Interference effects in multi-attribute advertising

INTERFERENCE EFFECTS IN MULTI-ATTRIBUTE ADVERTISING

By

HAYDEN NOEL

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FORIDA

2002

This work is dedicated to the memory of my deceased father, Peter A. Noel, and my
grandparents Victor and Thelma Noel also deceased. I love you all.

ACKNOWLEDGMENTS

This dissertation would not have been completed without the input of many

individuals. First and foremost, I am deeply indebted to the chair of my dissertation

committee, Chris Janiszewski. He has been an advisor, a role model, and a friend during

this arduous process. His guidance helped equip me with the research tools that I need to

successfully pursue a career in experimental research. Additionally, I am thankful for the

insightful comments and suggestions of the members of my dissertation committee Joe

Alba, Ira Fischler, Alan Sawyer, and Steve Shugan. I am especially grateful to Alan

Sawyer for his guidance and support beyond that of a dissertation committee member. I

would also like to thank Joel Cohen, Richard Lutz, and Barton Weitz for their advice and

encouragement during my doctoral studies. In addition to the mentoring provided by

faculty, I could not have made it through the program without the help of my colleagues;

to this end I would like to thank Americus Reed, Amitav Chakravarti, Marcus Da Cunha,

and Eduardo Andrade for providing me with feedback and advice when I needed it. I also

thank Tony Gonzalez, Ricky Lovell, Coswell De Peza, Marlon Morris, Stephen Eugene

and Fulton Wilson for their unflagging support. Last, but by no means least, I would like

to thank my family in the United States and in Trinidad especially my mother Mona

Noel, and my sister Joanne Noel. Without their love, support and prayers I would not

have been able to complete my studies.

TABLE OF CONTENTS

Page

ACKNOW LEDGEM ENTS ...................................... ... ............ iii

ABSTRA CT ....................... ................................................... vi

CHAPTER

1 INTRODUCTION .................... ... ............ 1

2 THEORETICAL BACKGROUND ............................................. 4

A advertising Repetition ............................................. ............ 4
Multi-brand / Multi-attribute Advertising ................... ... ........... 8
Summ ary and Hypotheses ......................................... ............. 12

3 EXPERIMENT 1: THE INFLUENCE OF GROUPING ...................... 14

Stim uli and Procedure ............................................... ........... 15
Predictions ................... .... .... .. 17
R results ............................................................... ............ 20
Discussion ............. ........... .......... .. ....... .... .. 21

4 EXPERIMENT 2: THE EFFECT OF WITHIN-CATEGORY
CO M PETITIO N ................................. ............ ......... .. ........... 23

Stim uli and Procedure ........................... ....... ................. ...... 24
Predictions ........ ......................... ............. .... .... 24
R results ............................................................... ............ 26
Discussion ............ ............ ............ ........ .... .. 27

5 EXPERIMENT 3: SINGLE ATTRIBUTE VS. MULTIPLE
ATTRIBUTES ......... .................................... 29

Stim uli and Procedure ............... ........ .............................. 29
Predictions ........................................................... ........... 31
Results .......................................................................... 32
D discussion ....................................... .................. ............. 33

6 EXPERIMENT 4: THE EFFECT OF RELATED ATTRIBUTES .......... 35

Stimuli and Procedure ......... .............................................. 35
Predictions ........... .............. ........ .... .. 37
R results ................................................................ .......... 38
D discussion .......................................................... ........... 39

7 GENERAL DISCUSSION .............. ......................... ............ 40

8 LIMITATIONS AND FUTURE RESEARCH ............................ 42

Limitations ............................... ..... ...... ................ 42
Future R research ........... ..... ................. .... .. ............ 43

APPENDIX

A STIM ULI AND DESIGN ......................................... .............. 45

B PREDICTIONS AND RESULTS ..... .......................... 49

REFERENCES ................................. 57

BIOGRAPHICAL SKETCH .............. ...................................... 61

v

Abstract of Dissertation Presented to the Graduate School of the University of Florida in
Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

INTERFERENCE EFFECTS IN MULTI-ATTRIBUTE ADVERTISING

By

Hayden Noel

December 2002

Chairman: Christopher A. Janiszewski
Major Department: Marketing

When people view two different ads for the same brand, the first ad interferes

with memory for the product-attribute information in the second ad (i.e., proactive

interference); and the second ad interferes with memory for the product-attribute

information in the first ad (i.e., retroactive interference). The implication is that multi-

attribute brand advertising reduces consumer memory for known attributes with each

additional attribute advertised. One potential solution for this memory interference

problem is to present ads for a brand's multiple attributes close together in time.

Although this recommendation is inconsistent with anecdotal advertising advice to limit

an ad's content to a single benefit and with empirical findings suggest that dispersed

presentations are more effective at enhancing memory access, the benefits of grouping

brand-attribute presentations appear to be rather robust. Implications for ad execution and

ad flighting are discussed.

CHAPTER 1
INTRODUCTION

Advertising repetition can be used to achieve two related goals. First, ad repetition

can increase the likelihood that a brand name is remembered (Craig, Sternthal, and

Leavitt 1976; Pechmann and Stewart 1988; Ray and Sawyer 1971; Unnava and Burnkrant

1991). Second, ad repetition can strengthen an association between a brand name and a

product benefit (Burke and Srull 1988). The greater the memory for a brand name or its

product benefits, the greater the likelihood that the brand will enter into a consumer's

consideration set (Hauser and Wernerfelt 1990; Hutchinson, Raman, and Mantrala 1994).

Changes in consumer consideration set inclusion are directly related to changes in market

share.

Although one might assume that the benefits of advertising repetition are

pervasive, evidence suggests that the benefits are limited to a very specific set of

circumstances. For example, Burke and Srull (1988) found that when people view two

different ads for the same brand, the first ad interferes with memory for the product-

attribute information in the second ad (i.e., proactive interference) and the second ad

interferes with memory for the product-attribute information in the first ad (i.e.,

retroactive interference). In a later study, Burke and Srull (1988) also found that

competing ads from the same product category create memory interference and that

repetition of the target ad is only effective when there is a single competing ad (also see

Keller 1987). Thus, existing empirical evidence suggests that advertising is less effective

if a brand advertises on more than one attribute or if a brand operates in a product

category where more than one brand advertises. In other words, advertising is fairly

ineffective for most brands in most product categories since most brands advertise more

than one attribute and most brands have competitors that also advertise.

Recommendations about how to solve the memory interference problem inherent

in a competitive advertising environment have been limited. The best advice for limiting

memory interference has been offered by Keller (1987) who argues that advertisers must

find unique cues (e.g., spokespeople) to associate with their brand attributes. The unique

cue can form an independent association with product attributes or can combine with the

brand name to provide two cues for recall of the attribute information. Keller, Heckler,

and Houston (1998) argue that suggestive brand names (e.g., PicturePerfect televisions)

can also serve as unique cues, but that the cue is only effective for semantically related

benefits. Unfortunately, suggestive brand names also hinder recall of subsequently

advertised, semantically unrelated benefits.

In this paper, I argue that there are situations in which repetition can be used to

enhance recall in multi-attribute advertising or multi-competitor advertising contexts. I

begin with a review of evidence related to brand name and product-attribute recall. Then

I discuss explanations of memory inhibition and facilitation in the context of repeated

exposure to material. I consider three different theories that could explain the memory

processes involved in learning repeatedly presented material encoding variability

theory, reconstruction theory, and study-phase retrieval. Then through a series of four

experiments, I examine competing predictions from these theories to determine which

theory or theories have the most predictive power in a given context. My goal is to show

3

that brands can use multiple ads to advertise multiple attributes without incurring a

memory interference penalty and that ad repetition strategies exist that are more or less

effective in competitive advertising environments.

CHAPTER 2
THEORETICAL BACKGROUND

Advertising Repetition

Studies examining the influence of advertising repetition on memory can be

categorized into two separate, but related advertising goals. First, brand managers are

interested in increasing the accessibility of brand names in consumers' memory. Second,

brand managers are interested in increasing the strength of the associations between

brand names and product attributes or benefits.

Brand Recall

There is considerable evidence that ad repetition increases the recall of brand

names (Craig, Stemthal, and Leavitt 1976; Singh and Rothschild 1983; Singh,

Rothschild, and Churchill 1989; Singh et al. 1994; Unnava and Burnkrant 1991). Craig et

al. (1976) find that increasing the number of ad repetitions increases the recall of the

brand names mentioned in the ads, but that the effect is most robust with long delays

between ad exposures and test. Singh and Rothschild (1983) show that increasing the

number of television commercial exposures from one to two to four increases recognition

of brand names, product claims, and product packages. Singh et al. (1988) showed that

increasing the number of television commercial exposures from one to two increases the

recall and recognition of the product classes and brand names mentioned in the

commercials. In general, ad repetition enhances memory for information contained in the

ad.

5

Repeated exposure to advertising may increase memory for brand names because

each ad exposure involves a different encoding context, hence each exposure creates

additional retrieval cues (Singh et al. 1994; Unnava and Burnkrant 1991). The encoding

variability hypothesis states that the more different the first presentation (PI) is from a

second presentation (P2) of a stimulus, the more paths there are to retrieval at test

(Glenberg 1976; Melton 1970). For example, Unnava and Bumkrant (1991) had people

view a shampoo ad with two different executions or identical executions. People had

higher unaided and aided recall of the brand name when the ad executions were different

than when they were the same. Unnava and Burnkrant (1991) attributed the increased

recall to the variable encoding associated with the different meanings in the two ad

executions (e.g., the picture and tag lines accompanying the brand name varied in each

execution of the ad).

Similarly, Singh et al. (1994) exposed people to two presentations of an identical

television commercial with a short lag (e.g., one intervening commercial) or a long lag

(e.g., four intervening commercials). They found increased recall of the commercial in

the long lag condition provided there was a delay between exposure and test. Singh et al.

(1994) argued that the increase in time between the two presentations resulted in a greater

difference in contexts at the time of encoding and that this increased the likelihood that

one of these contexts would be similar to the context at a delayed test. Finally, Unnava

and Sirdeshmukh (1994) developed two approaches to counter the detrimental effects of

competitive advertising based on the encoding variability hypothesis. They found that

competitive advertising had less influence on people experiencing varied ad executions

than on people experiencing constant ad executions.

Support for the encoding variability theory in advertising studies has not been

accompanied by an equivalent level of support in verbal learning studies. Madigan (1969)

presented variable (fever-chill, snow-chill) or constant (fever-chill, fever-chill) cues with

to-be-remembered words and found a weaker spacing effect in the varied cue condition.

D'Agostino and De Remer (1973, Experiment 2) showed similar results with cues

embedded in constant or variable sentences. In addition, manipulations of context and

meaning in different stimulus presentations often show no improvement, and sometimes

show a decrease, in recall (Dempster 1987; Murdoch and Babick 1961; Postman and

Knecht 1983). Thus, there is evidence that repeated presentations of an ad can lead to

better recall for brand names, but there is some debate about whether the encoding

variability hypothesis best explains the observed memory facilitation effects.

Brand Associations

A second advertising research stream investigates how people use brand names as

cues to recall information about a brand. Brand managers advertise product features

because they want to position their brand relative to other brands and because they want

consumers to retrieve attribute information and to conclude that their brand is superior to

competitor brands. Encouraging consumers to remember more information about a brand

may lead the consumer to conclude that the brand has more benefits, or performs better

on key benefits. Similarly, if consumers cannot remember much about competing brands,

they may infer that the competing brands are inferior on the missing information

(Simmons and Lynch 1991).

Brand associations are sensitive to interference from additional presentations of

competing ads in the same product category and additional ads by the same brand that

advertise a different product attribute. For example, Keller (1987) studied competitive

interference using four brands (B,) positioned using four sets of unique product claims

(e.g., B1 -- A,, B2 -4 A2, B3 A3, B4 -4 A4). Subjects who saw ads for two of these

brands were more likely to remember the attributes paired with brand 1 than were

subjects who saw four ads. Thus, repetition from competing ads interfered with recall of

the attributes associated with the target brand. Burke and Srull (1988) argued that this

same interference process could occur when a target brand did additional advertising of a

second attribute. Burke and Srull (1988) replicated Keller by presenting two ads for two

competing brands advertising two competing attributes (BI -) A,, B2 -) A2) and two ads

for the same brand advertising two competing attributes (BI -4 A1, B1 -- A2). Both

conditions showed the same pattern of interference, thus a second ad for a brand

interfered with recall of attributes from the first adjust as much as a second ad for a

competing brand.

There have been a number of recommendations about how to increase the

effectiveness of brand name cues. First, consumers can be encouraged to elaborate a

brand name and the brand benefit, thus increasing the strength of the association between

the brand and the benefit (Pechmann and Stewart 1988). Elaboration can be encouraged

via repetition. For example, Rethans et al. (1986) found that repeated exposures to an ad

increased familiarity with both a new product and the ad, and that recall of the ad content

also increased with frequency of exposure. Elaboration can also be encouraged via

processing goal. For example, Burke and Srull (1988) had some of their subjects indicate

their interest in the ads they viewed while others indicated brand purchase likelihood.

Subjects in the brand purchase likelihood condition showed no retroactive interference

from additional competitive or same-brand ads, apparently because the processing goal

encouraged elaboration of the brand-attribute association at Time 1. Second, advertisers

can limit the number of advertised attributes to a single feature. As the number of

advertised attributes increases within or across ads, recall of any feature declines (Burke

and Srull 1988; Keller 1987). Third, brand names can be accompanied by a distinctive

cue (Keller 1987). A unique cue can join with the brand name to create a stronger

activation of the product attribute or a unique cue can form independent association to the

product attributes. For example, Keller (1987) showed that reproducing the photo and

headline from an ad on a product package increased the likelihood that the product

attribute would be recalled. Likewise, Keller et al. (1998) showed that a unique,

suggestive brand name increased the recall of semantically related product attributes.

Multi-Brand/Multi-Attribute Advertising

Keller (1987) and Burke and Srull (1988) found that advertising multiple

attributes across multiple ads results in lower recall of a specific attribute than if the

advertising had been limited to a single attribute, assuming that exposure is equal. As the

number of attributes advertised by a brand increases, recall of any one attribute should

decrease. Similarly, as the number of competing brands increases, recall of the attributes

advertised for any one brand should decrease. This should be so unless the advertiser can

encourage the person to elaborate an ad, a difficult prospect in a passive message delivery

environment.

Insights into how to encourage elaboration in multi-ad/multi-attribute advertising,

and hence remove the interference associated with advertising more than a single

attribute, may be found in the spacing literature. Research on the spacing effect

compares a massed condition, where repeated items are presented one after the other,

with a distributed condition, where repeated items are separated by intervening items,

tasks, or the passage of time (Hintzman 1974; Singh et al. 1994). In general, a distributed

presentation schedule results in better memory for the items than a massed presentation

schedule, a result commonly called the spacing effect. The spacing of stimulus

presentations has been shown to enhance memory for nonsense syllables (e.g.,

Ebbinghaus 1985/1913), words (e.g., Glenberg and Lehmann 1980), sentences (e.g.,

Rothkoph and Coke 1966), pictures (e.g., Hintzman and Rogers 1973), and faces (Comell

1980).

Explanations of the spacing effect can be divided into two categories: deficient

processing explanations and enhanced processing explanations. Deficient processing

explanations propose that the massed presentation schedule reduces processing of the

first presentation of a stimulus (PI) or the second presentation of a stimulus (P2).

Enhanced processing explanations propose that the distributed presentation schedule

provides an opportunity to engage in additional processing during P1 or P2. The two

enhanced processing explanations that specifically address the increased elaboration of

material at P 1 or P2 are the retrieval hypothesis and the reconstruction hypothesis.

Retrieval hypothesis. The retrieval hypothesis predicts that spacing effects are

directly related to the study-phase retrieval of Pl and that the primary function of P2 is to

serve as a cue for retrieval of PI (Braun and Rubin 1998; Thios and D'Agostino 1976).

For example, a person may be able to access more information at test if they bring an

earlier elaborated instance of the stimulus (P ) into consciousness, as opposed to

attempting to elaborate the current instance (P2) during study. Perceiving P1 should

become more difficult as the number of intervening items increases and this difficulty of

perception should facilitate memory access at a later time (Jacoby 1974; Thios and

D'Agostino 1976). It should be noted that perception is an involuntary operation that

occurs when an individual encounters a stimulus. This process enables the individual to

discriminate common or familiar objects from novel objects. For example, when a subject

sees a stimulus for the second time, she/he will automatically try to bring a previous

instance of that object into consciousness. This becomes more difficult if the previous

presentation has occurred farther back in time. When one is presented with a cue at time

of recall, it is the additional processing that occurs at P2 while attempting to bring an

item into the realm of consciousness that leads to enhanced memory, and not the success

of identifying a previously seen stimulus.

The retrieval hypothesis predicts that the key to an effective repetition strategy is

to encourage elaborated processing of an ad at P1 and not at P2 so that there would be

successful, but effortful, attempts at perceiving P1 when P2 is presented. For example,

this hypothesis predicts that the common practice of having an initial flight of 60-second

commercials at the start of an ad campaign followed by a subsequent flight of 15-second

reductions of these commercials may be more effective at building memory traces than

two flights of 60-second commercials. Fifteen-second commercials contain fewer

retrieval cues, and as a consequence, successful perception of information from the 60-

second commercial becomes more difficult.

Reconstruction (accessibility) hypothesis. The reconstruction hypothesis

predicts that the spacing effect depends on whether or not P can be reconstructed at P2.

Reconstructive memory theories assume there is no fixed memory structure, just a

recreation of subjective perception given activation in various parts of the brain caused

by external and internal cues (Braun and Rubin 1998). The hypothesis assumes that

people have the option of accessing P1 from short-term memory or, in an attempt to bring

PI into consciousness, involuntarily reconstructing P1 when they encounter P2 (Jacoby

1978). If an item is repeated while a previous representation is still accessible in short-

term memory, then there is no need to reconstruct it. If the repetition is delayed and the

PI presentation begins to fade, then it is necessary to go through a reconstruction process

that infers the missing portions of P Reconstructing a stimulus is thought to increase

retention because the reconstructed portions of the stimulus become likely to be

reconstructed again in the future, similar to a perceptual bias (Jacoby and Craik 1978;

Lockhart, Craik, and Jacoby 1976).

Like retrieval, reconstruction is an involuntary operation that helps the person

discriminate common or familiar objects from novel objects. For example, when a subject

sees a stimulus for the second time, she/he will automatically assess whether the object

has been seen before in an attempt to bring the object into consciousness. This

assessment depends on the reconstruction of the stimulus with an easier reconstruction

leading to the inference that the stimulus is familiar. The reconstruction becomes more

difficult as the previous presentation occurs farther back in time. When one is presented

with a cue at time of recall, it is not successful perception, but the act of attempting to

reconstruct the stimulus during the process of perceiving it that creates greater access to

that stimulus and that leads to enhanced memory at test. Thus, the reconstruction

hypothesis predicts that reconstruction of PI depends on elaboration of P2 when an

individual is trying to bring a stimulus into consciousness.

Summary and Hypotheses

The retrieval hypothesis and the reconstruction hypothesis both predict that

repetition improves memory when the pattern of repetition encourages additional

processing of P or P2 that, in turn, leads to retrieval or reconstruction. The retrieval

hypothesis predicts that it is the difficulty of perceiving P1 that strengthens the memory

trace, whereas the reconstruction hypothesis predicts it is the elaboration of the P2 cue

used to retrieve the P1 item that influences memory performance. Although verbal

learning studies have attempted to differentiate between these two theories, this was not

necessary for my purposes, since these two views both yield similar predictions. They

both suggest that structural characteristics of a presentation may encourage more effortful

perceptual processes. For example, in a cued recall task (i.e., where a cue was provided at

test to aid recall of a specific target), Cuddy and Jacoby (1982) presented subjects with

repeated pairs of related words in which the second member was presented either intact

or with letters missing. They found that using intervening material that was similar to to-

be-remembered items resulted in better recall than using different intervening material. In

this instance, the presentation of similar intervening items led to greater interference. This

made for more effortful attempts at perception, since it would then be more difficult to

bring an item into consciousness with similar items appearing before P2. This implies

that if a single brand wants to advertise multiple benefits, it may be advantageous to

group the presentation of these brand-attribute pairs close together in time. Grouping

brand-attribute pairs close together in time would mean that there would be contiguous

presentation of attributes that are all related to the same brand. Attempts at perceiving the

first instance of the pairing at P2 would be more effortful, and would create additional

accessibility to the material at test. Thus, according to the retrieval and reconstruction

hypotheses:

HI: When a brand advertises multiple attributes, grouping these brand-
attribute presentations close together in time will facilitate cued recall.

The hypothesized benefits of grouping different ads for a brand are not a foregone

conclusion. First, the spacing literature shows that massed (i.e., consecutive)

presentations of a stimulus result in less recall of the stimulus and/or its associates.

Second, the encoding variability hypothesis predicts that massing the presentation of

stimuli reduces the variability of the encoding contexts accompanying the presentation of

the stimuli and limits memory access. How would this impact a cued recall task? In a

cued recall task, the predominant source of trace activation is the descriptive components

of the stimulus (in this instance, the brand and its specific unique elements). The context

in which the brand is presented would impact the nature of descriptive components that

are stored. Reduced variability in encoding contexts would lead to reduced variability of

descriptive components. In other words, massing the presentations limits the opportunity

of unique contextual or descriptive cues to aid in the prediction of a brand's associate.

Thus, according to the encoding variability hypothesis:

H2: When a brand advertises multiple attributes, grouping these brand-
attribute presentations close together in time will inhibit cued recall.

CHAPTER 3
EXPERIMENT 1: THE INFLUENCE OF GROUPING

The objective of Experiment 1 was to determine if spacing of repeated stimuli

would interact with different presentation schedules in a manner that would enhance

recall of a brand's multiple attributes. Research by Burke and Srull (1988) and Keller

(1987) showed that advertising multiple attributes across several ads results in reduced

recall of individual attributes. Spacing of repeated attributes and using different

presentation schedules might actually encourage elaboration, and thus eliminate the

interference that is normally associated with multiple attribute advertising. The spacing

effect theories that were outlined earlier provide insight into how this could be achieved.

Both the reconstruction and retrieval hypotheses, henceforth called the R&R

hypotheses, predict that the use of intervening material that is similar to the repeated

stimuli would result in better recall than if different intervening items were used. Thus,

grouping the presentation of brand-attribute pairs may lead to more effortful perceptual

processes and enhance recall. However, the encoding variability theory predicts that

grouping may reduce variability and actually hurt, and not enhance, recall. Thus, in this

experiment, I tested the influence of grouping multiple brand-attribute presentations on

the cued recall of the attributes. This allowed examination of a hypothesized mechanism

for reducing interference in multiple brand-attribute presentations. Also, because of the

differentiating predictions of the hypotheses being used, the results would enable us to

identify a theory or theories that best explain the underlying processes involved

I used a completely within-subject design, with two factors spacing and

grouping. The two levels of the spacing factor were the massed and spaced conditions

often used in spacing experiments. The two levels of the grouping factor were grouped

(different attributes related to the same brand were all presented in a contiguous manner)

and dispersed (different attributes related to the same brand were separated by

intervening brand-attribute presentations from other brands).

Stimuli and Procedure

The experimental design was a two (massed/spaced) by two (dispersed/grouped)

within-subject design with four category replicates (e.g., cameras, automobiles, cell

phones, and televisions). In addition, each condition had four unique attributes paired

with a single brand. Thus, there were 16 brand names (i.e., four per product category) and

64 product attributes (i.e., four per brand). In the massed-dispersed condition, there were

massed brand-attribute presentations for an attribute, but dispersed brand-attribute

presentations for different attributes. For example, as displayed in Table A-1, if letters are

brands and numbers are attributes, the presentation sequence for the massed-dispersed

condition was Al, Al, B5, B5, C9, C9, D13, D13, A2, A2, B6, B6, C10, C10, D14, D14,

A3, A3, B7, B7, Cl 1, C11, D15, D15, A4, A4, B8, B8, C12, C12, D16, D16. Note that

there are back-to-back presentations of any brand stimulus pair (e.g., Al, Al), but that

presentations of that same brand with a different attribute (e.g., A2, A3, A4) occur after a

delay. In the spaced-dispersed condition, there were no contiguous brand-attribute

presentations for an attribute (e.g., E17, F21, G25, H29, E18, F22, G26, H30, E19, F23,

G27, H31, E20, F24, G28, H32, E17, F21, G25, H29, E18, F22, G26, H30, E19, F23,

G27, H31, E20, F24, G28, H32). As in the massed-dispersed condition, all presentations

of a brand-stimulus pair (e.g., E17, E18, El9, E20) occur twice, but they occur far apart

in time.

Two additional conditions grouped the brand-attribute presentations into blocks.

In the massed-grouped condition, the brand-attribute presentations occurred contiguously

(e.g., 133, 133, 134,134, 135, 135, 136,136, J37, J37, J38, J38, J39, J39, J40, J40, K41,

K41, K42, K42, K43, K43, K44, K44, L45, L45, L46, L46, L47, L47, L48, L48). Note

that there are back-to-back presentations of any brand stimulus pair (e.g., 133, 133) and all

presentations for a given brand (e.g., 133, 134, 135, 136) occur in a block. In the spaced-

grouped condition, there were contiguous brand presentations but no contiguous brand-

attribute presentations (e.g., M49, M50, M51, M52, N53, N54, N55, N56, 057, 058,

059, 060, P61, P62, P63, P64, M49, M50, M51, M53, N53, N54, N55, N56, 057, 058,

059, 060, P61, P62, P63, P64). As in the massed-grouped condition, all presentations of

a brand-stimulus pair (e.g., M49, M50, M51, M52) occur twice, but they occur far apart

in time.

Pretests were used to select unfamiliar, neutral brand names from a list of foreign

brand names found on the Internet. This was done to limit the possibility that subjects

would use pre-existing associations to brand names to aid the recall of attributes.

Additional pre-testing was conducted to select product attributes with which subjects

were moderately familiar. In general, attributes that scored lower than 3 or higher than 5

on a 7-point familiarity scale were not used. However, there were a few instances where

attributes that were rated higher than five were chosen. These were all randomly

distributed among the different conditions. Subjects were invited into a lab and told that

they would be viewing a series of brand-attribute pairs and that they would be given a

recall test at the end of the session. Then, the brand-attribute pairs were presented on a

computer screen using an Authorware program. The screen contained a product category

label, and the brand-attribute pair. Each screen was displayed for 4 seconds and was

separated by an inter-stimulus interval of .5 seconds. Immediately after the presentation

of the 32 brand-attribute pairs in a condition, subjects were prompted with a category

label and brand name and asked to list any attributes that were associated with the brand.

After entering one attribute, a new screen appeared that prompted the subject to enter

another attribute. If subjects recalled an attribute, then this process continued until all

four attributes were entered. Subjects were allowed 18 seconds to begin typing an

attribute; otherwise they were prompted to go on to the next brand. The assignment of

brand name to a set of four product attributes within a product category was

counterbalanced. The order of presentation of the conditions was counterbalanced. The

order of presentation of brands in the recall task was randomly determined. Note that one

brand and four attributes from each of the four product categories appeared in each of the

four conditions, thus there could be no category by condition confound.

Predictions

Predictions about the influence of grouping can be made relative to the dispersed

conditions. Recall that the dispersed conditions are representative of the stimulus

presentation schedules observed in most spacing studies; hence the R&R and the

encoding variability hypotheses predict that recall will be higher in the spaced-dispersed

compared to the massed-dispersed condition. In order to examine the prediction of the

R&R hypotheses for example, we can compare Gosen (Brand A in the massed-dispersed

condition) with Kunnan (Brand E in the spaced-dispersed condition) (Table A-l). In the

massed-dispersed condition, on the first presentation of the first Gosen brand-attribute

pairing (P1-Al), one would encounter some difficulty in trying to perceive the stimulus.

Since the individual would have recently been exposed to the same stimulus, this effort

would be reduced on an immediate repetition of the pairing (P2-Al). This same sequence

of "increased effort" perceptual processes followed by "reduced effort" perceptual

processes would occur for subsequent massed presentations of the Gosen brand-attribute

pairings (i.e., A2, A2,...A3, A3,...A4, A4). Compared to the massed condition, the

spaced presentations would undergo more effortful attempts at perception for each

presentation of a brand-attribute pairing. Thus, both P1 and P2 would undergo "increased

effort" perceptual processes. The retrieval and reconstruction hypotheses predict that this

would create additional accessibility to the spaced material at test and result in the typical

spacing effect (Figure B-lb). The encoding variability hypothesis predicts that the

spacing of repetitions would result in the second presentation of a brand-attribute pairing

occurring in a different encoding context. This variability in encoding would lead to

enhanced recall in the spaced-dispersed condition relative to the massed-dispersed

condition, where brand-attribute pairs are repeated within the same encoding context

(Figure B-la).

Grouping of brand-attribute pairs would yield somewhat different predictions.

Note that the grouped conditions create a more contiguous presentation of brand-attribute

pairs. In effect, grouping creates additional massing; hence most explanations of the

spacing effect predict that grouping will hurt recall. For example, the encoding variability

hypothesis predicts that moving from dispersed to a grouped presentation schedule

reduces variability from Brand-attribute 1 to Brand-attribute 2, hence there should be

fewer descriptive and contextual cues that encourage the recall of any attribute

(Hypothesis 2). Therefore, encoding variability theory predicts a positive effect of

spacing and a negative effect of grouping on recall (Figure B-la).

It is also possible that grouping can help cued recall of brand attributes. The R&R

hypotheses predict that grouping should result in greater difficulty in trying to perceive

brand attributes because of similar material appearing immediately before the repeated

stimuli. For example, with the first repetition of a Gamo brand-attribute pair (i.e., 133,

133), we would undergo the same type of processes in trying to perceive the item as in the

massed-dispersed case greater difficulty in perceiving the first instance (PI-133) and

reduced difficulty in perceiving the second (P2-I33). However, unlike in the massed-

dispersed presentation schedule, when trying to perceive the second attribute (134), there

would be competition generated by the recently presented brand-attribute pairings which

attempted to form associates to the same brand. This would lead to a more difficult

perception opportunity at P1-134. This competition would also exist at P2-I34, resulting

in increased difficulty in perceiving that stimulus as well. This would occur for all

subsequent pairings for the Gamo brand in the massed-grouped condition. For example,

at P2-I36, even though this pairing appears immediately after PI-136, there would be

interference created by the presentation of 133, 134, and 135 immediately before 136.

Therefore, it would be difficult to bring the P2-I36 brand-attribute pairing into

consciousness. This presentation schedule creates competition from similar items when

trying to perceive later occurrences of brand-attribute pairs related to the same brand.

This leads to greater accessibility to the material at a later stage. Thus, at test when a cue

is presented, the result would be a facilitating effect on memory in the massed-grouped

condition relative to the massed-dispersed condition (Hypothesis 1).

The R&R hypotheses do not predict a facilitation effect of grouping in the spaced

condition. In the spaced-grouped condition, even though there are similar items appearing

before presentations, e.g., M49, M50, and M51 appear before M52, this competition is

not as great as in the latter stages of the massed-grouped case. Thus, while there is some

difficulty in perceiving later presentations, e.g. M52, this might not be enough to

overwhelm the impact of spacing. Thus, the spaced-dispersed and spaced-grouped

conditions should only benefit from the impact of spacing, and the grouping variable

would have little or no impact. As far as the impact of spacing, these two groups both

have approximately the same number of items between repetitions; hence there should be

little difference in recall between these two groups (Figure B-lb).

Results

Forty-six undergraduate students participated in the experiment for extra credit. A

repeated-measure MANOVA found no interaction of the brand name or condition order

counterbalance factors with the spacing or grouping manipulations (all F < 1.0). A test

for an interaction between spacing and grouping variables was significant (F(l, 45) =

6.35, p <.05). In the dispersed condition, the percentage of attributes correctly recalled

was significantly greater in the spaced condition (p= .50) and the massed condition (p =

.38; F(l, 45) = 6.38, p <.05). In the grouped condition, the percentage of attributes

correctly recalled did not differ between the spaced condition (p = .51) and the massed

condition (p= .49; F (1, 45) = 1.58,p > .05). A test comparing the grouped (p= .49) to

the dispersed (p= .38) presentation schedule in the massed condition was statistically

significant (F(1, 45) = 6.32,p <.05). In the spaced condition however, there was no

significant difference between the grouped condition (p= .51) and the dispersed

condition (p= .50, F (1, 45) = 0.82, p > .05).

Discussion

The key finding of Experiment I was that the massed-grouped presentation

schedule improved recall relative to the massed-dispersed presentation schedule, even

though the net effect of grouping was to further mass the presentations. The encoding

variability hypothesis predicted that additional massing would hurt recall, whereas the

R&R hypotheses predicted that the extra massing would help recall.

The results have some practical implications for advertising scheduling. For

example, suppose that the massed presentation of my stimuli (e.g., Al, Al) is equivalent

to a 60-second advertisement within which a brand name and a product attribute are

repeatedly paired. Results in the massed-dispersed condition show that dispersing the

presentations of ads promoting different attributes will hurt the recall of each individual

brand attribute. In contrast, if these four 60-second ads promoting four different attributes

were shown close together in time (i.e., same commercial block), there would be no such

interference in memory for the brand attributes. The results also show that when ad

length is lessened (e.g., 30 seconds) so that there is less repetition in an individual ad, but

the frequency of this advertising is increased, both dispersed and massed presentation

schedules will encourage the recall of brand attributes.

There is an alternative explanation for the results. Grouping may not only lead to

an interference effect and greater degree of difficulty in perceiving the stimuli, but

grouping could also lead to greater elaboration of the brand-attribute pairs. If this is so,

22

then a voluntary attention hypothesis could also explain the results of Experiment 1. The

voluntary attention hypothesis predicts that people voluntarily pay less attention to P2

when it occurs shortly after P1. Zechmeister and Shaughnessy (1980) argued that a

massed presentation schedule gives people a false sense of confidence about the stimulus

at P2, hence they ignore P2. In the massed-grouped condition of Experiment 1, it could

be argued that people no longer ignore P2 because they can elaborate about how it

combines with all of the other attributes associated with the brand. Given the possibility,

Experiment 2 compares the predictions of the R&R hypotheses with a voluntary attention

hypothesis.

CHAPTER 4
EXPERIMENT 2: THE EFFECT OF WITHIN-CATEGORY COMPETITION

Experiment 2 examined the impact of grouping in a competitive environment. In

Experiment 1, the brands used in each condition were not from the same product

category. In each condition, four category replicates (e.g., cameras, automobiles, cell

phones, and televisions) were used, and only one brand was associated with each

category. Previous research in marketing has already demonstrated that competing ads

from the same category can generate interference that impairs memory for a brand's

attributes (Keller 1987). We have already shown that grouping could enhance memory

when multiple attributes are associated with a brand. In this instance, we would also

expect grouping to enhance memory for the multiple attributes. This is predicted by both

the R&R hypotheses and the attention hypothesis when items are massed. However, the

two theories make differentiating predictions when items are spaced. The R&R

hypotheses predict an advantage of grouping and the attention hypothesis does not. Thus,

Experiment 2 allows an examination of a context in which the R&R hypotheses and the

voluntary attention hypothesis made different predictions about the cued-recall of brand

attributes. Second, the experimental context used was one in which different brands

shared the same attributes, a common occurrence in markets consisting of many

competitors.

Stimuli and Procedure

This experiment contained four within-subject conditions, and used an eight-cell

design. Four of the cells were repeated from Experiment 1. Therefore, I simply replicated

the stimuli from that experiment, but used half as many brand replicates (e.g., compare

the C and D stimuli in Table A-2 to C and D stimuli in Table A-1). In four new

conditions, I changed the brand name presented during P2. For example, the Al, Al, B5,

B5 stimulus stream in the massed-dispersed condition became Al, QI, B5, R5. Again, I

used only two brand replicates per category replicate. Thus, I added the same/different

brand name cue factor by altering the P2 brand name for half of the brand replicates used

in Experiment 1 (Table A-2).

Again, I measured recall of brand attributes given a brand name cue. For

attributes paired with two brand names, each brand name was given as a cue and the

subject's attribute response was scored in each case. This created an opportunity for twice

as much recall relative to the same brand name cue conditions, so I will focus my

analysis on differences in the patterns of recall across the same and different brand name

cue conditions.

Predictions

When the same brand name was paired with the same attribute at P2, the design

was identical to the design used in Experiment 1, thus I should observe the same pattern

of means as in Experiment 1. The massed-grouped presentation schedule should improve

recall relative to the massed-dispersed presentation schedule, and there should not be a

significant difference in recall in the spaced-grouped and the spaced-dispersed

conditions. However, when two brands are paired with the same attribute (different

brand cue condition), the hypotheses make both common and differentiating predictions.

First, both hypotheses predict that recall should decline when a second presentation of an

attribute is accompanied by a different brand name. There is one less chance for the

association between the brand name and attribute to strengthen. In addition, the voluntary

attention hypothesis predicts that attention to the second presentation of the attribute

should increase owing to the change in brand name. Thus, the voluntary attention

hypothesis predicts a potential decrease in recall owing to the weaker brand-attribute

association and a potential increase in recall owing to the increased attention to the

attribute at P2, but no difference in the pattern of recall (e.g., no three-way interaction of

spacing, grouping, and brand name cue consistency.

Although the R&R hypothesis also predicts a decline in recall owing to the

change in brand names at P2, it also predicts a three-way interaction. Wherein the

dispersed conditions in Experiment I showed a spacing effect, they should now show a

null effect for the different brand name cue stimuli. In the dispersed condition, each item

is perceived as unique and the same amount of processing is involved for each item,

whether it is massed or spaced. Thus, there should be no spacing effect. In addition,

wherein the grouped conditions in Experiment 1 showed a null effect, they should now

show a spacing effect when different brand name cues are used. In the different cue

condition, the changed brand name results in the paired associates appearing to be

unique, once-presented items. When these items are grouped, the contiguity of

presentation helps the subject elaborate on the manner in which these once-presented

stimulus-response pairs are all associated. This elaboration of PI or P2 promotes retrieval

and/or reconstruction. This is easier to do in the spaced-grouped condition (e.g., M49, M50.

M51, M52), than in the massed-grouped condition (e.g., 133, U33,134, U34,135, U35, 136, U36).

Hence, spaced-grouping leads to elaboration and promotes retrieval at test.

Results

Thirty undergraduate students participated in the experiment for extra-credit. A

repeated-measure MANOVA found no interaction of the brand name or condition order

counterbalance factors with the spacing, grouping, or brand name cue consistency

manipulations (all F < 1.0). A test for a three-way interaction of spacing, grouping, and

brand name cue consistency was statistically significant (F(l, 29) = 6.27,p <.05).

The test for a spacing by grouping interaction in the same brand name cue

condition was significant (F(l, 29) = 6.47, p < .05). In the dispersed attributes condition,

the percentage of attributes correctly recalled was significantly greater in the spaced

condition (p= .69) than in the massed condition (p= .50; F(I, 29) = 6.30, p < .05). In the

grouped attributes condition, the percentage of attributes correctly recalled did not differ

between the spaced condition (p= .71) and the massed condition (p= .67; F(1, 29)=

2.21, p > .05). These results replicate the results of Experiment 1.

The test for a spacing by grouping interaction in the different brand name cue

condition was significant (F(l, 29) = 5.48, p < .05). As predicted by the R&R hypotheses,

in the dispersed condition, the percentage of attributes correctly recalled did not differ

between the spaced condition (p= .45) and the massed condition (p= .42; F(1, 29)=

1.58, p > .05). Additionally, as predicted by the R&R hypotheses, in the grouped

condition, the percentage of attributes correctly recalled was significantly greater in the

spaced condition (p= .60) than in the massed condition (p= .48; F (1, 29) = 5.42, p <

.05).

Discussion

These results are consistent with the R&R hypotheses. Experiment 1 showed that

the massed presentation of brand-attribute pairings could be beneficial if the brand was

paired with multiple attributes in a concentrated period of time (e.g., the massed-grouped

condition). Experiment 2 showed that this strategy was not effective if competing brands

are advertising the same attributes. The spaced exposure combined with grouped-

attributes presentation schedule was shown to be more effective at promoting brand

learning. The spaced-grouped presentation schedule limited the interference from

competing brands and encouraged elaboration of the brand-attribute pairs for specific

brands.

Again, the results have potential implications for advertising practice. First, they

confirm that competitor advertising hurts cued-recall of brand attributes. I extend the

findings of Keller (1987) and Burke and Srull (1988) by showing this interference also

occurs when the competing ads mention the same attribute. A second important result is

that one method to combat this interference is to block a brand's advertisements on

multiple attributes and schedule these blocks at different time than competitor's

advertisement. This second conclusion is interesting because theories of proactive and

retroactive memory interference would argue that blocking multiple brand-attribute ads

for a single brand will always reduce memory for each attribute paired with the brand.

However, Experiment 2 shows that grouping ads for a brand can benefit memory in some

circumstances.

Thus far, I have evidence that grouping the presentations of brand's ads can aid

recall of the attributes presented in the ads. Yet, there are two more issues to be resolved.

First, I have argued that grouping can help memory access when a brand advertises

multiple attributes. Yet, there is no evidence that moving from one to four attributes

enhances the recall of any one attribute. My evidence only shows that some brand -

multiple attribute presentation schedules are better than others. The grouping in the first

two experiments may simply be lessening the amount of interference. Pairing a brand

with four attributes could be much worse than pairing a brand with a single attribute.

Spacing and grouping may only partially mitigate the interference problem.

Second, I have assumed that it is the difficulty of retrieving/reconstructing a PI

presentation of a stimulus that leads to increased memory and I have also assumed that

greater elaboration at P /P2 supports retrieval/reconstruction. If this were true, then

techniques that either enhance or reduce elaboration at P1/P2 should lead to enhanced or

impoverished recall; providing further support for the memory accessibility characteristic

of the R&R hypotheses. I handle this latter problem in Experiment 4, leaving the former

problem for Experiment 3.

CHAPTER 5
EXPERIMENT 3: SINGLE ATTRIBUTE VS. MULTIPLE ATTRIBUTES

Experiment 3 investigates the degree of interference when brands are paired with

a single attribute, two attributes, or four attributes. Recall that Burke and Srull (1988)

showed that two brand ads for the same brand advertising two competing attributes (e.g.,

BI -4 A1, B1 -) A2) reduced memory for both of the attributes. Burke and Srull used a

spaced presentation schedule and found that the interference effect was mitigated when

subjects were encouraged to elaborate on the advertisement. Given my claim that

grouping brand ads that promote different attributes encourages elaboration, which in

turn improves recall, I expected that grouping would mitigate any interference effect

associated with a brand advertising competing attributes.

Stimuli and Procedure

The experimental design was a two (massed vs. spaced) by two (dispersed vs.

grouped) by three (1 vs. 2 vs. 4 attributes) within-subject design with seven product

category replicates (e.g., personal computers, cordless phones, computer printers,

cameras, automobiles, cell phones, and televisions). The design of Experiment 3 was

similar to Experiment 1, but included an additional variable number of attributes

associated with a brand. For Experiment 3, the stimuli were designed to limit

confounding variables. Four unique brands were paired with four unique attributes, two

unique brands were paired with two sets of unique attributes, and one brand was paired

with four attributes. In this way, the subject had the opportunity to recall four attributes

for the single attribute brands, four attributes for the double-attribute brands, four

attributes for the quad-attribute brand. Within the one and the two attribute conditions,

brand-attribute pairs came from different product categories. In other words, there were

four unique brand/product categories in the one attribute condition, (Brands A, B, C, and

D), and two unique brand/product categories in the two attribute condition (Brands E and

F) (Table A-3). The design included the same four conditions as were used in Experiment

I with an adjustment for the attribute manipulation. In the massed-dispersed condition

(Table A-3), subjects saw Al, Al, E5, E5, G9, G9, B2, B2, E6, E6, G10, G10, C3, C3,

F7, F7, Gl11, G11, D4, D4, F8, F8, G12, G12. In the spaced-dispersed condition, subjects

saw H13, L17, N21, 114, L18, N22, J15, M19, N23, K16, M20, H13, L17, N21,114, L18,

N22, J15, M19, N23, K16, M20. Note that the number of intervening items for the single

attribute brands (e.g., H, I, J, K), the double-attribute brand (e.g., L, M) and the quad-

attribute brand (e.g., N) was a constant two items.

In the massed-grouped condition, subjects saw 025, 025, P26, P26, Q27, Q27,

R28, R28, S29, S29, S30, S30, T31, T31, T32, T32, U33, U33, U34, U34, U35, U35,

U36, U36). In the spaced grouped condition, subjects saw V37, V38, X39, Y40, Z41,

Z42, AA43, AA44, AB45, AB46, AB47, AB48, V37, V38, X39, Y40, Z41, Z42, AA43,

AA44, AB45, AB46, AB47, AB48. The order of the single-attribute brands, the double-

attribute brands, and the quad-attribute brand was counterbalanced in the grouped

conditions to control for primacy and recency effects. The condition presentation order

was counterbalanced.

Predictions

When the brand is paired with four attributes, the design is similar to the design

used in Experiment 1. Therefore, the R&R hypotheses would predict similar results to

Experiment 1 for the massed-grouped and spaced-grouped cells. In the two-attribute

condition, the design is similar to that of the four-attribute condition. But when brand-

attribute pairs are grouped, the two-attribute pairings yield a different presentation format

than the four-attribute pairings. In this instance, fewer brand-attribute pairs that are

associated with the same brand are presented contiguously in the two-attribute condition

(e.g., J39, J39, J40, J40) than in the four-attribute condition (e.g., K41, K41, K42, K42,

K43, K43, K44, K44). Thus, I would expect there to be "increased effort" perceptual

processes in the grouped four-attribute compared to the grouped two-attribute condition.

This would occur since there are a greater number of similar items appearing

immediately before the repeated stimuli. This would lead to higher levels of recall in the

four-attribute condition (Cell A) than in the two-attribute condition (Cell B) (Figure B-

5b).

As discussed in Experiment 1, the dispersed conditions are representative of the

presentation schedules observed in most spacing studies; hence there should be higher

recall in the spaced-dispersed than in the massed-dispersed condition whether or not two

or four attributes are used. The findings of Keller (1987) and Burke and Srull (1988)

imply that as the number of attributes advertised by a brand increase, recall of any one

attribute should decrease. Thus, in the massed-dispersed condition, we would expect a

decrease in recall when moving from two attributes (Cell E) to four attributes (Cell F)

(Figure B-5b). In the one-attribute condition, the presentation schedules are the same in

both the grouped and dispersed conditions; therefore the R&R hypotheses would predict

no difference in the pattern of results in these two conditions. I would expect a spacing

effect here since this presentation format is also similar to those used in spacing studies.

As there is no additional elaboration generated from the massed-dispersed presentation

format, I would expect increased levels of interference when the number of attributes

increases. Thus, there would be lower levels of recall for conditions with a greater the

number of attributes (Cell D 4 Cell E 4 Cell F).

The encoding variability hypothesis would predict a different pattern of results.

This hypothesis would predict an advantage of spaced stimuli over massed stimuli in all

conditions. As in Experiment 1, moving from dispersed to a grouped presentation

schedule would be predicted to have a negative impact on recall as there would be less

varied descriptive and contextual cues. In addition, the encoding variability hypothesis

would predict that increasing the number of attributes associated with a brand would

further impoverish recall. This theory predicts that the retrieval process is subject to cue

overload, in that a brand would lose its effectiveness as the number of attributes with

which it is associated increases (Glenberg 1979). Thus, moving from one to two, and then

four attributes would lead to increasingly lower levels of recall (Figure 5).

Results

Sixty-three undergraduate students participated in the experiment for extra-credit.

A repeated-measure MANOVA found no interaction of the brand name or condition

order counterbalance factors with the spacing, grouping and attribute number

manipulation (all F < 1.0). A test of the three-way interaction of spacing, grouping, and

number of attributes was statistically significant (F (1, 61)=4.57, p < .05). In the massed-

grouped condition, there was no significant difference between recall in the two-attribute

condition (p= .67) compared to the four-attribute condition (p= .73; F(1, 61) = 2.16,p >

.05). However, compared to the one-attribute condition (p= .52), recall was significantly

greater in both the two-attribute (F (1, 61) = 6.89, p < .05) and the four-attribute case (F

(1, 61) = 7.84, p < .05. There was no significant difference in the one-attribute condition

when comparing the grouped and dispersed case (F (1, 61) = 0.18, p> .05). In the

dispersed condition, there was no significant difference between recall in the two-

attribute (p= .43) compared to the four-attribute condition (p = .39; F (1, 61) = 1.48, p >

.05). But in a result that was opposite to the grouped case, when items were massed and

dispersed, recall was significantly lower for the two (F(1, 61) = 3.86,p <.05) and four-

attribute (F (1, 61) = 4.69,p < .05) conditions compared to the one-attribute condition (p

=.51).

There was a significant difference between grouped and dispersed stimuli.

A significantly greater number of grouped attributes (p= .68) were recalled compared to

dispersed attributes (p= .57; F (1, 61) = 38.19, p <.05). Additionally, there was a

significant difference between recall in the different attribute conditions, a greater

number of attributes were recalled in the two (p = .63) and the four-attribute (p= .66)

conditions compared to the one-attribute condition (p= .515; F (1, 61) = 9.51, p <.05).

Discussion

I have argued thus far that grouping can help memory access when a brand

advertises multiple attributes. The results of this experiment confirm this and also

provide evidence that grouping enhances memory for individual attributes even when we

move from one to four attributes in the massed-grouped condition. The results also

provide some support for the explanations of Keller (1987) and Burke and Srull (1988) -

moving from one to four attributes results in increased interference and leads to reduced

recall of individual attributes in the massed-dispersed condition. These results are

consistent with the R&R hypotheses, which predicted that grouping would positively

interact with an increase in the number of attributes. The encoding variability hypothesis

predicted incorrectly that grouping would have a deleterious effect on recall as the

number of attributes increased.

In terms of advertising practice, these results would also be of some significance.

They confirm that an increase in the number of competing attributes associated with a

specific brand leads to greater interference and reduced recall. But the results illuminate

several avenues that could reduce said interference. As in the previous experiments,

further massing, in the form of a grouped presentation schedule, enhances recall. An

advertising parallel is the repeated exposure of distinct ads that promote each of these

attributes close together in time (such as in the same commercial block). As in

Experiment 1, the results also confirm that reducing ad length along with increasing ad

frequency both in dispersed and grouped schedules would also enhance recall. One

other interesting finding is that when a product uses a unique positioning (i.e. promotes

only one attribute), recall is enhanced mainly when shorter, more frequently presented

ads are used.

CHAPTER 6
EXPERIMENT 4: THE EFFECT OF RELATED ATTRIBUTES

Experiment 4 investigates whether manipulating variables that promote

elaboration at PI or P2 enhances recall. I have assumed in my previous experiments that

greater elaboration at PI/P2 supports retrieval/reconstruction. If this were true, then

techniques that either enhance or reduce elaboration at PI/P2 should lead to enhanced or

impoverished recall. One such variable is the relatedness of attributes that are associated

with a particular brand. Attributes can have some semantic link (e.g., Horch Automobile

Side airbags, Antilock brakes, Rear seatbelt; all attributes related to safety) or they may

not possess any link (e.g., Adler television Channel Block, Mute function, Closed

caption; no clear relationship exists among these attributes). When a set of related

attributes is presented, we would expect different predictions from the R&R hypotheses

and the other major theory, the encoding variability hypothesis. The encoding variability

theory predicts that in a cued recall task, relatedness of attributes would have little or no

effect. However, the R&R hypotheses predict that there would be a positive impact of

grouping when attributes possess some semantic relationship this would occur in both

the massed and spaced conditions.

Stimuli and Procedure

The context used here was one in which some brands were associated with

attributes that possessed a semantic link, and other brands did not.

Recall in Experiment 2, the contiguity of presentation of once-presented items

lead to greater elaboration and enhanced recall. When subjects are presented with items

that are related in some manner, I would expect some elaboration as the possible linkage

between the attributes is processed. I used a mixed design for this experiment with 3

independent variables spacing and attribute relatedness (within) and grouping

(between).

In each condition, four unique brands were paired with three unique attributes.

Two of the brands were paired with related attributes and two were paired with unrelated

attributes. Attributes were pretested to determine whether subjects perceived them to be

related in some way. In the dispersed condition, when items were massed subjects saw

Al, Al, B4, B4, C7, C7, DIO 0, D10, A2, A2, B5, B5, C8, C8, D11, Dll 1, A3, A3, B6, B6,

C9, C9, D12, D12 (Table A-4). When items were spaced they saw E13, F16, G19, H22,

E14, F17, G20, H23, E15, F18, G21, H24, E13, F16, G19, H22, E14, F17, G20, H23,

E15, F18, G21, H24. In both instances two brands possessed attributes that were related

(e.g., attributes 1, 2, 3 and 13, 14, 15); while the other attributes were not (e.g., attributes

7, 8, 9 and 19, 20, 21). In the grouped condition, when items were massed subjects saw

125, 125, 126, 126,127,127, J28, J28, J29, J29, J30, J30, K31, K31, K32, K32, K33, K33,

L34, L134, L35, L135, L36, L136. When items were spaced they saw M37, M38, M39,

N40, N41, N42, 043,044, 045, P46, P47, P48, M37, M38, M39, N40, N41, N42, 043,

044, 045, P46, P47, P48. The order of the presentation of brands with related or

unrelated attributes was counterbalanced to control for order effects. As in Experiment 3,

the assignment of conditions to product categories and the condition presentation order

were counterbalanced.

Predictions

When unrelated attributes are presented, the design is similar to that used in

Experiment 1, thus I should observe the same pattern of results enhanced memory in

the massed-grouped condition, and the typical spacing effect in the dispersed condition.

However, when related attributes are presented, the R&R hypotheses would yield

different predictions. I would expect a similar pattern of means in the massed condition

for both related and unrelated attributes massed-grouped demonstrating greater recall

than massed-dispersed. However, when items are spaced, there should be additional

elaboration in both the grouped and dispersed cases. Subjects would go through

additional processing of the common thread connecting the different attributes. This

would be facilitated to a greater extent in the grouped case (M37, M38, M39, N40, N41,

N42, 043, 044, 045, P46, P47, P48, M37, M38, M39...) than in the dispersed case,

where the lack of contiguous presentation would not lead to the same level of elaboration

(E13, F16, G19, H22, E14, F17, G20, H23, E15, F18, G21, H24...).

Encoding variability theory would make different predictions. In the unrelated

condition, I would expect similar predictions to those made for Experiment I a main

effect of grouping and a main effect of spacing. In the related condition I would also

expect these results. Using semantically related attributes should give subjects another

retrieval cue, namely, the feature shared by all the attributes. But this additional cue

would have a positive effect on recall only in a free recall task. In a cued recall task the

best cue is the brand or item presented at test, thus additional cues would have minimal

impact (Kahana and Greene 1993).

Results

Seventy-one undergraduate students participated in the experiment for extra

credit. A mixed design was used with two within subject variables spacing and attribute

relatedness and one between subjects variable- grouping. A repeated-measure

MANOVA found no interaction of the brand name or condition order counterbalance

factors with the spacing, grouping and attribute number manipulation al F < 1.0). A test

for a three-way interaction of spacing, attribute relatedness and grouping was not

significant (F(l, 69) = 3.24, p > .05). The test for a spacing by grouping interaction was

not significant (F (1, 69) = 0.31, p > .05). However, for unrelated attributes the pattern of

means did approximate that obtained in Experiment 1. In the dispersed condition, the

percentage of attributes correctly recalled was significantly greater in the spaced

condition (p= .66) than in the massed condition (p= .52; F(l, 69) = 5.45, p < .05). In the

grouped condition, the percentage of attributes correctly recalled did not differ between

the spaced condition (p= .71) than in the massed condition (p= .64; F(1, 69) = 2.48, p >

.05). In the attribute related condition, when a dispersed presentation schedule was used,

the percentage of attributes correctly recalled was significantly greater in the spaced

condition (p= .80) than in the massed condition (p= .51; F (1, 69) = 27.15, p < .05). In

the grouped condition, the percentage of attributes correctly recalled was also

significantly greater in the spaced condition (p = .90) than in the massed condition (p=

.64; F(l, 69) = 32.07, p < .05). There was higher recall when a grouped presentation

schedule was used (p= .72) than a dispersed schedule (p= .62; F(1, 69) = 12.78,p <

.05). Additionally, a greater number of related attributes (p= .71) were recalled than

unrelated attributes (p= .63; F(l, 69) = 11.79,p < .05).

Discussion

These results provide additional support for the R&R hypotheses. The core

assumption of these hypotheses is that enhanced memory for attributes results from

greater elaboration at P or P2. Using a manipulation, such as relatedness of attributes,

which would lead to greater elaboration, did lead to enhanced recall.

Therefore, this result provides support for the memory accessibility characteristic of the

R&R hypotheses. Encoding variability theory predicted a spacing effect, but did not

predict differential recall for related vs. unrelated attributes when a cued recall task was

used.

These results also have implications for advertising practice. I have shown that a

grouped presentation schedule can help combat the interference generated when multiple

attributes for a brand are presented. However, when related attributes are associated with

a brand, then a spaced presentation schedule is preferred. In terms of ad scheduling, this

would mean that a company (e.g., Volvo) presenting a unique positioning for a brand

(e.g., safety), but using multiple related attributes (e.g., antilock brakes, side airbags, rear

seatbelts) would generate better recall of the brand's attributes by using shorter ads (30

seconds) and presenting these ads in the same commercial block. Each of these ads

would present a unique attribute related to safety. As the previous experiments

demonstrated, when attributes are unrelated, a massed presentation schedule would be

just as effective.

CHAPTER 7
GENERAL DISCUSSION

These four studies collectively provide strong support for the retrieval and the

reconstruction hypotheses as appropriate explanations for the benefits of certain patterns

of repetition and spacing. Both of these theories predict that repetition improves memory

when the presentation schedule encourages elaboration. Elaboration can occur as a result

of different manipulations. In Experiment 1, I showed that grouping of attributes

enhanced elaboration primarily when attributes were massed, thus reducing interference.

In Experiment 2, I showed that this strategy was not as effective in enhancing recall as a

spaced-grouped presentation schedule when competing brands are advertising the same

attributes. Experiments 3 and 4 examined possible problems associated with my

hypothesis, where elaboration was a central component.

In Experiment 3, I showed that grouping enhances memory for individual

attributes even when an increasing number of attributes is presented. Finally, in

Experiment 4, I demonstrated that using a manipulation (relatedness of attributes) that

would lead to greater elaboration at P or P2 enhances recall. All these findings were

predicted by the retrieval and reconstruction hypotheses and not the attention or encoding

variability hypothesis. This result should be of interest to consumer behavior researchers

since encoding variability theory has been a dominant explanation of spacing effects

within the consumer behavior literature (Schumann et al. 1990; Singh et al. 1994; Unnava

and Burnkrant 1991). These theories are well established in the verbal learning literature

but have not been central in our efforts to understand repetition and advertising memory

effects. Additionally, the dispersed conditions in my experiments were similar to the

presentation of stimuli in traditional spacing experiments. Research findings in this area

would indicate that dispersed presentations would be more effective at enhancing

memory access. However, the additional massing that resulted from the grouping

manipulation consistently generated greater recall across the four experiments in this

dissertation. This, finding certainly is surprising given existing empirical findings.

These results have significant implications for advertising scheduling. Schedules

that promote additional elaboration would lead to reduced interference and enhance recall

of attributes. Scheduling different ads for the same brand in the same commercial block

can actually help and not hurt recall of individual attributes. This finding runs counter to

anecdotal advertising advice to limit the amount of attributes advertised for a brand. In

addition to the presentation of multiple attributes for the same brand, interference can

also occur when competing ads mention the same attribute. The results do indicate that a

possible solution is to block a brand's advertisements on multiple attributes and schedule

at different times than competitors' advertisements. Additionally, blocking of shorter

commercials which present different, but related, attributes in each ad would also reduce

interference and lead to optimal recall.

CHAPTER 8
LIMITATIONS AND FUTURE RESEARCH

Limitations

This dissertation makes a significant contribution towards identifying ways in

which advertisers could reduce interference in a multi-attribute, multi-product

environment and also towards our understanding of the underlying processes. However

there are several issues that need further examination. First, the methods used limit the

generalizability of the findings. For example, all of the studies used were restricted to

shopping goods, (e.g., cameras, automobiles, televisions). However, a considerable

amount of advertising spending is also done on ads for convenience goods (e.g., soda)

and services (e.g., investment banking). This lack of different type of replicates limits the

generalizability of the findings discussed in these studies. Another characteristic that

limits the generalizability is the recall task used. The predictions are made relative to a

cued recall task. However, consumers are sometimes faced with memory-based choice

situations that, in an experimental context, would be better represented by a free recall

task (Alba et al. 1991). In other words, cues related to a specific brand are not always

available to consumers in choice situations; an issue that is not addressed in the four

studies presented in this dissertation.

Finally, the stimuli and paired associate task used more closely resemble a strict

verbal learning task than the type of task that is typically used in consumer behavior

research. However, one of the objectives of this research was to identify the underlying

processes involved that lead to enhanced recall for brand attributes. In the few instances

that the spacing effect was examined in marketing, ads were used (Singh et al. 1994,

Unnava and Burnkrant 1991). Using these stimuli in this instance might have increased

the noise in the experimental setting and not allowed us to develop differentiating

predictions using the spacing effect theories.

Future Research

There are two potentially useful extensions of this research. As discussed in the

limitations, sometimes consumers make choices without the benefit of cues related to a

specific brand. Experiments using a free recall task would address this issue. Both the

reconstruction and retrieval hypotheses are applicable in this experimental context and

make predictions relative to free recall (Greene 1989, Rose 1984). Both theories state that

memory would be enhanced in a free recall, as well as in a cued recall, environment. A

second, and related, avenue of research is identifying possible dissociations between the

reconstruction and the retrieval hypotheses. In a recent meta-analysis of ninety-seven

studies both of these explanations were most consistent with the meta-analytic findings

(Janiszewski, Noel, and Sawyer 2002). However, the retrieval hypothesis focuses on the

study-phase retrieval of first occurrence information and predicts that recall of P1 should

increase with lag; but central to the reconstruction hypothesis is the elaboration of stimuli

at P2 (D'Agostino and DeRemer 1973, Jacoby 1978). This difference in the focal point of

elaboration would enable us to make differentiating predictions for the two theories.

However, one caveat to understanding the spacing effect in marketing is that we may not

be able to identify a solitary theory that accounts for all instances where the effect is

observed. Several studies have shown that some combination of the more promising

44

explanations is necessary to explain the spacing effect given its sheer ubiquity (Challis

1993, Greene 1989). Thus, any attempt to establish dissociations between the

reconstruction and retrieval hypotheses and recommend one as a more appropriate

explanation might only shed limited light on the underlying mechanisms behind this

phenomenon.

APPENDIX A

STIMULI AND DESIGN

i0tsUl -1. u&Aptlt liciut I

Massed -Dispersed

Cue Attribute

Gosen Red-eye reduction

Gosen Red-eye reduction

Hale Voice mail

Hale Voice mail

Adler Closed caption

Adler Closed caption

Lancia Leather seats

Lanca Leather seats

Gosen Rechargeable flash

Gosen Rechargeable flash

Hale Web browser

Hale Web browser

Adler HDTV model

Adler HDTV model

Lancia CD-player

Lancia CD-player

Gosen Light-weight

Gosen Light-weight

Hale Vibrating alert

Hale Vibrating alert

Adler Headphone jack

Adler Headphonejack

Lancia Power windows

Lancia Power windows

Gosen Auto-film reload

Gosen Auto-film reload

Hale Paging function

Hale Paging action

Adler Channel block

Adler Channel block

Lanca Keyless entry
Lancia Keyless entry

Spaced -Dispersed

Cue Attribute

Kunnan Remote shutter 133

Avada Digital 133

Jensen Stereo sound 134

Volga Daytime running lights 134

Kunnan Auto film rewind 135

Avada 3-way cahug 135

Jensen Mute function 136

Volga 4-wheel drive 136

Kunnan 35mom film J37

Avada Voice-activated dialing J37

Jensen Flat screen J38

Volga V-8 engine J38

Kunnan Focus lock J39

Avada Headset connector J39

Jensen Second language signal J40

Volga Heated seats J40

Kunnan Remote shutter K41

Avada Digital K41

Jensen Stereo sound K42

Volga Daytime running hghts K42

Kunan Auto film rewind K43

Avada 3-way calling K43

Jensen Mute function K44

Volga 4-wheel drive K44

Kunnan 35mm film L45

Avada Voice-actated dialing L45

Jensen Flat screen L46

Volga V-8 engine L46

Kunnan Focus lock L47

Avada Headset connector L47

Jensen Second language signal L48
Volga Heated seats L48

Massed -Grouped

Cue Attribute

Gamo Viewfinder M49

Gamo Viewfinder M50

Gamo Zoom lens M51

Gasm Zoom lens M52

Gamo Autofocus N53

Gamo Autofocus N54

Gamo Wide-angle view N55

Gamo Wide-angle view N56

Lamber Analog 057

Lamber Analog 058

Laimber 24-hr. battery 059

Lamber 24-hr battery 060

Lamber Car adapter P61

Lamber Car adapter P62

Lamber Caller id P63

Lamber Caller id P64

Titlis 19-inch screen M49

Tillins 19-mch screen M50

Tillins Picture-i-picture M51

Tillins Picture-in-picture M52

Tilins Surround sound N53

Tilnts Surround sound N54

Titlins Universal remote N55

Tilihns Universal remote N56

Horch Anti-ock brakes 057

Horch Anti-lock brakes 058

Horch Power mirrors 059

Horch Power mirrors 060

Horch Rust proofing P61

lorch Rust proofing P62

Horch Air conditioning P63
Horch Air conditioning P64

Spaced -Grouped

Cue Attribute

Harrows Date-statmp

Harrows Weatherproof

Harrows Point-and-shoot

Harrows 6 ft. Flash range

Prolon Folding-case

Prolon Speed dial

Prolon Roam warning

Prolon One-touch real

Aston Parental control

Aston Remote locator

Aston Sleep timer

Aston 32-inch screen

Satra Paint protectant

Tatra Heated mirrors

Tatra Side air-bags

Tatra Child seat

arrows Date-stamp

Harrows Weatherproof

Harrows Pomt-aendshoot

Harrows 6 ft. Flash range

Prolon Folding-case

Prolon Speed dial

Prolon Roam warning

Prolon One-touch redial

Aston Parental control

Astoni Remote locator

Aston Sleep timer

Aston 32-rmbh screen

Tatra Paint protectant

Tatra Heated mirrors

Tatra Side air-bags
Tatra Child seat

Table A-2. Experiment 2

Massed -Dispersed

Cue Attribute

Gosen Red-eye reduction

Trabant Red-eye reduction

Hale Voice .mail

Thuhnli Voice mail

Adler Closed caption

Adler Closed caption

Lancia Leather seats

Lancia Leather seats

Gosen Rechargeable flash

Trabant Rechargeable flash

Hale Web browser

Thulin Web browser

Adler HDTV model

Adler HDTV model

Lancia CD-player

Lancia CD-player

Gosen Light-weight

Trabant Light-weight

Hale Vibrating Alert

Thuhn Vibrating Alen

Adler Headphone jack

Adler Headphone jack

Lancia Power windows

Laneia Power windows

Gosen Auto-film reload

Trabant Autm-film reload

Hale Paging function

Thulin Paging function

Adler Clhannel block

Adler Channel block

Lancia Keyless entry

Lancia Keyless entry

Spaced -Dispersed

Cue Attribute

Kuiman Remote shutter 133

Avada Digital U33

Jensen Stereo sound 134

Volga Daytime running lights U34

Kunnan Auto film rewind 135

Avada 3-way calling U35

Jensen Mute function 136

Volga 4-wheel drive 36

Kounan 35mm film J37

Avada Voice-activated dialing J37

Jensen Flat screen J38

Volga V-8 engine J38

Kunnan Focus lock J39

Avada Headset connector J39

Jensen Second language signal J40

Volga Heated seats J40

Talbot Remote shutter K41

Borgan Digital V41

Jensen Stereo sound K42

Volga Daytime running lights V42

Talbot Auto film rewind K43

Borgan 3-way calling V43

Jensen Mute function K44

Volga 4-wheel drive V44

Talbot 35mmn film L45

Borgan Voice-activated dialing L45

Jensen Flat screen L46

Volga V-8 engine L46

Talbot Focus lock L47

Borgan Headset connector L47

Jensen Second language signal L48

Volga Heated seats L48

Massed -Grouped Spaced -Grouped

Cue Attribute Cue Attribute

Game Viewfinder M49 Harrows Date-stamp

Willys Viewfinder M50 Harrows Weatherproof

Gamo Zoom lens M51 Harrows Point-and-shoot

Willys Zoom lens M52 Harrows 6 ft flash range

Gamo Autofocus N53 Prolon Foldmg-case

Willys Atofocus N54 Prolon Speed dial

Game Wide-angle view N55 Prolon Roam warning

Willys Wide-angle view N56 Prolon One-touchredial

Tillins 19-inch screen 057 Aston Parental control

Tilhms 19-inch screen 058 Aston remote locator

Tillins Picture-in-picture 059 Aston Sleep timer

Tillms Picture-in-picture 060 Aston 32-inch screen

Tillms Surround sound P61 Tatra Paint protectant

Tillms Surround sound P62 Tatra Heated mirrors

Tillms Universal remote P63 Tatra Side air-bags

TIliins Universal remote P64 Tatra Child seat

Lamber Analog W49 Lara Date-stanmp

Tucker Analog WSO Lara Weatherproof

Lamber 24-bhr. battery W51 Lara Point-and-shoot

Tucker 24-hr battery W52 Lara 6 ft, flash range

Lamber Car adapter X53 Scana Folding-case

Tucker Car adapter X54 Scana Speed dial

Lamber Caller id X55 Scana Roam warning

Tucker Caller id X56 Scana One-touch redial

Horch Anti-lock brakes 057 Aston Parental control

Horch Anti-lockbrakes 058 Aston Remote locator

Horch Power mirrors 059 Aston Sleep timer

Horch Power mirrors 060 Aston 32-inch screen

Horch Rust proofing P61 Tatra Paint prtectant

Horch Rust proofing P62 Tatra Heated mirrors

Horch Air conditioning P63 Tatra Side air-bags

Horch Air conditioning P64 Tatra Child seat

Table A-3. Experiment 3

Massed-Dispersed

Cue Attribute

Al Austin 21" monitor

Al Austin 21" monitor

ES Hale Voice mail

E5 Hale Voice mail

G9 Lancia Leather seats

G9 Lancia Leather seats

B2 Gamao Zoom lens

B2 Gamo Zoom lens

E6 Hale Web browser

E6 Hale Web browser

G10 Lancia Pant-protectant

G10 Lancia Paint-protectant

C3 Edsel Surge protector

C3 Ecsel Surge protector

F7 Adler Closed caption

F7 Adler Closed caption

Gll Lancia Powerwmindows

Gl Lancia Power windows

D4 Fortl Inkjet

D4 Fortin Inkjet

F8 Adler Channel block

F8 Adler Channel block

G12 Lancia Keylessentry

G12 Lancia Keylessentry

Spaced-Dispersed

Cue Attribute

H13 Morris CD tom

L17 Prolon Folding-case

N21 Volga Daylime run

114 Harrows Date stamp

LIS Prolon Speed dial

N22 Volga 4-wheel driv

Ji5 Morgan 2-way interc

MI9 Jensen Stereo sound

N23 Volga V-8 engine

KI6 Panhard High-resolut

M20 Jensen Mute functio

N24 Volga Heated seats

HI3 Morris CD rom

L17 Prolon Folding-case

N21 Volga Daytime run

114 Harrows Date stamp

L18 Prolon Speed dial

N22 Volga 4-wheel drive

J15 Morgan 2-way intercc

MI9 Jensen Stereo sound

N23 Volga V-8 engine

K16 Panhard High-resoluti

M20 Jensen Mute flnctioe

N24 Volga Heated seats

ning lights

Solo

Ion color

ting lights T32

T32

U33

e U33

m 1134

U34

U35

on color U35

n U36

Massed-Grouped

Cue Attribute

025 Bristol 10 gig hard drive V37

025 Bristol 10 gig hard-drive W38

P26 Willys Viewfinder X39

P26 Willys Viewfinder V40

Q27 Abarth Speakerphone Z41

Q27 Abanth Speakerphone Z42

Opel 5 pgs per minute AA43

Opel 5 pgs per minute AA44

LamberN24hr battery AB45

Lamber24-hr. battery AB46

LamberCar adapter AB47

LamberCar adapter AB48

Tilins Picture-in-pictue V37

Tillins Picture-in-pictureW38

Tillms Headphone jack X39

Tillins Headphone jack YV4

Horch Anti-lock brakes Z41

Horch Anti-lock brakes Z42

Horch Power nmrrors AA43

Horch Power mirrors AA44

Horch Rust proofing AB45

Horch Rust proofing AB46

Horch Side arbags AB47

U36 Horch Side airbags AB48 Tatra Child seat

Spaced--Grouped

Cue Attribute

Dusen 256 mb ram

Kunnan Auto-film rewind

Kaiser Wall mountable

Stoewer Water-resistant ink

Avada 3-way calling

Avada Digital

Aston Parental control

Aston Flat screen

Tatra CD player

Tatra Heated mirrors

Tatra Air conditioning

Tatra Child seat

Dusen 256mb ram

Kunnan Auto-film rewind

Kaiser Wall mountable

Stoewer Water-resistant ink

Avada 3-way calling

Avada Digital

Aston Parental control

Aston Flat screen

Tatra CD player

Tatra Heated mirrors

Tatra Air conditioning

Table A-4. Experiment 4

Group I

Massed-Dispersed Spaced-Dispersed

Cue Attribute Cue Attribute

Al Gamo Digital E13 Lancia Leather seats

Digital.

Side airbags

Side airbags

Analog

Analog

19" screen

19" ncmein

Video output

Video output

Anti-lock brakes

Anti-lock brakes

24-hr. battery

24-hr. battery

Surround sound

Surround sound

4 nib mnemry

4 mb memory

Rear seatbelts

Rear seatbelts

Car adapter

Car adapter

Universal remote
Universalremote

Disposable

3-way calling

Channel block

CD player

Lightweight

Caller id

Mute function

Heated mirrors

Compact

Headset

Closed caption

Leather seats

Disposable

3-way calling

Channel block

CD Player

Lightweight

Caller id

Mute function

Heated mirrors

Compact

Headset

Closed caption

Massed-Grouped

Cue Attribute

125 Gamo Digital.

125 Gamo Digital

126 Gamo Video output

126 Gamo Video output

127 Gamo 4 mb memory

127 Gamo 4 mb memory

J28 Horch Side aerbags

J28 Horch Side airbags

J29 Horch Anti-lock brakes

J29 Horch Anti-lock brakes

J30 Horch Rear seatbelts

J30 Horch Rear seatbelts

K31 Lamber Analog

K31 Lamber Analog

K32 Lamber 24-hr. battery

K32 Lamber 24-hr. battery

K33 Lamber Car adapter

K33 Lamber Car adapter

L34 Tilhns 19" screen

L34 Tillins 19" screen

L35 Tillns Surround sound

L35 Tillins Surround sound

L36 Tillins Universal remote

L36 Tilihs Universal remote

Group 2

Spaced--Grouped

2ue Attribute

cia Leather seats

cia CD player

cia Heated mirrors

en Disposable

en Lightweight

en Compact

da 3-waycalling

da Caller ID

da Headset

er Channel block

er Mute function

er Closed caption

cla Leather seats

cia CD player

cia Heated mirrors

en Disposable

en Lightweight

en Compact

da 3-way calling

da Caller d

da Headset

er Channel block

:r Mute function

:r Closed caption

APPENDIX B
PREDICTIONS AND RESULTS

Massed Spaced Massed Spaced
Presentation Schedule Presentation Schedule

--- Dspersed -+- Grouped -a- Dispersed -- Grouped

Figure B-1. Experiment 1 predictions. A) Encoding variability hypothesis.
B) Reconstruction hypothesis

50

0.6

0.55
S 0.49 0.51
05 5----^
0.500.45
0 0.4

0 38
0.35 0.38

0.3
Massed Spaced
Presentation Schedule

-- -Dispersed Grouped

Figure B-2. Experiment 1 results

51

A B

.. ........ "" -

Massed Spaced Massed Spaced
Presentation Schedule Presentation Schedule
..----- p ..S. ....Gr..S.. D.p....... S.,.-- .Gro-Gp. S.-..
DlsperseD, ir ferflnt -.-.. .Grolupe Different |* Disperse Dffe t -, Groped Dff et

Figure B-3. Experiment 2 predictions. A) Voluntary attention hypothesis.
B) Reconstruction hypothesis

52

0.80

0.70 0.67 0.71
0.71
0.67

060 0.50 0.60

5 0.50
Io ,8 ..
0.40 0.42 0.45
0.42

0.30
Massed Spaced
Presentation Schedule

D- Dspersed Same -*-- Grouped Same
-..- Dispersed Different o *. Grouped Different

Figure B-4. Experiment 2 results

Massed Spaced
Presentation Schedule

B

A-
A -----------

, 7

Massed

Presentation Schedule

Grouped 1 Attnbute --- Grouped 2 Attnbutes ..Grouped 1 Attribute Grouped 2 Attributes
-*-Grouped 4 Attributes a- Dispersed 1 Attribute -+- Grouped 4 Attributes Dispersed 1 Attribute
-- Dispersed 2 Attributes -*- Dispersed 4 Atributes Dispersed 2 Attributes Dispersed 4 Attributes

Figure B-5. Experiment 3 predictions. A) Encoding Variability hypothesis.
B) Reconstruction hypothesis

A

Spaced

0.9 -
0.78
0.8 0.73 U 0.73
0 0.72
0.7 0.73

00.6 -0-
S 0.52 ... 0.6
= 0.5 0.5 -- --------
0.4 o10. -
0.3 o
< 0.2
0.1
0
Massed Spaced
Presentation Schedule

*. Grouped 1 Attribute -- Grouped 2 Attributes
Grouped 4 Attributes ,--. Dispersed 1 Attribute
Dispersed 2 Attributes ----- Dispersed 4 Attributes

Figure B-6. Experiment 3 results

Massed Spaced
Presentation Schedule

- Grouped Related --4- -Grouped Unrelated
--- DispersedRelated ----DispersedUnrelate

Massed

Spaced

Presentation Schedule

- --Grouped Related -*- Grouped Unrelated
-- Dispersed Related -!- Dispersed Unrelate

Figure B-7. Experiment 4 predictions. A) Encoding Variability hypothesis.
B) Reconstruction hypothesis

56

0.89
0.9 7 o.
0. .64 0.80
60.8

0.60.4
0.7...--- .. .-8
0 0.6 4 __-
6 0 ,52 6
0 o.5 0-52
*C" 051
0.4
0.3
0.2
Massed Spaced
Presentation Schedule
-s- Grouped Related -+ Grouped Unrelated
-- Dispersed Related -w Dispersed Unrelatec

Figure B-8. Experiment 4 results

REFERENCES

Alba, Joseph W., J. Wesley Hutchinson, and John G. Lynch Jr. (1991), "Memory and
Decision Making," in Robertson and Kassarjian (Eds.) Handbook of Consumer
Behavior, New York: Prentice Hall.

Braun, Kelly Ann and David C. Rubin (1998), "A Retrieval Model of the Spacing
Effect," Memory, 6 (January), 37-65.

Burke, Raymond. R. and Thomas. K. Srull (1988), "Competitive Interference and
Consumer Memory for Advertising," Journal of Consumer Research, 15 (June),
55-68.

Challis, Bradford H. (1993), "Spacing Effects on Cued-Memory Tests Depend on Level
of Processing," Journal of Experimental Psychology: Learning, Memory and
Cognition, 19 (March), 389-396.

Cornell, Edward H. (1980), "Distributed Study Facilitates Infants' Delayed Recognition
Memory," Memory and Cognition, 8 (November), 539-542.

Craig, Samuel C., Brian Stemthal, and Clark Leavitt (1976), "Advertising Wearout -
Experimental Analysis," Journal of Marketing Research, 13 (November), 375-
372.

Cuddy, Lauren J., and Larry L. Jacoby (1982), "When Forgetting Helps Memory: An
Analysis of Repetition Effects," Journal of Verbal Learning and Verbal Behavior,
21 (August), 451-467.

Dempster, Frank N. (1987), "Effects of Variable Encoding and Spaced Presentations on
Vocabulary Learning," Journal of Educational Psychology, 79 (June), 162-170.

D'Agostino, Paul R. and Paula DeRemer (1973), "Repetition Effects as a Function of
Rehearsal and Encoding Variability," Journal of Verbal Learning and Verbal
Behavior, 12 (February), 108-113.

Ebbinghaus, H. (1913), Memory (H. A. Ruger and C. E. Bussenius, Trans.). New York:
Teachers College. (Original work published 1885) (Paperback ed., New York:
Dover, 164).

Elmes, David G., Craig J. Dye and N. J. Herdelin (1983), "What is the Role of Affect in
the Spacing Effect?" Memory and Cognition, 11 (March), 144-151.

Glenberg, Arthur M. (1976), "Monotonic and Nonmonotonic Lag Effects in Paired-
Associate and Recognition Memory Paradigms," Journal of Verbal Learning and
Verbal Behavior, 14, 1-16.

---- (1979), "Component-levels Theory of the Effects of Spacing of Repetitions on
Recall and Recognition," Memory and Cognition, 7 (March), 95-112.

Glenberg, Arthur M. and Thomas S. Lehmann ( i 'i. "Spacing Repetitions Over 1
Week," Memory and Cognition, 8 (November), 528-538.

Greene, Robert L. (1989), "Spacing Effects in Memory: Evidence for a Two-Process
Account," Journal of Experimental Psychology: Learning, Memory, & Cognition,
15 (May), 371-377.

Hauser, John R. and Berger Wernerfelt (1990), "An Evaluation Cost Model of
Consideration Sets," Journal of Consumer Research, 16 (March), 393-408.

Hintzman, Douglas L. (1974), "Theoretical Implications of the Spacing Effect," in
Theories in Cognitive Psychology: The Loyola Symposium, ed. Robert L. Solso,
Hillsdale, NJ: Erlbaum, 77-99.

Hintzman, Douglas L. and Miriam K. Rogers (1973), "Spacing Effects in Picture
Memory," Memory and Cognition, 1 (October), 430-434.

Hutchinson, J. Wesley, Kalyan Raman, and Murali K. Mantrala (1994), "Finding Choice
Alternatives in Memory: Probability Models of Brand Name Recall, Journal of
Marketing Research, 31 (November), 441-461.

Jacoby, Larry L. (1974), "The Role of Mental Contiguity in Memory: Registration and
Retrieval Effects," Journal of Verbal Learning and Verbal Behavior, 13
(October), 483-496.

---- (1978), "On Interpreting That Effects of Repetition: Solving a Problem Versus
Remembering a Solution," Journal of Verbal Learning and Verbal Behavior, 17
(December), 649-667.

---- and Fergus I. M. Craik (1978), "Effects of Elaboration of Processing at Encoding
and Retrieval: Trace Distinctiveness and Recovery on Initial Context," in L. S.
Cermak and Fergus I. M. Craik (Eds.), Levels of Processing and Human Memory,
Hillsdale, NJ: Erlbaum.

Janiszewski, Chris, Hayden Noel, and Alan Sawyer (2002), "A Meta-Analysis of the
Spacing Effect in Verbal Learning: Implications for Research on Advertising
Repetition and Consumer Memory," Working Paper. University of Florida.

Kahana, Michael J. and Robert L. Greene (1993), "Effects of Spacing on Memory for
Homogeneous Lists," Journal of Experimental Psychology: Learning, Memory,
and Cognition, 19, 159-162.

Keller, Kevin L. (1987), "Memory Factors in Advertising: The Effect of Advertising
Retrieval Cues on Brand Evaluations," Journal of Consumer Research, 14
(December), 316-333.

---, Susan E. Heckler, and Michael J. Houston (1998), "The Effects of Brand Name
Suggestiveness on Advertising Recall," Journal of Marketing, 62 (January), 48-
59.

Lockhart, Richard S., Fergus 1. M. Craik, and Larry L. Jacoby (1976), "Depth of
Processing, Recognition, and Recall: Some Aspects of a General Memory
System," in J. Brown (Ed.) Recall and Recognition, London: Wiley.

Madigan, Stephen A. (1969), "Intraserial Repetition and Coding Processes in Free
Recall," Journal of Verbal Learning and Verbal Behavior, 8, 828-835.

Melton, A.W. (1970), "The Situation with Respect to the Spacing of Repetitions and
Memory," Journal of Verbal Learning and Verbal Behavior, 9, 596-606.

Murdock, Jr. Bennet, B. and Arthur J. Babick (1961), "The Effect of Repetition on the
Repetition of Individual Words," American Journal of Psychology, 74, 596-601.

Pechmann, Cornelia and David W. Stewart (1988), "Advertising Repetition: A Critical
Review of Wearin and Wearout," Current Issues and Research in Advertising, 11
(2), 285-330.

Postman, Leo and Kitty Knecht (1983), "Encoding Variability and Retention," Journal of
Verbal Learning and Verbal Behavior, 22 (April), 133-152.

Ray, Michael L. and Alan Sawyer (1971), "Frequency Effects Revisited," Journal of
Advertising Research, 11 (February), 14-20.

Rethans, Arno J., John L. Swasy, and Lawrence J. Marks (1986), "Effect of Television
Commercial Repetition, Receiver Knowledge, and Commercial Length: A Test of
the Two-Factor Model," Journal of Marketing Research, 23, (February), 50-56.

Rose, Robert J. (1984), "Processing Time for Repetitions and the Spacing Effect,"
Canadian Journal of Psychology, 38 (December), 669-680.

Rothkopf, Ernst Z. and Esther U. Coke (1966), "Variations in Phrasing, Repetition
Intervals, and the Recall of Sentence Material," Journal of Verbal Learning and
Verbal Behavior, 5, 86-91.

Schumann, David W., Richard E. Petty, and D. Scott Clemons (190), "Predicting the
Effectiveness of Different Strategies of Advertising Variation: A Test of the
Repetition Variation Hypothesis," Journal of Consumer Research, 17
(September), 192-202.

Simmons, Carolyn. J., and John. G. Lynch (1991), "Inference Effects without Inference
Making Effects of Missing Information on Discounting and Use of Presented
Information," Journal of Consumer Research, 17 (March), 477-491.

Singh, Surendra, Sanjay Mishra, Neeli Bendapudi, and Denise Linville (1994),
"Enhancing Memory of Television Commercials Through Message Spacing,"
Journal of Marketing Research, 31 ( August), 384-392.

---- and Michael. L. Rothschild (1983), "Recognition as a Measure of Learning from
Television Commercials," Journal of Marketing Research, 20 (August), 235-248.

---, Michael. L. Rothschild, and Gilbert. A. Churchill (1988), "Recognition versus
Recall as Measures of Television Commercial Forgetting," Journal of Marketing
Research, 25 (February), 72-80.

Thios, Samuel J. and Paul R. D'Agostino (1976), "Effects of Repetition as a Function of
Study-Phase Retrieval," Journal of Verbal Learning and Verbal Behavior, 15
(October), 529-536.

Unnava, H. Rao and Robert E. Burnkrant (1991), "Effects of Repeating Varied Ad
Executions on Brand Name Memory," Journal of Marketing Research 28
(November), 406-416.

Unnava, H. Rao and D. Sirdeshmukh (1994), "Reducing Competitive ad Interference"
Journal of Marketing Research 31 (August), 403-411.

Zechmeister, Eugene B. and John J. Shaughnessy (1980), "When You Know That You
Know and When You Think That You Know but You Don't," Bulletin of the
Psychonomic Society, 15 (January), 41-44.

BIOGRAPHICAL SKETCH

Hayden Noel was born in Port-of-Spain, Trinidad on February 12, 1964. He

attended Richmond Street Boys E.C. School where his grandfather, Victor Noel, was

principal. He then pursued his secondary education at Queen's Royal College in St. Clair

Trinidad; a school that several generations of Noels attended including his grandfather,

his three uncles, his father and his elder brother. He obtained his bachelor's degree in

Management and Economics in 1988 from the University of the West Indies in Mona,

Jamaica. After working at Canadian Imperial Bank of Commerce in a managerial

capacity, he completed a Master of Business Administration degree at Pace University in

New York City. He then re-entered the corporate world for a stint in advertising at the

Bozell Worldwide advertising agency. His time there spurred his curiosity and he became

very interested in advertising research. He then left Bozell to pursue a Ph.D. in Marketing

at the University of Florida. He completed his degree in December 2002 and accepted a

position as an Assistant Professor at the Bernard M. Baruch School of Business, City

University of New York.

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.

M stopher anis ski, Chairman
Professor of Marketing

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.

ifeph W. Alba
Distinguished Professor of Marketing

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.

Stelken M. Shugan /
Russell Berrie Eminent Scdolar of
Marketing

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.

Professor of Marketing

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.

Ira S. Fischler
Professor of Psychology

This dissertation was submitted to the Graduate Faculty of the Department of
Marketing in the College of Business Administration and to the Graduate School and was
accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy.

December, 2002

Dean, Graduate School

LD

1780

20 8556 6437

3 1262 08556 B437

	Title Page
	Dedication
	Acknowledgement
	Table of Contents
	Abstract
	Introduction
	Theoretical background
	Experiment 1: The influence of...
	Experiment 2: The effect of...
	Experiment 3: Single attribute...
	Experiment 4: The effect of related...
	General discussion
	Limitations and future researc...
	Appendix A: Stimuli and design
	Appendix B: Predictions and...
	References
	Biographical sketch

MILLISECOND	CLASS.METHOD	MESSAGE
0	sobekcm_page_globals.constructor
0	sobekcm_page_globals.constructor	Application State validated or built
0	sobekcm_database.verify_item_lookup_object
0	sobekcm_page_globals.constructor	Navigation Object created from URI query string
0	sobekcm_database.verify_item_lookup_object
0	sobekcm_page_globals.display_item	Retrieving item or group information
0	sobekcm_page_globals.get_entire_collection_hierarchy	Retrieving hierarchy information
0	sobekcm_assistant.get_entire_collection_hierarchy
0	cached_data_manager.retrieve_item_aggregation
0	cached_data_manager.retrieve_item_aggregation	Found item aggregation on local cache
0	item_aggregation_builder.get_item_aggregation	Found 'all' item aggregation in cache
0	system.web.ui.page.page_load (ufdc.page_load)
0	sobekcm_page_globals.constructor.on_page_load
0	html_echo_mainwriter.add_style_references	Adding style references to HTML
0	html_echo_mainwriter.add_text_to_page	Reading the text from the file and echoing back to the output stream
56	html_echo_mainwriter.add_text_to_page	Finished reading and writing the file