Predicting EFFIE, Gouden Loeki and Loden Leeuw Winning
TV Ads by EEG Time Series Analysis
Yuhee Kim (10859969)
MSc in Brain and Cognitive Sciences, University of Amsterdam, Cognitive Neuroscience Track
Supervisor: dhr. dr. H.S. (Steven) Scholte
Co-assessor / UvA Representative: dhr. prof. dr. V.A.F. (Victor) Lamme
University of Amsterdam, Psychology / Brain & Cognition
Abstract
The aim of this research is to analyze the changes of EEG frontal and parietal activity during the natural observation of TV ads that won an EFFIE, Gouden Loeki or Loden Leeuw (renowned commercial awards in the Netherlands) or never won any commercial awards in past years. In detail, we were interested in analyzing the dynamics of brain activity of remembered and forgotten, likable and annoying, effective and ineffective TV ads that were explicitly rated in questionnaires by participants. We would like to link significant variation of EEG signals with a degree of attention, emotion, memory and the sense of effectiveness aroused by different TV ads, thus, we compared changes of EEG signal to results from questionnaires rating subjective memorability, likability and effectivity. From the continuous EEG data collected from 64 EEG channels, we chose 6 cortical ROIs (Visual Detection, Auditory Detection, Central Attention, Attention Control, Right Frontal, Left Frontal) and extracted AUC, area under the curve, information of each ROI based on three time domains (Early, Middle and Late) for every commercial for all subjects. All AUCs were averaged out among subjects and the mean AUC were used to perform the Kruskal-Wallis testing and Mann-Whitney testing to examine the difference between awards categories. Additionally, 2 independent sample t-tests were employed to find a correlation of mean AUC and a degree of attention, emotion and sense of effectiveness. The results revealed that Gouden Loeki winning TV ads caused the biggest mean AUC compared to other categories. Besides that, the result showed a significantly high mean AUC in all ROIs while watching remembered TV ads compared to forgotten TV ads. However, no significant difference were found from likable and annoying, effective and ineffective TV ads, hence it was not feasible to find a distinct EEG signal for these categories in the present experiment design.
1. Introduction
TV as the most effective advertising medium
Revenue from TV advertising worldwide will reach 172.5 billion USD and are anticipated to grow to 204.1 billion USD by 2019 according to Statista. These figures reflect how crucial TV advertisement is as a final product of strategic marketing planning. Within 15 seconds to a minute, it must grab the viewer’s attention as well as imprint a product and brand into their memory in order to make viewers become potential and loyal consumers. In spite of rapid changes of media platforms, TV advertising still appears
to be as effective as ever and operates a crucial role to enhance brand awareness to consumers (Rubinson, 2009). Therefore, advertisers and their agencies seek to assess the influence of their TV ads since their success can be a potential indicator of the success of the product and brand.
Prize for commercials
The enormous impact of TV advertisement encourages to establish the awards for well-made commercials. It is assumed that around 500 advertising awards are presented worldwide per year (Shamoon, 1987) but it has been criticized that most of advertising awards focus on its creativity only like “beauty contests” (Moriarty, 1996) instead of evaluate the actual influence of TV ads. This finding motivated the establishment of the Effectiveness awards (EFFIE), which not only judge creativity of TV ads but also “recognizes any and all forms of marketing communication that contribute to a brand's success” according to Effie Worldwide. Some studies prove that there is still a high correlation between award winning TV ads and creativity index (Ang & Low, 2000; Goldberg & James, 1994; Stone, Besser, & Lewis, 2000) but EFFIE is the only award that judges the overall effectiveness derived from entire marketing and advertising processes.
Gouden Loeki and Loden Leeuw are domestic TV advertisement awards in the Netherlands and both appraise the pleasantness (Gouden Loeki) or annoyingness (Loden Leeuw) caused from TV ads by votes from general public. Those awards simply consider the assessment of aroused emotion by merely watching the TV ads, so they are not entirely sufficient to measure the overall achievement as EFFIE is. In other words, a Gouden Loeki does not always guarantee the success of brands and products and a Loden Leeuw does not always assure the failure of brands and products. For example, in 2011, a German online shoe store “Zalando” won an EFFIE and Loden Leeuw altogether. Zalando’s Dutch TV ads succeeded to increase brand awareness in the market even though the TV ads turned out to irritate public according to PROFNEWS.
Those findings reconfirm why advertisers and their agencies invest a considerable amount of money and efforts on TV ads and their pre-test before launching.
Traditional methodologies to measure commercial influence
Advertisement pre-test is usually performed before its launching in public to predict the prospect. Conventionally, questionnaires and verbal interviews have been conducted right after the subject’s exposure to new commercials to estimate the success, but it is now recognized that traditional methodologies contain inevitable flaws. For example, non-completion of questionnaires or insincere response leads to the collection of
inaccurate data, which may impair the reliability of the results. Moreover, the process and analysis of interviews have been criticized because of their artificiality and subjectivity (Drennan, 2003). In addition, depending on the various verbal ability of the respondents, the real response could be underestimated or overestimated. Hence, these techniques are, potentially, uncertain to anticipate the success of TV advertisements.
Brain imaging to measure commercial influence
Neuroscience techniques like fMRI and EEG have received attention as promising alternatives for advertisement pre-testing instead of conventional techniques. In fact, it has been well known that hemodynamic measurements like fMRI can capture brain activity from deep structures like amygdala and nucleus accumbens on the level of cubic mm. However, fMRI is an expensive technique and hard to set a natural and viewer-friendly condition for the subject due to its mechanical construction. In contrast, EEG is relatively cheap and convenient to set the equipment under any circumstances. However, a main concern of using EEG technique is that it is limited to measure activity generated on the cortical structure of the brain and impractical to capture activity from subcortical structure through superficial EEG electrodes (Urbano et al., 1998) To overcome its weakness, many researchers have attempted to find a distinctive cortical activity which can be detected through high resolution EEG while the subject is watching TV ads. Eventually it was revealed that attention and emotional engagement of the viewers form a certain attitude toward TV ads and the newly formed attitude have an impact on memory formation (Klucharev, Smidts, & Fernandez, 2008). Hence, researchers are mainly focused on finding a correlated cortical brain activity with an increase of attention, emotional engagement and memory during the observation of TV advertisement using EEG (De Vico Fallani et al., 2008; Thompson, 2014; Vecchiato et al., 2010, 2011). In 2008, Fallani and his colleagues found out that the successful memory encoding are regulated by different frequency bands in the cortex. Vecchiato and his colleagues also observed an asymmetrical increase of theta and alpha activity in the left hemisphere related to the observation of pleasant TV ads. These results suggested that EEG is a suitable technique to measure genuine responses of viewers regard to TV ads and which can lead to a following research to measure the dynamics of cortical activation elicited by prize winning TV ads through EEG.
In this scenario, its purpose is to investigate the variation of cortical activation in frontal and parietal lobe depending on different awards winning commercials. In particular, the aim of this study is to trace the dynamics of the frontal and parietal cortex
while subjects watching TV ads, and to find a correlation between cortical activation and different awards to discover distinctive neuronal features. Moreover, we would like to examine the enhanced activation of the left frontal cortex compare to the right frontal cortex when participants watching Gouden Loeki or EFFIE TV advertisements to confirm if the present study can imitate the result of previous study (Vecchiato et al., 2011) since the asymmetrical activation of frontal lobe with positive engagement from TV ads was discovered.
Therefore, the main questions in this study are addressed as below:
(1) Is it possible to find a specific cortical activation that can predict the types of TV advertisement awards reflecting pleasantness, annoyingness or effectiveness?
(2) Is there any higher EEG left frontal cortex activity when experiencing of pleasantness or effectiveness from TV ads?
2. Material and Method
Participants
10 male and 10 female healthy undergraduate and graduate students of the University of Amsterdam whose average age was 22 (SD = 2.59) were participated in the study. They were not aware of the purpose of the experiment but instructed that they will watch their favorite TV sitcom during the experiment. A list of sitcoms was provided to the potential subjects in advance and we only recruited the subjects showing their interests to the sitcoms in the list. Written informed consent was gathered from each subject which was approved by ethical committee of the Faculty of Science, University of Amsterdam. These subjects have no personal history of neurological or psychiatric disorder. They had normal or corrected-to-normal vision and were free from medications, alcohol or drug abuse. For the EEG data acquisition, subjects were comfortably seated on a reclining chair in an electrically shielded room and watched stimuli which presented on a 1920*1080 resolution HD screen monitor. They were exposed to the series of sitcoms and commercials for about 90 minutes and asked to pay attention to the stimuli. Subjects did not know that they had to fill in the questionnaires after the EEG recording.
The stimuli of the experiment were programmed by Presentation® (Neurobehavioralsystems, Berkely, US) and the procedure consisted in observing 9 video clips that are edited from 3 episodes of the sitcom. In addition, 10 advertising breaks were inserted between the video clips to provide a viewer-friendly environment to participants for maintaining an ecological validity. On the presentation program, every stimulus was set to send a trigger on the onset and at the end of their displaying to recognize the presence of each stimulus in EEG records. Moreover, 3 seconds black screens were embedded between all video clips and commercials to clearly separate each stimulus in EEG data.
Sitcoms that have been televised on Dutch TV channels were used (“Family Guy”, “How I Met Your Mother”, “Seinfeld”, “The Big Bang Theory”, “Two and a Half Men”) and subjects could choose one of the sitcoms they wanted to watch during the experiment. Top 5 funniest episodes of each sitcom according to the Internet Movie Database (IMDb, www.imdb.com) were used in order to keep viewer’s concentration. Each episode lasted on average 20 minutes and was edited to three video clips for an average of 7 minutes. The order of sitcom video clips was identical for all subjects to follow the storyline.
A total of 52 commercials were showed during 10 advertising breaks and randomly played for each subject. Among these 52 commercials, 13 commercials are winners of EFFIE (EF), another 13 commercials of Gouden Loeki (GL), and the third 13 commercials of Loden Leeuw (LL) between 2009 and 2014. Lastly, the other 13 commercials which had never won any awards beforehand were included as “Neutral (NT)” condition. Most of the commercials were composed of different brands but some of them overlapped, thus, we restricted the whole commercials to include a maximum of two commercials of the same brand. In addition, every commercial has a different duration: for example, the shortest commercial lasts 15 seconds whereas the longest 60 seconds. To rule out the effect from differences in duration, an independent sample t-test for average duration between groups was conducted which showed no significant difference confirmed. Every advertising break was composed of 5 randomly assigned commercials by Presentation® except the very first break which consisted of 7 random commercials and it was consistent for all participants.
After the end of the EEG recording, each experimental subject was asked to fill in two types of questionnaires. To minimize the influence of face to face interview, the experimenter was only involved in explaining the purpose of the questionnaires at the beginning and subjects filled in the questionnaires freely with sufficient amount of time. In this session, the experimenter required the subjects to recall the commercials they remembered and to judge them on 5-point scales. The first questionnaire is to assess a
degree of subjects’ spontaneous recall about commercials. Subjects were asked to describe the commercials inserted between video clips as many as they remembered in a written form in terms of brand name, kind of products, and scenery. To encourage less pressure and more freedom, the length of description was unlimited. Successively, the second questionnaire is to estimate an extent of subjective perception of viewers about TV ads. As a cue to help recollection, a collection of images related to commercials were provided when subjects answered and rated the questionnaire. Cues were composed of three captured images of 52 commercials used in the EEG recording session and three images of 48 random commercials which we did not use previously as distractors. All images were presented on Microsoft PowerPoint presentation program and pages which contained three sequential images of each commercial is randomly presented. Subjects were required to answer three questions for every TV ad. First, we asked if they remember watching the commercial during EEG recording. If they did, they also had to answer the next two questions. If they did not, however, they could skip the next two questions and move to the questions for next commercial. For TV ads subjects answered they had remembered watching, we asked them to rate ranging from 1 to 5 according to the level of likability they felt during the observation (1, annoying; 3, indifferent; 5, likable). Additionally, we also requested them to score between 1 and 5 according to the level of effectiveness they perceived (1, ineffective; 3, indifferent; 5, effective).
EEG recordings and signal processing
EEG was amplified using the BIOSEMI Active-Two amplifier system and recorded at a sampling rate of 512 Hz from 64 pin-type active electrodes mounted in an elastic cap. Electrodes positions were evenly distributed over frontal, central, parietal, occipital, and temporal areas according to the extended 10–20 EEG system (Oostenveld & Praamstra, 2001). These 64 electrodes included conventional midline sites with FPz, AFz, Fz, FCz, Cz, CPz, Pz, POz, Oz, Iz electrodes; Fp1, AF3, AF7, F1, F3, F7, FC1, FC3, FC5, FT7, C1, C3, C5, T7, CP1, CP3, CP5, TP7, P1, P3, P5, P7, P9, PO3, PO7, O1, I1 electrodes in the left hemisphere; the homolog/even recording sites in the right hemisphere. Two other electrodes, the common mode sense [CMS] active electrode and the driven right leg [DRL] passive electrode, were used as a reference and a ground electrode, respectively (http://www.biosemi/faq/cms&drl.htm). Furthermore, six flat-type active electrodes were placed over right and left earlobes, right and left temples for monitoring horizontal eye movements, and above the left eyebrow and beneath the left eye for monitoring vertical eye movements and blinks.
The raw data were converted into the Brain Vision Analyzer software version 2.0.2 (Brain Products GmbH, Munich) in order to perform signal pre-processing such as artefacts detection, filtering and segmentation. The raw data for each subject was digitally filtered with a 0.1–128 Hz band-pass filter and the Ocular Correction Independent Component Analysis and Raw Data Inspection were applied to detect and remove unnecessary components such as eye movements, blinks and muscular artefacts. Then, the data were segmented to obtain EEG activity during the observation of each TV commercial. Segments included the original length of each commercial as well as 2 seconds before the onset of the commercial and 2 seconds after the end of the commercial to confirm that they contain the onset and the end markers (For example, if a duration of commercial A is 40 second, a length of EEG segment for commercial A is 44 seconds). However, these unnecessary parts were excluded when data was exported.
Lastly, we generated pools, as ROI, from the existing channels to validate the elevated activation within the designated cortical areas. Hence, each EEG segment was pooled and grouped in BVA according to following ROIs; Visual Attention or VA (Oz, O1, O2, Iz, I1, I2); Auditory Attention or AA (T7, T8); Central Attention or CA (P1, P2, P3, P4); Attention Control or AC (Fz, FCz, Cz, FC3, FC4, C3, C4); Right Frontal or RF (Fp2, AF8, AF4, F8, F6, F4, F2); Left Frontal or LF (Fp1, AF7, AF3, F7, F5, F3, F1).
Exporting and Segmenting Data
We computed an absolute value of area under the curve, AUC (µV * ms), information of 6 ROI (Visual Attention, Auditory Attention, Central Attention, Attention Control, Right Frontal, Left Frontal) based on three time domains (Early, Middle and Late) for every commercial for all subjects and exported them into numeric form. According to previous studies, analysis of the area under the curve in specific time ranges can provide significant differences (Merzagora et al, 2007). The area under the curve is a useful statistic for the performance analysis, especially, of classification, ranking algorithms and critical features (Hill et al., 2006). For example, AUC scores from EEG provide a quantitative indicator for accuracy when knowledge tracing was measured or status of coma was tested (Lee et al., 2010; Xu, Chang, Yuan, & Mostow, 2014).
When AUC of each ROI is computed, it was exported into three datasets depending on different time windows as “Early”, “Middle” and “Later”. Importantly, the first and last two seconds were excluded and only the original commercial duration was considered since they were added to check the markers in the early stage. “Early” data contained the first one third portion of the segment, “Middle” the second, and “Later” the last. The reason we divided each segment into three different categories is the viewer’s engagement in
advertising which follows cognitive processes like orienting responses, selective listening, processing level, cognitive elaboration and persuasion as time passes (Greenwald & Leavitt, 1984). Therefore, we assume that brain activation will be actively changed depending on viewer’s involvement in advertising with the passage of time.
EEG signals were also segmented and classified according to EF, GL, LL and NT conditions to find a neuronal signature of different awards. Furthermore, for an in-depth analysis, EEG signals were divided into six different conditions in line with answers from two questionnaires. The first condition was related to EEG signals that collected from the TV ads most of subjects correctly remembered in the questionnaire, and this was titled REMEMBER. The second condition was EEG signals that captured from the TV ads subjects forgot and rarely answered in the questionnaire and it was called FORGET. The third condition was formed by the brain activity captured from the TV ads that recognized as likable and was named LIKABLE. The fourth condition consisted of the brain activity collected from the TV ads that considered as annoying and was called ANNOYING. The fifth condition comprised EEG signals from the TV ads that regarded as effective and was named EFFECTIVE. Lastly, the sixth condition was composed of EEG signals from the TV ads considered as ineffective and we referred this condition as INEFFECTIVE.
Statistical Analysis
For EEG analysis, 18,720 (52 commercials * 20 subjects * 6 pools * 3 time windows) data values were created from a previous stage. The averaged AUC value from 20 subjects was calculated using Microsoft Excel program and the averaged values were shifted to SPSS program for further analysis. In order to evaluate the variations of cortical activation during the observation of the different award winning commercials, Kruskal-Wallis testing was employed to test a significant difference among all four groups and Mann-Whitney testing with a significant level of 0.05 was employed to compare between two groups. In addition, 2 independent sample t-tests were applied to evaluate the differences between groups of REMEMBER-FORGET / LIKABLE – ANNOYING / EFFECTIVE – INEFFECTIVE with a significant level of 0.05. Same statistical analysis methods were conducted in all three time groups.
For questionnaire analysis, only 14 samples were collected out of 20 subjects. For spontaneous recall data analysis, subjects remembered a different number of commercials and sometimes failed to recall them correctly like mismatching a brand and a product or mixing two commercials in one description. Those incorrect answers were excluded and only correctly recalled commercials were counted for analysis. The most frequently reported commercial was mentioned 10 times and the least frequently
answered commercial answered 0 time. Therefore, commercials that were recalled 8, 9 and 10 times belonged to REMEMBER condition and commercials answered 0, 1 and 2 times belonged to FORGET condition. For subjective judgment data analysis, scores for each questions were averaged out within subjects and mean numbers were calculated. Commercials with a mean score of more than 4 points belonged to LIKABLE and EFFECTIVE conditions and commercials with mean scores of less than 2 points belonged to ANNOYING and INEFFECTIVE conditions.
3. Result
As described previously, the mean AUC from all subjects was obtained in each award condition (EF, GL, LL and NT) and in each questionnaire condition (REMEMBER-FORGET / LIKABLE – ANNOYING / EFFECTIVE – INEFFECTIVE) for 6 ROIs (Visual Attention, Auditory Attention, Central Attention, Attention Control, Right Frontal, Left Frontal, Right Frontal) for 3 time windows (Early, Middle and Later). With the analysis of variation of the average values, we drew several results below.
First, the result of Kruskal Wallis analysis indicates that there is a significant difference between award categories in late time domain (VA:
(3) = 9.93, p = .019; AA: H (3) =
10.97, p = .012; CA: (3) = 9.56, p = .023; AC: (3) = 9.31, p = .025; RF: (3) = 11.02, p = .012; LF: (3) = 9.43, p = .024), with a mean of 36.69, 37.23,36.46, 36.77, 37.54 and 37.31 for each ROI of GL condition and of 19.31, 20.85, 19.46, 19.92, 19.00 and 20.92 for each ROI of EF condition. (Figure 1). To be specific, a mean of GL, NT, LL and EF winning TV ads were ranked in the order named and this order was consistent in most of ROIs in late time domain (Figure 2). To compare the difference between award conditions clearly, Mann-Whitney testing was performed between 2 awards in late time domain (EF vs GL / EF vs LL / EF vs NT / GL vs LL / GL vs NT / LL vs NT). The result indicates that GL winning TV ads showed a significantly higher mean AUC in all ROIs over LL (VA: U = 37, p = .014; AA: U = 29, p = .003; CA: U = 38, p = .016; AC: U = 39, p = .019; RF: U = 28, p = ,003; LF: U = 27, p = .002) and EF(VA: U = 32, p = .006; AA: U = 30, p = .004; CA: U = 31, p = .005; AC: U = 29, p = .003; RF: U = 32, p = ,006; LF: U = 39, p = .019) winning TV ads. However, no significant difference was detected in any other comparisons consisting only within NT, LL and EF winning commercials. Therefore, we confirmed that GL winning TV ads has a distinctive AUC signature over the other TV ads. More importantly, this results were only found in “Later” time window.Secondly, TV ads in the REMEMBER condition regardless of awards have significantly higher mean AUC in late time domain (VA: M = 29.06, SD = 10.21 ; AA: M = 33.70, SD = 12.87; CA: M = 36.22, SD = 11.48; AC: M = 33.42, SD = 10.39; RF: M = 37.20, SD = 10.87; LF:
M = 35.48, SD = 13.85) compared to the FORGET condition AUC (VA: M = 21.27, SD = 4.80;
AA: M = 20.76, SD = 4.67; CA: M = 24.06, SD = 5.31; AC: M = 22.41, SD = 5.17; RF: M = 26.92,
SD = 6.74; LF: M = 26.11, SD = 6.58) in all ROIs according to 2 independent sample t-test
(VA: t(13.48) = 2.34, p < 0.05; AA: t(12.09) = 3.16, p < 0.05; CA: t(13.36) = 3.25, p < 0.05; AC: t(23) = 3.47, p < 0.05; RF: t(23) = 2.90, p < 0.05; LF: t(23) = 2.24, p < 0.05). Specifically, central attention ROI showed the highest different mean AUC between conditions and these result is also verified in the late time window only that is in line with the first result. Therefore, no significant difference of mean AUC between REMEMBER condition and FORGET condition in any ROIs is found in “Early” and “Middle” time window (figure 3).
Discussion
This study is aimed to evaluate the changes in the frontal and parietal cortical activation which can be distinguished by the AUC collected from EEG while participants watch different award winning TV ads. We assumed that award winning TV ads could capture more attention and emotional engagement of viewers and different kinds of awards could generate a various degree of attention arousal, emotional engagement, memory capacity and sense of effectiveness that could be controlled in cortical level. With the experimental design adopted, the changes of AUC were clearly confirmed in the frontal and parietal cortex while subjects experience 1) Gouden Loeki winning commercials and 2) spontaneously remembered commercials. These two conditions can be characterized by a significantly higher AUC of frontal and parietal cortex compared to the other conditions. It is worth of note that the cortical activity reflected by AUC in frontal and parietal cortex is strongly related to the observation of commercials that have been judged funny and catchy or naturally remembered by the participants. These results are congruent with the finding of Fallani and his colleagues (2008) which confirmed the increase of connectivity from Brodmann areas 5(5_L and 5R) to Brodmann area 7_L while subjects watching memorable TV ads. Moreover, the other important finding of this study is EFFIE winning TV ads were not sufficient to boost the AUC of frontal and parietal cortex as Goden Loeki winning commercials were. This result can imply that AUC could be modulated by the pleasantness of TV ads but not the effectiveness of TV ads. However, some studies warned a ‘vampire effect’, a situation that viewers get distracted when processing brand benefits
by humors in commercials (Eisend, 2011). Moreover, even though EFFIE winning commercials did not elicit the AUC from the cortex enough, 4 EFFIE commercials used in the experiment were recalled 8.5 times by subjects in the spontaneous questionnaire. These findings may suggest that favorability caused by humors in TV ads is critical to increase the AUC in cortical level but it can distract the viewer’s attention in some cases and there should be other factors which also could imprint TV ads on viewer’s mind.
There are some new findings and limitations of the present study.
First, a new finding is that a significant difference was only demonstrated in the “Later” time domain. Regardless of the duration of commercials, difference was not existed in the “Early” and “Middle” time domain which could be interpreted that cognitive perception of commercials can be fluctuated while watching TV ads and completed at a later stage. This finding is in line with argue of Greenwald (Greenwald & Leavitt, 1984) that viewer’s involvement in advertising develops as time passes. Previous studies did not divide the commercials according to the lapse of time, therefore, considering the division of time lapse is recommended for future research with commercials.
Second, some limitations exited in the present research. Previous studies used TV ads that participants never had watched. Therefore, stimuli were under the controlled condition and the result can be interpreted clearly without prejudice. However, the present study used stimuli that may have a chance to be seen by subjects before. This can be interpreted that the number of exposures of each commercial can be different among subjects. Moreover, there are brands which appeared twice in the experiment since they won the awards and this can affect to memorization of commercials as well.
Lastly, previous studies implemented a time frequency analysis to prove EEG frontal asymmetries and Vecchiato and colleagues (2011) confirmed the increased theta and alpha band for the dislike commercial condition. Similar result was shown in the present study by a mapping view in BVA but an absolute mean value of AUC in the left frontal and right frontal ROIs does not reveal any difference and it is not feasible to extract the different frequency from AUC data. Hence, it is impossible to replicate the result of previous research from the present study. Additionally, comparing a degree of likable/annoying and effective/ineffective conditions with the mean AUC was not performed due to the small number of qualified data for the sake of many neutral responses to the questionnaires.
In conclusion, in the light of the result we obtained, the questions posed at the beginning can be answered by stating that there is an AUC signature which can be elicited by a certain kind of prize winning TV ads and captured in the cortical level. Moreover,
high AUC were confirmed by memorable commercials regardless of awards they won previously. Therefore, we could state that a degree of perceived likability and a recall capacity evoked by nature of TV ads can be predicted by AUC and this methodology could provide more intrinsic character about TV ads to their advertisers and agencies than conventional research methods do.
Figure 1. Mean AUC (µV * ms) of GL and EF in 6 ROIs (Visual Attention / Auditory Attention / Central Attention / Attention Control / Right frontal / Left Frontal) in late time window.
Figure 2. Mean AUC (µV * ms) of 4 awards (EF / GL / LL / NT) condition in 6 ROIs (Visual Attention / Auditory Attention / Central Attention / Attention Control / Right frontal / Left Frontal) in 3 time windows (Early / Middle / Late)
Figure 3. Mean AUC (µV * ms) of remembered and forgotten TV ads in 6 ROIs (Visual Attention / Auditory Attention / Central Attention / Attention Control / Right frontal / Left Frontal) in 3 time windows (Early / Middle / Late)
Reference
Ang, S. H., & Low, S. Y. M. (2000). Exploring the Dimensions of Ad Creativity. Psychology
& Marketing, 17(10), 835–854.
http://doi.org/10.1002/1520-6793(200010)17:10<835::AID-MAR1>3.0.CO;2-
De Vico Fallani, F., Astolfi, L., Cincotti, F., Mattia, D., Marciani, M. G., Gao, S., … Babiloni, F. (2008). Structure of the cortical networks during successful memory encoding in TV commercials. Clinical Neurophysiology : Official Journal of the International
Federation of Clinical Neurophysiology, 119(10), 2231–2237.
http://doi.org/10.1016/j.clinph.2008.06.018
Drennan, J. (2003). Cognitive interviewing: Verbal data in the design and pretesting of questionnaires. Journal of Advanced Nursing, 42, 57–63.
http://doi.org/10.1046/j.1365-2648.2003.02579.x
Eisend, M. (2011). How humor in advertising works: A meta-analytic test of alternative models. Marketing Letters, 22, 115–132. http://doi.org/10.1007/s11002-010-9116-z
Goldberg, S. M., & James, W. L. (1994). Creativity Vs . Journal of Advertising.
Greenwald, A. G., & Leavitt, C. (1984). Audience Involvement in Advertising: Four Levels.
Journal of Consumer Research, 11(1), 581. http://doi.org/10.1086/208994
Hill, J., Lal, T. N., Schroeder, M., Hinterberger, T., Widman, G., Elger, C., … Birbaumer, N. (2006). Classifying Event-Related Desynchronization in EEG, ECoG and MEG
signals. Pattern Recognition, 4174, 404–413. http://doi.org/10.1007/11861898_41 Merzagora, A. C., Izzetoglu, M., Bunce, S., & Onaral, B. (2007, August). Time-domain
analysis of EEG during Guilty Knowledge Test: investigation of epoch extraction criteria. In 2007 29th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (pp. 1302-1305). IEEE.
Klucharev, V., Smidts, A., & Fernandez, G. (2008). Brain mechanisms of persuasion: how “expert power” modulates memory and attitudes. Social Cognitive and Affective
Neuroscience, 3(4), 353–366. http://doi.org/10.1093/scan/nsn022
Lee, Y. C., Phan, T. G., Jolley, D. J., Castley, H. C., Ingram, D. a, & Reutens, D. C. (2010). Accuracy of clinical signs, SEP, and EEG in predicting outcome of hypoxic coma: a meta-analysis. Neurology, 74(7), 572–80.
http://doi.org/10.1212/WNL.0b013e3181cff761
Moriarty, S. E. (1996). Effectiveness, Objectives, and the Effie Awards. Journal of
Advertising Research, 36(4), 54–65.
Oostenveld, R., & Praamstra, P. (2001). The five percent electrode system for high-resolution EEG and ERP measurements. Clinical Neurophysiology, 112(4), 713–719. http://doi.org/10.1016/S1388-2457(00)00527-7
PROFNEWS. (2012). Zalando wins bronze Effie and gets Loden Leo. Retrieved from http://www.profnews.nl/985162/zalando-wint-bronzen-effie-en-krijgt-loden-leeuw
Rubinson, J. (2009). Empirical Evidence of TV Advertising Effectiveness. Journal of
Advertising Research, 49(2), 220–226.
http://doi.org/10.2501/S0021849909090321
Shamoon, S. (1987). “The Great Advertising Award Machine.” Winners (February). Statista. (2015). Global television advertising revenue 2015. Retrieved from
http://www.statista.com/statistics/237803/global-tv-advertising-revenue/ Stone, G., Besser, D., & Lewis, L. E. (2000). Recall, liking, and creativity in TV
commercials: A new approach. Journal of Advertising Research, 40(3), 7–16. Thompson, C. J. (2014). Journal of Consumer Research, Inc., 31(4), 162–180.
Urbano, A., Babiloni, C., Onorati, P., & Babiloni, F. (1998). Dynamic functional coupling of high resolution EEG potentials related to unilateral internally triggered one-digit movements. Electroencephalography and clinical Neurophysiology, 106(6), 477-487.
Vecchiato, G., Astolfi, L., De Vico Fallani, F., Cincotti, F., Mattia, D., Salinari, S., … Babiloni, F. (2010). Changes in brain activity during the observation of TV commercials by using EEG, GSR and HR measurements. Brain Topography, 23(2), 165–179. http://doi.org/10.1007/s10548-009-0127-0
Vecchiato, G., Toppi, J., Astolfi, L., De Vico Fallani, F., Cincotti, F., Mattia, D., … Babiloni, F. (2011). Spectral EEG frontal asymmetries correlate with the experienced
pleasantness of TV commercial advertisements. Medical & Biological Engineering &
Computing, 49(5), 579–583. http://doi.org/10.1007/s11517-011-0747-x
Xu, Y., Chang, K., Yuan, Y., & Mostow, J. (2014). EEG Helps Knowledge Tracing. …
Workshop on Utilizing EEG Input in …. Retrieved from
http://www.cs.cmu.edu/afs/cs/Web/People/listen2/pdfs/EEG-KT-camera-ready.pdf
