Factors Affecting the Use of Conceptually Driven and Data Driven Processing

Joan Meyers-Levy, UCLA
ABSTRACT - When apprehending new information, people may engage in data driven processing, which involves a detailed analysis of the incoming information, or they may pursue conceptually driven processing whereby apprehension is more cursory and largely guided by preexisting knowledge. A study is reported that examines the effects of the level and salience of cue incongruity on males' and females' choice of processing strategies. It is found that unless motivated by unequivocal evidence of incongruity, males favored the use of conceptually driven processing. Females, however, reliably employed data driven processing even when cue incongruity was less easily detected. The implication of these findings for consumer behavior is discussed.
[ to cite ]:
Joan Meyers-Levy (1988) ,"Factors Affecting the Use of Conceptually Driven and Data Driven Processing", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 169-173.

Advances in Consumer Research Volume 15, 1988      Pages 169-173

FACTORS AFFECTING THE USE OF CONCEPTUALLY DRIVEN AND DATA DRIVEN PROCESSING

Joan Meyers-Levy, UCLA

ABSTRACT -

When apprehending new information, people may engage in data driven processing, which involves a detailed analysis of the incoming information, or they may pursue conceptually driven processing whereby apprehension is more cursory and largely guided by preexisting knowledge. A study is reported that examines the effects of the level and salience of cue incongruity on males' and females' choice of processing strategies. It is found that unless motivated by unequivocal evidence of incongruity, males favored the use of conceptually driven processing. Females, however, reliably employed data driven processing even when cue incongruity was less easily detected. The implication of these findings for consumer behavior is discussed.

INTRODUCTION

A recurrent theme in cognitive psychology is that the processing of informal ion can be thought of as occurring in two modes (Bobrow and Norman 1975). One mode involves performing a detailed analysis of incoming stimulus cues in an effort to assemble cues into a coherent higher order representation. This analysis, which is referred to as data driven processing (DDP), typically results in a highly veridical representation (Glass, Holyoak and Santa 1979). The other mode of processing simplifies operations by relying more heavily upon existing schematic knowledge. Referred to as conceptually driven processing (CDP), this mode of processing entails drawing on existing schematic knowledge to form expectations about incoming information. These expectations are then used to guide processing such that incoming cues are coalesced with memory-based information, thereby further instantiating preexisting knowledge. These two processing strategies can be thought of as falling on a continuum in which both are often simultaneously activated (Anderson 1983) but a relative emphasis is devoted either to CD or DDP depending on the prevailing task conditions [While it is understood that processing typically entails some degree of both CD and DD modes of analysis, for ease of exposition these terms will be used to connote a relative emphasis on one of these strategies.].

The use of CDP has virtues as well as liabilities. Schematic knowledge that is activated during such processing provides a ready structure to impose on stimuli that might otherwise overwhelm the processor. Thus, CDP is believed to offer substantial survival value (Bahrick 1984) and is thought to be cognitively efficient because it circumvents more effortful detailed analysis of the stimulus (Taylor and Crocker 1981). At the same time, CDP has been found to impair accurate memory for the particular information presented at acquisition because stimulus cues are subjected to considerable assimilation and abstraction to accommodate them with prior knowledge (Alba and Hasher 1983). Moreover, schema-based inferences generated during CDP are sometimes confused with actual stimulus features leading to false intrusions of never-presented but schematically congruent stimulus features (Bower, Black and Turner 1979).

Given that CDP is characterized by both assets and liabilities, it follows that some conditions are likely to prompt the use of DD processing, which at times may be-more adaptive. Recent inquiry in social cognition implies that schema-incongruent or discrepant stimulus cues can motivate such processing (Hastie 1980, 1981; Srull 1981). In this research, subjects are typically asked to form an impression of a person and are led to expect that this individual possesses a particular trait. Subjects then are presented with a list of various behaviors engaged in by the person of which some are congruent and others are incongruent with schematic knowledge implied by the trait.

The recurring finding in these studies is that the presence of incongruent cues enhances accurate recall of all stimulus information, though the proportion of incongruent cues recalled typically exceeds that of congruent cues. This outcome is thought to occur because in an effort to make sense of the incongruous cues, individuals consider them in relation to congruent stimulus items that already have been encountered. Hence, associative linkages are formed between incongruent and congruent cues, which benefit memory for all the stimulus information. Thus, the view that emerges is that cue incongruity serves as a motivational device to prompt DDP, which is manifested in substantial accurate recall of the stimulus information. Moreover, the incidence of false intrusions during DDP is minimal (c.f Mandler 1980).

Cue incongruity is often used in advertisements directed at consumers. For example, an ad that introduces desert spice scent Sure deodorant depicts the physical product protruding from a deodorant container in an incongruent form that resembles desert terrain. An ad for Pall Mall cigarettes with filters heralds the product as "a mild breakthrough" by showing an oversized package of the cigarettes breaking though the roof of a barn. Given the frequent use of incongruity in ads, it is not surprising that consumer researchers have examined the effect of cue incongruity on processing. Similar conclusions obtain as those implied by the social cognition studies. For example, Sujan (1985) found that when product information was entirely consistent with subjects' preconceptions, processing appeared to be CD in that it centered on the rehearsal of subjects' prior product knowledge. But when the message contained incongruent information, individuals' processing appeared to be more DD: Relative to subjects in the former condition, those exposed to incongruity engaged in greater cognition and produced both more attribute specific thoughts. These effects, however, emerged only among subjects who were experts concerning the product category. Apparently because novices' limited knowledge of the category impaired their detection of the incongruity, novices were not motivated to employ DDP.

The preceding analysis suggests that message inclusion of incongruent information may not always stimulate DDP. People may fail to detect cue incongruity for a variety of reasons and therefore may not engage in DDP. For example, incongruity might not be detected if the incongruent cues are only mildly discrepant with other stimulus cues or if contextual factors reduce the attention devoted to the incongruent cues. In addition, individual differences in processing proclivities may limit the use of DD analysis.

This article investigates the conditions under which people invoke CD and DDP strategies. Two factors thought to moderate the influence of cue incongruity on processing strategy were examined: the difficulty of detecting cue incongruity and general differences in the genders' processing proclivities. These factors are of interest to consumer researchers because, as discussed earlier, incongruity is often employed in messages targeted at males and/or females. Thus, it would seem important to determine whether some incongruent cues may go relatively unnoticed and fail to induce DDP, and whether the genders might sometimes be differentially responsive to incongruity. A discussion of these moderating factors follows.

Factors Affecting the Detection of Cue Incongruity

An assumption in the incongruity research is that the processor necessarily detects cue incongruity and this in turn motivates DDP. However research by Hastie and colleagues (cited in Hastie 1980) suggests that one's choice of processing strategy may be contingent upon the level of cue incongruity. Cues that are low in incongruity may fail to motivate the use of DDP because the modestly aberrant content of these cues may be perceived not as incongruent but rather as fairly irrelevant to the stimulus-implied schema. As Alba and Hasher (1983) have noted, irrelevant cues may receive little or no elaborative processing. These cues can be assimilated quite readily then with schematically congruent cues in a CD fashion. Thus, cues that are modestly atypical but not in clear opposition to the schema may fail to invoke the associative activity that characterizes DD processing. This implies that DDP will occur when the content of selected stimulus cues is relatively high in incongruity but CDP will occur when this content is low in incongruity.

Detection of cue incongruity may also be influenced by the contextual salience of the incongruent cues. It would seem possible that even cues low in incongruity might be perceived as incongruent if they were made highly salient to the processor. Suppose, for example, that these low incongruity cues were positioned in a text-like message as discrete nonsequential cues (e.g. two incongruent items separated by two intervening congruent cues). The temporally distinct multiple exposures to the incongruent information might promote the detection and elaboration of the cues' incongruous qualities such that DDP would be stimulated. By contrast, if these low incongruity cues were presented contiguously in the stimulus message, thereby appearing as a unitary mass of incongruent information, the processor might be especially likely to overlook this singular unitized instance of low incongruity information and process the information in a CD manner [This hypothesis concerning cue positioning is based on the results of studies that have examined the "spacing effect." Such studies consistently find that memory for a repeated item is enhanced when repetitions are discretely spaced rather than massed (Hintzman 1974).]. Hence, contextual factors such as the positioning of incongruent cues within a message also are hypothesized to influence the choice of processing strategy.

Individual Differences with Regard to Gender

Finally, there is reason to believe that individual differences in terms of gender enhance the likelihood that either CD or DDP will be employed. Studies in the gender literature suggest that males may exhibit a greater propensity to engage in CDP relative to females, who may favor DD processing. For example, in a study by Christensen and Rosenthal (1982) males and females were provided with an expectancy about a person prior to their interviewing this individual. It was found that males' judgments were largely CD, reflecting the predominate use of the expectancy information. Females' judgments were more sensitive to information encountered during the interview, implying the use of DDP. Virtually identical outcomes have been reported by Farina (1982) who investigated the effect of stereotypes on the genders' judgments. Male job interviewers were very unfavorable toward a confederate interviewee who was described as a former mental patient relative to one who was normal. However, for females, the interviewee's history made no difference and instead judgments reflected the confederate's performance in the interview.

Research by Cupchik and Poulos (1984) also maintains this view concerning the genders' processing proclivities: While females tend to make fine discriminations that are based on both stimulus information and expectations, males render judgments that "easily fulfill their broadly defined expectations,' (p. 438). Thus again, females are seen as more predisposed to using DDP than are males, who favor CDP.

The study that follows examines the conditions under which individuals are likely to use CD and DD strategies. Males and females were presented with a description of a new in-depth television news program that contained a majority of features consistent with those possessed by such existing programs. Three independent variables were introduced in the context of a between-subjects factorial design. These included two levels of gender, two levels of contextual positioning of the incongruent cues (unitized, discrete) and three levels of cue incongruity (low, moderate, high). Subjects' choice of processing strategy was detected by examining treatment differences in accurate recall of the stimulus information as well as the incidence of false intrusions that were congruent with schematic knowledge but had not appeared in the stimulus.

On the basis of the preceding analysis, the use of CD and DDP was expected to vary depending upon the treatment conditions. Males' proclivity toward the use of CDP implies that they will engage in DDP only when cue incongruity is readily detected, thereby motivating DDP. Such conditions should prevail when cue incongruity is moderate or high. Males might also engage in DDP when cue incongruity is low if these cues are discrete (i.e. nonsequential) in their positioning. This discrete positioning may sufficiently emphasize cue incongruity such that males are motivated to engage in more DD analysis. However, predictions differ for males exposed to the low incongruity cues that are unitized in their positioning (i.e. sequential pairing). Because the salience of cue incongruity is understated in this treatment, males are expected to resort to CDP. By contrast, females' general proclivity toward DDP suggests that they will employ a DD strategy regardless of variations in the content or the positioning of the incongruent cues. Hence, females in all treatments are expected to engage in DDP.

METHOD

Subjects

Sixty-one males and 60 females, age 18 to 55, were recruited to participate in a new product study. Subjects were run in small groups and were screened such that they all watched 15 or more hours of television each week.

Stimulus

Subjects were randomly assigned different versions of booklets that contained the experimental materials. They were told that heir reactions were sought to a new television program that might be introduced by a major network. No reference was made to the recall task later administered. Subjects then read a description of a proposed new-half hour in-depth news program to be aired after the evening news. The description of this program closely resembled ABC's Nightline with Ted Koppel, a well known exemplar of the in-depth news program category. Information presented to subjects concerning the stimulus program's format and host was based on that of Nightline, rendering this stimulus information congruent with the in-depth news show category (e.g. the half hour in-depth news program would feature a well-known host who would moderate the discussion of various newsworthy topics and experts on featured issues would be presented).

Also included in the stimulus description were eight topical issues that were said to be scheduled for discussion on the show. Common to all stimulus descriptions were six issues that were found in pretests to be congruent with program content on in-depth news shows. These issues were: medical malpractice, United States presence in El Salvador, America's problem of drug infiltration, drug abuse, combating terrorism, and worldwide hunger. Also mentioned were two additional issues that varied in their degree of incongruity with the program category. Pretests indicated that these critical issues or cues were either of low (M=4.61), moderate (M=6.22), or high (M=8.15) incongruity (ratings were obtained on 11 point scales anchored as extremely consistent/inconsistent with items featured on in-depth news shows; higher numbers indicate greater incongruity). No gender differences were evident in these pretest ratings. The pairs of issues, which represented the critical incongruent stimulus cues, were "national weather forecasts" and "movie reviews" in the low incongruity condition, "healthful nutrition" and "organizing family finances" in the moderate incongruity condition, and "magic performances" and "poetry readings" in the high incongruity condition.

Another independent variable was the contextual positioning of the incongruent cues. Either the incongruent cues were unitized such that they were positioned together and presented consecutively in the stimulus description or they were discrete and positioned as two independent cues.

Dependent Measures

After reading the program description, subjects answered a series of questions concerning their general television viewing habits. These questions were included to limit short term memory effects.

Subjects then performed a recall task. They were asked to recall as accurately and completely as possible all statements presented in the stimulus material. No time limit was imposed. Finally, to determine subjects' familiarity with in-depth news shows such as the stimulus program, the frequency with which subjects watched Nightline was assessed because this was the show upon which the stimulus message was based.

RESULTS AND DISCUSSION

Confounding Checks

To ascertain whether gender was confounded with familiarity of in-depth news programs, an ANOVA was performed on the frequency with which subjects watched Nightline, the in-depth news program most like the stimulus program. This analysis revealed that Nightline was watched frequently (M=2.3 times per week, SD=1.61) and both genders watched the program with equal frequency (p<.94). Hence, gender differences in familiarity with shows similar to the stimulus program can be ruled out as a causal explanation for treatment effects on the other dependent measures.

Recall

To assess treatment effects on recall, analyses of variance were performed on the 2 (gender) by 2 (positioning of incongruent cues: discrete or unitary) by 3 (cue incongruity: low, moderate, and high) factorial design. Two planned orthogonal contrasts were performed on the 3 levels of cue incongruity: a comparison of the low versus moderate incongruity cue conditions and a comparison of the low versus high incongruity cue conditions. Treatment effects involving these two contrasts tended to be the same because, as anticipated, effects obtained only within the low incongruity cue condition. However, effects involving the low versus moderate incongruity cue contrast were stronger than were those involving the low versus high incongruity cue contrast, perhaps because the main effect of the latter contrast dominated the interaction. Hence, in reporting the results, focus will center on the effects as they pertain to the low versus moderate incongruity cue contrast, though all effects will be reported.

Based on the logic presented earlier, it was anticipated that detection of subjects' processing strategy would be revealed by examining the total number ofstimulus items accurately recalled and the number of false intrusions that were congruent with the in-depth news show category but did not appear in the stimulus. The use of DDP was expected to result in substantial recall of all stimulus items. Subjects adhering to this strategy should devote extra processing to the aberrant incongruent cues and, in an effort to explain them, relate these cues to the congruent stimulus items. Because detailed consideration would be afforded both types of stimulus cues, subjects' recall of these items should be high but the incidence of congruent false intrusions should be minimal. By contrast, the use of a CD strategy should produce a relative deficit in recall of stimulus items but a sizeable number of false intrusions. Subjects who use to this strategy should filter out the modestly aberrant stimulus input, thereby obviating detailed consideration of the incongruent and congruent items. Instead, processing would center on rehearsal of preexisting knowledge that is consistent with the stimulus theme. Hence recall of all stimulus items should be poor but false intrusions should be substantial.

Using a gist criterion, separate measures were coded for the number of stimulus items recalled and the number of congruent false intrusions. These recall classifications were analyzed as a proportion of the total number of all items recalled. Treatment means for each of these measures are reported in Table 1.

Accurate recall of the stimulus items. An analysis on accurate recall of all stimulus items revealed a marginally significant main effect of the low versus high incongruity cue contrast (F(1,108)=3.45, p<.06) and a main effect of positioning of the incongruent cues (F(1,108)=4.24, p<.04). Accurate recall of the stimulus items was greater in the high than in the low incongruity cue condition. This finding is consistent with the notion that highly incongruent stimulus cues are more likely to prompt DDP as the processor attempts to understand the incongruent cues relative to other available stimulus items. Moreover, the main effect of cue positioning serves as a manipulation check, verifying that the unitized positioning of the incongruent cues diminished the salience of these cues relative to when they were discretely presented, thereby undermining stimulus recall. These effects were qualified by an interaction of gender by positioning of incongruent cues by the low versus moderate incongruity cue contrast (F(1,108)=3.48, p<.05). The analogous interaction involving the low versus high incongruity cue contrast approached significance (F(1,108)=3.31, p<.07). As was anticipated, these effects can be attributed to a gender by positioning of incongruent cues interaction that appeared in the low incongruity cue condition (F(1,108)=4.94, p<.03) but not in the moderate or high incongruity cue conditions (Ps<1.00).

As can be seen in Table 1, in the low incongruity cue condition males evidence I lower recall of the stimulus items when the positioning of incongruent cues was unitized rather than discrete (F(1,108)=8.92, pc.003), while females manifested equivalent recall of congruent stimulus items across both incongruent cue positioning conditions (F<1). Furthermore, males' recall of the stimulus items in the low incongruity cue condition was lower than was that of females when the positioning of incongruent cues was unitized (F(1,108)=3.85, p<.OS), but not when incongruent cue positioning was discrete (po.22).

Thus, consistent with predictions, when low incongruity cues were presented as a unitized mass, males' accurate recall of the stimulus information was impaired, implying that these males used CDP in apprehending stimulus material. Males in the remaining experimental conditions and females in all conditions exhibited substantial recall of the stimulus items, which suggests that they engaged in DD processing.

TABLE 1

TREATMENT MEANS FOR PROPOSITIONAL RECALL OF ALL STIMULUS ITEMS AND CONGRUENT FALSE INTRUSIONS

Recall of false intrusions. Analysis on recall of congruent false intrusions revealed an inverse patterns of effects. A main effect of the low versus high incongruity cue contract (F(1,108)=8.51, p<.004) suggested that recall of congruent false intrusions was greater in the low than in the high incongruity cue condition. This finding suggests that overall, message inclusion of high incongruity cues is more likely to stimulate DDP relative to inclusion of low incongruity cues, which may generate CDP.

The interaction of gender by positioning of incongruent cues by low versus moderate incongruity cue contrast approached significance (pc.10) as did the analogous interaction involving the low versus high incongruity cue contrast (p<.12). These interactions were explored further because predictions relevant to them were posited a priori. The interaction of gender by positioning of incongruent cues was marginally significant in the low incongruity cue condition (F(1,108)=3.08, p<.08) but not in the moderate or high incongruity cue conditions (Fs<1).

As suggested by the means reported in Table 1, in the low incongruity cue condition males demonstrated greater recall of false intrusions when the positioning of incongruent cues was unitized rather than discrete (F(1,108)=4.94, p<.03), whereas females manifested a constant level of distorted recall across both cue positioning conditions (F<1). Tests of the remaining simple effects failed to reach significance (Fs < 1.14).

Though effects observed on this measure are somewhat attenuated, they converge with recall of all stimulus items in the interpretation they imply. The observation that males exhibited substantial incidence of false intrusions in their recall when the low incongruity cues were unitized implies that males employed CDP in this treatment. Males' relatively low incidence of false intrusions in the remaining conditions and females' generally infrequent incidence of false intrusions in all conditions implies that these subjects engaged in DDP of the stimulus information.

GENERAL DISCUSSION

The data from this study provides insight into when individuals use CD and DDP in apprehending message information. As existing research suggests, when individuals detect cue incongruity, they are likely to be motivated to engage in DDP. The observation that both genders exhibited substantial recall of the stimulus information and minimal incidence of false intrusions when the stimulus contained unambiguously (moderate and highly) incongruent cues implies that these subjects engaged in DDP. Apparently, detailed consideration was devoted to the stimulus cues in these conditions.

When detection of cue incongruity is undermined, however, CDP seems to be encouraged. The current study suggests that when cues are only modestly incongruent and are positioned as single units, the cues' incongruous characteristics are more difficult to detect, thereby encouraging the use of CDP. And because males seem to possess a greater proclivity to engage in such processing than do females, this strategy was apparently used by males in this condition. This is implied by males' less accurate recall of the stimulus items and their elevated incidence of false intrusions. Such deficits were absent in females' recall, implying that females engaged in DDP. These deficits were also absent in the recall of both genders when low incongruity cues were rendered more salient due to their discrete positioning in the message. This implies that both genders detected the modestly incongruous cues when they were positioned as discrete entities and thus engaged in DDP.

Taken together, these findings imply that cue incongruity need be quite pronounced to motivate males to employ DDP. Thus, in designing ads targeted at males, marketers who wish to encourage detailed processing of the message may want to accentuate product features that represent a point of difference. As demonstrated in the present study, points of difference that are relatively subtle or modestly incongruent might be accentuated contextually by their positioning in the message. Other methods of emphasizing low incongruity cues might include the use of visual devices (e.g. color, bold typeface, etc.) that enhance cues' attention-getting properties. Embellishment of incongruous product features may be less necessary when targeting females. Because females seem to favor the use of DDP in general, they appear to pursue such processing even when messages contain only modestly incongruent items. Whether this implies that special measures might be necessary to induce females to employ CDP when incongruent features are absent is an issue that must await future investigation.

REFERENCES

Alba, J.W., and Hasher, L. (1983), "Is Memory Schematic?" Psychological Bulletin, 93 (March), 203-231.

Anderson, I.R. (1983). The Architecture of Cognition. Cambridge, Mass: Harvard University Press.

Bahrick, H.P. (1984), "Semantic Memory Content in Permastore: Fifty Years of Memory for Spanish Learned in School," Journal of Experimental Psychology: General, 13, 1-29.

Bobrow, D. G. and Norman, D.A. (1975). "Some Principles of Memory Schemata," in Representation and Understanding: Studies in Cognitive Science, eds. D. G. Bobrow and A. Collins, N.Y.: McGraw-Hill.

Bower, G.H., Black, J.B., and Turner, TJ. (1979), "Scripts in Memory for Text," Cognitive Psychology, 11 (April), 177-220.

Christensen, D. and Rosenthal, R. (1982), "Gender and Nonverbal Decoding Skill as Determinants of Interpersonal Expectancy Effects," Journal of Personality and Social Psychology, 42 (January), 75-87.

Cupchik, G. and Poulos, C.X. (1984), "Judgments of Emotional Intensity in Self and Others: The Effect of Stimulus Context, Sex, and Expressivity," Journal of Personality and Social Psychology, 46 (February), 431-439.

Farina. A. (1982). 'The Stigma of Mental Disorders,"  In the Eye of the Beholder: Contemporary Issues in Stereotyping, ed. A.G. Miller, N.Y.: Praeger, 305-363.

Glass, A.L., Holyoak, KJ., and Santa, J.L. (1979), Cognition, Reading, Mass.: Addison-Wesley

Hastie, R. (1980), "Memory for Behavioral Information that Confirms or Contradicts a Personality Impression," in Person Memory: The Cognitive Basis of Social Perception, eds. R. Hastie, T.M. Ostrom, E.B. Ebbesen, R.S. Wyer, Jr., D.L. Hamilton, and D.E. Carlston, Hillsdale, NJ.: Lawrence.

Hastie, R. (1981), "Schematic Principles in Human Memory," in Social Cognition: The Ontario Symposium, eds. E.T. Higgins, C.P. Herman, and M.P. Zanna, Hillsdale, N.J.: Erlbaum.

Hintzman, D.L. (1974), "Theoretical Implications of the Spacing Effect." in Theories in Cognitive Psychology : The Loyola Symposium, ed. R. L. Solso, Potomac, M.D.: Erlbaum.

Mandler, J.N. (1980), "Categorical and Schematic Organization in Memory," in Memory Organization and Structure, ed. C. R. Puff, N.Y.: Academic Press.

Srull, T.K. (1981), "Person Memory: Some Tests of Associative Storage and Retrieval Models," Journal of Experimental Psychology, 110 (December), 440-463.

Sujan, M. (1985), "Consumer Knowledge: Effects of Evaluation Strategies Mediating Consumer Judgments, Journal of Consumer Research, 12 (June), 31-46.

Taylor, S.E., and Crocker, J. (1981), "Schematic Bases of Social Information Processing," in Social Cognition: The Ontario Symposium, eds. E.T. Higgins, C.P. Herman and M.P. Zanna, Hillsdale, NJ.: Erlbaum.

----------------------------------------