New & Existing Brand

Journal of Marketing Communications, Vol. 11, No. 2, 113–128, June 2005

Advertising for New and Existing Brands: The Impact of Media Context and Type of Advertisement WIM JANSSENS & PATRICK DE PELSMACKER Universiteit Antwerpen, Faculteit Toegepaste Economische Wetenschappen, Antwerp, Belgium

ABSTRACT The effects of advertisement and context type on the responses to advertisements for different brands of new and existing products were tested. In the first experiment (243 graduate students) a positive emotional advertisement and a non-emotional advertisement for a well-known and a new brand of printer were tested in a positive emotional context and a nonemotional media context. In the second experiment (206 graduate students) positive emotional and non-emotional advertisements for new brands of watches and healthy drinks were tested in an emotional and a non-emotional context. The type of context moderated the responses to advertisements for the well-known and new products: a positive emotional context led to a more positive attitude towards the advertisement and the brand and purchase intention for the wellknown brand than for the new brand. A non-emotional context led to more positive responses for the new brand than for the well-known brand. In general, emotional advertisements led to more positive affective reactions and non-emotional advertisements led to more positive cognitive reactions. However, the type of advertisement did not have a moderating effect on the responses to advertising for the new or well-known brands or different product types. The studies illustrated the relevance of media context for advertising new versus existing products. KEY WORDS: Advertising effectiveness, media context effects, advertisement type effects, new brands

Introduction The effects of the characteristics of advertising stimuli on responses to advertisements are well documented. More particularly, the impact of (positive) emotional and nonemotional (rational) appeals has been extensively studied (Weinberger and Gulas, 1992; LaTour and Henthorne, 1994; De Pelsmacker and Geuens, 1996; De Pelsmacker et al., 2002). Advertisements can be classified according to their main focus (cognition or affect) (Swaminathan et al., 1996). In this respect, emotional advertisements can be defined as advertisements serving mainly to elicit affective Correspondence Address: Patrick De Pelsmacker, Universiteit Antwerpen, Faculteit Toegepaste Economische Wetenschappen, Prinsstraat 13, 2000 Antwerpen, Belgium; Email: patrick.depelsmacker@ ua.ac.be 1352-7266 Print/1466-4445 Online/05/020113–16 # 2005 Taylor & Francis Group Ltd DOI: 10.1080/1352726042000306847

114

W. Janssens & P. De Pelsmacker

responses (Aaker and Stayman, 1992). Most emotional advertisements contain positive emotional appeals. Non-emotional or rational messages, on the other hand, can be defined as containing features, practical details and verifiable, factually relevant cues that can serve as evaluative criteria (Puto and Wells, 1984; Belch and Belch, 1998). Most studies have indicated that positive emotional appeals lead to more positive advertisement and brand responses in general, although cognitive reactions to advertisements are more positively affected by non-emotional advertising appeals than by emotional ones (De Pelsmacker et al., 1998). Media context can be defined as the characteristics of the content of the medium in which an advertisement is inserted (e.g. among articles in a magazine) as the individuals who are exposed to it perceive them. Although some studies have not found any effect of context on advertising effectiveness (e.g. Derks and Arora, 1993), Perry et al. (1997) found that some types of context may be more supportive for some advertisements than others. This is attributed to the priming principle (Herr, 1989; Yi, 1990, 1993; Lee and Sternthal, 1999; Lynch and Stipp, 1999): a specific context can serve as a primer to make consumers more susceptible to a certain advertisement, as a result of which the advertisement is processed more intensively. Advertisements that are relevant for or congruent with the mood of a subject at that moment may be more easily accessed and processed. This effect is called the mood congruency–accessibility hypothesis (Goldberg and Gorn, 1987; Perry et al., 1997). The types of advertisements and media contexts can make certain needs more salient and can also stimulate the motivation to pay attention to the product in the advertisement (Petty and Cacioppo, 1986; MacInnis and Jaworski, 1989). The link between type of advertisement and product type has been studied in some studies (Weinberger and Campbell, 1991; Alden and Hoyer, 1993). For instance, Chattopadhyay and Basu (1990) and Weinberger and Gulas (1992) concluded that the use of humorous appeals was more effective for existing products and/or when there was already a favourable attitude towards the brand. However, overall only a limited number of studies have focused upon the interaction effects between product types on the one hand and advertisement and context types on the other. The objective of this study was to explore the impact of the type of advertisement (positive emotional versus non-emotional) and context type (emotional versus nonemotional) on the responses of consumers to advertisements. Two experiments were conducted. In the first one advertisements for a well-known and a new brand of printer were compared. In the second experiment advertisements for a new brand of watch and a new healthy drink brand were studied. More particularly, the interaction effects between type of advertisement and context type on the one hand and newness of the brand on the other were investigated in the first experiment. The interaction between advertisement and context types on the one hand and the type of new products (food and durable) on the other hand were studied in the second experiment. Research Questions and Hypotheses Experiment 1 In the first experiment positive emotional and non-emotional advertisements for a new and an existing brand of printer were shown in positive emotional and

Advertising for New and Existing Brands

115

non-emotional media contexts. As far as the main effects of advertisement and context type and newness of the brand are concerned, it could be expected that advertising an existing and well-known brand would lead to a more favourable attitude towards the advertisement and the brand and a more positive purchase intention than advertising a new brand. Furthermore, positive emotional advertising in general could be expected to lead to more positive affective responses and less positive cognitive responses than non-emotional advertisements (De Pelsmacker and Geuens, 1996; De Pelsmacker et al., 1998). As far as media context is concerned, the expected effect is unclear. On the one hand, a positive emotional context could be expected to lead to more positive responses (Perry et al., 1997; De Pelsmacker et al., 2002). On the other hand, a printer is a relatively ‘non-emotional’ product. As a result, the similarity between a non-emotional context and a non-emotional product might lead to more positive responses (Yi, 1990, 1993; Perry et al., 1997; De Pelsmacker et al., 2002). Possibly the two context effects could counterbalance one another. This leads to the following hypotheses. H1: Advertisements for a well-known brand lead to more positive attitudes towards the advertisement and the brand and to more positive purchase intentions than advertisements for a new brand. H2: Positive emotional advertising stimuli lead to more positive feelings towards the advertisement than non-emotional stimuli. Non-emotional stimuli lead to more positive cognitive responses towards the advertisement than positive emotional stimuli. H3: The attitude towards advertisements and brands and the purchase intention do not differ between positive emotional and non-emotional media contexts. Chattopadhyay and Basu (1990) and Weinberger and Gulas (1992) found that a particular type of emotional appeal (humour) resulted in more favourable responses in the case of existing products, while this was not the case for new brands. Indeed, it can be assumed that consumers are more interested in rational, objective product information when they do not know the product. In the case of an existing and wellknown brand, a positive emotional message might be more effective in supporting the brand. The consumer is not so much interested in learning about the brand and the involvement with the brand is less explicit. In these circumstances, inducing a good feeling by means of positive emotional appeals might lead to more positive responses (Alden and Hoyer, 1993; Lee and Sternthal, 1999; Lynch and Stipp, 1999). Similarly, a non-emotional media context could serve as a primer to make consumers more motivated to pay attention to information about a new brand, as a result of which an advertisement for this new brand could be processed more intensively. On the contrary, a positive emotional context might serve as a more effective peripheral cue in the case of an existing brand (Petty and Cacioppo, 1986; MacInnis and Jaworski, 1989; Yi, 1990, 1993). This leads to the following hypotheses.

116

W. Janssens & P. De Pelsmacker

H4: A positive emotional advertising appeal leads to a more positive attitude towards the advertisement and the brand and a more positive purchase intention for an existing brand than for a new brand. A non-emotional appeal leads to a more positive attitude towards the advertisement and the brand and a more positive purchase intention for a new brand than for an existing brand. H5: A positive emotional media context leads to a more positive attitude towards the advertisement and the brand and a more positive purchase intention for an existing brand than for a new brand. A non-emotional media context leads to a more positive attitude towards the advertisement and the brand and a more positive purchase intention for a new brand than for an existing brand.

Experiment 2 In the second experiment emotional and non-emotional advertisements for a new brand of watch and a new brand of healthy drink were shown in emotional and nonemotional media contexts. In this experiment only new brands in different product categories were studied. As in the first experiment, positive emotional advertising could be expected to lead to more positive affective responses and less positive cognitive responses than non-emotional advertisements (De Pelsmacker and Geuens, 1996; De Pelsmacker et al., 1998). Hence, the following hypothesis can be formulated. H6: Positive emotional advertising stimuli lead to more positive feelings towards the advertisement than non-emotional stimuli. Non-emotional stimuli lead to more positive cognitive responses towards the advertisement than positive emotional stimuli. In this experiment both the watch and the healthy drink were new brands. As in experiment 1, it could be expected that a non-emotional context would be more supportive for the new brand advertisements than an emotional context (Yi, 1990, 1993; Perry et al., 1997; De Pelsmacker et al., 2002). However, since the luxurious watch tested here was more of a status symbol than the drink, it could be expected that an emotional context would also be supportive for the watch advertisement (Perry et al., 1997; De Pelsmacker et al., 2002), in which case the two effects could counterbalance each other. This leads to hypothesis 7. H7: The attitude towards advertisements for a healthy drink, the attitude towards the healthy drink brand and the purchase intention are more positive in non-emotional media contexts than in emotional media contexts. The attitude towards advertisements for a watch, the attitude towards the watch brand and the purchase intention are not influenced by the media context.

Advertising for New and Existing Brands

117

Experiment 1: New and Existing Brand of Printers Research Design In the first experiment a 26262 full factorial between-subjects design was used in which the type of advertisement, context type and brand type (new and existing) were manipulated as independent variables. Brand type. The product category studied was a printer. This type of product is sufficiently known to the participants studied. The distinction was made between an existing brand that the study confirmed to be universally recognised (HewlettPackard) and a non-existing (new) brand dubbed PowerPrint. Type of advertisement. A positive emotional advertisement and a non-emotional print advertisement were developed for both the Hewlett-Packard and PowerPrint printers. Apart from the product name and the logo on the printer, the advertisements were exactly the same for the Hewlett-Packard and PowerPrint printers. The non-emotional advertisement listed a number of objective characteristics of the printer. The headline focused on one of them. The emotional advertisement showed a printed image of a mother holding her baby and the headline focused upon ‘happy memories’. Media context. Two types of mock magazine contexts were developed. The nonemotional context contained financial and business facts and figures. The positive emotional context consisted of two articles on relationships and love. Eight double page magazine pages were composed, consisting of 75% context and 25% advertising space. Each page was a different combination of a type of context and a type of advertisement for the Hewlett-Packard and PowerPrint printers. In all cases the advertisement was positioned at the top of the right page. Dependent variables. Apart from a number of manipulation check items (six items for the type of advertisement and six items for context type), the attitude towards the advertisement (17 items) and the brand (five items) and purchase intention (six items) were measured, using seven-category Likert-type scales (De Pelsmacker et al., 2002). Participants. The study was based on a sample of 243 graduate students in business administration that was randomly assigned to the eight experimental conditions to form eight groups of approximately equal size. They were told that they were participating in a test of a new magazine and were allowed to look at the pages for 30 seconds. The pages were then collected, after which they were asked to complete a questionnaire containing the various scales. Experiment 2: New Brands of Watches and Healthy Drinks Research Design In the second experiment for each of two products (a healthy soya drink and watch) a 262 full factorial between-subjects design was used in which the type of

118

W. Janssens & P. De Pelsmacker

advertisement (positive emotional and non-emotional) and the context type (emotional and non-emotional front page of a newspaper) were manipulated as independent variables. Product type. The two products were a food product (soya drink) and a non-food product (watch), respectively. Both products were new (non-existing brand names at the time of the research). The watch tested in this experiment was assumed to be an emotionally oriented product type (a status symbol). Type of advertisement. A positive emotional advertisement and a non-emotional print advertisement were developed for both products. The emotional advertisement for the soya drink (‘Provamel’) showed an image of a mother with a young child and was headlined ‘because I love him’. The non-emotional advertisement showed an image of a glass in which soya milk was poured and was headlined ‘100% natural, 0% fat, 0% cholesterol’. The emotional advertisement for the watch (‘Berlucci’) showed a father with his daughter on the beach with the headline ‘Where has time gone?’ The non-emotional advertisement for the watch was shown on a white background, with a slogan referring to its qualitative and technical aspects. Media context. Two types of newspaper front pages were developed. Existing newspapers were used and their typical layout was taken into account in developing the manipulated front pages. One of the two newspapers is known for its more emotional approach to the news (in the experiment this was manipulated by means of articles with headlines such as ‘Chaos and tears’ and ‘Father heard how daughter was hit by car’ and was also manipulated by emotional photographs of the ones involved). The other newspaper is known for its more rational approach (this was manipulated by articles with headlines such as ‘Unemployment rises with 17 000’ and ‘Sabena on the ground’ and was also manipulated by objective photographs). Each front page consisted of 92% context and 8% advertising space. Each page was a different combination of a type of context and a type of advertisement for the soya drink and the watch. In all cases the advertisement was positioned at the bottom right of the front page. Dependent variables. Apart from a number of manipulation check items (ten items to measure involvement, six items for the type of advertisement and six items for context type), as in experiment 1, the attitude towards the advertisement (17 items) and the brand (five items) and purchase intention (six items) were measured using seven-category Likert-type scales (De Pelsmacker et al., 2002). Participants. The study involving the soya drink was based on a sample of 90 graduate students in business administration randomly assigned to the four experimental conditions to form four groups of approximately equal size. The same procedure was applied for the watch (sample size of 116). All the participants were told that they were participating in a test of evaluating front pages of known newspapers and were allowed to look at this front page for 30 seconds. After the collection of these pages they were asked to complete a questionnaire containing the various scales.

Advertising for New and Existing Brands

119

Results Experiment 1 The two scales used for checking the manipulation were factor analysed. In both cases all items loaded on one factor. The Cronbach’s a value for the advertisement type manipulation check was 0.96 and that for the context type manipulation check was 0.95. In checking the manipulation the mean scores of the two scales were used. The participants did indeed find the emotional advertisements more emotional than the non-emotional advertisements (mean score54.95 versus 2.10, t519.17 and p,0.001). Similarly, the emotional content was considered as more emotional than the non-emotional one (mean score55.02 versus 2.09, t522.45 and p,0.001). A principal components analysis performed on the 17 items measuring the attitude towards the advertisement yielded three underlying dimensions: feeling, knowing and understanding. This was in line with earlier results (De Pelsmacker et al., 2002). The attitude towards the advertisement items with a factor loading of more than 0.6 on the corresponding factor and less than 0.4 on any other factor were brought into a structural equation model in order to perform a confirmatory factor analysis with three latent variables using AMOSTM (SPSS). The analysis led to the omission of a number of items, after which the model showed acceptable results. Although the Bollen–Stine bootstrapped p-value was lower than the cut-off value of 0.05, the corrected x2 index was 2.54 (CFI, Comparative Fit Index50.94, TLI, Tuker Lewis Index50.92, and RMSEA, Root Mean Square Error of Approximation50.08), thereby supporting the unidimensionality of the constructs. All factor regression coefficients were significant and higher than 0.50. These results lend support to the convergent validity of the constructs (Hildebrandt, 1987; Steenkamp and Van Trijp, 1991). The reliability of the constructs was examined by means of three indicators. The composite construct reliabilities should have exceeded 0.70, the variances extracted should have been at least 0.50 and Cronbach’s a values should have exceeded 0.70 (Steenkamp and Van Trijp, 1991). As can be seen in Table 1, all the constructs met these criteria. Finally, the highest correlation coefficient between the constructs was 0.38, indicating discriminant validity. A confirmatory factor analysis was also performed on the items measuring the attitude towards the brand and purchase intention with each latent variable. The Bollen–Stine bootstrapped p-value was 0.38 and the corrected x2 index was 1.51 (CFI50.99, TLI50.99 and RMSEA50.046), thereby confirming the unidimensionality of the attitude towards the brand construct. A confirmatory factor analysis was performed for the purchase intention three-item construct, but no model fit measures could be obtained because of a lack of degrees of freedom. However, all the factor regression coefficients were significant and higher than 0.50, thereby supporting convergent validity. The three reliability indicators of the attitude towards the brand construct as well as the purchase intention construct showed good results and all the cut-off criteria were met (Table 1). The mean scores of the remaining items for each of the five constructs were used in the subsequent analysis (see Table 1). Multivariate analysis of variance revealed that the main effects of the type of advertisement and brand type were significant. More importantly, there was also a significant interaction effect between the brand and context type. The more detailed

120

W. Janssens & P. De Pelsmacker

Table 1. Reliability of the attitude toward the advertisement, attitude towards the brand and purchase intention scales – experiment 1 Attitude towards the advertisement feeling(construct reliability 0.87, variance extracted 0.62 and Cronbach’s a 0.87) This advertisement is beautiful This advertisement attracts attention This advertisement is remarkable This advertisement is original Attitude towards the advertisement knowing(construct reliability 0.83, variance extracted 0.55 and Cronbach’s a 0.82) This advertisement gives useful information This advertisement is believable This advertisement tells me something new This advertisement fits with the brand Attitude towards the advertisement understanding(construct reliability 0.82, variance extracted 0.70 and Cronbach’s a 0.75) It is not immediately clear which brand is advertised (2) You have to watch the advertisement frequently to know what it is exactly about (2) This advertisement is confusing (2) Attitude towards the brand(construct reliability 0.95, variance extracted 0.73 and Cronbach’s a 0.94) I like the brand I would recommend this brand to others I have a positive attitude towards this brand This brand is really something for me I would never use this brand (2) Purchase intention(construct reliability 0.9, variance extracted 0.91 and Cronbach’s a 0.90) Next time that I need such a product, I will choose the brand in the advertisement It is very likely that I will buy the advertised brand If I saw this brand in a shop, I would buy it

univariate effects are shown in Table 2, while the main effects are illustrated in Table 3. Hypothesis 1 is to a large extent confirmed. Advertisements for the existing and well-known brand led to a better understanding of the advertisement and to a more positive attitude towards the brand and Purchase Intention. In addition, to a certain extent the feeling towards the advertisement was more positive in the case of an existing brand. Only the attitude towards the advertisement-knowing factor was not affected. Hypothesis 2 was largely supported with respect to the effect of the type of advertisement on responses. Positive emotional appeals led to more positive attitude towards the advertisement-feeling scores, while non-emotional appeals led to more positive attitude towards the advertisement-knowing and understanding scores. However, the type of advertisement did not affect the attitude towards the brand and the purchase intention. Hypothesis 3 was confirmed. The context style as such did not have an impact on attitudes and purchase intentions. The interaction effect between the type of advertisement and the type of product was not significant. The use of emotional or non-emotional advertisements for new or existing products did not seem to make a difference for responses to the advertisements. Hypothesis 4 could not be confirmed. However, the interaction

Advertising for New and Existing Brands

121

Table 2. F-values for univariate tests – experiment 1 Attitude Attitude Attitude Attitude towards the towards the towards the advertisement advertisement advertisement towards the brand understanding knowing feeling

Effect Corrected model Advertisement Context Brand Advertisement6 Context Advertisement6 Brand Context6Brand Advertisement6 Context6Brand Significance:

****

0.1%,

11.130**** 66.612**** 0.677 2.763* 0.900

14.672**** 95.445**** 3.154* 1.240 1.253

6.297**** 9.120*** 0.360 26.398**** 0.122

Purchase intention

7.640**** 1.430 1.282 46.023**** 0.686

4.690**** 0.270 0.809 27.113**** 0.001

1.067

0.401

0.108

0.628

0.155

4.497** 0.057

3.484* 0.002

5.619** 3.870**

3.467* 0.790

4.933** 0.328

***

1%,

**

5%, *10%.

Table 3. Main effects of brand type, advertisement type and context type – experiment 1 Brand

Attitude towards the advertisement-feeling Attitude towards the advertisement-knowing Attitude towards the advertisement understanding Attitude towards the brand Purchase intention

Advertisement

Context

Existing

New

Emotional

Nonemotional

NonEmotional emotional

3.896

3.347

4.082

2.861

3.410

3.533

4.101

3.933

3.280

4.755

3.883

4.151

5.557

4.712

4.886

5.383

5.184

5.085

5.275

4.174

4.821

4.671

4.632

4.816

4.227

3.329

3.823

3.734

3.701

3.856

The cells are mean scores.

between the context type and product type did affect responses significantly (Figure 1). All three components of attitude towards the advertisement, attitude towards the brand and purchase intention were influenced in a similar way. A positive emotional context led to more positive responses to advertisements for existing products than for new products. A non-emotional context enhanced positive responses to advertisements for new products more than for existing products. Hypothesis 5 was fully confirmed.

122

W. Janssens & P. De Pelsmacker

Figure 1. Interaction effects between context type and product type.

Experiment 2 Both new products were expected to be moderate- to high-involvement goods. In order to check this manipulation the product involvement items were factor analysed. One factor emerged and the corresponding scale had a Cronbach’s a of 0.83. The scores for the healthy drink and the watch were not significantly different from each other (mean score55.74 versus 5.66, t50.446, and p50.656). The two scales used for checking the advertisement/context manipulation were also factor analysed. In both cases a one-factor solution emerged. The Cronbach’s a value for the type of advertisement manipulation check was 0.87 and that for the context type manipulation check was 0.96. The mean scores of the two scales were used for checking the manipulation. The participants did indeed find the emotional advertisements more emotional than the non-emotional advertisements (mean score53.56 versus 2.08, t57.89, and p,0.001). Similarly, the non-emotional newspaper front page was considered as more non-emotional than the emotional one (mean score emotional front page52.19 versus 5.56, t531.59, and p,0.001). As in experiment 1, a principal component analysis was performed on the 17 items measuring the attitude towards the advertisement, which again resulted in three underlying dimensions: feeling, knowing and understanding. The attitude towards the advertisement items with a factor loading of more than 0.50 on the corresponding factor and less than 0.35 on any other factor were brought into a structural equation model in order to perform a confirmatory factor analysis with three latent variables using AMOSTM (SPSS). The analysis led to the omission of a

Advertising for New and Existing Brands

123

number of items, after which the model showed acceptable results. The Bollen–Stine bootstrapped p-value was 0.11 and other model fit indices supported the unidimensionality of the constructs (corrected x2 index is 1.55, CFI50.95, TLI50.93. and RMSEA50.052). However, one of the standardized factor regression coefficients was 0.393, which is below the recommended value of 0.50 for giving support for convergent validity (Hildebrandt, 1987; Steenkamp and Van Trijp, 1991). This means that the item had to be left out of the model, which means that no fit measures can be given because deletion of this item caused a non-admissible solution, due to under-identification of the model. In order to check the reliability of the constructs only Cronbach’s a values were calculated (due to the absence of an admissible confirmatory factor analysis solution). As Table 4 shows, all Cronbach’s a values exceeded 0.70 (Steenkamp and Van Trijp, 1991), thereby indicating acceptable reliability. A confirmatory factor analysis was also performed on the items measuring attitude towards the brand and purchase intention with each latent variable. The Bollen–Stine bootstrapped p-value was 0.36 and the corrected x2 index was 1.42 (CFI50.99, TLI50.99 and RMSEA50.045), thereby confirming the unidimensionality of the attitude towards the brand construct. A confirmatory factor analysis indicated a six-item solution for the purchase intention. The Bollen–Stine bootstrapped p-value was 0.56, and the corrected x2 index was 1.47 (CFI50.99, TLI50.98 and RMSEA50.048), thereby supporting the unidimensionality criterion.

Table 4. Reliability of the attitude towards the advertisement, attitude towards the brand and purchase intention scales – experiment 2 Attitude towards the advertisement-feeling(Cronbach’s a 0.70) This advertisement is beautiful This advertisement attracts attention This advertisement is well made This advertisement is original Attitude towards the advertisement-knowing(Cronbach’s a 0.75) This advertisement gives useful information This advertisement is believable This advertisement tells me something new This advertisement is interesting Attitude towards the advertisement-understanding(Cronbach’s a 0.75) You have to watch the advertisement frequently to know what it is exactly about (2) This advertisement is confusing (2) Attitude towards the brand(Cronbach’s a 0.94) I have a positive attitude towards this brand This brand looks attractive I would recommend this brand to others This brand is really something for me Purchase intention(Cronbach’s a 0.73) Next time that I need such a product, I will buy the advertised brand I will certainly try this brand It is very likely that I will buy the advertised brand It is a good decision to buy the advertised brand

124

W. Janssens & P. De Pelsmacker

All the factor regression coefficients were significant and higher than 0.50, for both constructs, respectively, thereby supporting convergent validity. The cut-off criteria were met for the three reliability indicators of both constructs (Table 4). The mean scores of the remaining items for each of the five constructs were used in the subsequent analysis (see Table 4). A 262 multivariate analysis of variance was performed for each of the two products. There was a main effect of type of advertisement (p50.001) and a moderately significant interaction effect between advertisement and context type (p50.092) for the healthy drink advertisement. As can be noticed from the univariate results in Table 5, only the model for the attitude towards the advertisementknowing factor and purchase intention was significant. In these models it was the emotional/non-emotional advertisement factor that was significant. A more nonemotional advertisement was preferred over an emotional advertisement with regard to the constructs attitude towards the advertisement-knowing and purchase intention (Table 7). This is in line with hypothesis 6 stating that, in the case of new products, consumers are more eager to collect information. However, in this experiment the affective attitude component was not significantly influenced. Therefore, hypothesis 6 was only partly supported with respect to healthy drinks. The type of advertisement main effect for the watch advertisement was the only significant effect of the multivariate effects (p50.017). The univariate results (Table 6) showed that the advertisement type factor was only significant for the attitude towards the advertisement feeling variable. The cell means in Table 8 show that an emotional advertisement caused a higher attitude towards the advertisementfeeling score than a non-emotional advertisement. The non-emotional advertisement showed higher scores on the other four advertising effectiveness constructs. Although the effects measured were in the expected direction, none were significant. Therefore, hypothesis 6 was not supported by the results for the product type ‘watch’. No significant effects could be found for the effect of context type. Although hypothesis 7 was supported for the healthy drink, it was not confirmed for the watch.

Table 5. F-values for univariate tests – experiment 2: healthy drink

Effect

Attitude Attitude Attitude towards the towards the towards the advertisement advertisement advertisement understanding knowing feeling

Attitude towards the brand

Purchase intention

3.361** 7.488*** 1.204 1.199

0.899 2.147 0.262 0.295

2.224* 5.781** 0.017 0.347

Corrected model Advertisement Context Advertisement6 Context Significance:

****

0.1%,

1.091 2.89* 0.046 0.411 ***

1%,

**

5%, *10%.

1.34 0.105 0.371 3.485*

Advertising for New and Existing Brands

125

Table 6. F-values for univariate tests – experiment 2: watch Attitude Attitude Attitude Attitude towards the towards the towards the advertisement advertisement advertisement towards the brand understanding knowing feeling

Effect Corrected model Advertisement Context Advertisement6 Context Significance:

**

2.830** 6.443** 0.511 1.538

0.284 0.001 0.133 0.621

0.955 1.954 0.941 0.084

Purchase intention

1.074 0.832 2.285 0.165

0.402 0.825 0.323 0.004

5%.

Table 7. Main effects of advertisement type and context type – experiment 2: healthy drink Advertisement

Attitude towards the advertisement-feeling Attitude towards the advertisement-knowing Attitude towards the advertisement-understanding Attitude towards the brand Purchase intention

Context

Emotional

Non-emotional

Emotional Non-emotional

2.888

2.520

2.681

2.727

2.342

2.939

2.760

2.521

5.488

5.581

5.447

5.622

2.812 1.807

3.162 2.275

2.926 2.054

3.048 2.029

The cells are mean scores.

Discussion, Conclusions and Suggestions for Further Research In this study, context style as such and the interaction between context style and type of advertisement did not have a significant effect on the responses to advertisements. However, a positive emotional context appeared to be more supportive for advertisements for existing products, while a non-emotional context improved the responses to advertisements for new products. Apparently, a non-emotional context created the appropriate circumstances for consumers to be motivated to pay attention and to learn something about an advertisement for an unknown brand. On the other hand, a positive emotional context improved the responses to advertisements for well-known brands. This emotional context appeared to prime the feeling and knowledge structures that enabled the elaboration of messages about brands that were already known and that did not require an extensive cognitive elaboration. These results confirm earlier findings (Yi, 1990, 1993; Perry et al., 1997; De Pelsmacker et al., 2002). A similar effect of media context was not found for different types of new products. It may be that the two products tested were not so different in

126

W. Janssens & P. De Pelsmacker Table 8. Main effects of advertisement type and context type – experiment 2: watch Advertisement

Context

Emotional

Non-emotional

Emotional

Non-emotional

3.065

2.529

2.721

2.872

2.570

2.577

2.539

2.608

5.749

6.081

6.030

5.800

3.143

3.374

3.067

3.450

1.904

2.085

1.937

2.051

Attitude towards the advertisementfeeling Attitude towards the advertisementknowing Attitude towards the advertisementunderstanding Attitude towards the brand Purchase intention The cells are mean scores.

terms of their emotional and non-emotional characters and the fact that they were new brands was more important than the fact that they were different product categories. However, even in that case, one would expect to find a more positive effect of non-emotional contexts in general. Experiment 2 did not support the findings with respect to the context effect of new products that were found in experiment 1. The hypothesized effects of non-emotional and emotional types of advertisements were to a certain extent confirmed in both studies. Positive emotional advertisements led to more positive feelings towards the advertisement and non-emotional advertisements led to more positive cognitive reactions. This confirmed earlier results (De Pelsmacker et al., 1998). Again, this effect was more pronounced in experiment 1 than in experiment 2. One remarkable result was that an interaction effect on the responses to advertisements between the newness of the brand and the type of advertisement was not found. Consequently, the most important conclusion of this study is that media context characteristics can be at least as important for advertising effectiveness as the characteristics of the advertising stimulus itself. As Lynch and Stipp (1999) have already indicated, there is relatively little known about media context effects on advertising responses and more research is needed. The results of this study are encouraging in that they at least show the relevance of context effects for responses to advertisements. The managerial implications of this kind of research are also important. Advertising effectiveness can be greatly improved by putting the right advertisement for the right product in the right context. This seems to be all the more relevant in the case of new versus existing products. A major limitation of the studies is that they were conducted in a laboratory setting and in relatively small samples of graduate students in business administration. The laboratory setting can be justified by the imperative need for

Advertising for New and Existing Brands

127

exact and controllable measurements of effects. However, the artificial laboratory environment can be expected to have a negative impact on the external validity of the findings. In other words, individuals in a laboratory setting may react differently to advertisements than people in a real-life situation. Although the samples were large enough to allow statistically meaningful conclusions, the external validity of the results may be compromised by the nature of the subjects involved in the study (business administration students). Similar studies in other groups or in samples that are more representative of the general population could lead to different conclusions. Further research could be developed in various directions. Obviously, other types and combinations of contexts and advertisements could be investigated. The difference between various types of products should be further explored. Earlier research has also indicated the importance of product category involvement and context appreciation for context effects (De Pelsmacker et al., 2002) and of the level of context involvement for responses to advertisements (Norris and Colman, 1992; Gunter et al., 1997). More research that investigates the interaction between context involvement and appreciation, context/advertisement type similarity and type of product and product category involvement could be envisaged. Finally, the impact of the layout of the newspaper pages and the positioning of the advertisements on these pages could also be explored, for instance by means of eye-tracking measurement. References Aaker, D. A. & Stayman, D. M. (1992) Implementing the concept of transformational advertising, Psychology and Marketing, 9(3), pp. 237–253. Alden, D. L. & Hoyer, W. D. (1993) Identifying global and culture-specific dimensions of humor in advertising, Journal of Advertising, 22(2), pp. 29–37. Belch, G. E. & Belch, M. A. (1998) Advertising and Promotion: An Integrated Marketing Communications Approach, 4th edn (New York: Irwin McGraw-Hill). Chattopadhyay, A. & Basu, K. (1990) Humor in advertising: the moderating role of prior brand evaluation, Journal of Marketing Research, 27, pp. 466–476. De Pelsmacker, P. & Geuens, M. (1996) The communication effects of warmth, eroticism and humour in alcohol advertisements, Journal of Marketing Communications, 2(4), pp. 247–262. De Pelsmacker, P., Decock, B. & Geuens, M. (1998) Advertising characteristics and the attitude towards the ad – a study of 100 likeable TV commercials, Marketing and Research Today, 27(4), pp. 166–179. De Pelsmacker, P., Geuens, M. & Anckaert, P. (2002) Media context and advertising effectiveness: the role of context appreciation and context–ad similarity, Journal of Advertising, 31(3), pp. 25–37. Derks, P. & Arora, S. (1993) Sex and salience in the appreciation of cartoon humor, Humor–International Journal of Humor Research, 6(1), pp. 57–69. Goldberg, M. E. & Gorn, G. J. (1987) Happy and sad TV programs: how they affect reactions to commercials, Journal of Consumer Research, 14, pp. 387–403. Gunter, B., Beeson, C. & Furnham, A. (1997) Recall of television advertisements as a function of program evaluation, The Journal of Psychology, 131(5), pp. 541–553. Herr, P. M. (1989) Priming price: prior knowledge and context effects, Journal of Consumer Research, 16(June), pp. 67–75. Hildebrandt, L. (1987) Consumer retail satisfaction in rural areas: a reanalysis of survey data, Journal of Economic Psychology, 8, pp. 19–42. LaTour, M. S. & Henthorne, T. L. (1994) Female nudity in advertising, arousal and response: a parsimonious extension, Psychological Reports, 75, pp. 1683–1690.

128

W. Janssens & P. De Pelsmacker

Lee, A. & Sternthal, B. (1999) The effects of positive mood on memory, Journal of Consumer Research, 26, pp. 115–127. Lynch, K. & Stipp, H. (1999) Examination of qualitative viewing factors for optimal advertising strategies, Journal of Advertising Research, 39, pp. 7–19. MacInnis, D. J. & Jaworski, B. J. (1989) Information processing from advertisements: toward an integrative framework, Journal of Marketing, 53, pp. 1–23. Norris, C. & Colman, A. M. (1992) Context effects on recall and recognition of magazine advertisements, Journal of Advertising, 21(3), pp. 37–46. Perry, S. D., Jenzowsky, S. A., King, C. M. & Yi, H. (1997) Using humorous programs as a vehicle for humorous commercials, Journal of Communication, 47(1), pp. 20–39. Petty, R. E. & Cacioppo, J. T. (1986) Communication and Persuasion: Central and Peripheral Routes to Attitude Change (New York: Springer). Puto, C. & Wells, W. D. (1984) Information and transformational advertising: the differential efforts in time, Advances in Consumer Research, 11, pp. 638–643. Steenkamp, J. B. E. M. & Van Trijp, H. (1991) The use of Lisrel in validating marketing constructs, International Journal of Research in Marketing, 8, pp. 283–299. Swaminathan, V., Zinkhan, G. M. & Reddy, S. K. (1996) The evolution and antecedents of transformational advertising: a conceptual model, Advances in Consumer Research, 23, pp. 49–55. Weinberger, M. G. & Campbell, L. (1991) The use and impact of humor in radio advertising, Journal of Advertising Research, 30(6), pp. 44–51. Weinberger, M. G. & Gulas, C. S. (1992) The impact of humor in advertising – a review, Journal of Advertising, 18(2), pp. 35–59. Yi, Y. (1990) The effects of contextual priming in print advertisements, Journal of Consumer Research, 17(2), pp. 215–222. Yi, Y. (1993) Contextual priming in print advertisements: the moderating role of prior knowledge, Journal of Advertising, 22(1), pp. 1–10.

Notes on Contributors Wim Janssens is a doctoral student and teaching assistant at the faculty of applied economics at the University of Antwerp, Belgium. His research focuses on the effectiveness of different types of advertising in general and the impact of media context in particular and on ethical consumer behaviour. Patrick De Pelsmacker is a full professor of marketing at the faculty of applied economics at the University of Antwerp and part-time professor of marketing at the University of Ghent. His main research interest is the effectiveness of marketing communications instruments and consumer behaviour.