Aiming for the ‘golden’ spot:
Marketing effectiveness of official sponsorships during major sporting events, a
longitudinal analysis of durables, how to be effective?
By
Bart Mansier
University of Groningen
Faculty of Economics and Business
2 Preface
Firstly, I would like to thank my first supervisor Maarten Gijsenberg for sharing his knowledge during the process with a tremendously amount of precision and excellence. It was a great pleasure to be
supervised by Maarten. Maarten you are a great teacher and an excellent researcher.
Secondly, I want to thank my father Erik Mansier for all his love and support through the years.
3 Summary
4
List of figures:
P 11 Figure 1.Conceptual framework of the study.
P 29 Figure 2. Graphic representation of the marketing investments in the automobile industry. P 30 Figure 3. Total amount of sales in the automobile industry over a period of four years. P 33 Figure 4. Time varying mind-set metrics of Toyota of 48 months.
P 34 Figure 5. Major sporting events define event periods.
P 36 Figure 6. Drawing of a direct effect and an illustration of a simple mediation design. P 37 Figure 7. The Chow test formula (Chow, 1960)
P 39 Figure 8. The natural logarithm of variable (I) where I stands for one of the variables.
List of tables:
P 29 Table 1A. Marketing investments per year in millions of euros. P 31 Table 1B Sales per brand for each year in number of units. P 35 Table 2. Model specification.
P 41 Table 3. Fit of the model estimations, R-square and p-values.
P 43 Table 4. Direct effects of marketing investments on sales of durables. P 44 Table 5. Indirect effects of marketing investments on sales of durables. P 47 Table 6. Stage one and Stage two of the indirect route.
P 48 Table7. the effect of major event periods on the direct route.
P 50 Table 8. The indirect route, effects during event periods and during non-event periods. P 50 Table 9. The effect of major event periods on the direct and indirect route.
P 53 Table 10. Direct and indirect effects during official sponsorships.
5
Index
1. Introduction ... 7
2. Background information... 10
3. The direct route ... 11
3.1 Immediate effects ... 12
3.2 Delayed effects ... 13
4. The Indirect route ... 14
4.1. Immediate and delayed effects ... 15
4.1.1. Stage one ... 15
4.1.2. Stage two ... 17
5. The immediate and delayed effects of event periods ... 18
5.1. The direct route ... 19
5.2 The indirect route. ... 20
5.2.1. Stage one ... 21
5.2.2. Stage two ... 22
6. The immediate and delayed effect of an official sponsorship ... 23
6.1 The direct route ... 23
6.2 The indirect route ... 24
6.2.1. Stage one ... 25 6.2.2. Stage two ... 26 7. Methodology ... 27 7.1 Data collection ... 27 7.1.1. Marketing data ... 27 7.1.2. Sales data ... 29 7.1.3. Mind-set metrics ... 30 7.1.4. Event period ... 32 7.1.5. Sponsorship ... 32 7.2. Model building... 33 7.3 Data analysis ... 35
7.3.1. General mediation methodology ... 35
7.3.2. Data preparation ... 36
7.3.3. Model validation and robustness checks ... 39
8. Results ... 41
6 8.2. Indirect route ... 42 8.2.1. Stage one ... 43 8.2.2. Stage two ... 44 8.3. Major events ... 47 8.3.1 Direct route ... 47 8.3.2. Indirect route ... 47 8.3.2.1. Stage one ... 48 8.3.2.2. Stage two ... 48 8.4 Official sponsorships ... 51
8.4.1. The direct route ... 51
8.4.2. The indirect route ... 51
8.4.2.1. Stage one ... 52
8.4.2.2 Stage two ... 53
9. Discussion ... 55
10. Managerial implications ... 58
11. Limitations and implications for further research ... 60
12. Acknowledgements ... 61
13. Bibliography ... 62
7
1. Introduction
Watching big sport events on television is tremendously popular in The Netherlands. Especially during 2010, 2012, and 2014 big sport events generated approximately 70% of all television viewers, the percentage is based on the top 100 best viewed programs in The Netherlands.. Events, such as the winter and summer Olympics and the World and European Championship Football, are main drivers of this high percentage. With respect to these ‘big’ television years, the years 2011 and 2013 sport events still generate 30% of the top 100 best viewed programs (Stichting Kijkonderzoek, 2015).
These sport events gained a lot of media attention, which signals the importance of messages to consumers during these event periods. According to Kahneman and Tversky (1973), consumers are more likely to perceive advertising messages as more important and interesting during events. This may be one of the reasons event marketers invest heavily during event periods (International Events Group, 2013). This can be seen by the fact that 78% of all sponsorship managers invest as much in sponsorship activation as they tend to pay fees to acquire official sponsorship rights of the sport event. It is well known in marketing literature that when a brand has acquired the official sponsorship rights, extra investments are needed in order to clarify sponsorship motivations of brands towards consumers (Mazodier and Quester 2014) and so to leverage secondary associations (Keller, 1993). However, when it comes to the effectiveness of marketing investments, Gijsenberg (2014) showed that advertising effectiveness diminishes during these sport event periods, and is changing overtime. The contradictive findings above raise the following question: why are official sponsors still heavily investing during these events while they can be more effective during other periods of the year?
8 mind-set metrics (Srinivasan, Vanhuele and Pauwels 2010). Overall these soft mind-set metrics of brand equity seem to help to detect and understand the process from exposure of the brand to the purchase of a brand item by consumers (Keller And Lehmann 2006).
Recently, Hanssens, Pauwels, Srinivasan, Vanhuele and Yildirim (2014) showed that marketing investments are influencing sales of the fast moving consumer goods (FMCG) category in a direct (Transaction route) and indirect way (Mind-set route). The direct route is defined as: ‘The
direct effect of marketing investments on sales’ and the indirect route as: ‘The indirect effect of marketing investments through mind-set metrics on sales’. Whereas the indirect route consists of
two stages namely; (1) Marketing investments have an effect on mind-set metrics and; (2) Mind-set metrics have an effect on sales.
This study separates these two stages in order to provide insights into when a certain effect takes place and by which construct. This process of the direct and indirect route has not been studied often yet on a longitudinal basis related to marketing effectiveness and major event periods. Moreover, in a recent call for action by Gijsenberg (2014) it was urged that marketing effectiveness should be measured directly and indirectly through the presence of mind-set metrics for different periods for the durable category. Since the study of Gijsenberg (2014) focused on normal advertising of the FMCG market. It is argued that it would be conceptually interesting to investigate the link between official sponsorship and advertising effectiveness
9 (Lischka, Kienzler and Mellmann, 2014; Zhou and Ouyang 2003). Therefore I find it essential to distinct effects of marketing investments in immediate and delayed effects. Therefore the purpose of this study is to examine the following:
- First, this research paper aims to study the direct route of marketing investments on sales of the product category durables. The study aims to find immediate and delayed significant effects of marketing investments.
- Second, this research paper aims to study the first stage of the indirect route by studying the effects of marketing investments on the soft mind-set metrics regarding the product category durables. The study aims to find immediate and delayed significant effects of marketing investments.
- Third, this research paper studies the second stage of the indirect route by assessing the effects of soft mind-set metrics on sales of the product category durables. The study accounts for immediate and delayed effects on the relation of soft mind-set metrics on sales.
- Fourth, the study will combine the first and second path of the indirect route to find out whether the relation of marketing investments on sales is mediated by the presence of soft mind-set metrics.
- Fifth, this study aims to find out if the type of (event) period is influencing the relationship of marketing investments on sales (direct route) and if it influences the effect of marketing investments through soft mind-set metrics on sales (indirect route). Once more I accounts for the immediate and delayed effects of event periods when testing the direct and indirect route.
10
2. Background information
This part of the study explains some background information on marketing effectiveness and the impact that major sporting events and official sponsorships may have.
When referring to marketing effectiveness it is directly related to the need of marketers to be accountable, they have to be able to show how and when a certain marketing investment should take place and what the return on investment is. According to Srivastava, Tasadduq and Fahey (1998), marketing practitioners have found it difficult to measure and communicate to other executives and top management the value creation by investing in marketing actions. With all respect to the marketers, CFOs want to get ‘every bang for its buck’ (in a most efficient way) and besides they want to know which of the investments are most effective. This study describes effectiveness as the most efficient way marketing investments and all other related marketing efforts are made. Directly related, John Wanamaker once said, ‘Half my advertising is wasted, I just don't
know which half’ (Advertising Age, 1999). Nowadays, marketers need to be accountable (Rust,
Ambler, Carpenter, Kumar and Srivastava 2004). It is even cited by Stewart (2009): ‘The only question
is whether marketers will take responsibility for that accountability, or whether accountability will be imposed upon them by others’.
11
Figure 1.Conceptual framework of the study.
3. The direct route
12
3.1 Immediate effects
Years ago Assmus, Farley and Lehmann (1984) conducted the first empirical generalization study on advertising elasticity. The study represents a meta-analysis of 128 estimates of advertising elasticities from a total of 16 studies published between 1962 and 1981 and thereby providing guidelines as such for marketing effectiveness. This study indicated a significant positive immediate elasticity mean of 0.22. Besides, the authors denote that data for durable goods fit significantly less well with respect to their study (Assmus et al. 1984). In a more recent study of Sethuraman et al. (2011) a meta-analysis of 751 short-term direct-to-consumer brand advertising elasticities were estimated in 56 studies published between 1960 and 2008. The study found that the average immediate advertising elasticity is 0.12, which is substantially lower than the previous meta-analysis of Assmus et al. (1984). More specific, immediate elasticities for durables are approximately 0.29 (Sethuraman et al. 2011). The dataset of this study contains data for the automobile industry in The Netherlands, data for durables is a good bench mark. However, elasticities of car brands are positive and indicated a median elasticity of 0.07 (van Heerde, Srinivasan and Dekimpe 2010). Indicating that immediate elasticities of car brands are way lower compared to the meta-analysis of the durable category. This difference occurs most possibly through the effect of income since the acquisition of a car is way more expensive compared to other a durables, such as a television (McCollough, 2007). Besides, it was shown for durables that sales are directly affected by the economic situation (Dhawan and Jeske 2008; Deleersnyder, Dekimpe, Sarvary and Parker 2004).
As a result, these studies suggest that advertisement on consumers has a positive influence on consumers when it comes down to the durable category. However, when specifying towards the automobile industry it has been shown that for these types of products purchases are influenced by a number of factors. Considering this, it may be concluded that marketing effectiveness for the automobile industry is lower compared to other durable products.
13 advertising strategy, investing heavily in marketing efforts during contractions. It has been shown by (Dhalla, 1980; Srinivasan, Rangaswamy and Lilien 2005) that it is possible to improve advertising performance even during contraction periods. As a result of this literature review I specify the following hypothesis:
H1A: The immediate effect of marketing investments on sales is significantly positive for the product category durables.
3.2 Delayed effects
14
H1B: The delayed effect of marketing investments has a significant positive effect on sales for the product category durables.
4. The Indirect route
In this study the indirect route is defined as the effect of mind-set metrics that mediate the relationship of marketing investments on sales. The Indirect route as noted by Hanssens et al. (2014) consists of two important stages. Firstly, marketing investments are made to influence the minds and hearts of the consumers. And secondly, these changes in the minds and hearts of consumers should grow and change until they finally convert into sales. When addressing these stages of the indirect route, the theory of customer brand equity is highly applicable (Keller, 1993). Customer based brand equity is defined as the differential effect that brand knowledge has on customer response to the marketing of that brand.
The authors describe that brand knowledge is conceptualized according to an associative network memory model which consist of two components namely, (1) brand awareness and (2) brand image. Brand awareness is defined by brand recognition and recall, whereas brand image is defined as a set of brand associations indicating the quality, the uniqueness and favourability of the brand. Brand equity in this study is defined as soft mind-set metrics. Almost similar to this structure is the framework proposed by Vakratsas and Ambler (1999), who are explaining consumer behaviour by a number of steps. Firstly, advertising inputs are needed, which in the case resembles marketing efforts. Secondly, the authors explain that certain filters ‘kick in’ such as the motivation and involvement level of consumers. Thirdly, consumers are involved by cognition, affect and experience with the brand. And finally, consumer behaviour is affected by these three mind-set metrics. This final step can be seen as the second path of the indirect route.
15 total explained variance by mind-set metrics increases up to 16.3%. This underlines the importance of using mind-set metrics in market models. In a more recent study of Hanssens et al. (2014), it is shown that immediate effects of mind-set metrics are influencing four different product categories in the FMCG market. The mediating effects differ tremendously between the product categories, however mind-set metrics always indicate a significant effect with respect to marketing effectiveness. Therefore I define the following hypothesis:
H2: The relationship of marketing investments on sales is significantly positive mediated by mind-set metrics for the product category durables.
4.1. Immediate and delayed effects
4.1.1. Stage one
In stage one, the indirect route is explained as the effect of marketing investments on soft mind-set metrics. Immediate and delayed effects of marketing investments on soft mind-set metrics have indicated positive results (Hanssens et al. 2014). In order to work on a systematic basis, this study will follow the framework of brand knowledge proposed by Keller (1993). I further specify this framework by the use of several processing mechanisms proposed by Cornwell, Weeks and Roy (2005). They describe that there are three types of processing mechanisms which affect outcomes of a marketing activity. The processing mechanisms are defined as cognitive and affective (stage one) and behavioural outcomes (stage two). When focussing on the first stage; cognitive outcomes refer to brand image and brand awareness while affective outcomes refer to the liking of a brand and the preference of certain brands.
16 marketing investments are higher and thereby communicating the message, effects on soft mind-set metrics will increase due to the fact it is communicated. Besides, it is found by Esch, Langner, Schmitt and Geus (2006) that current purchases are affected mostly indirect by brand awareness and more often directly by brand image. The authors also highlight that future purchases are affected by brand knowledge and via brand relationship including brand satisfaction, brand trust and the attachment to the brand. Therefore, on the long-term (delayed effects) brand relationship factors are considered as well. Since relationship factors are expected to increase, future purchases on the long-term, it is more likely that when brand awareness is high (you need to be aware of the brand in order to start a relationship), the brand image (what do you feel about the brand) will finally decide to what extent a consumer opens up for a long-term relationship. Delayed effects of soft mind-set metrics have thereby indicated to be very effective. Probably due to the fact that consumers are more involved during the buying process of a durable good and that consumers search and process information more actively which could take consumers extra time and thereby indicate a long-term effect of the mind-set metrics (Kotler, Keller and Bliemel 2007). Delayed effects of marketing investments for the service market were found by Stahl et al. (2012). The authors found that mind-set metrics influence customer retention positively up to 23%. Thereby, the effects of mind-set metrics suggest to mediate the effects of marketing investments on sales. Using this knowledge the following hypotheses are defined with respect to the first stage of the indirect route.
H3A: The immediate effects of marketing investments have a significant positive effect on cognitive mind-set metrics for the product category durables.
H3B: The delayed effects of marketing investments have a significant positive effect on cognitive mind-set metrics for the product category durables.
H3C: The immediate effects of marketing investments have a significant positive effect on affective mind-set metrics for the product category durables.
17
4.1.2. Stage two
In stage two the indirect route is explained by the conversion rate or significant effects of soft mind-set metrics on sales. Does a change in the soft mind-mind-set metrics (i.e. cognitive or affective feelings) over time account for the sales of durables? Therefore I test the direct immediate and delayed effects of mind-set metrics (cognitive and affective) on the sales of durables.
Hanssens et al. (2014) demonstrate that in stage two mind-set metrics finally convert into sales, however this is for the FMCG product category. It seems that a conversion percentage of stage two is way lower compared to the percentage of change in the first stage of the indirect route while focussing on FMCGs. For this study I expect immediate conversion rates to be lower for the durable category compared to delayed conversion rates. One argument that supports this reasoning is that it takes time to build a customer relationship in order to create brand knowledge (Keller, 1993). Furthermore, I state that consumers need time to process information when considering and finally buying a durable product (Lischka et al. 2014). Following Hanssens et al. (2014), the theoretical foundation of the conversion originates from the attitude or planned behaviour theory. It implies that when defining the theory of planned behaviour, the intention to perform a given behaviour is a central factor (Ajzen, 1991).
18 H4a: Cognitive mind-set metrics have a significant positive immediate effect on the sales of durable
products.
H4B: Cognitive mind-set metrics have a significant positive delayed effect on sales of durable products.
H4C: Affective mind-set metrics have a significant positive immediate effect on the sales of durable products.
H4D: Affective mind-set metrics have a significant positive delayed effect on the sales of durable products.
5. The immediate and delayed effects of event periods
As noted earlier in The Netherlands sport event periods attract an enormous amount of television watchers during the years 2010, 2012 and 2014. Up to 70% of the top 100 broadcasting programs consist of sports during these years. Demonstrating that the summer and winter Olympics and the World and European football championships stand of high importance (Stichting Kijkonderzoek, 2015). Gijsenberg (2014) studied the effect of advertisement effectiveness for sport events before, during and after the sport events and compared them to non- event periods. The study indicates that brands’ advertising effectiveness strongly diminishes around major sport events, were especially short-term effects show a strong decline. With respect to the event definition of the study of Gijsenberg (2014), this study will combine the total of four periods into two periods. Namely the periods before and during will be defined as ‘during the event period’ and the periods after and non-event periods will be coded as ‘non-non-event periods’.
19
5.1. The direct route
When addressing the direct route it is important to explain the possible influence event periods can have on the relationship of marketing investments on sales. Event periods seem to have a diminishing effect on advertisement effectiveness (Gijsenberg, 2014). One of the important drivers of this diminishing effect is may be the effect of brand confusion (Poiesz and Verhallen 1989). When brands associate themselves with the event and the atmosphere they try to identify themselves with the consumers during the event period. It is expected that brand confusion will be larger due to the use of same colour, message and symbols (Keller, 1993). However, still a large number of brands advertise during the event period because of the mere exposure effects (Zajonc, 1968).
Besides, sporting events are inimitable in a sense that the sporting events are overloaded with a set of unique emotions. For example; sporting events unite people, and mediate conflicts, which was demonstrated by Nelson Mandela during the Rugby World Cup in 1995. Mandela (1995) said the following about sports: ‘It has the power to change the world, it has the power to unite
people in a way that little else does’. Exactly these types of strong emotions are providing a great
asset to build brands during these sporting events. It is clear that for this specific case the process takes weeks, months and perhaps even years to develop. Inducing that it takes time to process and transfer these strong emotions to the hearts and minds of the consumers. Besides, when several brands decide to advertise during a certain period at the same time, similar marketing investments will buy less advertising space since minutes on radio, television and on the internet are relatively more expensive. For example: an advertisement of 30 seconds during the Super Bowl costs about 2.3 million dollars in 2003, and increased up to 4.2 million dollars in 2014 (Kantar Media, 2015). The significant growth of costs during the Super Bowl Final suggests that event periods are of major importance.
20 investments mostly take place during the event and it is given that the product category is durables I formulate the following hypotheses.
H5A: The relationships of marketing investments have a significant positive effect on sales during ‘non-event’ periods.
H5B: The relationship of marketing investments on sales is lower ‘during event periods’ compared to ‘non-event’ periods.
5.2 The indirect route.
The indirect route is defined as, the mediating effect of soft mind-set metrics on the relation of marketing investments on sales while being moderated by an event period. Cornwell (2006) described these underlying effects of these mediating mind-set metrics as processing mechanisms which influence the cognitive and affective mind-set metrics. Besides, according to Speed and Thompson (2000) sponsorship is another area of marketing with source effects, store atmospheres, brand extensions and brand alliances. Whereby consumers are able to create and leverage a secondary association with the brand. It demonstrates that an official sponsorship influences people’s secondary association during an event period. However, to which costs? According to Valencia et al. (2003), well-established brands do not benefit greatly of advertisements during the Super Bowl Final of 2008. This type of advertisement is not an official sponsorship. However, according to Olson and Thjømøe (2009) TV-ads and sponsorships provide a comparable recognition in terms of value for the effects of recognition, liking and intention of TV advertising.
This study aims to find out whether a major event at itself is influencing the mind-set metrics. In a recent study, advertising effectiveness vanished during major event periods (Gijsenberg, 2013). However, Mazodier and Quester (2014) were able to show that brand affect is transferred by an official sponsorship to the consumer. Using this information I formulate the following hypothesis:
21
H6B: The effect of marketing investments through mind-set metrics is significantly larger during ‘non- event periods’ compared to ‘during event periods’ on sales.
5.2.1. Stage one
When relating the presence of event periods to the first stage of the indirect route I study the immediate and delayed effects. Addressing the first stage of the indirect effects I focus on the relationship of marketing investments related to the mind-set metrics. When focussing on stage one several processing mechanisms affect the cognitive and affective mind-set metrics (Cornwell et al. 2005). As described earlier the well-known mere exposure effect of Zajonc (1968) could lead to increased marketing investment effectiveness. Besides, messages during these event periods are gaining a lot of media attention and thereby they are likely to be perceived as more interesting and important (Kahneman and Tversky 1973). Another mechanism is the level of processing, were high involved consumers were influenced by strength of argument and low-involvement consumers by type of endorser (Petty, Cacioppo and Schuman 1983). Thereby this theory suggests that brands of durable products should use compelling arguments to buy their product instead of using some type of endorser. However, during event periods endorsers are leveraging secondary associations in order to create brand knowledge (Keller, 1993). As noted earlier, creating an immediate feeling or opinion about a certain brand is costless compared to the costly decision off buying a durable product. However, when building strong brands, Davis (2002) explains, ‘A strong brand position means having a unique, credible, sustainable, fitting, and valued place in the customers’ mind. When building brand knowledge for the long-term during a period while other brands increase marketing investments, the uniqueness of the cognitive and affective mind-sets of consumers will be of less strength compared to other unique marketing investments. Therefore this study sets the following hypothesis:
H7A: Marketing investments have a significant negative effect on cognitive mind-set metrics during an event period.
22
5.2.2. Stage two
When relating event periods to the second stage of the indirect route I focus on the immediate and delayed effects. During event periods the relationship of mind-set metrics on sales are negatively immediately influenced, most possibly in a negative manner due to several processing mechanisms and brand confusion (Cornwell, 2006; Keller, 1993). Moreover, consumers need time to process information of the advertisements when determining which car they want to buy or consider buying (Kotler, Keller and Bliemel, 2007). Therefore, I expect a negative immediate effect of major event periods on the relationship of mind-set metrics on sales.
However, delayed effects are most likely to occur during outside-event periods since the spill-over effects of marketing investments during an event period. An example: A car brand decides to invest during a certain event period; a number of steps are expected to follow. Firstly, the mind-set metrics are influenced immediately during event periods. Secondly, a consumer needs time to process a decision with respect to the high-involved product category. Finally, if a positive significant shift in the mind-set metrics occurs, it will most probably result in a delayed conversion off sales.
Positive conversion rates have been showed by Hanssens et al. (2014) for the fast moving consumer goods as an immediate effect. Where I expect a delayed positive effect with respect to event periods. In line with previous findings, Olson and Thjømøe (2009) find significant effects of sponsoring by a number of metrics namely; exposure, recognition, brand liking and purchase intention. Suggesting metrics are of importance during event periods. Besides, as noted earlier the processing mechanisms described by Cornwell (2006) are important drivers of a diverse set of metrics. With respect to stage two of the indirect route I formulate the following hypotheses:
H8A: The immediate effect of major event periods on the relationship of Cognitive mind-set metrics have significant negative immediate effect on sales during an event period.
H8B: The delayed effect of major event periods on the relationship of Cognitive mind-set metrics on sales is positive compared to non-event periods.
23
H8D: The delayed effect of major event periods on the relationship of Affective mind-set metrics on sales is positive compared to non-event periods.
6. The immediate and delayed effect of an official sponsorship
During major sporting events most of the time, an official sponsorship is represented by at least one car brand. This study knows one exception, namely the ice skating major event periods. It seems that there are three different types of brands who can be identified during sporting events. Following three studies during these events three type of groups can be defined (Cornwell et al. 2005; Gijsenberg, 2014; Mazodier, Quester and Chandon 2012). First, brands can be the official sponsor of an event. Secondly, brands try to actively stimulate the relationship with the event while not being an official sponsor (ambushers). And finally, a large group of brands try to lift on the increase of media attention and are defined as opportunists. This study focusses on official sponsors since I expect that these sponsors are able to transfer the positive feelings around the major events over to the consumers. Besides official sponsors are easier to monitor compared to the two other types of brands since they do not create official statements.
6.1 The direct route
How about marketing effectiveness? Should brands invest millions of dollars in order to get the ‘golden’ spot of an official sponsor, or should all marketing efforts be concentrated as an ambusher? Perhaps it is even better just to drop sporting events as a communication platform? According to (Helyar, 1997), up to two-thirds of the sponsors of the 1996 Olympics did not achieve their sales goals. However according to Herrmann, Walliser and Kacha (2011) a sponsorship reduces the number of main competitors when consumers make a decision between a number of brands. However, it is argued that when leveraging a sponsorship with a sales promotion more attention is drawn to the sponsorship and its commercial goals (Speed and Thompson 2000).
24 effects are of importance when consumers are affected by the direct route while it is being moderated by an official sponsor. Subsequently I define the following hypothesis:
H9A: The immediate effect of marketing investments on sales is significantly positive for the product category durables outside official sponsorship event periods.
H9B: The immediate effect of marketing investments on sales is significantly larger for official sponsorship period compared to non-official sponsorship periods.
6.2 The indirect route
However, when being an official sponsor it has been noticed by several studies that it is important to increase budgets in order to spread the word of you being an official sponsor (Gijsenberg, 2014; Mazodier and Quester 2014; McAlister, Kelly, Humphreys and Cornwell 2012). Most important reason for sponsors to increase their share of voice is that they need to emphasize the reason why they are an official sponsor. This enables the consumer to understand the sponsorship and if communicated well, the consumer will most probably adopt the send message. Besides, Sponsorship awareness levels increase over time. Walraven et al. (2014) showed that it takes two years before the metric awareness reaches a certain build-up point. For this study it is important that it has been shown that mind-set metrics are able to develop over time and that they contribute to the sales levels of the durable categories. When a major sporting event and a sponsor are matched based on image or a functional basis the process of building brand image is enhanced (Gwinner and Eaton 1999). Thereby, the mind-set metric on its own is able to transfer their value of the metric of a previous period up to the next. With respect to the literature study I specify the following hypotheses:
H10A: The effect of marketing investments through mind-set metrics on sales is significantly positive influenced during an official sponsorship period
25
6.2.1. Stage one
Most of present research indicates that whenever a company is sponsoring a big sporting event, purchase intentions increase significantly. Firstly, Madrigal (2000) noted that in line with the idea of in-group favouritism, higher levels of team identifications among attendees of sporting event appear to be positively related to intentions to purchase a sponsor’s product. Secondly, research has focused on the establishment of how sponsorship affects brand awareness, affect and persuade positive purchase intentions (Cornwall, 2008; Olson & Thjømøe, 2009; Quester & Thompson, 2001). Thirdly, in a recent study of Mazodier and Quester (2014) positive effects of sponsorship-linked marketing were shown on brand affect. It was McCracken (1989) who studied brand image transfer effects, he explains that the transfer is seen as the transfer of meaning. For example, when Max Verstappen is sponsored by RedBull his talent, youth and skills are possible secondary associations which could transfer to the brands image of RedBull. The variable official sponsorship is in nature closely related to a major event period. Therefore I emphasize that differences occur due to the fact a brand is an official sponsor and that when associated with the event images are able to transfer from the sponsored event to the consumer (Gwinner & Eaton, 1999).
Parker (1991) showed that spontaneous awareness which was generated by major sporting event and by an official sponsorship is initially low. However, delayed effects could be prevalent for months after the sponsorship. Besides, Quester and Farrelly (1998) reasoned that the impression of the sponsorship can be long-lasting once a meaningful association is developed. Furthermore, literature suggest that there were some compelling reasons for respondents to learn and remember the brands associated with the event (Quester & Farrelly, 1998). Based on the fact consumers learn from advertisement messages over time, it seems that consumers who have prior knowledge of an event are more likely to elaborate with a sponsorships message, and thereby its image transfer (Roy & Cornwell, 2004). Therefore I formulate the following hypotheses:
26
H11B: The effect of an official sponsorship has a significant positive immediate effect on the relationship of marketing investments on affective mind-set metrics.
6.2.2. Stage two
However, according to literature, it mostly depends on the goal of the brand when sponsoring an event when assessing marketing effectiveness, where mostly awareness have been a big objective of fast moving consumer goods (Cornwell et al., 2005; Valencia, Fortunato, Russell & Burns, 2003), other sponsorship goals have been set in terms of point-of –purchase sales promotions (Walraven et al., 2014). While CEOs of the automobile industry indicate that intention to purchase which is measured through the attitude of consumers is one of most important for their product category, while exposure has been seen as a lower priority (Smolianov & Aiyeku, 2009). Hanssens et al. (2014) show that mind-set metrics finally convert into sales. They indicate positive effects, since I expect that official sponsorships increase goodwill among consumers (Smith, 2004), due to the fact they are an official sponsor I formulate the following:
H12A: The immediate effect of cognitive mind-set metrics is significantly positive on the sales during an official sponsorship period.
H12B: The delayed effect of cognitive mind-set metrics is significantly positive on the sales during an official sponsorship period.
H12C: The immediate effect of affective mind-set metrics is significantly positive on the sales during an official sponsorship period.
27
7. Methodology
This study is a combination of an exploratory and a descriptive research design; I want to provide insights into several mechanisms which are accounting for the sales of durables over a longitudinal period. This section discusses the methodology of this study, consisting of two parts. Firstly, data collection methods are described in the first part. Secondly, data analysis and statistical procedures are discussed in the second part of the methodology.
7.1 Data collection
In the conceptual model a number of variables have been outlined. For each of these variables data has been gathered in the past four years. Firstly, I describe the marketing data which was obtained by the company Kien. Kien is a marketing research company that is specialized in quantitative marketing research. Secondly, sales data of the whole industry was obtained through the sales data of BOVAG. Thirdly, data for mind-set metrics were collected by Kien on a longitudinal basis. Fourth, data was collected to define and find out when certain event periods take place. Finally, during these event periods the official sponsor was defined.
7.1.1. Marketing data
28 an average of 1.3 million euros. The highest average was found for the brand Peugeot, this brand spend an average of 33.5 million euros. The highest overall measured marketing investments were in the year 2014, a total of 355.5 million euros was invested by all brands combined. The lowest level was found in the year 2013 with a total of 311.7 million euros. The total market indicated marketing investments of 1.37 billion euros over a period of four years. In figure 2 the total marketing investments are plotted, it shows that in the years 2012 and 2014 higher marketing investments are made compared to the years 2011 and 2013.
Brands 2011 2012 2013 2014 Total Average
Alfa Romeo 8.0 3.9 2.5 2.2 16.6 4.1 Audi 9.5 9.5 9.5 10.4 39.0 9.7 BMW 5.3 4.8 5.4 9.2 24.7 6.2 Chevrolet 9.7 13.4 2.8 0.0 25.8 6.5 Citroën 22.8 26.0 24.9 28.3 102.0 25.5 Dacia 1.8 6.5 6.9 10.1 25.3 6.3 Fiat 14.3 15.0 12.0 12.7 53.9 13.5 Ford 24.6 20.4 22.8 24.9 92.7 23.2 Honda 4.3 3.9 3.9 1.8 13.9 3.5 Hyundai 21.3 29.9 6.7 11.9 69.8 17.5 Kia 19.2 24.4 21.0 20.6 85.3 21.3 Mazda 2.9 4.1 1.5 3.5 12.0 3.0 Mercedes 7.9 8.5 11.0 7.4 34.8 8.7 Mini 1.6 1.1 1.2 1.5 5.4 1.3 Mitsubishi 3.2 2.8 8.8 14.9 29.7 7.4 Nissan 13.5 17.2 9.9 17.4 57.9 14.5 Opel 28.3 30.8 32.6 27.3 119.0 29.7 Peugeot 31.8 37.4 30.7 34.0 133.9 33.5 Renault 29.7 31.8 24.2 30.0 115.7 28.9 Seat 6.2 4.3 7.5 11.6 29.6 7.4 Skoda 13.2 12.7 12.4 18.5 56.7 14.2 Suzuki 8.7 7.6 6.5 3.4 26.2 6.6 Toyota 21.2 22.6 18.9 13.1 75.9 19.0 Volkswagen 20.8 20.6 18.9 25.7 86.1 21.5 Volvo 9.3 4.6 9.3 15.1 38.2 9.5 Total: 339.0 363.7 311.7 355.5 1370.0 342.5
29 300.000 350.000 400.000 450.000 500.000 550.000 600.000 2011 2012 2013 2014
Total amount of sales
Total amount of sales 250 300 350 400 2011 2012 2013 2014 Eu ro
Marketing Investments
Marketing InvestmentsFigure 2. Graphic representation of the total marketing investments in the automobile industry. 7.1.2. Sales data
Sales data was obtained by the branch organisation BOVAG. BOVAG is a Dutch platform for businesses in the mobility industry and assurances quality over the whole industry. Monthly sales data were obtained from January 2011 till January 2015. For each brand sales data was provided by the branch organisation. Data contained the total number of cars sold for a brand, the prices for all cars are not provided. Most important reason price is not used in the data is that car models change significantly in price due to different additional options and extra’s car models can have. In total the car industry sold approximately 1.8 million cars in a period of four years in The Netherlands. I find the highest numbers of car sales in the year 2011, a total of 547.211 cars were sold on the Dutch market. In the year 2014 the total amount of sold cars was the lowest by a total of 380.877 cars. The brand Volkswagen sold the most cars in a four year time period; the brand sold a total of 210.000 cars. The smallest amount of cars were sold by the brand Dacia, they sold a total of 3.546 cars. On average a car brand in The Netherlands sold a total of 18.307 cars each month.
Figure 3. Total amount of sales in the automobile industry over a period of four years.
30
Brands 2011 2012 2013 2014 Total Average
Alfa Romeo 5.565 3.648 1.824 1.633 12.670 3.168 Audi 18.614 17.266 17.796 15.658 69.334 17.334 BMW 16.006 18.657 19.035 15.499 69.197 17.299 Chevrolet 9.308 8.721 2.011 167 20.207 5.052 Citroën 29.564 26.150 19.709 19.304 94.727 23.682 Dacia 4.624 2.193 3.000 4.367 14.184 3.546 Fiat 29.398 18.953 14.271 14.653 77.275 19.319 Ford 42.413 36.322 32.220 21.091 132.046 33.012 Honda 4.906 2.974 2.047 1.528 11.455 2.864 Hyundai 23.637 21.845 11.276 8.301 65.059 16.265 Kia 23.209 27.373 23.974 17.651 92.207 23.052 Mazda 4.730 6.159 5.396 5.510 21.795 5.449 Mercedes 11.332 11.316 10.958 11.944 45.550 11.388 Mini 3.975 5.045 3.430 3.691 16.141 4.035 Mitsubishi 6.700 3.598 12.249 12.057 34.604 8.651 Nissan 12.335 9.439 6.795 11.157 39.726 9.932 Opel 41.353 35.534 23.690 23.544 124.121 31.030 Peugeot 46.272 40.400 30.421 37.345 154.438 38.610 Renault 44.431 43.883 37.204 30.559 156.077 39.019 Seat 18.590 12.658 9.341 11.019 51.608 12.902 Skoda 20.642 15.562 12.395 20.239 68.838 17.210 Suzuki 16.535 12.523 8.735 8.575 46.368 11.592 Toyota 35.485 33.744 28.786 20.477 118.492 29.623 Volkswagen 61.076 57.622 50.056 42.902 211.656 52.914 Volvo 16.511 19.975 24.444 22.006 82.936 20.734 Total: 547.211 491.560 411.063 380.877 1.830.711 457.678
Table 1B Sales per brand for each year in number of units. 7.1.3. Mind-set metrics
This study uses mind-set metrics as variable that may mediate the effect of marketing investments on sales. I use a total of five mind-set metrics which have been measured by Kien over a period of 48 months. For each month respondents were asked to fill in questionnaires, the number of respondents varied between approximately 700 up to 1.000 each month. Since I follow the brand knowledge structure of Keller (1993) I measured the cognitive mind-set metrics of brand awareness by the following constructs: Top of Mind, Familiarity, Spontaneous and Communication. I define the cognitive mind-set metrics as follows:
31 measure the effect of Top of Mind I added the total times a brand was noticed as a ‘Top of Mind’ brand and divided the total by the total number of respondents in the same time period.
- Familiarity was measured by the question whether consumers are familiar with the provided brand name. If a consumer was familiar with the brand name it was marked by a one and if not by a zero. In order to calculate the construct Familiarity I estimated the total amount of times a brand was familiar and divided it by the total respondents of the month in the same time period.
- Spontaneous was measured by asking consumers a number of brands which they name spontaneously. Each brand was dummy coded individually by a one if named and zero otherwise, the metric was measured over the first ten named brands. Once more, the total of named brands was added up and finally divided by the total number of respondents. - Communication was measured by asking respondents if they noted any advertisement of a
brand. For each brand the consumer named the brand was marked by a one and zero otherwise. The total number was added up and finally divided by the total number of respondents.
Unfortunately data on brand image is lacking in the provided dataset, I therefore focus on the awareness of the brands when referring to cognitive mind-set metrics. Besides the cognitive aspects of soft mind-set metrics affective metrics have been obtained. Defining cognitive mind-set metrics:
- Consideration was obtained by asking consumers if they consider buying that specific brand when choosing their favourable car brand. If so, the metric was dummy coded by a one and otherwise by a zero. Once more the total was added up and divided by the total amount of respondents for each month.
32 1,5 2,5 3,5 4,5 5,5 6,5 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 Par am e te r e sti m ate s
Mind-set metrics Toyota
Familiarity Communicated Consideration Top of Mind Spontaneous provides us elasticities when measuring effects. In figure 4 the varying soft mind-set metrics are showed of the randomly picked car brand Toyota.
Figure 4. Time varying mind-set metrics of Toyota of 48 months. 7.1.4. Event period
In order to define the event period it is crucial to define which sporting events are defined as an ‘event’ for this study. As noticed earlier, this study uses events on a single basis such as the Tour de France, Olympics and Football championships. An important driver of selecting events is significance of attention which is gathered. Therefore I select the Champions League final as a separate event since it receives an enormous amount of media attention each year. I use the top 100 best viewed programs in The Netherlands as a benchmark when defining the event periods which are shown in figure 5 (Stichting Kijkonderzoek, 2015). Event periods which do not receive high media attention are events that did not make it on the top 100 list that specific year. And therefore some events are left out of the analysis. Events are selected on basis of media attention, showing similarities with the selection of events of Gijsenberg (2014). Mostly event periods of major sporting events have a duration of 2 weeks, since I combine the pre- and during event period of I have monthly data and mark an event period by a one if so and otherwise by a zero.
7.1.5. Sponsorship
33 sponsor. Data is obtained by desk research in order to find which car brand was an official sponsor during the event period. Official sponsors are marked by a * in figure 5.
Figure 5. Major sporting events define event periods.* No official sponsor identified with the event.
7.2. Model building
For this study I specify three types of models. The first model tests whether the relationship of marketing investment on sales is mediated by mind-set metrics, which is specified by a simple mediation model. The second model I build is a moderated mediation model, were this study aims to find out if major event periods moderate one of the two routes (the direct and indirect route). And the final model is a pooled model for eight brands were I specify a moderated mediation model for official sponsorship periods. I pool these eight brands since they have been official sponsor of a major sporting event. The brands are: Audi, BMW, Ford, Hyundia, Kia, Škoda, Toyota and Volkswagen.
Event 2011 2012 2013 2014
Champions League Final EUFA Europa League Final
Ford (May)
Ford
(May) Ford (May) Seat (May) Tour de France
Škoda (June) World Cup Ice skating
World Championships in Athletics
(Feb)* (Feb)* (Jan-Feb)* Toyota (Aug) FIFA World Cup
Hyundai-Kia (June-July)
34 Simple mediation: 𝑀𝑏 = 𝑖1𝑟𝑏+ 𝑎1𝑟𝑏𝑋𝑏+ ∑𝑞𝑖=1,2,3𝑓𝑖=1,2,3,4+ 𝑒𝑚𝑏 𝑌𝑏 = 𝑖2𝑟𝑏+ 𝑐′1𝑏𝑋𝑏+ 𝑏𝑟𝑏𝑀𝑏+ ∑𝑞𝑖=1,2,3,4𝑔𝑖=1,2,3+𝑒𝑌𝑏 Moderated mediation: 𝑀𝑏 = 𝑖1𝑟𝑏+ 𝑎1𝑟𝑏𝑋𝑏 + 𝑎2𝑟𝑏𝑊 + 𝑎3𝑟𝑏𝑋𝑏𝑊 + ∑𝑞𝑖=1,2,3,4𝑓𝑖=1,2,3,4+ 𝑒𝑚𝑏 𝑌𝑏 = 𝑖2𝑟𝑏+ 𝑐′1𝑏𝑋𝑏+ 𝑐′2𝑏𝑊 + 𝑐′3𝑏𝑋𝑏𝑊 + 𝑏1𝑟𝑏𝑀𝑏+ 𝑏2𝑟𝑏𝑀𝑏𝑊 + 𝑀𝑏∑𝑞𝑖=1,2,3,4𝑔𝑖=1,2,3,4,+𝑒𝑌𝑏
Pooled model, Moderated mediation:
𝑀𝑝𝑜𝑜𝑙𝑒𝑑 = 𝑖1+ 𝑎𝑋𝑏 + 𝑎2𝑊 + 𝑎3𝑋𝑏𝑊 + ∑𝑞𝑖=1,2,3,4,5𝑓𝑖=1,2,3,4,5+ 𝑒𝑚𝑏
𝑌𝑝𝑜𝑜𝑙𝑒𝑑 = 𝑖2+ 𝑐′1𝑋𝑏+ 𝑐′2𝑊 + 𝑐′3𝑋𝑏𝑊 + 𝑏1𝑀𝑏+ 𝑏2𝑀𝑏𝑊 + 𝑀𝑏∑𝑞𝑖=1,2,3,4,5𝑔𝑖=1,2,3,4,5+𝑒𝑌𝑏
Model specification:
𝑀𝑏 Mediation effect of brand b (brands =1,2,3….25)
𝑀𝑝𝑜𝑜𝑙𝑒𝑑 Mediation effect of the eight pooled brands
𝑌𝑏 Effect on dependent variable Y (sales) of brand b (brands)
𝑌𝑝𝑜𝑜𝑙𝑒𝑑 Effect on dependent variable Y (sales) of the eight pooled brands
𝑖1𝑟𝑏 Regression intercept of mediator r (Mind-set metrics = Spontaneous, Top of Mind,
Communication, Familiarity and Consideration)of brand b (brands)
𝑖2𝑟𝑏 Regression intercept of mediator r (Mind-set metrics) of brand b (brands)
𝑎(1,2,3)𝑟𝑏 The regression coefficient of mediator r (Mind-set metrics) of brand b (brands)
𝑏(1,2)𝑟𝑏 The regression coefficient of mediator r (Mind-set metrics) of brand b (brands)
𝑐′(1,2,3,)𝑏 The direct effect of X on Y of brand b (brands)
𝑋𝑏 Independent variable marketing investments for brand b (brands)
∑𝑞𝑖=1,2,3,4𝑓𝑖=1,2,3,4 sum of effects for covariates q with effect coefficient f of covariates on the mediator
(mind-set metrics): ( (1) mind-set Metric (j-1), (2) Marketing Investments (j-1),(3) Sales (j-1)) and (4) (mind-set Metric j-1 * major event period) on the mediator (mind-set metric)
∑𝑞𝑖=1,2,3,4,5𝑔𝑖=1,2,3,4,5 sum of effects for covariates q with effect coefficient g of covariates on the dependent
variable (sales): ( (1) mind-set Metric (j-1), (2) Marketing Investments (j-1), (3) Sales (j-1)), (4) (mind-set Metric j-1 * major event period) and (5) (fixed effects of the 24 brands compared to the reference brand Alfa Romeo) on the dependent variable (Sales)
𝑒𝑚𝑏 Error term of mediator M (Mind-set metrics) for brand b (brands)
𝑒𝑌𝑏 Error term of dependent variable Y (sales)
W Moderator (Major event periods 1=yes 0=no)
Note: For the pooled model I used the same model specification as the moderated mediation model, however I specified effects for the total group of eight brands and I use official sponsorship periods as the W instead of a Major event period. Thereby coefficients are for the category instead of the brand level. The fixed effects of the pooled mediation model can be found in appendix A.
35
7.3 Data analysis
This study attempts to investigate when marketing investments are most efficient for the direct and the indirect route over time when considering durables. Therefore, this study uses data on a longitudinal basis in order to study immediate and delayed effects. Besides, our model needs to account for several direct and mediating effects, whereby this study is able to provide insights over time. Secondly, this study accounts for two different types of event periods, this in order to find out if marketing effectiveness increases during (1) sport events and during an (2) official sponsorship. In order to implement different types of event periods, a model accounting for a 2-way interaction is needed. The presence of a major event period and an official sponsorship will be dummy coded by a zero for outside event periods and a one during a major sport event periods and an official sponsorship for each brand.
7.3.1. General mediation methodology
According to Baron and Kenny (1986) a variable may function as a mediator when it accounts for the relation between the predictor and the criterion. Most importantly they describe that a mediator explains how a certain physical event takes place, whereas a moderator indicates when a certain event takes place. This mediation can be described as a direct effect of X on Y and secondly, X is hypothesized to exert an indirect effect on Y through M as displayed in figure 6.
Figure 6. Drawing of a direct effect and an illustration of a simple mediation design.
36 close related study by Hanssens et al. (2014) they use the Sobel mediation test in order to find mediation effects (Sobel, 1982). However, this type of analysis has been criticised along the years. According to Hayes (2009) there is one important assumption which makes the Sobel approach inappropriate. The Sobel test makes the assumption that there is a normal distribution of the sampling distribution of the indirect effect. However, sampling distribution of indirect effects tend to be asymmetric by kurtosis and nonzero skewness (Bollen & Stine 1990). One solution to this assumption is the use of bootstrapping. Bootstrapping allows researchers to take a subsample from the original sample in order to generate an empirical sampling distribution of the paths (ab), see figure 6 (Hayes, 2009). Researchers can then use the empirical (bootstrap) estimate of ab, along with the standard error of the bootstrap estimates, to compute a Z statistic or a confidence interval (Hayes, 2015).
The Sobel and bootstrapping approach has been compared recently by Mac Kinnon et al., (2002); the authors recommended using the bootstrapping method over the Sobel approach since bootstrapping increases the performance of the model and it takes out the variance. Therefore this study uses the advanced bootstrapping method (boot=5000). Due to the fact I have a limited number of data points (N=48) which represent a large time period I use a confidence interval of 90%. I therefore follow marginal significant effects (p < 0.10) as significant effects.
7.3.2. Data preparation
Since this study aims to make generalizable conclusions about the product category durables as a whole, I am willing to pool the data. However, according to Chow (1960) it is important to test equality of sets of coefficients in at least two regressions. I therefore perform a number of regressions and calculated the Chow statistic. The formula is displayed in figure 7.
37 The Chow test statistic was performed over a total of N= 1200 observations and indicated an F-statistic of (F=1.845; P<0.5) and is thereby significant since the P-value is smaller than 0.05. Thereby the Chow test indicates that it is not allowed to pool the data. Therefore estimates are made for each brand by using only one mind-set metric as a mediator due to the rule of thumb of 5 observations per parameter (Leeflang, Wieringa, Bijmolt & Pauwels 2015). The same holds for the Chow statistic when data is pooled while using a major event period as a moderator.
38 In order to measure the direct and indirect effects of marketing investments on sales, data will be transformed into the natural logarithm by the formula displayed in figure 8. This formula is built in order to create positive logarithms, therefore I multiplied the data points by one hundred and finally I add a one for each value. The natural logarithm is always one and thereby the new data is always equal or larger than zero (LN(1)≥0). Most compelling reason to use the natural logarithm of all variables is that interpretation of parameters becomes easier. After the log transformation, the estimated parameters can be interpreted as elasticities and thereby results are easier to analyse. Besides, in order to find effects of major sport events and official sponsorships data was coded with a one for events periods and official sponsorship periods for each brand, otherwise the data was coded with a zero.
Figure 8. The natural logarithm of variable (I) where I stands for one of the variables.
In order to provide generalizable insights over the product category I apply the added Z method (Rosenthal, 1991). This method allows for the combination of p-values across the 25 different car brand estimates. This method is used a total of five times since the study constructs a total of 5 models, one for each mind-set metric. For each car brand the z-score of the p-value is calculated (one-tailed standard- normal statistic). Further the Zs are summed up and I divide the sum by the square root of the number of included brands. Hereby, I create the new Z-score which is standard-normal distributed and thereby it allows to specify the associated p-values.
39
7.3.3. Model validation and robustness checks
In order to validate the model I use three robustness checks. First, I test if multicollinearity is an issue by analysing the VIF statisics of the direct route. Secondly, I measure a simple mediation model which we use as a model comparison to the more complete moderated mediation model. Thirdly, I was able to pool the data and build a moderated mediation model for car brands who are an official sponsor, results of that model will be a good benchmark.
In order to test if there is no multicollinearity, a number of logistic regression analyses are performed. Collinearity diagnostics are analysed for three randomly selected brands to find out if multicollinearity is present. Regression analysis is performed for the brands Alfa Romeo, Honda and Volkswagen and the model is specified using variables of the moderated mediation. All parameters of Alfa Romeo indicated VIF statistics between a minimum of 1.040 and a maximum of 4.297. VIF statistics of the brand Honda varied between 1.074 and 3.757. The brand Volkswagen showed VIF statistics with a minimum of 1.174 and a maximum of 10.965. The high VIF statistic of 10.965 was caused by the lagged mind-set metric Spontaneous. Although only one collinearity diagnostic indicates a value above 10 I understand this brings interpretation issues for the analysis as a whole. However, collinearity is not uncommon in mediation analysis, but it could lead to some interpretation problems. Therefore I follow the advice of Preacher and Hayes (2008) and keep constructs unique and simple instead of measuring all constructs in one model.
Secondly, I estimate VIF statistics for the pooled model due to the fact this is a different type of data set. I specified models using only one unique construct and VIF statistics indicated a minimum statistic of 1.037 and a maximum statistic of 1.632. The pooled model indicates stable and lower VIF statistics compared to the individual measured brands.
40
R R-sq Mean standard error F-statistic p-value Pooled model 0.9082 0.8248 0.2366 110.455 0.001
Simple mediation R R-sq s.e. Added Z p-value Familiarity 0.5390 0.3201 0.0003 <-5.999 0.001 Spontaneous 0.4940 0.2673 0.0003 <-5.999 0.001 Top of Mind 0.4319 0.2022 0.0003 -5.949 0.001 Communication 0.4940 0.2692 0.0003 <-5.999 0.001 Consideration 0.4687 0.2386 0.0003 <-5.999 0.001
Moderated mediation R R-sq s.e. Added Z p-value Familiarity 0.5900 0.3699 0.0025 <-5.999 0.001 Spontaneous 0.5731 0.3440 0.0822 <-5.999 0.001 Top of Mind 0,5098 0,2733 0,0066 <-5.999 0.001 Communication 0,5902 0,3668 0,0044 <-5.999 0.001 Consideration 0,5503 0,3191 0,0178 <-5.999 0.001
Table 3. Fit of the model estimations, R-square and p-values.
41
8. Results
In this section the results will be analysed of the three models which have been created. First, I estimate a simple mediation model, which is the benchmark model. Secondly, I estimate a moderated mediation model which I specified to test the hypotheses (H1-H8). Thirdly, I estimate a pooled moderated mediation model testing for (H9-H12).
8.1. Direct route
Results are generated for the 25 car brands where I estimate five types of models using one of the five different mind-set metrics each time. This results in a total of five unique models for 25 brands. Since this study wants to provide a general overview of the results the added Z method was used (Rosenthal, 1991). However, due to the fact i estimate the direct route five times for each brand the study estimated five models and I strive for a parsimonious model. I estimate an average beta, standard error and p-value. I therefore sum up all five values of the different models for one brand and divide them by the total number of estimated models, which is five. I replicate this process for all 25 brands and finally I perform the added Z method. In table 4 I show the results of the direct route. This table provides immediate direct effects and delayed direct effects. With respect to the first hypothesis I am interested in the following: H1A: The immediate effect of marketing investments on
sales is significantly positive for the product category durables. I am able to find positive significant
results (β=0.022 p<0.01). H1A is supported and I may infer that marketing investments have a positive effect on sales for durables.
With respect to the delayed effects I formalized the following hypotheses: H1B: The delayed
effect of marketing investments has a significant positive effect on sales for the product category durables. The added Z was insignificant (β=-0.0145 p=0.321) and I therefore check whether individual
42 for a total of 4 brands with respect to the metric models of Familiarity (β= -0.0892 p<0.10). I do find significant results, however these tend to be negative instead of positive which is in contrast to what was expected. Therefore I reject the hypothesis and infer that the delayed direct effects of marketing investments on sales tend to have a negative effect.
Table 4. Direct effects of marketing investments on sales of durables.
8.2. Indirect route
The following section describes results of the indirect route. The study measures whether a mind-set metric mediates the effect of marketing investments on sales. By the use of the mediation analysis confidence intervals are obtained instead of p-values, with a bootstrap of 5000.
Results are shown in table 5. First, according to this study the metric Top of Mind is not mediating the indirect route through stage one and stage two. Second, the metric Familiarity indicated one positive significant effect (β=0.07 P<0.10). Third, the metric Spontaneous indicated two significant positive effects (β=0.02 P<0.10) as a mediator. Fourth, for the metric models Communication results are most promising. A total of nine brands indicate a positive mediating effect on the relationship of marketing investments on sales (β=0.031 P<0.10). Finally, the models using the metric Consideration
Direct route
Immediate effects
(Marketing investments on sales)
Coefficient Standard error Number of Brands P-value
Combined metrics 0.0216 0.0343 25 0.013
Delayed effects
(Marketing investments (j-1) on sales)
Metric models Coefficient Number of Brands P-value
43 result in a total of four significant mediation effects with an average effect of (β=0.02 P<0.10). When addressing H2: The relationship of marketing investments on sales is significantly positive mediated
by mind-set metrics for the product category durables. The results partially support H2, since a
number of brands indicate significant positive effects.
Indirect route
Effect of marketing investments on sales.
Mediation effects
Metric models Number of brands Coefficient
Familiarity 1 0.07
Communication 9 0.03
Spontaneous 2 0.02
Top of Mind 0 0.00
Consideration 4 0.02
Table 5. Indirect effects of marketing investments on sales of durables. 8.2.1. Stage one
By the use of the mediation analysis of Hayes (2008) I am able to measure the effects of marketing investments on mind-set metrics (stage one). I report results in table 6 for H3 A, B, C and D in stage one and for H4 A, B, C and D in stage two. Once more estimates are calculated by the added Z method, if results of the added Z method are significant the numbers of brands are reported by 25. If not, the number of brands with significant effects are reported.
When addressing H3A: The immediate effects of marketing investments have a significant positive
effect on cognitive mind-set metrics for the product category durables.The hypothesis is supported
for the cognitive metrics Familiarity (β=0.001 p<0.10), Communication (β=0.024 p<0.10) and Spontaneous (β=0.003 p<0.10). However, I reject the hypothesis for the mind-set metric Top of Mind since the effect is negative. Therefore I conclude that marketing investments are influencing the cognitive mind-set metrics Familiarity, Communication and Spontaneous positively, whereas the mind-set metric Top of Mind are negatively influenced by marketing investments.
With respect to H3B: The delayed effects of marketing investments have a significant positive effect
