The Effects of Product Knowledge on the Comprehension, Retention, and Evaluation of Product Information

Joseph W. Alba, University of Florida
ABSTRACT - Subjects rated as high or low in their knowledge of stereos read, at their own pace, an extensive description of a fictitious stereo system and were later given a surprise recall test. Results showed that while knowledge level did not affect reading time, it did account for large differences in quantity and content of recall. Possible explanations and implications of these findings are discussed.
[ to cite ]:
Joseph W. Alba (1983) ,"The Effects of Product Knowledge on the Comprehension, Retention, and Evaluation of Product Information", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 577-580.

Advances in Consumer Research Volume 10, 1983      Pages 577-580


Joseph W. Alba, University of Florida


Subjects rated as high or low in their knowledge of stereos read, at their own pace, an extensive description of a fictitious stereo system and were later given a surprise recall test. Results showed that while knowledge level did not affect reading time, it did account for large differences in quantity and content of recall. Possible explanations and implications of these findings are discussed.


The issue of expertise - or more generally, topic familiarity - and its effects on comprehension and memory have long been subjects of interest to researchers in a variety of fields. Owing to the direct and indirect influence of Bartlett (1932) and the subsequent development of schema theory, concerns about the effects of prior knowledge on information processing have been dominant in the areas of educational, cognitive, and social psychology, and presently, consumer behavior research. Taken collectively, the resultant body of research generated from these diverse disciplines has shown that one's ability to comprehend and recall information relevant to some domain varies as a function of the amount of knowledge one possesses with respect to that same domain. What is especially noteworthy about this finding is the robustness of the relationship. Over a wide range of topics and experimental procedures it has consistently been demonstrated that as the amount of prior knowledge about a topic or the level of activation of that knowledge at the time of processing increases, so too do comprehension and recall of newly presented information pertaining to that topic (see, e.g., Ausubel 1968; Bransford & Johnson 1972; Chiesi, Spilich, & Voss 1979).

To date, this precise relationship between expertise and memory has not been demonstrated within the context of consumer behavior. The present study will attempt to fill this void in the literature. It is important to note that the issue of prior knowledge and its effects on memory should be of cent-al concern to consumer psychologists, especially in light of the distinction between memory-based and stimulus-based processing discussed recently by Lynch and Srull (1982). As they indicate, while it is true that brand comparisons are sometimes made in the presence of complete and tangible information, it is also true that some decisions are based primarily upon the information a consumer is able to retrieve from memory at the time of the decision. Accepting the validity of this premise, it is apparent that for some decisions the quality of choice may depend critically on the amount and content of retrieved relevant information. And, if the relationship between knowledge and recall holds in the domain of product information, it then follows that high-know] edge consumers should be capable of making better decisions than low-knowledge consumers. This hypothesis is examined in the study below.


A total of 79 undergraduates were asked to read a lengthy and somewhat technical description of a fictitious new stereo system. Subjects were instructed to read the "ad" at their own pace ar.d were told that their primary focus should be on comprehending the information. Comprehension was emphasized in order that subjects not consider the task a speed-reading one despite the obvious fact that they were being timed. After reading the final sentence, reading time was recorded. The reading task was followed by a series of questions designed to evaluate the stereo, the ad, and the knowledge and experience of the subjects with regard to stereos. Finally, a surprise recall test was given. Subjects were allowed 15 minutes to recall the information as completely and accurately as possible.


Recall and Beading Time

Knowledge level was self-assessed via a 7-point scale in which a "1" represented very low and "7" very high knowledge of stereos. Subjects rating the,r knowledge at one of the lowest three levels were classified as low in stereo knowledge; subjects rating themselves at one of the highest three levels were classified as high in stereo knowledge. Seventeen people rated themselves at the midpoint in the scale and were eliminated from the analysis. -- All analyses to be reported below employ knowledge level as the independent variable. However, similar patterns of results are found when subjects are separated according to self-assessed comprehension of the stimulus, i.e., low vs. high, and when they are separated according to whether or not they had searched for a stereo system in the last two years. Thus, it appears as though the present method of selecting subjects has some validity.

The stimulus was roughly broken down into 257 total idea units. An idea unit is a sentence or sentence fragment which conveys a single piece of information about the stereo. Of these, 116 were classified as representing "complex" or highly technical information. Reading time and recall scores for both high-knowledge and low-knowledge subjects are presented in Table 1. Reading time is expressed in minutes; recall is expressed in idea units.



Looking first at reading time, it is clear that knowledge level did not affect the amount of time required to read the information (F<1). This is somewhat surprising in light of the large body of research demonstrating that more easily comprehended information is read at a faster rate than less easily comprehended information (see, e.g., Haviland & Clark, 1974). It is the case, however, that unlike the present experiment, most of the research that has investigated reading time has done so at the individual sentence level, thereby making any strong comparisons impossible. It is also possible that our task instructions did not completely achieve their intended purpose. Subjects may have ignored the directions emphasizing comprehension over reading speed, thereby washing out any overall effect of Prior knowledge.

Since there was no effect of knowledge on reading time, its effect on recall will not be confounded bs differences in amount of study time. Table 1 shows the predicted effect of knowledge on recall. Consistent with previous research, high-know]edge subjects recalled significantly more idea units than low-knowledge subjects (p<.02). Thus, the conclusions drawn from studies tapping other domains of knowledge seem generalizable to product knowledge situations.

Perhaps more interesting than sheer quantity of recall is the nature of the recalled information. As Table 1 indicates, high-knowledge people recalled more complex information than low-knowledge people (p<.01) and that this difference completely accounts for the difference found in total recall (cf. Chiesi et al. 1979) The interaction between knowledge level and type of ideas recalled is represented in Figure 1, in which the ordinate denotes the proportion rather than the absolute number of idea units recalled.



In retrospect, these results are not surprising. The effects of prior knowledge should be observed only in chose cases in which there are differences in people' ability to comprehend information. Simpler ideas - ideas presumably which all subjects can understand such as facts related to styling and other less technical information about stereos should be recalled at equal rates across groups of people possessing different amounts of technical knowledge. On the other hard, since it is generally technical knowledge which separates high-knowledge from low-knowledge people, differences in recall should be observed when complex information is considered in isolation.


While the vast majority of recalled ideas can be considered to be accurate, if not verbatim, restatements of stimulus information, there was a small amount of "inaccurate" retrieval. Across groups, there were fewer than three intrusions per subject. Nevertheless, this amount was deemed substantial enough to warrant further analysis. Recall "errors" were placed into one of four categories: (1) generalizations - statements which combined separate facts true of individual components into more general assertions about the stereo system as a whole, (2) additions-statements about attributes never mentioned in the stereo description, (3) evaluations - opinion statements about the system, specific components, or the advertisement itself, and (4) confusions - statements which incorrectly attributed correct information about one component to a different component. The mean numbers of generalizations, additions, evaluations, and confusions per subject were 1.04, .61, .60, and .37, respectively. There was neither a main effect of knowledge level nor a knowledge level x error type interaction (both Fs<1). There was, however, a main effect of error type (p<.01) attributable to the fact that subjects made more generalizations than any other type of error.

Since generalizations were the most common type of error and since some of the evaluation errors could be considered generalizations of a sort (e.g., general statements about the overall quality of the stereo system), these results indicate that people infer, at times, to higher, more comprehensive levels of abstraction when presented with individual, basic level facts. This would seem to be a useful and perhaps necessary activity when the information needed to be comprehended and assimilated is both copious and complex, as it was in the present study.

Stimulus Evaluations

Subjects were required to evaluate the advertising copy and the product on a variety of dimensions prior to the surprise recall test. Each dimension was rated on a 7-point scale ranging from very low to very high levels of that dimension. As Table 2 indicates, several differences were found between the different knowledge groups.



With respect to the advertising copy, the most important finding regards comprehensibility. In keeping with previous research (see Alba & Hasher in press, for a review), high-knowledge people not only recalled more information, but they also found information related to their area of expertise to be more comprehensible than did people of low expertise. As will be discussed below, the relationship between recall and comprehension is more than correlational. Also, not surprisingly, high-knowledge people found the information to be more useful. One would expect a positive relationship between comprehensibility and perceived usefulness. Interestingly, high-knowledge people found the information to be marginally less believable. Since, however, the ad described a hypothetical product and was purposely constructed to portray a stereo which was extraordinarily good on many dimensions, it seems reasonable that the people with more stereo experience would be more skeptical of such extreme claims. There were no differences between knowledge groups on their perceptions of the amount of information presented or on the need for illustrations. Both groups thought that the information was slightly excessive and that illustrations would have been useful.

With respect to the stereo itself, only one difference was found: for no easily determined reason, low-knowledge subjects found the stereo's quality of construction to be higher than did high-knowledge subjects. Both groups perceived the product to be high in sound quality, technical sophistication, ease of operation., and expense. However. with the exception os ease of operation, values on these dimensions approached ceiling levels. Thus, while we can conclude that both groups thought highly of the product on these dimensions in an absolute sense, strong conclusions about the relative behavior of the two groups would not be valid.

A final question put to subjects was designed to evaluate the persuasiveness of the stimulus. Subjects were asked whether or not they would purchase the stereo given the opportunity, need, and resources to do so. Results show that 71% of the high and 72% of the low- knowledge subjects responded affirmatively. Thus, the advertisement seemed to be effective, yet did not affect the groups differently. This result, among with the results of the product related questions, shows that low-knowledge subjects did extract much information from the message. It is impossible at this point to determine the relative contributions of the simple and complex pieces of information to their final evaluations. It seems more than possible that consumers would form impressions about product attributes from information they do not fully comprehend because of biases operative during the processing of promotional information.


Prior Knowledge and Memory

To this point, all discussion of the effects of expertise on memory has involved free recall as the dependent variable. There is now, however. a growing body of literature indicating that these effects on memory can be eliminated on greatly attenuated when alternative retrieval measures are employed. Table 3 shows the results of an experiment by Alba, Alexander, Hasher, & Caniglia (1981) which are somewhat representative.



Using the classic stimuli of Bransford and Johnson (1972), it was again demonstrated that people whose prior knowledge is activated at the time of learning recall nearly twice as many ideas as do people whose knowledge is not activated. This difference in knowledge activation is also reflected in comprehension ratings. (On a 7-point scale, "1" signifies very low and "7" very high comprehensibility). However, when given a recognition test and asked to rate statements on a 7-Point scale (with "1" denoting "certainly new" and "7", "certainly old"), no differences were found across groups in ability to recognize formerly presented ideas or reject never presented ones. These and similar data are important because they demonstrate that knowledge exerts its influence at the point of retrieval rather than encoding. That is, people "learn" new information even when confused about its meaning. Memory appears to be considerably more accurate than one would suppose based only on the results of free recall experiments.

Explanations of the knowledge effect

Clearly, expertise has profound impact on both recall and comprehension. Yet, performance on recognition tests appears to be independent of prior knowledge. Three possible explanations of these findings are offered.

Hierarchical Cueing. If information is processed with particular goals or organizing concepts in mind, relevant individual items of information may then be "attached" to those goals or concepts during encoding. The goals or concepts may then serve as hierarchical cues which can be used to retrieve individual idea units at the time of recall. With stereos, for example, the names of individual components or particular dimensions such as sound quality, may serve as the higher order recall cues. In cases in which people possess no schematic knowledge, information will be difficult to understand, no particular processing strategies or organizing principles will exist, no hierarchical memory structures will obtain, and recall will suffer. Note that recognition should not be affected to any large extent since cues of the type discussed here will be most helpful in free recall situations.

Coherence. It has been demonstrated that when individual sentences or phrases possess a common referent, links between them are formed and recall is enhanced (see, e.g., Haviland & Clark 1974). Retrieval of one idea cues related ones. Often, messages are constructed in such a way that the referents of statements are not explicitly stated but are clear to anyone with the appropriate background knowledge. People without the requisite knowledge perceive the message to be a series of disconnected statements (cf. Bransford & Johnson 1972). Because the ideas are disconnected, recall is more difficult. Again, however, recognition should remain unaffected since inter-idea cueing is proportionately more important in free recall tasks.

Coherence will also obtain when two items are causally linked, i.e., when the first event is a necessary precondition for or helps explain the second. Recall is also enhanced when this variety of coherence is estabLished (Black & Bern 1981). In situations in which only high-knowledge individuals are able to recognize the causal link between two isolated ideas, only they should derive the benefit during free recall. As with referential coherence, recognition should remain unaffected.

Scripts/Frames. When people posses strong expectations about the structure of incoming information due to their well developed knowledge about that product class, they may not need to relearn the information. All they need do is note that their expectations were confirmed or denied. During recall, they will be able to consult their established knowledge bases (sometimes called scripts or frames) and use them as massive retrieval cues or checklists, examining each node and determining whether or not the recently encountered information contained ideas corresponding to those nodes. Low-knowledge people possess no such knowledge structures and therefore will not benefit from their cueing effects. Once more, recognition should be less affected since cueing is most important during free recall.


Returning to the distinction discussed at the outset between memory and stimulus-based processing, it seems quite apparent that in situations in which choice is determined by the quantity and quality of retrieved information, high-knowledge consumers will be at a distinct advantage. High-knowledge individuals will recall more total information and especially more sophisticated and perhaps more important information.

At this point, the findings concerning recognition are not entirely clear. While it is true that the data conclusively demonstrate that recall cannot be equated with memory, it is also true that memory cannot be equated with comprehension. People of low knowledge will be capable of recognizing a previously experienced advertising message, yet may remain incapable of comprehending it. Thus, the benefits of high recognition ability with regard to decision-making may be minimal.

There seem to be at least two potential benefits from being able to retain, if not recall, information which is not entirely comprehensible. First, since the information is in memory, during a second exposure to a stimulus more capacity may be devoted to comprehension and less to encoding. This may lead to comprehension increments over a longer period of time. Second, comparisons may be made across brands when only a single brand's data is in hand. Information from the brand which is present may serve as an extensive retrieval cue (cf. scripts) for otherwise inaccessible information concerning other brands. In these situations, the normally large advantage enjoyed by high-knowledge individuals in terms of amount of information available to them may dissipate. How this might affect the quality of the ultimate choice is still unclear and must await further research directed specifically at the issue of decision-making.


Alba, Joseph W., Alexander, Susan G., Hasher, Lynn, & Caniglia, Karen (1981), "The Role of Context in the Encoding of Information," Journal of Experimental Psychology: Human Learning and Memory, 7, 283-292.

Alba, Joseph W. & Hasher, Lynn (in press), "Is Memory Schematic?", Psychological Bulletin.

Ausubel, David P. (1968), Educational Psychology: A Cognitive View, New York: Holt, Rinehart, & Winston.

Bartlett, Frederick C. (1932), Remembering: A Study in Experimental and Social Psychology, Cambridge, England: Cambridge University Press.

Black, John B. and Bern, Hyman (1981), "Causal Coherence ar.d Memory for Events in Narratives," Journal of Verbal Learning and Verbal Behavior, 20, 267-275.

Bransford, John D. & Johnson, Marcia R. (1972), "Contextual Prerequisites for Understanding: Some Investigations of Comprehension and Recall," Journal of Verbal Learning and Verbal Behavior, 11, 717-726.

Chiesi, Harry L., Spilich. George J., & Voss, James F. (1979), "Acquisition of Domain-Related Information in Relation to High and Low Domain Knowledge," Journal of Verbal Learning and Verbal Behavior, 18, 257-274.

Haviland, Susan E. & Clark, Herbert H. (1974), "What's New? Acquiring New Information as a Process in Comprehension," Journal of Verbal Learning and Verbal Behavior, 13, 512-521.

Lynch, John G. & Srull, Thomas K. (1982), "Memory and Attentional Factors in Consumer Choice: Concepts and Research Methods," Journal of Consumer Research, 9, 18-37.