An Investigation of Consumer Product Warranties

as Market Signals of Product Reliability

Craig A. Kelley
California State University

Several explanations of the purpose jor consumer Act, warrantors and consumers still have the tendency
product warranties can be found in the marketing to perceive the purpose of consumer product warranties
literature. However, comparatively little research has been differently. Feldman (1976) speculated that consumers
done to develop and test a theoo' of the consumer product may perceive a warranty as a measure of assurance of
warranty. Recently the Market Signal Theory which posits product quality. Conversely, marketers seem to have a
that warranties serve as signals of product reliability, has myriad of reasons for offering a warranty. Warranties have
emerged in the economic, legal, and marketing literature. been used as an alternative to price competition as a
In this article, a test of the Market Signal Theory is promotional device or as a legal document designed to
conducted using pre- and post-Magnuson-Moss Act give the consumer certain rights extended in an implied
warranties. Marketing and public policy implications of warranty ( Wall Street Journal 1984).
the Market Signal Theory and directions for future Although the Magnuson-Moss Act impacts marketing
research are also discussed. decisions, little has been done to develop and test a theory
underlying the various uses of consumer product
warranties. This situation has changed in recent years as
renewed interest in theory construction has emerged in
economics, law, and marketing (Akerlof 1970; Gerner
INTRODUCTION and Bryant 1981; Kelley 1984; Priest 1981; Spence 1977;
Wiener 1985). The most prominent of the theories that
One of the important pieces of Federal consumer have been advanced is the Market Signal Theory.
protection legislation passed during the 1970s was the According to this theory, consumers make inferences
Magnuson-Moss Warranty Act of I975. The Act grew out concerning a product's reliability from the terms of the
of an increasing number of complaints from consumers product's warranty. Although there have been few tests
that warranties were being used as legal documents that of the Market Signal Theory, Wiener (1985) found some
limited a warrantor's responsibility for product service if evidence that post Magnuson-Moss Act warranties act
the product malfunctioned. The provisions of the Act as signals of product reliability. The purpose of this article
attempted to stimulate competition among marketers to is to extend the previous test of the Market Signal Theory
offer better written warranty terms. This objective was by examining the signaling capabilities of both preand
to be achieved by requiring the terms and conditions of post-Act warranties.
warranties to be disclosed to the consumer in simple and The remainder of the article is organized into four
understandable language before a purchase was made. sections. The first section describes the theoretical
The Act also required warranties to be labeled full or developments made in marketing economics and law
limited depending on what was covered by the warranty which attempt to explain consumer product warranties.
and encouraged warrantors to set up an arbitration system Next, the methodology is presented, followed by the
to settle warranty disputes quickly. research results. Finally, the marketing and public policy
Despite the standards set down in the Magnuson-Moss implications of this research are discussed.

THEORETICAL DEVELOPMENTS
9 1988,Academyof Marketing Science Three theories have been advanced that attempt to
Journalof the Academy of Marketing Science
Summer, 1988,Vol. 16, No. 2, 072-078 explain the Purpose of a consumer product warranty.
0092-0703/88/1602-0072 These are the Exploitation Theory, the Investment Theory,

JAMS 72 SUMMER, 1988

AN INVESTIGATION OF CONSUMER PRODUCT WARRANTIES AS MARKET SIGNALS OF PRODUCT RELIABILITY KELLEY

and the Market Signal Theory (Priest 1981). The provide better protection in terms of what is covered and
Exploitation Theory postulates that the relative market for how long would accompany the most reliable products.
power of various parties involved in an exchange Conversely, less reliable products would have less warranty
determines what warranty terms are offered. According protection. The relationship between warranty terms and
to this theory marketers use warranties as a means of product reUabillty would exist because it is costly for a
limiting their legal obligations by exploiting their market marketer to offer a warranty that provides extensive
power relative to the market power of the individual coverage over a lengthy period of time. Therefore the
consumer. When the balance of market power favors a marketer would have to make a corresponding investment
marketer, warranty terms will be very limited in coverage in product quality to increase reliability so that expensive
and duration. However, if consumers collectively have repairs that would have to be covered under the warranty
more market power relative to the marketer the warranty are minimized.
terms will provide greater coverage and be longer in Early tests of the Market Signal Theory did not include
duration. a measure of reliability for individual brands (Gerner and
The Exploitation Theory has been conceptualized and Bryant 1981; Priest 1981). Instead, an average service life
tested as an inverse relationship between a firm's market of the appliance category was used. Neither Gerner and
power and the amount of coverage offered in the firm's Bryant (1981) or Priest (1981) found support for the theory
warranty. Priest (1981) used major appliance industry using this aggregate measure of reliability. Kelley (1984)
concentration ratios (Herfindal Index)~ as a measure of argued that the theory might be better tested with brand
market power and found that the inverse relationship did specific information. A regression model that included a
not hold. Highly concentrated product groups did not brand's frequency of repair, basic warranty coverage and
differ from the least concentrated product group in the warranty coverage extended on selected parts was tested,
warrarty terms that were offered. In another test of the but the results did not support the Market Signal Theory.
theory, Kelley (1984) used the market shares of specific However, the results may have been influenced by the
brands of dishwashers as a surrogate measure of market small sample of brands included in the study.
power. Each brand's warranty was evaluated on an index Wiener (1985) fund mixed support for the theory when
composed of duration and coverage (called warranty post-Magnuson-Moss Act brands of cars, air conditioners,
quality) and then correlated with the brand's market share. washing machines, and clothes dryers with what
Counter to the Exploitation Theory, it was found that consumers may perceive to be superior warranty terms
brands possessing greater market power offered warranty were found to be more reliable than brands with less
terms of greater coverage and duration than brands with attractive warranty terms. Because of the mixed results,
less market power. additional tests of the Market Signal Theory involving
Priest (1981) put forth a second theory of the consumer additional product classes are needed to provide
product warranty called the Investment Theory. This confirmatory support for the theory. In addition, Wiener
theory explains the role of a warranty as a marketer's (1985) did not test the theory with pre-Act brands, but
investment in terms of an insurance policy and repair in his conclusion called for testing of the Market Signal
contract. According to the theory, a marketer's investment Theory comparing pre- and post-Act warranties. The
in warranty terms will be derived from that demanded contribution of the present article is to provide a
by the consumer. To the extent the consumer can repair comparative test which would yield important empirical
or replace certain parts of the product less expensively information on the relative success of the Magnuson-Moss
than the marketer, the warranty terms will be more limited. Act. If warranties are found to signal product reliability
If repairs can be made less expensively by the marketer, before but not after the Act, then the Act has not achieved
the warranty terms will offer greater coverage. one of its purposes of providing the consumer with useful
Gerner and Bryant (1981) conceptualized the Investment prepurchase product qualilty information. However, if
Theory as the cost of consumer search. They argued that warranties are fund to signal product reliability after but
increasing consumer research (i.e. the investment by the not before the Act, then the Act has had some positive
consumer) would reduce the variation in warranty affect in disclosing prepurchase information.
coverage offered by the marketers in an industry. When Three variables are needed to investigate the question
the investment in consumer search is reduced, an increase of whether pre- and post-Act warranties serve as market
in competition between marketers to offer more signals. According to the Market Signal Theory, brand
comprehensive warranty terms should result. reliability should be a function of warranty terms. In
Although the verbal arguments made in support of this addition, pre-and post-Magnuson-Moss Act measures of
theory seem plausible, the theory has nov been tested. brand reliability and warranty terms are needed to test
The absence of testing is most likely due to the lack of whether warranties signal brand reliability priori to and
necessary cost estimates of consumer search and industry after the Magnuson-Moss Act was passed. The relationship
service and repair data. between brand reliability, warranty and pre- and post-
The most developed and tested of the three theories Magnuson-Moss Act was written in the following form.
is the Market Signal Theory. A signal is a visible product where: R = f(W,T)
feature that consumers use to evaluate invisible product R = Brand reliability.
features. The Market Signal Theory posits that warranty W = Warranty terms.
terms are used by the consumer as a signal of product T = Time period warranty terms and brand reliability
reliability. According to this theory warranty terms that were measured.

JAMS 73 SUMMER, 1988

AN INVESTIGATION OF CONSUMER PRODUCT WARRANTIES AS MARKET SIGNALS OF PRODUCT RELIABILITY KELLEY

Observations prior to 1975 would be pre-Magnuson- on testing the Market Signal Theory. Firms might have
Moss Act. Observations after 1975 would be post- viewed the legislation as a window of opportunity in which
Magnuson-Moss Act. to establish a competitive advantage by improving their
warranty terms. Therefore only using observations from
1975, the year the Act was passed, and 1977 the year
METHODOLOGY the Act finally took effect, may find warranty terms in
a period of transition.
The relationship between brand reliability, warranty The four time periods, two pre- and two post-Act, used
terms, and time was analyzed with a log-linear approach. in the log-linear analysis were labeled first pre-MM, second
Log-linear analysis was selected over chi-square analysis pre-MM, first post-MM, and second post-MM observa-
because the former provides a more powerful test when tions rather than specific years. This allowed the
three or more categorical variables are involved (Kennedy categorization of data since the appliance classes were
I983). In addition log-linear analysis can be used to rated a few months apart but not in the same year. The
examine associations between two or more variables in four categories had the extra benefit of detecting any
an asymmetrical case (i.e: dependent relationship) like the changes in warranty terms that might have taken place
relationship stated above. over time. However omissions in the secondary data did
Observations from five major appliance product classes not allow for a longitudinal study for each brand. The
(refrigerators, clothes washers and dryers, air conditioners, number of brands for each product class across the four
and built-in dishwashers) constituted the sample since time periods are shown in Table 1. The selection of the
secondary data was available on brand reliability and data set resulted in an acceptable number of cells (two)
warranty terms in Consumer Reports magazine. Other with sampling zeros (Kennedy 1983).
appliance categories (e.g.. televisions) also were considered Two expert judges categorized each brand's warranty
but rapid technological changes in these categories did terms as basic to the product class if the terms were
not allow product reliability to be consistently measured. common to all manufacturers. If specific parts (e.g., an
Consumer Reports rates major appliances about every on-off switch or a refrigerator compressor) were covered
two years which required a decision concerning which for an additional period beyond the basic terms then the
time periods to use to select the data. Prior to 1971 and warranty was categorized as better. Interjudge reliability
after 1983, Consumer Reports used a different reliability was 0.98. Ties were decided by a third judge (Bruckes,
scale. This left roughly a decade between 1972 and 1983 Mitchell, and Staelin, 1984).
to select the data based on a rating of much worse than The frequency of repair reported by Consumer Reports
average to much better than average. Therefore the was used as a surrogate measure of product reliability.
decision was made to include two product class Every year the readership of Consumer Reports is used
observations that immediately preceded and two as a sample to gather data used to evaluate products.
observations that immediately proceeded the passage of The data is converted to an ordinal scale where each
the Magnuson-Moss Act. It was believed that selecting brand's repair record is categorized as much worse than
four time periods reduced the confounding effect that the average, worse than average, average, better than average,
turmoil surrounding the passage of the Act might have and much better than average.

TABLE 1
Brand Frequency by Product Class Over Time Periods
Time Period
First Second First Second
Pre-MM Pre-MM Post-MM Post-MM Total

Air Conditioners 17 17 16 14 64

Built-in Dishwashers 16 16 18 17 67

Clothes Dryers 14 12 9 12 47

Clothes Washers 15 13 10 10 48

Refrigerators 10 11 7 12 40
266

Note: MM represents 1975 - - the year that the Magnuson-Moss Warranty Act was passed. First pre-MM would be the observation from the earliest
time period prior to 1975. Second post-MM would be the observation from the latest time period after 1975.

JAMS 74 SUMMER, 1988

AN INVESTIGATIONOF CONSUMER PRODUCTWARRANTIESAS MARKET SIGNALSOF PRODUCTRELIABILITY KELLEY

TABLE 2
Residual and Component Chi-Square Analysis
Residuals
K-Factor* D.F. LR X2 Probability
0-Mean 39 197.71 .000**
1 31 30.95 .468
2 12 6.17 .907
3 0 0.0 1.0

Components
K-Factor* D.F. LR X2 Probability
1 8 166.76 .000"*
2 19 24.78 .167
3 12 6.17 .907

Note: LR X2= LikelihoodRatio, Chi Square

*The K-Factoris used by the BMDP/4F program to denotefull models. For examplethe main-marginalmodel is indicated by K=I which is
the mutuallyindependentmodel of lnFiji= h + hiT+ hjw+ h i p'.
**p < .05

ANALYSIS The next step in model selection was to look for

significant interactions by examining the partial and
The B M D P / 4 F computer program 2 was used to identify marginal association tests between the variables (Brown
what combination of variables, if any, would yield a log- 1983). These tests determine whether an interaction was
linear model that fit the data. Since the log of zero is equal to zero. If both tests were significant, the interaction
negative infinity, 0.5 was added to each cell. Goodman should be included in the model. Conversely, if both tests
(1970) reported that this addition would not unduly were insignificant, the interaction should not be included
influence the fitting of log-linear models. in the model. If one test was significant and the other
Since the Market Signal Theory relationship specified test was insignificant, further analysis would be needed
earlier was asymmetrical, the first log-linear model tested to determine whether or not the interaction should be
for parsimony included controlling for the time by included in the model. Table 3 summarizes the results
warranty variable combination. This model was written of the partial and marginal association analysis.
in the following form. It was apparent from the partial and marginal analysis
lnFijk = h + hiT+ hjw + hk R + hijTw that the warranty x reliability interaction effect should
be included in the final model (partial LR X2 = I1.01,
where: P < 0.026; marginal L R X2 = 11.83; p < 0.018). In addition,
T = Time (first pre-MM, second pre-MM, f'trst Post-MM, the time by warranty interaction must be included since
second Post-MM). this was an asymmetric log-linear analysis. Therefore, the
W = Warranty (basic or better). most parsimonious model that provided a good fit of the
R = Reliability (much worse, worse, average, better, much data was lnFijk = )kiT + hj W + hkR + hijTW'l" hjkWR'(LR
better). X2 = 15.69, df = 24, p < .899). According to this model,
warranty and reliability and warranty and time were
Table 2 summarizes the residual and component chi- conditionally dependent and time was independent of
square analysis used to screen possible higher order reliability. Thus it earl be concluded that consumer product
models. The purpose of the residual chisquare analysis warranties signal brand reliability and that this signaling
was to identify models that may provide an adequate fit occurred before and after the Magnuson-Moss Act.
of the frequency data. The residual chi-squares indicated Since reliability was a polytomous variable, it remained
that a model containing one or more main effects may to be determined which brands had the better warranty
be selected (LR X2 = 30.95 p < .468). Component chi- terms. The z scores presented in Table 4 indicate that
square analysis, which tests for significant differences in brands covered by only a basic warranty were significantly
residuals between groups of models, also indicated that more likely to have a much worse than average repair
a model containing one or more main effects may be record than brands covered by better warranty terms (z
selected (LR X2 = 166.76, p < .000). Although a main = 2.414 p < .05). Conversely, brands covered by better
effect model may provide an adequate fit, the asymmetric warranty terms possessed significantly better reliability
nature of the relationship specified at the outset indicates (z = 2.785, P < .05). These results support the Market
at least one two-variable combination (or interaction) must Signal Theory and clearly indicate that consumers could
be included in the t'mal model. use warranties as a signal of brand reliability.

JAMS 75 SUMMER, 1988

AN INVESTIGATION OF CONSUMER PRODUCT WARRANTIES AS MARKET SIGNALS OF PRODUCT RELIABILITY KELLEY

TABLE 3
Partial and Marginal Association Analysis
Partial Association Marginal Association
Effect D.F. LRx 2 Probability D.F. LR X2 Probability
W 1 29.57 .000
T 3 1.28 .735
R 4 135.91 .000
WT 3 2.62 .454 3 3.44 .328
WR 4 11.01 .026 4 11.83 .018"
TR 12 9.52 .658 12 10.34 .580
WTR 12 6.17 .907

W = Warranty
T = Time Period
R = Reliability

TABLE 4
Summary of Z Tests
Basic Better
Reliability fwR hwR Z fir hwR Z
Much worse 21 .544 2.414" 2 -.544 -2.414"
Worse 25 -.037 -.246 13 .037 .246
Average 95 .053 .495 43 -.053 -.495
Better 28 -.075 -.530 16 .075 -.530
Best 9 -.486 -2.785* 13 .486 2.785*

*wp< .05
f r~ = frequency

DISCUSSION coverage than the competition must also improve product

quality. For example, the upgrading of the power train
This study was concerned with theory justification. The warranties of General Motors and Ford Motor Co. to
results contributed to the growing body of knowledge six years/60,000 miles and Chrysler's power train warranty
concerning the purpose of a consumer product warranty. to seven years/70,000 miles must be accompanied by a
The log-linear anallysis used in the present study differed significant improvement in product reliability if the
from previous tests of the Market Signal Theory in that warranty signal is going to be effective. Otherwise, the
more information was provided by estimating the consumer will learn through experience not to rely on
parameter values for warranty quality and brand the warranty signal and the result will be a short-term
reliability. These estimates allowed for tests of differences gain for the automobile manufacturers.
between levels of each variable. In addition, the resulting Concomitently, improving product quality may require
log-linear model extended the research conducted by changes in product design and materials used in building
Wiener (1985) in two ways. First, the theory held for pre- the product. Such changes usually involve a substantial
Act warranties as well as post-Act warranties. Second, capital investment. Hence the price of the product may
the theory held for product classes other than those studied have to be increased to cover this investment. An increase
by Wiener(1985). in price may have a dual effect in that the price may
signal product quality and reinforce the warranty signal.
Marketing Implications The marketer should evaluate different combinations of
price levels and selected warranty terms that will produce
The Market Signal Theory holds several implications a desired product-reliability signal.
for marketing decisions. First, marketers may use Second the Market Signal Theory suggests marketers
warranties to differentiate and position their products. with warranty terms that are better than the competition
Marketers that seek to position their product as very should be aggressive in promoting the warranty. However
reliable by offering a warranty that provides greater marketers that promote their warranties to signal product

JAMS 76 SUMMER, 1988

AN INVESTIGATIONOF CONSUMERPRODUCTWARRANTIESAS MARKETSIGNALSOF PRODUCTRELIABILITY KELLEY

reliability must exercise care in drafting advertising copy result suggests that any improvement in warranty terms
and in selecting media vehicles to communicate the attributed to competition initiated by the Magnuson-Moss
warranty signal. This is because the Market Signal Theory Act may have been augmented by consumer product
assumes that consumers read and understand the terms warranties serving as market signals of product reliability
of a warranty. Of this assumption is not met, the warranty before the Act was passed. Given time, the market through
signal will fail to communicate or communicate inaccurate consumer demand may have ultimately led to competition
product reliability information. Therefore, the marketer among firms to improve and voluntarily present to the
may have to educate consumers as to differences that consumer warranty terms without the Magnuson-Moss
exist in warranties in order to insure consumers know Act. Therefore the major contributor of the Act may have
how to use this information. For example, the marketer been to provide consistency in the market signal by
may have to inform consumers as to the difference between standardizing the form which warranties may be written.
a full and limited warranty. The distinguishing feature It was not the intent of the present study to determine
of a full warranty is that it covers parts and labor expenses whether the Magnuson-Moss Act has resulted in greater
whereas a limited warranty covers only parts. A full competition among firms to increase warranty coverage.
warranty could then be used in a comparative advertisement Assuming for the moment that the Act has not stimulated
to signal a particular brand's reliability advantage. competition a m o n g manufacturers to offer better
Third, the signaling effect of a warranty may force a warranties, the reason could be that warranties may not
change in the way warranty repair service is handled. be a major factor in purchase decisions. For example,
Consumers may view better warranties as also guaranteeing a recent survey reported only 7% of major appliance
prompt warranty service. If this service meets consumer consumers purchased a brand primarily because of the
expectations, the warranty signal will be reinforced and warranty (Wall Street Journal 1984). However this survey
the consumer may be more likely to purchase other did not consider whether consumers know how to interpret
products produced by the manufacturer. Warranty service the warranty signal. An effort may be needed to educate
may be performed by a network of dealers or the consumers in how to interpret warranties and use them
manufacturer. It may be attractive to have dealers perform in their purchase decisions. Through this education
the service since they are convenient for the customer. consumers may become more aware of warranties and
However dealers may require additional incentives to increase the role warranties play in purchase decisions.
service more comprehensive warranties. Such incentives Third, the Market Signal Theory may be used to
may include reimbursement for labor used in performing evaluate other consumer legislation that involves the
repair work above what it may cost the manufacturer disclosure of prepurchase information. For example the
to make the repairs and/or periodic training of dealer theory may be extended to nutritional labeling truth-in-
service personnel. Controlling the network of dealers who lending disclosures, and energy labeling. Perhaps it will
perform warranty service is also paramount if the warranty be found that the provisions of these acts require
is going to be an accurate signal of product reliability. information to be disclosed as signals but the consumer
The affect of the signal will be reduced unless there is cannot interpret the signal or combines the signal with
a consistency between the consumer's perception of what other product cues. The theory may offer a new perspective
the warranty terms signal and the consumer's experience in exploring how consumer disclsure legislation may be
with warranty service. A warranty service audit may be more fully implemented.
used to evaluate and control the performance of dealers
in the network (Wilkes and Wilcox 1976). FUTURE RESEARCH
An alternative to a dealer service network would be
for the manufacturer to provide the repair service. Even though the Market Signal Theory was supported
Providing multiple service outlets would offer greater in the present study, several questions remain. First, it
customer service (e.g., less distance to transport the is still unclear how important warranties are in the
product) but would be more expensive to operate. purchase decision. Even though the results indicated
Conversely, having one or a few service centers would warranties signaled product reliability information, the
be less costly, but potentially increase customer study did not test whether consumers actually use
dissatisfaction. The number of service outlets would warranties as market signals. Consumers may rely on more
depend on the characteristics of the product (e.g., large salient product-related signals, (e.g. price and brand name)
or small) and the level of service objective of the when making purchase decisions. Future research needs
manufacturer. In any event, if the manufacturer to to examine what combination of information cues (e.g.
promotes the warranty as a signal of product reliability price, warranty, investment in brand name advertising,
the warranty service strategy involving either dealer or etc.) contributes to the bundle of attributes referred to
manufacturer service centers must be consistent with as product quality/reliability and whether the effect of
consumer expectations. each variable is the same in all purchase decisions.
Second because the overwhelming number of firms offer
Public Policy Implications basic warrarty terms it is unclear why more firms do not
seize the opportunity to gain a competitive advantage by
The Market Signal Theory also holds several implications offering better warranty terms. Perhaps it is the cost
for public policy. The results indicated both pre- and post- involved in increasing product reliability alone with
Act warranties acted as signals of product reliability. This escalating repair costs that makes offering better

JAMS 77 SUMMER, 1988

AN INVESTIGATION OF CONSUMER PRODUCT WARRANTIES AS M A R K E T SIGNALS OF PRODUCT RELIABILITY KELLEY

warranties unprofitable. A method for estimating the costs Consumer Reports. 1983. ~Dishwashers." 48 August: 406-411.
of service repair quality control, and consumer perceptions _ _ 1983. "Refrigerator/Freezers." 48 January: 21-25.
of less reliable products is needed. In addition, modeling _ _ 1982. "Washing Machines." 47 October: 508-512.
the relationship between these costs remains for future _ _ 1982. "Air Conditioners." 47 July: 356-361.
research. _ _ 1982. "Clothes Dryers." 47 January: 31-37.
Third, confirmation of the Market Signal Theory _ _ 1980. "Clothes Dryers. ~ 45 October: 623-627.
requires more than two studies. Additional studies are _ _ 1980. "Air Conditioners." 45 July: 428-433.
needed to add to the mounting evidence that supports _ _ 1980. "Builtin Dishwashers." 45 June: 366-370.
this theory. Such a study may compare the Exploitation _ _ 1979. "Washing Machines." 44 November: 650-654.
Market Signal and Investment theories. Competing _ _ 1976. ~Builtin Dishwashers."41 May: 280-287.
1975. "Clothes Dryers." 40 November: 686-689.
theories that incorporate other product cues and consumer _ _

1975. ~Washing Machines." 40 October: 611-615.

perceptions also should be developed. It is only through
_ _

1975. ~No Frost Refrigerators." 40 September:. 564-569.

a process of confirmatory and comparative testing that
_ _

_ 1975. "Room Air Conditioners. ~ 40 July: 410-416.

_

marketing knowledge of how consumer warranties are used _ _ 1974. "Washing Machines." 39 October: 721-727.
by manufacturers and consumers be expanded (Sternthal, _ _ 1974. "Room Air Conditioners. ~ 39 July: 516-522.
Tybout and Calder 1987). _ 1974. "Builtin Dishwashers." 39 May: 386-390.
_

_ _ 1973. "Refrigerators." 38 September: 545-552.

_ _ 1972. "Clothes Dryers." 37 September: 482-587.
SUMMARY Feldman, Laurence P. 1976. "New Legislation and the Prospects of Real
Warranty Reform." Journal of Marketing 40 July: 41-47.
Few attempts to develop and test a theory of the Gerner, Jennifer, and W. Keith Bryant. 1981. "Appliance Warranties as a
consumer product warranty exists in the marketing Market Signal?" Journal of Consumer Affairs 15 Summer: 75-86.
literature. The present study extended previous investi- Goodman, L. A. 1970. "The Multivariate Analysis of Qualitative Data:
Interactions Among Multiple Classifications." Journal of the American
gations of the Market Signal Theory when consumer Statistical Association 65 March: 226-256.
product warranties were found to signal product reliability Kelley, Craig A. 1984. "Extending the Understanding of Consumer Product
information prior to and after the Magnuson-Mass Act Warranty Theories Via Brand Specific Testing." In Beyond 1984 in
was passed. This evidence corroborated support of the Marketing Education Bruce J. Walker and David L. Kurtz (eds.). Western
Marketing Educators' Association.
theory found in previous research. The fact that warranties
Kennedy, John J. 1983. Analyzing Qualitative Data: Introductory Log-Linear
act as signals of product reliability has important Analysis for Behavioral Research. New York: Praeger Publishing.
implications for marketing practitioners and public Magnuson-Moss Warranty - - Federal Trade Commission Improvement Act.
policymakers. However additional theory development 15 U.S.C. 2301.
and testing should continue as other product cues may Priest, George. 1981. ~The Theory of the Consumer Product Warranty."
mediate or interact with a warranty to signal product Yale Law Journal 90 May: 1297-1352.
reliability. Spence, Michael. 1977. "Consumer Misperccptions, Product Failure, and
Producer Liability." The Review of Economic Studies 44 October: 561-
572.
Sternthal, Brian, Alice M. Tybout and Bobby J. Calder. 1987. "Confirmatory
FOOTNOTES Versus Comparative Approaches to Judging Theory Tests." Journal of
Consumer Research 14 June: 114-125.
~The Herfindal Index is used by economists to indicate the degree of Wall Street Journal. 1984. "Aggressive Use of Warranties is Benefiting Many
concentration in an industry. The index consists of the sum of the squares Concerns." April 5: 33.
of the market shares of each firm in an irdustry. Herfindal values range Wilkes, Robert E., and James B. Wilcox. 1976. "Consumer Perceptions of
from close to 0 (no concentration) to 1 (monopoly).
Product Warranties and Their Implications for Retail Strategy." Journal
2The BMDP/4F Program is used to evaluate multiway frequency tables.
of Business Research 4 February: 35-43.
Several log-linear model selection and estimate options are provided in the
program. Wiener, Joshua, L. 1985. "Are Warranties Accurate Signals of Product
Reliability?" Journal of Consumer Research 12 September: 245-250.

REFERENCES ABOUT THE AUTHOR

Akerlof, George. 1970. "The Market for Lemons: Quality Uncertainty and CRAIG A. KELLEY is Associate Professor of
the Market Mechanism." Quarterly Journal of Economics 84 3: 488-500. Marketing at the California State University, Sacramento.
Brown, Morton B. 1983. "Two-Way and Muhiway Frequency Tables - - He received his Ph.D. from Arizona State University. His
Measures of Association and the Log-Linear Model." In BMDP Statistical
Software. W. J. Dixon (ed.). Los Angeles: University of California Press.
research emphasis is in the area of the legal aspects of
Bruckes, Merrie, Andrew A. Mitchell, and Richard Staelin. 1984. "The Effect
marketing. He has had articles published in the Journal
of Nutritional Information Disclosure in Advertising. An Information of the Academy of Marketing Science, Journal of
Processing Approach." Journal of Public Policy and Marketing 3: 1-25. Marketing Education, and Journal of Services Marketing.

JAMS 78 S U M M E R , 1988