When your recall is better for first and last items on a list but poor for Middle items is called as?

ABSTRACT - Primacy effects have been considered as a general phenomenon, but not as an effect that might vary amongst individuals, or in this case, consumers. This study investigates the moderating effect of the level of consumer knowledge on primacy effects in brand recall. The study also extends past research results by considering a situation where consumers are learning about brands in one product category, instead of many categories. The results indicate that low knowledge consumers are more susceptible to primacy effects than high knowledge consumers.

Citation:

Elizabeth Cowley (1999) ,"Primacy Effects: When First Learned Is Best Recalled", in E - European Advances in Consumer Research Volume 4, eds. Bernard Dubois, Tina M. Lowrey, and L. J. Shrum, Marc Vanhuele, Provo, UT : Association for Consumer Research, Pages: 155-160.

European Advances in Consumer Research Volume 4, 1999      Pages 155-160

PRIMACY EFFECTS: WHEN FIRST LEARNED IS BEST RECALLED

Elizabeth Cowley, University of New South Wales, Australia

[This research was funded by an Australian Research Council grant, and supported by the School of Marketing at University of New South Wales, Sydney, Australia.]

ABSTRACT -

Primacy effects have been considered as a general phenomenon, but not as an effect that might vary amongst individuals, or in this case, consumers. This study investigates the moderating effect of the level of consumer knowledge on primacy effects in brand recall. The study also extends past research results by considering a situation where consumers are learning about brands in one product category, instead of many categories. The results indicate that low knowledge consumers are more susceptible to primacy effects than high knowledge consumers.

1.0 INTRODUCTION

A robst finding in memory research is that people tend to remember the things they learned first more easily than they remember the things they learned in the middle or at the end of the learning episode. Although this phenomenon, called a primacy effect, has often been demonstrated in cognitive psychology labs, there are only a few investigations of the effects with marketing information (for example, Burke and Srull 1988; Pieters and Bijmolt 1997). In both of these studies, primacy effects are considered in the context of advertising blocks: the length of the block and the placement of an advertisement in that block in particular. Two issues emerging from these studies are investigated here.

First, the primacy effects found in these studies are treated as a general phenomena. The potential for variation in the magnitude of the effects across consumers is not considered. The most common explanation for the primacy effect is that an individual has more opportunity to rehearse the first few items on a list. Rehearsal is one of the possible strategies employed when encoding new information, but clearly not the only effective strategy. Imagery (Miller and Marks 1992) and elaboration during encoding (MeyersBLevy 1990: Mick 1992), as well as integration into existing knowledge structures (Kardes and Kalyanaram 1992), have also been demonstrated to encourage learning and memory. Integrating information requires that links be created between brands with similar features or attributes, or between brands useful in the same usage situation. It is asserted here that the rehearsal strategy is most likely to be employed when there is limited opportunity to use other learning strategies such as integration. Studies in cognitive psychology indicate that when the to-be-learned information does not encourage integration, the probability of observing primacy effects increases (some examples are Craik and Watkins, 1973; Rundus, Loftus and Atkinson, 1970; Sato, 1990). Some speculation, and some evidence in the consumer research literature suggests that higher knowledge consumers are more able than lower knowledge consumers to integrate information during encoding (Alba and Hutchinson, 1987; Mitchell and Dacin, 1996; Srull, 1983). The argument presented here posits that those unlikely to integrate information during encoding, low knowledge consumers (LKCs), are more susceptible to primacy effects.

Second, consumers do not exclusively learn information about brands from commercials, there are other learning situations. For instance, consumers look in retail environments for information, as well as reading articles in magazines such as Consumer Reports. These information sources result in quite a different type of learning situation because consumers are exposed to information about many brands in the same product category. Consumers are also often presented with verbal presentations of a list of specials in restaurants, and available draught beers in pubs or bars. If an order effect in recall exists, it would be in the best interest of the marketer to be aware of the effect and conditions under which it is likely to occur. Whether the order of the information presented to the consumer matters to their ability to recall the information later has not been considered in marketing literature. The objective of this study is to investigate the knowledge level of the consumer as a potential moderator of primacy effects.

2.0 BRAND RECALL: REHEARSAL, DISTINCTIVENESS, AND INTEGRATION

How will brand recall be affected when integration does not occur at encoding? There is no study reported in the consumer behaviour literature that limits the opportunity to integrate brand information at study, and then tests for an effect on brand recall. There have been studies in cognitive psychology, however, that intentionally use lists of items that do not lend themselves to integration of the to-be-learned information (some examples are Craik, 1970; Craik and Watkins, 1973; Runds, Loftus and Atkinson, 1970; Rundus and Atkinson, 1970). In these studies, subjects see a list of relatively meaningless information such as numbers (Darley and Glass, 1975), or unrelated words (Bruce and Papay, 1970; Craik and Watkins, 1973; Rundus, Loftus and Atkinson, 1970; Sato, 1990) and after a brief study session, try to recall items from the list. Typically, the items presented at the beginning and at the end of the list are better recalled than the items presented in the middle of the list. The effects have been found both when subjects are aware of an impending memory test (Rundus, Loftus and Atkinson, 1970), and when they are not (Craik and Watkins, 1973). Many explanations exist for the effect, but two of the most often cited are the rehearsal explanation and the von Restorff effect.

2.1 The Rehearsal Explanation (Craik, 1970; Rundus, 1971; Rundus and Atkinson, 1970).

According to this explanation, individuals rehearse information presented on a list while it is held in working memory. The more often a piece of information is rehearsed, the greater the probability the information will be transferred to long term memory. As working memory reaches capacity, new information entering working memory supplants the incumbent information. Because the first pieces of information on the list are rehearsed longer, they have the highest probability of transfer to long term memory, and therefore, a greater probability of recall.

The rehearsal explanation is perhaps the most common explanation for the primacy effect, although many compelling studies question its explanatory power. Of these studies, some limit the opportunity to rehearse with time constraints (Kim and Suk-Oh, 1979; Leshowitz and Hanzi, 1974; Titus, 1991; Wixted and McDowell, 1989), instructions (Bruce and Papay, 1970; Titus, 1991) or distraction (Murdock, 1965), but still find primacy effects.

2.2 The von Restorff Effect (von Restorff, 1933).

According to this explanation, distinct items are better recalled than other items. Since the first items in the list are the most distinct then they should be recalled with a higher probability than subsequent items. Titus and Robinson (1973) offer support of this explanation when a shift of the voice in auditory list presentation resulted in a primacy effect for the items immediately following. Shiffrin (1970) suggested that primacy effects are the result of both rehearsal and the von Restorff effect, however, Rundus (1971) presents evidence that the von Restorff effect is due entirely to an increase in the amount of rehearsal for distinctive items (Rundus, 1971, Experiment 2).

Both of these explanations rely on the assumption that the individual processing the information is simply trying to encode the information without any real attempt to understand, integrate or group the items on the list. In this case, there should be a relationship between the absence of integration and the primacy effect. Tulving and Madigan (1970) suggest in their review of memory and verbal learning, that presentation order effects are related to instances where 'the barest minimum by way of learning’ occurs (page 454).

If the same logic is applied to consumer learning, consumers that process brand information in a non-integrative manner, are more likely to demonstrate primacy effects in recall than are consumers that integrate brand information present during the same learning episode. In fact, evidence in both person perception (Takahashi, 1977) and consumer behaviour studies (Kardes and Kayanaram, 1992) demonstrate that presenting items simultaneously, which effectively frees cognitive resources for integration (Craik and Lockhart 1972), attenuates the probability of primacy effects in recall. If the opportunity for integrating the information improves, then the probability of observing primacy effects during recall decreases. Logically then, when consumers are integrating new information into existing knowledge structures, they are muc less likely to demonstrate primacy effects in recall.

3.0 INTEGRATION AND THE LEVEL OF PRODUCT KNOWLEDGE

High knowledge consumers (HKCs) are often able to recall more brands than LKCs (Alba and Chattopadhyay, 1985; Bettman and Park, 1980; Hutchinson, 1983). The advantage of the HKC in recalling previously presented brand information is exaggerated when the presentation format is random instead of blocked (Srull, 1983). Presumably, low knowledge individuals find it difficult to organise the product information at learning, while high knowledge individuals are able to use schemas stored in memory to guide the organisation of new information (Chiesi, Spilich and Voss, 1979; Voss, Vesonder and Spilich, 1980). When learning new information, consumers that rate themselves as HKCSs, and have knowledge of the characteristics of a product, are more likely to draw comparisons between brands (Mitchell and Dacin, 1996). Mitchell and Dacin (1996) also found that the number of links or relations between brands and attributes increases with product knowledge, and that consumers with more knowledge are more likely to represent the level of the attribute, instead of just the presence of the attribute. The additional information about the quantity of an attribute presents further opportunity for integration, as relations between brands may also be based on the level of an attribute instead of just its presence. Taken together, HKCs should be more likely to integrate information during encoding, and therefore, less likely to evidence primacy effects in recall.

LKCs are less likely to have integrated knowledge structures. In this case, the structures are not conducive to integration during encoding, particularly integration based on attributes or usage situations. Instead, the LKC may organise their knowledge around the learning episode, and the brands that are strongly related to the episode are most likely to be recalled. The LKC, therefore, is more likely to show primacy effects in recall.

4.0 EMPIRICAL WORK

4.1 Study Design

In this study, subjects see two lists of brand attribute statements (first, second). There is a break in between, this is similar to a shopping trip where more than one retail outlet is visited with a short lag between learning episodes. The brands can be integrated over usage situation because three brands are associated to each usage situation. After a brief study session and a filler task, subjects try to recall as many of the brands as possible. Primacy effects will occur in the recall performance of consumers (subjects) that do not integrate the information. Primacy effects will not be observed if the consumer is integrating information during encoding.

The design includes 2 levels of consumer knowledge as a measured between subject factor. Subjects are randomly assigned to one of two versions of the lists that counterbalance the order of presentation of the brand names. The measure of recall performance is the probability of recalling brands from three positions on the list (the beginning, middle or end).

4.1.1 Subjects.

4.2 Stimulus Design

4.2.1 The Product Category.

4.2.2 The Brands.

The brand names are five letter nonsense words. The brand names were pretested to check for:

1] significant differences in the probability of recalling the different names,

2] associations between the brand names and existing camera brand names, and

3] associations between the brand names and any existing brand name.

Pretest OneBSimilarity to Existing Brand Names

Forty undergraduate students were shown each of twenty brand names with the following instructions:

i] list any characteristics associated with cameras that come to mind when you read each of the hypothetical brands,

ii] rate your familiarity with each of the hypothetical brand names,

iii] state whether the brand name reminds you of other existing brand names and,

iv] identify other similar and related brand names.

Two of the hypothetical names reminded subjects of existing camera brands. Three of the hypothetical brand names (including the two previous names) reminded subjects of brand names for other products. These brand names were not included in the main study. Many of the comments made during the pretest were speculations as to the country of manufacture or the language the word might be taken from. Although there was some consensus as to the country itself, there was no systematic consensus as to the meaning of the 'word’ (no more than one subject identified a particular meaning for the word).

Pretest TwoBMemorability Pretest

Twenty students were shown 16 brand names and asked to rate whether the camera sounded as though it might be 'easy to use’ on a scale anchored with 'easy to use’ and 'difficult to use’. Ten minutes later they were asked to recall as many of the brands as possible. The position on the list (beginning, middle or end) is counterbalanced between four lists. When position on the list is a covariate, there are no significant differences in the probability of recall between the brands.

4.2.3 The Attributes.

4.2.4 The Lists.

Target Statements.

Each brand, and each attribute was mentioned once during the study session. The lists were constructed such that integration was possible, but not easy. The following rules were adhered to during list construction:

1] brands associated to the same usage situation were not seen sequentially,

2] each brand was seen only once,

3] each attribute was seen only once, and

4] each usage situation appeared on both lists.

There was an opportunity for integration over usage situations however, as three of the attributes related to each of four usage situations.

Filler and Buffer Statements.

Generally, the results of studies investigating the influence of presentation order reveal a fairly smooth function declining from the relatively high probability of retrieval for the first few items. The filler statements served to segregate the lists such that the target items captured the beginning, middle and end of the serial position curve.

The order in which brands were presented was counterbalanced between the two sets of lists. Brands that were at the beginning of the list one, version one, were either in the middle or at the end of the list two, version two. An ANOVA reveals that the list version was not a significant factor in brand retrieval (F(1,47)=0.09, p>.10).

5.0 PROCEDURE

Subjects were randomly assigned to one of the two versions of the lists. Subjects were informed that they would be participating in a number of short tasks. The first would be to read brand information about cameras. They were warned that the brands were hypothetical, and that they should not be concerned that the brand names would be unfamiliar to them. The subjects saw the brand attribute statements projected on a screen, and heard the study administrator read the statements aloud. Each of the statements was seen individually. Subjects were encouraged to ask for more time to read the statements if the pacing of the presentation was too brief. No subjects complained of the pace. This presentation format was implemented to encourage the subject to pay attention to the information.

The remainder of the session is divided into three parts: two study sessions and one test session.

5.1 Study Session One.

Subjects saw the first list of ten statements about cameras projected on a screen with an overhead projector, while the administrator read the statements aloud. Each statement was visible for approximately ten seconds. Each statement was seen sequentially. Following the study session, subjects read unrelated material for five minutes.

5.2 Study Session Two.

Subjects saw the second list of ten statements about cameras projected on a screen with an overhead projector, while the administrator read the statements aloud. Each of the statements was seen sequentially. A second five minutes was spent reading unrelated material.

5.3 Test Session.

Subjects recalled as many of the brands as possible from the two lists. Following the free recall task, subjects completed a knowledge questionnaire, were debriefed and thanked for their participation.

5.4 Measures

5.4.1 Dependent Measure.

5.4.2 Independent Measures.

Objective Measures.

Subjective Measures.

A median split on the knowledge measure results in 14 HKCs and 14 LKCs. The coefficient alpha for the measure of knowledge is .89. All the correlations between the individual objective and subjective measures are significant at p<.01.

TABLE 1

ANOVA RESULTS

6.0 ANALYSIS AND RESULTS

The hypothesis for the study asserts that the presentation order of information at encoding influences the retrieval behaviour of LKCs more than HKCs. Specifically, LKCs will demonstrate primacy effects in recall.

The hypothesis is tested with a mixed ANOVA on retrieval probabilities, with knowledge level, and the list (first, second) as between subject factors and list placement (beginning, middle, end) as a within subject factor. As expected, the results of the ANOVA reveal non-significant results for the between subject factors of knowledge level (F(1, 156)=0.26, p>.10), list (F(1, 156)=0.26, p>.10), and the interaction of the two (F(1, 156)=0.06, p>.10). The probability of retrieval of brands was the same regardless of knowledge level or list. The interaction of importance is the knowledge level by list placement (or order) interaction which tests whether the order of presentation effects brand recall differently for high and low knowledge consumers. The ANOVA results indicate that this interaction is significant (F(2, 165)=4.4, p<.05). See the ANOVA Table (Table 1) for results.

To facilitate the interpretation of this interaction, separate ANOVAs were run for each knowledge level. The ANOVA for LKCs reveals that the list factor is non-significant (F(1, 78)=0.27, p>0.10), and the list by list placement (or order) interaction is non-significant (F(1, 78)=0.98, p>.10), but that list placement factor (or order) is significant (F(1, 78)=6.68, p<.005). In other words, brand recall was the same for both lists and the order of presentation effected brand recall in the same manner for both lists, primacy effects were found for both lists. The retrieval probabilities for LKCs for the items at the beginning of the list, are significantly greatr than the probabilities for items in the middle (t=-2.65, p<.01) or at the end of the list (t=-2.65, p<.01). Retrieval probabilities for HKCs are unchanged regardless of the list factor (F(1, 78)=0.04, p>.10), list placement or order (F(1,78)=0.13, p>.10), or the interaction of the two (F(1, 78)=2.14, p>.10). See Table 2 for the results.

The primacy effect is found in the recall of LKCs for each list. The list placement factor is measured for each list, and does not interact with the list factor. The data observed here suggests that instead of 'things I learned in the lab today’ or 'things I learned shopping today’, the learning episode is limited to 'things I learned on one list’ or 'things I learned in one store’.

7.0 DISCUSSION

The study is simple, but important. First, the results indicate that it is LKCs that are more likely to demonstrate primacy effects. The logic employed here suggests that the presence of primacy effects indicates non-integrative processing. However, the assertion can not be made that LKCs are not processing integratively unless processing is manipulated and tested with processing measures.

Murdock (1976) asserts that order effects should not be observed when the subject has the opportunity, and ability to associate on the basis of their content. If the information about brands was integrated during encoding, then the retrieval probabilities would vary in a manner uncorrelated to presentation order. This is the pattern of results for the HKC. Perhaps HKCs use knowledge structures from memory to organise the information and to aid in 'seeing’ the relations between items.

8.0 CONTRIBUTIONS AND LIMITATIONS

The contribution of these results is twofold. For marketers, the order in which less knowledgeable consumers are exposed to brand names affects their ability to remember the brands later. This has implications for any learning episode where information about more than one brand is communicated. Examples of this situation occur in shopping environments, where consumers may listen to a sales pitch, or be exposed to point-of-sale material, or when learning information about brands in a format where many brands are compared. It is important to keep in mind that LKCs may recall the first things they are told by a salesperson better than subsequent pieces of information.

The second contribution is toward an understanding of the way in which consumers process information during learning. Of course, these are only initial indications of processing styles, but the results are consistent with the idea that HKCs integrate information during encoding, and that LKCs do not integrate information. Instead the LKC learns the information as though it was a list of unrelated information. A future study could manipulate the learning session in order to test for primacy effects when: 1] integration is impossibleBperhaps by distracting the consumer and reducing the available resources, and 2] integration is encouraged by the format of the presentation or the task at study. Primacy effects should be apparent for HKCs in the first instance, and primacy effects should be attenuated for LKCs in the second instance.

There are important limitations to the study. First, the sample size is small. Although the results have been replicated in a study not reported here. Second, the study is performed in a laboratory situation without the creation of a retail environment, and all of the relevant variables found there. A valuable extension to the research would be to investigate the likelihood of detecting primacy effects in the field. It may also be interesting to know whether primacy effects would be found over different forms of media, in other words,whether the media bounds the learning episode. Another interesting area of research might be to investigate the interaction of involvement and knowledge on the probability of primacy effects in recall.

TABLE 2

RETRIEVAL PROBABILITIES

REFERENCES

Alba J. W. and Hutchinson, J. W. (1987), Dimensions of Consumer Expertise, Journal of Consumer Research, 13: 411-454.

Bettman, J. R. and Park, C. W. (1980), Effects of Prior Knowledge and Experience and the Phase of the Choice Process on Consumer Decision Processes: A Protocol Analysis, Journal of Consumer Research, 7: 234-248.

Bruce, D. and Papay, J. P. (1970), Primacy Effect in Single-Trial Free Recall, Journal of Verbal Learning and Verbal Behavior, 9, 473-486.

Burke, R. R. and Srull, T. K. (1988), Competitive Interference and Consumer Memory for Advertising, Journal of Consumer Research, 15: 55-68.

Chiesi, H. L., Spilich, G. J. and Voss, J. F. (1979), Acquisition of Domain-Related Information in Relation to High and Low Domain Knowledge, Journal of Verbal Learning and Verbal Behavior, 18, 257-273.

Craik, F. I. M. (1970), The Fate of Primary Memory Items in Free Recall, Journal of Verbal Learning and Verbal Behavior, 9, 143-148.

Craik, F. I. M. and Lockhart, R. S. (1972), Levels of Processing: A Framework for Memory Research, Journal of Verbal Learning and Verbal Behavior, 11, 671-684.

Craik F. I. M and Watkins, M. J. (1973), The Role of Rehearsal in Short-Term Memory, Journal of Verbal Learning and Verbal Behavior, 12, 599-607.

Darley, C. F. and Glass, A. L. (1975), Effects of Rehearsal and Serial List Position on Recall, Journal of Experimental Psychology: Learning, Memory and Cognition, 104 (4), 453-458.

Hutchinson, J. W. (1983), Expertise and the Structure of Free Recall, in Richard Bagozzi and Alice M. Tybout, Advances in Consumer Research, Vol. 10. Ann Arbor, MI: Association for Consumer Research.

Kardes, F. R. and Kalyanaram, G. (1992), Order-of-entry Effects on Consumer Memory and Judgment: An Information Integration Perspective, Journal of Marketing Research, 293, 343-357.

Kim, J. and Suk-Oh. M. (1979), Psychological Mechanisms which Determine Short-term Memory Capacity, Korean Journal of Psychology, 2(4), 200-216.

Leshowitz, B. and Hanzi, R. (1974), Serial Position Effects for Tonal Stimuli, Memory and Cognition, 2,112-116.

Meyers-Levy, J. (1990) Elaborating on elaboration: The distinction between relational and item-specific elaboration. Journal of Consumer Research, 18, 358-367.

Mick, D. G. (1992), Levels of Subjective Comprehension in Advertising Processing and Their Relations to Ad Perceptions, Attitudes, and Memory, Journal of Consumer Research, 18(4): 411-424.

Miller, D. W. and Marks, L. J. (1992), Mental imagery and sound effects in radio commercials, Journal of Advertising, 21(4): 83-93.

Mitchell, A. A., & Dacin, P. F. (1996) The assessment of alternative measures of consumer expertise. Journal of Consumer Research, 23 (3), 219-240.

Murdock, B. B. (1965), Effects of Subsidiary Task on Short-Term Memory, British Journal of Psychology, 56, 413-419.

Murdock, B. B. (1976), Item and Order Information in Short-Term Serial Memory, Journal of Experimental Psychology; General, Vol. 105 (2), 191-216.

Murdock, B. B. (1995), Primacy and Recency in the Chunking Model, in Charles A. Weaver III, Suzanne Mannes and harles R. Fletcher (eds.), Discourse Comprehension: Essays in Honor of Walter Kintsch, 49-63.

Pieters, R. G. M. and Bijmolt, T. H. A. (1997), Consumer memory for Television Advertising: A Field Study of Duration, Serial Position, and Competitive Effects, Journal of Consumer Research, 23: 362-372.

Rundus, D. (1971), Analysis of Rehearsal Processes in Free Recall, Journal of Experimental Psychology, 89, 63-77.

Rundus, D. and Atkinson, R. C. (1970), Rehearsal processes in free recall: A Procedure for direct observation, Journal of Verbal Learning and Verbal Behavior, 9, 99-105.

Rundus D., Loftus, G. R. and Atkinson, R. C. (1970), Immediate Free Recall and Three-Week Delayed Recognition, Journal of Verbal Learning and Verbal Behavior, 9, 684-688.

Sato, K. (1990), Recency Effects and Temporal Distinctiveness: Dissociation Between Free Recall and Memory for Position/Order, Perceptual and Motor Skills, 71, 1339-1351.

Shiffrin, R. M. (1970), Memory Search, in D.A. Norman (Ed), Models of Human Memory. New York: Academic Press.

Srull, T. K. (1983), The Role of Prior Knowledge in the Acquisition, Retention, and Use of New Information, in Richard Bagozzi and Alice M. Tybout (eds.), Advances in Consumer Research, Vol. 10. Ann Arbor, MI: Association for Consumer Research, 572-576.

Takahashi, S. (1977), Context Effects as a Function of Serial Position of Test Objective in Person Description, Japanese-Psychological-Research, 19(3), 114-120.

Titus, T. G. (1991), Effects of Rehearsal Instructions on the Primacy Effect in Free Recall, Psychological Reports, 68, 1371-1377.

Titus, T. G. and Robinson, J. A.(1973), Pseudo-primacy Effects in Free Recall, Perceptual and Motor Skills, 37, 891-899.

Tulving, E. and Madigan, S. A. (1970), Memory and Verbal Learning, Annual Review of Psychology, Vol. 21, 437-484.

Voss, J. F., Vesonder, G. T. and Spilich, G. J. (1980), Text Generation and Recall by High-Knowledge and Low-Knowledge Individuals, Journal of Verbal Learning and Verbal Behavior, 19: 651-667.

von Restorff, H. (1933), Uber die Wirkung Von Bereichsbildungen in Spurenfeld, Psychogisch Forschung, 18, 299-342.

Wixted, J. T. and McDowell, J. J. (1989), Contributions to the Functional Analysis of Single-Trial Free Recall, Journal of Experimental Psychology: Learning, Memory and Cognition, 15, 685-697.

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

What is it called when you recall the first and last items on a list better than the middle of the list?

What is it called when you remember the first and last things on a list?

When you have better recall for items at the beginning or end of a list it called Group of answer choices?

Why do we tend to remember the first and last items in a list?