European Advances in Consumer Research Volume 1, 1993 Pages 160-167
THE OPERATIONALIZATION OF MOTIVATION, CAPACITY AND OPPORTUNITY TO PROCESS AN ADVERTISING MESSAGE
Henry S.J. Robben, Tilburg University, The Netherlands
Theo B.C. Poiesz, Tilburg University, The Netherlands
[We like to thank Sjaak Bloem and Jolande van der Valk for their substantial help in the preparation and execution of the reported research, and Irwin Levin for his comments on an earlier version of this article.]
General level operationalizations of three factors thought to affect information processing of commercial stimuli were proposed. This approach contrasts with more specific operationalizations that are encountered frequently in the advertising effectiveness literature. The present investigation tested two hypotheses regarding the effects of motivation, capacity and opportunity on processing information on photographed billboards. Two main findings emerged: (a) motivation, capacity and opportunity were viewed as relatively independent factors; and (b) motivation and capacity exerted significant main effects on recall and recognition scores, whereas the main effect of opportunity on the dependent variables was marginal. Opportunity affected brand and message awareness through two-way and three-way interactions.
In the past decade, several authors have attempted to integrate the existing research on specific determinants of advertising effects. The most well-known result of such an effort is the Elaboration Likelihood Model by Petty and Cacioppo (1986). This model identifies two generic determinants of message elaboration: motivation and ability. Both variables can be viewed as capable of summarizing the effects of many more specific determinants of motivation and ability. Among many other examples of determinants, need for cognition, source attractiveness, the use of color, and the use of humor may affect motivation, and, for example, mental fatigue and product related knowledge can both be subsumed under the more generic variable ability.
If models on advertising processing would attempt to include all these highly specific determinants, they would be too complex to handle, even when ignoring possible interactions. Therefore, it is not possible to design a generally applicable model on advertising effects in which all possibly relevant variables are specified at a detailed level. One way of dealing with this limitation is to depart from models or sets of hypotheses with regard to partial phenomena. Another way is to identify variables at a more general level, at the cost of losing specificity.
Stewart and Koslow (1989) attempted to come to an understanding of advertising effects at both a specific and at a general level by using an analytical rather than a theoretical approach. In their study, they coded 1,017 television commercials for content and had these subsequently rated by their respondents. Message characteristics were related to recall, comprehension and persuasion. Results were presented with regard to two different levels of generality or specificity. At the more specific level, individual message characteristics (such as, for example, the use of an outdoor versus indoor setting in the ad; a male versus a female character) were correlated with the dependent variables. At the more general level, two variables were proposed that seemed capable of summarizing the effects at the more specific level, namely (1) a strong brand/product focus, and (2) a brand differentiating message (clear statement of benefits).
While Petty and Cacioppo (e.g., 1986) use two generic variables, motivation and ability to process an ad, other authors have suggested the use of three general variables. In his TRIAD model, Poiesz (1989) distinguishes motivation, opportunity and capacity. Andrews (1988) and MacInnis and Jaworski (1989) refer to motivation, opportunity and ability (MOA). Poiesz' (1989) model can be theoretically positioned as an elaboration of the Elaboration Likelihood model and the MOA-models.
The differences between these various approaches may relate to the concepts employed, and/or to the level of generality or specificity of the variables.
Table 1 provides an overview of different definitions suggested by various authors. The authors of the studies identified in Table 1 do not justify the use of either two or three generic variables. Even though the choice between two and three generic alternatives may be trivial from one point of view, the choice for either option may possibly affect conceptual clarity, operationalization possibilities and problems, and relevance for methodological and application questions.
Ability versus Capacity and Opportunity
In essence, given the fact that motivation is included indiscriminately in the various approaches, the question centers around the ability concept: to what extent would it be useful to distinguish between capacity and opportunity? For a discussion of this question, it is necessary to first consider the general advantages and disadvantages associated with different levels of generality or specificity of variables.
Variables that are the object of research on the effects and the effectiveness of advertising vary, apart from their conceptual content, with regard to the level of generality or specificity. More general or generic variables subsume variables at a more specified level. Highly specific variables relate to particular consumer, message, medium, and/or situational aspects. Examples of the latter type of variables are the consumer's product interest, the message source credibility, the size and the format of the ad, and the presence or absence of distracting stimuli other than the ad. Each of these variables deals with a particular, detailed aspect of the total exposure situation.
Other studies depart from variables at a more general, and thus less specific level. An example of such a variable is the attitude toward the ad (e.g., Lutz, MacKenzie and Belch 1983). On the one hand, this concept is specific in that it relates to ad-based motivation. On the other hand, it is a generic variable to the extent that it is capable of subsuming and summarizing many specific consumer and message variables, including consumer product interest and a variety of ad execution variables. To extend the latter example, if the basis of motivation would not be specified, the level of generality would be even higher.
Whether a variable is located at a more generic or more specific level of analysis can not be determined by objective criteria. Variable specificity is a relative characteristic at best as, in principle, each single individual variable can be dissected into other, more specific variables. With regard to variable generality, there are fewer degrees of freedom. The most generic variable is a single variable, comprising all the variables at all more specific levels. We will return to this issue later.
CONCEPTS SUGGESTED BY DIFFERENT AUTHORS
The fact that each specific variable can be assumed to comprise other, more specific variables renders the label 'specific' somewhat ambiguous. However, the distinction between levels of variables may be useful for theoretical, methodological and possibly also practical reasons. We will start by discussing the differences between relatively generic and relatively specific variables. For simplicity, we will only refer to a comparison of more generic and more specific variables, and will ignore possible finer differentiations between these two crude levels:
- Generic variables have the advantage of being generalizable to other advertising situations. Specific variables only apply to the very situation in which they have been assessed. The use of specific variables is often accompanied by the implicit or explicit use of the ceteris paribus clause. Generic variables may be more relevant for inclusion in models that are intended to be applicable for different messages, different consumers, and different exposure situations.
- The advantage of generic variables in the previous point is counterbalanced by the obvious disadvantage that they are not able to provide a precise understanding of what exactly produces an advertising effect. By definition, generality implies lack of specificity, and only specificity reveals the detailed reasons why an advertising effect occurs or fails to occur. At each level of generality or specificity comprehensiveness and exactness of a variable are complementary characteristics.
- The set of specific variables has as an obvious disadvantage that the overview may be easily lost, especially at high levels of specificity. An explanation of an advertising effect on the basis of specific variables may easily be confronted with an alternative and equally plausible explanation using other specific variables.
The choice of a particular level of generality or specificity seems to be a rather trivial one, as it is simply dictated by the goal of the individual researcher. For example, if the goal is to study the relationship between, on the one hand, source credibility in a particular tv-commercial on a particular brand of a particular product and, on the other hand, ad processing by individual consumers belonging to a particular target group, the research goal does not require the specification of generic variables such as motivation and ability to process. However, if the research goal is to study the 'likelihood of ad elaboration' on the basis of generic explanatory variables (such as motivation and ability) it may not be relevant nor efficient to simultaneously study the effects of a multitude of 'miniature' determinants of elaboration. The choice of a particular generality or specificity level is therefore up to the individual researcher, for whom the choice is unlikely to create a dilemma. Yet, of course, this choice still has to be made, in which we may run into questions relating to conceptual parsimony, operationalization possibilities and limitations, and empirical outcomes in terms of explained variance.
Returning to the issue of the dissection of the ability variable into capacity and opportunity, the reasons against making such a differentiation are obvious. The first reason is dictated by the rule of parsimony: two variables are preferred to more variables if two variables can explain the phenomenon under consideration. The second reason is that ability is a conceptually clear concept, well capable of covering the effects due to internal and external processing related factors.
Some objections might be raised against these justifications, however. First, with regard to the rule of parsimony, it must be noted that the mere number of explanatory variables cannot provide the only criterion. If this were so, the use of a single explanatory variable would also be preferable to the use of two explanatory variables. A subjective rating of the single variable 'probability to show behavior X' might be an example of such an ultimately integrating variable. However, by using a single variable we would lose explanatory power, maybe not in the statistical sense, but in the sense of understanding the psychological determinants of behavior X.
Therefore, the rule of parsimony should be in balance with the researcher's need for explanatory power. This might justify the use of more variables than a strict application of the rule of parsimony would require.
Several arguments may be presented, in random order, for dividing the ability concept into capacity and opportunity if we define capacity as person-related ability and opportunity as environment-related ability. The first argument is that, from a psychological point of view, the distinction between internal and external causes of behavior is a fundamental one.
A second argument is that a separate assessment of capacity and opportunity may be helpful in determining to what extent message processing is hampered by person-related aspects or aspects in the communication environment. In the former case, increasing message simplicity might improve processing, while in the second case it might not. Conversely, choosing a medium that allows more processing time may stimulate processing, but only if the message itself can be understood. For example, simply increasing exposure time - opportunity - may not be enough if the message recipient does not have the knowledge to understand the technical language (person related capacity) employed in the message.
Consistent with this reasoning, there is also a methodological argument for distinguishing between capacity and opportunity. A person's ability to process ad information should always be considered with reference to the conditions under which processing is to take place. Asking a person to what extent a message is understood may produce different answers depending upon the time allowed for processing and processing difficulty as determined by, for example, extraneous distraction. If the opportunity aspect is not dealt with separately, it may be easily overlooked by researchers in laboratory settings, as has been done frequently in many studies performed in the past.
Operationalizing Generic Variables
It is noteworthy that even though several authors have suggested the use of generic variables motivation, capacity/ability or opportunity, they have not suggested operationalizations for these variables. Generic variables seem to have been proposed, thus far, for theory development only, and not for operational use in advertising research. If generic variables were operationalized at all, they were operationalized with the help of elaborate questionnaires on specific level variables. For example, motivation has been operationalized by way of multi-item involvement scales (see, e.g., Zaichkowsky 1985). However, for several reasons generic level operationalizations may be helpful. The first reason is a theoretical one. Generic level operationalizations would make it possible to research the interdependencies of determinants of behavior at a very general level, and to assess their relevance in the explanation and prediction of advertising processing. Academic advertising research is often associated with the investigation of highly specific variables associated with micro phenomena. Yet, this does not preclude the possibility and potential relevance of research on generic level variables. The second reason is a practical one: the pre-testing of advertising messages is often hampered by the lack of clarity as to what should be measured and by the abundance of possible items. A simple instrument with a limited set of items that each cover one very general advertising processing condition would be very functional.
In the remainder of this paper we will describe an attempt to produce operationalizations of motivation, capacity and opportunity, using the following points of departure:
- Motivation, capacity and opportunity are taken as individual measures. Subjects should assess the degree to which each concept applies to them personally, given a particular advertising message;
- Motivation, capacity and opportunity should be measured in an independent way. Ideally, for conceptual and operational clarity, no overlap between the concepts should exist;
- In line with the need for general level variables, general level operationalizations should be applied. There is no sense in specifying generic variables with elaborate and detailed operationalizations. For this reason, we will choose operationalizations that in their level of generality match the three proposed concepts. In fact, single item operationalizations will be used. Items will be phrased so that subjects will perceive them as referring to distinct and separate aspects of the total exposure situation. Also for practical reasons single item operationalizations might be preferable to multi-item batteries.
Pretests of Operationalizations
Before executing the main experiment, two pretest studies were conducted. The goal of the first pretest was to provide an initial test of operationalizations of the three generic variables based on the points of departure described earlier. In order to create advertising situations in which opportunity would really matter we selected a medium for which time constraints (i.c. opportunity constraints) are particularly relevant: the billboard.
In many outdoor settings, billboards are located in conditions that are favorable for exposure but not necessarily favorable for processing. Time available for processing is often limited, especially when the potential recipient is moving relative to the billboard (e.g., in traffic situations). In billboard exposure situations distractions can easily create opportunity limitations such as the need to watch traffic, and the presence of other commercial signs and ads attracting attention. Consistent with this argument is the choice of the dependent variables in this study. Brand and message awareness measures reflect the types of advertising impact that are most likely achieved by billboards. Such measures need to be contrasted with measures like persuasion or purchase intention, which reflect deeper and more specific processing of brand related information. Rossiter, Percy and Donovan (1991) view brand awareness as a precursor for the development of brand attitude.
This discussion may not be taken to signify that recall and recognition of brand information can be used as if they were interchangeable. Recall and recognition may be related measures. Jones (1987) discusses a dual mechanism of recall structure in which he distinguishes direct recall and indirect recall. Direct recall is independent from recognition in that it represents the result of accessing information in memory by retrieval information with which it has a direct link. Indirect recall is in fact a two-stage process; the first is independent from recognition, whereas the second stage in fact constitutes recognition. This indirect recall route is consistent with the generation-recognition model endorsed by Bahrick (1970, and see Mandler 1980). Given this dissimilarity between both measures, it seems likely that they may be affected differently and by different variables.
The choice of the dependent variables influenced the operationalizations of the predictor variables. For the three generic variables the following operationalizations [The operationalizations presented here are translations from the originally Dutch ones.] were used:
Motivation: 'To what extent do you think the billboard is interesting?';
Capacity: 'To what extent do you think the advertiser's message about his brand is understandable?'; and
Opportunity: 'To what extent do you have enough time to absorb the billboard (that is, if you would want to do so)?'
It was assumed that interest can subsume all motivation related specific variables, including product and message related interests. (An operationalization at the generic level runs the risk of other generic variables entering in, as the following example with regard to the variable motivation may show. An alternative operationalization of this variable is the amount of attention that a consumer is willing to spend on a commercial. However, willingness to pay attention is not only a matter of motivation but also a matter of capacity, and possibly a matter of opportunity as well.)
Capacity was operationalized in terms of understandability. In order to avoid the possibility that a consumer might consider only a fraction of the total message, such as the head line, a reference to the advertiser's goal was included. Understandability may obtain a high score in a situation where ability to process is low. In such a case, opportunity is limited.
Opportunity is a matter of processing determinants related to the actual environment of the individual, including distraction and time available for processing. Opportunity was operationalized in terms of time, assuming that perceived adequacy of processing time is capable of including all environment related determinants. Distraction, for example, can be overcome by increasing the amount of processing time. More processing time is needed if the level of distraction increases.
In the present study, a within-subjects design would be associated with several complexities. In such a design a subject would first have to judge the stimuli (ads) with regard to their position on the three independent variables, and subsequently rate them again to provide the dependent measure(s) per stimulus. Thus, stimuli would have to be presented twice. The second stimulus exposure might affect the critical variables motivation, capacity and opportunity, however. For example, opportunity to process the message may be expected to be rated differently at the first and at the second exposure.
For this reason, a between-subjects design was employed involving different groups of participants in each of the phases of the current investigation. The investigation can be divided into two parts. The first part comprised two pretest studies, the second part contained the main study. In the pretest studies, two groups of subjects rated advertising messages with regard to motivation, capacity and opportunity. The main study required participants to evaluate a number of billboards with regard to three dimensions, and recall and recognition scores were obtained.
Stimulus Selection and Control Group Ratings
The goal of this part of the investigation was to assess the three dimensions for advertising messages, and to make a selection from these messages for the second phase, the assessment of the dependent variables. In the first part an initial selection was made out of 47 slides containing traffic situations with billboards. No particular a priori criteria were used for the selection of these billboards, nor for set size.
From the total set, a subset of 16 stimuli was selected, two stimuli for each of the 8 possible stimulus categories with two levels (relatively high and relatively low) per dimension (motivation, capacity, and opportunity).
Subjects of Pretest 1 and Pretest 2. Nineteen undergraduate students of Tilburg University, 9 men and 10 women took part in pretest 1; 16 undergraduate students at the university, 11 male and 5 female, participated in pretest 2.
Material. For financial reasons it was not possible to have new billboards designed for the present study only. However, because of the nature of the present study, subjects should not be confronted with already familiar billboards. A solution to this problem was to present Belgian (Flemish) billboards to Dutch subjects (who are usually not familiar with such billboards). Apart from some differences in pronunciation, there are no critical language differences between Dutch and Flemish, at least as far as the present studies are concerned.
In this type of research, special attention must be given to the issue of external validity. In the present study, billboards were presented as a part of a slide depicting a 'traffic situation'. Slides of the billboards were taken from the position of a car driver.
Procedure. Each of the pretests was announced as a study on the evaluation of traffic situations. Subjects were seated between 6 and 15 feet from the screen on which the slides were shown. They received written instructions on how to interpret the key words motivation, capacity, and opportunity in the postexperimental questionnaire. Motivation was thus explained as referring to an attractive, captivating billboard. Capacity reflected an understanding of the information contained in the billboard, including verbal and visual information. Finally, opportunity was phrased in terms of the amount of time subjects had to absorb the information on the billboard.
Each billboard had to be rated on the three dimensions using a 5-point scale, with a '1' indicating low motivation, low capacity or low opportunity. A '5' represented high motivation, high capacity or high opportunity. These questions thus captured participants' subjective experiences of the motivation, capacity and opportunity dimensions as represented on the slides.
Prior to showing the slides, five practice slide were presented, each for 10 seconds. Stimulus exposure time in the first pretest was 10 seconds. Subjects subsequently responded to the three questions after each billboard slide; they had the opportunity to ask additional questions. The order of the motivation, capacity and opportunity questions were alternated over consecutive stimuli.
In the second pretest, motivation was manipulated by warning subjects by flashing a red light before the presentation of certain slides. It was explained to the participants that when the red light flashed, they had to answer questions about those particular slides after the presentation. This procedure sought to establish a situation of increased motivation to process the presented information, as compared to the standard situation which did not require subjects to pay special attention to the slides. Slides were either presented during 6 seconds or 14 seconds, thus creating low or high opportunity to process the information.
Results of the Pretests
Subjects' responses to the subjective experience questions were used to categorize the experimental stimuli as representing low or high motivation, low or high capacity and low or high opportunity. The critical scale limits for both levels of each of the dimensions were determined based upon distribution characteristics of the response variables. The empirical limits suggested that the cut-off points centered around the midpoints of the scales employed. We like to emphasize that this procedure represented a categorization into relative levels of the predictor variables rather than absolute levels.
The categorization procedure yielded an indication of how stimuli were distributed over the levels of motivation, capacity and opportunity. The numbers in parentheses represent the distribution of the stimuli as obtained in the second pretest. There were 15 (11) high and 32 (36) low motivation stimuli, 32 (31) high and 15 (16) low capacity stimuli, and 36 (24) high and 8 (23) low opportunity stimuli. Due to partial nonresponse, totals did not add up to 47 stimuli in pretest 1. The number of stimuli assigned to each category is presented in Table 2 for both pretest studies. A difficulty with the present selection is that it does not provide a perfect pair of stimuli for each of the stimulus categories. For example, the M+ C+ O- category does not contain any stimuli at all that meet the set criteria. In spite of the limitations suggested by Table 2, the best possible pairs of stimuli per category were selected to feature in the main experiment, in order to execute the experimental procedure as planned.
Subjects. Seventy-five students in economics or psychology at Tilburg University volunteered in the experiment; there were 35 male and 40 female subjects. The students were told that they participated in a traffic investigation. Subjects' ages ranged from 19 to 32 years.
CLASSIFICATION OF THE STIMULI OVER THEORETICALLY DERIVED CATEGORIES (M=MOTIVATION, C=CAPACITY, O=OPPOTUNITY, - = LOW LEVEL, + = HIGH LEVEL)
Design. The 16 stimuli selected in the second pretest were mixed with an identical number of slides that pictured typical traffic scenes. The latter slides did not contain any billboards or other forms of commercial communication, and did not differ from the experimental stimuli in terms of photographic angle or perspective. This inclusion was used to enhance subjects' perception of participating in a traffic study. The slides were projected in random order.
Procedure. The procedure for the main experiment did not differ from the one used previously. Slides were projected one at a time, and participants answered three questions after each slide. These questions were designed to assess subjects' perceptions of the presence in the slide of the factors motivation, capacity and opportunity. A '1' per dimension indicated that the slide was respectively not interesting, incomprehensible or provided insufficient time to process the brand information; likewise, a '5' on a dimension indicated that the stimulus situation was respectively interesting, comprehensible or provided sufficient time to process the information on the slide. The sequence in which these dependent variables were measured was varied in order to avoid any systematic presentation effects. After seeing all slides, a filler task consisting of simple arithmetics was performed. The research instrument then required that subjects noted down the brand names they could recall, and the corresponding brand message. Subsequently they were to indicate on a list containing brand names whether they recognized these and the certainty of that answer. This certainty score was captured by a 5-point scale with a '1' indicating complete uncertainty and a '5' indicating complete certainty. The recognition scores were multiplied by the certainty scores to obtain the final recognition measures to be used in the subsequent analyses.
Two main hypotheses, each referring to generic variables and operationalizations, were tested:
H1: The independent variables motivation, capacity and opportunity can be independently manipulated. This means that the intercorrelations between these factors will be low.
H2: The independent variables motivation, capacity and opportunity will exert significant main effects on the recall, recognition and message content data.
The data collected in the main experiment were subjected to the following analyses to provide a test of the hypotheses.
Preparation of the Analyses
To test the hypotheses, two types of analyses were performed. Only the responses to the experimental stimuli were processed. The first analysis assessed the independence among subjects' perceptions of the predictor variables motivation, capacity and opportunity. To this end, subjects' responses on the subjective experience variables were correlated for each billboard. In total, 16 * 3 = 48 coefficients of correlation obtained.
To answer the second hypothesis, the data obtained in the experiment were prepared for multivariate analysis of variance (MANOVA) as follows. The recall measurements of the bill board advertisements were summed per cell, as the individual scores showed little variation in general. The same procedure applied to the message scores. The recognition scores for each billboard were multiplied by the certainty scores indicated by the subjects. The resultant combined scores were also summed for each experimental cell. These summed scores for the recall, recognition and message measurements for each individual were treated as repeated measurements in the MANOVA.
The billboards were classified into appropriate categories of the independent variables based on the subjective experiences of the subjects. Each cell of the design contained two billboards.
Independence of Predictor Variables
To assess the independence of the predictor variables, the scores on items measuring the perceptions of these variables were correlated per stimulus. To simplify the presentation, the median of the coefficients of correlation obtained in each phase of the investigation are presented in Table 3.
Table 3 provides a general overview with regard to the assumed independence of the predictor variables. The variation among the calculated coefficients per study was substantial. This finding suggests that, even per pair of stimuli in each experimental cell, the empirical correlations between the independent factors varies. In this fashion, Table 3 presents evidence about the relative success of the operationalizations of the predictor variables over the experimental stimuli.
Given the sizes of the median coefficients, is seems that the assumption of independence between the predictor variables cannot be dismissed immediately. The coefficients are uniformly low, and hence indicate that the pairs of variables share only a marginal portion of the variance; this portion is 17% at most. So these coefficients are taken to indicate that in terms of subjective experiences, subjects judged these factors to be independent from each other. Consistent with this reasoning, the significance of the contents of the correlation is given precedence over any statistical significance. Hypothesis 1, therefore, received empirical support in the present investigation.
MEDIANS OF COEFFICIENTS OF CORRELATION BETWEEN THE PREDICTOR VARIABLES IN THE DIFFERENT PHASES OF THE INVESTIGATION
RESULTS OF OMNIBUS TESTS OBTAINED IN MULTIVARIATE ANALYSES OF VARIANCE
Results of Multivariate Analysis of Variance
Table 4 contains the results of the multivariate analyses of variance. The multivariate effects of the motivation and the capacity variable were highly significant (p < .001), where the opportunity variable did not exert a significant main effect. These effects can be interpreted by inspecting the means for the levels of the independent variables in Table 5. Higher motivation to process brand related information on the billboard led to increased recall and recognition of the information, as well as a more accurate description of the message contents. Identical effects were found for the capacity variable, indicating that increased capacity to process brand related information yielded more accurate recall and recognition. The results of the opportunity variable seem to indicate that the means of the dependent variables did not differ substantially between the low and high opportunity conditions. Inspection of Table 5, however, shows that the differences among the means were in the expected direction. In sum, these findings might hint at an operationalization of the opportunity variable which potentially requires reconsideration. It should be remembered here that subjects were presented with slides that were categorized using data collected in the pretest studies. Based on the results of the first pretest, the opportunity manipulation in the second pretest was changed into 14 seconds of exposure of the slide in the high opportunity condition and to 6 seconds for exposure in the low opportunity condition. The absence of a significant effect might reflect a weak manipulation, but we would like to suggest an alternative explanation. It is possible that even the exposure time in the low opportunity condition already exceeded a certain critical threshold. The lower limit of six seconds may be too high to generate sufficient variance in the dependent measures.
The interaction terms of the independent variables yielded mixed effects; Table 6 contains the summary statistics of the multivariate and univariate test of these effects. The two-way interactions all affected the recognition variable, and one affected recall of brand name as well. The three-fold interaction influenced only the message recall criterion. Given these mixed effects, the interaction terms are not considered fully here. Table 7 contains the necessary data to evaluate the interactions. A short description of the effects of the two-way interactions is provided in order to get an appreciation of what the results indicate. The motivation * capacity interaction indicated that brand recognition is enhanced by increasing capacity in a situation of high motivation. The motivation * opportunity interaction showed that increasing opportunity in a low motivation condition improved brand recognition. The capacity * opportunity interaction suggested that brand name recall and brand name recognition are superior when high opportunity is available in a high capacity situation. That finding is contrary to the result found in the low capacity situation, where low opportunity seems to improve performance. The second order interaction for recall of message contents indicated that the capacity * opportunity interaction differed for the low and high motivation conditions. As can be seen from the last row in Table 7, the direction of the differences between the means for the opportunity conditions reverses with the low and high capacity conditions.
The data depicted in Tables 4 and 5 yield substantive support for the second hypothesis, which stated that the independent variables motivation, capacity and opportunity will exert significant main effects on the recall, recognition and message content data. Significant main effects were found for the motivation and capacity factors on all dependent variables in the expected directions. The opportunity factor only influenced the recognition variable. The latter finding suggests that the difference between both opportunity levels may not have been large enough to induce the expected difference in information processing.
DISCUSSION AND CONCLUSIONS
For the sake of brevity and clarity, we will present some concluding observations in separate points:
- It seems possible to not only conceive of generic variables, but also to operationalize them in a way that allows for useful discriminations among different advertisements.
- A full within-subjects approach would be preferable to the between group comparison presented in the present paper. Here, the categorization of stimuli was derived from a number of pretests and applied in the main experiment which involved different subjects. In future studies attention will have to be paid to the development of more adequate within-subjects design.
- It is important to relate the operationalizations of the predictor variables to the specific advertising effectiveness or advertising impact variables that are used as criterion measures in advertising research. Differences in dependent variables may be reflected in the operationalizations of the independent variables.
- In future studies research also the operationalization of the three generic dimensions will have to be refined. A special point worth considering in such studies is that the independence of the operationalizations may not be a matter of semantics only, but also a matter of the time allowed for processing and elaboration. By increasing exposure or elaboration time, the likelihood that the three variables may influence each other may increase, which then would be expressed in higher intercorrelations among the three variables.
- Of course, the present study dealt with the extent of ad processing, not with the extent to which the advertiser's goal was met. The two effects can overlap but not necessarily so.
- In spite of some conceptual and methodological questions that were raised by this research, the present approach seems worth pursuing further as an approach next the one identified in the introduction, which focused upon variables on a more specific level only.
CELL MEANS AND STANDARD DEVIATIONS FOR THE DEPENDENT VARIABLES
OVERVIEW OF MULTIVARIATE AND UNIVARIATE TESTS OF INTERACTION EFFECTS
CELL MEANS AND STANDARD DEVIATIONS FOR THE DEPENDENT VARIABLES IN THE EXPERIMENTAL CELLS
Andrews, J. Craig (1988), "Motivation, Ability, and Opportunity to Process Information: Conceptual and Experimental Manipulation Issues," Advances in Consumer Research, 15, 219-225.
Bahrick, Harry P. (1970), "Two-phase model for prompted recall," Psychological Review, 77, 215-222.
Jones, Gregory V. (1987), "Independence and Exclusivity Among Psychological Processes: Implications for the Structure of Recall," Psychological Review, 94, 229-235.
MacInnis, Deborah J. and Bernard J. Jaworski (1989), "Information Processing from Advertisements: Toward an Integrative Framework," Journal of Marketing, 53, 1-23.
Lutz, Richard J., Scott B. MacKenzie, and George E. Belch (1983), "Attitude toward the Ad as a Mediator of Advertising Effectiveness: Determinants and Consequences," Advances in Consumer Research, 10, 532-539.
Mandler, George (1980), "Recognizing: The Judgment of Previous Occurrence," Psychological Review, 87, 252-271.
Petty, Richard E. and John T. Cacioppo (1986), Central and Peripheral Routes to Attitude Change, New York: Springer-Verlag.
Poiesz, Theo B. C. (1989), De transformatie van een karikatuur [The transformation of a caricature], Tilburg: Tilburg University.
Rossiter, John R., Larry Percy and Robert J. Donovan (1991), "A Better Advertising Planning Grid," Journal of Advertising Research, 32, 11-21.
Stewart, David W. and Scott Koslow (1989), "Executional Factors and Advertising Effectiveness: A replication," Journal of Advertising, 18, 21-32.
Zaichkowsky, Judith Lynne (1985), "Measuring the Involvement Construct," Journal of Consumer Research, 12, 341-352.