113 celebrity endorsement and stresses the need for thorough testing before celebrity endorsement is to be undertaken. To this, the personality fit of the endorser (and the ICS model) may be utilized by managers, to provide an indication of the effectiveness of a potential endorsement. However, it will likewise be necessary to include other factors that may have an impact, such as attractiveness, expertise and credibility.
The weighted importance of personality fit was measured against the three other main attributes found in the celebrity endorsement literature (expertise, physical attractiveness and credibility).
While the test was too weak to draw any statistically proven conclusions, theoretical arguments and rational answers from each of the methods used indicate that personality fit as being the most important attribute, when rating likeability and brand congruency. Based on the results and lessons learned from this study, the authors have presented a model for defining different forms of celebrity endorsement collaborations (see fiFigure 16 - Celebrity Endorsement congruency forms).
Though only postulated this model may provide an understanding of how ‘Identification’ and
‘Credibility’ helps the consumer transfer the meaning to themselves, which they may express through ‘Self-expression’.
Personality fit – an important piece of the puzzle.
114 Denmark is indeed a part of the Western culture, Gruens et. al. (2009) was actually not empirically proven for the Danish society as such. Furthermore, only one item was used to measure each personality dimension, even though the actual work of Gruens et. al (2009) included entire pools of items. This was however consciously chosen, so as to keep the perception of burden down, but in the end this will have weakened the validity of the results. This further as the items in the survey was stated in English, which is not the first language of most Danes, which may have caused them to misinterpret the meaning of the items (ibid). Further, as the definition of words may be different across cultures. The word Simple (which is one of the utilized personality dimensions) may in the Danish society for instance be positively associated with the simple and esteemed Scandinavian design, or associated with the similar word meaning ‘dull’. As such, much was left to the individual’s interpretation of the items. This may have affected the respondent’s ratings and made the personality dimensions less valid.
The issue with definitions and cultural differences is further found in a significant amount of the used theory. As highlighted in this study the definitions of attributes vary from theory to theory, making it difficult to compare and make strong arguments hereof. At the same time some of the theories used are based on different cultures, which may affect the comparison of results. This limitation is however found in the majority of academic work and is difficult to completely remove, as researchers generally test within their relevant context.
The number of items used was not only limited for the personality dimensions. All constructs in general had a limited amount of items, which thus limits the validity of the results, especially as the subject of personality is rather abstract. This was done to keep the perception of burden down, as the survey was already rather long, and it was viewed necessary to gain a certain quantity of respondents in order to test the conceptual model. Especially the limited amount of items within the involvement constructs may have influenced the results. If all 20 items, that previously has been found to make up involvement, had been used it is possible that the outcome would have been different. This limitation also explains why the emotion measurements (ad and brand
‘feelings’) were taking out of the conceptual model and instead accounted for theoretically. If the emotion constructs had been implemented this could possibly have showed slightly different results, but more importantly it might have provided deeper understanding of how personality affect emotions and which ones, as well as its impact hereof. Though it should be noted, as it is
115 argued in the study, the conceptual model and its results can be used without directly measuring the impact of emotions, as its effect is shown in measured attitude. In general it can be deemed, that for each more directly and rational questioned given, as a result of less used items due to constraints, the more understanding of its effect is lost. This though, does not mean that results found under constraints necessarily are less correct.
The respondents used in this study influenced by the authors limited means and reach, resulting in an fairly young average age group of 30 years old where the majority hereof highly likely have been students or newly graduated, which also is highlighted in the results section. It can be argued when the majority of the respondents have the same amount of life experience it can have an effect on the answers given. This also limits several other academic studies, and even though it is noteworthy, the findings are still deemed valid. Furthermore the chosen broad variety of respondents can have affected the results in relation to a narrower target group. If a narrow target group with higher interest in the Rolex brand had been used, the impact on brand attitude assumable had been altered and total effects of the used celebrities might have differentiated more.
In addition the results could have been to more use for Rolex, however this was not the purpose of this study.
The chosen target group involved both genders and the likely influence hereof were tested. The results showed to some extend the opposite of what was assumed, as it was believed prior to the test that females would have a higher positive attitude towards Hugh Grant than males had. This could be seen in the total effects, as there was a 46.9% difference in the effect on brand uplift between George Clooney (highest) and Hugh Grant within the female target group and only 2.77%
difference in the male target group. While it still confirms the estimated model, it is limited, as it is not possible to say if the difference is due to the used celebrities or that personality means more for women than men. While in this study it is assumed to be caused by the given celebrities, but it can also be argued that personality attribute is more important for women and thereby affect the attitude to a higher degree. Further research to cover this question is therefore suggested.
The results section showed an evident limitation as the unknown model (John smith) had gained misinterpreted answers by the respondents, which biased the results. This was thus also discussed during the presented results. Whereas this limits the possibility to statistically benchmark and prove how personality of a celebrity can have a more positive effect then an unknown model, this
116 study still provided strong indication of this conclusion and has been reinforced by other referenced theories.
Throughout the study several theoretical arguments have been made. While there are used a high amount of difference theories to make arguments as strong as possible and thereby valid, it must still be seen as a limitation, when it is not empirically tested. This also, as it must be assumed that the arguments are affected by the limitations in the different theories used to create some of these arguments. This form of limitation also is present in the theoretical suggested model of how celebrity endorsement works (figure Figure 9 - Conceptual Model).
The sample size has not been as immense as one would wish, when trying to say something generally. The sample is also grossly represented by young academics, which may have biased the results, and limited its validity.
Other factors have been mentioned numerous times and it is certain that other factors influence brand attitude and brand uplift. Its impact has both been theoretically argued and empirically tested. Though this has its limitations as the test is based on weak statistical performance and the results here from should, at its best, only be seen as indicators. The indicators are though backed up through the theoretical arguments, which make them relevant as a presented discovery. This limitation is highly relevant and the findings provide an indication of this thesis only reinforce the importance of further research into this area.