UNIVERSITY OF GRONINGEN
SUPERVISOR: PROF. DR. T.H.A. BIJMOLT
SECOND SUPERVISOR: A. MINNEMA
To Zap or Not to Zap: that
is not always the question
Zapping and brand exposure as determinants of
brand recall
R.M. Koole 7-7-2013
1
The commercial advertising world is a multi-billion dollar industry, so the stakes of its effectiveness are high. The aim of commercial advertising is to increase sales and long-term brand equity by generating and sustaining brand awareness, associations, and attitudes (Texeira, Wedel, and Pieters 2010). However, the advertising landscape changed dramatically over the last decades and with that the recipe for effective commercial advertising. First of all, although official numbers do not exist, it is widely believed that consumers are being exposed to more and more advertisements. For example, a business report stated that in the 70's an average customer was exposed to 500 to 2.000 ads per day, while in 2004 this number increased to between 3.000 and 5.000 ads (Wegert 2004). So it is safe to say that consumers are overloaded with advertising clutter (Rumbo 2002).Breaking through this clutter to reach customers is a difficult challenge for advertisers these days. One of their worries is commercial avoidance strategies, a defense mechanism that is activated by the overload of exposures (Speck and Elliot 1997). Several technical developments made it easier for consumers to execute such strategies. For example, the emergence of the Videocassette Recorder (VCR), Remote control devices, and more recently Interactive Television (iTV) changed the role of consumers when watching television commercials from passive to active viewers. So an increase in commercial avoidance need combined with an increase in commercial avoidance possibilities makes avoidance strategies an important phenomenon in commercial advertising. The effect of this phenomenon on advertising effectiveness has been of interest for both practitioners and academics.
One of the components of advertising effectiveness that received significant attention is brand recall. Brand recall is believed to be an essential part of the consumers’ decision-making process, since a brand needs to be considered first in order to be chosen (Baker et al. 1986). In other words, it can be seen as a vital first step towards desirable outcomes. For example, an increase in recall is associated with an increase in perceived quality (Hoyer and Brown 1990), purchase intention (Nedungandi 1990), and actual sales (Woodside and Wilson 1985). Therefore, brand recall and its determinants are an important topic in both practice and research.
2
performed on a commercial level due to the influence of commercial overload. Non-zapping respondents are exposed to more commercials and might therefore remember a smaller percentage of the brands they saw. Whereas zapping respondents are exposed to less commercials and thus experience less overload, leading to a higher score on brand recall for the commercials they did not zap. However, since zapping respondents saw a smaller amount of commercials in absolute terms, their commercial block scores on brand recall might be quite similar to the non-zapping respondents. So the aggregation might positively bias the results and therefore the true relation on a commercial level is not reflected.An important part of the effect of avoidance on brand recall is not yet examined. This is remarkable since brand recall of a commercial, rather than of a commercial block, is of specific interest for research and practice due to several developments over the last years. First of all, many networks nowadays engage in so-called “road-blocking” of commercials, the time-synchronization of commercials on different channels. This holds that when a consumer zaps channels, it zaps from one commercial into another (Texeira, Wedel, and Pieters 2010). This implies that consumers are exposed to several individual commercials instead of commercial blocks. Another development is the emergence of online advertising. Online advertising holds individual commercials that are shown before, during, or after streaming. These developments demand a shift from a commercial block focus to an individual commercial focus. Therefore, examining the relationship between avoidance and recall on a more detailed level is of value. Developments in measurement methods make it possible to collect moment-to-moment data on zapping, a well-known avoidance strategy. This allows us to study the effect on brand recall on a more detailed level. To our knowledge, this study will be the first to examine the effect of zapping and brand recall on a commercial level.
3
the effect of brand exposure on brand recall. Brand exposure is an important determinant of brand recall (Stewart and Koslow 1989; Texeira, Wedel, and Pieters 2010). However, its effect is often studied using forced exposure, thereby not permitting avoidance strategies. This is remarkable since avoidance directly influences whether a respondent is exposed to the brand. It will therefore be of interest to study the effect of brand exposure on brand recall in a more naturalistic setting that allows avoidance. By examining both avoidance and brand exposure we can study the effect of different partial exposure forms (i.e. partial exposure with or without brand exposure). Thereby, this study will allow us to examine the effect of different forms of partial exposure on recall instead of treating it as one global concept.The research question this paper will address is:
What is the influence of zapping and brand exposure per commercial on brand recall per commercial?
This research question will contribute to the existing literature in several ways. We will examine the influence of both zapping and brand exposure on brand recall. Thereby, it will add to the knowledge about the important topic of commercial advertising effectiveness and its determinants. By including these two determinants of brand recall, we will also be able to examine their relative effect and thus their importance. This will provide more information about the value of partial commercial exposure as well as about the relative differences between partial exposures. Another gap that is addressed by this study is related to the measurement method being used. By using both moment-to-moment zapping data and brand recall per commercial, we will zoom in on a more detailed level and expand the existing knowledge.
Also, previous studies often used self-reported measurements of zapping behavior. This has the disadvantage of being slow, intrusive, difficult to measure continuously, and may lead to mere measurement effects (Texeira, Wedel, and Pieters 2012). In this study we will use automated zapping detection that provides information on a moment-to-moment (per second) basis. This allows us to overcome the self-measurement bias, which is another advantage of this study.
4
that zapping is not a negative event per se, but should be examined in the light of different forms of partial exposure.Managerial relevance
Besides its contribution to literature, this study will also be of value for practitioners such as advertisers and television networks. Advertisers are mainly interested in the brand recall of their specific commercial rather than that of the whole commercial block. This research will be the first to provide such detailed information. Our results show that zapping can indeed lead to a decrease in recall on a commercial level, but only when it occurs before brand exposure. This indicates the importance of focusing on both zapping and brand exposure. In order to optimize brand recall the consumer needs to be exposed to at least audio brand exposure before zapping occurs. So advertisers should aim for this goal when creating commercials.
For television networks it is important to know what influences commercial effectiveness when setting the broadcasting time prices. Our results show that partial exposure of commercial can indeed be as effective in terms of brand recall as complete exposure. This indicates that also partially viewed commercials should be taken into account when setting the prices.
The following section will provide a short theoretical background about brand recall, zapping and, brand exposure. In this section we will also present our hypotheses. Then our methodology and model are discussed, followed by our results. We will conclude this paper with a discussion and suggestions for future research.
THEORETICAL BACKGROUND
Advertising effectiveness
5
The measurement of cognition is often done by using recall as a dependent variable. (Stewart et al. 1985; Bellman, Schweda, and Varan 2010).Aided and free recall
In this study we will use recall as a dependent variable. Recall refers to the respondents' memory for a particular advertising execution. The type of recall can be classified on two dimensions: firstly, the numbers and types of stimulus available to the respondent and secondly, the criterion for an acceptable response (Stewart et al. 1989). In this study we will use two different forms of recall, namely free and aided brand recall. For free recall the individual is asked to describe a stimulus that is not present, while for aided recall the stimulus that is verbally present to the individual must be identified as being seen or heard previously (Moorman, Neijens, and Smit 2007). The criterion for an acceptable response for both free and aided recall is whether the brand name can be recalled. This is a less strict criterion than what is used by for example proved recall, where the commercials needs to be described correctly in sufficient detail to be classified as being recalled (Brown 1985). Aided recall requires less learning than free recall and can therefore be achieved at a lower cost. This is supported by evidence that aided recall scores are significantly higher than free recall scores (Singh and Rothschild 1983). The acceptance of both memory measurements differs per field, and even per advertising form. For example in printed advertising, aided recall traditionally has been the dominant measurement method, while for broadcast media this method is less accepted (Singh, Rothschild, and Churchill 1988).
6
Moorman, Neijens, and Smit 2007). Therefore, in this study we will include both aided and free recall as dependent variables.Avoidance strategies
There are several variables that can influence the brand recall of a commercial. As already been mentioned, one of those variables that conceived significant attention over the last decades is commercial avoidance. There are several commercial avoidance strategies, which can be classified in three different forms; physical avoidance, mechanical avoidance (Abernethy 1990), and cognitive avoidance (Speck and Elliott 1997). Physical avoidance includes leaving the room when the commercials are on, mechanical avoidance includes switching channels and fast forwarding, and cognitive avoidance includes consumers being in the room or even watching the commercials, but being cognitively distracted. An example of cognitive avoidance is multitasking during commercial viewing, like reading or doing chores (Ehrenberg and Twymann 1967). With physical avoidance there is no exposure to the commercial, while for mechanical and cognitive avoidance partial exposure is permitted (Bellman, Schweda, and Varan 2010). In this study we are interested in the different forms of (partial) exposure caused by the moment of avoidance and their effect on brand recall. With cognitive avoidance partial exposure is possible, however, this is difficult to measure. For zapping, a form of mechanical avoidance, this is easier since it can be automatically detected when a button is pushed to stop watching. Therefore, it is possible to collect exposure information on a moment-to-moment basis. Since one of our goals is to examine the value of partial exposure, zapping is the avoidance strategy of interest for this study.
Zapping and brand recall
As already mentioned, there have been several studies that examined the influence of avoidance strategies on advertising effectiveness. The results of those studies show major differences in outcomes, not only between the types of avoidance strategies and advertising effectiveness included, but also between the same types studied as can be seen in table 1. Logic suggests that avoidance leads to lower commercial effectiveness due to a decrease in exposure. Therefore, it is believed that advertising effectiveness is in danger due to the increase in advertising avoidance (Gustafson and Siddarth 2007; Texeira, Wedel, and Pieters 2010). This is supported by several studies that report a significant decline in brand recall when commercial avoidance is present (e.g. Moorman, Neijens, and Smit 2007; Stout and Burda 1989).
7
much lower than was expected. He proposed that this was caused by the so-called “Providential Attention Back on Set”. This holds that “a commercial viewer has to really watch the set to see/know/perceive what she or he is doing and ends up with more commercial exposure value than we have been prone to grant” (Greene 1988, p.15). This implies that zapping might not be as negative to brand recall as is often believed, since zapping asks for attention to the screen and therefore leads to exposure. Bellman et al. (2010) found that the effect of zapping on brand recall was non-significant, which aligns with the theory of “Providential Attention Back on Set”. Based on the conflicting outcomes we can conclude that there is no consensus about the influence of avoidance on commercial effectiveness. Some claim the effect is negative, due to the decrease in exposure that is the consequence of zapping. Others claim that avoidance is an active decision that increases the attention towards a commercial, leading to positive outcomes. When reviewing the literature there are several components that might influence the outcomes and are therefore important to consider. First of all, the usage of forced versus free exposure. In the Bellman et al. (2010) study the non-zapping respondents were all in the forced exposure condition. The zapping respondents were placed in a free exposure condition and were told that it was possible to use zapping. However, the participants that decided not to zap any commercials were removed from the sample. Instead, the respondents in the forced exposure condition were indicated as non-zappers. Forced exposure respondent might be more reluctant and therefore less motivated to be involved in the commercials. This difference in involvement rather than the avoidance might explain the insignificance of the relationship with brand recall. Other studies provide evidence that forced exposure indeed leads to differences in results in comparison to free exposure (Texeira, Wedel, and Pieters 2010). Therefore, in this study we will only use free exposure in order to avoid involvement bias and to more accurately replicate real-life situations.Another component is the level of analysis being used. As we already discussed, this study will be the first to examine the relationship between zapping and recall on a commercial level. We expect a negative effect of zapping on brand recall, since we expect that less exposure leads to lower brand recall. This is contrary to some other studies that show a non-significant or even positive effect. However, we believe that the aggregation of the data leads to a positive bias. We do not have evidence that the relationship between zapping and brand recall is different for aided or free recall. Therefore, we propose the following hypothesis:
8
TABLE 1Relationship between avoidance strategies and recall
Study Hypothesis Method Avoidance measure effectiveness Advertising measure
Level of
analysis Commercial exposure Sample size Result Bellman,
Schweda and Vuran 2010
+/- Experiment Skipping: Automatic Zapping: Automatic Mute Audio only Zipping Aided recall Prompted recall Commercial
block Free and forced 86 n.s. n.s. n.s. - - Krugman, Cameron and White 1995
- Field study Eyes on screen:
observation Aided recall Free recall Commercial block Free 64 - Greene 1988 +/- Field study Zapping: Single-item
scale Proven recall Commercial block Free 4000 -
Moorman, Neijens and
Smit 2007
- Field study Attention: Single-item
scale aided recall Corrected
Free recall Proven recall
Commercial
block Free 344 -
Stout and
Burda 1989 - Experiment Zipping Aided recall Free recall Commercial block Forced 163 - Ehrenberg
and Twyman 1967
- Field study Cognitive avoidance: Survey Physical avoidance:
Survey
Free recall Commercial
block Free 3600 - -
Thorson and
Zhao 1997 - Experiment Eyes on screen: observation Aided recall Free recall Commercial block Forced 200 - Koole 2013 -/n.s. Experiment Zapping: Automatic Free recall
9
Brand exposure and recall
As already mentioned, brand exposure is another important determinant of brand recall. Branding activity is the exposure of a brand identity symbol in a commercial (e.g. brand name, logo, typeface, trademark, soundtrack). Based on these exposures the viewer is explicitly exposed to the brand. While other forms of brand identities are also possible, in this study we only take into consideration brand exposure: the revelation of the brand name. Brand exposure is very important for brand recall, since it helps viewers to call upon experiences and memories of the brand and view the commercial in that context (Texeira, Wedel, and Pieters 2010). In other words, brand exposure serves as an identification tool. Therefore, in general we do expect a positive effect of brand exposure on recall.
Audio and visual exposure
A brand can be exposed during a commercial in several ways. For this study we will make the distinction between visual and audio brand exposure, since during a commercial the brand is often exposed both visually and aurally. We expect that both audio and visual brand exposure positively influence recall. The inclusion of zapping allows for several possible brand exposure conditions: visual-only, audio-only, and both. This study will also examine the relative effect of these different conditions on recall in order to determine the superior brand exposure condition.
The dual-coding model suggests that exposure to both the audio and visual form is superior to only visual or only audial information, since the combined audio-visual format is easier to be retrieved from memory (Paivio 1986). This is confirmed by studies that found that visual-audio exposures are superior in terms of recall compared to visual-only exposures (Law and Braun 2000; Brennan and Babin 2004). Therefore, we believe that visual-audio brand exposure is superior to visual-only brand exposure and audio-only brand exposure.
According to the dual-coding model visual stimuli are more memorable because they are believed to activate both audio and visual codes, thereby providing more elaborate encoding and thus facilitating recall. Audio stimuli were indicated as being less elicited to memory (Paivio and Csapo 1969). This phenomenon of “picture superiority effect” is supported by other researchers as well (e.g. Babbit 1982; Childers and Houston 1984) and was also found to hold for dynamic stimuli such as commercials (Bryce and Yalch 1993).
10
stimulants than audio stimulants, therefore the addition of a visual stimulant (e.g. visual brand exposure) might have less power to the memory than an audio stimulant (e.g. audio brand exposure). Thus the “picture superiority effect” might not hold. This is supported by a study of Gupta and Lord (1998), who found that audio-only product messages manifest significant higher levels of recall than did subtle visual-only product messages, implying the superiority of audio exposure. However, this study was done in the field of brand exposure during movies rather than during commercials. Since brand exposure during commercials (intentional learning) is fundamentally different from brand exposure during movies (incidental learning) (Karrh, McKee, and Pardun 2003) we do not necessarily expect the same results.We conclude that there is no consensus in the literature about the superiority of either visual-only or audio-only. Hence, we will interpret it as an indication of the importance of studying the relative influence of audio-only and visual-only brand recall. Exploratory research is needed to examine which of the two forms is superior. Also we will need empirical evidence to find out whether the effects are the same for aided and free brand recall. Our hypotheses are:
H2: (a) Audio (b) visual brand exposure has a positive effect on free brand recall.
H2c: Audio and visual brand exposure has a stronger effect on free brand recall than audio-only or visual-only brand exposure.
H3: (a) Audio (b) visual brand exposure has a positive effect on aided brand recall.
H3c: Audio and visual brand exposure has a stronger effect on aided brand recall than audio-only or visual-only brand exposure.
Zapping and brand exposure
While the form of recall offers no basis for predicting differences in the separate effects of zapping and brand exposure, we do expect a difference when it comes to their relative effect. As already mentioned before, free and aided recall are two different forms of measuring memory learning. The memory learning process can be classified in three subsequent stages reflecting the various degrees of learning; encoding, storage, and retrieval (Tulving and Thompson 1973; Lang 1995). Aided recall indexes the second stage; how well the information is storaged and free recall is a measure of the third stage; the amount of information available for retrieval. Free recall therefore asks for a more complex learning process than aided recall since more stages need to be completed (Singh, Rothschild, and Churchill 1988).
11
the activated node. Also the probability of excitation is stronger for nodes that are related to a larger number of connecting nodes. The creation of nodes and their connection is accomplished by exposure to information, such as commercials. For free recall the number of nodes and their association is more important than for aided recall since for retrieval the specific nodes are more difficult to be excited because no cues are available (Srull 1981).When a commercial is zapped, the consumer is exposed to less information and thus creates less nodes and numbers of links to other nodes, which is especially harmful for free recall (Schmitt, Tavassoli, and Millard 1993). In order for the retrieval stage to be completed, a significant amount of information is needed and therefore an increase in exposure to information is beneficial. Brand exposure already includes some information, however if a respondent zaps there is still other information of the commercial left unexposed. Therefore, we expect that non-zapping leads to higher probability of free recall than zapping even if the zapping occurs after the brand exposure. Thus, we propose that zapping is a stronger determinant of free recall than brand exposure.
As already mentioned, for aided recall a less complex learning process in needed, since a cue is presented and thus only the storage stage has to be completed (Lang 1995). We expect that this stage can already be completed by brand exposure and therefore full exposure to a commercial is not a necessity. Partial exposure in which the brand is exposed might already be sufficient for aided recall, making brand exposure a stronger determinant than zapping.
The difference in the relative effect is supported by a study that found that exposure to more advertising components was especially valuable for the retrieval process and thus for free recall, while this same beneficial effect was not found for aided recall (Schmitt, Tavassoli, and Millard 1993). Our hypotheses are:
H4a: Zapping is a stronger determinant of free brand recall than brand exposure. H4b: Brand exposure is a stronger determinant of aided brand recall than zapping.
METHOD
Participants and data collection
12
respondents were included in the sample (40,2% female, 59,8% male; aged 18 till 70, average age 31). The commercials were showed in random order, to prevent primacy and recency effects. At each moment during a commercial, the respondents were allowed to zap to the next commercial by clicking on the "next" button on the screen. One commercial held only visual brand exposure and no audio brand exposure. The models were estimated with and without this commercial and yielded no significant differences. In our model we used the seven commercials, since we are interested in the effect of both audio and visual brand exposure.Measures
Free recall. The dependent variable free recall was measured by asking the respondents which
of the commercial they could remember seeing without giving any cues.
Aided recall. The dependent variable aided recall was measured by presenting the respondents
a list of 16 commercials, half of which had appeared and half which had not. The respondents were asked to indicate per commercial whether they recalled seeing it. (1: definitely, 2: not sure, 3: definitely not). We only classified the “definitely recalled” answer as aided recall to be more certain that the commercial was recalled.
Zapping. Zapping was measured by the recorded avoidance decision on a moment-to-moment
basis (per second). It indicates per second per commercial whether a respondent decides to stop watching the commercial and go to the next one (0: no zapping, 1: zapping). This variable was used to create a dummy variable per respondent per commercial (0: no zapping, 1: zapping).
Brand exposure. Each commercial was watched to determine the moment of audio and visual
brand exposure. Subsequently, it was determined per respondent per commercial whether or not he or she exposed to the brand (zapped before brand exposure=0, zapped after brand exposure or did no zap=1). This was done for both audio and visual brand exposure, to create two variables.
Brand exposure condition. The variables audio and visual brand exposure just indicate
13
Brand exposure Zapping condition. A Chi-square test showed that the variables zapping and
brand exposure are significantly related (χ=326.083, p<0.001). Their dependency will lead to multicollinearity issues, which will violate our results. The dependency can be explained by the commercials that are not zapped. If a respondent did not zap both visual and audio brand exposure will always be 1, thus they are dependent. In order to solve this issue we created a new variable using five categories representing the five possible conditions: (0: No audio brand exposure, no visual brand exposure, zapped, 1: No audio brand exposure, visual brand exposure, zapped, 2: Audio brand exposure, no visual brand exposure, zapped, 3: Audio brand exposure, visual brand exposure, zapped, 4: Audio brand exposure, visual brand exposure, not zapped).
Commercial number. To control for the influence of specific commercial characteristics we
also included a dummy variable representing each commercial. This allows us to take into account commercial specific characteristics that might influence brand recall (e.g. humor, visual attraction, product category, length of the commercial). Summary information per commercial can be found in table 2.
Demographics. For each respondent, information was available about gender (0:male,
14
TABLE 2
Means per commercial
Commercial
Number Commercial Name Free Recall Aided Recall Zapping percentage Brand exposure percentage Audio Brand Exposure Visual Brand Exposure Duration
1 ABN (a) Mean .310 c .471 cefg .695 .420 bdefg 31000 31000 35135
SD .464 .501 .462 .495
2 Activia (b) Mean .333 c .603 f .581 .678 aeg 22000 19000 25152
SD .473 .491 .495 .469
3 Citroen (c) Mean .488 abdg .626 a .684 .546 eg 19000 18000 30047
SD .501 .485 .466 .499
4 Dyson (d) Mean .253 c .598 f .603 .649 aeg 13000 15000 20127
SD .436 .492 .491 .479
5 Nikon (e) Mean .368 .707 a .592 1.00 abcdfg 17000 0 20127
SD .484 .457 .493 0
6 Pickwick (f) Mean .391 .764 abd .649 .661 aeg 8000 19000 30175
SD .489 .426 .479 .475
7 SENSEO (g) Mean .259 c .718 a .655 .839 abcdef 5000 20000 30047
SD .439 .451 .476 .369
Total Mean .343 .641 .637 .685 16428 17428 27258
SD .475 .480 .481 .465 8142 8503 5242
15
MODEL
In order to analyze our models we will use logistic regression. This is the preferred form of analysis since our dependent variables, free and aided recall, are binary. Therefore, the relationship is bounded by 0 and 1 and a logistic curve is needed to represent the relationship between the independent and dependent variables. Also, logistic regression can deal with the violations of assumptions of multiple regressions, such as the binomial distribution of the error term and the non-constant variance of the dependent variable (Hair et al. 2010).
Maximum likelihood is used as the estimation technique for logistic regression and this requires larger sample sizes than multiple regression. There are different numbers mentioned in the literature about the appropriate sample size. Hosmer and Lemeshow (2000) recommend to use sample sizes larger than 400. Our sample size holds 1218 observations; therefore, we conclude that we meet the overall sample size requirement. Another important criterion is the sample size per category of the dependent variable. The recommended sample size for each group is at least ten observations per estimated parameter. Table 3 shows the sample size per group. As can be seen, all the groups are larger than the required minimum of 160 (16 parameters times 10). A final consideration is the sample sizes per category of the independent variable. These should not be too small (n>20) since the model then may have trouble converging and reaching a solution. Therefore, we checked the sizes of the subgroups, which can be seen in table 4 below. The subgroup sizes were all sufficient. Thus, we can conclude that the requirements regarding the sample size are all met.
TABLE 3
Sample size per category dependent variable
No free recall 800 No aided recall 437
Free recall 418 Aided recall 781
TABLE 4
Sample sizes per category explanatory variable
No audio, no visual, zapping No audio, visual, zapping Audio, no visual, zapping Audio, visual,
16
In order to test our hypotheses we estimated several different models. Below you can find an overview of the models used.TABLE 5
Summary of models
Model Dependent variable Explanatory variable Hypothesis
1 (a) Free recall
(b) Aided recall Zapping 1a/1b
2 (a) Free recall
(b) Aided recall Visual brand exposure Audio brand exposure 2a/2b/3a/3b 3 (a) Free recall
(b) Aided recall Brand exposure condition 2c/3c
4 (a) Free recall
(b) Aided recall Brand exposure Zapping Condition 4a/4b
The following control variables are included in all the models: commercial number, age, education, and gender.
Model estimation fit and predictive accuracy
Since model 4 is our most elaborate model we will only describe the fit and accuracy of that model in more detail below. The results with regard to the fit and accuracy of model 1, 2, and 3 can be found in the result section.
18
TABLE 6 TABLE 7
Model 1: Estimation results logistic regression zapping and recall Model 2: Estimation results logistic regression brand exposure and recall
Free Recall Aided Recall Free Recall Aided Recall
Explanatory variables Parameter (SE) Odds Wald Statistic Parameter (SE) Odds Wald Statistic Explanatory variables Parameter (SE) Odds Wald Statistic Parameter (SE) Odds Wald Statistic
Constant -.847** .429 5.266 -.314 .731 .706 Constant -2.041*** .130 29.957 -1.677*** .187 19.749
Zapping -.963*** .382 55.205 -1.257*** .285 73.912 Visual brand exposure .260 1.297 1.766 .814*** 2.257 14.934
Commercial Number 34.807*** 43.917*** Audio brand exposure 1.547*** 4.698 60.565 1.327*** 3.769 46.795
Commercial 2 -.007 .993 .001 .445* 1.1561 3.772 Commercial Number 47.849*** 26.290***
Commercial 3 .784** 2.189 11.644 .688** 1.989 8.944 Commercial 2 -.112 .894 .200 .277 1.319 1.212 Commercial 4 -.399 .671 2.268 .458** 1.580 3.975 Commercial 3 .678** 1.970 7.694 .544** 1.723 4.689 Commercial 5 .165 1.179 .496 .969*** 2.635 16.725 Commercial 4 -.704** .495 7.398 .191 1.210 .581 Commercial 6 .327 1.387 1.981 1.364*** 3.911 30.867 Commercial 5 .071 1.073 .075 .548* 1.730 3.773 Commercial 7 -.314 .730 1.645 1.112*** 3.042 21.783 Commercial 6 .030 1.030 .015 1.230*** 3.420 21.236 Gender .058 1.060 .184 .139 1.149 .988 Commercial 7 -.868** .420 10.863 .590** 1.804 4.867 Age .008 1.008 2.982 .023 1.023 21.595 Gender .009 1.009 .004 .071 1.073 .228 Education .083* 1.086 3.778 .065*** 1.067 2.268 Age .000 1.000 .001 .013** 1.013 6.435 Education .071 1.074 2.582 .046 1.047 .997
Cox and Snell R² Nagelkerke R² -2 Log Likelihood % correct classification χ² .077 .106 1469.521 68.2 18.031 .124 .169 1429.385 69.5 8.545
19
TABLE 9
Free Recall: Results logistic regression brand exposure
on recall per baseline condition Aided Recall: Results logistic regression brand exposure on recall per baseline condition Baseline condition 1 No exposure 2 Visual-only 3 Audio-only 4 Both 1 No exposure 2 Visual-only 3 Audio-only 4 Both 1 No exposure β 1.312*** 2.358*** 2.085*** 1.450*** 1.879*** 2.213*** Odds 3.712 10.567 8.042 4.261 6.544 9.142 Wald 17.781 71.134 118.439 23.194 45.472 175.682 2 Visual-only β -1.312*** 1.046** .773** -1.450*** .429 .763** Odds .269 2.847 2.166 .236 1.536 2.146 Wald 17.781 9.247 8.970 22880 1.284 7.989 3 Audio-only β -2.358*** -1.046** -.273 -1.879*** -.429 .334 Odds .095 .351 .761 .153 .651 1.397 Wald 71.134 9.247 1.409 45.472 1.284 1.490 4 Both β -2.085*** -.773** .273 -2.213*** -.763** -.334 Odds 1.314 .462 .124 .109 .466 .716 Wald 118.439 8.970 1.409 175.682 7.989 1.490 TABLE 8
Model 3: Estimation results logistic regression brand exposure condition and recall
Free Recall Aided Recall
Explanatory variables Parameter (SE) Odds Wald Statistic Parameter (SE) Odds Wald Statistic
Constant -.123 .844 .098 -.581 1.678 1.644
Brand exposure condition 130.549*** 183.935***
No exposure -2.085*** .124 118.439 -2.213*** .109 175.682 Visual-only -.773** .462 8.970 -.763** .466 7.989 Audio-only .273 1.314 1.409 -.334 .716 1.490 Commercial Number 57.154*** 22.561** Commercial 2 -.287 .751 1.224 .147 1.159 .329 Commercial 3 .665** 1.945 7.019 .530** 1.699 4.380 Commercial 4 -.864** .421 10.442 .106 1.112 .174 Commercial 5 -.384 .681 1.882 .147 1.159 .233 Commercial 6 -.185 .831 .502 1.127*** 3.088 17.221 Commercial 7 -1.206*** .299 18.201 .351 1.420 1.563 Gender .019 1.019 .017 .076 1.079 .259 Age -.001 .999 .037 .013** 1.013 5.709 Education .076* 1.079 2.924 .047 1.048 1.062
20
TABLE 10Model 4: Estimation results logistic regression
Free Recall Aided Recall
Explanatory variables Parameter (SE) Odds Wald Statistic Parameter (SE) Odds Wald Statistic
Constant -.056 .946 .023 .567 1.763 2.220
Brand exposure Zapping Condition 131.106*** 183.957***
No audio, no visual, zapping -2.138*** .118 114.507 -2.187*** .112 149.957
No audio, visual, zapping -.808** .446 9.606 -.745** .475 7.409
Audio, no visual, zapping .231 1.259 .970 -.310 .734 1.222
Audio, visual, zapping -.196 .822 .908 .104 1.110 .161
Commercial Number 57.918*** 22.557*** Commercial 2 -.300 .741 1.334 .152 1.165 .351 Commercial 3 .680** 1.974 7.309 .525** 1.691 4.283 Commercial 4 -.870** .419 10.561 .107 1.113 .176 Commercial 5 -.450 .667 2.078 .157 1.170 .263 Commercial 6 -.200 .819 .582 1.131*** 3.098 17.316 Commercial 7 -1.216*** .296 18.448 .353 1.423 1.582 Gender -.020 .980 .021 -.074 .929 .248 Age -.001 .999 .024 .013** 1.013 5.664 Education .076* 1.079 2.932 .047 1.048 1.065
Cox and Snell R² .155 .211
21
TABLE 11Free Recall: Results logistic regression per baseline condition Aided Recall: Results logistic regression per baseline condition
22
The parameter estimates are measures of the change in the ratio of the probabilities. Those are appropriate when interpreting the direction of the relationship, however when interpreting the magnitude of the relationship these parameters are less useful. They reflect the change in the logged odds, which is a unit of measure not understandable in depicting how much the probabilities actually change. Therefore, when examining the magnitudes of the effects we will look at the exponentiated logistic coefficients, which reflect the changes in the odds.RESULTS
The results of the logistic regression can be found in Table 6 till table 11. Below we will present our results per hypothesis.
H1a: Zapping has a negative effect on free brand recall & H1b: Zapping has a negative effect on aided brand recall. For our first two hypotheses we found support. The results show that
zapping has a significant negative influence on the probability of both free (β=-0.963, odds=.382, p<0.001) and aided brand recall (β=-1.257, odds=.285, p<0.001).
H2a: Audio brand exposure has a positive effect on free brand recall & H2b: Visual brand exposure has positive effect on free brand recall. Our results show that we only found support for
hypothesis H2a; audio brand exposure does have a significant positive effect on free brand recall (β=1.547, odds=4.698, p<0.001). We did not find support for hypothesis H2b; visual brand exposure did not have a significant effect on brand recall probability (β=.260, p=n.s).
H2c: Audio and visual brand exposure has a stronger effect on free brand recall than audio-only or only brand exposure. We found partial support for this hypothesis. Exposure to
23
H3a: Audio brand exposure has a positive effect on aided brand recall & H3b: Visual brand exposure has a positive effect on aided brand recall. Our results show that we found support for
both hypothesis H3a and H3b; both audio and visual brand exposure did have a significant positive effect on the probability of aided recall (respectively β=1.327, odds=3.769, p<0.001 and β=.814, odds=2.257, p<0.001).
H3c: Audio and visual brand exposure has a stronger effect on aided brand recall than audio-only or visual-only brand exposure. Hypothesis 3c is only partially supported. We did indeed find a
significantly lower probability for the visual-only condition compared to the both exposure condition (β=-.763, odds=.466, p<0.05). This indicates that the aided brand recall probability in the visual-only condition is 53.4% lower than in the both exposure condition. No significant difference was found between the audio-only and the both exposure condition (β=-.334, p=n.s.). This implies that the both exposure condition is only superior in terms of aided brand recall when compared to the visual-only condition but not to the audio-only condition. Our results show that audio-only brand exposure is not superior to visual-only brand exposure (β=.429, p=n.s.). Therefore, we can conclude that the both exposure condition is superior to the visual-only condition. This superiority did not hold to the audio-visual-only condition. So even though the audio-only condition was not significantly superior to the visual-only condition, we can still define audio-only exposure as a somewhat more powerful exposure than visual-only exposure with regard to its effect on aided brand recall.
H4a: Zapping is a stronger determinant of free brand recall than brand exposure. This hypothesis
24
condition of brand exposure is superior to the no brand exposure condition. Therefore, we can conclude that we even found contradicting evidence for our hypothesis, namely that brand exposure is a stronger determinant of free brand recall than zapping.
H4b: Brand exposure is a stronger determinant of aided brand recall than zapping. Our
hypothesis is supported when the brand exposure conditions are all significantly superior to the no brand exposure conditions, while the non-zapping condition is not significantly superior to all the non-zapping conditions. Our results show that the no brand exposure condition leads to a significant lower aided recall probability compared to the visual-only (β=-1.422, odds=.236, p<0.001), audio-only (β=-1.877, odds=.153, p<0.001), both exposure zapping (β=-2.291, odds=.101, p<0.001), and both exposure non-zapping condition (β=-2.187, odds=.112, p<0.001). The non-zapping condition only leads to a significant higher aided brand recall probability compared to the no brand exposure (β=2.187, odds=8.907, p<0.001) and the visual-only brand exposure condition (β=.745, odds=2.106, p<0.05). No significant difference was found between the non-zapping condition and the audio-only (β=.310, p=n.s.) and both exposure zapping condition (β=-.104, p=n.s.). Therefore, we can conclude that brand exposure is a stronger determinant of aided brand recall than zapping and thus hypothesis 4b is supported.
The control variables will be discussed in the light of the results of model 4a and 4b, since those are the most elaborate models.
Commercial number. The specific commercial characteristics as measured by the commercial
dummy have a significant effect on both free (p<0.001) and aided recall (p<0.001). In comparison to commercial 1 (our baseline), commercial 3 (β=.680, odds=1.974, p<0.05 and β=.525, odds=1.691, p<0.05) has significantly higher probability of both free and aided brand recall. Commercial 4 (β=-.870, odds .419, p<0.05) and commercial 7 (β=-1.216, odds=.296, p<0.001) have a significantly lower probability of free brand recall than commercial 1. For aided recall commercial 6 scores significantly higher than commercial 1 (β=1.131, odds=3.098, p<0.001).
Gender, age, and education. Gender did not have a significant effect on either free brand recall
25
a respondent is one level higher, the probability of free brand recall is 7.9% higher. For aided recall this relation was not significant (β=.046, p=n.s.).We also found supporting evidence that we did not encounter multicollinearity issues in model 4a and 4b, since the highest VIF 1.135 < 10.0 (Hair et al. 2010).
To conclude we found that for free recall, zapping in comparison to non-zapping is only significantly negative when it occurs before any brand exposure or leads to visual-only brand exposure. The absence of any brand exposure was found to be significantly more harmful to free brand recall than visual-only brand exposure, indicating that brand exposure always leads to a significant increase in free brand recall probability. Audio-only brand exposure was found to be a significantly stronger determinant of free brand recall than visual-only brand exposure, indicating that at least audio brand exposure is needed to optimize free brand recall. For aided recall we found the same results, except that for this form of recall audio-only brand exposure was not found to be a significantly stronger determinant than visual-only brand exposure.
DISCUSSION AND IMPLEMENTATION
Our study reveals some important implications for both research and practice. First of all, it shows that including partial exposure as a global concept might be too short-sighted and can lead to distorted conclusions. When we included only zapping as a determinant of recall its effect was significantly negative. This might lead to the conclusion that zapping is a negative event per se. However, when including brand exposure as an additional determinant, zapping is only significantly negative under certain conditions.
It is important to note that a respondent’s brand exposure does depend on its zapping behavior. If a respondent zaps before the brand has been exposed, the probability of brand recall is significantly lower. If a respondent zaps after audio or both forms of brand exposure, the probability of brand recall is not significantly different than when no zapping occurs. This indicates that zapping is only harmful when it takes place before at least audio brand exposure. Therefore, zapping and its determinants remain an important topic for both practitioners and academics. However, this research shows that avoidance strategies should not solely been seen as a global concept and should be studied in the light of other determinants of brand recall that differentiate partial exposure. This also emphasizes the added value of collecting moment-to-moment data per commercial, since it enables the possibility to collect the information needed for such analyses.
26
effect of zapping on recall can be contributed to differences in brand exposure and not to differences in commercial overload per commercial block.The results of our study provide a possible explanation for the inconsistency in findings between previous research about the effect of zapping on recall. It might be that the studies finding minor negative or non-significant effects had a higher percentage of brand exposure than the studies finding a significant negative effect.
There are some differences in results for free and aided recall which aligns with previous research that they are two different concepts (Shapiro and Krishnan 2001; Moorman, Neijens, and Smit 2007). This has implications for both academics and practitioners. First of all, it confirms the value of including both measures of recall when examining advertising effectiveness’ concept of learning. Secondly, for practitioners it shows that depending on the preferred outcome a different focus might be considered. We will discuss the implications for each form of recall in more depth.
For free recall it was found that zapping only has a significantly negative effect when it occurs after the brand is only exposed visually of before any brand exposure. There was no significant difference in free brand recall between zapping occurring after at least audio brand exposure and no zapping behavior. This conditional significance of zapping on free recall is not in line with our hypothesis. We expected that full exposure would lead to significantly higher free recall than partial exposure in which brand exposure occurs. However, our results show that when at least audio brand exposure occurs, the retrieval stage needed for free recall already is as likely to be completed as after full exposure.
Zapping after only visual brand exposure was found to be significantly negatively different from no zapping. However, this condition was found to already be a significant improvement from no brand exposure at all. This indicates that although not optimal, visual brand exposure already significantly increases brand recall.
27
The chance of zapping without any brand exposure thus increases and this has a significant negative effect on free recall, making mystery ads less recommendable.Furthermore, our results also show that free recall is a complex phenomenon. Simply being exposed to the whole commercial instead of partial exposure does not significantly increase its likelihood.
As already been mentioned, our results indicate that when free recall is the aimed form of effectiveness it is significantly beneficial to focus on brand exposure, while also paying attention to determinants of zapping behavior. However, brand exposure might lead to commercial avoidance, providing an important dilemma for advertisers. The presence of a brand conveys information and if they are present too prominently this might lead to information overload which results in commercial avoidance (Woltman-Elpers, Wedel, and Pieters 2003). Also if the brand presence is received as being an aggressive persuasion expression, people will tend to resist from it resulting in avoidance (Aaker and Bruzzone 1985). Texeira, Wedel, and Pieters (2010) found that avoidance rates could be decreased by changing the pattern of brand exposure while keeping the brand activity level per commercial the same. This implies that it is important to pay attention to the pattern, while it is not necessarily the amount of brand exposure that is leading to commercial avoidance. They found that the most optimal strategy is the so-called pulse strategy in which brands are shown frequently for short time periods instead of infrequent longer brand exposure.
Based on our results we suggest using audio brand exposure in the beginning of the commercial. This is not likely to be perceived as being an aggressive persuasion expression since the visual part of the commercial can be non-branding. This strategy might lead to audio brand exposure while minimizing avoidance, an optimal combination for a higher probability of free recall.
28
The dominance of audio exposure might be explained by several factors. First of all, visual brand exposure has a higher probability of being implicitly exposed. For example, the product showed in the commercial might serve as an implicit form of visual brand exposure. For audio exposure there are less implicit substitutes of brand exposure. Therefore, even if the brand is not yet explicitly visually exposed, it can already be implicitly exposed. So in the audio-only condition, visual exposure can still have occurred implicitly whereas the opposite is less likely to occur. Another explanation might be that in general, commercials are more visually focused than aurally. In other words, there are more visual than audio stimulants present in a commercial (Lang 1995). When an audio stimulant is present such as audio brand exposure, this might be more notable than visual brand exposure, leading to a stronger memory effect.Our result also holds an important implication for the negotiation of networks about broadcasting prices for commercials. Our study shows that partial exposures can indeed be as beneficial in terms of brand recall as full exposures under certain conditions of brand exposure. So for networks our research implies that partial exposure rates should be taken into account when setting the broadcasting prices. For advertising agencies our research also holds important information. Brand exposure is partly under control of the advertisers, meaning they are not completely powerless in the competition for recall.
LIMITATIONS AND FUTURE RESEARCH
This study has several limitations. First of all, even though the respondents were able to participate in this study at home, the setting could be even more naturalistic. The respondents were asked to view commercials, which makes it possible they paid more attention to the commercials than they would normally do. It is therefore unlikely that they participated in other forms of avoidance than zapping, such as cognitive or physical avoidance. Even though this has the advantage of being able to measure the effect of zapping in more isolation, it does not reflect the natural setting of commercial viewing in which more avoidance strategies are likely to be exhibited.
Road-29
blocking is the time-synchronization of commercials on different channels. This holds that when a consumer zaps channels, it zaps from one commercial into another (Texeira, Wedel, and Pieters 2010). Therefore, we do not view the sequential presentation of the commercials as a major limitation. However, for future research it is advised to use a more natural television setting.Another limitation is that we only used commercials of well-known brands. Our respondents most likely already had prior experience with the brands and partial exposure might therefore serve as a reminder that influences recall (Bellman, Schweda, and Varan 2010). This might underestimate the effect of zapping on recall. For future research it is therefore advisable to also include less well-known brands to examine the generalizability of our findings.
For future research it would be of interest to examine the influence of zapping and brand exposure on other components of advertising effectiveness as well (e.g. persuasion, comprehension, brand choice, purchase intention). Zufryden et al. (1993) already found a positive relation of zapping on purchase behavior. This indicates that the effect of zapping might differ between the components of effectiveness and is thus relevant to study.
Another suggestion for future research is to include other avoidance strategies as well, such as zipping, muting, and cognitive avoidance. As already found by Bellman et al. (2010) the effect on brand recall is not the same across all avoidance strategies.
In this study we only included two possible determinants of brand recall, namely zapping and brand exposure. As already been mentioned, brand recall is a complex phenomenon that consists of many possible determinants (Stewart and Furse 1986). Therefore, for future research it is suggested to include more determinants. For example, it would be of interest to examine the influence of different forms of brand exposure in terms of prominence. This will shed light on whether the brand exposure needs to be prominent or whether subtle exposure is already sufficient in order for the brand to be recalled. This will be valuable information since several studies already showed that (too) prominent brand exposure leads to avoidance (Woltman-Elpers, Wedel, and Pieters 2003; Aaker and Bruzzone 1985). In our study we only included explicitly forms of brand exposure. So for future research it will be interesting to include implicit forms of brand exposure as well. Especially since, as we already mentioned in the discussion, the implicit forms of visual exposure might explain the superiority of audio brand exposure over visual brand exposure.
30
2005), program involvement (Moorman, Neijens, and Smit 2007) will be of interest to study in the light of zapping. This will enable a deeper understanding of the relative effect of zapping. To analyze our data we used a simple logit model. However, our data contains a hierarchical relationship between the respondents and the commercials, as data is pooled across respondents. Our model does not account for this hierarchy, since commercials are treated as independent observations. As a consequence, the risk of Type 1 error is inflated and standard errors might be too small (Pieters and Bijmolt 1997). However, it is important to note that our main findings are related to the lack of significance differences between the exposure conditions, making the risk of a Type 1 error less influential for our results. Nevertheless, we do acknowledge that this is a limitation of our research and it is suggested that for future research a multilevel logit model is used. These models explicitly account for the variance and effects on both the respondent and commercial level.Another research opportunity is to examine the wear in time of exposure. In previous studies the zapping function was disabled the first five seconds of the commercials to ensure that the zapping decision was based on partial exposure. However, in our study we decided not to use disabling at any point. We believe that partial exposure happens the first second one sees the commercial. For example, at one of our commercials the brand was already exposed at the first second of the commercial. To be sure we also run our analyses by leaving out exposures of less than five seconds, however the results showed that it did not significantly change our results. Therefore, we believe that exposure can already be defined as partial exposure in the first few seconds of the commercial. For future research there is an opportunity to investigate the wear in time of exposure in order to further examine after what time we can assume that the partial exposure of a consumer is of influence for brand recall.
31
References ,
Aaker, David A. and Donald E. Bruzzone (1985), “Causes of Irritation in Advertising,” Journal of
Marketing, 49 (2), 47-57.
Abernethy, Avery M. (1990), “Television Exposure: Programs versus advertising,” Journal of
Current Issues and Research in Advertising, 13 (1), 61-77.
Babbit, Brian C. (1982), “Effect of Task Demands on Dual Coding of Pictorial Stimuli,” Journal of
Experimental Psychology, 8 (1), 73-80.
Baker, William E., Heather Honea, and Christel A. Russell (2004), “Do not Wait to Reveal the Brand Name: The effect of brand-name placement on television advertising effectiveness,”
Journal of Advertising, 33 (3), 77-85.
Baker, William E., Wesley J. Hutchinson, D Moore, and Prakash Nedungadi (1986), “Brand Familiarity and Advertising: Effect on the evoked set and brand preferences.” Advances in
Consumer Research, 13, 637-42.
Bellman, Steven, Anika Schweda, and Duane Varan, D (2010), “The Residual Impact of Avoided Television Advertising,” Journal of Advertising, 39 (1), 67-81.
Bettman, James R. (1979), “An Information Processing Theory of Consumer Choice,” Reading, MA: Addison Wesley Publishing Company.
Brennan, Ian and Laurie A. Babin (2004), “Brand Placement Recognition: The influence of presentation mode and brand familiarity,” Journal of Promotion Management, 10, 185-202. Brown, Gene (1985), "Tracking Studies and Sales Effects: A U.K. perspective," Journal of
Advertising Research, 25, 52-64.
Bryce, Wendy J. and Richard F. Yalch (1993), “Hearing versus Seeing: A comparison of consumer learning of spoken and pictorial information in television advertising,” Journal of Current Issues
and Research in Advertising, 15 (1), 1-20.
Childers, Terry L. and Micheal J. Houston, (1984), “Conditions for a Picture Superiority Effect on Consumer Memory,” Journal of Consumer Research, 11, 643-54.
Ehrenberg, Andrew S. C., and W.A. Twyman (1967), “On Measuring Television Audiences,”
Journal of the Royal Statistical Society, 130 (1), 1-60.
Fazio, Russel H., Paul M. Herr, and Martha C. Powell (1992), “On the Development and Strength of Category-Brand Associations in Memory: The case of Mystery Ads,” Journal of Consumer
Psychology, 1 (1), 1-13.
Greene, William F. (1988), “Maybe the Valley of the Shadow isn't so Dark after all,” Journal of
Advertising Research, 28, 11-15.
32
Gustafson, Paul and Sivaramakrishnan Siddarth (2007), “Describing the Dynamics of Attention to TV Commercials: A hierarchical Bayes analysis of the time to zap an ad,” Journal of AppliedStatistics, 34 (5), 585-609.
Hair, Joseph F., William C. Black, Barry J. Babin, and Rolph E. Anderson (2010), Multivariate Data
Analysis. 7th edition, New Jersey: Pearson Education.
Hosmer, David W. and Stanley Lemeshow (2000), Applied Logistic Regression. New York: Wiley. Hoyer, Wayne D. and Steven P. Brown (1990), “Effect of Brand Awareness on Choice for a Common, Repeat-purchase Product,” Journal of Consumer Research, 17, 141-48.
Kardes, Frank R. (1988), “Spontaneous Inference Processes in Advertising: The effects of conclusion omission and involvement on persuasion,” Journal of Consumer Research, 15, 225-33. Karrh, James A., Kathy B. McKee, and Carol J. Pardun (2003), “Practitioners’ Evolving Views on Product Placement Effectiveness,” Journal of Advertising Research, 43 (2), 138-48.
Krugman, Dean M., Glen T. Cameron, and Candace M. White (1995), “Visual Attention to Programming and Commercials: The use of in-home observations,” Journal of Advertising, 24, 1-12.
Lang, Annie (1995), “Defining Audio/Video Redundancy from a Limited-Capacity Information Processing Perspective,” Communication Research, 22, 85-115.
Lavidge, Robert J. and Gary A. Steiner (1961), "A Model of Predictive Measurements of Advertising Effectiveness," Journal of Marketing, 25 (6), 59-62.
Law, Sharmistha and Kathryn A. Braun (2000), “I’ll Have what She’s Having: Gauging the impact of product placement on viewers,” Psychology and Marketing, 17, 1059-75.
Moorman, Marjolein, Peter C. Neijens, and Edith G. Smit (2007), “The Effects of Program Involvement on Commercial Exposure and Recall in a Naturalistic Setting,” Journal of
Advertising, 36 (1), 121-37.
Nedungadi, Prakash (1990), “Recall and Consumer Consideration Sets: influencing choice without altering brand evaluations,” Journal of Consumer Research, 17, 263-76.
Paivio, Allan (1986), Mental Representations: A Dual Coding Approach. New York: Oxford University Press.
Paivio, Allan and Kal Csapo (1969), “Concrete Image and Verbal Memory Codes,” Journal of
Experimental Psychology, 80 (5), 279-85.
Pieters, Rik G.M. and Tammo H.A. Bijmolt (1997), “Consumer Memory for Television Advertising: A field study of duration, serial position, and competition effects,” Journal of
Consumer Research, 23 (4), 362-72.
Rossiter, John R. Larry Percy, and Robert J. Donovan (1991) “A Better Advertising Grid” Journal
of Advertising Research, 31 (5), 11-21.
33
Shapiro, Stewart and Shanker Krishnan (2001), “Memory-Based Measures for AssessingAdvertising Effects: A comparison of explicit and implicit memory effects,” Journal of
Advertising, 30 (3), 1-13.
Singh, Surendra N., and Michael L. Rothschild (1983), “Recognition as a Measure of Learning from Television Commercials,” Journal of Marketing Research, 20, 235-48.
Singh, Surendra N., Michael L. Rothschild, and Gilbert A. Churchill (1988), “Recognition versus Recall as Measures of Television Commercial Forgetting,” Journal of Marketing Research, 25 (1), 72-80.
Speck, Paul S. and Michael T. Elliot (1997), “Predictors of Advertising Avoidance in Print and Broadcast Media,” Journal of Advertising, 26, 61-76.
Srull, Thomas K. (1981), “Person Memory: Some tests of associative storage and retrieval models,” Journal of Experimental Psychology, 7, 440-63.
Stewart, David W. and David H. Furse (1986), Effective Television Advertising: A study of 1000
commercials. Lexington, MA: Lexington Books.
Stewart, David W. and Scott Koslow (1989), “Executional Factors and Advertising Effectiveness: A replication,” Journal of Advertising, 18 (3), 21-32.
Stewart, David W., Connie Peckmann, Srinivasan Ratneshwar, Jon Stroud, and Beverly Bryant (1985), “Methodological and Theoretical Foundations of Advertising Copytesting: a review,”
Current Issues and Research in Advertising, 2, 1-74.
Stout, Patricia A. and Benedicta L. Burda (1989), “Zipped Commercials: Are they effective?,”
Journal of Advertising, 18 (4), 23-32.
Texeira, Thales, Michel Wedel, and Rik M. Pieters (2010), “Moment-to-Moment Optimal
Branding in TV Commercials: Preventing Avoidance by Pulsing,” Marketing Science, 29 (5), 783-804.
Texeira, Thales, Michel Wedel, and Rik M. Pieters (2012), “Emotion-Induced Engagement in Internet Video Advertisements,” Journal of Marketing Research, 49 (2), 144-59.
Thorson, Esther and Xinshu Zhao (1997), “Television Viewing Behavior as an Indicator of Commercial Effectiveness,” in Measuring Advertising Effectiveness, Wells, W.D. New Jersey: Lawrence Erlbaum.
Till, Brian D. and Daniel W. Baack (2005), “Recall and Persuasion: Does creative advertising matter?,” Journal of Advertising, 34 (3), 47-57.
Tulving, Endel and Donald M. Thomson (1973), “Encoding Specificity and Retrieval Processes in Episodic Memory, Psychological Review, 80 (5), 352-73.
Wegert, Tessa (2004), “When Consumers Love Advertising,” ClickZ Networks, (accessed on March 18, 2012), [available at http://www.clickz.com/clickz/column/1696299/when-consumers-love-advertising].
Woltmans-Elpers, Josephine L.C.M., Michel Wedel, and Rik M. Pieters (2003), “Why do
34
Zielske, Hubert A. (1982), “Does Day-After Recall Penalize 'Feeling' Ads?,” Journal of AdvertisingResearch, 22, 19-22.
