Before learning how to apply social psychological research and theory in real life scenarios, it is important to be able to synthesize complex information and relay this informati

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 �����������216.54.120.76 on Mon, 22 May 2023 15:13:35 +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 �����������216.54.120.76 on Mon, 22 May 2023 15:13:35 +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 �����������216.54.120.76 on Mon, 22 May 2023 15:13:35 +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 �����������216.54.120.76 on Mon, 22 May 2023 15:13:35 +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 �����������216.54.120.76 on Mon, 22 May 2023 15:13:35 +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-