Citation
Examining Purchase and Non-Redemption of Mail-in Rebates: The Impact of Offer Variables on Consumers' Subjective and Objective Probability of Redeeming

Material Information

Title:
Examining Purchase and Non-Redemption of Mail-in Rebates: The Impact of Offer Variables on Consumers' Subjective and Objective Probability of Redeeming
Creator:
SILK, TIMOTHY GUY ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Consumer research ( jstor )
Coupons ( jstor )
Deadlines ( jstor )
Marketing ( jstor )
Movies ( jstor )
Procrastination ( jstor )
Purchasing ( jstor )
Rate bases ( jstor )
Sales rebates ( jstor )
Websites ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Timothy Guy Silk. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
4/30/2004
Resource Identifier:
55802504 ( OCLC )

Downloads

This item has the following downloads:

silk_t ( .pdf )

silk_t_Page_53.txt

silk_t_Page_71.txt

silk_t_Page_12.txt

silk_t_Page_63.txt

silk_t_Page_82.txt

silk_t_Page_36.txt

silk_t_Page_08.txt

silk_t_Page_89.txt

silk_t_Page_20.txt

silk_t_Page_88.txt

silk_t_Page_84.txt

silk_t_Page_50.txt

silk_t_Page_23.txt

silk_t_Page_72.txt

silk_t_Page_61.txt

silk_t_Page_64.txt

silk_t_Page_42.txt

silk_t_Page_62.txt

silk_t_Page_29.txt

silk_t_Page_16.txt

silk_t_Page_79.txt

silk_t_Page_54.txt

silk_t_Page_76.txt

silk_t_Page_22.txt

silk_t_Page_09.txt

silk_t_Page_85.txt

silk_t_Page_78.txt

silk_t_Page_69.txt

silk_t_Page_21.txt

silk_t_Page_38.txt

silk_t_Page_25.txt

silk_t_Page_66.txt

silk_t_Page_93.txt

silk_t_Page_33.txt

silk_t_Page_05.txt

silk_t_Page_14.txt

silk_t_Page_97.txt

silk_t_Page_41.txt

silk_t_Page_83.txt

silk_t_Page_45.txt

silk_t_Page_32.txt

silk_t_Page_58.txt

silk_t_Page_30.txt

silk_t_Page_28.txt

silk_t_Page_70.txt

silk_t_Page_74.txt

silk_t_Page_13.txt

silk_t_Page_15.txt

silk_t_Page_90.txt

silk_t_Page_01.txt

silk_t_Page_40.txt

silk_t_Page_17.txt

silk_t_Page_07.txt

silk_t_pdf.txt

silk_t_Page_47.txt

silk_t_Page_80.txt

silk_t_Page_44.txt

silk_t_Page_04.txt

silk_t_Page_11.txt

silk_t_Page_73.txt

silk_t_Page_31.txt

silk_t_Page_18.txt

silk_t_Page_19.txt

silk_t_Page_67.txt

silk_t_Page_06.txt

silk_t_Page_02.txt

silk_t_Page_26.txt

silk_t_Page_94.txt

silk_t_Page_77.txt

silk_t_Page_92.txt

silk_t_Page_27.txt

silk_t_Page_57.txt

silk_t_Page_95.txt

silk_t_Page_75.txt

silk_t_Page_49.txt

silk_t_Page_81.txt

silk_t_Page_43.txt

silk_t_Page_35.txt

silk_t_Page_65.txt

silk_t_Page_39.txt

silk_t_Page_91.txt

silk_t_Page_86.txt

silk_t_Page_87.txt

silk_t_Page_60.txt

silk_t_Page_52.txt

silk_t_Page_96.txt

silk_t_Page_34.txt

silk_t_Page_24.txt

silk_t_Page_03.txt

silk_t_Page_46.txt

silk_t_Page_10.txt

silk_t_Page_51.txt

silk_t_Page_68.txt

silk_t_Page_59.txt

silk_t_Page_56.txt

silk_t_Page_48.txt

silk_t_Page_55.txt

silk_t_Page_37.txt


Full Text











EXAMINING PURCHASE AND NON-REDEMPTION OF MAIL-IN REBATES:
THE IMPACT OF OFFER VARIABLES ON CONSUMERS' SUBJECTIVE AND
OBJECTIVE PROBABILITY OF REDEEMING














By

TIMOTHY GUY SILK


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 FLORIDA


2004































Copyright 2004

by

Timothy Guy Silk


































This dissertation is dedicated to my parents.
















ACKNOWLEDGMENTS

I am deeply indebted to my advisor, Chris Janiszewski, who committed himself to

my development from the day I arrived in the program and supported me in pursuing the

topics that excite me. I would also like thank my committee members Richard Lutz and

Joe Alba for their extensive comments which contributed a great deal to the conceptual

development of the dissertation, Alan Cooke for his conceptual guidance as well as his

generous assistance in developing the website used in the experiments, and Mahendrara

Nimalendran for his support and feedback at all stages of the dissertation. I am also

indebted to Joel Cohen, who encouraged me throughout the course of my doctoral

education. Finally, I would like to thank my family for their encouragement, particularly

Alvin Silk for his guidance.




















TABLE OF CONTENTS

Page


ACKNOWLEDGMENT S .............. .................... iv


LI ST OF T ABLE S ................. ................. viii............


LIST OF FIGURES .............. .................... ix


AB STRAC T ................ .............. xi


CHAPTER


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


The Need for Research .............. ...............1.....
Goals of the Dissertation ................. ...............3.......... .....


2 TYPOLOGY OF REBATE PURCHASES .............. ...............4.....


How Rebates Differ From Coupons .............. ...............4.....
Classifying Purchases ................. ........... ...... ....................5
Implications for Measuring Redemption Rates .............. ...............5.....
Implications for Measuring Breakage .............. ...............7.....
Influencing Breakage ................. ...............7.................
Research on Breakage............... ...............8


3 CONCEPTUAL AND THEORETICAL FOUNDATIONS .............. ....................13


Pre-Purchase Factors .............. ...............14....
Redemption Confidence ................. ......... ...............14......
D esirability .............. ...............15....
Temporal Distance............... ...............16
Post-Purchase Factors ................. ...............16.................
Procrastination ................. ...............17.................

Prospective Forgetting ................. ...............18.................
Redemption Effort ................. ...............19.................
Repeat-Purchase Factors............... ...............20
Past Redemption Behavior .............. ...............20....


4 INTRODUCTION TO EXPERIMENTS .............. ...............22....













5 EXPERIMENT 1 .............. ...............23....


Design ................. ...............23.................
Procedure ................ ...............23.................

Phase 1: Purchase .............. ...............23....

Phase 2: Redemption ............... .... .. .......... ...............26......

Phase 3: Responding to Rebate Applications ................. .......... ...............27

Dependent Measures ................. ...............27.................
Purchase .................. ........... ...............27.......

Breakage and Redemption............... ...............2
Procrastination ................. ..... ......... ........... .............2

Redemption Effort Manipulation Check .............. ...............28....
Re sults ................ ...............29.................

Purchase ................. ...............29.................

Re dempti on ................. ...............3.. 0......... ....
Procrastination ................. ...............3.. 0..............

D discussion ................... .......... ...............3.. 1....

Purchase and Redemption .............. ...............3 1....

Insights Regarding Breakage ................. ...............32........... ....


6 EXPERIMENT 2 ................ ...............37........... ....


Design and Procedure ................. ...............38........... ....
Re sults ................ ...............39.................

Purchase ................. ...............39.................

Re dempti on ................. ...............39........... ....
Procrastination ................. ...............40.................

Probability Estimates ................. ...............40.................
Discussion ................. ...............42.................


7 EXPERIMENT 3 .............. ...............47....


Design and Procedure ................. ...............47........... ....
Re sults ................. ...............48.................

Repeat Purchase ................. ...............48.................
Re dempti on ................. ...............48........... ....
Procrastination ................. ...............49.................

Repeat Breakage ................. ...............50.................
Discussion ................. ...............51.................


8 POST TEST SURVEY ................. ...............55................


Re sults .................. ........... ...............55.......

Reasons for Breakage ................. ...............56........... ....
Redeemers vs. Non-redeemers .............. ...............57....
Non-Redeemers ............. .............58......












9 EXPERIMENT 4 ................ ...............62........... ....


Design and Procedure ................. .......... ...............62......
Dependent Measures and Analysis............... ...............63
Re sults ................ ...............64.................

Re dempti on ................. ...............64........... ....
Procrastination ................. ...............64.................
Discussion ................. ...............64.................


10 GENERAL DI SCUS SSION ................. ...............68.......... ....


Alternative Explanations for Breakage ................. ...............69................
Implications for Regulators .............. ...............70....
Curbing Breakage ................. ...............70.................

Implications for M managers ................. .... ........ ...............71......
Trend toward Shorter Redemption Periods .............. ...............71....
The Role of Redemption Effort in Breakage ................. ................. ....__72
Repeat Purchase Behavior............... ...............72
Future Research ................... ... ...............7
Effects of Effort on Redemption .............. .. ......._ ......__ ...........7
Differential Effects of Reward on Purchase and Redemption. ..........................74


APPENDIX : POST TEST SURVEY ITEMS .............. ...............75....


LIST OF REFERENCES ............. ...... ._. ...............80..


BIOGRAPHICAL SKETCH .............. ...............85....

















LIST OF TABLES


Table pg

1. Self-reported reasons for not making a repeat rebate purchase in experiment 3 ...........54

2. Self-reported reasons for breakage by non-redeemers ....._____ ... .....___ ..............60

3. Post-test responses by redeemers and non-redeemers ...........__... ........ ..........._.60

4. Post-test responses by non-redeemers only .............. ...............61....


















LIST OF FIGURES


Figure pg

1. Typology of rebate purchases ........... ......__ ...............11.

2. Impact of post-purchase outcomes on redemption rates ......___ ....... ...__...........12

3. Factors that contribute to breakage ................. ...............21........... ..

4. Purchase opportunity choice set ............. ...............33.....

5. Proportion of experiment 1 participants purchasing rebate and non-rebate offers by
condition............... ...............3

6. Proportion of experiment 1 rebate buyers that applied for the rebate by condition ......34

7. Experiment 1 login delay by condition............... ...............3

8. Experiment 1 mail delay by condition ................. ...............35..............

9. Experiment 1 breakage by stage of redemption (n=184) ................. ............ .........36

10. Proportion of experiment 2 participants purchasing rebate and non-rebate offers by
condition............... ...............4

1 1. Proportion of experiment 2 rebate buyers that applied for the rebate by condition ....43

12. Experiment 2 login delay by condition ...._.._.._ ... ......_.._......_ ..........4

13. Experiment 2 mail delay by condition ...._.._.._ ... ......_.._......_ .........4

14. Experiment 2 confidence and base rate estimates by choice .............. ....................45

15. Experiment 2 confidence and base rate estimates by reward .............. ...................45

16. Experiment 2 confidence and base rate estimates by redemption period ........._.._.......46

17. Experiment 2 confidence and base rate estimates of redeemers and non-redeemers ..46

18. Proportion of rebate buyers making a repeat purchase of the rebate offer in
experiment 3 by condition............... ...............5










19. Proportion of experiment 3 rebate buyers that applied for the rebate by condition ....52

20. Experiment 3 login delay by condition ....__. ................. ..... .......... ..... 5

21. Experiment 3 mail delay by condition ........._._.._......_.. ....... ... ........5

22. Proportion of experiment 4 coupon holders that applied for the rebate by condition .66

23. Experiment 4 login delay by condition ....__. ................. .... ........... .... 6

24. Experiment 4 mail delay by condition ........._._. ........... .......... ..........6
















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

EXAMINING PURCHASE AND NON-REDEMPTION OF MAIL-IN REBATES:
THE IMPACT OF OFFER VARIABLES ON CONSUMERS' SUBJECTIVE AND
OBJECTIVE PROBABILITY OF REDEEMING

By

Timothy Guy Silk

May 2004

Chair: Chris Janiszewski
Major Department: Marketing

Mail-in rebates are popular with retailers and manufacturers because they achieve

attractive price points and increase sales while limiting the number of consumers that

redeem the rebate to obtain the price discount. This research examines a phenomenon

known as breakage, which occurs when consumers are enticed to purchase as a result of a

rebate offer but subsequently fail to apply for the rebate.

Borrowing from research on overconfidence, procrastination and prospective

forgetting, this research examines underlying psychological mechanisms that lead to

breakage and how they exert influence via changes in offer characteristics. A series of

experiments demonstrate that breakage can be attributed to the discrepancy between

consumers' subjective probability of redeeming (i.e., redemption confidence) and their

obj ective probability of redeeming (i.e., actual redemption behavior). Increasing

redemption confidence increases the likelihood of purchasing a rebate offer. However,

high redemption confidence does not necessarily translate into successful redemption.









The experiments produced five key results. First, increasing the rebate reward

increases redemption confidence and the proportion of consumers that purchase the

rebate offer, but has a weaker effect on the proportion of buyers that apply for the rebate.

Second, increasing the length of the redemption period increases redemption confidence

and the proportion of consumers that purchase the rebate offer, but also increases

procrastination and breakage. Third, increasing the effort required to redeem the rebate

results in a "backlash" effect where the redemption rate is marginally higher for

consumers that experience a high-effort redemption process compared to those that

experience a low-effort redemption process. Fourth, failing to redeem a rebate or

experiencing a high-effort redemption process dramatically decreases the proportion of

rebate buyers that purchase the offer again. Finally, structuring redemption procedures to

encourage procrastination can have a greater impact on breakage than increasing the

effort required to redeem the rebate.

These results have maj or implications for marketers in demonstrating that shorter

redemption periods decrease sales and increase redemption applications. Results also

suggest that excessively effortful redemption procedures should not be viewed as a

minimum requirement for breakage and that firms may be over-attributing the role of

effort in creating breakage.















CHAPTER 1
INTTRODUCTION

Mail-in rebates (hereafter rebates) are one of the most popular pricing mechanisms

used by manufacturers and retailers to increase sales and discriminate among consumers

who vary in their price sensitivity (Blattberg and Neslin 1990; Jolson, Weiner, and

Rosecky 1987; Soman and Gourville 2002). Rebates are frequently used in the consumer

goods sector (Tat and Schwepker 1998) and are the most common promotion tactic used

in consumer electronics (Lanctot 2002). Rebates are popular because they can be used to

lower a product' s price and increase sales while limiting the number of consumers that

redeem the rebate to obtain the price discount (Lanctot 2002). As explained by one

retailer, "Manufacturers love rebates because redemption rates are close to none ... they

get people into stores, but when it comes time to collect, few people follow through. And

this is just what the manufacturer has in mind" (Greenman 1999, p. Gl). This research

examines this phenomenon, widely referred to as breakage, which occurs when

consumers are enticed to purchase as a result of the rebate offer but subsequently fail to

apply for the rebate (Jolson et al. 1987).

The Need for Research

The reasons for breakage are not well understood. Despite the attention rebates

have received in the business and popular press (e.g., McGinn 2003; McLaughlin 2002),

and despite calls for regulation to protect consumers (e.g., Shim 2002; Spencer 2002),

research on consumer response to rebates is extremely limited. Most academic rebate

research has relied on consumer surveys and economic models to address other important










rebate issues. For example, research has examined optimum rebate reward amounts (Ali,

Jolson, and Darmon 1994), consumer perceptions of manufacturers' motives for offering

rebates (Avila, Chapman, and Avila 1989; Tat, Cunningham, and Babakus 1988),

consumer attributions of satisfaction with rebate shopping experiences (Hunt and

Keaveney 1992; Hunt, Keaveney, and Lee 1995), consumer perceptions of the

redemption process (Jolson, Weiner, and Rosecky 1987; Tat et al. 1988), and consumer

motives toward rebate redemption (Tat 1994; Tat and Schwepker 1998). With the

exception of the research on delayed incentives (Soman 1998) and correlates of rebate

proneness (Jolson et al. 1987), there have been no investigations of the reasons for

breakage.

Practitioner-driven research on breakage is also severely limited. Interviews with

manufacturers and marketing service firms reveal that practitioner-driven research has

been hindered by four factors.l First, as will be explained, marketers experience

considerable difficulty in isolating and measuring breakage. Second, the use of rebate

promotions is driven primarily by sales managers aiming to increase sales during

promotion periods dictated by key account retailers. As one manager explained, "If your

key account retailer gives you a promotion window and you are looking for a quick and

easy way to increase sales during that window, you are going to put together a rebate

promotion before you use a coupon or some other promotion tactic."2 As a result,

manufacturers are more concerned with meeting sales targets and maintaining downward

pressure on redemption rates than investigating factors that influence breakage. Third,


SSource: BDS Marketing, Consumer Electronics Association, DDS Right Choice, GeigerDonnelly
Marketing, Promotion Marketing Association, TCA Fulfillment Services.

2 Source: BDS Marketing.










manufacturers are extremely reluctant to conduct market-based experiments that

manipulate offer characteristics due to fears that doing so will adversely affect sales

and/or increase redemption rates. Fourth, industry associations and the vast maj ority of

manufactures do not maintain data on rebate promotions, limiting the availability of

market data. In summary, trends in offer characteristics such as decreasing the length of

time consumers are given to redeem and increasing the effort required to redeem have

been guided primarily by industry intuition about effective strategy as opposed to formal

research.

Goals of the Dissertation

Borrowing from research on overconfidence, procrastination and prospective

forgetting, this research examines the underlying psychological mechanisms that lead to

breakage. I will show that changes in offer characteristics have the potential to influence

both purchase and redemption, and that these influences can be opposing or synergistic.

As a consequence, understanding breakage requires an understanding of the behavioral

processes that influence the expectation of redemption at the time of purchase and the

initiation and completion of the redemption process. The dissertation proceeds as follows:

Chapter 2 introduces a typology of rebate purchases to illustrate how breakage differs

from other post-purchase outcomes and discusses managerial issues pertaining to the

design, implementation, and evaluation of rebate promotions. Chapter 3 introduces

psychological theory that predicts how rebate offer characteristics influence the purchase

and redemption behaviors that contribute to breakage. Chapters 4 through 9 present a

series of experiments that test the hypotheses by examining the purchase and post-

purchase behavior of rebate buyers and investigating how and when breakage occurs.

Finally, chapter 10 discusses the findings and implications for regulators and managers.















CHAPTER 2
TYPOLOGY OF REBATE PURCHASES

How Rebates Differ From Coupons

Rebates offer consumers an opportunity to receive a monetary reward for buying a

promoted product, provided they expend some effort to receive the reward (Jolson et al.

1987; Rothschild 1987; Tat et al. 1988). Previous research has treated rebates and

coupons as similar tactics for discriminating among consumers who differ in their price

sensitivity (Blattberg and Neslin 1990; Rothschild 1987). Although rebates may be

similar to coupons in their ability to price discriminate, the temporal arrangement of

purchase, effort, and receipt of the reward differ for rebates and coupons (Soman and

Gourville 2002). For example, consumers who use a typical coupon have already

expended the effort needed to redeem it (e.g., obtaining the coupon from packaging or

advertising), and will receive the reward at the time of purchase. In contrast, consumers

whose choice is influenced by a rebate have not yet expended the effort required to obtain

the reward, and will receive the reward only if they expend the necessary effort after

purchase. Thus, rebates have the added features of a delay between purchase and

completion of the effort, which I term redemption delay, and a delay between completion

of the effort and receipt of the reward, which I term reward delay. The presence of these

delays has resulted in rebates being referred to as delayed incentives (Soman 1998). The

separation of purchase and redemption is an important distinguishing factor between

rebates and coupons because it presents the opportunity for consumers to purchase a

rebate offer but not redeem it (i.e., breakage).









Classifying Purchases

Classifying rebate purchases provides the starting point for understanding

breakage. As shown in figure 1, rebate purchases can be classified into rebate-dependent

purchases (i.e., incremental sales that occur due to the presence of the rebate), and rebate-

independent purchases (i.e., baseline sales that occur even when the rebate is absent).

Rebate-dependent purchases can be further segmented into purchases by buyers who

estimate their probability of redeeming is sufficient to merit purchase (i.e., probabilistic

redeemers) and purchases by buyers who are certain of redemption (i.e., intended

redeemers that estimate their probability of redemption is 100%). Rebate-independent

purchases can be further segmented into purchases by buyers who have no intention to

redeem the rebate (i.e., intended non-redeemers) and purchases by buyers who will take

advantage of the rebate if the opportunity becomes available (i.e., opportunistic

redeemers) .

Implications for Measuring Redemption Rates

The post-purchase outcomes of each purchase classification are important in

understanding how redemption rates are calculated and how they can be non-diagnostic

(and potentially misleading) as a measure of rebate program effectiveness. The rebate

fulfillment industry relies on redemption rates (i.e., the number of redemptions divided

by total units sold) to monitor the effectiveness of their rebate programs and to create

guidelines for program implementation. Rebate promotions that achieve sales obj ectives

and generate redemption rates below guidance are deemed successful, while promotions

that generate redemption rates above guidance are deemed unsuccessful and tend to be









discontinued immediately.l For example, consumer electronics manufacturers are often

told to expect redemption rates of 30% to 35% and hope for rates as low as 20% (Norr

2000), and to discontinue promotions that generate redemption rates above 50% (Norberg

2002).

As illustrated in figure 1, intended redemption and opportunistic redemption are

both included in the calculation of redemption rates, while breakage and intended non-

redemption contribute to lower redemption rates. The inclusion of opportunistic

redemption in the calculation of redemption rates reveals how redemption rates can be

non-diagnostic and potentially misleading as a measure of rebate effectiveness. Consider

the three rebate offers shown in figure 2. These offers generate incremental sales of 20%,

40%, and 60% as a percentage of total sales, and exhibit redemption rates of 30%, 30%,

and 40% respectively. According to industry rules of thumb, offer C is least desirable due

to its high redemption rate, while offers A and B are equally desirable since they yield

equivalent redemption rates. Yet, these conclusions are incorrect. In fact, offer C is most

desirable since it contributes the highest proportion of incremental sales and exhibits the

lowest redemption rate on incremental purchases. Furthermore, in addition to having a

lower redemption rate on incremental purchases, offer B is more desirable than offer A

since it does not pay rebate rewards on base sales to opportunistic redeemers.

This example reveals that redemption rates can be a non-diagnostic and potentially

misleading measure of rebate effectiveness whenever redemption rates are a) diluted by

purchases by intended non-redeemers, and/or b) inflated by redemptions by opportunistic

redeemers. Given that advertising effects and habitual purchases generate a substantial

SSource: BDS Marketing, DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing
Association, TCA Fulfillment Services.









number of rebate-independent purchases (i.e., consumers purchasing due to influences

other than the rebate itself), it is likely that redemption rates observed in the marketplace

are affected by some degree of dilution and/or inflation.2 Although these shortcomings

are difficult to remedy, they are mentioned to emphasize that the typology of rebate

purchases in Eigure 1 can aid managers in recognizing that the industry's preoccupation

with redemption rates can be problematic.

Implications for Measuring Breakage

Breakage is an important outcome for marketers and the rebate fulfillment industry

because it represents incremental sales revenue that does not require the payment of

redemption rewards. It is also an important outcome for consumer protection advocates

who wish to determine if breakage occurs on a scale that warrants regulation. However,

research has yet to obtain accurate measures of breakage. Marketers currently use

redemption rates as a proxy for breakage because aggregate sales data cannot distinguish

rebate-dependent purchases from rebate-independent purchases. As discussed previously,

redemption rates can be a non-diagnostic and potentially misleading measure of rebate

effectiveness. Thus, one of the goals of this research is to experimentally isolate rebate-

dependent purchases to obtain direct measures of breakage.

Influencing Breakage

From the perspective of the issuing firm, successful rebate programs should

increase the number of rebate-dependent purchases (i.e., incremental sales) without

encouraging redemptions by opportunistic redeemers. Successful rebate programs should

also be structured in a manner that limits intended redemption (i.e., allow for breakage).



2 Redemption rates typically range from 2% to 40% depending on the product category.









Attempts to influence breakage usually involve manipulations of rebate redemption

periods (i.e., the amount of time given to redeem the rebate), and redemption effort (i.e.,

the amount of effort required to redeem). For example, over the past 20 years, the rebate

industry has shortened redemption periods from a mode of 60 days to a mode of 30 days

in an effort to increase breakage.3 This trend is continuing with 15-day and 7-day

redemption periods becoming increasingly common. Another trend designed to increase

breakage is increasing redemption effort (e.g., requiring multiple proofs of purchase,

making redemption procedures more complex, etc.). As discussed previously, these

trends have been guided primarily by industry intuition about effective strategy as

opposed to formal research.

Research on Breakage

Although there has been no direct documentation of breakage, three pieces of

evidence are consistent with breakage and highlight the need for additional research.

First, Jolson et al. (1987) surveyed 294 consumers and report that 72 % of "light" rebate

users (people who reported one or two rebate-triggered purchases during the past 12

months) and 39.5% of frequent rebate users (people who reported more than two rebate-

triggered purchases during the past 12 months) admit to breakage on over 30% of their

rebate-triggered purchases of small home appliances costing less than $100. The results

suggest that breakage is a common and recurring problem among rebate buyers.

However, given that these data are self-reported, memory biases (i.e., forgetting) and

self-presentation biases (i.e., not wanting to admit to breakage) are likely to have created

a downward bias in reporting breakage. Conversely, it is questionable whether


3 Source: BDS Marketing, DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing
Association, TCA Fulfillment Services.










respondents were capable of retrospectively separating rebate-triggered (i.e., rebate-

dependent) purchases from rebate-independent purchases, which may have created an

upward bias in reporting breakage.

Second, Dhar and Hoch (1996, Study 1), find that in-store coupons result in a 309%

average increase in incremental sales across five product categories, but that coupon

redemption averaged only 73% of the incremental sales. The implication is that there is a

minimum of 27% breakage. However, the behaviors leading to breakage are

fundamentally different for rebates and coupons due to the difference in the temporal

order of effort and reward discussed earlier. Third, Soman (1998) finds that people are

willing to forego a $1 payment for completing a survey in order to receive a larger

delayed incentive payment (either $2 or $4 depending on the condition) for completing a

second survey (either four or eight pages), yet fail to return the second survey at a later

date (i.e., in exactly two weeks, exactly four weeks, or at any time within 4-weeks).

Redemption rates ranged from a low of 23.5% for those who were offered $4 for

completing an 8-page survey, to a high of 44.5% for those who were offered $4 for

completing a 4-page survey. Soman (1998) proposed that people underweight the future

effort required to redeem relative to the savings offered by the reward. However, two

aspects of the study differ from most real-world rebate promotions. First, participants

were aware of the effort requirements (i.e., the second survey) when deciding whether to

choose the delayed incentive offer, whereas most rebate offers do not disclose the effort

requirements until after purchase. This may have generated a downward bias in the

proportion of participants that agreed to the delayed incentive offer, and raises questions

about how and when effort influences breakage. Second, no financial investment was










required from participants who chose the delayed incentive offer (i.e., forgoing a $1

reward does not involve the same financial cost as purchasing a product offering a

rebate). Thus, participants may not have had the same motivation to redeem as those who

purchase real-world rebates. This may have generated a downward bias in the redemption

rate and raises questions about how the cost/reward relationship influences breakage.

Based on the research conducted to date, the behaviors surrounding breakage are

not well understood. Several explanations for breakage have been suggested in passing,

including overconfidence, procrastination, and forgetting, yet research has yet to directly

examine these explanations in detail. It is also unclear whether consumers repeatedly fall

victim to breakage. Although there is some evidence to suggest that breakage is a

recurring problem for some consumers (e.g., Jolson et al. 1987), the evidence implies that

rebate buyers do not learn from past experience and/or minimize the relevance of failing

to redeem previous offers when considering the purchase of a rebated product. The

current research will address these issues.



















REBATE-INDEPENDENT PURCHASE


I I


OPPORTUNISTIC REDEEMERS


INTENDED NON-REDEEMERS


Included in redemption rate


Figure 1. Typology of rebate purchases


OPPORTUNISTIC REDEMPTION INTENDED NON-REDEMPTION













Offer A
PUR RDN

.7 0




.0 .0




.2 .2


1.0 .0


Offer B
PUR RDN

.6 0






.2 .1



.2 .2


1.0 .0


Offer C
PUR RDN

.4 0O






.4 .2



.2 .2


1.0 .0


Classification


Segment


Intended
Non-Redeemers


Opportunistic
IRedeemers

Probabilistic
Redeemers


Intended
1Redeemers


Rebate-independent
purchases
(Base sales)




Rebate-dependent
purchases
(Incremental sales)


Total


Incremental sales
as a percentage of
total sales

Redemption rate

Redemption rate
on incremental
sales

Breakage


40%


30%


75%


25%


60%


40%


67%


33%


30%


100%


0%


PUR represents each segment's contribution to total purchases.
RDN represents each segment's contribution to total redemptions


Figure 2. Impact of post-purchase outcomes on redemption rates















CHAPTER 3
CONCEPTUAL AND THEORETICAL FOUNDATIONS

One conceptual insight into understanding breakage is provided by Soman's (1998)

assertion that the decision to purchase a product offering a delayed incentive can be

independent of the decision to redeem the delayed incentive, and that the structure of a

delayed incentive offer encourages people to overweight the benefits (i.e., the reward),

and underweight the costs (i.e., the effort). Although it is questionable whether the

behavioral decision theory perspective of "weighting" is central to explaining breakage, I

adopt the perspective that purchase and redemption represent separate behaviors and

consider psychological mechanisms that influence each of these behaviors.

Explanations for breakage were obtained from the survey study conducted by

Jolson et al. (1987), the delayed incentives experiments (Soman 1998), and the numerous

cases of breakage reported in the popular press. A literature review was conducted to

identify relevant psychological theory that could explain how rebate offer characteristics

influence consumers' purchase and redemption behavior to create breakage. In

accordance with the definition of breakage, and as illustrated in figure 3, changes in

breakage rates may be attributed to three types of factors. First, there may be pre-

purchase factors that encourage consumers to make rebate-dependent purchases by

influencing consumers' subj ective probability of redeeming the rebate (i.e., redemption

confidence), but that have little influence on the obj ective (i.e., actual) probability of

redeeming the rebate. Second, there may be post-purchase factors that directly affect the

obj ective probability of redeeming a rebate by influencing the likelihood of initiating










and/or completing the redemption process. Third, there may be repeat-purchase factors

that affect how past experience with rebate redemption influences the likelihood of

making repeat rebate-dependent purchases.

Pre-Purchase Factors

Redemption Confidence

Research in psychology has demonstrated that people are often highly confident

when assessing the extent of their own knowledge, when assessing their traits and

abilities, and when making predictions about the future (see Alba and Hutchinson 2000

for a review of knowledge calibration). Most relevant for understanding rebate-dependent

purchase behavior is the tendency for people to be highly confident about the likelihood

of performing future tasks (e.g., Griffin, Dunning, and Ross 1990; Hoch 1985; Vallone,

Griffin, Lin, and Ross 1990). For example, when people are asked to provide a subj ective

probability estimate (i.e., a confidence estimate) of performing a future task, the average

subj ective probability estimate of respondents is often considerably higher than the

obj ective response rate (i.e., the proportion of respondents that actually perform the

behavior). This difference between confidence and the obj ective response rate is

overconfidence.

Two key findings in the overconfidence literature can inform us about rebate

purchase and redemption behavior. First, people are often willing to use their subj ective

probability estimates (i.e., confidence) to support their decisions to engage in a future

behavior (e.g., Vallone et al. 1990). This implies that influencing prospective buyers'

subj ective probability of redeeming a rebate (i.e., redemption confidence) may be an

effective means of encouraging rebate-dependent purchases. Second, high confidence is

often driven by contextual factors that increase people's confidence estimates above the










obj ective response rate. Put differently, contextual factors are capable of influencing

confidence estimates while having little effect on the obj ective probability of performing

the behavior (e.g., Griffin et al. 1990; Vallone et al. 1990). Stated formally,

H1: Increasing a consumer' s redemption confidence will increase the probability of
rebate-dependent purchase.

H2: IHCreaSing a consumer's redemption confidence will result in a less than
proportionate increase in the obj ective probability of redeeming the rebate.

A review of the overconfidence literature identified two contextual antecedents of

overconfidence that correspond to rebate offer characteristics and may encourage

rebate-dependent purchases: a) the desirability of future outcomes, and b) temporal

distance.


Desirability

Studies in overconfidence have demonstrated that people become more confident in

their personal forecasts as outcomes become more desirable (e.g., Pulford and Colman

1996). It is believed that the "wishful thinking" evoked by an optimistic bias toward a

desirable outcome may restrict or interfere with one's ability to consider situational

factors that could interfere with achieving the desired outcome (Griffin et al. 1990; Hoch

1985; Vallone et al. 1990). Applied to rebate-dependent purchase, desirability predicts

that redemption confidence will vary as a function of the attractiveness of the rebate

offer. Thus, as the size of reward increases (relative to the list price), so should the

consumer' s redemption confidence.

H3: IHCreaSing the size of the reward will increase
a. the consumer's redemption confidence.
b. the probability rebate-dependent purchase.










Temporal Distance

Overconfidence research has demonstrated that people become more confident in

their personal forecasts as the temporal distance (i.e., delay) between their prediction and

the occurrence of the predicted event increases (e.g., Gilovich, Kerr, and Medvec 1993;

Milburn 1978; Wright and Ayton 1992). The influence of temporal distance may be a

consequence of people' s failure to consider potential barriers to future success. For

example, research has shown that predictions made at a temporal distance generate more

reasons for why one may succeed than for why one may fail, with the pattern reversing

when people make predictions on the day of the task (Gilovich et al. 1993). This pattern

is consistent with findings generated by research on temporal construal (Liberman and

Trope 1998; Trope and Liberman 2000). Applied to rebate-dependent purchase, temporal

distance predicts that redemption confidence will vary as a function of the amount of time

between purchase and the deadline for redemption (i.e., the length of the redemption

period). Thus, as the length of the redemption period increases, so should the consumer' s

redemption confidence.

H4: IHCreaSing the length of the redemption period will increase
a. the consumer's redemption confidence.
b. the probability of rebate-dependent purchase.

Post-Purchase Factors

Post-purchase factors influence the consumer's likelihood of initiating and/or

completing the redemption process and directly influence the obj ective probability of

redeeming. Post-purchase factors include procrastination, prospective forgetting, and

redemption effort.









Procrastination

Procrastination occurs when individuals postpone tasks until tomorrow without

foreseeing that when tomorrow comes, the required action will be delayed yet again

(Akerlof 1991; Silver 1974). Procrastination has been reported to be pervasive within

everyday activities (Ellis and Knaus 1977) and among normal adult populations (Harriott

and Ferrari 1996). Furthermore, procrastination has been cited as a potential cause of

breakage (e.g., Jolson et al. 1987; Soman 1998), although it has yet to be formally

investigated in a rebate context. One explanation for task-induced procrastination is that

people avoid tasks that are perceived as more aversive or less attractive than alternative

uses of one' s time (Sigall, Kruglanski, and Fyock 2000). Research has demonstrated that

procrastinators make poor estimates about the amount of time needed to complete

activities (Buehler, Griffin, and Ross 1994; Lay 1988), commence tasks later than non-

procrastinators (Ferrari 1993; Pychyl, Morin, and Salmon 2000; Sigall et al. 2000), take

longer to complete tasks (Holmes 2002; Lay 1988), and fail to act on their intentions to

perform tasks (Lay and Bums 1991). Procrastination is believed to persist because people

(a) have a desire to avoid the activity, (b) make a decision to delay, (c) promise

themselves to get to it later, (d) engage in substitute diversionary activities, and (e) make

excuses to justify delays and exonerate themselves from blame (Knaus 2000).

Research shows that deadlines help to mitigate procrastination by facilitating the

completion of tasks. For example, Ariely and Wertenbroch (2002) observed that students

were willing to self-impose deadlines to overcome procrastination, but that performance

was higher when deadlines were externally imposed by the instructor than when self-

imposed by students. Research has also demonstrated that the longer people are given to

complete a task, the greater the likelihood of procrastination and failing to complete the









task. For example, Tversky and Shafir (1992) offered students $5 for answering and

returning a questionnaire within 5 days, 3 weeks, or within no deadline and observed

return rates of 60%, 42%, and 25% respectively. Applied to rebate redemption, these

findings predict that imposing immediate redemption deadlines will mitigate

procrastination and increase the obj ective probability of redeeming. Put differently,

imposing less immediate deadlines (i.e., increasing the length of the redemption period)

will foster procrastination and decrease the obj ective probability of redeeming.

Hg: Increasing the length of the redemption period will
a. increase procrastination.
b. decrease the obj ective probability of redeeming the rebate.

Prospective Forgetting

Prospective forgetting examines people's ability to retrieve and act on previously

formed intentions (e.g., Krishnan and Shapiro 1999; Marsh, Hicks, and Watson 2002;

Shapiro and Krishnan 1999). This research has shown that consumers frequently forget to

act on previously formed intentions and that successful completion of intentions is

contingent on a prospective memory component (i.e., remembering that an intention was

formed) and a retrospective component (i.e., remembering the content of the intention).

This research is relevant to rebate redemption because it has shown that delays that

separate the formation of an intention and an opportunity to act on the intention are likely

to foster forgetting (Shapiro and Krishnan 1999). Thus, the delay that is imposed between

the purchase and redemption of a rebate offer provides the opportunity for prospective

forgetting to play a role in breakage. Research has shown that increasing the length of the

delay can encourage forgetting (Scher and Ferrari 2000). Applied to rebate redemption,

prospective forgetting predicts that increasing the delay between purchase and










redemption (i.e., increasing the length of the redemption period) will increase the

likelihood of prospective forgetting and decrease the obj ective probability of redeeming.

H6: IHCreaSing the length of the redemption period will
a. increase prospective forgetting.
b. decrease the obj ective probability of redeeming the rebate.

Redemption Effort

Redemption effort (i.e., the effort required to successfully redeem a rebate) is often

cited as a reason why consumers fail to redeem rebate offers (e.g., Jolson et al. 1987;

McLaughlin 2002; Norr 2000; Shim 2002; Soman 1998; Spencer 2002). Redemption

effort can contribute to breakage by causing consumers to give up before completing the

redemption process and by preventing redemption when consumers make simple errors

such as discarding packaging or sales receipts that are required to redeem. In addition to

contributing to breakage, redemption effort can also lower redemption rates by increasing

rej section rates on submitted applications (i.e., failing to submit UPC codes). Interviews

with marketing service firms confirm that increasing redemption effort is a popular

strategy for encouraging breakage2. In fact, it is widely believed that redemption effort is

the greatest contributing factor in creating breakage. As explained by a former marketing

consultant, "If you have to take a knife and cut through heavy cardboard to get a bar

code, redemption rates drop precipitously" (Norr 2000, p. 21). As stated by Federal Trade

Commission Director J. Howard Beals, "Some companies are quick to offer attractive

rebates, but often make them so difficult to redeem that consumers simply give up"

SRejection rates typically range between 1% and 10% of submitted applications and can reach as high as
25%. Rejected rebate applications do not qualify as breakage because breakage is defined as failing to
submit a rebate application. Rejected applications are subtracted from the total number of applications
when calculating redemption rates.

2 Source: BDS Marketing, DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing
Association, TCA Fulfillment Services.










(Shim 2002, p.1). There is also experimental evidence showing that increasing

redemption effort increases task aversiveness and breakage (Soman 1998).

H7: Increasing redemption effort will decrease the obj ective probability of redeeming
the rebate.

Repeat-Purchase Factors

Repeat-purchase factors influence how past experience with rebate redemption

affect the likelihood of making repeat rebate-dependent purchases. Although there is

evidence to suggest that breakage is a frequent and recurring problem for some

consumers (e.g., Jolson et al. 1987), research has yet to examine whether people make

repeat rebate-dependent purchases after failing to redeem a previous offer.

Past Redemption Behavior

One factor which may influence repeat purchase behavior is the extent to past

redemption behavior (i.e., redeeming or failing to redeem a previous offer) influences a

consumer' s repeat purchase behavior. Successful redemption may encourage repeat

rebate-dependent purchases. Conversely, failing to redeem may discourage repeat rebate-

dependent purchases.

Hs: Failing to redeem a previous offer will decrease the probability of repeat rebate-
dependent purchase.















I I


REBATE-DEPENDENT PURCHASE


I I


I I


PROBABILISTIC REDEEMERS i


INTENDED REDEEMERS


1


POST-PURCHASE FACTORS
Procrastination
Prospective forgetting
Redemption effort


PRE-PURCHASE FACTORS
Redemption Confidence
a) Desirability
b) Temporal distance


REPEAT-PURCHASE FACTORS


Past redemption behavior


REBATE-INDEPENDENT PURCHASE


BREAKAGE


INTENDED REDEMPTION


Included in redemption rate


Figure 3. Factors that contribute to breakage


OPPORTUNISTIC REDEEMERS INTENDED NON-REDEEMERS


OPPORTUNISTIC REDEMPTION INTENDED NON-REDEMPTION















CHAPTER 4
INTRODUCTION TO EXPERIMENTS

Four experiments were conducted to examine consumers' purchase and post-

purchase behavior with a real rebate offer and observe how reward, redemption period,

and redemption effort influence purchase, procrastination, prospective forgetting,

redemption and breakage. The experiments isolate rebate-dependent purchases and

measure breakage by presenting consumers with an opportunity to purchase a rebate offer

that is unattractive relative to other purchase alternatives unless the consumer redeems

the rebate (i.e. the only reason to purchase the offer is to take advantage of the rebate,

thus eliminating rebate-independent purchases).

Experiment 1 examined how manipulating offer characteristics (i.e., reward,

redemption period and redemption effort) influences purchase, procrastination, breakage,

and redemption. Experiment 2 examined the relationship between consumers' redemption

confidence and the likelihood of purchase and redemption. Experiment 3 was conducted

with rebate buyers from experiment 1 and examined how past experience with rebate

redemption influences the likelihood of repeat rebate-dependent purchase. A post-test

survey was administered to participants of experiments 1 and 3 to obtain descriptive data

on rebate buyers' purchase and redemption behavior. Finally, experiment 4 was

conducted to assess the direct effect of reward on redemption behavior.















CHAPTER 5
EXPERIMENT 1

Design

Experiment 1 consisted of an opportunity to make a rebate-dependent purchase and

an opportunity to redeem the rebate. To maximize the external validity of the experiment,

consumers were shown the same information as is shown with a typical in-store rebate

promotion. The reward and redemption deadline were clearly indicated on the offer, but

no information regarding the redemption process (i.e., redemption effort) was provided.

As a result, the purchase phase of the experiment employed a two (reward: $6 or $9) by

three (redemption period: 1-day, 7-days, or 21-days) between-subjects design, producing

six experimental conditions. Consistent with typical rebate offers, consumers became

aware of the effort required to redeem upon initiating the redemption process. Thus, the

redemption phase of the experiment crossed the reward and redemption period factors

with an additional between-subj ects factor (redemption effort: low or high), resulting in

12 experimental conditions.

Procedure

Phase 1: Purchase

Twelve hundred and thirty-three students from an undergraduate subj ect pool

participated in the experiment for course credit and were randomly assigned to one of the

six purchase conditions, resulting in 205 or 206 participants per cell. A large sample size

was required since it was expected that a relatively small proportion of participants would

purchase the rebate offer, resulting in smaller sample sizes for the redemption analysis.










The purchase opportunity followed a bogus computer-admini stered market research

questionnaire that was used to disguise the purpose of the experiment and prevent

hypothesis guessing. Upon entering the experiment lab, participants were informed that a

well-known Hollywood movie studio had experienced considerable financial losses due

to movies that did poorly at the box office, and was conducting a pilot study to learn

about people's movie preferences. A cover story that was developed to disguise the study

and justify the need for some participants to return for a second and third session, which

would serve as experiment 3 and the post-test respectively. The cover story read as

follows:

"A2 well-known Hollywood movie studio is conducting this pilot study at five
colleges across the country to get a better idea of the tyipes of movies college
audiences want to see. You will be asked a series of questions about your movie
preferences, such as your favorite genres, actors, and movies. Your responses will
be aggrega~clted and used to develop 10 new movie concepts to be tested in two
follow-up studies to takettttt~~~~~~~tttttt place later this semester. Fewer people are needed for
the follow-up studies, so the computers will randomly select people at the end of
today 's study and ask them to participate in the follow-up studies. If you are
asked to participate, please indicate aI hehrl~ll or not you want to takettttt~~~~~~~tttttt part in the
two follow-up studies. Remember that selection is random, so please do not feel
badly if you are not selected.

Finally, the studio is offering pairs of discounted movie tickets to all participants
at the end of the study as a thank you for participating today. Simply indicate
which ticket offer you wish to purchase, or choose "decline if you do not wish to
purchase tickets. The tickets are validfor one year, and are validfor any movie
and show time at the local movie theatre. Please remember that purchase is
optional and is completely up to you. "

After the completing the bogus market research study, participants were presented

with the computer-admini stered purchase opportunity shown in figure 4. The choice set

consisted of three options: a) purchase two movie tickets for a discounted price of $11i, b)

purchase two movie tickets for $13 less a $6 (or $9) mail-in rebate, or c) decline both

offers and purchase no tickets. Purchasing the rebate offer represented a rebate-dependent










purchase since the offer was attractive only if redeemed (i.e., the $13 before-rebate price

that one pays if the rebate offer is not redeemed is greater than the $11 non-rebated offer

and the $12 regular price that students pay at the box onfce). Redemption period was

manipulated as one, seven, or 21 days from the date of purchase by changing the "rebate

must be redeemed by" date that was clearly marked on the rebate offer. The two ticket

offers were presented on the top and the middle of the computer screen and the placement

was randomized to guard against order effects.

After making their choice, participants were presented with a confirmation screen

that emphasized the Einancial consequences of their decision and allowed them to choose

a different option. For example, those who chose the rebate offer were shown the offer a

second time, accompanied by the message "You have chosen the rebate offer. You will

pay $13 today. Please confirm your choice." Following the choice confirmation,

participants who declined to purchase tickets or who chose the $11 non-rebated offer

completed a short computer-administered filler task while the computer asked

participants who chose the rebate offer to participate in the follow up studies to be

conducted later that semester. Once completed, participants who declined to purchase

tickets exited the room and those who chose the $11 offer paid for and received their

tickets and exited the room. Those who chose the $13 rebate offer paid for and received

their tickets. After making their payment, rebate buyers in each condition were randomly

assigned to either the low-effort or high-effort redemption effort condition and were

given a rebate purchase receipt that was required to redeem the rebate. This produced a

balanced number of rebate buyers in the high-effort and low-effort cells of each purchase

condition.










Phase 2: Redemption

The redemption procedure was designed to measure redemption, breakage, and the

extent to which rebate buyers procrastinated at each stage of the redemption process. This

was achieved by using a rebate redemption website which, unbeknownst to participants,

tracked each buyer's progress with the redemption process. To maintain external validity,

the rebate redemption website and redemption requirements were modeled after the

rebate redemption website of a well-known national electronics retailer. The purchase

receipt given to each rebate buyer listed the expiry date for mailing the rebate

applications and instructed the buyer to visit the rebate website listed on the receipt to

obtain the redemption form.

To redeem the rebate, rebate buyers were required to access the rebate website and

enter their receipt code. Participants in the low-effort condition were required to complete

five steps: 1) complete the form by typing their name, address, ticket numbers, receipt

code, and purchase date, 2) print the completed form, 3) complete the on-line customer

service survey, which served as a manipulation check of redemption effort, 4) photocopy

the UPC code on the back of both movie tickets, and 5) mail the redemption form,

original receipt and photocopies to the address listed on the redemption form prior to the

expiry date indicated on their purchase receipt (i.e., either 1, 7, or 21 days from the date

of purchase). Participants in the high-effort condition were required to complete the same

five steps as those in the low-effort condition, but were also required to read a terms and

conditions agreement and re-enter the information in step 1 for verification purposes,

effectively requiring step one to be completed twice.


SThe expiry date printed on the purchase receipt matched the "must be redeemed on or before" date of the
rebate offer shown in the choice set.









Phase 3: Responding to Rebate Applications

All rebate applications were examined to ensure that they conformed to the

redemption requirements listed on the website. Applicants who successfully redeemed

prior to the deadline were accepted and received a rebate check ($6 or $9) with a brief

letter stating that their application had been accepted. All acceptance letters were mailed

five business days after the postmark date of the application to standardize the response

time across participants. Applicants who submitted their application prior to the deadline

but who failed to include either the original purchase receipt or the photocopy of the UPC

code received a letter informing them that they had been rej ected and inviting them to re-

apply with the required materials prior to the deadline indicated on their purchase receipt.

All rej section letters were mailed two business days after the postmark date of their

application to standardize the response time and maximize participants' chances for re-

application. Applicants who submitted their requests after the deadline received a letter

indicating that their request had been rej ected since it was postmarked after the deadline.

Dependent Measures

Purchase

Consumers' choice of ticket offer (rebate offer / non-rebated offer / decline) was

recorded and analyzed using multinomial logistic regression.

Breakage and Redemption

Login (yes / no) measured whether rebate buyers initiated the redemption process

by logging on to the website and provided a means of measuring the degree of breakage

attributable to failing to initiate the redemption process. Applied (yes / no) measured

whether rebate buyers completed the redemption process by mailing a rebate application

and provided a means of measuring the degree of breakage attributable to failing to










complete the redemption process. A third measure, accepted (yes / no), measured whether

the rebate application conformed to the redemption requirements and provided a means

of measuring the overall redemption rate as well as the rej section rate on rebate

applications.2 Login and applied are of most interest since breakage is defined as failing

to apply for the rebate. There are no a-priori hypotheses regarding errors in rebate

applications that could affect redemption and rej section rates. Login, applied and accepted

were analyzed using binary logistic regression.

Procrastination

Login delay measured procrastination associated with initiating the redemption

process by measuring the delay (in days) between the purchase date and the date on

which the buyer logged on to the rebate website. Mail delay measured procrastination

associated with completing the redemption process by measuring the delay (in days)

between logging on to the website and the postmark date of the mailed rebate application.

Login delay and mail delay were analyzed using ANOVA.

Redemption Effort Manipulation Check

Three measures of redemption effort were obtained to determine the efficacy of the

redemption effort manipulation. After completing and printing the redemption form on

the rebate website, rebate buyers responded to two "customer service" questions on a

seven-point semantic differential scale: "Using this website was": 1 = easy / difficult;

and 1 = not at all effortful / 7 = extremely effortful. A third measure was obtained by

tracking the elapsed time each participant spent on the website. A MANOVA revealed a

significant effect of redemption effort on the difficulty rating (Mzow efor, = 2.08, May~,i effort


2 The discrepancy between "applied" and "accepted" is caused by rejected (late or incomplete) rebate
applications.










= 3.65) (F (1, 137) = 21.34, p < .001), effort rating (Most- efOrt = 2.86,,11as,i efort = 4.75) (F

(1, 137) = 29.92, p < .001), and time spent on the website (2ost- go;.,= 4.9 minutes, 2 11,,g.,

sort, = 9.0 minutes) (F (1, 137) = 59.33, p < .001). Reward, redemption period and all

interactions were not significant for all three measures (all p's >.40), confirming that the

reward and redemption period manipulations did not influence perceptions of redemption

effort.

Results

Purchase

The purchase results for experiment 1 are shown in Eigure 5. A higher proportion of

consumers purchased the rebate offer when the reward increased from $6 to $9 and when

the redemption period increased from 7 to 21 days. Multinomial logistic regression3

indicated that the reward by redemption period interaction was not significant (X2 = 1.40,

p > .80). Further analysis revealed significant main effects of reward (X2 = 15.11i, p <

.001) and redemption period (X2 = 12.82, p < .05). The odds ratio for reward indicates that

the estimated odds of purchasing the rebate offer are 1.9 times higher when the rebate is

$9 than when it is $6. Similarly, the odds ratio for redemption period indicates that the

estimated odds of purchasing the rebate offer are 2.0 times higher when the redemption

period is 21 days than when it is either seven days or one day.






3 In accordance with the MNL model specification guidelines proposed by Ben-Akiva and Lerman (1985),
a nested logit model was specified before graduating to a MNL model to ensure that the MNL model was a
better predictor of purchase compared to the nested model, and to ensure that the error terms of the three
purchase alternatives satisfied the IID (independent and identically distributed) assumption. The nested
model specified "purchase" and "no purchase" as discrete alternatives, with the rebate and non-rebate
offers nested under "purchase."









Redemption

Redemption analyses were conducted on the 184 consumers who purchased the

rebate offer. Sixty-seven percent of rebate buyers applied for the rebate, resulting in a

breakage rate of 33%. Eleven percent of the applications (8% of rebate buyers) had their

rebate applications rej ected, resulting in a redemption rate of 59%. The proportion of

rebate buyers in each condition that applied for the rebate is shown in Eigure 6. Logit

analyses found no significant two-way or three-way interactions on the proportion of

buyers that logged on to the website, applied for the rebate, or were accepted. As

predicted, increasing the length of the redemption period decreased the proportion of

rebate buyers that applied for the rebate (X2 = 5.12, p < .05). Thus, increasing the length

of the redemption period has a positive influence on purchase behavior and a negative

influence on redemption behavior. Unexpectedly, increasing redemption effort generated

a marginally significant increase in the proportion of rebate buyers that applied for the

rebate (X2 = 3.03, p < .08). As predicted, increasing the reward did not have a significant

effect on the proportion of rebate buyers that logged on to the website (X2 = 0. 14, p >

.70), applied for the rebate (X2 = 0.01, p > .90), or were accepted (X2 = 0.50, p > .40). The

estimated odds of applying for the rebate are 2.4 times higher when the redemption

period is one day compared to 21 days, and 1.8 times higher when the redemption period

is seven days compared to 21 days, and 1.7 times higher when consumers experience the

high-effort redemption process compared to the low-effort redemption process.

Procrastination

Procrastination analyses were conducted on the 149 rebate buyers that initiated the

redemption process by logging on to the website and the 123 rebate buyers that

completed the redemption process by mailing a rebate application. The results for login










delay and mail delay are shown in figures 7 and 8 respectively. As predicted, increasing

the length of the redemption period significantly increased the extent to which rebate

buyers procrastinated before logging on to the website (F (2, 137) = 20.07, p < .001).

Planned contrasts revealed a significant difference between all levels of redemption

period (all p's < .01). As predicted, increasing the reward had no effect on login delay (F

(1, 137) = 0.23, p > .60). The effect of redemption effort on login delay was also non-

significant (F (1, 137) = 0.05, p > .80), which was expected since effort was not

manipulated until after rebate buyers logged on to the website.

With regard to mail delay, the ANOVA revealed a significant redemption period by

redemption effort interaction (F (2, 111) = 49. 16, p < .001). As shown in figure 8,

increasing the redemption period increased the extent to which rebate buyers in the low-

effort condition procrastinated before mailing the rebate application. In contrast, rebate

buyers in the high-effort condition tended to mail the rebate application within a few of

days of visiting the website, regardless of the amount of time that was left before the

redemption deadline. As with login delay, size of reward did not have a significant effect

on mail delay (F (1, 111) = 0.07, p > .80).

Discussion

Purchase and Redemption

Increasing the reward increased the number of rebate-dependent purchases,

supporting H3b, but did not influence redemption, consistent with H12. Increasing the

length of the redemption period from 7 days to 21 days also increased the number of

rebate-dependent purchases, supporting H4b, and any increase in the redemption period

increased procrastination and decreased the proportion of buyers that applied for the

rebate, supporting HS5. It appears that shorter redemption deadlines mitigate










procrastination and decrease breakage by facilitating the completion of redemption

applications that would otherwise go unfinished.

Insights Regarding Breakage

As discussed earlier, the challenges associated with isolating rebate-dependent

purchases has made it difficult to determine the extent to which breakage occurs and how

it is influenced by offer characteristics. The breakage rate of 33% indicates that breakage

is a common post-purchase outcome that can lower redemption rates in a significant

manner. As illustrated in figure 9, 19% of rebate buyers failed to log on to the website

and an additional 14% failed to mail their applications. Thus, failing to initiate the

redemption process accounts for about 58% of breakage. This finding is important

because increasing redemption effort does not influence this form of breakage (i.e., rebate

buyers did not become aware of the effort required until after logging on to the website).

In fact, in opposition to H7, increasing redemption effort caused rebate buyers to

complete the redemption process more quickly and generated a marginal increase in

rebate applications. This suggests buyers in the high effort condition may have been more

motivated to apply for the rebate in response to the high-effort redemption requirements.

Failing to mail redemption applications is also a significant problem for consumers,

and the breakage rate would have been greatly reduced had consumers been given the

option of completing the entire redemption process on-line (i.e., no mail-in requirement).





































4 Reward was manipulated as either $6 (as shown) or $9, resulting in an after-rebate price of either $7 (as shown) or $4. Redemption period was manipulated by
setting the "must be redeemed by" date to 1, 7, or 21 days after the date of purchase.


Figure 4. Purchase opportunity choice set4


Ilr~I3Y I~M












30%





25% 6-

=0%
1-Da 7-Dy 2-a









90%

i-D5% 7-a 7% Da


.$9 60% t 6 eb





50%



40%%



1-Day 7-Day 21-Day
Redemption Period

S$9 High effort $6 High effort
+ $9 Low effort $6 Low effort



Figure 6. Proportion of experiment 1 rebate buyers that applied for the rebate by
condition


























o.4,-
0.3~


1-Day


7-Day
Redemption Period


21-Day


S$9 High effort
+ $9 Low effort




Figure 7. Experiment 1 login delay by condition


14


$6 High effort
$6 Low effort


12.4

S12.0


5.8,'

,i 5.2

2.8


1-Day


7-Day
Redemption Period


21-Day


S$9 High effort
- + $9 Low effort


$6 High effort
$6 Low effort


Figure 8. Experiment 1 mail delay by condition












Buyers









/
N =184
(1 00%)


Login to Website







Yes =149
(81% of buyers)


Applied




Yes = 123

/ (67% ofbuyers)


Accepted

Yes = 109
(59% of buyers)




Rejected =14
(8% of buyers)


No = 26
(14% of buyers)


No = 35
(19% of buyers)


Breakage
Breakage = 19% Breakage = 14%
by stage

Cumulative
Total breakage = 19% Total breakage = 33%
breakage

Contribution
to breakage % of breakage = 58% % of breakage = 42%


Figure 9. Experiment 1 breakage by stage of redemption (n=1 84)















CHAPTER 6
EXPERIMENT 2

Although experiment 1 demonstrated how rebate offer characteristics influence

purchase, procrastination and redemption, it did not directly test the relationship between

consumers' subj ective probability of redeeming and the likelihood of purchase and

redemption as outlined in hypotheses H1-H4. Experiment 2 was designed to test this

relationship by replicating the procedure used in experiment 1 and measuring consumers'

subj ective probability of redemption In doing so, experiment 2 will determine the extent

to which consumers making rebate-dependent purchases are confident of redeeming, and

whether manipulating reward and redemption period influence redemption confidence.

Most importantly, experiment 2 will determine how the redemption confidence of rebate

buyers relates to their redemption behavior (testing H2), and whether redeemers are any

more or less confident of redeeming at the time of purchase than non-redeemers.

Experiment 2 was also designed to obtain consumers' base rate estimates of

successful rebate redemption (i.e., objective probability estimates). Although H1 predicts

that redemption confidence (i.e., the subj ective probability of redeeming) drives rebate-

dependent purchases, it would be informative to learn how consumers' redemption

confidence compares to their base rate estimates of successful redemption. In particular,

it would be understandable for consumers to be highly confident of redeeming if they

believe the base rate of redemption is high as well (i.e., a false consensus effect).


SSubjective probability estimates were not obtained in experiment 1 because doing so would have
compromised the external validity of the procedure.









However, if consumers' base rate estimates are relatively accurate, it would suggest that

high levels of redemption confidence stem from an overly optimistic view of one' s

personal probability of redeeming rather than inaccurate perceptions of the base rate of

redemption.

Design and Procedure

Experiment 2 used a design and procedure identical to that used in experiment 1

with two exceptions. First, the 7-day redemption period conditions were excluded from

the design in an effort to reduce the number of subjects required for the study. This

change was considered minor since retaining the 1-day and 21-day redemption period

conditions would accommodate the full range of effects observed in experiment 1. This

resulted in a two (reward: $6 / $9) by two (redemption period: 1-day / 21-day) by two

(redemption effort: low / high) between-subj ects design. As in experiment one, the

redemption effort manipulation was delayed until rebate buyers logged on to the website,

resulting in four purchase conditions. Four hundred subj ects were randomly assigned to

one of the four purchase conditions resulting in 100 subj ects per cell. The redemption

procedure used in experiment 2 was identical to that used in experiment 1.

The second exception to the experiment 1 design was the addition of two measures

administered to participants immediately following their purchase decision. The first

additional measure was a subj ective probability estimate adapted from Vallone et al.

(1990). Those purchasing the rebate offer were asked, "How confident are you that you

will redeem the rebate offer?" while those not purchasing the rebate offer were asked,

"How confident are you that you would redeem the rebate if you had purchased the

rebate offer?" Participants recorded their responses on a probability scale anchored by

"0%: I am completely certain that I will (would) not redeem the rebate" and "100%: I am









completely certain that I will (would) redeem the rebate." The second measure was a base

rate estimate. Participants were asked, "Please estimate the percentage of rebate buyers

that you think will redeem the rebate." Participants recorded their responses on a

probability scale anchored by "0%: None of the rebate buyers will redeem the rebate" and

"100%: All of the rebate buyers will redeem the rebate."

Results

Purchase

As illustrated in figure 10, the purchase results replicated the effects observed in

experiment 1. Multinomial logistic regression indicated that the reward by redemption

period interaction was not significant (X2 = 1.29, p > .50). Further analysis revealed a

significant main effect of reward (X2 = 7.30, p < .05) and a marginally significant main

effect of redemption period (X2 = 4.52, p < .10). The estimated odds of purchasing the

rebate offer are 2.8 times higher when the rebate is $9 than when it is $6, and 2.4 times

higher when the redemption period is 21 days than when it is one day.

Redemption

Redemption analyses were conducted on the 54 participants who purchased the

rebate offer. The results replicated the effects observed in experiment 1. Seventy-four

percent of rebate buyers applied for the rebate, resulting in a breakage rate of 26%. Eight

percent of the applications (6% of rebate buyers) had their rebate applications rej ected,

resulting in a redemption rate of 69%. The proportion of rebate buyers in each condition

that applied for the rebate is shown in figure 11. Logit analyses found no significant two-

way or three-way interactions on the proportion of buyers that logged on to the website,

applied for the rebate, or were accepted. As in experiment 1, increasing the length of the

redemption period resulted in a significant decrease in the proportion of rebate buyers









that applied for the rebate (X2 = 3.93, p < .05), and increasing redemption effort resulted

in a marginally significant increase in the proportion of rebate buyers that applied for the

rebate (X2 = 3.55, p < .06). Increasing the reward did not have a significant effect on the

proportion of rebate buyers that logged on to the website (X2 = 0. 17, p > .60), applied for

the rebate (X2 = 0. 14, p > .90), or were accepted (X2 = 0.03, p > .80). The estimated odds

of applying for the rebate are 5.9 times higher when the redemption period is one day

compared to 21 days, and 3.8 times higher when consumers experience the high-effort

redemption process compared to the low-effort redemption process.

Procrastination

Procrastination analyses were conducted on the 47 rebate buyers that logged on to

the website and the 40 rebate buyers that mailed a rebate application. The results for login

delay and mail delay are shown in figures 12 and 13 respectively. As in experiment 1,

increasing the length of the redemption period significantly increased the extent to which

rebate buyers procrastinated before logging on to the website (F (1, 39) = 21.53, p <

.001). As expected, the effects of reward (F (1, 39) = 0.01, p > .90) and redemption effort

(F (1, 39) = 0.02, p > .80) on login delay were not significant. As in experiment 1, a

significant redemption period by redemption effort interaction (F (1, 32) = 14.01, p <

.001) was observed for mail delay. As shown in figure 13, increasing the redemption

period reduced procrastination in the high-effort condition relative to the low-effort

condition. Size of reward did not have a significant effect on mail delay (F (1, 32) = 0.59,

p >.40).

Probability Estimates

Subj ective probability (i.e., confidence) and base rate estimates for experiment 2

are shown in figure 14. Results indicate that rebate buyers are highly confident of










redeeming at the time of purchase despite generating rather accurate estimates of the base

rate of redemption. Rebate buyers' confidence estimates (M~= 93.5%) ranged from 50%

to 100%, and over 60% of rebate buyers reported being 100% confident of redeeming the

rebate.2 The mean base rate estimate of rebate buyers (M~= 53.7%) was close to the 60%

redemption rate observed in experiment 1, and did not differ significantly from the base

rate estimates of decliners (M~= 5 1.0%) or of those who purchased the non-rebated offer

(M~= 55.6%), (F (2, 397) = 0.62, p > .50).

As shown in figure 15, increasing the size of reward significantly increased

consumers' subj ective probability of redeeming from a mean of 5 1.8% to a mean of

58.5%, (F (1, 396) = 4.28, p < .05), supporting H1 and H13. As shown in figure 16,

increasing the length of redemption period also significantly increased consumers'

subjective probability of redeeming from a mean of 51.5% to a mean of 58.8%, (F (1,

396) = 5.10, p < .05), supporting H1 and H14. Increasing reward (F (1, 396) = 3.50, p <

.06) and redemption period (F (1, 396) = 3.21, p < .08) resulted in marginally significant

increases in consumers' base rate estimates.

With regard to the relationship between redemption confidence and redemption,

binary logistic regression revealed that redemption confidence did not have a significant

effect on whether rebate buyers logged in, (P = -.02, X2 = 0. 14, p > .70), applied (P = -.08,

X2 = 2.08, p > .10) or were accepted (P = -. 10, X2 = 0. 10, p > .70), supporting H12. In fact,

as shown in figure 17, non-redeemers were just as confident at the time of purchase (M~=

94. 1%) as redeemers (M~= 93.2%), (F (1, 52) = 0.09, p > .70). Base rate estimates were


2 The mean confidence estimate of rebate buyers (M~= 93.5%) was significantly higher than that of those
who declined the offer (M~= 49.6%) and who purchased the non-rebated offer (M~= 38.5%), (F (2, 397) =
55.00, both p's < .001), and multinomial logistic regression confirmed that confidence has a significant
positive effect on the likelihood of purchasing the rebate offer (13 = .10, X2 = 36.94, p < .001).









also similar of redeemers (M~= 54.0%) and non-redeemers (M~= 53.1%), (F (1, 52) =

0.04, p > .80). Finally, a median split of rebate buyers based on their confidence estimates

revealed similar redemption rates for the higher-confidence (67%) and lower-confidence

(70%) groups (X2 = 0.09, p > .70).

Discussion

In addition to replicating the results of experiment 1, experiment 2 provides direct

evidence of the relationship between consumers' redemption confidence and their

purchase and redemption behavior. As predicted by H1, H3 and H4, increasing reward

and redemption period increases redemption confidence and the probability of rebate-

dependent purchase. However, as predicted by H2, increasing redemption confidence

does not generate a proportionate increase in the obj ective probability of redeeming.

Non-redeemers were just as confident of redeeming as redeemers, revealing that some

poorly calibrated consumers fall victim to redemption overconfidence by making rebate-

dependent purchases that they are unlikely to redeem. Interestingly, buyers are highly

confident despite rather accurate base rate estimates. Although it is questionable whether

consumers consider base rates during purchase, the results at least suggests that high

redemption confidence does not stem from a false consensus effect. Thus, it is possible

that rebate buyers exhibit a "better than average" effect in believing that their personal

probability of redeeming is greater than that of others.

















































S70%
c
F io


1-Day 21-Day
Redemption Period

$$9 Rebate $$6 Rebate
$$11 Non-Rebate --$11 Non-Rebate



Figure 10. Proportion of experiment 2 participants purchasing rebate and non-rebate
offers by condition


100% ~- 10096


90%83


1155%
S50% 5~

40%
1-Day 21-Day
Redemption Period

S$9 High effort $6 High effort
5- $9 Low effort $6 Low effort



Figure 11i. Proportion of experiment 2 rebate buyers that applied for the rebate by
condition


12%


7 3%b
67%


43


30%

S25%



1 0%

t~ 5%

10%











































# 5.5


03


1-Day


21-Day


Redemption Period


S$9 High effort
- $9 Low effort


$6 High effort
$6 Low effort


Figure 12. Experiment 2 login delay by condition


8


2oY


1-Day


21-Day


Redemption Period


S$9 High effort
- + $9 Low effort


$6 High effort
$6 Low effort


Figure 13. Experiment 2 mail delay by condition













100-
93.5

90-

3 80-




S60 37 55.6
4 9.6-51
S503 -5


30

Confidence Estimate Base Rate Estimate

O Rebate Buyers O Decliners Regular Buyers



Figure 14. Experiment 2 confidence and base rate estimates by choice


100-

90-

a, 80-



60 58.5
2 1. 53.4
g 49.7
a. 50-

40-

30
Confidence Estimate Base Rate Estimate

O $6 O $9



Figure 15. Experiment 2 confidence and base rate estimates by reward



















51.5


Confidence Estimate Base Rate Estimate

01-Day 021-Day


Figure 16. Experiment 2 confidence and base rate estimates by redemption period


93.2 94.1


54.0 53.1


Confidence Estimate Base Rate Estimate

O Redeemers O Non-Redeemers


Figure 17. Experiment 2 confidence and base rate estimates of redeemers and non-
redeemers


53.3


49.8















CHAPTER 7
EXPERIMENT 3

Although there is evidence to suggest that breakage is a recurring problem for some

consumers (e.g., Jolson et al. 1987), this implies that rebate buyers do not learn from past

experience and/or minimize the relevance of past redemption failures when considering

the purchase of a rebate offer. Research has yet to experimentally investigate the repeat

purchase behavior of rebate buyers to determine the extent to which breakage is a

recurring outcome or to identify the conditions that lead to repeat instances of breakage.

Experiment 3 was designed to address these questions.

Design and Procedure

Approximately five weeks after completing experiment 1, the 184 rebate buyers

from experiment 1 were assigned to the same experimental conditions they experienced

in experiment 1 and were provided with another opportunity to purchase and redeem the

same offer. The experimental procedure and analyses were the same as in experiment 1,

with four exceptions. First, the bogus market research survey that preceded the purchase

opportunity consisted of new questions about movie preferences to maintain the face

validity of the cover story. Second, participants were now aware of the redemption

requirements if they had attempted to redeem the rebate in experiment 1 (i.e., rebate

buyers that did not log on to the website in experiment 1 remained unaware of the

redemption requirements). Third, those declining repeat purchase were asked why they

did not purchase the rebate offer. Finally, the analysis included an additional predictor









variable, past redemption behavior, which indicated whether the rebate buyer redeemed

the rebate in experiment 1.

Results

Repeat Purchase

As shown in figure 18, the purchase results closely resemble the redemption results

of experiment 1. Multinomial logistic regression found no significant interactions on the

proportion of buyers making a repeat purchase. Further analysis revealed significant main

effects of past redemption behavior (X2 = 32.71, p < .001) and redemption effort (f =

9.88, p < .01). The estimated odds of repeat purchase are 10.8 times higher for buyers

that redeemed the offer in experiment 1 than for buyers that did not redeem, and 3.2 times

higher for buyers that experienced the low-effort redemption process than for buyers that

experienced the high-effort redemption process. Thus, failing to redeem the offer and

experiencing high-effort redemption requirements drastically reduced the likelihood of

repeat purchase. In fact, only 8% of non-redeemers made a repeat purchase versus 41%

of redeemers. The self-reported reasons for not making a repeat purchase were analyzed

using multinomial logistic regression and are summarized in table 1. Past redemption

behavior was the only factor to have a significant effect on the reasons given for not

making a repeat purchase (X2 = 55.28, p < .001). Failing to redeem the previous offer was

the most common response given by non-redeemers who declined repeat purchase.

Redemption

Redemption analyses were conducted on the 51 rebate buyers that made repeat

purchases. Eighty-four percent of rebate buyers applied for the rebate, resulting in a

breakage rate of 16%, well below the rates of breakage observed in experiments 1 (33%)

and 2 (26%). Four percent of the applications (5% of repeat rebate buyers) had their









rebate applications rej ected, resulting in a redemption rate of 80%, well above the

redemption rates observed in experiments 1 (59%) and 2 (69%). Despite the higher

redemption rate and lower rate of breakage, the redemption results illustrated in figure 19

are consistent with the effects observed in experiments 1 and 2. Logit analyses found no

significant interactions on the proportion of buyers that logged on to the website, applied

for the rebate, or were accepted. Increasing the length of the redemption period decreased

the proportion of rebate buyers that applied for the rebate (X2 = 4.84, p < .05), and

increasing redemption effort generated a marginally significant increase in the proportion

of repeat buyers that applied for the rebate (X2 = 2.59, p < .10). Increasing the reward did

not have a significant effect on the proportion of rebate buyers that logged on to the

website (X2 = 0.01, p > .90), applied for the rebate (X2 = 0.37, p > .50), or were accepted

(X2 = 0.48, p > .40). The estimated odds of applying for the rebate are 5.1 times higher

when the redemption period is one day compared to 21 days, 2.9 times higher when the

redemption period is seven days compared to 21 days, and 6.3 times higher when

consumers experience the high-effort redemption process compared to the low-effort

redemption process.

Procrastination

Procrastination analyses were conducted on the 46 repeat buyers that initiated the

redemption process by logging on to the website and the 43 repeat buyers that completed

the redemption process by mailing a rebate application. The results for login delay and

mail delay replicated the effects observed in the previous experiments and are shown in

figures 20 and 21 respectively. Increasing the length of the redemption period

significantly increased the extent to which rebate buyers procrastinated before logging on

to the website (F (2, 34) = 18.79, p < .001). Planned contrasts revealed a significant









difference between the seven-day and 21-day redemption periods (p < .001), but the

difference between the one-day and seven-day redemption periods was not significant (p

> .10). Interestingly, redemption effort did not have a significant effect on login delay

despite the vast maj ority of repeat buyers being aware of the redemption requirements (F;

(1, 34) = 1.29, p > .20). Aversiveness to effort would predict repeat buyers in the high-

effort condition to procrastinate longer than those in the low-effort condition, yet this was

not the case. The effect of reward on login delay was not significant (F (1, 34) = 0.22, p >

.60). Analysis of mail delay revealed a significant redemption period by redemption

effort interaction (F (2, 31) = 7.04, p < .01). Consistent with the results of experiments 1

and 2, increasing the redemption period reduced procrastination in the high-effort

condition relative to the low-effort condition. As before, size of reward did not have a

significant effect on mail delay (F (1, 31) = 0.06, p > .80).

Repeat Breakage

The opportunity to observe repeat instances of breakage was limited by the small

proportion of non-redeemers that made a repeat purchase (only six of the 75 non-

redeemers from experiment 1 made a repeat purchase). However, repeat breakage was

observed for three of the six non-redeemers that made a repeat purchase. Two of the three

repeat non-redeemers failed to log on to the website on both occasions, while the third

logged on to the website but did not mail an application on both occasions. All three

repeat non-redeemers were in the low-effort, 21-day redemption period condition, and

two had purchased the $9 rebate offer. Given the small sample of repeat non-redeemers,

valid inferences cannot be made regarding the conditions that increase the likelihood of

repeat breakage. However, self-reports obtained from the post-test reveal that two repeat

non-redeemers procrastinated and eventually forgot to log on to the website on both










occasions. The third non-redeemer did not realize the redemption form had to be

submitted by mail on the first occasion, and lost the purchase receipt required for

redemption on the second occasion. Thus, although it is hard to estimate the scope of

repeat breakage, it certainly is a problem for some consumers.

Discussion

The results of experiment 3 suggest that rebate buyers avoid offers that are known

to require a high degree of effort and offers that they failed to redeem in the past, and that

moderate rates of breakage can still occur among successful redeemers that make repeat

rebate-dependent purchases. The breakage rate among repeat buyers that redeemed in

experiment 1 was 12%, and increasing the redemption period significantly increased

procrastination and breakage despite these buyers having full knowledge of the

redemption process. The low rate of repeat purchase among non-redeemers is

understandable considering that experiment 3 was conducted within five weeks of

experiment 1 and involved the same offer. The rate of repeat purchase and breakage

among non-redeemers may have been greater if more time had elapsed between the two

experiments or if the second offer had been more attractive than the first offer.

Nonetheless, experiment 3 represents a conservative test of repeat breakage and suggests

that repeat breakage is limited by the low probability of repeat rebate-dependent purchase

among non-redeemers.




















































` 4 71%
67%





1-Day 7-Day 21-Day
Redemption Period


-

-


~25% 25%


1 3%


S40%


2 0%

1 %
20%


4.44%


1-Day


7-Day
Redemption Period


21-Day


S$9 High effort
- + $9 Low effort


$6 High effort
$6 Low effort


Figure 18. Proportion of rebate buyers making a repeat purchase of the rebate offer in
experiment 3 by condition


00o%


100%


100%


100%


S$9 High effort
- + $9 Low effort


$6 High effort
$6 Low effort


Figure 19. Proportion of experiment 3 rebate buyers that applied for the rebate by
condition








53




10














1-Day 7-Day 21-Day
Redemption Period

S$9 High effort $6 High effort
- $9 Low effort $6 Low effort



Figure 20. Experiment 3 login delay by condition


10
9.0

-7-3...-.- 7.0
~~6.5








0 0 50 5
1-Day 7-Day 21-Day
Redemption Period

S$9 High effort $6 High effort
+ $9 Low effort $6 Low effort


Figure 21. Experiment 3 mail delay by condition








54






Table 1. Self-reported reasons for not making a repeat rebate purchase in experiment 3
Non- Total
Reason Rde rs Redeemers Responses
n =64n =69 n =133

I still haven't used the tickets I bought last time 54.7% 13.0% 33.1%

I didn't redeem last time 0.0% 47.8% 24.8%

It's too much workleffort to redeem the rebate 21.9% 23.2% 22.6%

I don't have the money / can't afford it 15.6% 10.2% 12.8%

I don't go to the movies that often 4.7% 2.9% 3.7%

Other 3.1% 2.9% 3.0%

Total 100% 100% 100%















CHAPTER 8
POST TEST SURVEY

Experiments 1, 2 and 3 were not able to separate the effects of procrastination and

prospective forgetting on breakage, nor were they able to identify other potential factors

that may have interfered with redeeming. The post-test survey was designed to address

these issues. The survey was administered to the 184 rebate buyers that participated in

experiments 1 and 3 approximately five weeks after the study 3 purchase opportunity.

Rebate buyers responded to items that solicited retrospective descriptions of the purchase

decision and efforts to redeem the rebate, as well as items that measured redemption

confidence and base rate estimates, perceived fairness of the rebate offer, actual

redemption effort relative to expectations of redemption effort at the time of purchase,

attitudes toward rebate offers, and the likelihood of purchasing a rebate offer in the

future. In addition to these items, non-redeemers (i.e., rebate buyers that failed to redeem

the rebate in experiment 1 and/or experiment 3) explained why they did not redeem the

rebate offer. The post-test survey items are listed in the appendix.

Results

All 184 rebate buyers (112 redeemers and 72 non-redeemers) completed the post-

test. Ninety-eight percent of participants believed that the series of studies was about

measuring movie preferences, indicating that the cover story was successful. One-

hundred percent of participants indicated that they had purchased the rebate offer to

obtain the rebate, confirming that the procedure was successful in isolating rebate-

dependent purchases.









Reasons for Breakage

Non-redeemers' self-reported reasons for breakage are summarized in table 2.

Procrastination and forgetting accounted for a greater proportion of breakage (59.7%)

than redemption effort (20.8%) and procedural constraints (19.5%) combined. As

expected, redemption effort was a factor only among those who initiated redemption,

while procrastination interfered with initiating anzd completing redemption. These results

are consistent with the procrastination and redemption patterns observed in the

experiments and confirm that procrastination can have a greater influence on breakage

than procedural constraints and redemption effort. One explanation for the strong

influence of procrastination on breakage is that procrastination invites a multitude of

other factors to interfere with redemption. Responses to open-ended questions about why

participants failed to redeem revealed that those who procrastinated maintained the

intention to redeem the rebate, but either engaged in diversionary activities or were

distracted by other daily tasks and commitments. This behavior is consistent with Knaus'

(2000) self-perpetuating explanation of procrastination. It is also worth noting that

although procrastination leads to forgetting in some cases, this accounted for only 5.6%

of total breakage.

With respect to the relative influences of procrastination and prospective forgetting,

the self reports suggest that procrastination contributes more to breakage than prospective

forgetting. Prospective forgetting accounted for 22.5% of the breakage among rebate

buyers that failed to initiate the redemption process, but did not contribute to breakage

among those who initiated the redemption process. Apparently, those who forgot to

redeem did so early in the process and consequently "did nothing" after purchase. Further

analysis revealed that the proportion of respondents who reported forgetting to redeem as









the reason for breakage increased with the length of the redemption period, although this

difference was not significant. Thus, there is not sufficient evidence to support H6. One

explanation for the relatively weak influence of prospective forgetting is that the purchase

receipt may have served as an effective memory cue which reminded rebate buyers of

their intention to redeem the rebate.

Redeemers vs. Non-redeemers

Response frequencies to the items administered to redeemers and non-redeemers

are summarized in table 3. Consistent with the findings of experiment 2, retrospective

confidence estimates confirmed that rebate buyers were highly confident of redeeming at

the time of purchase (M~= 93.3%) and that redeemers (M~= 93.5%) and non-redeemers

(M~= 92.9%) were equally confident of redeeming (F (1, 170) = 0.06, p > .80). Consistent

with other studies on overconfidence in personal forecasts (e.g., Hoch 1985), the maj ority

(67%) of rebate buyers did not consider factors that could have potentially interfered with

redeeming. The proportion of rebate buyers that considered such factors was similar for

redeemers (P = 32.8%) and non-redeemers (P = 32.8%), (X2 = 0.01, p > .90), which is

understandable given the similarity in their confidence estimates. As argued in the

overconfidence literature (e.g., Griffin and Varey 1996; Hoch 1985; Pulford and Coleman

1996; Vallone et al.1990), failing to consider factors that could interfere with achieving a

desired outcome is likely to contribute to overconfidence.

Perceived fairness of the rebate offer was measured by asking rebate buyers to

indicate whether they perceived the offer as "fair" or "deceptive". The proportion of

rebate buyers that perceived the rebate offer as "fair" was significantly higher for

redeemers (P = 90.4%) than for non-redeemers (P = 76.1%), (X2 = 6.43, p < .05). Further

analysis revealed that the difference between redeemers and non-redeemers can be









attributed to rebate buyers whose rebate applications were rej ected for not complying

with the terms and conditions of the offer (i.e., late or incomplete submissions).

Changes in attitudes toward rebate offers were measured by asking rebate buyers to

indicate their general attitude toward rebates before and after participating in the study.

Rebate buyers were also asked to estimate their probability of purchasing a rebate offer in

the future. Changes in attitudes and the probability of future purchase were analyzed

using MANOVA. The analysis tested the effects of reward, redemption period,

redemption effort, redemption behavior (redeeming/not redeeming the offer), and relative

redemption effort (a measure of the actual redemption effort relative to rebate buyers'

expectations at the time of purchase). As shown in table 3, changes in attitudes (F (1,

170) = 14.33, p < .001) and the probability of future purchase (F (1, 170) = 4.89, p < .05)

were significantly different for redeemers and non-redeemers. Relative redemption effort

also had a significant negative effect on changes in attitudes (F (1, 170) = 34.06, p <

.001) and the probability of future purchase (F (1, 170) = 21.15, p < .001). In summary,

failing to redeem the rebate and redemption procedures that are more effortful than

expected can have negative effects on consumers' attitudes and purchase intentions.

These findings are consistent with the repeat purchase effects observed in experiment 3,

and provide support for H18.

Non-Redeemers

Response frequencies to the items administered exclusively to non-redeemers are

summarized in table 4. Non-redeemers were asked whether they considered themselves

"better off", "no better or worse off', or "worse off' for not redeeming the rebate. Only

35.9% of non-redeemers considered themselves "worse off." Multinomial logistic

regression revealed a significant effect of redemption effort on responses (X2 = 6.39, p <










.05), and the estimated odds of being "worse off' were 3.9 times higher for those in the

high-effort condition compared to those in the low-effort condition. In explaining their

responses, many non-redeemers mentioned that they still benefited from their purchase

(i.e., seeing a movie) despite paying a higher price. This suggests that some non-

redeemers may use product utility or product satisfaction as a post-hoc justification for

failing to redeem a rebate offer.

Non-redeemers were also asked which behavior, "purchasing the rebate offer" or

"not redeeming the rebate offer", was their bigger mistake. Interestingly, the vast

maj ority (68.8%) regarded not redeeming as their bigger mistake. This implies that the

maj ority of rebate buyers believe that their purchases were justified and that the main

culprit in breakage is faulty post-purchase behavior. In contrast, the results of experiment

2 suggest that the main culprit in breakage is redemption overconfidence that leads

consumers to purchase rebate offers that they are unlikely to redeem. Thus, most non-

redeemers appear to be unaware of their redemption overconfidence or that redemption

overconfidence contributes to breakage.

Finally, non-redeemers were asked whether failing to redeem the rebate was their

fault or the fault of the rebate offer. Seventy percent of non-redeemers attributed their

redemption failure to themselves. Binary logistic regression revealed a significant effect

of relative redemption effort on responses (X2 = 9.02, p < .01), indicating that non-

redeemers are more likely to attribute their redemption failure to the rebate offer when

redeeming the rebate requires more effort than is expected at the time of purchase.












Table 2. Self-reported reasons for breakage by non-redeemers

Buyers who Buyers who
did not initiate did initiate Total
Reason
redemption redemption Responses
n=40 n=32 n=72

Procrastination and forgetting:


1 procrastinated and the deadline passed before I applied

I procrastinated and eventually forgot to apply

I completely forgot to apply

Redemption effort & deciding to not redeem:

I tried to redeem but it was too much work so I gave up

I decided the money saved was not worth the effort
required

The money saved didn't seem as valuable when it came
time to redeem

Procedural constraints:

I lost my receipt / tickets and could not apply

I used the tickets before I could photocopy them and could
not apply

Total


50.0%

10.0%

12.5%


43.8%


47.2%

5.6%

6.9%



12.5

8.3


28.1%

18.8%


10.0%

17.5%


100%


5.6

13.9


100%


9.4%


100%


Table 3. Post-test responses by redeemers and non-redeemers

Redeemers Nn
Item n =112 Redeemers Test Result
n =72


93.5% 92.9% 0 0 ,7> 80


63.5% 62.7% 0 0 ,7> 90

32.8% 33.0% X2 =0.01, p>.90

90.4% 76.1% X2 =6.43, p<.05

-.08 -26.0 14 3 < 001


81.0% 70.6% 8 < ,105


Mean subjective probability of redemption (redemption confidence):


Mean base rate estimate for redemption:

Proportion that foresaw problems that could interfere with redeeming:

Proportion indicating the rebate offer was a fair offer:

Change in attitude toward rebates (attitude before attitude after):


Probability of purchasing a rebate offer in the future:








61



Table 4. Post-test responses by non-redeemers only
Item

How would you categorize yourself as a result of buying the rebate offer?



Which behavior do you regard as the bigger mistake?


All things considered, not redeeming the rebate was:


Non-Redeemers n = 72

Better off: No better Worse off:
21.9% or worse 35.9%
off: 42.2%

Buying the rebate Not redeeming:
31.3% 68.8%

The offer's fault: My fault:
29.7% 70.3%















CHAPTER 9
EXPERIMENT 4

Experiments 1 and 2 demonstrated that increasing reward has a stronger influence

on purchase behavior than redemption behavior. However, since participating in the

redemption phase of the experiments was conditional on purchasing the rebate offer, it is

possible that the weak (i.e., null) effect of reward on redemption behavior is an artifact of

self-selection bias (i.e., those who did not purchase the rebate offer may have responded

differently than rebate buyers had they participated in the redemption phase of the

experiments). It is important to note that any self-selection that may have occurred in the

previous experiments also exists in the marketplace since the opportunity to redeem is

always conditional on purchase. Thus, any threat posed by self-selection cannot affect the

external validity of the results observed in the previous experiments. However, to make

the assertion that manipulating reward exerts a relatively weak influence directly on

redemption behavior, an alternative procedure was developed which excluded the

purchase phase used in the in the previous studies and randomly assigned participants

directly to the two reward conditions of the redemption phase of the procedure.

Design and Procedure

Experiment 4 used a simple two-cell between-subj ects design (reward: $6 or $9).

Redemption period was fixed at 21 days and redemption effort was fixed using the low-

effort redemption procedure used in the previous experiments. One hundred and thirty-

four students from an undergraduate subject pool participated in the experiment for

course credit and were randomly assigned to either the $6 or $9 reward condition,










resulting in 67 participants per cell. Upon entering the experiment room, participants

were informed of the market research survey on movie preferences and told that they

would inherit a rebate offer coupon from a previous marketing campaign after completing

the survey. The following cover story was used to introduce the market research survey

and explain why the rebate coupons were being distributed:

"Earlier this summer, to kick-off the start of the "summer blockbuster season, a
well-known cinema chain launched a rebate promotion that was later
discontinued after learning that the promotion was unprofitable due to the
unexpectedly high number of rebate redemptions. However, due to a recent FTC
ruling outlawing the premature cancellation of rebate promotions once they are
launched, the cinema chain is obligated to distribute the remaining rebate
coupons to the general public and mail rebate checks to those who redeem the
rebate. To kill two birds nI ithr one stone, the cinema has decided to use this
opportunity to learn more about the tyipes of movies college audiences want to
see, and to distribute the remaining rebate coupons. After completing the market
research survey on movie preferences, you will receive a rebate coupon that you
are free to redeem for the cash reward prior to the redemption deadline stated on
the coupon. "

Following the completion of the bogus market research survey, rebate coupons

were randomly distributed to participants and the participants exited the room. The

redemption procedure that followed was identical to that used in experiments 1, 2, and 3,

with the exception that the UPC code was printed on the rebate coupon since movie

tickets were not used in the study.

Dependent Measures and Analysis

With the exception of the purchase dependent measure which was omitted from the

study, experiment 4 used the same dependent measures as the previous experiments.

Redemption was analyzed using binary logistic regression and procrastination was

analyzed using ANOVA.









Results

Redemption

Redemption analyses were conducted on the 134 participants that received rebate

coupons. Approximately 7.5% of coupon holders applied for the rebate, resulting in a

breakage rate of 92.5%. Thirty percent of the applications (2.3% of coupon holders) had

their rebate applications rej ected, resulting in a redemption rate of 5.2%. The proportion

of coupon holders in each condition that applied for the rebate is shown in figure 22.

Consistent with the results of the previous experiments, logit analyses revealed that

increasing the reward did not have a significant effect on the proportion of coupon

holders that logged on to the website (X2 = 0.06, p > .80), applied for the rebate (X2 = 0.00,

p = 1.00), or were accepted (X2 = 0. 15, p > .70).

Procrastination

Procrastination analyses were conducted on the 21 coupon holders that logged on

to the website and the 10 coupon holders that mailed a rebate application. The results for

login delay and mail delay are shown in figures 23 and 24 respectively. Consistent with

the results of the previous experiments, increasing the reward did not have a significant

effect on the extent to which coupon holders procrastinated before logging on to the

website (F (1, 19) = 1.05, p >.30) or mailing the rebate application (F (1, 8) < 1, p >.80).

Discussion

Although the null effect of reward on redemption behavior observed in experiments

1, 2 and 3 was replicated in experiment 4, the prudent interpretation is that the results

were inconclusive regarding the direct effect of reward on redemption behavior due to the

extremely low response rate (i.e., only 21 of the 134 coupon holders initiated the

redemption process and only 10 of the coupon holders applied for the rebate). One










explanation for the low response rate is that excluding the purchase phase of the

procedure effectively eliminated the up-front financial investment that usually

accompanies the purchase of a rebate offer, thereby diminishing participants' motivation

to redeem the rebate coupon. Alternatively, participants may not have believed that they

were being given an opportunity to receive a cash reward with "no strings attached." The

failure of experiment 4 to produce conclusive results dictates that claims cannot be made

regarding the direct influence of reward on redemption behavior. Nonetheless, the

assertion that increasing reward has a stronger influence on purchase behavior than

redemption behavior when the opportunity to redeem is contingent on purchase is

maintained.
















10%


8%
-


t 6%


O 4%
-




0%


7.5%


Redemption Period


O $6 H $9


Figure 22. Proportion of experiment 4 coupon holders that applied for the rebate by
condition

1.6


1.2
1.2



S0.8-

0.5

0.4-





Redemption Period

O $6 g $9


Figure 23. Experiment 4 login delay by condition








67




2.5

2.0
2.0


1.5 -


1.0 -





0.0

Redem ption Period

O $6 g $9


Figure 24. Experiment 4 mail delay by condition















CHAPTER 10
GENERAL DISCUSSION

The psychological mechanisms proposed in this article are applicable not only to

real-world mail-in rebates, but also to other paradigms where the opportunity for

breakage looms (i.e., when the attainment of a delayed incentive is conditional on the

performance of some future effort). This research offers four contributions to marketing's

understanding of breakage. First, by isolating rebate-dependent purchases, the

experiments establish that breakage is indeed a common post-purchase outcome. Second,

by tracking post-purchase behavior, the experiments go beyond redemption rates to

examine how and when breakage occurs. Although exact levels of breakage are

dependent on the characteristics of the offer and the redemption process, results suggest

that much of breakage can be attributed to psychological mechanisms that exert influence

before consumers initiate the redemption process (i.e., redemption overconfidence,

procrastination, and to a lesser extent, forgetting). This finding is important because it

demonstrates that breakage is largely determined before consumers become aware of the

effort required to redeem a rebate. Third, the results provide direct evidence of the

relationship between consumers' redemption confidence and their purchase and

redemption behavior. Redemption confidence can be manipulated to encourage purchase

by increasing the size of a reward and the length of a redemption period, but high

confidence in redeeming does not necessarily translate into actual redemption behavior.

Fourth, by examining repeat purchase behavior, the experiments reveal that rebate buyers

making rebate-dependent purchases avoid offers that are known to require a high degree









of effort and offers that they failed to redeem in the past, and that moderate rates of

breakage can still occur among successful redeemers that make repeat rebate-dependent

purchases. Although breakage can be a recurring problem for some consumers, the

likelihood of repeat breakage is limited by the low probability of repeat purchase among

failed redeemers.

Alternative Explanations for Breakage

It is evident from the results that there are many factors which influence purchase

and redemption behavior to create breakage. In addition to the factors identified by this

research, there are two additional explanations for breakage that may have exerted

effects. One of these explanations is reward discounting. Research on value discounting

has established that people often apply discount rates that diminish the perceived value of

future payments (see Frederick, Loewenstein, and O'Donoghue 2002 for a review). If

reward discounting were to occur after purchase, the reward could appear less valuable at

the time of redemption than it does at the time of purchase, which would encourage

breakage. Although evidence supporting reward discounting was not observed, it may

still have exerted an effect. The absence of a reward by redemption period interaction on

redemption in experiments 1, 2, 3 and 4 suggests that the influence of reward discounting

is relatively minor. Furthermore, none of the respondents reported that reward appeared

less valuable after making their purchases. Nevertheless, reward discounting cannot be

ruled out as a contributor to breakage.

A second alternative explanation is that consumers frame the value of the reward

relative to the list price at the time of purchase and later reframe the value of the reward

relative to redemption effort after purchase. Consequently, the cost of the future effort is

underweighted at the time of purchase and more accurately weighted after purchased.










The post test results provide some support for this explanation. The low proportion of

rebate buyers that foresaw problems with redeeming at the time of purchase (P = 33%)

suggests that redemption effort is not prominent in the minds of most buyers at the time

of purchase. This portion of the reframing argument is essentially the same as the

argument for redemption overconfidence (i.e., people are overconfident because they fail

to consider factors that could potentially interfere with redeeming). With regard to the

reframing of reward after purchase, one would expect a) the maj ority of breakage to

occur once buyers become aware of the redemption effort, and b) for rebate buyers to

report redemption effort and/or insufficient reward as the reason for breakage. Although

this occurred, it was not the case for the maj ority of cases. Experiments 1, 2, and 3

showed that the majority of breakage occurred before buyers became aware of

redemption effort, and the self-reported reasons for breakage indicate that redemption

effort and/or insufficient reward accounted for only 20.8% of breakage. In summary,

although reframing may certainly occur, it does not appear to account for the maj ority of

breakage.

Implications for Regulators

Curbing Breakage

From a regulatory perspective, the results indicate that a substantial proportion of

rebate buyers can fall victim to breakage, and that breakage is caused, in part, by

consumers being highly confident of redeeming a rebate that they are obj ectively unlikely

to redeem. For those interested in curbing breakage, two approaches may be effective.

One approach is to focus on curbing consumers' redemption confidence at the time of

purchase. This approach is based on the notion that a consumer' s obj ective probability of

redeeming is largely predetermined, and that curbing breakage is a function of









discouraging rebate-dependent purchases among consumers who are unlikely to redeem

the rebate. As noted earlier, a common behavioral phenomenon is that people tend to

overlook situational factors that could potentially interfere with achieving a desired

outcome. Thus, regulators may wish to conduct research that examines how to overcome

this phenomenon in the context of rebates and other forms of delayed incentives. For

example, regulators may wish to explore how regulating the presentation of rebate offers

(e.g., explicitly listing redemption requirements on the price tag) influences breakage. A

second approach to curbing breakage is to focus on post-purchase behavior. This assumes

that breakage can be reduced by simplifying the redemption process. Research could

examine whether regulatory efforts to simplify redemption are effective in mitigating

procrastination and procedural constraints that contribute to breakage. This approach is

currently reflected in pending legislation in several regions (e.g., California Bill 1154).

Implications for Managers

Trend toward Shorter Redemption Periods

From a managerial perspective, four implications are noteworthy. The first

implication relates to the trend toward shorter redemption periods. Twenty years ago,

consumers were typically given 60 to 90 days to redeem a rebate offer. Today, 30-day,

14-day and 7-day redemption periods are the norm in many product categories.l

Marketers report that this trend has been guided by two forces: the desire to complete and

assess promotions in a timely manner, and the conventional wisdom that giving

consumers less time to redeem will decrease redemption rates. Although this intuition

seems reasonable, the results have maj or implications for marketers in demonstrating that



SSource: DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing Association.









shorter redemption periods decrease sales and increase redemption applications. As

discussed earlier, there are several factors that have hindered research on breakage.

Consequently, many managers may be unaware of these effects.

The Role of Redemption Effort in Breakage

The maj ority of breakage appears to occur before consumers become aware of the

effort required to redeem. Even if this result is attributed to consumers' expectations of

redemption effort (which self-reports suggest is not the case), manipulating redemption

effort would not have affected this result. Furthermore, increasing redemption effort

decreased procrastination and marginally increased redemption applications. Although

redemption effort did create breakage among those who initiated redemption,

procrastination interfered with initiating and completing redemption and accounted for a

greater proportion of breakage than redemption effort. These results have three

implications. First, excessively effortful redemption processes should not be viewed as a

minimum requirement for breakage. Second, increasing redemption effort to increase

breakage can have an upper limit and can increase redemption rates. Third, although

exact levels of breakage are dependent on the characteristics of a firm' s redemption

process, redemption processes that provide multiple opportunities to procrastinate can

create more breakage than moderate levels of redemption effort. Given the conventional

wisdom that breakage is caused primarily by redemption effort, firms may be over-

attributing the role that effort plays in creating breakage.

Repeat Purchase Behavior

Results reveal that rebate buyers avoid high-effort offers and offers that they failed

to redeem in the past, and that moderate rates of breakage can still occur among

successful redeemers that make repeat rebate-dependent purchases. This implies that










high-effort redemption requirements can steer consumers away from rebates and generate

less breakage in the long run. Consequently, moderate-effort offers may be more optimal

than high-effort offers since moderate-effort offers can take advantage of breakage

without risking higher redemption rates or decreases in future sales.

Future Research

Effects of Effort on Redemption

Increasing redemption effort was shown to reduce procrastination and increase

(marginally) the proportion of buyers that applied for the rebate. This result is opposite to

the "aversiveness of effort" effect observed in earlier experiments on delayed incentives

(e.g., Soman 1998). The apparent inconsistency between these effects may be explained

by the timing of the effort manipulations. The delayed incentives study manipulated

effort at the time of choice by showing subj ects the length of the survey that was to be

completed to receive the reward. In contrast, the current study did not manipulate effort

until after subj ects had initiated the redemption process. Post-test responses revealed that

the maj ority of consumers in the high-effort condition felt that the redemption procedures

were more effortful than they had expected at the time of purchase. Thus, the quicker

response and higher redemption rate among these participants could be interpreted as a

backlash effect. However, the reasons for this effect are not clear. In particular, three

explanations are plausible. First, the result could be a sunk cost effect whereby effort

invested in stage one of a task increases one's motivation to complete subsequent stages

of the task. Second, the result could be a retribution effect whereby one' s motivation to

redeem increases when redeeming is framed as an opportunity to seek retribution for

completing an unnecessarily effortful task. Third, the result could be a salience effect

whereby increasing the effort increases the salience of the task and the likelihood of










completing it before a deadline. Future research could examine these explanations and

their implications.

Differential Effects of Reward on Purchase and Redemption

Increasing the reward was shown to exert a stronger effect on purchase than

redemption. Although the null effect of reward on redemption is unlikely to hold at larger

reward values (i.e., at some point, a reward will be sufficiently large to generate a

significant effect on redemption), it is uncertain whether the differential strength of the

effect will hold at larger reward values (i.e. will offering a $100 reward still have a larger

influence on purchase than redemption?). Reports of large sales increases and moderate

redemption rates for large reward values (i.e., $400) suggest that the differential effect

may hold (e.g., Spencer 2002). Given the difficulties of studying large-ticket purchases in

a laboratory environment, this issue would be best addressed by future field studies that

manipulate offer characteristics and control for market factors.















APPENDIX
POST TEST SURVEY ITEMS

Instructions
The series of studies that you completed for us were not about movies, but were
actually about rebate offers. This rebate study was conducted for the Federal Trade
Commission to find out if rebates constitute a risk to consumer welfare.

We will ask you some questions about how you decided to purchase the rebated
movie tickets, and about your attempts) to redeem the rebate offer.
It is very important that you answer all questions honestly so that we can
accurately assess how consumers respond to rebate offers.


Questions
*Denotes questions that were administered to non-redeemers only.

(1) Did you know at the time of doing the movie rating studies that this was an
experiment on rebate offers? (Please be honest)
I Yes
O No

(2) If you did know, please tell us how you knew. If you did not know, click
continue.

(3) When you purchased the tickets, how confident were you that you would redeem
the rebate?


Your confidence estimate should correspond to the probability that you would
redeem the rebate. For example, "100% confident" means that you were
completely certain that you would redeem the rebate. "50% confident" means
that you were completely uncertain about whether you would redeem (e.g., a
50/50 chance you would redeem). Any number below 50% means that you
believed it was more likely that you would not redeem. For exam le, "0%
confident" means that you were com letely certain you would not redeem the
rebate. Please indicate your confidence estimate using the scale below.
Remember, your answer should be based on what you thought at the time of
purchase.l



i Probability scale instructions adapted from Vallone et al. 1990.









0% confidence 100% confidence [sliding scale]

(4) Please estimate the percentage of rebate buyers that you thought would redeem
the rebate when you purchased the tickets.

Your estimate should correspond to the percentage of rebate buyers you thought
would redeem the rebate. For example, an estimate of 100% means that you
thought all of the rebate buyers would redeem the rebate. An estimate of 50%
means that you thought half of the rebate buyers would redeem the rebate. An
estimate of 0% means that you thought none of the rebate buyers would redeem
the rebate. Please indicate your estimate below. Remember, your answer should
be based on what you thought at the time of purchase.

0% 100% [sliding scale]

(5) At the time of purchasing the tickets, did you foresee any problems that could
interfere with redeeming the rebate?
If yes, please type "yes" and describe what you thought about when you
purchased the tickets, and why you still chose to buy the tickets.
If no, please type "no" and describe what you thought about when you purchased
the tickets.


(6) How was the experience of redeeming the rebate compared to what you expected
it would be when you purchased the tickets?

-100 (more effort than I expected) to +100 (less effort than I expected) [sliding
scale]

(7) Please describe the actions you took (if any) to redeem the rebate, and any
obstacles that you encountered. For example, did you logon to the rebate
website? Did you complete the redemption form? What obstacles did you
encounter? (e.g., procrastination, no postage stamps, etc.)

(8) *In your own words, please describe why you didn't redeem the rebate. Please be
as specific as possible.

(9) *Below are some common reasons why people don't redeem rebates. Please
select the reason that best describes your situation:
0 I mailed my rebate application but I was rejected.
O I lost my receipt / tickets and could not apply.
O I used the tickets before I could photocopy them and could not apply.
O I procrastinated and the deadline passed before I applied.
O I procrastinated and eventually forgot to apply.
O I completely forgot to apply.










0 I tried to redeem but it was too much work so I gave up.
O I decided the money saved was not worth the effort required.
O The money saved didn't seem as valuable when it came time to redeem.
(10) *If we gave you another chance to redeem the rebate, would you take it?
O Yes
O No

(11) *Please explain the reason for your answer in the box below.

(12) *All things considered, how would you categorize yourself as a result of buying
the rebate offer?
O I am better off for buying the rebate offer.
O I am no better or worse off for buying the rebate offer.
O I am worse off for buying the rebate offer.

(13) *Please explain the reason for your answer in the box below.

(14) *Which behavior do you regard as the bigger mistake?
O Buying the rebate offer was the bigger mistake.
O Not redeeming the rebate offer was the bigger mistake.

(15) *Please explain the reason for your answer in the box below.

(16) *All things considered, not redeeming the rebate was:
0 My fault
0 The fault of the offer

(17) *Please explain the reason for your answer in the box below.

(18) All things considered, the rebate offer was:
0 A fair offer
I Not a fair offer (deceptive)

(19) Please explain the reason for your answer in the box below.

(20) Before participating in this study, how many times in the past have you purchased
a rebate offer with the intention of redeeming the rebate?
O Never
I Once
0 Twice
0 3 times
O 4 times










0 5- 10 times
0 More than 10 times

(21) Of the purchases you indicated above, approximately what proportion of them did
you successfully redeem?


0% (none of them) 100% (all of them) [sliding scale]

(22) Before this study, what was your general attitude toward rebate offers?
Use the scale below to record your response. A rating above "O" represents a
positive attitude toward rebates. A rating of "O" represents a neutral attitude.
A rating below "O" represents a negative attitude toward rebates.

-100 (generally negative) to +100 (generally positive) [sliding scale]

(23) Since participating in this study, what is your general attitude toward rebate
offers?
Use the scale below to record your response. A rating above "O" represents a
positive attitude toward rebates. A rating of "O" represents a neutral attitude.
A rating below "O" represents a negative attitude toward rebates.

-100 (generally negative) to +100 (generally positive) [sliding scale]

(24) What is the likelihood that you will purchase another rebate offer (for some other
product) in the future?

Your estimate should correspond to the probability that you will buy a rebate
offer in the future. For example, an estimate of 100% means that you are
completely certain that you will buy another rebate offer in the future. An
estimate of 50% means that you are completely uncertain about whether you will
buy another rebate offer in the future. Any number below 50% means that you
believed it is more likely that you will not buy another rebate offer in the future.
For example, 0% means that you are completely certain that you will not buy
another rebate offer in the future.


0% (not at all likely) 100% (very likely) [sliding scale]

(25) [Those who did not successfully redeem were given the opportunity to purchase
tickets again using the purchase opportunity choice set.]











(26) *How confident are you that you will redeem the rebate?

Your confidence estimate should correspond to the probability that you would
redeem the rebate. For example, "100% confident" means that you are
completely certain that you will redeem the rebate. "50% confident" means that
you are completely uncertain about whether you will or will not redeem (e.g., a
50/50 chance you will redeem). Any number below 50% means that you believe
it is more likely that you will not redeem. For example, "0% confident" means
that you are completely certain you will not redeem the rebate. Please indicate
your confidence estimate using the scale below.


0% confidence 100% confidence [sliding scale]

(27) *Please estimate the percentage of rebate buyers that you think will redeem the
rebate.


Your estimate should correspond to the percentage of buyers that will redeem the
rebate. For example, an estimate of 100% means that you think all of the rebate
buyers will redeem the rebate. An estimate of 50% means that you think half of
the rebate buyers will redeem the rebate. An estimate of 0% means that you think
none of the rebate buyers will redeem the rebate. Please indicate your estimate
below.


0% 100% [sliding scale]
















LIST OF REFERENCES


Akerlof, George A. (1991), "Procrastination and Obedience," American Economic
Review, 81 (May) 1-19.

Alba, Joseph W. and J. Wesley Hutchinson (2000), "Knowledge Calibration: What
Consumers Know and What They Think They Know," Journal of Consumer
Research, 27 (September), 123-156.

Ali, Abdul, Marvin A. Jolson, and Rena Y. Darmon (1994), "A Model for Optimizing the
Refund Value in Rebate Promotions," Journal of Business Research, 29 (March),
239-245.

Ariely, Dan and Klaus Wertenbroch (2002), "Procrastination, Deadlines, and
Performance: Self-Control by Precommitment," Psychological Science, 13 (May),
219-224.

Avila, Ramon A., Joseph D. Chapman, and Teresa K. Avila (1989), "An Exploratory
Study on Consumer' s Attitudes Toward Rebates," in Developments in Marketing
Science, 12, 273-77.

Ben-Akiva, Moshe E. and Steven R. Lerman (1985), Discrete Choice Analysis: Theory
and Applications to Travel Demand, Cambrid ge, MA: MIT Press.

Blattberg, Robert and Scott Neslin (1990), Sales Promotions: Concepts, M~ethods and'
Strategies, Englewood Cliffs, NJ: Prentice-Hall.

Buehler, Roger, Dale Griffin, and Michael Ross (1994), "Exploring the 'Planning
Fallacy': Why People Underestimate Their Task Completion Times," Journal of
Personality and' Social Psychology, 67 (September), 366-3 81.

Dhar, Sanjay K. and Stephen J. Hoch (1996), "Price Discrimination using In-Store
Merchandising," Journal of2arketing, 60 (January), 17-30.

Ellis, Albert and William J. Knaus (1977), Overcoming Procrastination, New York:
Institute for Rational Living.

Ferrari, Joseph R. (1993), "Christmas and Procrastination: Explaining Lack of Diligence
at a 'Real-World' Task Deadline," Personality and'Individ'ual Differences, 14
(January), 25-33.










Frederick, Shane, George Loewenstein, and Ted O'Donoghue (2002), "Time Discounting
and Time Preference: A Critical Review," Journal of Econontic Literature,
40(June), 351-401.

Gilovich, Thomas, Margaret Kerr, and Victoria Husted Medvec (1993), "Effect of
Temporal Perspective on Subj ective Confidence," Journal ofPersonality and'
Social Psychology, 64 (April), 552-560.

Greenman, Catherine (1999), "The Trouble with Rebates," The New York Times,
September 16, 1999, Gl.

Griffin, Dale W., David Dunning, and Lee Ross (1990), "The Role of Construal
Processes in Overconfident Predictions about the Self and Others," Journal of
Personality and' Social Psychology, 59 (December), 1 128-1 139.

Griffin, Dale W. and Carol A. Varey (1996), "Towards a Consensus on Overconfidence,"
Organizational Behavior and Human Decision Processes, 65 (March), 227-23 1.

Harriott, Jesse and Joseph R. Ferrari (1996), "Prevalence of Chronic Procrastination
Among Samples of Adults," Psychological Reports, 78 (April), 61 1-616.

Hoch, Stephen J. (1985), "Counterfactual Reasoning and Accuracy in Predicting Personal
Events," Journal ofExperintental Psychology: Lea-l~rning. Mentory and' Cognition,
11 (4), 719-731.

Holmes, Richard Alan (2002), "The Effect of Task Structure and Task Order on
Subj ective Distress and Dilatory Behavior in Academic Procrastinators," Doctoral
Dissertation, Hofstra University.

Hunt, Kenneth A. and Susan M. Keaveney (1992), "Sales Promotion Wars: A Strategic
Application of Attribution Theory," Journal of Pronzotion Management, 1 (3), 3-
20.

Hunt, Kenneth A., Susan M. Keaveney, and Moonkyu Lee (1995), "Involvement,
Attributions and Consumer Responses to Rebates," Journal of Business and'
Psychology, 9 (Spring), 273-297.

Jolson, Marvin, Joshua Weiner, and Richard Rosecky (1987), "Correlates of Rebate
Proneness," Journal of Advertising Research, 27 (Feb-March), 33-43.

Knaus, William J. (2000), "Procrastination, Blame and Change," Journal of Social
Behavior andPersonality, 15 (Special Issue), 153-166.

Krishnan, H. Shanker and Stewart Shapiro (1999), "Prospective and Retrospective
Memory for Intentions: A Two-Component Approach," Journal of Consunter
Psychology, 8 (2), 141-166.









Lanctot, Roger (2002), "Rebates Rule in the Retail Computer Software Market," Business
Wire, November 12, 2002.

Lay, Clarry H. (1988), "The Relation of Procrastination and Optimism to Judgments of
Time to Complete an Essay and Anticipation of Setbacks," Journal of Social
Behavior and Personality, 3 (September), 201-214.

Lay, Clarry H. and Peter Burns (1991), "Intentions and Behavior in Studying for an
Examination: The Role of Trait Procrastination and its interaction with Optimism,"
Journal of Social Behavior and Personality, 6 (September), 605-617.

Liberman, Nira and Yaacov Trope (1998), "The Role of Feasibility and Desirability
Considerations in Near and Distant Future Decisions: A Test of Temporal
Construal Theory," Journal ofPersonality and' Social Psychology, 75 (July), 5-18.

Marsh, Richard L., Jason L. Hicks, and Valerie Watson (2002), "The Dynamics of
Intention Retrieval and Coordination of Action in Event-Based Prospective
Memory," Journal ofExperintental Psychology: Learning ~Mentoz and' Cognition,
28 (July), 652-659.

McGinn, Daniel (2003), "Let's Make a (Tough) Deal," Newsweek, June 23, 2003, 48-49.

McLaughlin, Katy (2002), "Cranky Customer: Claiming That Holiday Rebate," The Wall
Street Journal, December 3, 2002, D2.

Milburn, Michael (1978), Sources of Bias in the Prediction of Future Events,"
Organizational Behavior and Human Performance, 21 (February), 17-26.

Norberg, Bob (2002), "Rebate Debate: Manufacturers Betting Consumers Won't Apply
for Refunds," The Gainesville Sun, December 16, 2002, WorkLife 6.

Norr, Henry (2000), "The How and Why of Rebates," San Francisco Chronicle,
December 18, 2000, T21.

Pulford, Briony D. and Andrew M. Colman (1996), "Overconfidence, Base Rates, and
Outcome Positivity/Negativity of Predicted Outcomes," British Journal of
Psychology, 87 (August), 431-445.

Pychyl, Timothy A., Richard W. Morin, and Brian R. Salmon (2000), "Procrastination
and the Planning Fallacy: An Examination of the Study Habits of University
Students," Journal of Social Behavior and Personality, 15 (Special Issue), 135-150.

Rothschild, Michael (1987), Marketing Conanunications: From Fundamentalsd~~~dd~~~ddd~~ to
Strategies, Lexington, MA: D. C. Heath.

Scher, Steven J. and Joseph R. Ferrari (2000), "The Recall of Completed and
Noncompleted Tasks Through Daily Logs to Measure Procrastination," Journal of
Social Behavior and Personality, 15 (Special Issue), 255-265.










Shapiro, Stewart and H. Shanker Krishnan (1999), "Consumer Memory for Intentions: A
Prospective Memory Perspective," Journal ofExperintental Psychology: Applied, 5
(June), 169-189.

Shim, Richard (2002), "Philips Settles Rebate Complaint," The New York Tintes.conz,
August 7, 2002.

Sigall, Harold, Arie Kruglanski, and Jack Fyock (2000), "Wishful Thinking and
Procrastination," Journal of Social Behavior and Personality, 15 (Special Issue),
283-296.

Silver, Maury (1974), "Procrastination," Centerpoint, 1, 49-54.

Soman, Dilip (1998), "The Illusion of Delayed Incentives: Evaluating Future Effort-
Money Transactions," Journal of2arketing Research, 3 5 (November), 427-437.

Soman, Dilip and John T. Gourville (2002), "Mail-in Rebates and Consumer Choice,"
Working Paper, Harvard Business School, Harvard University.

Spencer, Jane (2002), "Rej ected! Rebates Get Harder to Collect," The Wall Street
Journal, June 11, 2002, D1.

Tat, Peter (1994), "Rebate Redemption: A Motivational Perspective," Psychology &
Marketing, 11 (Jan/Feb), 15-26.

Tat, Peter, William Cunningham III, and Erin Babakus (1988), "Consumer Perceptions of
Rebates," Journal ofAdvertising Research, 28 (Aug-Sep), 45-50.

Tat, Peter and Charles Schwepker (1998), "An Empirical Investigation of the
Relationship Between Rebate Redemption Motives: Understanding How Price
Consciousness, Time and Effort, and Satisfaction Affect Consumer Rebate
Redemptions," Journal of2arketing Theory and Practice, 6 (Spring), 61-71.

Trope, Yaacov and Nira Liberman (2000), "Temporal Construal and Time-Dependent
Changes in Preference," Journal ofPersonality and Social Psychology, 79
(December), 867-889.

Tversky, Amos and Eldar Shafir (1992), "Choice Under Conflict: The Dynamics of
Deferred Decisions," Psychological Science, 3 (November), 358-361.

Vallone, Robert. P., Dale W. Griffin, Sabrina Lin, and Lee Ross (1990), "The
Overconfident Prediction of Future Action Outcomes by Self and Others," Journal
ofPersonality and Social Psychology, 58 (April), 582-592.

Weiner, Bernard (2000), "Attributional Thoughts about Consumer Behavior," Journal of
Consumer Research, 27 (December), 3 82-3 87.










Wright, George and Peter Ayton (1992), "Judgmental Probability Forecasting in the
Immediate and Medium Term," Organizational Behavior and' Human Decision
Processes, 51 (April), 344-363.
















BIOGRAPHICAL SKETCH

Timothy Guy Silk received his BBA from Wilfrid Laurier University in Waterloo,

Ontario, Canada, in 1997. He j oined the University of Western Ontario as an instructor,

where he taught at the Ivey School of Business and traveled to Russia and Ukraine to

teach local business managers as part of the LEADER economic development program.

Tim moved to the United States in the fall of 2000 to pursue his Ph.D. in marketing at the

University of Florida, and will j oin the University of South Carolina as an assistant

professor of marketing in August 2004.




Full Text

PAGE 1

EXAMINING PURCHASE AND NON-REDEMPTION OF MAIL-IN REBATES: THE IMPACT OF OFFER VARIABLES ON CONSUMERS SUBJECTIVE AND OBJECTIVE PROBABILITY OF REDEEMING By TIMOTHY GUY SILK A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by Timothy Guy Silk

PAGE 3

This dissertation is dedicated to my parents.

PAGE 4

ACKNOWLEDGMENTS I am deeply indebted to my advisor, Chris Janiszewski, who committed himself to my development from the day I arrived in the program and supported me in pursuing the topics that excite me. I would also like thank my committee members Richard Lutz and Joe Alba for their extensive comments which contributed a great deal to the conceptual development of the dissertation, Alan Cooke for his conceptual guidance as well as his generous assistance in developing the website used in the experiments, and Mahendrara Nimalendran for his support and feedback at all stages of the dissertation. I am also indebted to Joel Cohen, who encouraged me throughout the course of my doctoral education. Finally, I would like to thank my family for their encouragement, particularly Alvin Silk for his guidance. iv

PAGE 5

TABLE OF CONTENTS Page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT.......................................................................................................................xi CHAPTER 1 INTRODUCTION........................................................................................................1 The Need for Research.................................................................................................1 Goals of the Dissertation..............................................................................................3 2 TYPOLOGY OF REBATE PURCHASES..................................................................4 How Rebates Differ From Coupons.............................................................................4 Classifying Purchases...................................................................................................5 Implications for Measuring Redemption Rates............................................................5 Implications for Measuring Breakage..........................................................................7 Influencing Breakage....................................................................................................7 Research on Breakage...................................................................................................8 3 CONCEPTUAL AND THEORETICAL FOUNDATIONS......................................13 Pre-Purchase Factors..................................................................................................14 Redemption Confidence......................................................................................14 Desirability..........................................................................................................15 Temporal Distance...............................................................................................16 Post-Purchase Factors.................................................................................................16 Procrastination.....................................................................................................17 Prospective Forgetting.........................................................................................18 Redemption Effort...............................................................................................19 Repeat-Purchase Factors.............................................................................................20 Past Redemption Behavior..................................................................................20 4 INTRODUCTION TO EXPERIMENTS...................................................................22 v

PAGE 6

5 EXPERIMENT 1........................................................................................................23 Design.........................................................................................................................23 Procedure....................................................................................................................23 Phase 1: Purchase................................................................................................23 Phase 2: Redemption...........................................................................................26 Phase 3: Responding to Rebate Applications......................................................27 Dependent Measures...................................................................................................27 Purchase...............................................................................................................27 Breakage and Redemption...................................................................................27 Procrastination.....................................................................................................28 Redemption Effort Manipulation Check.............................................................28 Results.........................................................................................................................29 Purchase...............................................................................................................29 Redemption..........................................................................................................30 Procrastination.....................................................................................................30 Discussion...................................................................................................................31 Purchase and Redemption...................................................................................31 Insights Regarding Breakage...............................................................................32 6 EXPERIMENT 2........................................................................................................37 Design and Procedure.................................................................................................38 Results.........................................................................................................................39 Purchase...............................................................................................................39 Redemption..........................................................................................................39 Procrastination.....................................................................................................40 Probability Estimates...........................................................................................40 Discussion...................................................................................................................42 7 EXPERIMENT 3........................................................................................................47 Design and Procedure.................................................................................................47 Results.........................................................................................................................48 Repeat Purchase...................................................................................................48 Redemption..........................................................................................................48 Procrastination.....................................................................................................49 Repeat Breakage..................................................................................................50 Discussion...................................................................................................................51 8 POST TEST SURVEY...............................................................................................55 Results.........................................................................................................................55 Reasons for Breakage..........................................................................................56 Redeemers vs. Non-redeemers............................................................................57 Non-Redeemers...................................................................................................58 vi

PAGE 7

9 EXPERIMENT 4........................................................................................................62 Design and Procedure.................................................................................................62 Dependent Measures and Analysis.............................................................................63 Results.........................................................................................................................64 Redemption..........................................................................................................64 Procrastination.....................................................................................................64 Discussion...................................................................................................................64 10 GENERAL DISCUSSION.........................................................................................68 Alternative Explanations for Breakage.......................................................................69 Implications for Regulators........................................................................................70 Curbing Breakage................................................................................................70 Implications for Managers..........................................................................................71 Trend toward Shorter Redemption Periods.........................................................71 The Role of Redemption Effort in Breakage.......................................................72 Repeat Purchase Behavior...................................................................................72 Future Research..........................................................................................................73 Effects of Effort on Redemption.........................................................................73 Differential Effects of Reward on Purchase and Redemption.............................74 APPENDIX : POST TEST SURVEY ITEMS..................................................................75 LIST OF REFERENCES...................................................................................................80 BIOGRAPHICAL SKETCH.............................................................................................85 vii

PAGE 8

LIST OF TABLES Table page 1. Self-reported reasons for not making a repeat rebate purchase in experiment 3...........54 2. Self-reported reasons for breakage by non-redeemers...................................................60 3. Post-test responses by redeemers and non-redeemers...................................................60 4. Post-test responses by non-redeemers only...................................................................61 viii

PAGE 9

LIST OF FIGURES Figure page 1. Typology of rebate purchases........................................................................................11 2. Impact of post-purchase outcomes on redemption rates................................................12 3. Factors that contribute to breakage................................................................................21 4. Purchase opportunity choice set 4 ...................................................................................33 5. Proportion of experiment 1 participants purchasing rebate and non-rebate offers by condition...................................................................................................................34 6. Proportion of experiment 1 rebate buyers that applied for the rebate by condition......34 7. Experiment 1 login delay by condition..........................................................................35 8. Experiment 1 mail delay by condition...........................................................................35 9. Experiment 1 breakage by stage of redemption (n=184)...............................................36 10. Proportion of experiment 2 participants purchasing rebate and non-rebate offers by condition...................................................................................................................43 11. Proportion of experiment 2 rebate buyers that applied for the rebate by condition....43 12. Experiment 2 login delay by condition........................................................................44 13. Experiment 2 mail delay by condition.........................................................................44 14. Experiment 2 confidence and base rate estimates by choice.......................................45 15. Experiment 2 confidence and base rate estimates by reward......................................45 16. Experiment 2 confidence and base rate estimates by redemption period....................46 17. Experiment 2 confidence and base rate estimates of redeemers and non-redeemers..46 18. Proportion of rebate buyers making a repeat purchase of the rebate offer in experiment 3 by condition........................................................................................52 ix

PAGE 10

19. Proportion of experiment 3 rebate buyers that applied for the rebate by condition....52 20. Experiment 3 login delay by condition........................................................................53 21. Experiment 3 mail delay by condition.........................................................................53 22. Proportion of experiment 4 coupon holders that applied for the rebate by condition.66 23. Experiment 4 login delay by condition........................................................................66 24. Experiment 4 mail delay by condition.........................................................................67 x

PAGE 11

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 EXAMINING PURCHASE AND NON-REDEMPTION OF MAIL-IN REBATES: THE IMPACT OF OFFER VARIABLES ON CONSUMERS SUBJECTIVE AND OBJECTIVE PROBABILITY OF REDEEMING By Timothy Guy Silk May 2004 Chair: Chris Janiszewski Major Department: Marketing Mail-in rebates are popular with retailers and manufacturers because they achieve attractive price points and increase sales while limiting the number of consumers that redeem the rebate to obtain the price discount. This research examines a phenomenon known as breakage, which occurs when consumers are enticed to purchase as a result of a rebate offer but subsequently fail to apply for the rebate. Borrowing from research on overconfidence, procrastination and prospective forgetting, this research examines underlying psychological mechanisms that lead to breakage and how they exert influence via changes in offer characteristics. A series of experiments demonstrate that breakage can be attributed to the discrepancy between consumers subjective probability of redeeming (i.e., redemption confidence) and their objective probability of redeeming (i.e., actual redemption behavior). Increasing redemption confidence increases the likelihood of purchasing a rebate offer. However, high redemption confidence does not necessarily translate into successful redemption. xi

PAGE 12

The experiments produced five key results. First, increasing the rebate reward increases redemption confidence and the proportion of consumers that purchase the rebate offer, but has a weaker effect on the proportion of buyers that apply for the rebate. Second, increasing the length of the redemption period increases redemption confidence and the proportion of consumers that purchase the rebate offer, but also increases procrastination and breakage. Third, increasing the effort required to redeem the rebate results in a backlash effect where the redemption rate is marginally higher for consumers that experience a high-effort redemption process compared to those that experience a low-effort redemption process. Fourth, failing to redeem a rebate or experiencing a high-effort redemption process dramatically decreases the proportion of rebate buyers that purchase the offer again. Finally, structuring redemption procedures to encourage procrastination can have a greater impact on breakage than increasing the effort required to redeem the rebate. These results have major implications for marketers in demonstrating that shorter redemption periods decrease sales and increase redemption applications. Results also suggest that excessively effortful redemption procedures should not be viewed as a minimum requirement for breakage and that firms may be over-attributing the role of effort in creating breakage. xii

PAGE 13

CHAPTER 1 INTRODUCTION Mail-in rebates (hereafter rebates) are one of the most popular pricing mechanisms used by manufacturers and retailers to increase sales and discriminate among consumers who vary in their price sensitivity (Blattberg and Neslin 1990; Jolson, Weiner, and Rosecky 1987; Soman and Gourville 2002). Rebates are frequently used in the consumer goods sector (Tat and Schwepker 1998) and are the most common promotion tactic used in consumer electronics (Lanctot 2002). Rebates are popular because they can be used to lower a products price and increase sales while limiting the number of consumers that redeem the rebate to obtain the price discount (Lanctot 2002). As explained by one retailer, Manufacturers love rebates because redemption rates are close to none they get people into stores, but when it comes time to collect, few people follow through. And this is just what the manufacturer has in mind (Greenman 1999, p. G1). This research examines this phenomenon, widely referred to as breakage, which occurs when consumers are enticed to purchase as a result of the rebate offer but subsequently fail to apply for the rebate (Jolson et al. 1987). The Need for Research The reasons for breakage are not well understood. Despite the attention rebates have received in the business and popular press (e.g., McGinn 2003; McLaughlin 2002), and despite calls for regulation to protect consumers (e.g., Shim 2002; Spencer 2002), research on consumer response to rebates is extremely limited. Most academic rebate research has relied on consumer surveys and economic models to address other important 1

PAGE 14

2 rebate issues. For example, research has examined optimum rebate reward amounts (Ali, Jolson, and Darmon 1994), consumer perceptions of manufacturers motives for offering rebates (Avila, Chapman, and Avila 1989; Tat, Cunningham, and Babakus 1988), consumer attributions of satisfaction with rebate shopping experiences (Hunt and Keaveney 1992; Hunt, Keaveney, and Lee 1995), consumer perceptions of the redemption process (Jolson, Weiner, and Rosecky 1987; Tat et al. 1988), and consumer motives toward rebate redemption (Tat 1994; Tat and Schwepker 1998). With the exception of the research on delayed incentives (Soman 1998) and correlates of rebate proneness (Jolson et al. 1987), there have been no investigations of the reasons for breakage. Practitioner-driven research on breakage is also severely limited. Interviews with manufacturers and marketing service firms reveal that practitioner-driven research has been hindered by four factors. 1 First, as will be explained, marketers experience considerable difficulty in isolating and measuring breakage. Second, the use of rebate promotions is driven primarily by sales managers aiming to increase sales during promotion periods dictated by key account retailers. As one manager explained, If your key account retailer gives you a promotion window and you are looking for a quick and easy way to increase sales during that window, you are going to put together a rebate promotion before you use a coupon or some other promotion tactic. 2 As a result, manufacturers are more concerned with meeting sales targets and maintaining downward pressure on redemption rates than investigating factors that influence breakage. Third, 1 Source: BDS Marketing, Consumer Electronics Association, DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing Association, TCA Fulfillment Services. 2 Source: BDS Marketing.

PAGE 15

3 manufacturers are extremely reluctant to conduct market-based experiments that manipulate offer characteristics due to fears that doing so will adversely affect sales and/or increase redemption rates. Fourth, industry associations and the vast majority of manufactures do not maintain data on rebate promotions, limiting the availability of market data. In summary, trends in offer characteristics such as decreasing the length of time consumers are given to redeem and increasing the effort required to redeem have been guided primarily by industry intuition about effective strategy as opposed to formal research. Goals of the Dissertation Borrowing from research on overconfidence, procrastination and prospective forgetting, this research examines the underlying psychological mechanisms that lead to breakage. I will show that changes in offer characteristics have the potential to influence both purchase and redemption, and that these influences can be opposing or synergistic. As a consequence, understanding breakage requires an understanding of the behavioral processes that influence the expectation of redemption at the time of purchase and the initiation and completion of the redemption process. The dissertation proceeds as follows: Chapter 2 introduces a typology of rebate purchases to illustrate how breakage differs from other post-purchase outcomes and discusses managerial issues pertaining to the design, implementation, and evaluation of rebate promotions. Chapter 3 introduces psychological theory that predicts how rebate offer characteristics influence the purchase and redemption behaviors that contribute to breakage. Chapters 4 through 9 present a series of experiments that test the hypotheses by examining the purchase and post-purchase behavior of rebate buyers and investigating how and when breakage occurs. Finally, chapter 10 discusses the findings and implications for regulators and managers.

PAGE 16

CHAPTER 2 TYPOLOGY OF REBATE PURCHASES How Rebates Differ From Coupons Rebates offer consumers an opportunity to receive a monetary reward for buying a promoted product, provided they expend some effort to receive the reward (Jolson et al. 1987; Rothschild 1987; Tat et al. 1988). Previous research has treated rebates and coupons as similar tactics for discriminating among consumers who differ in their price sensitivity (Blattberg and Neslin 1990; Rothschild 1987). Although rebates may be similar to coupons in their ability to price discriminate, the temporal arrangement of purchase, effort, and receipt of the reward differ for rebates and coupons (Soman and Gourville 2002). For example, consumers who use a typical coupon have already expended the effort needed to redeem it (e.g., obtaining the coupon from packaging or advertising), and will receive the reward at the time of purchase. In contrast, consumers whose choice is influenced by a rebate have not yet expended the effort required to obtain the reward, and will receive the reward only if they expend the necessary effort after purchase. Thus, rebates have the added features of a delay between purchase and completion of the effort, which I term redemption delay, and a delay between completion of the effort and receipt of the reward, which I term reward delay. The presence of these delays has resulted in rebates being referred to as delayed incentives (Soman 1998). The separation of purchase and redemption is an important distinguishing factor between rebates and coupons because it presents the opportunity for consumers to purchase a rebate offer but not redeem it (i.e., breakage). 4

PAGE 17

5 Classifying Purchases Classifying rebate purchases provides the starting point for understanding breakage. As shown in figure 1, rebate purchases can be classified into rebate-dependent purchases (i.e., incremental sales that occur due to the presence of the rebate), and rebate-independent purchases (i.e., baseline sales that occur even when the rebate is absent). Rebate-dependent purchases can be further segmented into purchases by buyers who estimate their probability of redeeming is sufficient to merit purchase (i.e., probabilistic redeemers) and purchases by buyers who are certain of redemption (i.e., intended redeemers that estimate their probability of redemption is 100%). Rebate-independent purchases can be further segmented into purchases by buyers who have no intention to redeem the rebate (i.e., intended non-redeemers) and purchases by buyers who will take advantage of the rebate if the opportunity becomes available (i.e., opportunistic redeemers). Implications for Measuring Redemption Rates The post-purchase outcomes of each purchase classification are important in understanding how redemption rates are calculated and how they can be non-diagnostic (and potentially misleading) as a measure of rebate program effectiveness. The rebate fulfillment industry relies on redemption rates (i.e., the number of redemptions divided by total units sold) to monitor the effectiveness of their rebate programs and to create guidelines for program implementation. Rebate promotions that achieve sales objectives and generate redemption rates below guidance are deemed successful, while promotions that generate redemption rates above guidance are deemed unsuccessful and tend to be

PAGE 18

6 discontinued immediately. 1 For example, consumer electronics manufacturers are often told to expect redemption rates of 30% to 35% and hope for rates as low as 20% (Norr 2000), and to discontinue promotions that generate redemption rates above 50% (Norberg 2002). As illustrated in figure 1, intended redemption and opportunistic redemption are both included in the calculation of redemption rates, while breakage and intended non-redemption contribute to lower redemption rates. The inclusion of opportunistic redemption in the calculation of redemption rates reveals how redemption rates can be non-diagnostic and potentially misleading as a measure of rebate effectiveness. Consider the three rebate offers shown in figure 2. These offers generate incremental sales of 20%, 40%, and 60% as a percentage of total sales, and exhibit redemption rates of 30%, 30%, and 40% respectively. According to industry rules of thumb, offer C is least desirable due to its high redemption rate, while offers A and B are equally desirable since they yield equivalent redemption rates. Yet, these conclusions are incorrect. In fact, offer C is most desirable since it contributes the highest proportion of incremental sales and exhibits the lowest redemption rate on incremental purchases. Furthermore, in addition to having a lower redemption rate on incremental purchases, offer B is more desirable than offer A since it does not pay rebate rewards on base sales to opportunistic redeemers. This example reveals that redemption rates can be a non-diagnostic and potentially misleading measure of rebate effectiveness whenever redemption rates are a) diluted by purchases by intended non-redeemers, and/or b) inflated by redemptions by opportunistic redeemers. Given that advertising effects and habitual purchases generate a substantial 1 Source: BDS Marketing, DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing Association, TCA Fulfillment Services.

PAGE 19

7 number of rebate-independent purchases (i.e., consumers purchasing due to influences other than the rebate itself), it is likely that redemption rates observed in the marketplace are affected by some degree of dilution and/or inflation. 2 Although these shortcomings are difficult to remedy, they are mentioned to emphasize that the typology of rebate purchases in figure 1 can aid managers in recognizing that the industrys preoccupation with redemption rates can be problematic. Implications for Measuring Breakage Breakage is an important outcome for marketers and the rebate fulfillment industry because it represents incremental sales revenue that does not require the payment of redemption rewards. It is also an important outcome for consumer protection advocates who wish to determine if breakage occurs on a scale that warrants regulation. However, research has yet to obtain accurate measures of breakage. Marketers currently use redemption rates as a proxy for breakage because aggregate sales data cannot distinguish rebate-dependent purchases from rebate-independent purchases. As discussed previously, redemption rates can be a non-diagnostic and potentially misleading measure of rebate effectiveness. Thus, one of the goals of this research is to experimentally isolate rebate-dependent purchases to obtain direct measures of breakage. Influencing Breakage From the perspective of the issuing firm, successful rebate programs should increase the number of rebate-dependent purchases (i.e., incremental sales) without encouraging redemptions by opportunistic redeemers. Successful rebate programs should also be structured in a manner that limits intended redemption (i.e., allow for breakage). 2 Redemption rates typically range from 2% to 40% depending on the product category.

PAGE 20

8 Attempts to influence breakage usually involve manipulations of rebate redemption periods (i.e., the amount of time given to redeem the rebate), and redemption effort (i.e., the amount of effort required to redeem). For example, over the past 20 years, the rebate industry has shortened redemption periods from a mode of 60 days to a mode of 30 days in an effort to increase breakage. 3 This trend is continuing with 15-day and 7-day redemption periods becoming increasingly common. Another trend designed to increase breakage is increasing redemption effort (e.g., requiring multiple proofs of purchase, making redemption procedures more complex, etc.). As discussed previously, these trends have been guided primarily by industry intuition about effective strategy as opposed to formal research. Research on Breakage Although there has been no direct documentation of breakage, three pieces of evidence are consistent with breakage and highlight the need for additional research. First, Jolson et al. (1987) surveyed 294 consumers and report that 72 % of light rebate users (people who reported one or two rebate-triggered purchases during the past 12 months) and 39.5% of frequent rebate users (people who reported more than two rebate-triggered purchases during the past 12 months) admit to breakage on over 30% of their rebate-triggered purchases of small home appliances costing less than $100. The results suggest that breakage is a common and recurring problem among rebate buyers. However, given that these data are self-reported, memory biases (i.e., forgetting) and self-presentation biases (i.e., not wanting to admit to breakage) are likely to have created a downward bias in reporting breakage. Conversely, it is questionable whether 3 Source: BDS Marketing, DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing Association, TCA Fulfillment Services.

PAGE 21

9 respondents were capable of retrospectively separating rebate-triggered (i.e., rebate-dependent) purchases from rebate-independent purchases, which may have created an upward bias in reporting breakage. Second, Dhar and Hoch (1996, Study 1), find that in-store coupons result in a 309% average increase in incremental sales across five product categories, but that coupon redemption averaged only 73% of the incremental sales. The implication is that there is a minimum of 27% breakage. However, the behaviors leading to breakage are fundamentally different for rebates and coupons due to the difference in the temporal order of effort and reward discussed earlier. Third, Soman (1998) finds that people are willing to forego a $1 payment for completing a survey in order to receive a larger delayed incentive payment (either $2 or $4 depending on the condition) for completing a second survey (either four or eight pages), yet fail to return the second survey at a later date (i.e., in exactly two weeks, exactly four weeks, or at any time within 4-weeks). Redemption rates ranged from a low of 23.5% for those who were offered $4 for completing an 8-page survey, to a high of 44.5% for those who were offered $4 for completing a 4-page survey. Soman (1998) proposed that people underweight the future effort required to redeem relative to the savings offered by the reward. However, two aspects of the study differ from most real-world rebate promotions. First, participants were aware of the effort requirements (i.e., the second survey) when deciding whether to choose the delayed incentive offer, whereas most rebate offers do not disclose the effort requirements until after purchase. This may have generated a downward bias in the proportion of participants that agreed to the delayed incentive offer, and raises questions about how and when effort influences breakage. Second, no financial investment was

PAGE 22

10 required from participants who chose the delayed incentive offer (i.e., forgoing a $1 reward does not involve the same financial cost as purchasing a product offering a rebate). Thus, participants may not have had the same motivation to redeem as those who purchase real-world rebates. This may have generated a downward bias in the redemption rate and raises questions about how the cost/reward relationship influences breakage. Based on the research conducted to date, the behaviors surrounding breakage are not well understood. Several explanations for breakage have been suggested in passing, including overconfidence, procrastination, and forgetting, yet research has yet to directly examine these explanations in detail. It is also unclear whether consumers repeatedly fall victim to breakage. Although there is some evidence to suggest that breakage is a recurring problem for some consumers (e.g., Jolson et al. 1987), the evidence implies that rebate buyers do not learn from past experience and/or minimize the relevance of failing to redeem previous offers when considering the purchase of a rebated product. The current research will address these issues.

PAGE 23

11 REBATE-INDEPENDENT PURCHASE Included in redemption rate OPPORTUNISTIC REDEEMERS INTENDED NON-REDEEMERS OPPORTUNISTIC REDEMPTION INTENDED NON-REDEMPTION PROBABILISTIC REDEEMERS INTENDED REDEEMERS INTENDED REDEMPTION BREAKAGE REBATE-DEPENDENT PURCHASE Figure 1. Typology of rebate purchases

PAGE 24

12 Offer A Offer B Offer C Classification Segment PUR RDN PUR RDN PUR RDN Intended Non-Redeemers .7 .0 .6 .0 .4 .0 Rebate-independent purchases (Base sales) Opportunistic Redeemers .1 .1 .0 .0 .0 .0 Probabilistic Redeemers .0 .0 .2 .1 .4 .2 Rebate-dependent purchases (Incremental sales) Intended Redeemers .2 .2 .2 .2 .2 .2 Total 1.0 .0 1.0 .0 1.0 .0 Incremental sales as a percentage of total sales 20% 40% 60% Redemption rate 30% 30% 40% Redemption rate on incremental sales 100% 75% 67% Breakage 0% 25% 33% PUR represents each segments contribution to total purchases. RDN represents each segments contribution to total redemptions Figure 2. Impact of post-purchase outcomes on redemption rates

PAGE 25

CHAPTER 3 CONCEPTUAL AND THEORETICAL FOUNDATIONS One conceptual insight into understanding breakage is provided by Somans (1998) assertion that the decision to purchase a product offering a delayed incentive can be independent of the decision to redeem the delayed incentive, and that the structure of a delayed incentive offer encourages people to overweight the benefits (i.e., the reward), and underweight the costs (i.e., the effort). Although it is questionable whether the behavioral decision theory perspective of weighting is central to explaining breakage, I adopt the perspective that purchase and redemption represent separate behaviors and consider psychological mechanisms that influence each of these behaviors. Explanations for breakage were obtained from the survey study conducted by Jolson et al. (1987), the delayed incentives experiments (Soman 1998), and the numerous cases of breakage reported in the popular press. A literature review was conducted to identify relevant psychological theory that could explain how rebate offer characteristics influence consumers purchase and redemption behavior to create breakage. In accordance with the definition of breakage, and as illustrated in figure 3, changes in breakage rates may be attributed to three types of factors. First, there may be pre-purchase factors that encourage consumers to make rebate-dependent purchases by influencing consumers subjective probability of redeeming the rebate (i.e., redemption confidence), but that have little influence on the objective (i.e., actual) probability of redeeming the rebate. Second, there may be post-purchase factors that directly affect the objective probability of redeeming a rebate by influencing the likelihood of initiating 13

PAGE 26

14 and/or completing the redemption process. Third, there may be repeat-purchase factors that affect how past experience with rebate redemption influences the likelihood of making repeat rebate-dependent purchases. Pre-Purchase Factors Redemption Confidence Research in psychology has demonstrated that people are often highly confident when assessing the extent of their own knowledge, when assessing their traits and abilities, and when making predictions about the future (see Alba and Hutchinson 2000 for a review of knowledge calibration). Most relevant for understanding rebate-dependent purchase behavior is the tendency for people to be highly confident about the likelihood of performing future tasks (e.g., Griffin, Dunning, and Ross 1990; Hoch 1985; Vallone, Griffin, Lin, and Ross 1990). For example, when people are asked to provide a subjective probability estimate (i.e., a confidence estimate) of performing a future task, the average subjective probability estimate of respondents is often considerably higher than the objective response rate (i.e., the proportion of respondents that actually perform the behavior). This difference between confidence and the objective response rate is overconfidence. Two key findings in the overconfidence literature can inform us about rebate purchase and redemption behavior. First, people are often willing to use their subjective probability estimates (i.e., confidence) to support their decisions to engage in a future behavior (e.g., Vallone et al. 1990). This implies that influencing prospective buyers subjective probability of redeeming a rebate (i.e., redemption confidence) may be an effective means of encouraging rebate-dependent purchases. Second, high confidence is often driven by contextual factors that increase peoples confidence estimates above the

PAGE 27

15 objective response rate. Put differently, contextual factors are capable of influencing confidence estimates while having little effect on the objective probability of performing the behavior (e.g., Griffin et al. 1990; Vallone et al. 1990). Stated formally, H 1 : Increasing a consumers redemption confidence will increase the probability of rebate-dependent purchase. H 2 : Increasing a consumers redemption confidence will result in a less than proportionate increase in the objective probability of redeeming the rebate. A review of the overconfidence literature identified two contextual antecedents of overconfidence that correspond to rebate offer characteristics and may encourage rebate-dependent purchases: a) the desirability of future outcomes, and b) temporal distance. Desirability Studies in overconfidence have demonstrated that people become more confident in their personal forecasts as outcomes become more desirable (e.g., Pulford and Colman 1996). It is believed that the wishful thinking evoked by an optimistic bias toward a desirable outcome may restrict or interfere with ones ability to consider situational factors that could interfere with achieving the desired outcome (Griffin et al. 1990; Hoch 1985; Vallone et al. 1990). Applied to rebate-dependent purchase, desirability predicts that redemption confidence will vary as a function of the attractiveness of the rebate offer. Thus, as the size of reward increases (relative to the list price), so should the consumers redemption confidence. H 3 : Increasing the size of the reward will increase a. the consumers redemption confidence. b. the probability rebate-dependent purchase.

PAGE 28

16 Temporal Distance Overconfidence research has demonstrated that people become more confident in their personal forecasts as the temporal distance (i.e., delay) between their prediction and the occurrence of the predicted event increases (e.g., Gilovich, Kerr, and Medvec 1993; Milburn 1978; Wright and Ayton 1992). The influence of temporal distance may be a consequence of peoples failure to consider potential barriers to future success. For example, research has shown that predictions made at a temporal distance generate more reasons for why one may succeed than for why one may fail, with the pattern reversing when people make predictions on the day of the task (Gilovich et al. 1993). This pattern is consistent with findings generated by research on temporal construal (Liberman and Trope 1998; Trope and Liberman 2000). Applied to rebate-dependent purchase, temporal distance predicts that redemption confidence will vary as a function of the amount of time between purchase and the deadline for redemption (i.e., the length of the redemption period). Thus, as the length of the redemption period increases, so should the consumers redemption confidence. H 4 : Increasing the length of the redemption period will increase a. the consumers redemption confidence. b. the probability of rebate-dependent purchase. Post-Purchase Factors Post-purchase factors influence the consumers likelihood of initiating and/or completing the redemption process and directly influence the objective probability of redeeming. Post-purchase factors include procrastination, prospective forgetting, and redemption effort.

PAGE 29

17 Procrastination Procrastination occurs when individuals postpone tasks until tomorrow without foreseeing that when tomorrow comes, the required action will be delayed yet again (Akerlof 1991; Silver 1974). Procrastination has been reported to be pervasive within everyday activities (Ellis and Knaus 1977) and among normal adult populations (Harriott and Ferrari 1996). Furthermore, procrastination has been cited as a potential cause of breakage (e.g., Jolson et al. 1987; Soman 1998), although it has yet to be formally investigated in a rebate context. One explanation for task-induced procrastination is that people avoid tasks that are perceived as more aversive or less attractive than alternative uses of ones time (Sigall, Kruglanski, and Fyock 2000). Research has demonstrated that procrastinators make poor estimates about the amount of time needed to complete activities (Buehler, Griffin, and Ross 1994; Lay 1988), commence tasks later than non-procrastinators (Ferrari 1993; Pychyl, Morin, and Salmon 2000; Sigall et al. 2000), take longer to complete tasks (Holmes 2002; Lay 1988), and fail to act on their intentions to perform tasks (Lay and Burns 1991). Procrastination is believed to persist because people (a) have a desire to avoid the activity, (b) make a decision to delay, (c) promise themselves to get to it later, (d) engage in substitute diversionary activities, and (e) make excuses to justify delays and exonerate themselves from blame (Knaus 2000). Research shows that deadlines help to mitigate procrastination by facilitating the completion of tasks. For example, Ariely and Wertenbroch (2002) observed that students were willing to self-impose deadlines to overcome procrastination, but that performance was higher when deadlines were externally imposed by the instructor than when self-imposed by students. Research has also demonstrated that the longer people are given to complete a task, the greater the likelihood of procrastination and failing to complete the

PAGE 30

18 task. For example, Tversky and Shafir (1992) offered students $5 for answering and returning a questionnaire within 5 days, 3 weeks, or within no deadline and observed return rates of 60%, 42%, and 25% respectively. Applied to rebate redemption, these findings predict that imposing immediate redemption deadlines will mitigate procrastination and increase the objective probability of redeeming. Put differently, imposing less immediate deadlines (i.e., increasing the length of the redemption period) will foster procrastination and decrease the objective probability of redeeming. H 5 : Increasing the length of the redemption period will a. increase procrastination. b. decrease the objective probability of redeeming the rebate. Prospective Forgetting Prospective forgetting examines peoples ability to retrieve and act on previously formed intentions (e.g., Krishnan and Shapiro 1999; Marsh, Hicks, and Watson 2002; Shapiro and Krishnan 1999). This research has shown that consumers frequently forget to act on previously formed intentions and that successful completion of intentions is contingent on a prospective memory component (i.e., remembering that an intention was formed) and a retrospective component (i.e., remembering the content of the intention). This research is relevant to rebate redemption because it has shown that delays that separate the formation of an intention and an opportunity to act on the intention are likely to foster forgetting (Shapiro and Krishnan 1999). Thus, the delay that is imposed between the purchase and redemption of a rebate offer provides the opportunity for prospective forgetting to play a role in breakage. Research has shown that increasing the length of the delay can encourage forgetting (Scher and Ferrari 2000). Applied to rebate redemption, prospective forgetting predicts that increasing the delay between purchase and

PAGE 31

19 redemption (i.e., increasing the length of the redemption period) will increase the likelihood of prospective forgetting and decrease the objective probability of redeeming. H 6 : Increasing the length of the redemption period will a. increase prospective forgetting. b. decrease the objective probability of redeeming the rebate. Redemption Effort Redemption effort (i.e., the effort required to successfully redeem a rebate) is often cited as a reason why consumers fail to redeem rebate offers (e.g., Jolson et al. 1987; McLaughlin 2002; Norr 2000; Shim 2002; Soman 1998; Spencer 2002). Redemption effort can contribute to breakage by causing consumers to give up before completing the redemption process and by preventing redemption when consumers make simple errors such as discarding packaging or sales receipts that are required to redeem. In addition to contributing to breakage, redemption effort can also lower redemption rates by increasing rejection rates on submitted applications 1 (i.e., failing to submit UPC codes). Interviews with marketing service firms confirm that increasing redemption effort is a popular strategy for encouraging breakage 2 In fact, it is widely believed that redemption effort is the greatest contributing factor in creating breakage. As explained by a former marketing consultant, If you have to take a knife and cut through heavy cardboard to get a bar code, redemption rates drop precipitously (Norr 2000, p. 21). As stated by Federal Trade Commission Director J. Howard Beals, Some companies are quick to offer attractive rebates, but often make them so difficult to redeem that consumers simply give up 1 Rejection rates typically range between 1% and 10% of submitted applications and can reach as high as 25%. Rejected rebate applications do not qualify as breakage because breakage is defined as failing to submit a rebate application. Rejected applications are subtracted from the total number of applications when calculating redemption rates. 2 Source: BDS Marketing, DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing Association, TCA Fulfillment Services.

PAGE 32

20 (Shim 2002, p.1). There is also experimental evidence showing that increasing redemption effort increases task aversiveness and breakage (Soman 1998). H 7 : Increasing redemption effort will decrease the objective probability of redeeming the rebate. Repeat-Purchase Factors Repeat-purchase factors influence how past experience with rebate redemption affect the likelihood of making repeat rebate-dependent purchases. Although there is evidence to suggest that breakage is a frequent and recurring problem for some consumers (e.g., Jolson et al. 1987), research has yet to examine whether people make repeat rebate-dependent purchases after failing to redeem a previous offer. Past Redemption Behavior One factor which may influence repeat purchase behavior is the extent to past redemption behavior (i.e., redeeming or failing to redeem a previous offer) influences a consumers repeat purchase behavior. Successful redemption may encourage repeat rebate-dependent purchases. Conversely, failing to redeem may discourage repeat rebate-dependent purchases. H 8 : Failing to redeem a previous offer will decrease the probability of repeat rebate-dependent purchase.

PAGE 33

Redemption Confidence a) Desirability b) Temporal distance PRE-PURCHASE FACTORS REBATE-INDEPENDENT PURCHASE REBATE-DEPENDENT PURCHASE POST-PURCHASE FACTORS Procrastination Prospective forgetting Redemption effor t OPPORTUNISTIC REDEEMERS INTENDED NON-REDEEMERS OPPORTUNISTIC REDEMPTION INTENDED NON-REDEMPTION PROBABILISTIC REDEEMERS INTENDED REDEEMERS INTENDED REDEMPTION BREAKAGE REPEAT-PURCHASE FACTORS Past redemption behavior 21 Included in redemption rate Figure 3. Factors that contribute to breakage

PAGE 34

CHAPTER 4 INTRODUCTION TO EXPERIMENTS Four experiments were conducted to examine consumers purchase and post-purchase behavior with a real rebate offer and observe how reward, redemption period, and redemption effort influence purchase, procrastination, prospective forgetting, redemption and breakage. The experiments isolate rebate-dependent purchases and measure breakage by presenting consumers with an opportunity to purchase a rebate offer that is unattractive relative to other purchase alternatives unless the consumer redeems the rebate (i.e. the only reason to purchase the offer is to take advantage of the rebate, thus eliminating rebate-independent purchases). Experiment 1 examined how manipulating offer characteristics (i.e., reward, redemption period and redemption effort) influences purchase, procrastination, breakage, and redemption. Experiment 2 examined the relationship between consumers redemption confidence and the likelihood of purchase and redemption. Experiment 3 was conducted with rebate buyers from experiment 1 and examined how past experience with rebate redemption influences the likelihood of repeat rebate-dependent purchase. A post-test survey was administered to participants of experiments 1 and 3 to obtain descriptive data on rebate buyers purchase and redemption behavior. Finally, experiment 4 was conducted to assess the direct effect of reward on redemption behavior. 22

PAGE 35

CHAPTER 5 EXPERIMENT 1 Design Experiment 1 consisted of an opportunity to make a rebate-dependent purchase and an opportunity to redeem the rebate. To maximize the external validity of the experiment, consumers were shown the same information as is shown with a typical in-store rebate promotion. The reward and redemption deadline were clearly indicated on the offer, but no information regarding the redemption process (i.e., redemption effort) was provided. As a result, the purchase phase of the experiment employed a two (reward: $6 or $9) by three (redemption period: 1-day, 7-days, or 21-days) between-subjects design, producing six experimental conditions. Consistent with typical rebate offers, consumers became aware of the effort required to redeem upon initiating the redemption process. Thus, the redemption phase of the experiment crossed the reward and redemption period factors with an additional between-subjects factor (redemption effort: low or high), resulting in 12 experimental conditions. Procedure Phase 1: Purchase Twelve hundred and thirty-three students from an undergraduate subject pool participated in the experiment for course credit and were randomly assigned to one of the six purchase conditions, resulting in 205 or 206 participants per cell. A large sample size was required since it was expected that a relatively small proportion of participants would purchase the rebate offer, resulting in smaller sample sizes for the redemption analysis. 23

PAGE 36

24 The purchase opportunity followed a bogus computer-administered market research questionnaire that was used to disguise the purpose of the experiment and prevent hypothesis guessing. Upon entering the experiment lab, participants were informed that a well-known Hollywood movie studio had experienced considerable financial losses due to movies that did poorly at the box office, and was conducting a pilot study to learn about peoples movie preferences. A cover story that was developed to disguise the study and justify the need for some participants to return for a second and third session, which would serve as experiment 3 and the post-test respectively. The cover story read as follows: A well-known Hollywood movie studio is conducting this pilot study at five colleges across the country to get a better idea of the types of movies college audiences want to see. You will be asked a series of questions about your movie preferences, such as your favorite genres, actors, and movies. Your responses will be aggregated and used to develop 10 new movie concepts to be tested in two follow-up studies to take place later this semester. Fewer people are needed for the follow-up studies, so the computers will randomly select people at the end of todays study and ask them to participate in the follow-up studies. If you are asked to participate, please indicate whether or not you want to take part in the two follow-up studies. Remember that selection is random, so please do not feel badly if you are not selected. Finally, the studio is offering pairs of discounted movie tickets to all participants at the end of the study as a thank you for participating today. Simply indicate which ticket offer you wish to purchase, or choose decline if you do not wish to purchase tickets. The tickets are valid for one year, and are valid for any movie and show time at the local movie theatre. Please remember that purchase is optional and is completely up to you. After the completing the bogus market research study, participants were presented with the computer-administered purchase opportunity shown in figure 4. The choice set consisted of three options: a) purchase two movie tickets for a discounted price of $11, b) purchase two movie tickets for $13 less a $6 (or $9) mail-in rebate, or c) decline both offers and purchase no tickets. Purchasing the rebate offer represented a rebate-dependent

PAGE 37

25 purchase since the offer was attractive only if redeemed (i.e., the $13 before-rebate price that one pays if the rebate offer is not redeemed is greater than the $11 non-rebated offer and the $12 regular price that students pay at the box office). Redemption period was manipulated as one, seven, or 21 days from the date of purchase by changing the rebate must be redeemed by date that was clearly marked on the rebate offer. The two ticket offers were presented on the top and the middle of the computer screen and the placement was randomized to guard against order effects. After making their choice, participants were presented with a confirmation screen that emphasized the financial consequences of their decision and allowed them to choose a different option. For example, those who chose the rebate offer were shown the offer a second time, accompanied by the message You have chosen the rebate offer. You will pay $13 today. Please confirm your choice. Following the choice confirmation, participants who declined to purchase tickets or who chose the $11 non-rebated offer completed a short computer-administered filler task while the computer asked participants who chose the rebate offer to participate in the follow up studies to be conducted later that semester. Once completed, participants who declined to purchase tickets exited the room and those who chose the $11 offer paid for and received their tickets and exited the room. Those who chose the $13 rebate offer paid for and received their tickets. After making their payment, rebate buyers in each condition were randomly assigned to either the low-effort or high-effort redemption effort condition and were given a rebate purchase receipt that was required to redeem the rebate. This produced a balanced number of rebate buyers in the high-effort and low-effort cells of each purchase condition.

PAGE 38

26 Phase 2: Redemption The redemption procedure was designed to measure redemption, breakage, and the extent to which rebate buyers procrastinated at each stage of the redemption process. This was achieved by using a rebate redemption website which, unbeknownst to participants, tracked each buyers progress with the redemption process. To maintain external validity, the rebate redemption website and redemption requirements were modeled after the rebate redemption website of a well-known national electronics retailer. The purchase receipt given to each rebate buyer listed the expiry date for mailing the rebate application 1 and instructed the buyer to visit the rebate website listed on the receipt to obtain the redemption form. To redeem the rebate, rebate buyers were required to access the rebate website and enter their receipt code. Participants in the low-effort condition were required to complete five steps: 1) complete the form by typing their name, address, ticket numbers, receipt code, and purchase date, 2) print the completed form, 3) complete the on-line customer service survey, which served as a manipulation check of redemption effort, 4) photocopy the UPC code on the back of both movie tickets, and 5) mail the redemption form, original receipt and photocopies to the address listed on the redemption form prior to the expiry date indicated on their purchase receipt (i.e., either 1, 7, or 21 days from the date of purchase). Participants in the high-effort condition were required to complete the same five steps as those in the low-effort condition, but were also required to read a terms and conditions agreement and re-enter the information in step 1 for verification purposes, effectively requiring step one to be completed twice. 1 The expiry date printed on the purchase receipt matched the must be redeemed on or before date of the rebate offer shown in the choice set.

PAGE 39

27 Phase 3: Responding to Rebate Applications All rebate applications were examined to ensure that they conformed to the redemption requirements listed on the website. Applicants who successfully redeemed prior to the deadline were accepted and received a rebate check ($6 or $9) with a brief letter stating that their application had been accepted. All acceptance letters were mailed five business days after the postmark date of the application to standardize the response time across participants. Applicants who submitted their application prior to the deadline but who failed to include either the original purchase receipt or the photocopy of the UPC code received a letter informing them that they had been rejected and inviting them to re-apply with the required materials prior to the deadline indicated on their purchase receipt. All rejection letters were mailed two business days after the postmark date of their application to standardize the response time and maximize participants chances for re-application. Applicants who submitted their requests after the deadline received a letter indicating that their request had been rejected since it was postmarked after the deadline. Dependent Measures Purchase Consumers choice of ticket offer (rebate offer / non-rebated offer / decline) was recorded and analyzed using multinomial logistic regression. Breakage and Redemption Login (yes / no) measured whether rebate buyers initiated the redemption process by logging on to the website and provided a means of measuring the degree of breakage attributable to failing to initiate the redemption process. Applied (yes / no) measured whether rebate buyers completed the redemption process by mailing a rebate application and provided a means of measuring the degree of breakage attributable to failing to

PAGE 40

28 complete the redemption process. A third measure, accepted (yes / no), measured whether the rebate application conformed to the redemption requirements and provided a means of measuring the overall redemption rate as well as the rejection rate on rebate applications. 2 Login and applied are of most interest since breakage is defined as failing to apply for the rebate. There are no a-priori hypotheses regarding errors in rebate applications that could affect redemption and rejection rates. Login, applied and accepted were analyzed using binary logistic regression. Procrastination Login delay measured procrastination associated with initiating the redemption process by measuring the delay (in days) between the purchase date and the date on which the buyer logged on to the rebate website. Mail delay measured procrastination associated with completing the redemption process by measuring the delay (in days) between logging on to the website and the postmark date of the mailed rebate application. Login delay and mail delay were analyzed using ANOVA. Redemption Effort Manipulation Check Three measures of redemption effort were obtained to determine the efficacy of the redemption effort manipulation. After completing and printing the redemption form on the rebate website, rebate buyers responded to two customer service questions on a seven-point semantic differential scale: Using this website was: 1 = easy / 7=difficult; and 1 = not at all effortful / 7 = extremely effortful. A third measure was obtained by tracking the elapsed time each participant spent on the website. A MANOVA revealed a significant effect of redemption effort on the difficulty rating (M low effort = 2.08, M high effort 2 The discrepancy between applied and accepted is caused by rejected (late or incomplete) rebate applications.

PAGE 41

29 = 3.65) (F (1, 137) = 21.34, p < .001), effort rating (M low effort = 2.86, M high effort = 4.75) (F (1, 137) = 29.92, p < .001), and time spent on the website (M low effort = 4.9 minutes, M high effort = 9.0 minutes) (F (1, 137) = 59.33, p < .001). Reward, redemption period and all interactions were not significant for all three measures (all ps >.40), confirming that the reward and redemption period manipulations did not influence perceptions of redemption effort. Results Purchase The purchase results for experiment 1 are shown in figure 5. A higher proportion of consumers purchased the rebate offer when the reward increased from $6 to $9 and when the redemption period increased from 7 to 21 days. Multinomial logistic regression 3 indicated that the reward by redemption period interaction was not significant ( 2 = 1.40, p > .80). Further analysis revealed significant main effects of reward ( 2 = 15.11, p < .001) and redemption period ( 2 = 12.82, p < .05). The odds ratio for reward indicates that the estimated odds of purchasing the rebate offer are 1.9 times higher when the rebate is $9 than when it is $6. Similarly, the odds ratio for redemption period indicates that the estimated odds of purchasing the rebate offer are 2.0 times higher when the redemption period is 21 days than when it is either seven days or one day. 3 In accordance with the MNL model specification guidelines proposed by Ben-Akiva and Lerman (1985), a nested logit model was specified before graduating to a MNL model to ensure that the MNL model was a better predictor of purchase compared to the nested model, and to ensure that the error terms of the three purchase alternatives satisfied the IID (independent and identically distributed) assumption. The nested model specified purchase and no purchase as discrete alternatives, with the rebate and non-rebate offers nested under purchase.

PAGE 42

30 Redemption Redemption analyses were conducted on the 184 consumers who purchased the rebate offer. Sixty-seven percent of rebate buyers applied for the rebate, resulting in a breakage rate of 33%. Eleven percent of the applications (8% of rebate buyers) had their rebate applications rejected, resulting in a redemption rate of 59%. The proportion of rebate buyers in each condition that applied for the rebate is shown in figure 6. Logit analyses found no significant two-way or three-way interactions on the proportion of buyers that logged on to the website, applied for the rebate, or were accepted. As predicted, increasing the length of the redemption period decreased the proportion of rebate buyers that applied for the rebate ( 2 = 5.12, p < .05). Thus, increasing the length of the redemption period has a positive influence on purchase behavior and a negative influence on redemption behavior. Unexpectedly, increasing redemption effort generated a marginally significant increase in the proportion of rebate buyers that applied for the rebate ( 2 = 3.03, p < .08). As predicted, increasing the reward did not have a significant effect on the proportion of rebate buyers that logged on to the website ( 2 = 0.14, p > .70), applied for the rebate ( 2 = 0.01, p > .90), or were accepted ( 2 = 0.50, p > .40). The estimated odds of applying for the rebate are 2.4 times higher when the redemption period is one day compared to 21 days, and 1.8 times higher when the redemption period is seven days compared to 21 days, and 1.7 times higher when consumers experience the high-effort redemption process compared to the low-effort redemption process. Procrastination Procrastination analyses were conducted on the 149 rebate buyers that initiated the redemption process by logging on to the website and the 123 rebate buyers that completed the redemption process by mailing a rebate application. The results for login

PAGE 43

31 delay and mail delay are shown in figures 7 and 8 respectively. As predicted, increasing the length of the redemption period significantly increased the extent to which rebate buyers procrastinated before logging on to the website (F (2, 137) = 20.07, p < .001). Planned contrasts revealed a significant difference between all levels of redemption period (all p's < .01). As predicted, increasing the reward had no effect on login delay (F (1, 137) = 0.23, p > .60). The effect of redemption effort on login delay was also non-significant (F (1, 137) = 0.05, p > .80), which was expected since effort was not manipulated until after rebate buyers logged on to the website. With regard to mail delay, the ANOVA revealed a significant redemption period by redemption effort interaction (F (2, 111) = 49.16, p < .001). As shown in figure 8, increasing the redemption period increased the extent to which rebate buyers in the low-effort condition procrastinated before mailing the rebate application. In contrast, rebate buyers in the high-effort condition tended to mail the rebate application within a few of days of visiting the website, regardless of the amount of time that was left before the redemption deadline. As with login delay, size of reward did not have a significant effect on mail delay (F (1, 111) = 0.07, p > .80). Discussion Purchase and Redemption Increasing the reward increased the number of rebate-dependent purchases, supporting H3b, but did not influence redemption, consistent with H2. Increasing the length of the redemption period from 7 days to 21 days also increased the number of rebate-dependent purchases, supporting H4b, and any increase in the redemption period increased procrastination and decreased the proportion of buyers that applied for the rebate, supporting H5. It appears that shorter redemption deadlines mitigate

PAGE 44

32 procrastination and decrease breakage by facilitating the completion of redemption applications that would otherwise go unfinished. Insights Regarding Breakage As discussed earlier, the challenges associated with isolating rebate-dependent purchases has made it difficult to determine the extent to which breakage occurs and how it is influenced by offer characteristics. The breakage rate of 33% indicates that breakage is a common post-purchase outcome that can lower redemption rates in a significant manner. As illustrated in figure 9, 19% of rebate buyers failed to log on to the website and an additional 14% failed to mail their applications. Thus, failing to initiate the redemption process accounts for about 58% of breakage. This finding is important because increasing redemption effort does not influence this form of breakage (i.e., rebate buyers did not become aware of the effort required until after logging on to the website). In fact, in opposition to H7, increasing redemption effort caused rebate buyers to complete the redemption process more quickly and generated a marginal increase in rebate applications. This suggests buyers in the high effort condition may have been more motivated to apply for the rebate in response to the high-effort redemption requirements. Failing to mail redemption applications is also a significant problem for consumers, and the breakage rate would have been greatly reduced had consumers been given the option of completing the entire redemption process on-line (i.e., no mail-in requirement).

PAGE 45

33 Figure 4. Purchase opportunity choice set 41 4 Reward was manipulated as either $6 (as shown) or $9, resulting in an after-rebate price of either $7 (as shown) or $4. Redemption period was manipulated by setting the must be redeemed by date to 1, 7, or 21 days after the date of purchase.

PAGE 46

34 17%24%16%3%5%16%10%8%5%6%6%0%5%10%15%20%25%30%1-Day7-Day21-DayRedemption PeriodProportion that Purchased $9 Rebate $6 Rebate $11 Non-Rebate $11 Non-Rebate Figure 5. Proportion of experiment 1 participants purchasing rebate and non-rebate offers by condition 77%64%69%75%65%52%70%63%50%81%80%75%40%50%60%70%80%90%1-Day7-Day21-DayRedemption PeriodProportion that Applied $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 6. Proportion of experiment 1 rebate buyers that applied for the rebate by condition

PAGE 47

35 1.24.31.72.13.22.90.34.00.40.13.8024681012141-Day7-Day21-DayRedemption PeriodDays $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 7. Experiment 1 login delay by condition 5.22.40.72.62.80.82.312.41.312.05.81.4024681012141-Day7-Day21-DayRedemption PeriodDays $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 8. Experiment 1 mail delay by condition

PAGE 48

36 Buyers Buyers Login to Website Login to Website Applied Applied Accepted Accepted Yes = 109 Yes = 109 (59% of buyers) (59% of buyers) Yes = 123 Yes = 123 (67% of buyers) (67% of buyers) Yes = 149 Yes = 149 (81% of buyers) (81% of buyers) Rejected = 14 Rejected = 14 (8% of buyers) (8% of buyers) N = 184 N = 184 (100%) (100%) No = 26 No = 26 (14% of buyers) (14% of buyers) No = 35 No = 35 (19% of buyers) (19% of buyers) Breakage Breakage by stage by stage Breakage = 19% Breakage = 19% Breakage = 14% Breakage = 14% Cumulative Cumulative breakage breakage Total breakage = 19% Total breakage = 19% Total breakage = 33% Total breakage = 33% Contribution Contribution to breakage to breakage % of breakage = 58% % of breakage = 58% % of breakage = 42% % of breakage = 42% Figure 9. Experiment 1 breakage by stage of redemption (n=184) Figure 9. Experiment 1 breakage by stage of redemption (n=184) 36

PAGE 49

CHAPTER 6 EXPERIMENT 2 Although experiment 1 demonstrated how rebate offer characteristics influence purchase, procrastination and redemption, it did not directly test the relationship between consumers subjective probability of redeeming and the likelihood of purchase and redemption as outlined in hypotheses H1-H4. Experiment 2 was designed to test this relationship by replicating the procedure used in experiment 1 and measuring consumers subjective probability of redemption 1 In doing so, experiment 2 will determine the extent to which consumers making rebate-dependent purchases are confident of redeeming, and whether manipulating reward and redemption period influence redemption confidence. Most importantly, experiment 2 will determine how the redemption confidence of rebate buyers relates to their redemption behavior (testing H2), and whether redeemers are any more or less confident of redeeming at the time of purchase than non-redeemers. Experiment 2 was also designed to obtain consumers base rate estimates of successful rebate redemption (i.e., objective probability estimates). Although H1 predicts that redemption confidence (i.e., the subjective probability of redeeming) drives rebate-dependent purchases, it would be informative to learn how consumers redemption confidence compares to their base rate estimates of successful redemption. In particular, it would be understandable for consumers to be highly confident of redeeming if they believe the base rate of redemption is high as well (i.e., a false consensus effect). 1 Subjective probability estimates were not obtained in experiment 1 because doing so would have compromised the external validity of the procedure. 37

PAGE 50

38 However, if consumers base rate estimates are relatively accurate, it would suggest that high levels of redemption confidence stem from an overly optimistic view of ones personal probability of redeeming rather than inaccurate perceptions of the base rate of redemption. Design and Procedure Experiment 2 used a design and procedure identical to that used in experiment 1 with two exceptions. First, the 7-day redemption period conditions were excluded from the design in an effort to reduce the number of subjects required for the study. This change was considered minor since retaining the 1-day and 21-day redemption period conditions would accommodate the full range of effects observed in experiment 1. This resulted in a two (reward: $6 / $9) by two (redemption period: 1-day / 21-day) by two (redemption effort: low / high) between-subjects design. As in experiment one, the redemption effort manipulation was delayed until rebate buyers logged on to the website, resulting in four purchase conditions. Four hundred subjects were randomly assigned to one of the four purchase conditions resulting in 100 subjects per cell. The redemption procedure used in experiment 2 was identical to that used in experiment 1. The second exception to the experiment 1 design was the addition of two measures administered to participants immediately following their purchase decision. The first additional measure was a subjective probability estimate adapted from Vallone et al. (1990). Those purchasing the rebate offer were asked, How confident are you that you will redeem the rebate offer? while those not purchasing the rebate offer were asked, How confident are you that you would redeem the rebate if you had purchased the rebate offer? Participants recorded their responses on a probability scale anchored by %: I am completely certain that I will (would) not redeem the rebate and %: I am

PAGE 51

39 completely certain that I will (would) redeem the rebate. The second measure was a base rate estimate. Participants were asked, Please estimate the percentage of rebate buyers that you think will redeem the rebate. Participants recorded their responses on a probability scale anchored by %: None of the rebate buyers will redeem the rebate and %: All of the rebate buyers will redeem the rebate. Results Purchase As illustrated in figure 10, the purchase results replicated the effects observed in experiment 1. Multinomial logistic regression indicated that the reward by redemption period interaction was not significant ( 2 = 1.29, p > .50). Further analysis revealed a significant main effect of reward ( 2 = 7.30, p < .05) and a marginally significant main effect of redemption period ( 2 = 4.52, p < .10). The estimated odds of purchasing the rebate offer are 2.8 times higher when the rebate is $9 than when it is $6, and 2.4 times higher when the redemption period is 21 days than when it is one day. Redemption Redemption analyses were conducted on the 54 participants who purchased the rebate offer. The results replicated the effects observed in experiment 1. Seventy-four percent of rebate buyers applied for the rebate, resulting in a breakage rate of 26%. Eight percent of the applications (6% of rebate buyers) had their rebate applications rejected, resulting in a redemption rate of 69%. The proportion of rebate buyers in each condition that applied for the rebate is shown in figure 11. Logit analyses found no significant two-way or three-way interactions on the proportion of buyers that logged on to the website, applied for the rebate, or were accepted. As in experiment 1, increasing the length of the redemption period resulted in a significant decrease in the proportion of rebate buyers

PAGE 52

40 that applied for the rebate ( 2 = 3.93, p < .05), and increasing redemption effort resulted in a marginally significant increase in the proportion of rebate buyers that applied for the rebate ( 2 = 3.55, p < .06). Increasing the reward did not have a significant effect on the proportion of rebate buyers that logged on to the website ( 2 = 0.17, p > .60), applied for the rebate ( 2 = 0.14, p > .90), or were accepted ( 2 = 0.03, p > .80). The estimated odds of applying for the rebate are 5.9 times higher when the redemption period is one day compared to 21 days, and 3.8 times higher when consumers experience the high-effort redemption process compared to the low-effort redemption process. Procrastination Procrastination analyses were conducted on the 47 rebate buyers that logged on to the website and the 40 rebate buyers that mailed a rebate application. The results for login delay and mail delay are shown in figures 12 and 13 respectively. As in experiment 1, increasing the length of the redemption period significantly increased the extent to which rebate buyers procrastinated before logging on to the website (F (1, 39) = 21.53, p < .001). As expected, the effects of reward (F (1, 39) = 0.01, p > .90) and redemption effort (F (1, 39) = 0.02, p > .80) on login delay were not significant. As in experiment 1, a significant redemption period by redemption effort interaction (F (1, 32) = 14.01, p < .001) was observed for mail delay. As shown in figure 13, increasing the redemption period reduced procrastination in the high-effort condition relative to the low-effort condition. Size of reward did not have a significant effect on mail delay (F (1, 32) = 0.59, p > .40). Probability Estimates Subjective probability (i.e., confidence) and base rate estimates for experiment 2 are shown in figure 14. Results indicate that rebate buyers are highly confident of

PAGE 53

41 redeeming at the time of purchase despite generating rather accurate estimates of the base rate of redemption. Rebate buyers confidence estimates (M = 93.5%) ranged from 50% to 100%, and over 60% of rebate buyers reported being 100% confident of redeeming the rebate. 2 The mean base rate estimate of rebate buyers (M = 53.7%) was close to the 60% redemption rate observed in experiment 1, and did not differ significantly from the base rate estimates of decliners (M = 51.0%) or of those who purchased the non-rebated offer (M = 55.6%), (F (2, 397) = 0.62, p > .50). As shown in figure 15, increasing the size of reward significantly increased consumers subjective probability of redeeming from a mean of 51.8% to a mean of 58.5%, (F (1, 396) = 4.28, p < .05), supporting H1 and H3. As shown in figure 16, increasing the length of redemption period also significantly increased consumers subjective probability of redeeming from a mean of 51.5% to a mean of 58.8%, (F (1, 396) = 5.10, p < .05), supporting H1 and H4. Increasing reward (F (1, 396) = 3.50, p < .06) and redemption period (F (1, 396) = 3.21, p < .08) resulted in marginally significant increases in consumers base rate estimates. With regard to the relationship between redemption confidence and redemption, binary logistic regression revealed that redemption confidence did not have a significant effect on whether rebate buyers logged in, ( = -.02, 2 = 0.14, p > .70), applied ( = -.08, 2 = 2.08, p > .10) or were accepted ( = -.10, 2 = 0.10, p > .70), supporting H2. In fact, as shown in figure 17, non-redeemers were just as confident at the time of purchase (M = 94.1%) as redeemers (M = 93.2%), (F (1, 52) = 0.09, p > .70). Base rate estimates were 2 The mean confidence estimate of rebate buyers (M = 93.5%) was significantly higher than that of those who declined the offer (M = 49.6%) and who purchased the non-rebated offer (M = 38.5%), (F (2, 397) = 55.00, both ps < .001), and multinomial logistic regression confirmed that confidence has a significant positive effect on the likelihood of purchasing the rebate offer ( = .10, 2 = 36.94, p < .001).

PAGE 54

42 also similar of redeemers (M = 54.0%) and non-redeemers (M = 53.1%), (F (1, 52) = 0.04, p > .80). Finally, a median split of rebate buyers based on their confidence estimates revealed similar redemption rates for the higher-confidence (67%) and lower-confidence (70%) groups ( 2 = 0.09, p > .70). Discussion In addition to replicating the results of experiment 1, experiment 2 provides direct evidence of the relationship between consumers redemption confidence and their purchase and redemption behavior. As predicted by H1, H3 and H4, increasing reward and redemption period increases redemption confidence and the probability of rebate-dependent purchase. However, as predicted by H2, increasing redemption confidence does not generate a proportionate increase in the objective probability of redeeming. Non-redeemers were just as confident of redeeming as redeemers, revealing that some poorly calibrated consumers fall victim to redemption overconfidence by making rebate-dependent purchases that they are unlikely to redeem. Interestingly, buyers are highly confident despite rather accurate base rate estimates. Although it is questionable whether consumers consider base rates during purchase, the results at least suggests that high redemption confidence does not stem from a false consensus effect. Thus, it is possible that rebate buyers exhibit a better than average effect in believing that their personal probability of redeeming is greater than that of others.

PAGE 55

43 14%22%6%12%3%2%4%3%0%5%10%15%20%25%30%1-Day21-DayRedemption PeriodProportion that Purchased $9 Rebate $6 Rebate $11 Non-Rebate $11 Non-Rebate Figure 10. Proportion of experiment 2 participants purchasing rebate and non-rebate offers by condition 100%73%100%83%86%55%67%50%40%50%60%70%80%90%100%1-Day21-DayRedemption PeriodProportion that Applied $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 11. Proportion of experiment 2 rebate buyers that applied for the rebate by condition

PAGE 56

44 3.20.43.40.73.30.63.20.3024681-Day21-DayRedemption PeriodDays $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 12. Experiment 2 login delay by condition 2.62.45.56.70.30.70.30.5024681-Day21-DayRedemption PeriodDays $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 13. Experiment 2 mail delay by condition

PAGE 57

45 93.553.749.651.038.555.630405060708090100Confidence EstimateBase Rate EstimateProbability Estimate Rebate Buyers Decliners Regular Buyers Figure 14. Experiment 2 confidence and base rate estimates by choice 51.849.758.553.430405060708090100Confidence EstimateBase Rate EstimateProbability Estimate $6 $9 Figure 15. Experiment 2 confidence and base rate estimates by reward

PAGE 58

46 51.549.858.853.330405060708090100Confidence EstimateBase Rate EstimateProbability Estimate 1-Day 21-Day Figure 16. Experiment 2 confidence and base rate estimates by redemption period 93.254.094.153.130405060708090100Confidence EstimateBase Rate EstimateProbability Estimate Redeemers Non-Redeemers Figure 17. Experiment 2 confidence and base rate estimates of redeemers and non-redeemers

PAGE 59

CHAPTER 7 EXPERIMENT 3 Although there is evidence to suggest that breakage is a recurring problem for some consumers (e.g., Jolson et al. 1987), this implies that rebate buyers do not learn from past experience and/or minimize the relevance of past redemption failures when considering the purchase of a rebate offer. Research has yet to experimentally investigate the repeat purchase behavior of rebate buyers to determine the extent to which breakage is a recurring outcome or to identify the conditions that lead to repeat instances of breakage. Experiment 3 was designed to address these questions. Design and Procedure Approximately five weeks after completing experiment 1, the 184 rebate buyers from experiment 1 were assigned to the same experimental conditions they experienced in experiment 1 and were provided with another opportunity to purchase and redeem the same offer. The experimental procedure and analyses were the same as in experiment 1, with four exceptions. First, the bogus market research survey that preceded the purchase opportunity consisted of new questions about movie preferences to maintain the face validity of the cover story. Second, participants were now aware of the redemption requirements if they had attempted to redeem the rebate in experiment 1 (i.e., rebate buyers that did not log on to the website in experiment 1 remained unaware of the redemption requirements). Third, those declining repeat purchase were asked why they did not purchase the rebate offer. Finally, the analysis included an additional predictor 47

PAGE 60

48 variable, past redemption behavior, which indicated whether the rebate buyer redeemed the rebate in experiment 1. Results Repeat Purchase As shown in figure 18, the purchase results closely resemble the redemption results of experiment 1. Multinomial logistic regression found no significant interactions on the proportion of buyers making a repeat purchase. Further analysis revealed significant main effects of past redemption behavior ( 2 = 32.71, p < .001) and redemption effort ( 2 = 9.88, p < .01). The estimated odds of repeat purchase are 10.8 times higher for buyers that redeemed the offer in experiment 1 than for buyers that did not redeem, and 3.2 times higher for buyers that experienced the low-effort redemption process than for buyers that experienced the high-effort redemption process. Thus, failing to redeem the offer and experiencing high-effort redemption requirements drastically reduced the likelihood of repeat purchase. In fact, only 8% of non-redeemers made a repeat purchase versus 41% of redeemers. The self-reported reasons for not making a repeat purchase were analyzed using multinomial logistic regression and are summarized in table 1. Past redemption behavior was the only factor to have a significant effect on the reasons given for not making a repeat purchase ( 2 = 55.28, p < .001). Failing to redeem the previous offer was the most common response given by non-redeemers who declined repeat purchase. Redemption Redemption analyses were conducted on the 51 rebate buyers that made repeat purchases. Eighty-four percent of rebate buyers applied for the rebate, resulting in a breakage rate of 16%, well below the rates of breakage observed in experiments 1 (33%) and 2 (26%). Four percent of the applications (5% of repeat rebate buyers) had their

PAGE 61

49 rebate applications rejected, resulting in a redemption rate of 80%, well above the redemption rates observed in experiments 1 (59%) and 2 (69%). Despite the higher redemption rate and lower rate of breakage, the redemption results illustrated in figure 19 are consistent with the effects observed in experiments 1 and 2. Logit analyses found no significant interactions on the proportion of buyers that logged on to the website, applied for the rebate, or were accepted. Increasing the length of the redemption period decreased the proportion of rebate buyers that applied for the rebate ( 2 = 4.84, p < .05), and increasing redemption effort generated a marginally significant increase in the proportion of repeat buyers that applied for the rebate ( 2 = 2.59, p < .10). Increasing the reward did not have a significant effect on the proportion of rebate buyers that logged on to the website ( 2 = 0.01, p > .90), applied for the rebate ( 2 = 0.37, p > .50), or were accepted ( 2 = 0.48, p > .40). The estimated odds of applying for the rebate are 5.1 times higher when the redemption period is one day compared to 21 days, 2.9 times higher when the redemption period is seven days compared to 21 days, and 6.3 times higher when consumers experience the high-effort redemption process compared to the low-effort redemption process. Procrastination Procrastination analyses were conducted on the 46 repeat buyers that initiated the redemption process by logging on to the website and the 43 repeat buyers that completed the redemption process by mailing a rebate application. The results for login delay and mail delay replicated the effects observed in the previous experiments and are shown in figures 20 and 21 respectively. Increasing the length of the redemption period significantly increased the extent to which rebate buyers procrastinated before logging on to the website (F (2, 34) = 18.79, p < .001). Planned contrasts revealed a significant

PAGE 62

50 difference between the seven-day and 21-day redemption periods (p < .001), but the difference between the one-day and seven-day redemption periods was not significant (p > .10). Interestingly, redemption effort did not have a significant effect on login delay despite the vast majority of repeat buyers being aware of the redemption requirements (F (1, 34) = 1.29, p > .20). Aversiveness to effort would predict repeat buyers in the high-effort condition to procrastinate longer than those in the low-effort condition, yet this was not the case. The effect of reward on login delay was not significant (F (1, 34) = 0.22, p > .60). Analysis of mail delay revealed a significant redemption period by redemption effort interaction (F (2, 31) = 7.04, p < .01). Consistent with the results of experiments 1 and 2, increasing the redemption period reduced procrastination in the high-effort condition relative to the low-effort condition. As before, size of reward did not have a significant effect on mail delay (F (1, 31) = 0.06, p > .80). Repeat Breakage The opportunity to observe repeat instances of breakage was limited by the small proportion of non-redeemers that made a repeat purchase (only six of the 75 non-redeemers from experiment 1 made a repeat purchase). However, repeat breakage was observed for three of the six non-redeemers that made a repeat purchase. Two of the three repeat non-redeemers failed to log on to the website on both occasions, while the third logged on to the website but did not mail an application on both occasions. All three repeat non-redeemers were in the low-effort, 21-day redemption period condition, and two had purchased the $9 rebate offer. Given the small sample of repeat non-redeemers, valid inferences cannot be made regarding the conditions that increase the likelihood of repeat breakage. However, self-reports obtained from the post-test reveal that two repeat non-redeemers procrastinated and eventually forgot to log on to the website on both

PAGE 63

51 occasions. The third non-redeemer did not realize the redemption form had to be submitted by mail on the first occasion, and lost the purchase receipt required for redemption on the second occasion. Thus, although it is hard to estimate the scope of repeat breakage, it certainly is a problem for some consumers. Discussion The results of experiment 3 suggest that rebate buyers avoid offers that are known to require a high degree of effort and offers that they failed to redeem in the past, and that moderate rates of breakage can still occur among successful redeemers that make repeat rebate-dependent purchases. The breakage rate among repeat buyers that redeemed in experiment 1 was 12%, and increasing the redemption period significantly increased procrastination and breakage despite these buyers having full knowledge of the redemption process. The low rate of repeat purchase among non-redeemers is understandable considering that experiment 3 was conducted within five weeks of experiment 1 and involved the same offer. The rate of repeat purchase and breakage among non-redeemers may have been greater if more time had elapsed between the two experiments or if the second offer had been more attractive than the first offer. Nonetheless, experiment 3 represents a conservative test of repeat breakage and suggests that repeat breakage is limited by the low probability of repeat rebate-dependent purchase among non-redeemers.

PAGE 64

52 25%20%20%13%25%44%29%28%50%38%31%18%0%10%20%30%40%50%60%1-Day7-Day21-DayRedemption PeriodProportion that Purchased $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 18. Proportion of rebate buyers making a repeat purchase of the rebate offer in experiment 3 by condition 80%80%71%80%67%60%100%100%100%50%60%70%80%90%100%1-Day7-Day21-DayRedemption PeriodProportion that Applied $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 19. Proportion of experiment 3 rebate buyers that applied for the rebate by condition

PAGE 65

53 2.04.00.03.05.41.75.70.30.05.30.40.402468101-Day7-Day21-DayRedemption PeriodDays $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 20. Experiment 3 login delay by condition 3.01.52.02.07.37.06.59.00.51.00.60.502468101-Day7-Day21-DayRedemption PeriodDays $9 High effort $6 High effort $9 Low effort $6 Low effort Figure 21. Experiment 3 mail delay by condition

PAGE 66

54 Table 1. Self-reported reasons for not making a repeat rebate purchase in experiment 3 Reason Redeemers n = 64 Non-Redeemers n = 69 Total Responses n = 133 I still havent used the tickets I bought last time 54.7% 13.0% 33.1% I didnt redeem last time 0.0% 47.8% 24.8% Its too much work/effort to redeem the rebate 21.9% 23.2% 22.6% I dont have the money / cant afford it 15.6% 10.2% 12.8% I dont go to the movies that often 4.7% 2.9% 3.7% Other 3.1% 2.9% 3.0% Total 100% 100% 100%

PAGE 67

CHAPTER 8 POST TEST SURVEY Experiments 1, 2 and 3 were not able to separate the effects of procrastination and prospective forgetting on breakage, nor were they able to identify other potential factors that may have interfered with redeeming. The post-test survey was designed to address these issues. The survey was administered to the 184 rebate buyers that participated in experiments 1 and 3 approximately five weeks after the study 3 purchase opportunity. Rebate buyers responded to items that solicited retrospective descriptions of the purchase decision and efforts to redeem the rebate, as well as items that measured redemption confidence and base rate estimates, perceived fairness of the rebate offer, actual redemption effort relative to expectations of redemption effort at the time of purchase, attitudes toward rebate offers, and the likelihood of purchasing a rebate offer in the future. In addition to these items, non-redeemers (i.e., rebate buyers that failed to redeem the rebate in experiment 1 and/or experiment 3) explained why they did not redeem the rebate offer. The post-test survey items are listed in the appendix. Results All 184 rebate buyers (112 redeemers and 72 non-redeemers) completed the post-test. Ninety-eight percent of participants believed that the series of studies was about measuring movie preferences, indicating that the cover story was successful. One-hundred percent of participants indicated that they had purchased the rebate offer to obtain the rebate, confirming that the procedure was successful in isolating rebate-dependent purchases. 55

PAGE 68

56 Reasons for Breakage Non-redeemers self-reported reasons for breakage are summarized in table 2. Procrastination and forgetting accounted for a greater proportion of breakage (59.7%) than redemption effort (20.8%) and procedural constraints (19.5%) combined. As expected, redemption effort was a factor only among those who initiated redemption, while procrastination interfered with initiating and completing redemption. These results are consistent with the procrastination and redemption patterns observed in the experiments and confirm that procrastination can have a greater influence on breakage than procedural constraints and redemption effort. One explanation for the strong influence of procrastination on breakage is that procrastination invites a multitude of other factors to interfere with redemption. Responses to open-ended questions about why participants failed to redeem revealed that those who procrastinated maintained the intention to redeem the rebate, but either engaged in diversionary activities or were distracted by other daily tasks and commitments. This behavior is consistent with Knaus (2000) self-perpetuating explanation of procrastination. It is also worth noting that although procrastination leads to forgetting in some cases, this accounted for only 5.6% of total breakage. With respect to the relative influences of procrastination and prospective forgetting, the self reports suggest that procrastination contributes more to breakage than prospective forgetting. Prospective forgetting accounted for 22.5% of the breakage among rebate buyers that failed to initiate the redemption process, but did not contribute to breakage among those who initiated the redemption process. Apparently, those who forgot to redeem did so early in the process and consequently did nothing after purchase. Further analysis revealed that the proportion of respondents who reported forgetting to redeem as

PAGE 69

57 the reason for breakage increased with the length of the redemption period, although this difference was not significant. Thus, there is not sufficient evidence to support H6. One explanation for the relatively weak influence of prospective forgetting is that the purchase receipt may have served as an effective memory cue which reminded rebate buyers of their intention to redeem the rebate. Redeemers vs. Non-redeemers Response frequencies to the items administered to redeemers and non-redeemers are summarized in table 3. Consistent with the findings of experiment 2, retrospective confidence estimates confirmed that rebate buyers were highly confident of redeeming at the time of purchase (M = 93.3%) and that redeemers (M = 93.5%) and non-redeemers (M = 92.9%) were equally confident of redeeming (F (1, 170) = 0.06, p > .80). Consistent with other studies on overconfidence in personal forecasts (e.g., Hoch 1985), the majority (67%) of rebate buyers did not consider factors that could have potentially interfered with redeeming. The proportion of rebate buyers that considered such factors was similar for redeemers (P = 32.8%) and non-redeemers (P = 32.8%), ( 2 = 0.01, p > .90), which is understandable given the similarity in their confidence estimates. As argued in the overconfidence literature (e.g., Griffin and Varey 1996; Hoch 1985; Pulford and Coleman 1996; Vallone et al.1990), failing to consider factors that could interfere with achieving a desired outcome is likely to contribute to overconfidence. Perceived fairness of the rebate offer was measured by asking rebate buyers to indicate whether they perceived the offer as fair or deceptive. The proportion of rebate buyers that perceived the rebate offer as fair was significantly higher for redeemers (P = 90.4%) than for non-redeemers (P = 76.1%), ( 2 = 6.43, p < .05). Further analysis revealed that the difference between redeemers and non-redeemers can be

PAGE 70

58 attributed to rebate buyers whose rebate applications were rejected for not complying with the terms and conditions of the offer (i.e., late or incomplete submissions). Changes in attitudes toward rebate offers were measured by asking rebate buyers to indicate their general attitude toward rebates before and after participating in the study. Rebate buyers were also asked to estimate their probability of purchasing a rebate offer in the future. Changes in attitudes and the probability of future purchase were analyzed using MANOVA. The analysis tested the effects of reward, redemption period, redemption effort, redemption behavior (redeeming/not redeeming the offer), and relative redemption effort (a measure of the actual redemption effort relative to rebate buyers expectations at the time of purchase). As shown in table 3, changes in attitudes (F (1, 170) = 14.33, p < .001) and the probability of future purchase (F (1, 170) = 4.89, p < .05) were significantly different for redeemers and non-redeemers. Relative redemption effort also had a significant negative effect on changes in attitudes (F (1, 170) = 34.06, p < .001) and the probability of future purchase (F (1, 170) = 21.15, p < .001). In summary, failing to redeem the rebate and redemption procedures that are more effortful than expected can have negative effects on consumers attitudes and purchase intentions. These findings are consistent with the repeat purchase effects observed in experiment 3, and provide support for H8. Non-Redeemers Response frequencies to the items administered exclusively to non-redeemers are summarized in table 4. Non-redeemers were asked whether they considered themselves better off, no better or worse off, or worse off for not redeeming the rebate. Only 35.9% of non-redeemers considered themselves worse off. Multinomial logistic regression revealed a significant effect of redemption effort on responses ( 2 = 6.39, p <

PAGE 71

59 .05), and the estimated odds of being worse off were 3.9 times higher for those in the high-effort condition compared to those in the low-effort condition. In explaining their responses, many non-redeemers mentioned that they still benefited from their purchase (i.e., seeing a movie) despite paying a higher price. This suggests that some non-redeemers may use product utility or product satisfaction as a post-hoc justification for failing to redeem a rebate offer. Non-redeemers were also asked which behavior, purchasing the rebate offer or not redeeming the rebate offer, was their bigger mistake. Interestingly, the vast majority (68.8%) regarded not redeeming as their bigger mistake. This implies that the majority of rebate buyers believe that their purchases were justified and that the main culprit in breakage is faulty post-purchase behavior. In contrast, the results of experiment 2 suggest that the main culprit in breakage is redemption overconfidence that leads consumers to purchase rebate offers that they are unlikely to redeem. Thus, most non-redeemers appear to be unaware of their redemption overconfidence or that redemption overconfidence contributes to breakage. Finally, non-redeemers were asked whether failing to redeem the rebate was their fault or the fault of the rebate offer. Seventy percent of non-redeemers attributed their redemption failure to themselves. Binary logistic regression revealed a significant effect of relative redemption effort on responses ( 2 = 9.02, p < .01), indicating that non-redeemers are more likely to attribute their redemption failure to the rebate offer when redeeming the rebate requires more effort than is expected at the time of purchase.

PAGE 72

60 Table 2. Self-reported reasons for breakage by non-redeemers Reason Buyers who did not initiate redemption n = 40 Buyers who did initiate redemption n = 32 Total Responses n = 72 Procrastination and forgetting: I procrastinated and the deadline passed before I applied 50.0% 43.8% 47.2% I procrastinated and eventually forgot to apply 10.0% -5.6% I completely forgot to apply 12.5% -6.9% Redemption effort & deciding to not redeem: I tried to redeem but it was too much work so I gave up -28.1% 12.5 I decided the money saved was not worth the effort required -18.8% 8.3 The money saved didnt seem as valuable when it came time to redeem ---Procedural constraints: I lost my receipt / tickets and could not apply 10.0% -5.6 I used the tickets before I could photocopy them and could not apply 17.5% 9.4% 13.9 Total 100% 100% 100% Table 3. Post-test responses by redeemers and non-redeemers Item Redeemers n = 112 Non-Redeemers n = 72 Test Result Mean subjective probability of redemption (redemption confidence): 93.5% 92.9% F (1, 170) = 0.06, p > .80 Mean base rate estimate for redemption: 63.5% 62.7% F (1, 170) = 0.01, p > .90 Proportion that foresaw problems that could interfere with redeeming: 32.8% 33.0% 2 = 0.01, p>.90 Proportion indicating the rebate offer was a fair offer: 90.4% 76.1% 2 = 6.43, p<.05 Change in attitude toward rebates (attitude before attitude after): -.08 -26.0 F (1, 170) = 14.33, p < .001 Probability of purchasing a rebate offer in the future: 81.0% 70.6% F (1, 170) = 4.89, p < .05

PAGE 73

61 Table 4. Post-test responses by non-redeemers only Item Non-Redeemers n = 72 How would you categorize yourself as a result of buying the rebate offer? Better off: 21.9% No better or worse off: 42.2% Worse off: 35.9% Which behavior do you regard as the bigger mistake? Buying the rebate 31.3% Not redeeming: 68.8% All things considered, not redeeming the rebate was: The offers fault: 29.7% My fault: 70.3%

PAGE 74

CHAPTER 9 EXPERIMENT 4 Experiments 1 and 2 demonstrated that increasing reward has a stronger influence on purchase behavior than redemption behavior. However, since participating in the redemption phase of the experiments was conditional on purchasing the rebate offer, it is possible that the weak (i.e., null) effect of reward on redemption behavior is an artifact of self-selection bias (i.e., those who did not purchase the rebate offer may have responded differently than rebate buyers had they participated in the redemption phase of the experiments). It is important to note that any self-selection that may have occurred in the previous experiments also exists in the marketplace since the opportunity to redeem is always conditional on purchase. Thus, any threat posed by self-selection cannot affect the external validity of the results observed in the previous experiments. However, to make the assertion that manipulating reward exerts a relatively weak influence directly on redemption behavior, an alternative procedure was developed which excluded the purchase phase used in the in the previous studies and randomly assigned participants directly to the two reward conditions of the redemption phase of the procedure. Design and Procedure Experiment 4 used a simple two-cell between-subjects design (reward: $6 or $9). Redemption period was fixed at 21 days and redemption effort was fixed using the low-effort redemption procedure used in the previous experiments. One hundred and thirty-four students from an undergraduate subject pool participated in the experiment for course credit and were randomly assigned to either the $6 or $9 reward condition, 62

PAGE 75

63 resulting in 67 participants per cell. Upon entering the experiment room, participants were informed of the market research survey on movie preferences and told that they would inherit a rebate offer coupon from a previous marketing campaign after completing the survey. The following cover story was used to introduce the market research survey and explain why the rebate coupons were being distributed: Earlier this summer, to kick-off the start of the summer blockbuster season, a well-known cinema chain launched a rebate promotion that was later discontinued after learning that the promotion was unprofitable due to the unexpectedly high number of rebate redemptions. However, due to a recent FTC ruling outlawing the premature cancellation of rebate promotions once they are launched, the cinema chain is obligated to distribute the remaining rebate coupons to the general public and mail rebate checks to those who redeem the rebate. To kill two birds with one stone, the cinema has decided to use this opportunity to learn more about the types of movies college audiences want to see, and to distribute the remaining rebate coupons. After completing the market research survey on movie preferences, you will receive a rebate coupon that you are free to redeem for the cash reward prior to the redemption deadline stated on the coupon. Following the completion of the bogus market research survey, rebate coupons were randomly distributed to participants and the participants exited the room. The redemption procedure that followed was identical to that used in experiments 1, 2, and 3, with the exception that the UPC code was printed on the rebate coupon since movie tickets were not used in the study. Dependent Measures and Analysis With the exception of the purchase dependent measure which was omitted from the study, experiment 4 used the same dependent measures as the previous experiments. Redemption was analyzed using binary logistic regression and procrastination was analyzed using ANOVA.

PAGE 76

64 Results Redemption Redemption analyses were conducted on the 134 participants that received rebate coupons. Approximately 7.5% of coupon holders applied for the rebate, resulting in a breakage rate of 92.5%. Thirty percent of the applications (2.3% of coupon holders) had their rebate applications rejected, resulting in a redemption rate of 5.2%. The proportion of coupon holders in each condition that applied for the rebate is shown in figure 22. Consistent with the results of the previous experiments, logit analyses revealed that increasing the reward did not have a significant effect on the proportion of coupon holders that logged on to the website ( 2 = 0.06, p > .80), applied for the rebate ( 2 = 0.00, p = 1.00), or were accepted ( 2 = 0.15, p > .70). Procrastination Procrastination analyses were conducted on the 21 coupon holders that logged on to the website and the 10 coupon holders that mailed a rebate application. The results for login delay and mail delay are shown in figures 23 and 24 respectively. Consistent with the results of the previous experiments, increasing the reward did not have a significant effect on the extent to which coupon holders procrastinated before logging on to the website (F (1, 19) = 1.05, p >.30) or mailing the rebate application (F (1, 8) < 1, p >.80). Discussion Although the null effect of reward on redemption behavior observed in experiments 1, 2 and 3 was replicated in experiment 4, the prudent interpretation is that the results were inconclusive regarding the direct effect of reward on redemption behavior due to the extremely low response rate (i.e., only 21 of the 134 coupon holders initiated the redemption process and only 10 of the coupon holders applied for the rebate). One

PAGE 77

65 explanation for the low response rate is that excluding the purchase phase of the procedure effectively eliminated the up-front financial investment that usually accompanies the purchase of a rebate offer, thereby diminishing participants motivation to redeem the rebate coupon. Alternatively, participants may not have believed that they were being given an opportunity to receive a cash reward with no strings attached. The failure of experiment 4 to produce conclusive results dictates that claims cannot be made regarding the direct influence of reward on redemption behavior. Nonetheless, the assertion that increasing reward has a stronger influence on purchase behavior than redemption behavior when the opportunity to redeem is contingent on purchase is maintained.

PAGE 78

66 7.5%7.5%0%2%4%6%8%10%1Redemption PeriodProportion that Applied $6 $9 Figure 22. Proportion of experiment 4 coupon holders that applied for the rebate by condition 1.20.500.40.81.21.6Redemption PeriodDays $6 $9 Figure 23. Experiment 4 login delay by condition

PAGE 79

67 2.01.80.00.51.01.52.02.5Redemption PeriodDays $6 $9 Figure 24. Experiment 4 mail delay by condition

PAGE 80

CHAPTER 10 GENERAL DISCUSSION The psychological mechanisms proposed in this article are applicable not only to real-world mail-in rebates, but also to other paradigms where the opportunity for breakage looms (i.e., when the attainment of a delayed incentive is conditional on the performance of some future effort). This research offers four contributions to marketings understanding of breakage. First, by isolating rebate-dependent purchases, the experiments establish that breakage is indeed a common post-purchase outcome. Second, by tracking post-purchase behavior, the experiments go beyond redemption rates to examine how and when breakage occurs. Although exact levels of breakage are dependent on the characteristics of the offer and the redemption process, results suggest that much of breakage can be attributed to psychological mechanisms that exert influence before consumers initiate the redemption process (i.e., redemption overconfidence, procrastination, and to a lesser extent, forgetting). This finding is important because it demonstrates that breakage is largely determined before consumers become aware of the effort required to redeem a rebate. Third, the results provide direct evidence of the relationship between consumers redemption confidence and their purchase and redemption behavior. Redemption confidence can be manipulated to encourage purchase by increasing the size of a reward and the length of a redemption period, but high confidence in redeeming does not necessarily translate into actual redemption behavior. Fourth, by examining repeat purchase behavior, the experiments reveal that rebate buyers making rebate-dependent purchases avoid offers that are known to require a high degree 68

PAGE 81

69 of effort and offers that they failed to redeem in the past, and that moderate rates of breakage can still occur among successful redeemers that make repeat rebate-dependent purchases. Although breakage can be a recurring problem for some consumers, the likelihood of repeat breakage is limited by the low probability of repeat purchase among failed redeemers. Alternative Explanations for Breakage It is evident from the results that there are many factors which influence purchase and redemption behavior to create breakage. In addition to the factors identified by this research, there are two additional explanations for breakage that may have exerted effects. One of these explanations is reward discounting. Research on value discounting has established that people often apply discount rates that diminish the perceived value of future payments (see Frederick, Loewenstein, and ODonoghue 2002 for a review). If reward discounting were to occur after purchase, the reward could appear less valuable at the time of redemption than it does at the time of purchase, which would encourage breakage. Although evidence supporting reward discounting was not observed, it may still have exerted an effect. The absence of a reward by redemption period interaction on redemption in experiments 1, 2, 3 and 4 suggests that the influence of reward discounting is relatively minor. Furthermore, none of the respondents reported that reward appeared less valuable after making their purchases. Nevertheless, reward discounting cannot be ruled out as a contributor to breakage. A second alternative explanation is that consumers frame the value of the reward relative to the list price at the time of purchase and later reframe the value of the reward relative to redemption effort after purchase. Consequently, the cost of the future effort is underweighted at the time of purchase and more accurately weighted after purchased.

PAGE 82

70 The post test results provide some support for this explanation. The low proportion of rebate buyers that foresaw problems with redeeming at the time of purchase (P = 33%) suggests that redemption effort is not prominent in the minds of most buyers at the time of purchase. This portion of the reframing argument is essentially the same as the argument for redemption overconfidence (i.e., people are overconfident because they fail to consider factors that could potentially interfere with redeeming). With regard to the reframing of reward after purchase, one would expect a) the majority of breakage to occur once buyers become aware of the redemption effort, and b) for rebate buyers to report redemption effort and/or insufficient reward as the reason for breakage. Although this occurred, it was not the case for the majority of cases. Experiments 1, 2, and 3 showed that the majority of breakage occurred before buyers became aware of redemption effort, and the self-reported reasons for breakage indicate that redemption effort and/or insufficient reward accounted for only 20.8% of breakage. In summary, although reframing may certainly occur, it does not appear to account for the majority of breakage. Implications for Regulators Curbing Breakage From a regulatory perspective, the results indicate that a substantial proportion of rebate buyers can fall victim to breakage, and that breakage is caused, in part, by consumers being highly confident of redeeming a rebate that they are objectively unlikely to redeem. For those interested in curbing breakage, two approaches may be effective. One approach is to focus on curbing consumers redemption confidence at the time of purchase. This approach is based on the notion that a consumers objective probability of redeeming is largely predetermined, and that curbing breakage is a function of

PAGE 83

71 discouraging rebate-dependent purchases among consumers who are unlikely to redeem the rebate. As noted earlier, a common behavioral phenomenon is that people tend to overlook situational factors that could potentially interfere with achieving a desired outcome. Thus, regulators may wish to conduct research that examines how to overcome this phenomenon in the context of rebates and other forms of delayed incentives. For example, regulators may wish to explore how regulating the presentation of rebate offers (e.g., explicitly listing redemption requirements on the price tag) influences breakage. A second approach to curbing breakage is to focus on post-purchase behavior. This assumes that breakage can be reduced by simplifying the redemption process. Research could examine whether regulatory efforts to simplify redemption are effective in mitigating procrastination and procedural constraints that contribute to breakage. This approach is currently reflected in pending legislation in several regions (e.g., California Bill 1154). Implications for Managers Trend toward Shorter Redemption Periods From a managerial perspective, four implications are noteworthy. The first implication relates to the trend toward shorter redemption periods. Twenty years ago, consumers were typically given 60 to 90 days to redeem a rebate offer. Today, 30-day, 14-day and 7-day redemption periods are the norm in many product categories. 1 Marketers report that this trend has been guided by two forces: the desire to complete and assess promotions in a timely manner, and the conventional wisdom that giving consumers less time to redeem will decrease redemption rates. Although this intuition seems reasonable, the results have major implications for marketers in demonstrating that 1 Source: DDS Right Choice, GeigerDonnelly Marketing, Promotion Marketing Association.

PAGE 84

72 shorter redemption periods decrease sales and increase redemption applications. As discussed earlier, there are several factors that have hindered research on breakage. Consequently, many managers may be unaware of these effects. The Role of Redemption Effort in Breakage The majority of breakage appears to occur before consumers become aware of the effort required to redeem. Even if this result is attributed to consumers expectations of redemption effort (which self-reports suggest is not the case), manipulating redemption effort would not have affected this result. Furthermore, increasing redemption effort decreased procrastination and marginally increased redemption applications. Although redemption effort did create breakage among those who initiated redemption, procrastination interfered with initiating and completing redemption and accounted for a greater proportion of breakage than redemption effort. These results have three implications. First, excessively effortful redemption processes should not be viewed as a minimum requirement for breakage. Second, increasing redemption effort to increase breakage can have an upper limit and can increase redemption rates. Third, although exact levels of breakage are dependent on the characteristics of a firms redemption process, redemption processes that provide multiple opportunities to procrastinate can create more breakage than moderate levels of redemption effort. Given the conventional wisdom that breakage is caused primarily by redemption effort, firms may be over-attributing the role that effort plays in creating breakage. Repeat Purchase Behavior Results reveal that rebate buyers avoid high-effort offers and offers that they failed to redeem in the past, and that moderate rates of breakage can still occur among successful redeemers that make repeat rebate-dependent purchases. This implies that

PAGE 85

73 high-effort redemption requirements can steer consumers away from rebates and generate less breakage in the long run. Consequently, moderate-effort offers may be more optimal than high-effort offers since moderate-effort offers can take advantage of breakage without risking higher redemption rates or decreases in future sales. Future Research Effects of Effort on Redemption Increasing redemption effort was shown to reduce procrastination and increase (marginally) the proportion of buyers that applied for the rebate. This result is opposite to the aversiveness of effort effect observed in earlier experiments on delayed incentives (e.g., Soman 1998). The apparent inconsistency between these effects may be explained by the timing of the effort manipulations. The delayed incentives study manipulated effort at the time of choice by showing subjects the length of the survey that was to be completed to receive the reward. In contrast, the current study did not manipulate effort until after subjects had initiated the redemption process. Post-test responses revealed that the majority of consumers in the high-effort condition felt that the redemption procedures were more effortful than they had expected at the time of purchase. Thus, the quicker response and higher redemption rate among these participants could be interpreted as a backlash effect. However, the reasons for this effect are not clear. In particular, three explanations are plausible. First, the result could be a sunk cost effect whereby effort invested in stage one of a task increases ones motivation to complete subsequent stages of the task. Second, the result could be a retribution effect whereby ones motivation to redeem increases when redeeming is framed as an opportunity to seek retribution for completing an unnecessarily effortful task. Third, the result could be a salience effect whereby increasing the effort increases the salience of the task and the likelihood of

PAGE 86

74 completing it before a deadline. Future research could examine these explanations and their implications. Differential Effects of Reward on Purchase and Redemption Increasing the reward was shown to exert a stronger effect on purchase than redemption. Although the null effect of reward on redemption is unlikely to hold at larger reward values (i.e., at some point, a reward will be sufficiently large to generate a significant effect on redemption), it is uncertain whether the differential strength of the effect will hold at larger reward values (i.e. will offering a $100 reward still have a larger influence on purchase than redemption?). Reports of large sales increases and moderate redemption rates for large reward values (i.e., $400) suggest that the differential effect may hold (e.g., Spencer 2002). Given the difficulties of studying large-ticket purchases in a laboratory environment, this issue would be best addressed by future field studies that manipulate offer characteristics and control for market factors.

PAGE 87

APPENDIX POST TEST SURVEY ITEMS Instructions The series of studies that you completed for us were not about movies, but were actually about rebate offers. This rebate study was conducted for the Federal Trade Commission to find out if rebates constitute a risk to consumer welfare. We will ask you some questions about how you decided to purchase the rebated movie tickets, and about your attempt(s) to redeem the rebate offer. It is very important that you answer all questions honestly so that we can accurately assess how consumers respond to rebate offers. Questions *Denotes questions that were administered to non-redeemers only. (1) Did you know at the time of doing the movie rating studies that this was an experiment on rebate offers? (Please be honest) Yes No (2) If you did know, please tell us how you knew. If you did not know, click continue. (3) When you purchased the tickets, how confident were you that you would redeem the rebate? Your confidence estimate should correspond to the probability that you would redeem the rebate. For example, % confident means that you were completely certain that you would redeem the rebate. % confident means that you were completely un certain about whether you would redeem (e.g., a 50/50 chance you would redeem). Any number below 50% means that you believed it was more likely that you would not redeem. For example, % confident means that you were completely certain you would not redeem the rebate. Please indicate your confidence estimate using the scale below. Remember, your answer should be based on what you thought at the time of purchase 1 1 Probability scale instructions adapted from Vallone et al. 1990. 75

PAGE 88

76 0% confidence 100% confidence [sliding scale] (4) Please estimate the percentage of rebate buyers that you thought would redeem the rebate when you purchased the tickets. Your estimate should correspond to the percentage of rebate buyers you thought would redeem the rebate. For example, an estimate of 100% means that you thought all of the rebate buyers would redeem the rebate. An estimate of 50% means that you thought half of the rebate buyers would redeem the rebate. An estimate of 0% means that you thought none of the rebate buyers would redeem the rebate. Please indicate your estimate below. Remember, your answer should be based on what you thought at the time of purchase 0% 100% [sliding scale] (5) At the time of purchasing the tickets, did you foresee any problems that could interfere with redeeming the rebate? If yes, please type yes and describe what you thought about when you purchased the tickets, and why you still chose to buy the tickets. If no, please type no and describe what you thought about when you purchased the tickets. (6) How was the experience of redeeming the rebate compared to what you expected it would be when you purchased the tickets? -100 (more effort than I expected) to +100 (less effort than I expected) [sliding scale] (7) Please describe the actions you took (if any) to redeem the rebate, and any obstacles that you encountered. For example, did you logon to the rebate website? Did you complete the redemption form? What obstacles did you encounter? (e.g., procrastination, no postage stamps, etc.) (8) *In your own words, please describe why you didnt redeem the rebate. Please be as specific as possible. (9) *Below are some common reasons why people dont redeem rebates. Please select the reason that best describes your situation: I mailed my rebate application but I was rejected. I lost my receipt / tickets and could not apply. I used the tickets before I could photocopy them and could not apply. I procrastinated and the deadline passed before I applied. I procrastinated and eventually forgot to apply. I completely forgot to apply.

PAGE 89

77 I tried to redeem but it was too much work so I gave up. I decided the money saved was not worth the effort required. The money saved didnt seem as valuable when it came time to redeem. (10) *If we gave you another chance to redeem the rebate, would you take it? Yes No (11) *Please explain the reason for your answer in the box below. (12) *All things considered, how would you categorize yourself as a result of buying the rebate offer? I am better off for buying the rebate offer. I am no better or worse off for buying the rebate offer. I am worse off for buying the rebate offer. (13) *Please explain the reason for your answer in the box below. (14) *Which behavior do you regard as the bigger mistake? Buying the rebate offer was the bigger mistake. Not redeeming the rebate offer was the bigger mistake. (15) *Please explain the reason for your answer in the box below. (16) *All things considered, not redeeming the rebate was: My fault The fault of the offer (17) *Please explain the reason for your answer in the box below. (18) All things considered, the rebate offer was: A fair offer Not a fair offer (deceptive) (19) Please explain the reason for your answer in the box below. (20) Before participating in this study, how many times in the past have you purchased a rebate offer with the intention of redeeming the rebate? Never Once Twice 3 times 4 times

PAGE 90

78 5 10 times More than 10 times (21) Of the purchases you indicated above, approximately what proportion of them did you successfully redeem? 0% (none of them) 100% (all of them) [sliding scale] (22) Before this study, what was your general attitude toward rebate offers? Use the scale below to record your response. A rating above represents a positive attitude toward rebates. A rating of represents a neutral attitude. A rating below represents a negative attitude toward rebates. -100 (generally negative) to +100 (generally positive) [sliding scale] (23) Since participating in this study, what is your general attitude toward rebate offers? Use the scale below to record your response. A rating above represents a positive attitude toward rebates. A rating of represents a neutral attitude. A rating below represents a negative attitude toward rebates. -100 (generally negative) to +100 (generally positive) [sliding scale] (24) What is the likelihood that you will purchase another rebate offer (for some other product) in the future? Your estimate should correspond to the probability that you will buy a rebate offer in the future. For example, an estimate of 100% means that you are completely certain that you will buy another rebate offer in the future. An estimate of 50% means that you are completely uncertain about whether you will buy another rebate offer in the future. Any number below 50% means that you believed it is more likely that you will not buy another rebate offer in the future. For example, 0% means that you are completely certain that you will not buy another rebate offer in the future. 0% (not at all likely) 100% (very likely) [sliding scale] (25) [Those who did not successfully redeem were given the opportunity to purchase tickets again using the purchase opportunity choice set.]

PAGE 91

79 (26) *How confident are you that you will redeem the rebate? Your confidence estimate should correspond to the probability that you would redeem the rebate. For example, % confident means that you are completely certain that you will redeem the rebate. % confident means that you are completely un certain about whether you will or will not redeem (e.g., a 50/50 chance you will redeem). Any number below 50% means that you believe it is more likely that you will not redeem. For example, % confident means that you are completely certain you will not redeem the rebate. Please indicate your confidence estimate using the scale below. 0% confidence 100% confidence [sliding scale] (27) *Please estimate the percentage of rebate buyers that you think will redeem the rebate. Your estimate should correspond to the percentage of buyers that will redeem the rebate. For example, an estimate of 100% means that you think all of the rebate buyers will redeem the rebate. An estimate of 50% means that you think half of the rebate buyers will redeem the rebate. An estimate of 0% means that you think none of the rebate buyers will redeem the rebate. Please indicate your estimate below. 0% 100% [sliding scale]

PAGE 92

LIST OF REFERENCES Akerlof, George A. (1991), Procrastination and Obedience, American Economic Review, 81 (May) 1-19. Alba, Joseph W. and J. Wesley Hutchinson (2000), Knowledge Calibration: What Consumers Know and What They Think They Know, Journal of Consumer Research, 27 (September), 123-156. Ali, Abdul, Marvin A. Jolson, and Rena Y. Darmon (1994), A Model for Optimizing the Refund Value in Rebate Promotions, Journal of Business Research, 29 (March), 239-245. Ariely, Dan and Klaus Wertenbroch (2002), Procrastination, Deadlines, and Performance: Self-Control by Precommitment, Psychological Science, 13 (May), 219-224. Avila, Ramon A., Joseph D. Chapman, and Teresa K. Avila (1989), An Exploratory Study on Consumers Attitudes Toward Rebates, in Developments in Marketing Science, 12, 273-77. Ben-Akiva, Moshe E. and Steven R. Lerman (1985), Discrete Choice Analysis: Theory and Applications to Travel Demand, Cambridge, MA: MIT Press. Blattberg, Robert and Scott Neslin (1990), Sales Promotions: Concepts, Methods and Strategies, Englewood Cliffs, NJ: Prentice-Hall. Buehler, Roger, Dale Griffin, and Michael Ross (1994), Exploring the Planning Fallacy: Why People Underestimate Their Task Completion Times, Journal of Personality and Social Psychology, 67 (September), 366-381. Dhar, Sanjay K. and Stephen J. Hoch (1996), Price Discrimination using In-Store Merchandising, Journal of Marketing, 60 (January), 17-30. Ellis, Albert and William J. Knaus (1977), Overcoming Procrastination, New York: Institute for Rational Living. Ferrari, Joseph R. (1993), Christmas and Procrastination: Explaining Lack of Diligence at a Real-World Task Deadline, Personality and Individual Differences, 14 (January), 25-33. 80

PAGE 93

81 Frederick, Shane, George Loewenstein, and Ted ODonoghue (2002), Time Discounting and Time Preference: A Critical Review, Journal of Economic Literature, 40(June), 351-401. Gilovich, Thomas, Margaret Kerr, and Victoria Husted Medvec (1993), Effect of Temporal Perspective on Subjective Confidence, Journal of Personality and Social Psychology, 64 (April), 552-560. Greenman, Catherine (1999), The Trouble with Rebates, The New York Times, September 16, 1999, G1. Griffin, Dale W., David Dunning, and Lee Ross (1990), The Role of Construal Processes in Overconfident Predictions about the Self and Others, Journal of Personality and Social Psychology, 59 (December), 1128-1139. Griffin, Dale W. and Carol A. Varey (1996), Towards a Consensus on Overconfidence, Organizational Behavior and Human Decision Processes, 65 (March), 227-231. Harriott, Jesse and Joseph R. Ferrari (1996), Prevalence of Chronic Procrastination Among Samples of Adults, Psychological Reports, 78 (April), 611-616. Hoch, Stephen J. (1985), Counterfactual Reasoning and Accuracy in Predicting Personal Events, Journal of Experimental Psychology: Learning, Memory and Cognition, 11 (4), 719-731. Holmes, Richard Alan (2002), The Effect of Task Structure and Task Order on Subjective Distress and Dilatory Behavior in Academic Procrastinators, Doctoral Dissertation, Hofstra University. Hunt, Kenneth A. and Susan M. Keaveney (1992), Sales Promotion Wars: A Strategic Application of Attribution Theory, Journal of Promotion Management, 1 (3), 3-20. Hunt, Kenneth A., Susan M. Keaveney, and Moonkyu Lee (1995), Involvement, Attributions and Consumer Responses to Rebates, Journal of Business and Psychology, 9 (Spring), 273-297. Jolson, Marvin, Joshua Weiner, and Richard Rosecky (1987), Correlates of Rebate Proneness, Journal of Advertising Research, 27 (Feb-March), 33-43. Knaus, William J. (2000), Procrastination, Blame and Change, Journal of Social Behavior and Personality, 15 (Special Issue), 153-166. Krishnan, H. Shanker and Stewart Shapiro (1999), Prospective and Retrospective Memory for Intentions: A Two-Component Approach, Journal of Consumer Psychology, 8 (2), 141-166.

PAGE 94

82 Lanctot, Roger (2002), Rebates Rule in the Retail Computer Software Market, Business Wire, November 12, 2002. Lay, Clarry H. (1988), The Relation of Procrastination and Optimism to Judgments of Time to Complete an Essay and Anticipation of Setbacks, Journal of Social Behavior and Personality, 3 (September), 201-214. Lay, Clarry H. and Peter Burns (1991), Intentions and Behavior in Studying for an Examination: The Role of Trait Procrastination and its interaction with Optimism, Journal of Social Behavior and Personality, 6 (September), 605-617. Liberman, Nira and Yaacov Trope (1998), The Role of Feasibility and Desirability Considerations in Near and Distant Future Decisions: A Test of Temporal Construal Theory, Journal of Personality and Social Psychology, 75 (July), 5-18. Marsh, Richard L., Jason L. Hicks, and Valerie Watson (2002), The Dynamics of Intention Retrieval and Coordination of Action in Event-Based Prospective Memory, Journal of Experimental Psychology: Learning Memory, and Cognition, 28 (July), 652-659. McGinn, Daniel (2003), Lets Make a (Tough) Deal, Newsweek, June 23, 2003, 48-49. McLaughlin, Katy (2002), Cranky Customer: Claiming That Holiday Rebate, The Wall Street Journal, December 3, 2002, D2. Milburn, Michael (1978), Sources of Bias in the Prediction of Future Events, Organizational Behavior and Human Performance, 21 (February), 17-26. Norberg, Bob (2002), Rebate Debate: Manufacturers Betting Consumers Wont Apply for Refunds, The Gainesville Sun, December 16, 2002, WorkLife 6. Norr, Henry (2000), The How and Why of Rebates, San Francisco Chronicle, December 18, 2000, T21. Pulford, Briony D. and Andrew M. Colman (1996), Overconfidence, Base Rates, and Outcome Positivity/Negativity of Predicted Outcomes, British Journal of Psychology, 87 (August), 431-445. Pychyl, Timothy A., Richard W. Morin, and Brian R. Salmon (2000), Procrastination and the Planning Fallacy: An Examination of the Study Habits of University Students, Journal of Social Behavior and Personality, 15 (Special Issue), 135-150. Rothschild, Michael (1987), Marketing Communications: From Fundamentals to Strategies, Lexington, MA: D. C. Heath. Scher, Steven J. and Joseph R. Ferrari (2000), The Recall of Completed and Noncompleted Tasks Through Daily Logs to Measure Procrastination, Journal of Social Behavior and Personality, 15 (Special Issue), 255-265.

PAGE 95

83 Shapiro, Stewart and H. Shanker Krishnan (1999), Consumer Memory for Intentions: A Prospective Memory Perspective, Journal of Experimental Psychology: Applied, 5 (June), 169-189. Shim, Richard (2002), Philips Settles Rebate Complaint, The New York Times.com, August 7, 2002. Sigall, Harold, Arie Kruglanski, and Jack Fyock (2000), Wishful Thinking and Procrastination, Journal of Social Behavior and Personality, 15 (Special Issue), 283-296. Silver, Maury (1974), Procrastination, Centerpoint, 1, 49-54. Soman, Dilip (1998), The Illusion of Delayed Incentives: Evaluating Future Effort-Money Transactions, Journal of Marketing Research, 35 (November), 427-437. Soman, Dilip and John T. Gourville (2002), Mail-in Rebates and Consumer Choice, Working Paper, Harvard Business School, Harvard University. Spencer, Jane (2002), Rejected! Rebates Get Harder to Collect, The Wall Street Journal, June 11, 2002, D1. Tat, Peter (1994), Rebate Redemption: A Motivational Perspective, Psychology & Marketing, 11 (Jan/Feb), 15-26. Tat, Peter, William Cunningham III, and Erin Babakus (1988), Consumer Perceptions of Rebates, Journal of Advertising Research, 28 (Aug-Sep), 45-50. Tat, Peter and Charles Schwepker (1998), An Empirical Investigation of the Relationship Between Rebate Redemption Motives: Understanding How Price Consciousness, Time and Effort, and Satisfaction Affect Consumer Rebate Redemptions, Journal of Marketing Theory and Practice, 6 (Spring), 61. Trope, Yaacov and Nira Liberman (2000), Temporal Construal and Time-Dependent Changes in Preference, Journal of Personality and Social Psychology, 79 (December), 867-889. Tversky, Amos and Eldar Shafir (1992), Choice Under Conflict: The Dynamics of Deferred Decisions, Psychological Science, 3 (November), 358-361. Vallone, Robert. P., Dale W. Griffin, Sabrina Lin, and Lee Ross (1990), The Overconfident Prediction of Future Action Outcomes by Self and Others, Journal of Personality and Social Psychology, 58 (April), 582-592. Weiner, Bernard (2000), "Attributional Thoughts about Consumer Behavior," Journal of Consumer Research, 27 (December), 382-387.

PAGE 96

84 Wright, George and Peter Ayton (1992), Judgmental Probability Forecasting in the Immediate and Medium Term, Organizational Behavior and Human Decision Processes, 51 (April), 344-363.

PAGE 97

BIOGRAPHICAL SKETCH Timothy Guy Silk received his BBA from Wilfrid Laurier University in Waterloo, Ontario, Canada, in 1997. He joined the University of Western Ontario as an instructor, where he taught at the Ivey School of Business and traveled to Russia and Ukraine to teach local business managers as part of the LEADER economic development program. Tim moved to the United States in the fall of 2000 to pursue his Ph.D. in marketing at the University of Florida, and will join the University of South Carolina as an assistant professor of marketing in August 2004. 85