauwels et al. (2011). The (number of) categories of consumer segments will be determined by a segmentation based on the media usage of consumers during their pre-purchase process across online and offline media channels.
Figure 1
Conceptual Framework
* Discussion boards also include forums and blogs.
** Review websites include both product review websites and product comparison websites.
a Based on Bellman et al. (1999), Donthu and Garcia (1999), Häubl and Trifts (2000), Lohse et al. (2000), Kwak et al. (2002), Ahuja et al. (2003), Grant et al. (2007), Jepsen (2007), Riegner (2007), Verhoef et al. (2007), Huang et al. (2009), Cheema and Papatla (2010), Zhu and Zhang (2010).
b Based on Kwak et al. (2002), Rubinson (2009), Cheema and Papatla (2010).
c Based on Wolfinbarger and Gilly (2001), Verhoef et al. (2007), Sands et al. (2010), Pauwels et al. (2011).
d Based on Beatty and Smith (1987), Verhoef et al. (2007), Rubinson (2009).
e Based on Enoch and Johnson (2010), Zigmond and Stipp (2010).
f Based on Saeed et al. (2003), Chang and Thorson (2004), Tsao and Sibley (2004), Xiao and Benbasat (2007), Wakolbinger et
al. (2009), Pergelova et al. (2010), Pfeiffer and Zinnbauer (2010).
g Based on Laurent and Kapferer (1985), Kwak et al. (2002), Heo and Cho (2009).
h Based on Nelson (1970), Degeratu et al. (2000), Swaminathan (2003), Riegner (2007), Weathers et al. (2007), Huang et al. (2009), Pauwels et al. (2011).
i Based on Raj (1982), Ahuja et al. (2003), Chu et al. (2008).
Moderators
Media Involvementg Product Typeh Consumer Segmenti Online Media Usage
in the Pre-Purchase Stages f Social Media Discussion Boards* Review Websites** Company Websites
| 19 |
3 | Research Design
3.1 | Data Collection and Measurement
To examine the hypotheses posited in the previous chapter, online and offline questionnaires are used. The online questionnaire is made using Qualtrics online survey software, whereas the offline paper-based questionnaire is a print-out of the online version. Each questionnaire consists of 24 questions. A convenience sample of train passengers on the Zwolle and the Groningen-Leeuwarden routes and college students attending the Rijksuniversiteit Groningen (RuG) or the Hanzehogeschool Groningen was asked to fill out a paper questionnaire. Also, friends, family, and acquaintances were asked to fill out an online version of the questionnaire, and to forward the questionnaire link to others. All samples were taken in January and February 2012 and are drawn from the same population, viz. the Dutch people.
3.1.1 | Questionnaire
In the first part of the questionnaire, several demographic questions are asked, viz. the participant’s gender, age, education level, most important daily activity, annual household income level before taxes, household size, and the number of children (under 18 years of age) in the household. Only for income level, a non-response option (do not know / do not wish to say) is included. In the second part, participants are asked to report their online media usage time in (half) hours for each of the four specific media (social media, discussion boards, forums and blogs, product review and comparison websites, and company websites). Additionally, they are asked how much of that time they spend looking for information on products or brands. The same is asked for offline media, which includes television, radio, newspapers, and magazines.
The next part of the questionnaire includes questions on simultaneous media usage. Firstly, participants are asked how often they use multiple media at the same time. Secondly, they are asked which combinations of media they use simultaneously, and which combinations of media they generally use on the same day.
| 20 | product categories are clothing (Citrin, Stem, Spangenberg, and Clark, 2003; Pauwels et al., 2011) and
liquor (Nelson, 1970), whereas the non-sensory product categories are consumer electronics (Burke, 2002; Citrin et al., 2003; Pauwels et al., 2011), and DVDs and CDs (Burke, 2002; Citrin et al., 2003; Pauwels et al., 2011). These categories all include products which are regularly bought by most consumers.
In the next part of the questionnaire, participants’ media involvement is measured. Because no single indicator of involvement can satisfactorily describe, explain or predict consumer involvement, Laurent and Kapferer (1985) recommend measuring an involvement profile, i.e. using several separate variables to measure involvement. The Personal Involvement Inventory (PII) published by Zaichkowsky (1985) was developed to capture the concept of involvement for products. A revised ten-item inventory was developed by McQuarrie and Munson (1992), and improved the usability and validity of the original inventory.
This Revised Product Involvement Inventory (RPII) is used in this study to assess online and offline media involvement. Participants are asked to rate the different media included in this study on ten items measured on a five-point Likert-type scale. Using feedback from the pre-test version of the questionnaire, this number is decreased to seven items in order to improve its accessibility. These seven items are: importance, personal meaning, excitement, coolness, interestingness, fun, and appeal. Each participant’s rating on the online media for each of the seven items is added up to an online media involvement score. Likewise, the ratings for online media are added up to form an offline media involvement score.
| 21 | The English to Dutch translation of the psychographic variables can be a reason for some of the
variety in the scale reliability per item for this research compared to Konuş et al. (2008). Deleting one or more variables from the categories did not increase scale reliabilities, except for a slight increase in motivation to conform. However, it was decided not to delete any of the variables in order to stay close to the Konuş et al. (2008) study on which the psychographic variable categories are based. According to Nunnally (1978), a Cronbach’s Alpha over 0.7 is desirable, but what a satisfactory level of reliability is depends on how a measure is being used. In this research, the psychographics are not a vital part but are mainly included to be able to describe the participants in more detail.
The questionnaire (only in Dutch) can be found in appendix 1.
3.1.2 | Sample Characteristics
A total of 217 questionnaires are collected. Of these 217 responses, five are excluded because these responses are incomplete and unusable. The final analysis sample is 212 (114 males, 98 females; M = 23.49 years of age, SD = 8.536). Unfortunately, despite measures to obtain a diverse, representative sample of the Dutch population, students are overrepresented, whereas people with lower education levels are underrepresented in the current sample. Refusals are the primary source of non-response in this survey. Sample characteristics are shown in table 1.
Table 1 Sample Characteristics (N = 212) N % M SD Demographics Gender Male 114 53.8 Female 98 46.2 Age 23.49 8.54 Level of education Primary school 1 0.5 MAVO 4 1.9 HAVO 11 5.2 VWO 8 3.8 MBO 10 4.7 HBO 104 49.1 University 74 34.9
Most important daily activity
| 22 |
Below average 19 9.0
Average 25 11.8
Above average 20 9.4
Two times average or higher 13 6.1
Do not know / Do not wish to say 57 26.9
Household size 3.11 2.02
Number of children in the household 0.48 0.85
Psychographics Innovativeness 2.57 0.69 Loyalty 3.32 0.60 Motivation to conform 3.01 0.64 Shopping enjoyment 3.27 1.04 Time pressure 3.32 0.94 Price consciousness 3.47 0.79
3.2 | Data Analysis: Regression
Several regression models on online spending were constructed in order to test all hypotheses. In total, four regression models are presented: a base model, a model for simultaneous media usage, a model for media involvement, and a model for product type. Variance inflation factors (VIF) and tolerance levels are used to detect multicollinearity.
The base model is a regression model performed on online spending in general in order to test the relationships between online media, offline media, online spending and offline spending. This model is also the basis for the other models.
In the simultaneous model, the effect of simultaneous media usage is measured with a regression analysis on online spending in general. This regression model consists of the base model extended with the simultaneous media usage variables and interactions between simultaneous media usage and total time spent looking for information on products or brands online and offline in order to test if simultaneous media usage influences consumers with a high pre-purchase online or offline media usage1.
Likewise, the involvement model is also performed on online spending in general. This model tests the effects of media involvement in itself and the effects of an interaction between media involvement and pre-purchase media usage by adding these variables to the base model in order to measure the (moderating) effects of a high or low media involvement.
| 23 | The product model is a regression model on online spending for all product categories in order to
measure the (moderating) effects of product type. Similar to the previous two models, the product model extends the base model with product type in itself as well as with interactions between product type and total time spent looking for information on products or brands online and offline.
An overview of the different regression models is given in table 2 below. In the simultaneous model and the involvement model, total time spent looking for information on products or brands online and offline is deleted from the regression because of very high correlations with the interaction variables. In the simultaneous model, these total time variables had VIF-values of 9.92 (online) and 77.6 (offline), and the interaction variables had VIF-values of 10.2 (online) and 77.1 (offline), whereas in the involvement model, VIF-values of 180.8 (online) and 124.4 (offline) were found for the total time variables, and VIF-values of 183.6 (online) and 124.5 (offline) for the interaction variables.
Because spending is measured in percentages, and online and offline spending combined has to add up to 100 percent, the regression outcomes for offline spending are exactly the same as for online spending but negative beta values are now positive and vice versa. The only differences lie in the beta-coefficient of the constant and the values of the associated standard error, T-value, 95 percent confidence interval, and significance level.
Table 2
Overview Regression Model Sets
Model 1 Model 2 Model 3 Model 4
Name Base Simultaneous Involvement Product
Dependent
variable Online spending in general Online spending in general Online spending in general Online spending for all product categories
Hypotheses H1, H2 H3a, H3b H4a, H4b, H4c, H4d H5a, H5b, H5c, H5d
Remarks Extension of base
model Extension of base model Extension of base model
Added explanatory variables Demographics, psychographics, media usage variables, and media usage time variables Simultaneous media usage, and interactions between simultaneous media usage and total time spent online/offline on products or brands Online/offline media involvement, and interactions between media involvement and total time spent online/offline on products or brands
Product type variable in which 0 = non-sensory and 1 = sensory product, and interactions between product type and total time spent online/offline on products or brands Excluded explanatory variables*
Total time spent online/offline on products or brands
Total time spent online/offline on products or brands
| 24 | PASW® Statistics 18.0 is used to run the regression models.
3.2.1 | Operationalization
Before constructing the regression models for answering the hypothesis, certain explanatory variables were tested in different ways. Initially, education was divided into eight categories, namely primary school, LTS, MAVO, HAVO, VWO, MBO, HBO, and university, with primary school functioning as the base category. Because none of the participants answered LTS as their highest form of education, this category was excluded from the study. Likewise, most important daily activity was also initially divided into eight categories, namely salaried employment, self-employment, volunteer work, housekeeping, school or studies, unemployed or disabled, retired, and other as the base category. Because none of the participants answered that they were retired, this category was excluded from the study.
Due to the very limited amounts of cases for some categories, and because several education categories were highly correlated with several daily activity categories (with VIF-values around 30 and 40, among others), it was decided to reduce the number of categories for both variables mentioned above. In effect, the number of education categories was decreased to four, with MBO, HBO, and university in the regression model, and a new combined category of primary school, MAVO, HAVO, and VWO functioning as the base category, and most important daily activity was also re-divided into four categories, with housekeeping, school or studies, and a new combined category of salaried employment, self-employment, and volunteer work in the regression, and unemployed or disabled and other functioning as the base category. Besides reducing the number of dependent variables in the regression, this also solved the multicollinearity problem.
The last of the discrete descriptive variables, income, was initially divided into minimum, below average, average, above average, and two times average or higher, with “do not know or do not wish to say” as the base category, but in order to reduce the number of dependent variables in the regression, income was re-divided into four categories: minimum and below average incomes were combined, as were above average and two times average or higher incomes.
| 25 | simultaneous media usage was divided into four usage volume categories, namely “every day”, “one
or more times per week”, “one or more times per month”, and “less often or never”, with the latter functioning as the base category.
In all models, two levels of the demographic variable most important daily activity, viz. work and school or studies, still show slightly higher VIF-values ranging from 6.1 to 7.3. A possible explanation is the large amount of cases for these two levels compared to much fewer cases for the other levels in most important daily activity. However, as a rule of thumb, multicollinearity may not be a serious issue if VIF-values do not exceed 10. Furthermore, O’Brien (2007) suggests that this rule of thumb needs to be interpreted in the context of other factors that influence the variance of the regression coefficients. It is therefore decided not to delete or combine these levels.
Because of the VIF-issues in this study and because all of the constructs are measured through the same survey, it is possible that a common method bias exists. Common method variance refers to variance that is attributable to the measurement method rather than to the constructs of interest (Bagozzi and Yi, 1991). This is a potential problem in behavioural research because it is one of the main sources of measurement error, which threatens the validity of the conclusions about the relationships between measures (Podsakoff, MacKenzie, Lee, and Podsakoff, 2003).
Harman’s single-factor test is used to detect common method variance. In this test, all variables in a study are loaded into an exploratory factor analysis and the unrotated factor solution is examined for either a single factor that emerges from the factor analysis or one general factor that accounts for the majority (≥ 50 percent) of the variance, in which case a substantial amount of common method variance is present (Podsakoff et al., 2003). Principal axis factoring is used as the extraction method.
In the results of Harman’s single-factor test, neither a single factor nor a general factor that accounts for the majority of the variance is found. The largest percentage of variance explained by a single factor is 10.12 percent. While these results do not preclude the possibility of common method bias, they do suggest that it is not of great concern and unlikely to confound the interpretations of the results of this study.
3.2.2 | Specification
| 26 | product categories is the dependent variable. The dependent variable in the base model, the
simultaneous model, and the involvement model is online spending in general.
Online Spendingi = β0 + β1GENi + β2AGEi + β3MBOi + β4HBOi + β5UNIi + β6WORKi + β7HOUSi + β8STUDi + β9BAIi + β10AIi + β11AAIi + β12HHi + β13CHILi + β14INNOi + β15LOYi + β16MTCi + β17SEi + β18TPi + β19PCi + β20SUi + β21DFBUi + β22PRCUi + β23CUi + β24TVUi + β25RUi + β26NUi + β27MUi + β28TTONi + β29TTOFi + β30PTONi + β31PTOFi + εi
Online Spendingi = β0 + β1GENi + β2AGEi + β3MBOi + β4HBOi + β5UNIi + β6WORKi + β7HOUSi + β8STUDi + β9BAIi + β10AIi + β11AAIi + β12HHi + β13CHILi + β14INNOi + β15LOYi + β16MTCi + β17SEi + β18TPi + β19PCi + β20SUi + β21DFBUi + β22PRCUi + β23CUi + β24TVUi + β25RUi + β26NUi + β27MUi + β28TTONi + β29TTOFi + β30DSMUi + β31WSMUi + β32MSMUi + β33ISPNi + β34ISPFi + εi
Online Spendingi = β0 + β1GENi + β2AGEi + β3MBOi + β4HBOi + β5UNIi + β6WORKi + β7HOUSi + β8STUDi + β9BAIi + β10AIi + β11AAIi + β12HHi + β13CHILi + β14INNOi + β15LOYi + β16MTCi + β17SEi + β18TPi + β19PCi + β20SUi + β21DFBUi + β22PRCUi + β23CUi + β24TVUi + β25RUi + β26NUi + β27MUi + β28TTONi + β29TTOFi + β30ONIi + β31OFIi + β32INPNi + β33IFPFi + εi
Online Spendingi = β0 + β1GENi + β2AGEi + β3MBOi + β4HBOi + β5UNIi + β6WORKi + β7HOUSi + β8STUDi + β9BAIi + β10AIi + β11AAIi + β12HHi + β13CHILi + β14INNOi + β15LOYi + β16MTCi + β17SEi + β18TPi + β19PCi + β20SUi + β21DFBUi + β22PRCUi + β23CUi + β24TVUi + β25RUi + β26NUi + β27MUi + β28TTONi + β29TTOFi + β30PTONi + β31PTOFi + β32TYPEi + β33ITPNi + β34ITPFi + εi
where, for respondent i:
GENi = gender (male = 0, female = 1)
AGEi = age
MBOi = dummy for MBO education (no = 0, yes = 1) HBOi = dummy for HBO education (no = 0, yes = 1) UNIi = dummy for university education (no = 0, yes = 1)
WORKi = dummy for work as the most important daily activity (no = 0, yes = 1)
HOUSi = dummy for housekeeping as the most important daily activity (no = 0, yes = 1) STUDi = dummy for school or studies as the most important daily activity (no = 0, yes = 1) BAIi = dummy for below average income (no = 0, yes = 1)
AIi = dummy for average income (no = 0, yes = 1) AAIi = dummy for above average income (no = 0, yes = 1)
HHi = household size
CHILi = number of children in the household INNOi = innovativeness score
LOYi = loyalty score
MTCi = motivation to conform score
SEi = shopping enjoyment score
TPi = time pressure score
PCi = price consciousness score
SUi = social media usage (no = 0, yes = 1)
DFBUi = discussion group, forum and blog usage (no = 0, yes = 1) PRCUi = product review and comparison website usage (no = 0, yes = 1)
(1)
(2)
(3)
| 27 | CUi = company website usage (no = 0, yes = 1)
TVUi = television usage (no = 0, yes = 1) RUi = radio usage (no = 0, yes = 1) NUi = newspaper usage (no = 0, yes = 1) MUi = magazine usage (no = 0, yes = 1)
TTONi = total time per week spent on online media TTOFi = total time per week spent on offline media
PTONi = total time per week spent looking for information on products or brands online PTOFi = total time per week spent looking for information on products or brands offline DSMUi = dummy for daily simultaneous media usage (no = 0, yes = 1)
WSMUi = dummy for simultaneous media usage one to six times per week (no = 0, yes = 1) MSMUi = dummy for simultaneous media usage one to three times per month (no = 0, yes = 1) ISPNi = interaction between simultaneous media usage and PTON
ISPFi = interaction between simultaneous media usage and PTOF ONIi = online involvement score
OFIi = offline involvement score
INPNi = interaction between online involvement score and PTON IFPFi = interaction between offline involvement score and PTOF TYPEi = product type (sensory = 0, non-sensory = 1)
ITPNi = interaction between product type and PTON ITPFi = interaction between product type and PTOF
εi = prediction error (residual)
3.3 | Data Analysis: Segmentation
The segmentation based on the media usage of consumers is used to test the moderating effects of consumer segments. In order to arrive at the segments, behavioural segmentation in terms of usage volume (Kotler, 1997) is used. In other words, how much time participants use different media is used to segment the market. This is measured as the participants’ total media usage per medium. Time spent looking for information on products or brands per medium as well as usage per medium (yes or no) were also tested as possible segmentation criteria, but did not lead to usable results.
| 28 | The variables of the base regression model are used for segmentation. Several segmentation models
in varying formations (with or without active covariates and with or without direct effects) were tested. The inclusion of the demographics or the psychographics as active covariates had little to no effect on the segments. However, including online spending as an active covariate in the model did lead to different, better results. Because there were some violations of the local independence assumption (LIA), meaning that there were some bivariate residual values substantially larger than 1, several models including direct effects to account for the residual association between some segmentation variables were tested.
Only the inclusion of a direct effect between the total amount of time spent on social media and on television (with a BVR value of 11.48) led to an improvement of the model. Furthermore, AWE values indicate that for each segmentation model, a two-segment solution is the best fit for the data, so regarding the number of segments there are little differences between the models.
The final segmentation model is a model with all eight media usage variables as indicators, online spending as an active covariate, all demographic variables and all psychographic variables as inactive covariates, and a direct effect between total time spent on social media and on television.
| 29 |
4 | Empirical Results
4.1 | Descriptive Statistics of Spending and Media Usage
Overall, online spending is quite low (M = 22.37, SD = 21.074). This is in line with Verhoef et al. (2007), who found a similar effect in that the physical store is preferred by consumers in both the search and the purchase stages. Looking at the product-specific spending, the results show a very low online spending for sensory products (M = 10.36, SD = 21.805), as expected, and a slightly higher but still quite low online spending for non-sensory products (M = 29.72, SD = 38.022).
Figure 2 shows the average usage times per medium, and figure 3 reveals the average time spent looking for information on products (or brands) for each medium. The red bars represent the online media, whereas the orange bars represent the offline media. Most notably, television is the most used medium in both categories, although it has to be noted that the distances in media usage time in figure 3 are relatively small. Also, online media are used much more for researching products, which is to be expected given the accessibility and availability of information on the internet.
Figure 2
Average Usage per Medium
Figure 3
