Matching Celebrities with Products:
The effects of congruence between celebrity endorsers and
products on consumer responses and the moderating role of
persuasion knowledge.
Master’s thesis at the Graduate School of Communication
Name Thijs Zuijderwijk
Student number 10013202
University University of Amsterdam
Faculty Social and Behavioral Sciences
Track Persuasive communication
Place Amsterdam
Name supervisor Dr. A. den Boon
Abstract
This study focuses on the question whether the degree of congruence between celebrity endorsers and products affects brand recall, brand attitude and purchase intentions. Persuasion knowledge is taken into account as a possible moderator in the relationship between congruence and two of these dependent variables. An online-experiment was conducted in which 133 people were used to analyze the data. Two different varieties of stimuli with different products and celebrity endorsers were used to test the hypotheses. The results show significantly higher brand recall scores in one of the varieties. The people who saw the congruent condition had a better remembrance of the brand, compared to people who saw the incongruent condition. No significant effects were found for the relationship between congruence and brand attitude and purchase intentions. In addition, no moderating role of persuasion knowledge was found in any of the cases. Limitations and practical implications are discussed.
Introduction
From a marketing communication perspective it is important for companies to find and use strategies that can lead to a competitive advantage for a particular product or service. Celebrity endorsement in advertising can be seen as one of these strategic tools. It is estimated that nowadays one in four advertisements use a form of celebrity endorsement (MarketWatch, 2006). Celebrity endorsement can be defined as a way to promote a brand or campaign that involves a well-known person using their status to help promote a brand, product or service. Companies invest large sums of money to get celebrities to endorse their products in the expectation the positive characteristics of these celebrities will be transferred to their brand, service or product. Celebrity endorsement is known to influence advertising effectiveness, brand memory, brand attitude, and purchase intentions (Amos, Holmes & Strutton, 2008; Silvera & Austad, 2004; Byrne, Whitehead & Breen, 2003; Erdogan, 1999; Till, Stanley & Priluck, 2008). However, due to the fact celebrity endorsement is a common used strategy it is hard to compete with other companies that use the same strategy. So for companies it is relevant to know how to use celebrity endorsement in order to still obtain the competitive advantage.
One of these advantages might be obtained by matching the celebrity endorser with the brand, product or service. The effects of investigating this matching in celebrity endorsement resulted in the match-up hypothesis, which shortly states that the degree of fit between an endorser and a brand can have cognitive and affective effects on brand responses (Till & Busler, 2000). However, when looking into this topic in the abundance of existing literature it becomes clear there is a gap. There is little research on the effects of matching celebrity endorsers with products. Especially with three different dependent variables on three different levels: brand recall
(cognitive), brand attitude (affective) and purchase intentions (conative). In most studies the focus is on matching the image of the celebrity with the image of the brand. In some other studies the focus is on congruence in relation to attractiveness or
expertise between celebrity endorsers and products or brands (Kahle & Homer, 1985; Kamins, 1990; Till & Busler, 2000). This study will focus solely on the level of congruence between particular products and celebrity endorsers on three different outcomes (brand recall, brand attitude and purchase intentions) and takes persuasion knowledge into account when studying these three different relationships. The literature shows a possible moderating role of persuasion knowledge on the relationships between the degree of congruence and the cognitive and conative outcomes (Matthes, Schemer & Wirth, 2007). To get a clear understanding what is investigated in this study the following research question is formulated: To what extent
does congruence between celebrity endorsers and products influence brand attitude, brand recall and purchase intentions? And how is this relationship moderated by persuasion knowledge?
Theoretical framework
The main focus in this study is the effect of congruence between celebrity endorsers and products on brand, - attitude, recall and purchase intentions. Although the link between celebrity endorsers and their effects on brand attitude and purchase intentions is researched extensively (Amos, Holmes & Strutton, 2008; Byrne, Whitehead & Breen, 2003; Silvera & Austad, 2004), there is scarce research on matching celebrity endorsers with products and the effects that might occur. However, researchers did look at the effects of matching an endorser or spokesperson with the brand. They investigated the ‘match-up hypothesis’. This hypothesis states that endorsers are more
effective when there is a “fit” between the endorser and the endorsed brand (Till & Busler, 2000). To get a more thorough understanding of the match-up hypothesis and its effects, it will be explained in more detail.
The core of the match-up hypothesis is about the impact different types of endorsers, often celebrity endorsers, have on the endorsed brand. The term ‘match-up hypothesis’ was first introduced by Kanungo and Pang (1973). In their research they paired males and females endorsers with different types of products and found that effects varied depending on the product with which the models were paired. The results were explained in terms of ‘fittingness’ between the endorsers and products. This idea of a fit between an endorser and a brand or product was later on captured by the term “match-up hypothesis” (Kahle & Homer, 1985; Kamins, 1990).
Once the term had been introduced into the scientific literature, researchers mainly investigated it in context of attractiveness or expertise (Kahle & Homer, 1985; Kamins, 1990; Till & Busler, 2000). In these studies the focus was on how the
attractive image of an endorser can be associated with a brand or product that enhances one’s attractiveness. Kahle and Homer (1985) were among the first who found
significant results for the match-up hypothesis in this context. They showed that people have a more favorable brand attitude when a brand that is associated with enhancing one’s attractiveness was paired with an attractive endorser, compared to an unattractive endorser. A more recent study showed expertise is possibly more
important compared to physical attractiveness for matching a brand with an
appropriate endorser (Till & Busler, 2000). Still, the question remains how this works when celebrity endorsers are matched with particular products in a different context.
Degree of congruence and brand recall
Associative learning theory
There are several theories that can be used to explain the effects of the match-up hypothesis on levels of recall. First the associative learning theory is a useful
framework. Associative learning is a mechanism people use to link different concepts to each other. An associative network structure of memory consists of patterns of nodes (concepts) linked together (Anderson, 1983). For both endorsers and brands we have these networks available in our brain. When we are confronted with a celebrity endorser, the associative network that we have available of that person is activated. When a celebrity endorses a product or brand, the nodes of that celebrity may be compared with the advertised attributes of the product for congruence or fit with their available schema. The degree of fit/congruence between the new information about the product and the existing information about the celebrity may then influence the level of recall. There are studies indicating congruent information is remembered better, compared to information that is not relevant with existing schema’s (Cantor & Mischel, 1979; Cohen, 1981; Taylor & Cracker, 1981).
The filtering model
This model explains and describes how the initial coding of schema-related
information is structured in a meaningful way (Taylor & Cracker, 1981). People have this information structured and probably have existing schemas for celebrities. When exposed to an ad featuring the celebrity, the existing schema relating to that celebrity, is likely to be evoked. This process takes place when the celebrity is used prominently, which is often the case. According to the filtering model, when the consumer is
schema of the celebrity would not be encoded as well as information congruent with that schema (Taylor & Cracker, 1981). Thus, information congruent with the celebrity schema should have higher levels of brand recall.
Associative network model
Although there are several studies indicating a link between congruent information and better remembrance of the information, there is also a theory that states the opposite. This theory suggests that information that is inconsistent with existing schemas is likely to be quite salient and informative. Therefore, it is possible that people attend to it more closely and the information could be processed more deeply (Craick &
Lockhart, 1972). Srull, Lichtenstein and Rothbart (1985) examined this process in more detail. They found that information inconsistent with the existing schemas would be linked to a larger number of items in the memory, compared to information
consistent with the schema. Thus, incongruent information will be processed more extensively and therefore it can be remembered better than congruent information (Srull et al., 1985). A more recent study in the context of congruence between advertisements and magazines showed similar results (Moorman, Neijens & Smit, 2002). The authors argue a phenomenon called “route to visibility”. They propose that incongruent advertisements might be more prominent and therefor provide a certain route to visibility. This in turn might lead to a better recall of the advertisement.
The majority of existing theories and empirical studies point in the direction of congruent information leading to higher levels of recall (Cantor & Mischel, 1979; Cohen, 1981; Taylor & Cracker, 1981; Anderson, 1983). However there are also theories predicting the opposite (Craick & Lockhart, 1972). Nevertheless, there is
more empirical evidence pointing in the direction of congruence leading to better brand recall. Therefore the first hypothesis will be as follows:
H1: Congruence between the celebrity endorser and the product will lead to a
better memory of the brand, compared to an ad with no congruence between the celebrity endorser and the product.
Degree of congruence and brand attitude
There are very few studies of congruence on affect. The question is whether the degree of congruence elicits affect. There are empirical studies which support the idea that increased elaboration associated with extreme incongruency has an adverse effect on evaluations (Lee & Mason 1999; Meyers-Levy & Ty-bout, 1989). In one study the focus was on the degree of congruence between the modality and the plot of a story in a product placement setting (Russell, 2002). In the study they argue that incongruence can lead to an increase in attention to the placement, this will lead to viewers thinking about the reason for the brand’s presence in the show. This so called brand-relevant thinking will result in corrective mechanisms, such as counter-arguing or reactance, if the placement is perceived as inappropriate (Friestad & Wright 1995). Therefore they state that congruous placements will be perceived as acceptable, while incongruous placements are likely to raise suspicion with viewers and eventually lead to counter-arguing.
When these results are reflected on matching celebrity endorsers and products it is to be expected that incongruency between the two, can lead to an increase in attention to the ad. The increase in attention can cause viewers to think on the reason why the celebrity endorsement is present in the advertisement. This in turn can activate
corrective mechanisms and people could start counter-arguing the message. Therefore, we expect that in conditions where celebrity endorsers and products are a match there will be better attitudinal evaluations compared to conditions where celebrity endorsers and products do not match. From these expectations, which were formed from
empirical findings, the second hypothesis is drawn:
H2: Congruence between the celebrity endorser and the product will lead to a
more positive brand attitude, compared to an ad with no congruence between the celebrity endorser and the product.
Degree of congruence and purchase intentions
Now the question remains how purchase intentions can be influenced by the degree of congruence between celebrity endorsers and products. This issue, addressing the effectiveness of a match between a celebrity endorser and a product was examined from an attractiveness/likability perspective (Kahle & Homer, 1985). In their study likability is defined as easy to like and having qualities that bring about a favorable regard, whether attractiveness is defined as providing pleasure or delight, especially in appearance or manner. The results showed that attitudes and purchase intentions changed due to source attractiveness, but not likability. These findings are consistent with the match-up hypothesis. They showed that the degree of attractiveness with the celebrity endorser was important if the product is attractiveness-related. In this study we solely look at congruence in terms of schematic resemblance between the celebrity endorser and the product. Nevertheless, we expect a similar effect to be found.
H3: Congruence between the celebrity endorser and the product will lead to a
higher purchase intention, compared to an ad with no congruence between the celebrity endorser and the product.
Moderating effect of persuasion knowledge
Assuming that all the effects discussed above have no parameters would be incorrect when looking into the existing literature. One important characteristic of receivers is their knowledge on persuasion attempts. In this case a celebrity endorser is used to endorse the brand. People might understand what is going on, when confronted with a message in which celebrity endorsement is used to persuade the viewers. Friestad and Wright (1994) were one of the first researchers who looked into this concept of persuasion knowledge. They created a Persuasion Knowledge Model that consists of three types of knowledge people can use when confronted with a persuasive attempt. First there is the agent knowledge, this is the knowledge receivers have about the sender (in their study described as the agent). Second there is the topic knowledge that can be defined as the knowledge receivers have on the topic about what the persuasive attempt is made. Third and last there is the persuasion knowledge. Persuasion
knowledge can be defined as a receiver’s perception of an agent's strategic behavior in presenting information designed to influence someone's beliefs, attitudes, decisions, or actions (Friestad & Wright, 1994). The question is what the effects of this persuasion knowledge are.
In another study conducted by Friestad and Wright (1995) they argue that people can detect a persuasive attempt. This leads to activation of persuasion knowledge and people consequently develop more negative attitudes towards the brand. Persuasion knowledge evolves from experiences with persuasion that may
differ among individuals (Boush, Friestad & Rose, 1994; Friestad & Wright, 1995). Therefore people with more persuasion knowledge have different attitudes after watching a particular persuasive message, compared to people with less persuasion knowledge. In fact, people who have developed a high degree of persuasion
knowledge during their media socialization are less likely to be influenced by celebrity endorsement via the congruence effect. In contrast, individuals who have no accessible and applicable persuasion knowledge for celebrity endorsements are unlikely to
employ strategies of counter-arguing. Thus, persuasion knowledge is a possible moderator that can be identified based on the existing literature.
In a study which looked at the moderating role of persuasion knowledge on the effects of brand placements on brand attitudes, the researchers found lower brand attitudes with people who have high persuasion knowledge, compared to people who have low persuasion knowledge (Matthes, Schemer & Wirth, 2007). According to these outcomes a moderating role of persuasion knowledge is expected on the
relationship between celebrity/product congruity and brand attitudes. It is expected that congruence between celebrity endorsers and products have a positive effect on brand attitudes when persuasion knowledge is low. For people with high persuasion
knowledge it is expected that the attitude towards the brand will be less positive or even negative. For purchase intentions we expect the same directions as for brand attitudes.
However for our third dependent variable brand memory there is no literature to give any direction. In the study conducted by Matthes, Schemer and Wirth (2007), brand recognition was also measured as a dependent variable and they found no significant differences between respondents who had low or high persuasion
+
+
+ -
-
why persuasion knowledge would influence the memory of the brand. Therefore, there are no hypotheses drawn on the moderating role of persuasion knowledge on the relation between celebrity/product congruity and brand memory. From the
assumptions made about the effects on brand attitudes and purchase intentions, the last two hypotheses are drawn.
H4: Congruence between the celebrity endorser and the product will lead to a
more positive brand attitude, compared to an ad with no congruence between the celebrity endorser and the product, when persuasion knowledge is low.
H5: Congruence between the celebrity endorser and the product will lead to a
higher purchase intention, compared to an ad with no congruence between the celebrity endorser and the product, when persuasion knowledge is low.
Figure 1: Conceptual model
Congruence vs. Incongruence Brand recall Persuasion knowledge Brand attitude Purchase intentions
Method Design
To test the hypotheses two different varieties of stimulus sets were used. Selecting two celebrities and two different products from fictitious brands created two levels of relationship between the celebrity endorser and the product. In this study fictitious brand were chosen, so strong existing attitudes towards the brand could not have an effect on brand attitude. The set up was created in a way product 1 is congruent with celebrity endorser 1, but incongruent with celebrity endorser 2. Product 2 is congruent with celebrity endorser 2, but incongruent with celebrity endorser 1. In that way an experimental design was created in which both conditions can control for each other. See figure two for the configurations of the four conditions.
Figure 2: Configurations for experimental design
Celebrity endorser 1 Celebrity endorser 2 Product 1 Congruent 1 Incongruent 1
Product 2 Incongruent 2 Congruent 2
Participants and procedure
A total of 204 participants participated in the experiment. The participants were gathered using a non-probability convenience sampling method. Not all 204
participants could be used, because some of them were not familiar with the celebrity endorser or did not entirely complete the experiment. After cleaning the data and removing participants that could not be used in this study, a sample of 133 participants was used in data analyses. A slight majority of the sample is male (56%). The age of
the participants is between 13 and 73 years (M = 30.06, SD = 14.01). For most of the participants a Bachelor’s degree (30.8%) and a Master’s degree (32.3%) is the highest education they finished. The participants were randomly divided between the four conditions in a way every condition contains at least 30 participants (condition 1: 30, condition 2: 38, condition 3: 35, condition 4: 30). The four conditions were regrouped into two conditions: congruence (N = 65) and incongruence (N = 68). However the original conditions were also used to look at differences between the Jamie Oliver and Pharrel Williams varieties.
The study was conducted between the date of 18th of November and the 8th of December. To get access to the experiment participants clicked a link that directed them to the online experiment. The experiment was created in Qualtrics, an online questionnaire platform. The program enables researchers to conduct studies in an online environment and to quantitatively analyze statistical data. First the participants read a short introduction in which they were briefly told some information about the study. They were also told that answers they were about to give would be handled on an anonymous basis and the information obtained by this study will solely be used for this study and not for other purposes. After the participants read the introduction they were, dependent on in which condition they were, confronted with one of the four advertisements.
By randomizing the different advertisements in the online experiment, it was assured all participants were evenly random assigned to the four conditions. After being exposed to either a celebrity endorser matching with the product or a celebrity endorser that did not match with the product, they continued with a filler task. This filler task made sure participants were cognitively occupied with something else before they were asked if they could recall the brand. After the brand memory question,
participants were asked about their attitude towards the brand and the other constructs that are investigated in this study. Lastly, the participants ware asked to give some demographics.
Materials and measures
Independent measures
Advertisements. To investigate the effects of congruency between celebrity
endorsers and products on brand recall, brand attitude and purchase intentions, two different celebrities and products were used. The first is Pharrel Williams, a famous singer and producer of music. The second celebrity is Jamie Oliver, a famous cook. These celebrity endorsers, dependent on the condition the participants were in, present a new cooking pan or new headphones. These products both have fictitious brands: Cookpro and Soundpro. In all four conditions the presented advertisement contains the same amount of global information about the product and the layout of the ad is made as similar as possible. There is a picture of the celebrity endorsers face present in the advertisement, besides the product is shown with the name of the brand in the center. The text in the advertisement says: “[Celebrity endorser] uses this brand new
[product]” See Appendix A for the advertisements. There was also a question to check if the chosen celebrities in this study are known to be famous. In this way there is an additional control to check if the celebrities are also perceived as celebrities.
Congruence. To measure congruence a modified version of the already
validated scale by Speed and Thompson (2000) was used. Congruence was measured on 7-point Likert scale (1 = strongly disagree, 7 = strongly agree). Examples of items the participants had to answer were: “There is a logical connection between the celebrity and the product”, “The image of the celebrity and the image of the product
are similar”, “The celebrity and the product fit together well”, “The celebrity and the product stand for similar things”, and lastly “It makes sense to me that this product is endorsed by this celebrity” (Speed & Thompson, 2000). A reliability analyses was conducted to see whether the formed scale is reliable. Cronbach’s alpha showed a satisfying internal reliability (α = 0.95).
Dependent measures
Brand recall. This variable was measured explicitly. The variable consists of
concepts that explicitly activate the participants to think back on the stimuli they have just seen. Brand recall was measured by means of spontaneous recall. The participants were asked, by the use of an open question, to name the brand. If the brand name was incorrect it was coded ‘0’, when the brand name was correct it was coded ‘1’
(Wakolbinger, Denk & Oberecker, 2009).
Brand attitude. Brand attitude was measured with seven items on a 7-point
semantic differential scale. The participants were asked to indicate how they found the following characteristics would apply to the different brands: uninteresting –
interesting, stupid – like, negative – positive, difficult – easy, unpleasant – pleasant, unattractive – attractive, boring – fascinating (Matthes, Schemer & Wirth 2007). A reliability analyses was conducted to see whether the formed scale is reliable. Cronbach’s alpha showed a proficient internal reliability (α = .87). There was a possibility to make the scale even more reliable by deleting one item. However, this was a negligible improvement (α = 0.88), so it is decided to keep the item in the scale.
Purchase intention. To measure purchase intention participants were asked
itself. The respondents could answer this question on a scale from one to seven (1 = definitely will not, 7 = definitely will).
Persuasion knowledge. Persuasion knowledge was measured with five items.
The scale was administered with a five-point scale ranging from one ‘not agree at all’ to five ‘fully agree’. After collecting data these scores were divided dichotomously with ‘0’ (ranging from 0 to 2.49) and ‘1’ (ranging from 2.50 to 5). In order to capture the negative preoccupation, persuasion knowledge is operationalized as an activated attitude of how disturbing the celebrity endorsements were perceived (Friestad & Wright, 1995). Examples of statements were: “I think some persons will feel disturbed by the celebrity endorsement that was shown in the advertisement”, “I found the celebrity endorsement appearances in the story annoying” and “The celebrity
endorsement in the advertisement should have been cut out”. A reliability analyses was conducted to see whether the formed scale is reliable. The test shows that the
persuasion knowledge scale is reliable (α = 0.74).
Analysis
To find out whether congruence between celebrity endorsers and products effects brand recall, brand attitude and purchase intentions, both Independent Sample T-tests and one-way ANOVA analyses were conducted. In all cases a Levene's Test for
Equality of Variances was performed to check whether it could be assumed the
variance between groups is equal. Lastly Pearson R correlations were calculated to check whether background variables correlated with both independent and dependent variables. In this way it was checked if covariates were necessary to take into account when conducting the one-way ANOVA analyses.
Results Descriptive statistics
A total of 204 participants conducted the experiment, from which 133 were used to analyze the data. The participants in the sample had a moderate level of persuasion knowledge (M = 2.75, SD = 1.02). The participants had a fairly positive attitude towards both fictitious brands (M = 4.25, SD = 1.09). Furthermore, the results show the congruent and incongruent conditions were also perceived as intended. Participants who saw one of the congruent conditions scored significantly higher on the degree of congruence scale (M = 5.20, SD = 1.19), compared to participants who saw one of the incongruent conditions (M = 2.24, SD = 1.27), t (131) = 13.78, p < 0.001, CI [3.38, -2.53].
Congruence and brand recall
To find out whether there is a difference in brand recall scores between participants who saw a congruent condition or an incongruent condition, an Independent Sample
T-test was conducted. In the first T-test both the incongruent and congruent conditions
were compared. This test shows that participants who watched either the Jamie Oliver or Pharrel Williams congruent condition (M = .29, SD = .46), did not significantly differ in brand recall scores compared to participants who watched one of the
incongruent conditions (M = .23, SD = .43), t (131) = -.74, p = .459, CI, [-0.21, 0.09]. Second, the conditions were separated and a test was conducted to find the differences in brand recall scores between Jamie Oliver congruent and Jamie Oliver incongruent. The t-test shows that participants who watched the Jamie Oliver
congruent condition had a significantly higher brand recall score (M = .37, SD = .49), compared to participants who saw the Jamie Oliver incongruent condition (M = .10,
SD = .31), t (63) = 2.72, p < 0.01, CI [0.07, 0.47]. For the Pharrel Williams conditions,
another Independent Sample T-test was conducted. The test shows there were no significant differences in brand recall between the participants who watched the congruent condition (M = .20, SD = .41) and the participants who watched the incongruent condition (M = .34, SD = .48), t (66) = -1.32, p = .192, CI [-0.35, 0.07]. This means the first hypothesis is partially supported, because the significant
difference was only found in the Jamie Oliver variety.
Congruence and brand attitude with persuasion knowledge as moderator
To find out whether participants who saw a congruent condition had a better attitude towards the brand, compared to participants who saw an incongruent condition a
one-way ANOVA analyses was conducted. First a Levene's Test for Equality of Variances
was conducted. The test shows no significance, this means it can be assumed that the variance between the two conditions is not unequal, F (3, 129) = .51, p = 0.674. When controlling for background variables the results show there are no other variables that correlate with both the independent as the dependent variable. Therefore no variables there were included as covariates in the ANOVA.
The one-way ANOVA analysis shows there are no significant differences in attitude scores between the two conditions, F (1, 129) = .67, p = .416. The participants who saw one of the congruent conditions did not score significantly higher on brand attitude (M = 4.40, SD = 1.11), compared to the participants who saw one the
incongruent conditions (M = 4.12, SD = 1.06). This means the second hypothesis is not confirmed. When testing for an interaction effect, as expected, there is no moderating role of persuasion knowledge. Participants with low persuasion knowledge in a congruent condition did not have a significantly higher brand attitude (Mcongruence =
4.73, SD = .96), compared to participants with low persuasion knowledge in an incongruent setting (Mincongruence = 4.49, SD = 1.02). This means the fourth hypothesis is also not confirmed. The results from the one-way ANOVA analysis with brand attitude as dependent variable is presented in Table 1. Another result can be seen when looking at Table 1. Although no significant effects were found for congruence, there is a significant difference between people with low (M = 4.64, SD = .99) and high
persuasion knowledge (M = 3.98, SD = 1.09) on brand attitude, F (1, 129) = 10.66, p < 0.01, η2 = 0.075.
Table 1:
One-way ANOVA analyses with brand attitude as dependent variable
Variable df F p-value Congruence 1 .67 .416 Persuasion_knowledge 1 10.66 .001 Congruence * Persuasion_knowledge 1 0.27 .653
Note: Persuasion knowledge is measured as high (1) and low (0).
Congruence and purchase intention with persuasion knowledge as moderator
Again the background variables were checked for correlations with both the dependent as independent variables. There were no significant correlations, thus there were no background variables included as covariates. To test whether there are differences in purchase intentions between participants who saw a congruent condition and
conducted. The Levene's Test for Equality of Variances shows it can be assumed the variance is not unequal across groups, F (3, 129) = .79, p = .501.
The one-way ANOVA analyses shows there are no significant differences in purchase intentions between participants who saw the congruent condition (M = 3.25,
SD = 1.51), and participants who saw the incongruent condition (M = 3.03, SD = 1.61), F (1, 129) = .22, p = .639. This means the third hypothesis is not confirmed. When
testing for an interaction effect, again there is no moderating role of persuasion knowledge. Participants with low persuasion knowledge in a congruent condition did not have a significantly higher purchase intention (Mcongruence = 3.40, SD = 1.58), compared to participants with low persuasion knowledge in an incongruent setting (Mincongruence = 3.29, SD = 1.68). This means the fifth and last hypothesis is also not confirmed. The result of the one-way ANOVA analyses with purchase intention as dependent variable is presented in Table 2.
Table 2:
One-way ANOVA analyses with purchase intention as dependent variable
Variable df F p-value Congruence 1 .22 .639 Persuasion_knowledge 1 1.55 .216 Congruence * Persuasion_knowledge 1 0.004 .947
Discussion, limitations and implications
The first hypothesis is partially supported. The results showed that people recall the brand better when they saw Jamie Oliver in a congruent setting compared to people who saw Jamie Oliver in an incongruent setting. However this effect was not found for the Pharrel Williams variety. Apparently the expectations on brand recall differ
between the two varieties of celebrity endorsers and products that were chosen in this study. Only the outcomes for the Jamie Oliver advertisements are in line with the discussed theories (Cantor & Mischel, 1979; Cohen, 1981; Taylor & Cracker, 1981; Anderson, 1983). The reason for these different results between the two varieties might be due to the difference in schemas that people have. Jamie Oliver is clearly a cook and is only known for his profession as being a cook. Thus a cooking pan is a very strong fit with Jamie Oliver and it can be assumed that any other product that not fit in the cooking category or cooking schema is an obvious mismatch. For Pharrel Williams this might be slightly different. Besides a musician, Pharell Williams is also a fashion designer and a well-known producer for other artists. It is likely that people have a larger schema for Pharrel compared to Jamie. A cooking pan seems to be an obvious mismatch with Pharrel Williams. However, though a set of headphones might be perceived as a good fit it might not be perceived as good of a fit when it is
compared with Jamie Oliver and a cooking pan.
Another reason for the insignificant findings in the Pharrel Williams condition might be due to the fact opposite hypotheses can be formulated according to the
Associative Network Model (Craick & Lockhart, 1972; Srull et al., 1985). The
difference in brand recall scores can be insignificant because this theory suggests incongruent information will be processed more extensively and therefore it can be remembered better than congruent information. This can explain the fact there are no
significant differences between the congruent and incongruent settings in the Pharrel Williams variety. Still, it is remarkable this theory is not applicable in the Jamie Oliver condition. Apparently, the hypotheses that state each other’s opposite, work differently for different celebrity endorsers and products. Future research should look more deeply into these different outcomes for matching different celebrity endorsers with products.
The second and third hypotheses were both not confirmed. People who saw a congruent advertisement did not have a better brand attitude or higher purchase intention, compared to people who saw an incongruent advertisement. The
expectation, that incongruency leads to an increase in attention, which leads viewers thinking on the reasons why the celebrity endorsement is present, and this in turn can activate a counter arguing process is not found to be correct in this study (Lee & Mason 1999; Friestad & Wright 1995; Meyers-Levy & Ty-bout, 1989). The same goes for the expectation on the effects of congruence on purchase intentions. However, there is scarce theory and empirical evidence that gives direction for the difference in purchase intentions between congruent and incongruent advertisements.
When relating these findings to existing literature, it is hard to find an explanation for these results. However, the Theory of Planned Behavior can give implications for these results (Fishbein & Ajzen, 2010). According to this theory it seems logical that when brand attitude was not heightened by the degree of
congruence, purchase intentions are not heightened either. The theory of planned behavior gives an explanation on what steps are needed for people to perform a certain behavior. In this case buying a particular product can be described as the behavior. The theory states that first people form an attitude towards the behavior. This influences our intentions and lastly these intentions can lead to performing the behavior. Forming an attitude is a condition needed before a person goes into the next phase: changing the
intention (Fishbein & Ajzen, 2010). Thus, this can explain why purchase intention did not differ between the congruent and incongruent settings. However this explanation for the insignificant findings of purchase intentions still does not certify the
insignificant findings for the effects of congruence on brand attitude.
Although we controlled for existing attitudes towards the brands by making them fictitious, the celebrity endorsers used in this study were real. This limitation had to be taken for granted; else it would be practically impossible to look into the effects of congruence between celebrity endorsers and products. This does mean existing attitudes towards the celebrity endorsers could have an impact on the results. This could be an explanation for the insignificant findings. Another limitation that could have caused multiple insignificant results is the filler task that was used in this study. After participants saw the advertisement they were asked to watch an episode of ‘How it’s Made’. In this episode the participants watched a 5-minute video on how bowling balls are made. This topic was chosen because it was assumed the topic does not fit both schemas of celebrity endorsers and products. After watching the video some questions were asked about the content. In this way they would be cognitively
occupied with something else than the experimental manipulation. The video, because of its dull nature, could have caused irritated responses by some people.
The fourth and fifth hypotheses were also not confirmed. People who saw a congruent condition did not have a better brand attitude or higher purchase intention compared to people who saw an incongruent condition, when their persuasion knowledge is low. When relating these results to the rejection of hypotheses two and three it falls in line with what to expect. A moderator effect would only be possible if there is a significant relation between congruence and both dependent variables (brand attitude and purchase intention).
Nevertheless, there is a significant direct effect of persuasion knowledge on brand attitude. People with high persuasion knowledge had a significantly lower brand attitude, compared to people how have low persuasion knowledge. This outcome falls in line when related to the existing theories. The persuasion knowledge model argues that people with high persuasion knowledge are more likely to detect the persuasive attempt. This leads to activation of persuasion knowledge and people consequently develop more negative attitudes towards the brand (Friestad & Wright, 1995). However, because there was no direct effect of congruence on brand attitude, a
moderator effect in this study was not found. Future researchers could look into this in more detail, by using different validated brand attitude scales. Lastly, the sample in this study was relatively small and although each condition satisfies the minimum of 30 participants a larger sample could have given a significant result for the effect of congruence on brand attitude.
Besides some limitations, this study also has some strong points to take into consideration. All scales in this study were validated and had a satisfying reliability score. This makes the internal validity of this study strong. The fact a filler task was used makes the study results more generalizable, because the setting resembles that of a realistic setting in which people are confronted with multiple media and different advertisements. Because this was an online-experiment all external factors could not be controlled for. Nevertheless, this also makes it more realistic, because people could do the research in their own natural environment. This strengthens the external
validity.
The purpose of this study was to investigate the effects of congruence between celebrity endorsers and products on brand recall, brand attitude and purchase
intentions, with persuasion knowledge as a possible moderator. An online experiment in which 133 people participated was conducted. Every participant was confronted with an advertisement in which the celebrity endorser was congruent or incongruent with a product. Based on the results of this study the conclusion can be drawn that congruence between celebrity endorsers and products only effects brand recall under certain conditions (particular celebrity endorsers and products). In this case Jamie Oliver with a cooking pan (congruent) leads to higher brand recall, compared to Jamie Oliver with a set of headphones (incongruent). However, Pharell Williams with a set of headphones (congruent) did not lead to higher brand recall, compared to Pharell Williams with a cooking pan (incongruent). No differences in brand attitude and purchase intentions were found between the congruent and incongruent conditions for both varieties. There was also no moderating role of persuasion knowledge.
To summarize, no hypothesis was fully supported. Nevertheless, the result of congruence on brand recall provides implications for marketing purposes. Most research only focused on the match-up effects of endorsers and brands on recall or brand attitude. This study looked at dependent variables on three different levels and shed light on an important parameter. Now future researchers should focus more deeply on the different effects of matching multiple celebrity endorsers with different products and investigate why recall effects are only found in certain varieties. More important, the focus should be on the possible effects of schematic congruence between celebrity endorsers and products on brand attitude and purchase intentions.
Amos, C., Holmes, G. & Strutton, D. (2008). Exploring the relationship between celebrity endorser effects and advertising effectiveness. International Journal
of Advertising, 27(2), 209-234.
Anderson, J. R. (1983). A Spreading Activation Theory of Memory. Journal of Verbal
Learning and Verbal Behavior, 22(3), 261- 295.
Boush, D. M., Friestad, M. & Rose, G. M. (1994). Adolescent skepticism toward TV advertising and knowledge of advertiser tactics. Journal of consumer research, 165-175.
Byrne, A., Whitehead, M. & Breen, S. (2003). The naked truth of celebrity endorsement. British Food Journal, 105(5), 288-296.
Cantor, N. & Mischel, W. (1979). Prototypes in person perception. Advances in
experimental social psychology, 12, 3-52.
Cohen, C. E. (1981). Person categories and social perception: Testing some boundaries of the processing effect of prior knowledge. Journal of Personality and Social
Psychology, 40(3), 441-467.
Craick, F. I. & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of verbal learning and verbal behavior, 11(6), 671-684. Erdogan, B. Z. (1999). Celebrity endorsement: a literature review. Journal of
marketing management, 15(4), 291-314.
Fishbein, M. & Ajzen, I. (2010). Predicting and changing behavior: The reasoned
action approach (pp. 17-26). New York: Psychology Press.
Friestad, M. & Wright, P. (1994). The persuasion knowledge model: How people cope with persuasion attempts. Journal of Consumer Research, 21(1), 1-31.
Friestad, M. & Wright, P. (1995). Persuasion knowledge: lay people’s and researchers’ beliefs about the psychology of advertising. Journal of Consumer Research,
22, 62–74.
Kahle, L. R. & Homer, P. M. (1985). Physical attractiveness of the celebrity endorser: A social adaptation perspective. Journal of consumer research, 954-961. Kamins, M. A. (1990). An investigation into the “match-up” hypothesis in celebrity
advertising: When beauty may be only skin deep. Journal of Advertising,
19(1), 4-13.
Kanungo, R. & Pang, S. (1973). Effects of human models on perceived product quality. Journal of Applied Psychology, 57(2), 172-178.
Lee, Y. & Mason, C. (1999). Responses to information incongruency in advertising: The role of expectancy, relevancy, and humor. Journal of Consumer Research,
26, 156–169.
MarketWatch (2006), “A-list celebrity endorsements are failing to dazzle consumers”, MarketWatch: Global Round-Up 5(9), 29-30.
Matthes, J., Schemer, C. & Wirth, W. (2007). More than meets the eye: Investigating the hidden impact of brand placements in television magazines. International
Journal of Advertising, 26(4), 477-503.
Meyers-Levy, J. & Tybout, A. M. (1989). Schema congruity as a basis for product evaluation. Journal of consumer research, 39-54.
Moorman, M., Neijens, P. C. & Smit, E. G. (2002). The effects of magazine-induced psychological responses and thematic congruence on memory and attitude toward the ad in a real-life setting. Journal of Advertising, 31(4), 27-40.
Russell, C. A. (2002). Investigating the effectiveness of product placements in television shows: The role of modality and plot connection congruence on brand memory and attitude. Journal of consumer research, 29(3), 306-318. Silvera, D. H. & Austad, B. (2004). Factors predicting the effectiveness of celebrity
endorsement advertisements. European Journal of Marketing, 38(12), 1509-1526.
Speed, R. & Thompson, P. (2000). Determinants of sports sponsorship response.
Journal of the Academy of Marketing Science, 28(2), 226-238.
Srull, T. K., Lichtenstein, M. & Rothbart, M. (1985). Associative storage and retrieval processes in person memory. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 11(2), 316.
Taylor, S. E. & Crocker, J. (1981). Schematic bases of social information processing. In Social cognition: the Ontario symposium, 1, 89-134.
Till, B. D., & Busler, M. (2000). The match-up hypothesis: Physical attractiveness, expertise, and the role of fit on brand attitude, purchase intent and brand beliefs. Journal of Advertising, 29(3), 1-13.
Till, B.D., Stanley, S.M. & Priluck, R. (2008). Classical conditioning and celebrity endorsers: an examination of belongingness and resistance to extinction.
Psychology and Marketing 25(2), 173-189.
Wakolbinger, L.M., Denk, M. & Oberecker, K. (2009). The effectiveness of Combining online and print advertisements: is the whole better than the individual parts? Journal of Advertising Research, 49(3), 360-372.
Appendix A - Advertisements
Advertisement 1. Condition: Pharrel Williams congruent
Advertisement 3. Condition: Jamie Oliver congruent
Appendix B - Measurements
Brand recall: What brand did you see in the advertisement?
Brand attitude: To what extent do you find the following characteristics apply to the
brand that was shown in the advertisement?
1 (1) 2 (2) 3 (3) 4 (4) 5 (5) 6 (6) 7 (7) Stupid:Like (1) Pleasant:Unpleasant (2) Negative:Positive (3) Uninteresting:Interesting (4) Unattractive:Attractive (5) Easy:Difficult (6) Boring:Fascinating (7)
Purchase intentions: How likely is it that you will purchase the product that was
shown in the advertisement, the next time an opportunity presents itself?
Definitely will not (1) (2) (3) (4) (5) (6) Definitely will (7) (1)
Persuasion knowledge: To what extent do you agree with the following statements? Not agree at all (1) (2) (3) (4) Fully agree (5) I think some persons will feel disturbed by the celebrity endorsement that was shown in the advertisement. (1) I found the celebrity endorsement appearances in the advertisement annoying. (2) The celebrity endorsement in the advertisement should have been cut out.
(3)
Celebrity familiarity control question: Did you know the endorser in the
advertisement is a celebrity? Yes (1)
Congruence control question: Please indicate to what extent you agree with the following statements Totally Disagree (1) (2) (3) (4) (5) (6) Totally Agree (7) There is a logical connection between the celebrity and the product. (1) The image of the celebrity and the image of the product are similar. (2) The celebrity and the product fit together well. (3) The celebrity and the product stand for similar things. (4) It makes sense to me that this product is endorsed by this celebrity. (5)
Gender: What is your gender?
Male (1) Female (2)
Education: What is the highest level of education you have completed?[If you are
currently a student, please indicate the level you expect to complete.] No formal education (1)
Primary school / Junior High (2) Secondary school / High School (3)
Middelbaar Beroepsonderwijs (MBO) / Polytechnic Degree (4) Hogescholen (HBO) / College Applied Degree (5)
University Bachelor Degree (WO) (6) University Masters Degree (7) University PhD Degree (8)
