Consumers' Responses to Negative Word-of-Mouth Communication: An Attribution Theory Perspective
Author(s): Russell N. Laczniak, Thomas E. DeCarlo and Sridhar N. Ramaswami
Source: Journal of Consumer Psychology , 2001, Vol. 11, No. 1 (2001), pp. 57-73
Published by: Wiley
Stable URL: https://www.jstor.org/stable/1480311
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms
Wiley is collaborating with JSTOR to digitize, preserve and extend access to Journal of Consumer Psychology
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
JOURNAL OF CONSUMER PSYCHOLOGY, 11(1), 57-73 Copyright ? 2001, Lawrence Erlbaum Associates, Inc.
Consumers' Responses to Negative Word-of-Mouth Communication: An Attribution Theory Perspective
Russell N. Laczniak, Thomas E. DeCarlo, and Sridhar N. Ramaswami Department of Marketing
Iowa State University
Research on negative word-of-mouth communication (WOMC) in general, and the process by which negative WOMC affects consumers' brand evaluations in particular, has been limited. This study uses attribution theory to explain consumers' responses to negative WOMC. Experi- mental results suggest that (a) causal attributions mediate the negative WOMC-brand evalua- tion relation, (b) receivers' attributions depend on the manner in which the negative WOMC is conveyed, and (c) brand name affects attributions. Results also suggest that when receivers at- tribute the negativity of the WOMC message to the brand, brand evaluations decrease; however,
if receivers attribute the negativity to the communicator, brand evaluations increase.
Word-of-mouth communication (WOMC) is an important marketplace phenomenon by which consumers receive infor- mation relating to organizations and their offerings. Because WOMC usually occurs through sources that consumers view as being credible (e.g., peer reference groups; Brooks, 1957; Richins, 1983), it is thought to have a more powerful influ- ence on consumers' evaluations than information received
through commercial sources (i.e., advertising and even neu- tral print sources such as Consumer Reports; Herr, Kardes, & Kim, 1991). In addition, this influence appears to be asym- metrical because previous research suggests that negative WOMC has a stronger influence on customers' brand evalua- tions than positive WOMC (Amdt, 1967; Mizerski, 1982; Wright, 1974). Given the strength of negative, as opposed to positive WOMC, the study presented here focuses on the for- mer type of information.
Our research develops and tests, using multiple studies, a set of hypotheses that describes consumers' attributional and evaluative responses to different types of negative-WOMC messages. The hypotheses posit that consumers will generate predictable patterns of attributional responses to nega- tive-WOMC messages that are systematically varied in terms of information content. Furthermore, they predict that attributional responses will mediate the negative WOMC-brand evaluation relation. Finally, and similar to re- cent studies (cf. Herr et al., 1991), the hypotheses suggest
Requests for reprints should be sent to Russell N. Laczniak, Iowa State University, Department of Marketing, 300 Carver Hall, Ames, IA 50011-2065. E-mail: [email protected]
consumer responses to negative WOMC are likely to be influenced by strength of the targeted brand's name.
This study extends research on negative WOMC in two im-
portant ways. First, whereas previous studies have typically examined receivers' responses to a summary statement of a fo-
cal brand's performance (cf. Bone, 1995; Herr et al., 1991), it is
likely that the information contained in negative-WOMC mes-
sages is more complex than this. In this study, focal messages
are manipulated to include three components of information besides the communicator's summary evaluation (Richins, 1984). Messages include information about the (a) consensus of others' views of the brand (besides the communicator), (b) consistency of the communicator's experiences with the brand
over time, and (c) distinctiveness of the communicator's opin-
ions of the focal brand versus other brands in the category. In-
terestingly, these types of information correspond to the information dimensions examined in Kelley's (1967) seminal work dealing with attribution theory. It is also important to note
that although others have used this work to model individual responses to another's actions (e.g., observing someone's in- ability to dance), this study is the first that empirically extends
Kelley's research into a context in which consumers interpret a conversation about a brand.
Second, whereas other studies have posited the existence of a direct relation between negative WOMC and postexposure brand evaluations (e.g., Amdt, 1967; Haywood, 1989; Katz & Lazerfield, 1955; Morin, 1983), our investigation examines the attributional process that explains this association. This approach is consistent with the thinking of several researchers (i.e., Bone, 1995; Herr et al., 1991; Smith & Vogt, 1995) who posited that cognitive mechanisms
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
58 LACZNIAK, DECARLO, RAMASWAMI
are important, as they can more fully explain the negative WOMC-brand evaluation linkage. Furthermore, this re- search is consistent with other studies that suggest (but do not
test the notion) that receivers' cognitive processing of nega- tive WOMC involves causal attributional reasoning (cf. Folkes, 1988; Mizerski, Golden, & Kernan, 1979).
( Negative WOMC Information
Configuration
Causal Attributions }_( Brand Evaluationj
(Brand Name Strength t
THEORY AND HYPOTHESES
Negative WOMC FIGURE 1 Attributional process model for receivers of negative word-of-mouth communication.
Negative WOMC is defined as interpersonal communication concerning a marketing organization or product that deni- grates the object of the communication (Richins, 1984; Weinberger, Allen, & Dillon, 1981). Negative WOMC po- tentially has a more powerful influence on consumer behav- ior than print sources, such as Consumer Reports, because in-
dividuals find it to be more accessible and diagnostic (Herr et al., 1991). In fact, research has suggested that negative WOMC has the power to influence consumers' attitudes (Engel, Kegerreis, & Blackwell, 1969) and behaviors (e.g., Arndt, 1967; Haywood, 1989; Katz & Lazerfield, 1955).
Attributions as Responses to Negative WOMC
Because the transmission of negative WOMC involves inter- personal and informal processes, attribution theory appears to
be particularly helpful in understanding a receiver's interpre-
tation of a sender's motives for communicating such informa-
tion (Hilton, 1995). The central theme underlying attribution theory is that causal analysis is inherent in an individual's need to understand social events, such as why another person would communicate negative information about a brand (Heider, 1958; Jones & Davis, 1965; Kelley, 1967). For this study, causal attribution is defined as the cognition a receiver
generates to infer the cause of a communicator's generation of negative information (Calder & Burkrant, 1977).
Figure 1 illustrates the proposed process consumers use to deal with negative WOMC. Specifically, it proposes two im- portant influences on receivers' attributional responses to negative-WOMC communication. First, the information con- veyed by the sender in a negative-WOMC message is posited to influence receivers' causal attributions. Second, brand-name strength of the focal brand is also thought to di- rectly affect receivers' causal attributions. These attributional
responses, in turn, are expected to affect receivers' brand evaluations. Therefore, this study suggests that attributions mediate the presupposed negative-WOMC-brand evaluation relation. Such a model is consistent with theoretical frame-
works of interpersonal communication that suggest that attri- butions mediate an interpersonal message's effect on a receiver's evaluation of the focal object (e.g., Hilton, 1995).
There is additional support for the mediational role played
by attributions in influencing individuals' brand evaluations. For example, studies in the advertising literature have sug- gested that receivers generate causal attributions that in turn affect their evaluations of the advertised brand (e.g., Wiener & Mowen, 1986). In the performance evaluation literature, studies indicate that sales manager attributions of salesperson
performance shape their reactions toward a salesperson (e.g., DeCarlo & Leigh, 1996). Thus, the following is proposed for receivers of negative WOMC:
H1: Causal attributions will mediate the effects of
negative WOMC on brand evaluations.
Information Type and Causal Attributions
According to research in classical attribution theory (Kelley, 1967, 1973), the categories of causal attributions that people generate in response to information include: stimulus (i.e., brand, in this case), person (i.e., communicator, in this case), circumstance, or a combination of these three.1 The specific type of attributions generated by individuals, how- ever, depends on the manner in which information is con- veyed. According to attribution theory (Kelley, 1967) and other studies dealing with WOMC (e.g., Richins, 1984), a re- ceiver is likely to use three important information dimensions
to generate causal attributions: consensus, distinctiveness, and consistency. In a negative-WOMC context, the consen- sus dimension refers to the degree to which others are likely to
agree with the negative views of the communicator. The dis- tinctiveness dimension encapsulates the extent to which the communicator associates the negative information with a par- ticular brand but not other brands. Finally, the consistency di-
'Although attribution theory suggests that individuals have the potential
to generate multiple and interactive attributional responses, this study fo- cuses only on those attributions that are thought to have a significant impact on brand evaluations in the negative-WOMC context (i.e., brand and com- municator attributions).
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
CONSUMERS' RESPONSES 59
mension deals with the degree to which the communicator has had stable negative experiences with the brand across time and situations.
The three dimensions of information noted previously are typically viewed as being high or low in occurrence. For example, WOMC that is high on the consensus dimen- sion would suggest that others (beside the communicator) have had problems with the focal brand. Low-consensus information, on the other hand, would indicate that the communicator does not know anyone (besides him- or her- self) who has had problems with the focal brand. Com- bining the three information dimensions yields eight (2 x 2 x 2) potential configurations. It is commonly noted in attri- bution research (e.g., Hilton & Jaspars, 1987; Kelley, 1972; Teas & McElroy, 1986) that two (out of the eight possible) combinations provide theoretically unambigu- ous information (and thus are relevant to the study of nega- tive WOMC). These combinations are the high-consensus, high-distinctiveness, and high-consistency information configuration, as well as the low-consensus, low-distinc- tiveness, and high-consistency information configuration. A third combination-the low-consensus, high-distinc- tiveness, low-consistency information-has been posited to be the most ambiguous configuration (Hilton & Jaspars, 1987). Accordingly, the research presented here focuses on negative-WOMC messages utilizing these three infor- mation combinations.
Previous research has suggested that receivers will more likely attribute the consequences of an action (i.e., the negativity in a WOMC message) to the communication ob- ject (i.e., the focal brand) when information is configured as being high consensus, high distinctiveness, and high consis- tency, in comparison to the other information configurations
(Hilton & Jaspars, 1987; Kelley, 1967). Information con- tained in negative WOMC using this type of configuration will likely be viewed by receivers as more logical and well de- veloped than that configured as low consensus, low distinc- tiveness, and high consistency, or low consensus, high distinctiveness, and low consistency. Specifically, the infor- mation is considered to be logical because the communicator indicates that he or she has had repeated bad experiences with
the focal brand (high consistency), knows of many others who have had problems with the focal brand (high consen- sus), and also believes that most other brands are of high qual-
ity (high distinctiveness). After receiving such focused and cogent arguments, we believe that consumers will generate stronger brand attributions to it as compared to the other nega-
tive-WOMC configurations. Receivers of a low-consensus, low-distinctiveness, and
high-consistency configuration, on the other hand, should be inclined to direct the negativity of the information toward the communicator for a number of reasons. Information config- ured in this manner is less logical and persuasive than WOMC configured as high consensus, distinctiveness, and consis- tency because it provides an inconsistent and critical view of
not just the focal brand, but all brands in the product class. Therefore, the communicator may be viewed as being overly
negative and opens him- or herself to critical attributions. Furthermore, past research has indicated that individuals may consider a communicator's assessment as typical behavior in the absence of prior knowledge about the communicators' motives (Hilton, Smith, & Alicke, 1988). For negative WOMC configured as low consensus, low distinctiveness, and high consistency, the communicator's assessment may be perceived as containing more information about the com- municator than the focal brand of the conversation. Spe- cifically, the communicator bases the negative argument on his or her repeated bad experiences (high consistency), but when questioned, the communicator provides information that no one else he or she knows has had problems (low con- sensus) and that he or she feels all brands are of low quality (low-distinctiveness information). Thus, the communicator is likely to be viewed as contradictory and the receiver should
attribute the negativity toward the communicator.
This study utilizes negative WOMC configured as low consensus, high distinctiveness, and low consistency as a comparison to the other two scenarios for a number of rea- sons. First, there is significant evidence that this configura- tion is the most ambiguous with respect to brand and communicator attributions (cf. Hilton & Jaspars, 1987; Iacobucci & McGill, 1990). Second, research indicates that the perceived informativeness of this configuration about the
focal object or person is low (Hilton & Slugoski, 1986). Be- cause ofthis, we believe that the low-consensus, high-distinc- tiveness, low-consistency pattern is uniquely suited as a comparison configuration to the two patterns that yield unam-
biguous attribution patterns.2 Thus, all hypothesized compar-
isons of the strength of specific attributional responses made
by receivers of the high-consensus, high-distinctiveness, and high-consistency scenario versus the low-consensus, low-distinctiveness, and high-consistency scenario are im- plicitly made in comparison to this configuration as well. Therefore, the following comparison hypotheses for the three
negative-WOMC configurations are:
H2: Consumers exposed to negative WOMC config- ured as high consensus, high distinctiveness, and
high consistency will be more likely to attribute
the negativity of the message toward the brand than those receiving other configurations.
H3: Consumers exposed to negative WOMC config- ured as low consensus, low distinctiveness, and
high consistency will be more likely to attribute the negativity of the message toward the commu-
nicator than those receiving other configurations.
2As will be seen in Study 2, we examine responses to alternative informa-
tion configurations.
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
60 LACZNIAK, DECARLO, RAMASWAMI
Brand Name and Causal Attributions
Prior research has suggested that the effects of negative WOMC on brand evaluations are likely to be reduced when prior positive brand impressions exist in consumers' memo- ries (Herr et al., 1991). Consistent with this notion, we con- tend that the effect of a brand name is likely to influence consumers' attributional processing of negative WOMC. A more-favorable brand name is expected to reduce the persua- siveness of negative WOMC because impression-inconsis- tent information is typically deflected away from the brand and discounted (Hoch & Deighton, 1989). Such a view is con- sistent with research in attribution theory, which suggests that
attributions directed at the focal object are unlikely to be gen-
erated by receivers who have favorable associations with it (Harvey & Weary, 1984). Thus, based on the cognitive pro- cess mechanisms of attributional biasing (De Nisi, Cafferty, & Meglino, 1984; Regan, Straus, & Fazio, 1974) and dis- counting (Sanbonmatsu, Kardes, & Gibson, 1989), receivers should attribute negativity away from the focal brand when the negative information (about the focal brand) is inconsis- tent with a positive brand name. On the other hand, negative
WOMC is more likely to fit with consumers' associations for less-favorable brand names and should reinforce these asso-
ciations (cf. Wilson & Peterson, 1989). Therefore, the follow- ing hypothesis is proposed:
H4: Consumers exposed to negative WOMC about a more-favorable brand will be less likely to attribute
the negativity toward the brand than those receiv-
ing negative WOMC about a less-favorable brand.
However, if not toward the brand, to where (or to whom) do
receivers of negative WOMC about more-favorable brands at- tribute the negativity? It is our contention that receivers of neg-
ative information about favorable brands will be more likely to
attribute the negativity of the WOMC toward the communica-
tor. Hilton's (1995) model of social communication suggested that, all things being equal, a receiver assumes that a conveyor
of interpersonal information is trying to be helpful and conse-
quently should be positively disposed toward the communica- tor at the time of exposure. However, when contradictory information (i.e., negative information about a favorable brand
name) is presented by a communicator, the receiver will shift
his or her impressions toward the negative, leaving him or her
in a state of cognitive imbalance. Hilton's model suggested this imbalance will be overcome by the receiver's attributing the negativity of the message toward the communicator. Thus, we posit the following:
H5: Consumers exposed to negative WOMC about a more-favorable brand will be more likely to attribute the negativity toward the communica- tor than those receiving negative WOMC about a less-favorable brand.
Attributions and Brand Evaluation
The influential role of brand attributions on brand evalua-
tion is consistent with cognitive processing models of perfor- mance evaluation (e.g., DeCarlo & Leigh, 1996; De Nisi et al., 1984). In these models, there is a direct linkage between performance attributions of an employee and his or her per- formance evaluation. In other words, when supervisors attrib-
ute negative performance to the subordinate, evaluations tend
to be lower. Extending this notion to the negative-WOMC context, we posit that brand attributions will have a negative effect on brand evaluations. This is the case because as re-
ceivers link the negativity of WOMC messages to a brand (via
brand attributions), their evaluations should be negatively af-
fected. Specifically, we hypothesize the following:
H6: The strength of brand attributions generated in response to negative WOMC about a particular brand will be inversely related to brand attitudes.
We hypothesized that communicator attributions will be directly related to brand evaluations (i.e., a positive relation is
expected between communicator attributions and brand eval- uations). This view is consistent with theories of conversation
that suggest that the interpretation of social discourse requires
a form of cognitive balancing (cf. Brown & van Kleeck, 1989). In the process of communicating negative WOMC, the communicator is establishing a negative link between them- selves and the brand (in the eyes of the receiver). When the re-
ceiver attributes negativity toward the communicator, his or
her reasoning process will allow him or her to disassociate it from the brand. Once blame is assessed toward the communi-
cator, Hilton (1995) suggested that the receiver will rally to the defense of the brand and in fact be more supportive of it.
Thus, it is hypothesized that:
H7: The strength of communicator attributions gener-
ated in response to negative WOMC about a partic-
ular brand will be directly related to brand attitudes.
STUDY 1
Method
Overview
Participants were randomly assigned to one of six cells in a 3 (negative-WOMC information scenario) x 2 (brand-name strength) full-factorial experiment. The negative-WOMC message was manipulated to provide three information con- figurations, as noted previously. Strength of brand name was manipulated by providing negative WOMC about more- ver- sus less-favorable brands. Dependent measures included multi-item measures of causal attributions (i.e., measures of
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
CONSUMERS' RESPONSES 61
the strength of brand and communicator attributions) and postexposure brand evaluations. Several manipulation checks were also obtained.
Stimuli
Product. Personal computer (or PC) brands were fea- tured as the experimental stimuli for a number of reasons. First,
previous research has indicated that there must be sufficient
motivation for consumers to generate and receive negative WOMC (Richins, 1984). Because personal computers are complex and relatively expensive products, there is a reason- ably high probability that consumers may be involved with the
product class and thus be motivated to attend to and process negative WOMC about brands within the product class. Sec- ond, we believe that both high- and low-product knowledge in-
dividuals use personal computers; thus, even low-knowledge people might engage in negative WOMC.3 Finally, personal computers have been used as experimental stimuli in similar negative-WOMC studies (Herr et al., 1991). The potential ef- fects of product involvement and product knowledge are as- sessed by including them as covariates in the study.
Brand-name strength. The starting point for selecting brands that differed in name strength was to use the results of
a study sponsored by the Wall Street Journal (Pope, 1993) in the personal computer category. According to the results of this study, the brands Packard Bell and Everex ranked 9th and
10th, respectively, whereas IBM and Compaq were 1st and 2nd, respectively. Thus, two less-positive brand names (Packard Bell and Everex) and two more-positive brand names (IBM and Compaq) were pretested.
A number of brand-name strength measures were used in preexperimental pilot tests and the main experiments as ma- nipulation checks. The measures included scales of brand fa- miliarity, brand affect, brand awareness, favorable brand associations, and purchase intention. These measures were used in previous empirical research dealing with brand equity (Cobb-Walgren, Ruble, & Donthu, 1995). Pilot test results in- dicated that participants rated the Packard Bell and Everex computers significantly lower on these dimensions than IBM and Compaq computers. Furthermore, no significant differ-
30ur sample indicated having a high degree of involvement with the per-
sonal computer product class. Product class involvement was assessed with Zaichkowsky's (1985) scale. Factor analysis results suggest that this scale contained two dimensions: one appearing to be a utilitarian involvement fac-
tor (a =.78) and the other as an aesthetic involvement factor (a = .91). Partic- ipants' mean level of utilitarian involvement is 6.11 on a 7-point scale, rang- ing from 1 (low involvement) to 7 (high involvement), and the mean level of aesthetic involvement is 5.37 on a similar 7-point scale-well above the mid- points of each scale. In addition, the sample reported having a moderate level of personal computer knowledge (M= 6.07, SD = 3.21), with 10 being a per- fect score. Personal computer knowledge was assessed with a 10-item objec- tive measure developed by the authors (KR-20 = .91).
ences were found between the two less-positive brand names (i.e., Packard Bell and Everex). However, the within-cell analysis for the more-positive brand-name computers indi- cates a significant difference between IBM and Compaq. IBM was rated more positive in our pilot tests than Compaq. However, we selected the more moderately rated Compaq brand name for use in the main experiments. We believed that
using Compaq (vs. IBM) would be a stronger test of the power of brand name in consumers' processing of negative WOMC. Specifically, if differences are noted between Everex and Compaq brands in the main experiment, it would seem logical to conclude that the difference between Everex and IBM would be greater.
Negative-WOMC scenarios. Three distinct nega- tive-WOMC conditions were developed using high and low levels of consensus, distinctiveness, and consistency infor- mation, as previously discussed. These configurations were: (a) high consensus, distinctiveness, and consistency; (b) low consensus, low distinctiveness, and high consistency; and (c) low consensus, high distinctiveness, low consistency.
Each negative-WOMC scenario contained identical infor- mation with the exception of the statements of either high or low levels of consensus, consistency, and distinctiveness in- formation, and the brand described in the scenario. Care was
taken to ensure that relatively equal total word counts (high consensus, distinctiveness, and consistency = 208; low con- sensus, low distinctiveness, and high consistency = 210; and low consensus, high distinctiveness, low consistency = 206) were used for each scenario.
The negative-WOMC stimulus was a tape-recorded con- versation that occurred between two confederates. The con-
federates read the scenarios from a script provided by the authors. To ensure that the WOMC sounded natural, the con-
federates rehearsed the script a number of times. The confed-
erates included a man (Pat), a perceived computer expert and the provider of the negative comments, and a woman (un- named), the person interested in the other's opinion. (A tran-
script of each scenario is provided in the Appendix.)
Participants
A total of 192 male and female undergraduate students was recruited from various business classes at a major midwester
university. Participants received extra course credit for their participation in the study. Because of missing data, 11 partici-
pants were dropped from the analysis phase of the study. Thus, the final sample size was 181 (103 men and 78 women). The sample distribution across the six experimental cells ranged from a low of 26 (14%) to a high of 35 (19%) partici- pants. The use of a homogeneous sampling frame (e.g., un- dergraduate students) is appropriate as recommended theory test procedures require selection of respondent groups such that rigorous examinations can be conducted (Calder, Phil- lips, & Tybout, 1982).
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
62 LACZNIAK, DECARLO, RAMASWAMI
Procedure
Participants were randomly assigned to one of the six experi-
mental cells. They were handed a booklet containing instruc- tions, a transcript of the negative WOMC, and related ques- tionnaire items. The first page of the booklet contained instructions that were read aloud by a proctor who invited the participants to read along. This procedure was used to increase
participants' involvement with the conversation that is likely to occur in face-to-face conversations. The instructions read:
The University Microcomputer Products Center and Marketing Department have been gathering informa- tion from focus groups to determine what students look
for in computers. These focus groups are routinely taped to capture as much information as possible. One session included an interesting conversation that oc- curred between two students during a session break. Because the conversation took place during a break, the quality of the tape was poor. So, you are going to listen to a reenacted audio tape of this conversation.
PLEASE LISTEN TO THIS CONVERSATION
VERY CAREFULLY. While listening to this tape, keep in mind that we will be asking you to make judg- ments regarding the brand that these two people are talking about, the people who are having the conversa- tion, and the situation that surrounds this conversation.
So, please follow this conversation carefully, because we will be asking you your opinions about the brand, the people, and the situation. A transcript of this con- versation has been provided. This may help you more closely follow the conversation.
Although the experimental procedure used in this study did not allow participants to hear face-to-face negative com- ments about the experimental brands as was the case in previ- ous studies (cf. Herr et al., 1991), it did allow the experimenters to achieve greater control of message content. Given that the focus of this study was to determine consum-
ers' attributional and attitudinal responses to a variety of mes-
sages (and not on the accessibility of the message compared to memory information), we concluded that the tradeoff be-
tween the realism associated with a face-to-face message, versus a tape-recorded one with greater message content con- trol, was appropriate.
The second page of the booklet contained the transcript of the assigned negative-WOMC scenario. To ensure that partici- pants were actively involved in the experiment, they were told
to read the transcript while listening to the taped conversation.
Immediately following the conclusion of the tape, participants were asked to turn the page and read general questionnaire in- structions. The participants were then instructed to turn the page to the scaled causal attribution measures and read them prior to completing the scales. These instructions read:
Pat made some negative comments about (brand name inserted). There may be many reasons why he did so, but what we are interested in is your thoughts about why you think Pat made these comments. There are no
right or wrong reasons, we simply want to find out your
feelings for the statements below. You may refer back to the transcript on page 2 to help you respond.
Following these instructions, participants answered ques- tions regarding their evaluation of the scaled attributions, brand evaluation, and questions related to the effectiveness of
the manipulations. Three separate manipulation-check groups (comprised of
participants similar to those used in the main experiments) were used to test the efficacy of the brand-name manipula- tions and to check perceptions that each negative-WOMC scenario included consensus, consistency, and distinctive- ness information dimensions. In the first group, participants
(n = 79) were given the brand-name manipulation check mea- sures for familiarity and affect for each brand, as well as mea-
sures of brand evaluation. For the second pilot test, participants (n = 129) were provided with questionnaires that included four additional measures of brand-name strength: brand awareness, ad awareness, brand associations, and pur- chase intention (Cobb-Walgren et al., 1995). Participants completed these questionnaires for either the Compaq (n = 65) or Everex brands (n = 64). The third group (n = 68) lis- tened to the negative WOMC for the Compaq brand, and then
they were asked to complete the manipulation check for the consensus, consistency, and distinctiveness information in the negative-WOMC scenarios.
Measures
Causal Attributions
Multi-item scales were developed to assess communicator and brand attributions for the cause of the negative information
conveyed in the scenarios. The two measures were based on recommendations made by Lichtenstein and Bearden (1986) and Weiner (1980). In addition, the scale-development proce- dure used followed Churchill's (1979) paradigmatic recom- mendations. First, a large pool of items was developed to correspond to the brand and communicator attribution con- structs. To capture the richness of possible attributions, a num-
ber of theoretical perspectives from the literature were utilized.
Communicator attributions were developed from Wind's (1978) categories of people, which included demographic-so- cioeconomic factors, personality-lifestyle factors, and atti- tudes-behaviors toward a product. Brand attributions were developed with the aid of Enis and Roering's (1980) levels of product offerings. Two pilot tests (the first with n = 50 and the
second with n = 44) were then conducted to purify the mea- sures. As a result of the tests and pretest, participants' re- sponses to open-ended questions in test scenarios-several
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
CONSUMERS' RESPONSES 63
items (from each measure)-were eliminated, and others were revised to improve their specificity and precision.
For the final measures, in which we used 7-point scales ranging from 1 (strongly agree) to 7 (strongly disagree), ex- ploratory factor analysis of the nine causal attribution items yielded two factors accounting for 84% of the variance. The lowest item loading was .63. Inter-item correlations ranged from .66 to .91. The following header was used in measuring all causal attributions: “Pat said these negative things about
[insert brand name] because.” The following specific items were used to measure brand attributions: “This PC is an
inferior product,” “This PC is an unpopular brand,” “This PC's performance was poor,” “This PC is unusual,” and “This PC lacked the features Pat wanted or needed.” The fol-
lowing specific items were used to measure communicator at- tributions: “He doesn't know enough about using PCs,” “He does not appear to have the expertise to evaluate the product properly,” “He wanted to look smarter than he really is,” and
“He's the type of person who always says bad things about brands.” For the two scales, a = .95 (brand attributions) and a =95 (communicator attributions).
Attitude Toward the Brand
Three bipolar brand-attitude items on a 7-point scale, ranging
from 1 (bad, unfavorable, negative) to 7 (good, favorable, posi-
tive), were summed to form the postexposure brand evaluation
measure (Ajzen & Fishbein, 1980). Factor analysis of these items
yielded one factor explaining 91% of the variance. Loadings ranged from .92 to .95, and inter-item correlations ranged from
.88 to .93. For the brand evaluation measure, a = .95 (higher strength brand name) and a = .93 (lower strength brand name).
Manipulation Checks
Multiple items were used to ensure that the brand-name treatments provided participants with stimuli that varied in their degree of perceived strength. Measures were developed to tap the brand familiarity and affect dimensions of brand name. Factor analysis on items corresponding to both dimen- sions yielded a two-factor solution with expected item load- ings. Specific items for the affect dimension of the brand scale include: “My overall opinion about this brand of com- puter is very favorable,” “I have positive feelings about this brand,” “I really like this brand of computer,” and “Using this
brand of computer makes me feel good about myself.” Items for the familiarity dimension of the brand equity scale in- clude: “If I had to name a single brand to represent all comput-
ers, it would be this one,” “When I think of computers, this is the brand that comes to mind,” and “This brand is a very good example of my image of what a computer is.” The internal consistency of the two measures (i.e., familiarity and affect) was in the range of 0.68 and 0.79.
The measures of brand awareness, ad awareness, brand as- sociations, and purchase intention used in the post hoc test
were similar to those employed by Cobb-Walgren et al. (1995).
Specifically, brand awareness and ad awareness (“Have you heard of computers?” and “Have you ever seen any ad- vertising for ?”) were assessed with one-item, 3-point scales ranging from 1 (yes), 2 (no), to 3 (don 'tknow). Brand as-
sociations were obtained by asking participants the following question: “When you think of computers, what descrip- tive words, thoughts, symbols, or images come to mind? Please
list all that you can think of no matter how simple or irrelevant
they may seem to you.” Purchase intention (“Ifyou were to buy
a computer, how likely are you to purchase a computer?”) was assessed with one item on a 5-point scale ranging from 1 (very unlikely) to 5 (very likely).
Several measures were also gathered to ensure the efficacy
of the WOMC manipulations. The manipulation check for consensus information utilized a single-item measure: “Ac- cording to Pat, a number of people have had problems with
[insert brand name] every time they boot one up.” The consistency and distinctiveness manipulations used multi-item, 7-point scales ranging from 1 (strongly agree) to 7
(strongly disagree). Consistency items include: “Pat has had problems with [insert brand name]” and “Pat has prob- lems with [insert brand name] every time he boots one up.” Distinctiveness items include: “Pat usually does not say many good things about personal computers,” “Pat does not like many of the other computer brands in the market,” and “Pat likes other personal computer brands, but not [insert brand name]. Factor analysis and reliability indexes (a = .98 and a = .88 for the consistency and distinctiveness measures, respectively) indicate that these measures were satisfactory.
Results and Discussion
Brand Name and WOMC
Manipulation Checks
To ensure that the brand-name treatment was sufficiently
strong, 40 participants similar to those, but not included, in the main experiment were asked to respond to the brand scales discussed previously. Different participants were used to avoid contaminating the main experimental results from measurement context effects that might have arisen from these measures (Feldman & Lynch, 1988). Results suggest that the brand-name manipulation was appropriate because the affect (M = 4.19, SD = 1.09 for the more-positive brand name; M= 3.22, SD = 1.12 for the less-positive brand name), t(38) = 3.62,p < .001, and familiarity (M= 3.52, SD = 1.28 for the more-positive brand name; M = 2.25, SD = 1.22 for the less-positive name), t(39) = 4.29,p < .000, measures yielded significant differences in the expected direction.
A total of 68 participants, similar to those used in the main ex-
periment, were used to test the efficacy of the negative-WOMC
scenarios. Results of multivariate analysis of variance (MANOVA) procedures indicate that the three communication scenarios yielded differences across the dependent variables of
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
64 LACZNIAK, DECARLO, RAMASWAMI
perceptions of message consensus, consistency, and distinctive-
ness, F(6, 126) = 133.20,p < .001 (Wilks' X = .02). Subsequent one-way analyses of variance (ANOVAs) using Fisher's least significant difference test were run to assess the independent ef-
fects of the high and low levels of consensus, consistency, and
distinctiveness information in the three negative-WOMC sce- narios. The analyses show strong results for the consensus, F(2,
65) = 139.53,p < .001; consistency, F(2, 65) = 176.40,p < .001; and distinctiveness, F(2, 65) = 182.38, p < .001, manipulation checks. The cell means (see Table 1) for consensus, consistency,
and distinctiveness indicate significant contrasts in the expected direction for the three conditions.
The post hoc strength of brand-name analyses for brand awareness, X2(1) = 76.91, p < .01, indicates that more partici- pants were aware of Compaq (54 out of65 participants, 83.1%) as opposed to Everex (4 out of 64, 6.3%). The analyses for ad awareness, X2(1) = 47.29, p < .01, indicates that more partici- pants had seen advertising about Compaq (35 out of 65 partici-
pants, 53.8%) as opposed to the Everex brand (0 out of 64, 0.0%). Participants also had more brand associations (i.e., total thoughts about the brand) for Compaq (M= 2.49) as compared to the Everex brand (M= 1.55), t(127) = 2.89,p < .01. More- over, participants also had more-positive brand associations for Compaq (M= 0.92) as compared to the Everex brand (M= 0.14), t(127)= 4.32,p < .01. Finally, participants also indicated that they are more willing to purchase a Compaq (M= 2.71) as compared to an Everex (M= 1.92), F(, 127) = 14.87,p < .001. In sum, these analyses provide additional support that consum-
ers perceive Compaq to have a significantly more-positive brand name than Everex.
Tests of Hypotheses
Analysis procedures. The mediation hypothesis ofat- tributions on the negative WOMC-brand evaluation model (H 1) was assessed using hierarchical multiple regression pro- cedures as suggested by Baron and Kenny (1986). The overall effects of the negative-WOMC scenarios and brand-name strength on the two attribution measures were assessed with a
MANOVA and follow-up ANOVA procedures.4
4To further understand the nature of processing used by the participants,
cognitive-response data were also gathered. The cognitive responses were in-
dependently coded by three judges as (a) brand attributions, (b) person attribu-
tions, (c) situation attributions, or (d) nonattributional thoughts. A majority de-
cision was used to assign thoughts to final categories. Disagreements by all threejudges were resolved by the authors. Interjudge reliability was calculated
using the proportional-reduction-in-loss approach as suggested by Rust and Cooil (1994). The proportional-reduction-in-loss reliability was .82, which is
well above Rust and Cooil's recommended level of .70. The magnitude and complexity of these data preclude their detailed discussion in this article. How-
ever, they suggest that consumers actively engaged in attributional processing
of the negative WOMC. Participants generated a total of 511 classifiable cog- nitive responses, of which 169 (33%) were attributional responses. Further- more, the pattern of attributional responses within each of the cells was gener-
ally consistent with the scaled results reported here.
Mediation analyses. Because it is possible that prod- uct involvement and knowledge could alter the proposed rela- tions in the negative WOMC-causal attribution-brand evalu- ation model, we controlled for their effects. Involvement and
knowledge were included as covariates (in which the effects are accounted for before examining the effects of other inde-
pendent variables) in the hierarchical regression (Cohen & Cohen, 1983) procedures. To determine if the attribution variables mediated the relation between negative WOMC and brand attitude, Baron and Kenny (1986) suggested that sev- eral conditions need to be met. First, it would need to be deter-
mined that negative WOMC significantly influences the attri-
bution variables and postexposure brand evaluation. Second, the attribution variables would need to significantly affect postexposure brand evaluation. Finally, results would need to indicate that a statistically significant change in R2 would oc- cur when the attribution variables were added to the regres-
sion equation in which negative WOMC served as the inde- pendent variable and brand evaluation as the dependent variable. The mediation results for the study presented here are reported in Table 2.
1. The initial step in the mediation analysis is to determine
if negative WOMC significantly influences the attribution and
postexposure brand evaluation variables after controlling for the effects of several covariates. Given that there are two de-
pendent (attribution) variables, a multivariate coanalysis of variance (MANCOVA) is used to test the negative WOMC-at- tribution relation. MANCOVA results indicate that negative WOMC is significantly related to the attribution variables
(Wilks' X = .91), approximate F(6, 342) = 2.74,p < .05. To de- termine if the negative WOMC-postexposure brand evalua- tion linkage is significant, the three experimental treatment levels of negative WOMC were coded as two dummy-coded variables with low consensus, high distinctiveness, and low consistency as the comparison treatment (coded as zero and zero for the two dummy variables). Multiple hierarchical re- gression results suggest that the set of negative-WOMC vari- ables significantly influence postexposure brand attitudes (in- cremental R2 = .07), F(2, 175) = 7.68, p < .01.
2. The second step in the mediation analysis is to deter- mine if the attribution variables significantly affect postexposure brand evaluation. Regression results indicate that the set of attribution variables significantly influences postexposure brand evaluation after controlling for the covariates (incremental R2 =. 19), F(2, 173) = 21.84,p < .001.
3. In the final step of the mediation analysis, results indi-
cate a statistically significant change in R2 by adding the attri- bution variables to the regression equation in which the rela- tion WOMC dummy variables served as independent variables and brand evaluation as the dependent variable (in- cremental R2 = 0.15), F(2, 171) = 18.12, p < .001. Results show that brand attributions are negatively related to postexposure brand evaluation (P = -.35), t(175) = -4.96,p <
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
TABLE 1
Analysis of Variance of the Consensus, Distinctiveness, and Consistency Manipulation
Information Configuration
Dependent Variable LHL LLH HHH p Value df
Consensus Low level Low level High level M 6.50a 6.20a 1.41b 139.53* 2, 65 SD 0.74 1.41 1.14
N 22 24 22
Distinctiveness High level Low level High level M 6.04a 1.77b 5.80a 182.38* 2, 65 SD 0.82 0.81 0.94
N 22 24 22
Consistency Low level High level High level M 6.29a 1.43b 1.68b 176.40* 2, 65 SD 1.11 0.78 1.00
N 22 24 22
Note. LHL = low levels ofconsensus inforation, high levels ofdistinctiveness information, and low levels of consistency information. LLH = low levels of
consensus information, low levels of distinctiveness information, and high levels of consistency information. HHH = high levels of consensus information, high
levels of distinctiveness information, and high levels of consistency information. Superscript letters indicate significant means. Different letters (i.e., a vs. b) are
significantly different (p < .05 or better). Means with the same letter are not significantly different (p > .05). *p<.01.
TABLE 2
Hierarchical Regression Coefficients (t Values in Parentheses)
Dependent Variable: Brand Evaluation
Covariates, Negative Covariates, Attribution Variables Covariates Only Model WOMC Model Model Full Model
Covariates
Product knowledge .13* .13* .09 .10 (1.67) (1.75) (1.35) (1.40)
Aesthetic involvement .18** .15* .04 .04
(2.08) (1.76) (0.55) (.49) Utilitarian involvement -.05 -.03 -.03 -.01
(-.60) (-.39) (-.34) (-.18) Independent variables Negative WOMC (LLH) .01 -.04
(.03) -(-.53) Negative WOMC (HHH) -.28**** -.22***
— (-3.32) (-2.84) Brand attributions – -39**** -.35****
(-5.47) (-4.96) Communicator attributions -.16** .15**
-~- -~- (2.27) (2.21)
Adjusted R2 .03 .10 .22 .25 F 3.04 5.04 10.99 9.48
p < .05 .001 .001 .001
Incremental R2 .07 .19 .15
F change 7.68 21.84 18.12 p < .001 .001 .001
Note. WOMC = word-of-mouth communication. LLH = low levels of consensus information, low levels ofdistinctiveness information, and high levels ofcon-
sistency information. HHH = high levels of consensus information, high levels of distinctiveness information, and high levels of consistency information. *p < .10. **p < .05. ***p < .01. ****p < .001.
65
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
66 LACZNIAK, DECARLO, RAMASWAMI
.001, whereas communicator attributions are positively re-
lated to brand evaluations (P =. 15), t( 175) = 2.21 ,p < .05. The high-consensus, high-distinctiveness, and high-consistency treatment remains significantly related to brand evaluation in
the full model but at a weakened level (P = -0.22 vs. P = -0.28 in the original model).
These results appear to suggest that both brand and communi-
cator attributions meet the criteria for mediating the relation
between negative WOMC and postexposure brand evalua- tions.5 Thus, results support HI. The previous mediation re- sults add credence to Burnkrant's (1982) and Richins' (1984) contention that receivers employ attributional processing when encountering negative WOMC.
MANOVA results. MANOVA results indicate that
the WOMC scenarios (Wilks' X = .94), approximate F(4, 336) = 2.52,p < .05, and brand name (Wilks' X = .88), F(2, 168) = 10.59,p < .001, have a significant effect on attribu- tions. Results indicate a significant Brand Name x Nega- tive-WOMC interaction. (ANOVA results indicate that this interaction is significant only for communicator attri- butions.) Given that the interaction is not theoretically rel- evant (i.e., no interaction hypotheses were proposed), we report it only to be complete, and, thus, the interaction will not be discussed.
ANOVA results. H2 posits that participants exposed to high consensus, distinctiveness, and consistency WOMC will generate stronger brand attributions than those exposed to the other WOMC scenarios. The omnibus Ftest indicates a
main effect for negative WOMC on brand attributions, F(2, 178) = 4.77, p < .01. The a priori planned comparison test, t(178) = -2.85, p < .01, indicates that participants are more likely to attribute the negative information to the brand when
exposed to the high-consensus, distinctiveness, and consis- tency scenario (M= 4.51) as opposed to either the low-con- sensus, low-distinctiveness, and high-consistency scenario (M= 3.88) or low-consensus, high-distinctiveness, low-con- sistency scenario (M = 4.10). Thus, H2 is supported.
H3 posits that participants exposed to low-consensus, low-distinctiveness, and high-consistency negative WOMC
5Based on a comment by one of the reviewers, an additional mediation analysis was conducted using a summary attribution measure (i.e., a separate single item that was collected prior to the multiple item measures) for brand
and communicator attributions. The results are similar to those using the summated attribution measures. Hierarchical regression estimates, with in- volvement and knowledge as covariates, indicate that the added effects of the
single-item attribution measures on postexposure brand evaluations explain significantly more variance to the WOMC, brand-name, covariate model (in- cremental R2 =. 10), change inF= 11.99,p < .001. Results show that brand at-
tribution is negatively related to postexposure brand evaluations (p = -.28), t(175) = 3.96,p < .01, whereas communicator attribution is positively related to brand evaluations (B = .16), t(175) = 2.45,p < .05.
will generate stronger communicator attributions than those re-
ceiving high-consensus, distinctiveness, and consistency and low-consensus, high-distinctiveness, low-consistency negative WOMC. Omnibus F test results indicate a significant effect of
the negative-WOMC scenarios on communicator attributions, F(2, 178) = 2.38, p < .10. Planned comparison results suggest that participants are more likely to generate stronger communi-
cator attributions (M = 4.51) after receiving low-consensus, low-distinctiveness, and high-consistency negative WOMC as compared to high-consensus, distinctiveness, and consistency negative WOMC (M = 4.06) or low-consensus, high-distinc- tiveness, low-consistency negative WOMC (M= 4.07), t(178) = -2.18,p < .05. Thus, H3 is supported.
In combination, results of the test of Hypotheses 2 and 3 sug-
gest that the manner with which a communicator structures his
or her negative-WOMC message influences receivers' attributional responses. It appears that receivers attribute the
cause for a negative-WOMC message to the brand when the in-
formation is unambiguous and sufficiently strong (e.g., those
that are configured as high consensus, high distinctiveness, and
high consistency). On the other hand, receivers appear to attrib-
ute the negativity to the communicator when the information
conveyed is more ambiguous, less well developed, or both (e.g.,
those that are configured as low consensus, low distinctiveness,
and high consistency). Such results empirically support Richins'
(1984) suggestions that the structure of WOMC messages influ-
ence the manner in which receivers process such information.
Results also emphasize the need for researchers to conceptualize
and operationalize negative WOMC as something more than a summary negative statement about a brand.
H4 posits that stronger brand attributions are expected for
less-favorable, as compared to more-favorable, brands for re-
ceivers of all configurations of negative WOMC. ANOVA re- sults indicate a significant effect of brand name on brand attributions, F(l, 179) = 23.28,p < .001. As hypothesized (and in support of H4), the means indicate that participants are more
likely to make brand attributions for less-favorable brands (M
= 4.58) as opposed to more-favorable brands (M= 3.79). H5 posits that receivers will generate stronger communi-
cator attributions for more-favorable, as compared to less-fa-
vorable, brands for all configurations of negative WOMC. ANOVA results suggest a significant effect of brand name on communicator attributions, F(1, 179) = 7.71, p < .01. Mean comparisons indicate that participants are more likely to at- tribute the negativity of the WOMC to the communicator for
more-favorable brands (M = 4.48) as opposed to less-favor- able brands (M= 3.94). Thus, H5 is supported.
Results of the tests of Hypotheses 4 and 5 indicate that re- ceivers of negative WOMC are more likely to generate brand attributions for brands with less-favorable names, whereas receivers of negative WOMC featuring more-favorable names tend to attribute the negativity toward the communica- tor. Thus, more-positive brand names appear to be protected from the effects of negative WOMC, especially when weaker or ambiguous negative WOMC is communicated.
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
CONSUMERS' RESPONSES 67
H6 and H7 deal with the effects of attributions on brand
evaluations. Specifically, H6 posits that brand attributions generated in response to negative WOMC will be inversely related to postexposure brand evaluation, whereas H7 sug- gests a direct relation between communicator attributions and
brand evaluations. As indicated in Table 2, regression results show that brand attributions are negatively related to brand evaluation (p =-.32), t( 175) = -4.37,p < .001, supporting H6. Conversely, communicator attributions appear to be directly related to postexposure brand evaluations (P = .13), t(175) = 1.96, p < .05. Thus, H7 is supported.
Results for H6 indicate that negative WOMC will have the expected negative effect on receivers' postexposure brand evaluations when the negativity is associated with the brand. However, results for H7 suggest that when receivers associate the negativity of a WOMC message with the communicator, they will increase their evaluations of the focal brand. Such a
finding suggests that communicator attributions may be one mechanism that allows receivers of negative WOMC to dis- associate the message from the brand.
STUDY 2
H2 and H3, based largely on the covariation principle in attri- bution theory, posit specific relations between negative WOMC and receiver attributions. One inherent assumption with these hypotheses is that particular combinations of the information dimensions-consensus, distinctiveness, and consistency-would lead consumers to generate stronger brand and communicator attributions. Although these hy- potheses (and subsequent findings) provide interesting in- sights into receivers' responses to different configurations of negative WOMC, they do not generate conclusive evidence about why specific attributional responses emerged. For ex- ample, the finding that negative WOMC configured as high consensus, high distinctiveness, and high consistency would be more likely to lead consumers to generate stronger brand attributions than the other tested configurations, may have occurred because (a) high-consensus information was suffi- cient to elicit brand attributions (recall that the high consen-
sus, high distinctiveness, and high consistency configuration was the only one tested that included high-consensus infor- mation), or (b) high levels of all three information dimensions
led participants to generate a predominance of brand attribu- tions. Theoretically, both explanations are possible. Whereas Kelley's (1967) classic covariation model suggested that high levels of all three information dimensions are needed for par-
ticipants to generate strong brand attributions, the findings of Folkes and Kotsos (1986) indicated that high levels of con- sensus information may be sufficient to drive them. Study 2 was designed, in part, to determine which of these two expla- nations is more plausible.
In addition, results from Study 1 also indicate that receivers of low-consensus, low-distinctiveness, and high-consistency in-
formation generate stronger communicator attributions as com-
pared to the other two information configurations tested. Given that the low-consensus, low-distinctiveness, and high-consis-
tency WOMC configuration was the only one tested in Study 1 that contained low-distinctiveness information, it is possible that
low levels of this dimension are driving receivers of nega- tive-WOMC information to generate stronger communicator at-
tributions. Thus, Study 2 was designed to also investigate whether low-distinctiveness information was sufficient to drive
participants' generation of communicator attributions.
Method
Negative-WOMC Scenarios
Four negative-WOMC scenarios were tested in Study 2. They include (a) high consensus, high distinctiveness, and high consistency; (b) low consensus, high distinctiveness, and high consistency; (c) high consensus, low distinctive- ness, and high consistency; and (d) low consensus, low dis- tinctiveness, and high consistency. As illustrated in Tables 3 and 4, these configurations were systematically selected to test the notions discussed previously. If high-consensus in- formation is an important driver of brand attributions in nega-
tive-WOMC communication, we would expect a significant difference between the means on brand attributions for the
first contrast (i.e., high consensus, high distinctiveness, and high consistency vs. low consensus, high distinctiveness, and high consistency), given that consistency and distinctiveness information dimensions are held constant, whereas only con-
TABLE 3
Effect of Consensus Information on Brand Attributions
Brand
Information Dimensions Attribution
t P
Consensus Distinctiveness Consistency M SD Value Value
High High High 4.34 1.26 3.31 .001 Low High High 3.79 1.31 High Low High 3.73 1.00 -.29 .766 Low Low High 3.72 1.00
TABLE 4
Effect of Distinctiveness Information on Communicator Attributions
Communicator
Information Dimensions Attribution
t P Consensus Distinctiveness Consistency M SD Value Value
High High High 4.19 1.44 -3.33 .001 High Low High 5.08 1.46 Low Low High 4.67 1.49 2.47 .014 Low High High 4.02 1.58
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
68 LACZNIAK, DECARLO, RAMASWAMI
sensus information is varied. Similarly, we would expect re- ceivers to generate stronger brand attributions after receiving
high consensus, low distinctiveness, and high consistency as compared to the low-consensus, low-distinctiveness, and high-consistency information scenario in the second contrast.
Two a priori contrasts were also used to test whether distinc-
tiveness information was driving the generation of communicator
attributions as suggested by the results in Study 1. If low-distinc-
tiveness negative WOMC is driving communicator attributions,
we would expect a more-positive direct relation with communica-
tor attributions for the high-consensus, low-distinctiveness, and
high-consistency information configuration as compared to the
high-consensus, high-distinctiveness, and high-consistency con-
figuration. Similarly, consumers receiving negative WOMC con-
figured as low consensus, low distinctiveness, and high consistency would likely generate stronger communicator attri-
butions as compared to those receiving low-consensus, high-dis-
tinctiveness, and high-consistency negative WOMC.
Participants and Procedure
A total of 259 undergraduate students were recruited from two
introductory marketing classes from a large midwestem univer-
sity. Incomplete responses were obtained from 3 participants, so
the net sample was 256. The sample distribution across the four
cells ranged from a low of 56 (21%) to a high of71 (27%). Partici-
pation in the study fulfilled a course requirement in both classes.
None of the Study 2 participants took part in Study 1.
Participants were randomly assigned to one of the four ex-
perimental cells. The procedure, instructions, and scenarios (other than the systematic variation of the three information di-
mensions) were identical to those in Study 1. Brand-name strength was held constant in all four experimental cells. Given
our interest in the effect of the WOMC configurations on attri-
butions, Study 2 focuses on the Compaq brand name only.
Results and Discussion
Further manipulation checks on the WOMC scenarios were not completed because the manipulation check results in Study 1 indicated that the wording of the information config- urations is viable. MANCOVA results (with involvement and computer knowledge as covariates) indicate that the WOMC scenarios (Wilks' X = .88), approximate F(6, 494) = 5.1 1,p < .001, have a significant effect on brand and communicator at-
tributions. Follow-up tests including a priori planned com- parisons were used to test if (a) high consensus is sufficient to elicit brand attributions (see Table 3) and (b) low distinctive- ness is the primary information cue receivers use to generate communicator attributions (see Table 4). The omnibus F test for brand attributions indicates a significant effect for nega- tive WOMC, F(3, 252) = 4.73, p < .05. Moreover, the a priori contrast between the high-consensus, high-distinctiveness, and high-consistency configuration (M = 4.34) versus the low-consensus, high-distinctiveness, high-consistency con-
figuration (M = 3.79) is significant for brand attributions,
t(252) = 3.31, p > .01. The contrast between high-consensus, low-distinctiveness, and high-consistency configuration (M = 3.73) versus low-consensus, low-distinctiveness, and high-consistency configuration (M= 3.72) is not significant, t(252) = -.30, p > .10, for brand attributions. The results of these two contrasts suggest, although not conclusively, that receivers covary consensus information with distinctiveness and consistency in generating brand attributions.
To test this assertion more fully, a third contrast was con-
ducted. This contrast compared the high-consensus, high-dis-
tinctiveness, and high-consistency information (M = 4.34) with the high-consensus, low-distinctiveness, and high-con- sistency negative WOMC (M= 3.73) on brand attributions. It was indicated by t test results that the means are significantly
different and in the direction supporting the covariation hy- pothesis, t(252) = 2.59,p > .05. The pattern of results for these
three tests indicates that participants who received negative WOMC, configured as high in consensus, distinctiveness, and consistency, generated stronger brand attributions than all of the other tested scenarios. Such findings suggest that re-
ceivers covary all three information dimensions when gener- ating brand attributions toward the focal brand. Thus, in the
presence of varying levels of distinctiveness and consistency information, high-consensus information is not to be suffi- cient for receivers to attribute the negativity ofa WOMC mes-
sage toward the brand.
Results of the omnibus F test, comparing the information configurations on the dependent variable of communicator attributions, indicate a significant effect for negative WOMC,
F(5, 380) = 4.11 ,p < .01. The a priori contrast of the high-con-
sensus, high-distinctiveness, and high-consistency scenario (M = 4.19) versus the high-consensus, low-distinctiveness, and high-consistency configuration (M= 5.08) is significant for the communicator-attribution dependent variable, t(380) = -3.52, p > .001. In addition, the second contrast that com- pares the low-consensus, low-distinctiveness, and high-con- sistency configuration (M = 4.67) with the low-consensus, high-distinctiveness, and high-consistency scenario (M= 4.02) is also significant for communicator attributions, t(380) = 2.47, p > .05. Collectively, these results suggest that low levels of distinctiveness information may be an important factor in leading receivers of negative WOMC to attribute the negativity of a message toward the communicator.
GENERAL DISCUSSION, LIMITATIONS, AND IMPLICATIONS
General Discussion
The influence of negative WOMC on consumers is seemingly significant. Yet, empirical research dealing with this phenome-
non is surprisingly limited. In particular, there appears to be a dearth of research that deals with the manner in which consum-
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
CONSUMERS' RESPONSES 69
ers cognitively process negative WOMC. This study draws on prior conceptual evidence to develop and test hypotheses relat-
ing to the attributional responses used by consumers when en- countering negative WOMC. In general, the results support the
hypotheses and provide several key contributions. First, this research found that causal attributions mediate the
negative-WOMC-brand evaluation relation. Thus, not only do consumers generate causal attributions in response to negative
WOMC, but our results suggest that they are used to influence subsequent brand evaluations. Second, this research, using a broader and-according to Richins (1984)-richer conceptu- alization of the negative WOMC, found that different combi- nations of consensus, distinctiveness, and consistency information lead receivers to have differential responses. Third, this research also found that the strength of the focal
brand's name influenced consumers' responses to negative WOMC. The use of existing brand names adds realism to re- searchers' understanding of the WOMC phenomenon, which, up to this point, had been based on empirical investigations that
used fictitious brands (e.g., Bone, 1995; Herr et al., 1991). As noted previously, brand attributions mediated the rela-
tion between negative WOMC and brand evaluations for cer- tain information configurations. Specifically, the relation was fully mediated for negative WOMC that was configured as low consensus, low distinctiveness, and high consistency; and low consensus, high distinctiveness, and low consis- tency. However, the direct effects of high-consensus, high-distinctiveness, and high-consistency WOMC on brand evaluation are strongly, but not completely, attenuated when attributions are considered. Thus, it appears that well orga- nized and compelling negative WOMC can have a direct ef- fect on brand evaluations.
Although previous research emphasized that negative WOMC will have a negative impact on receivers' brand eval- uations, our study points out the nongeneralizability of this finding. For example, our results suggest that information configured as low-consensus, low-distinctiveness, and high-consistency negative WOMC may actually increase re- ceivers' brand evaluations. This finding suggests that com- municator attributions may be one mechanism that allows receivers of negative WOMC to disassociate the message from the brand. Specifically, receivers will appear to use attributional processes to deflect the negativity away from the
brand for this type of WOMC and “rally” to its defense and, as a result, increase evaluations. Consistent with theoretical models of conversational processing that deal with cognitive incongruity (cf. Hilton, 1995), receivers appear to search for a balance between themselves, the communicator, and the brand on receiving negative information from others.
These results may be also used to augment past findings. For example, whereas Herr et al. (1991) findings suggested that re- ceivers may not always find negative WOMC to be diagnostic (i.e., useful), results of this study provide a further explanation of
why this is the case. Specifically, our findings suggest that when negative WOMC is attributed toward the communicator (and
not the brand), receivers will not decrease their evaluations of
the brand (i.e., they will not use the negative information). Thus,
the nature of the causal attributions, generated in response to
negative WOMC, may be used as indicators of the extent to which consumers find this information to be diagnostic.
Another interesting finding ofthis study is that brand name
does have a direct effect on the attributions generated by re-
ceivers of negative WOMC. Although past research (cf. Herr et al., 1991) suggested that negative WOMC will not have a significant effect on brand evaluations when receivers have strong prior beliefs about the focal brand, it is our belief that brand name (and its association) influences brand evaluations through the cognitive mechanism of attributional processing.
Finally, the results from Study 2 yield some additional in-
sights into receivers' attributional responses to different con-
figurations of negative WOMC. First, the results indicate that
negative-WOMC messages containing only high-consensus information (but varying levels of distinctiveness, consensus information, or both) may not lead receivers to generate sig- nificantly strong brand attributions. Thus, receivers of nega-
tive-WOMC messages that include distinctiveness and consistency information dimensions are not likely to focus on
consensus information alone as a means to generate attribu- tions toward the brand. Second, the results also suggest that when high levels of distinctiveness and consistency are paired with high-consensus information, receivers will gen- erate stronger brand attributions as compared to the other in- formation configurations tested in this study. In combination,
these results provide additional support for the high-consen- sus, distinctiveness, and consistency information-brand attri-
bution relation obtained in Study 1. Negative-WOMC messages containing low-distinctiveness
information, on the other hand, may be sufficient to enable re-
ceivers to generate attributions toward the communicator. Al-
though future research is needed to examine the generalizability
of this finding in other contexts, it appears that low-distinctive-
ness information provides enough relevant evaluative informa- tion about the communicator for the receiver to blame him or her
for the negativity. Although not explicitly tested in this study, it
is possible that receivers are using the simplest attributional pro-
cess available; that is, in the absence of contradictory informa-
tion about the communicator's disposition, receivers apparently
discount or ignore other information about the brand (i.e., con-
sensus and consistency) and blame the communicator for the negativity of the message when it is not particularly distinctive.
It is interesting to note that a direct linkage between low distinc- tiveness and communicator attributions is consistent with find-
ings in other settings (e.g., Alicke & Insko, 1984; Bassili & Regan, 1977; Hansen, 1980).
Limitations
Not unlike other empirical research efforts, the results presented
in this study are limited by a number of factors-many of which can be addressed in future studies. First, this study dealt with neg-
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
70 LACZNIAK, DECARLO, RAMASWAMI
ative WOMC regarding a single product (personal computers). Although personal computers were purposefully chosen (as they represent a product class with which consumers would be ex-
pected to share WOMC), results may not be generalizable to other
product categories. Second, the negative-WOMC scenarios were
artificial. They were chosen to represent theoretically based infor-
mation configurations. Future research should attempt to investi-
gate actual negative WOMC to determine if it is configured in a
manner similar to that portrayed in our study. It is important to
note that previous studies investigating negative WOMC em- ployed a unidimensional view of the concept. The attempt made
in this study to examine negative-WOMC message structure could act as a catalyst for future research in this domain. Future re-
search could use different conceptualizations of message struc-
ture, such as argument type (e.g., benefit vs. attribute), order of
presentation, and message complexity. Third, the WOMC treat-
ments were presented to participants via a tape-recorded message
with explicit instructions as to how to listen. Such a method limits
participants' opportunities to ask questions and seek clarifications
of message points (Smith & Vogt, 1995), and it eliminates some
of the interpersonal flavor of WOMC. Thus, future research may
attempt to determine if participants' attributional responses to in- teractive WOMC situations are similar to those observed in this
study. Moreover, participants were not afforded the opportunity
to hear the negative WOMC on multiple occasions. Future re- search may be used to detennine if receiver responses may have
differed had they listened to multiple accounts of the information.
Fourth, in this study, brand-name strength was manipulated si-
multaneously on the two dimensions of affect and familiarity. Al-
though the focus here was to determine the overall effect of brand-name strength on consumers' responses to negative WOMC, future research could isolate the effects of each of these
brand-name strength dimensions. Finally, classic attribution the-
ory posits that certain information configurations would lead to
interactive attributional responses of receivers (e.g., Person x Brand attributions). We did not deal with these interactive attribu-
tions because theory does not suggest how they would affect our
ultimate dependent variable (i.e., brand evaluations); yet, future
researchers are advised to consider these types of responses to negative WOMC.
Implications
Findings of this study hold a number of implications for man-
agers. First, managers should be aware that consumers exposed to negative WOMC actively process the information and change their brand evaluations only under certain conditions. For example, it appears that negative WOMC configured in a strong and compelling manner negatively affects brand evalu-
ations. On the other hand, negative WOMC that is less compel- ling could even have a positive effect on consumers' evalua- tions of brands (depending on the attributions of the receiver). Thus, it appears that managers do not always need to take cor- rective action when negative WOMC is generated about their offerings. Corrective action, however, may be needed when
the negative-WOMC message states that others are in consen-
sus with the communicator's view, the focal brand has a unique problem, and it consistently performs poorly.
Second, study results also suggest that consumers have the ability to deflect the negativity of a WOMC message away from a brand. This may be the case as consumers rec- ognize that negative WOMC may be generated for reasons other than dissatisfaction with a brand (e.g., to obtain feel- ings of power or prestige, to reduce their own postpurchase dissonance, etc.). Managers therefore are encouraged to monitor WOMC, as it may not always prove to be harmful to their brand. Several mechanisms are available to monitor
negative WOMC. For example, Bolen (1994) encouraged marketers to listen actively to customer conversations and question them to determine underlying sources of dissatis- faction. We also believe that it would be useful for managers
to monitor WOMC by setting up focus groups with opinion leaders. Opinion leaders often initiate negative WOMC (Brooks, 1957) or are, at a minimum, aware of consumer complaints with regard to a particular brand. Marketers should question opinion leaders about the type of negative information that they received from others and details of the
situation surrounding the communication. Finally, we en- courage managers to monitor negative WOMC that occurs via Internet chat rooms. Hopefully, this and other informa- tion will aid managers in determining the degree to which WOMC is configured in a manner similar to high consensus, high distinctiveness, and high consistency.
Third, results suggest that managers may offset the poten-
tial effects of negative WOMC by gaining high levels of strength or equity for their brands. Such a view is consistent with theoretical notions (Haugtvedt, Leavitt, & Schneier, 1993; Keller, 1993) contending that high levels of equity al- low consumer perceptions about a brand to be resistant to ex-
ternal forces such as negative WOMC. Previous research suggested that higher levels of equity may be gained by en- hancing consumer's familiarity with a brand and reinforcing positive affect for that brand (Farquhar, 1989). Although it is
obvious that advertising may help managers increase the fa- miliarity of their brands, a well-chosen ad message could build and enhance consumers' affect toward the brand as well
(Haugtvedt et al., 1993). In addition, satisfactory brand per- formance should reinforce the positive associations that cus- tomers hold in their memories.
Finally, our findings illustrate that when brand attributions are made in response to strongly configured negative WOMC, evaluations will be reduced for all brands. It is im- portant to realize that strong and compelling negative WOMC will lead consumers to generate brand attributions, which, in turn, penalizes brand evaluations for high- and low-strength brands. Given the complex set of interrelations among brand-name strength, negative-WOMC configura- tions, and attributional processing, marketers may benefit from a systematic understanding of how (potential) custom- ers process all marketplace information about their brands.
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
CONSUMERS' RESPONSES 71
In conclusion, this article represents an initial attempt to
model a process consumers employ when receiving multidi- mensional negative-WOMC messages. The results indicate that such messages not only affect brand evaluations directly,
but also have indirect influences through causal attributions made by consumers. In addition, the results highlight the use-
fulness of brand factors in influencing the effects of negative
WOMC. Although the study yielded several interesting re- sults, a more systematic research effort regarding consumers'
processing of negative WOMC is needed before researchers can confidently make definitive conclusions about this com- mon marketplace phenomenon.
ACKNOWLEDGMENTS
We gratefully acknowledge the helpful comments of Les Carlson, DeAnna Kempf, James McElroy, R. Kenneth Teas, Paul Herr, and three anonymous reviewers.
REFERENCES
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes andpredicting so- cial behavior. Englewood Cliffs, NJ: Prentice-Hall.
Alicke, M. D., & Insko, C. A. (1984). Sampling of similar and dissimilar comparison objects as a function of the generality of attribution goals. Journal of Personality and Social Psychology, 46, 763-777.
Arndt, J. (1967). Role of product-related conversations in the diffusion of a
new product. Journal of Marketing Research, 4, 291-295. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable dis-
tinction in social psychological research: Conceptual, strategic, and sta-
tistical considerations. Journal of Personality and Social Psychology, 31, 1173-1182.
Bassili, J. N., & Regan, D. T. (1977). Attribution focus as a determinant of in-
formation selection. Journal of Social Psychology, 101, 113-121. Bolen, W. H. (1994). The role of word-of-mouth 'advertising' in retailing.
American Business Review, 5, 11-14.
Bone, P. F. (1995). Word-of-mouth effects on short-term and long-term product judgments. Journal of Business Research, 32, 213-223.
Brooks, R. C., Jr. (1957, October). 'Word-of-mouth' advertising in selling new products. Journal of Marketing, 22, 31-39.
Brown, R., & van Kleeck, R. (1989). Enough said: Three principles of expla- nation. Journal of Personality and Social Psychology, 57, 590-604.
Burnkrant, R. E. (1982). Influence of other people on consumer attributions
toward brand and person. Journal ofBusiness Research, 10, 319-3 3 8.
Calder, B. J., & Burkrant, R. E. (1977). Interpersonal influence on con- sumer behavior: An attribution theory approach. Journal of Consumer Research, 4, 29-38.
Calder, B. J., Phillips, L. W., & Tybout, A. M. (1982). The concept of exter- nal validity. Journal of Consumer Research, 9, 240-244.
Churchill, G. A. (1979). A paradigm for developing better measures of mar-
keting constructs. Journal of Marketing Research, 16, 64-73. Cobb-Walgren, C. J., Ruble, C. A., & Donthu, N. (1995). Brand equity, brand
preference and purchase intent. Journal ofAdvertising, 24, 41-54. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysisfor
the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
DeCarlo, T. E., & Leigh, T. W. (1996, April). Impact of salespersons' attrac- tion on sales managers' attributions and feedback. Journal of Mar- keting, 60, 47-66.
DeNisi, A. S., Cafferty, T. P., & Meglino, B. M. (1984). A cognitive view of the performance appraisal process: A model and research propositions. Organizational Behavior and Human Performance, 33, 360-396.
Engel, J. F., Kegerreis, R. J., & Blackwell, R. D. (1969, July). Word-of-mouth
communication by the innovator. Journal of Marketing, 33, 15-19.
Enis, B. M., & Roering, K. J. (1980). Product classification taxonomies: Syn-
thesis and consumer implications, theoretical development in market-
ing. In C. Lamb & P. Dunne (Eds.), Theoretical development in marketing (pp. 186-189). Chicago: American Marketing Association.
Farquhar, P. H. (1989). Managing brand equity. Marketing Research, 1, 24-33.
Feldman, J. M., & Lynch, J. G., Jr. (1988). Self generated validity and other effects of measurement on belief, attitude, intention, and behavior.
Journal of Applied Psychology, 73, 421-425. Folkes, V. S. (1988). Recent attribution research in consumer behavior: A re-
view and new directions. Journal of ConsumerResearch, 14, 548-565. Folkes, V. S., & Kotsos, B. (1986, April). Buyers' and sellers' explanations
for product failure: Who done it? Journal of Marketing, 50, 74-80. Hansen, R. D. (1980). Common sense attribution. Journal ofPersonality and
Social Psychology, 39, 996-1009. Harvey, J. H., & Weary, G. (1984). Current issues in attribution theory and
research. Annual Review of Psychology, 35, 427-459. Haugtvedt, C. P., Leavitt, C., & Schneier, W. L. (1993). Cognitive strength of
established brands: Memory, attitudinal and structural approaches. In D. A. Aaker & A. L. Biel (Eds.), Brand equity and advertising: Adver-
tising's role in building strong brands (pp. 247-262). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Haywood, K. M. (1989). Managing word of mouth communications. Journal of Services Marketing, 3, 55-65.
Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.
Herr, P. M., Kardes, F. R., & Kim, J. (1991). Effects of word-of-mouth and
product attribute information on persuasion: An accessibil- ity-diagnosticity perspective. Journal of Consumer Research, 17, 454-462.
Hilton, D. J. (1995). The social context of reasoning: Conversational infer- ence and rational judgement. Psychological Bulletin, 118, 248-271.
Hilton, D. J., & Jaspars, J. M. (1987). The explanation of occurrences and non-occurrences: A test of the inductive logic model of causal attribu-
tions. British Journal of Social Psychology, 26, 189-201.
Hilton, D. J., & Slugoski, B. R. (1986). Knowledge-based causal attribution: The abnormal conditions focus model. Psychological Review, 93, 75-88.
Hilton, D. J., Smith, R. H., & Alicke, M. D. (1988). Knowledge-based infor-
mation acquisition: Norms and the functions of consensus information. Journal of Personality and Social Psychology, 55, 530-540.
Hoch, S. J., & Deighton, J. (1989, April). Managing what consumers learn from experience. Journal of Marketing, 53, 1-20.
Iacobucci, D., & McGill, A. L. (1990). Analysis of attribution data: Theory testing and effects estimation. Journal of Personality and Social Psy- chology, 59, 426-441.
Jones, E. E., & Davis, K. (1965). From acts to dispositions: The attribution
process. In L. Berkowitz (Ed.), Advances in experimental social psy- chology (Vol. 2, pp. 219-266). New York: Academic.
Katz, E., & Lazerfield, P. (1955). Personal influence. Glencoe, IL: Free Press.
Keller, K. (1993, January). Conceptualizing, measuring, and managing cus- tomer-based brand equity. Journal of Marketing, 57, 1-22.
Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska Symposium on Motivation (pp. 192-238). Lincoln: University of Nebraska Press.
Kelley, H. H. (1972). Causal schema and the attribution process. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the causes of behavior (pp. 1151-1174). Morristown, NJ: General Learning Press.
Kelley, H. H. (1973). The process of causal attribution. American Psycholo- gist, 28(2), 107-128.
Lichtenstein, D. R., & Bearden, W. 0. (1986). Measurement and structure of
Kelley's covariance theory. Journal of Consumer Research, 13, 290-296.
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
72 LACZNIAK, DECARLO, RAMASWAMI
Mizerski, R. W. (1982). An attribution explanation of the disproportionate influence of unfavorable information. Journal of Consumer Research, 9, 301-310.
Mizerski, R. W., Golden, L. L., & Keran, J. B. (1979). The attribution pro- cess in consumer decision making. Journal of Consumer Research, 6, 123-137.
Morin, S. P. (1983, February 28). Influentials advising their friends to sell
lots of high-tech gadgetry. Wall Street Journal, p. 30.
Pope, K. (1993, October 15). Computers: They're no commodity. Wall Street Journal, p. B1.
Regan, D. T., Straus, E., & Fazio, R. F. (1974). Liking and the attribution pro- cess. Journal of Experimental Social Psychology, 10, 385-397.
Richins, M. L. (1983, Winter). Negative word-of-mouth by dissatisfied con- sumers: A pilot study. Journal of Marketing, 47, 68-78.
Richins, M. L. (1984). Word of mouth communication as negative informa-
tion. In T. Kinnear (Ed.), Advances in consumer research XI (pp. 697-702). Provo, UT: Association for Consumer Research.
Rust, R. T., & Cooil, B. (1994). Reliability measures for qualitative data: Theory and implications. Journal of Marketing Research, 31, 1-14.
Sanbonmatsu, D. M., Kardes, F. R., & Gibson, B. D. (1989). The impact of initial processing goals on memory-based brand comparisons. In T. K. Srull (Ed.), Advances in consumer research (Vol. 16, pp. 429-432). Ann Arbor, MI: Association for Consumer Research.
Smith, R. E., & Vogt, C. A. (1995). The effects of integrating advertising and
negative word-of-mouth communications on message processing and response. Journal of Consumer Psychology, 4, 133-151.
Teas, R. K., & McElroy, J. C. (1986, January). Causal attributions and expec-
tancy estimates: A framework for understanding the dynamics of salesforce motivation. Journal of Marketing, 50, 75-86.
Weinberger, M. G., Allen, C. T., & Dillon, W. R. (1981). The impact ofnega- tive marketing communications: The consumers' union/Chrysler con- troversy. Journal ofAdvertising, 10, 20-28+.
Weiner, B. (1980). A cognitive- (attribution-) emotion action model of moti-
vated behavior: An analysis of judgments. Journal of Personality and Social Psychology, 39, 186-200.
Wiener, J. L., & Mowen, J. C. (1986). Source credibility: On the independent effects of trust and expertise. In R. J. Lutz (Ed.), Advances in consumer
research (Vol. 13, pp. 306-310). Provo, UT: Association for Consumer Research.
Wilson, W. R., & Peterson, R. A. (1989). Some limits on the potency of word-of-mouth information. In T. K. Srull (Ed.), Advances in consumer
research (Vol. 16, pp. 23-29). Provo, UT: Association for Consumer Research.
Wind, Y. (1978). Issues and advances in segmentation research. Journal of Marketing Research, 15, 317-337.
Wright, P. L. (1974). The harassed decision-maker. Time pressures, distractions,
and the use of evidence. Journal ofApplied Psychology, 59, 555-561.
Zaichkowsky, J. L. (1985). Measuring the involvement construct. Journal of Consumer Research, 12, 341-352.
Accepted by Paul Herr.
APPENDIX
Introduction for All Three Scenarios:
Person 1: “Hey Pat I'm thinking about buying a computer and I noticed that you were able to answer those
computer questions pretty easily.” Pat: ” Yeah, it's sort of a hobby of mine.” Person 1: “Well, do you know anything about the
Compaq (Everex) PC?”
Pat: “Yeah, I know a lot about 'em. Why, what do ya' want to know?”
Person 1: “I know we've only got a couple of seconds before
we get started again, but I've been thinking about
buying a Compaq (Everex),what do you think?”
High Consensus, High Distinctiveness, High Consistency Scenario Manipulation
Pat: “Well, I don't like 'em and lot of other people I have talked to don't like Compaq (Everex) either.”
Person 1: “Really, like who?” Pat: “Well, a bunch of different people I know have said
they have had some sort of problem with 'em.”
Person 1: “Yeah, but what about you? Have you used 'em?”
Pat: “Yeah, I've used Compaq (Everex) a lot and ev- ery time something always went wrong. It just seemed like every time I boot one up I have some kind of a problem.”
Person 1: “So, you don't think much of Compaq (Everex) then, huh?”
Pat: “Well, no, I don't think much of Compaq (Everex). But, I could say some good things about some of the other brands out there. You
know, there are lots of other good PCs, but I don't think Compaq (Everex) is one of 'em.”
Low Consensus, Low Distinctiveness, High Consistency Scenario Manipulation
Pat: “Well, I don't like 'em. But, you know, a lot of other people I have talked to like Compaq (Everex).”
Person 1: “Really, like who?” Pat: “Well, a bunch of different people I know have
said they haven't had any problems with 'em.”
Person 1: “Yeah, but what about you? Have youused'em?” Pat: “Yeah, I've used Compaq (Everex) a lot and ev-
ery time something always went wrong. It just seemed like every time I boot one up I have some kind of a problem.”
Person 1: “So, you don't think much of Compaq (Everex) then, huh?”
Pat: “Well, no, I don't think much of Compaq (Everex). Though, to be honest, I can't say many good things about the other brands out there either. You know,
there aren't many good PCs, and I don't think Compaq (Everex) is a good one either.”
Low Consensus, High Distinctiveness, Low Consistency Scenario Manipulation
Pat: “Well, I don't like 'em. But, you know, a lot of other people I have talked to like Compaq (Everex).”
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
CONSUMERS' RESPONSES 73
Person 1: “Really, like who?” Pat: “Well, a bunch of different people I know have
said they haven't had any problems with 'em.” Person 1: “Yeah, but what about you? Have you used'em?” Pat: “Yeah, I've used Compaq (Everex) a lot, and
the last time something went wrong. But, you know, I hadn't had any problems with it except for that last time.”
Person 1: “So, you don't think much of Compaq (Everex) then, huh?”
Pat: “Well, no, I don't think much of Compaq (Everex). But, I could say some good things about some of the other brands out there.
You know, there are lots of other good PCs, but I don't think Compaq (Everex) is one of 'em.”
This content downloaded from �����������47.132.152.38 on Sun, 21 May 2023 12:03:55 +00:00������������
All use subject to https://about.jstor.org/terms
- Contents
- image 1
- image 2
- image 3
- image 4
- image 5
- image 6
- image 7
- image 8
- image 9
- image 10
- image 11
- image 12
- image 13
- image 14
- image 15
- image 16
- image 17
- Issue Table of Contents
- Journal of Consumer Psychology, Vol. 11, No. 1, 2001
- Front Matter
- Low-Involvement Learning: Repetition and Coherence in Familiarity and Belief [pp. 1 – 11]
- Processing Product Unique Features: Alignability and Involvement in Preference Construction [pp. 13 – 27]
- Service Experiences and Satisfaction Judgments: The Use of Affect and Beliefs in Judgment Formation [pp. 29 – 41]
- When Arousal Influences Ad Evaluation and Valence Does Not (And Vice Versa) [pp. 43 – 55]
- Consumers' Responses to Negative Word-of-Mouth Communication: An Attribution Theory Perspective [pp. 57 – 73]
- Back Matter
