Marsha L. Richins

Negative Word-of-Mouth byDissatisfied

Consumers: APilot Study

While marketing scholars have emphasized theimportance of customer satisfaction, few studieshave examined in detail consumers' responses todissatisfaction. This study examines correlates ofone possible response—telling others about thedissatisfaction—and identifies variables that dis-tinguish this response from others. Variables in-vestigated include the nature of the dissatisfac-tion, perceptions of blame for the dissatisfaction,and perceptions of retailer responsiveness. Mar-keting management and consumer behavior re-search implications are discussed.

Introduction

MARKETING firms have traditionally been in-terested in customer satisfaction, and with good

reason. Customers continue to purchase those prod-ucts with which they are satisfied, and in telling oth-ers about particularly pleasing products, they may in-fluence the brand perceptions of those with whomthey communicate.

Prior to the 1970s little was published in the mar-keting literature about customer satisfaction. Partiallyin response to the consumer movement, however, in-terest in this topic rapidly grew. Several conferenceswere held (e.g.. Day 1977, Hunt 1977) and articlesbegan appearing in the marketing literature. A numberof these studies discussed the appropriate ways tomeasure satisfaction levels (Andreasen 1977, West-brook 1980). Others investigated the causes andsources of dissatisfaction (Day and Landon 1976;Diamond, Ward and Faber 1976), and some ad-dressed the theoretical bases of satisfaction (Oliver1980, Swan and Combs 1976). While work pro-gressed in this area, however, much less attention was

Marsha L, Richins is Assistant Professor of Marketing, Louisiana StateUniversity.

given to consumers' reactions to dissatisfaction. Po-tential responses include (a) switching brands or re-fusing to repatronize the offending store, (b) makinga complaint to the seller or to a third party, and (c)telling others about the unsatisfactory product or re-tailer.

The potential impact of these responses on a firmcan be significant. One nationwide study (TechnicalAssistance Research Programs 1979) reported thatdepending on the nature of the dissatisfaction, from30 to over 90% of dissatisfied respondents did notintend to repurchase the brand involved. Data re-ported by Diener and Greyser (1978) indicated that34% of those dissatisfied with a personal care producttold others about their dissatisfaction. If the numberof consumers experiencing dissatisfaction is highenough, such responses may have lasting effects interms of negative image and reduced sales for thefirm.

Many firms have not worried about these negativeeffects because they believed few consumers weredissatisfied with their products. Managers tended touse complaint rates as dissatisfaction indicators andassumed that if complaint rates are low, overall dis-satisfaction is low and thus the negative effects de-scribed above are neglible. A number of studies (Bestand Andreasen 1977, Day and Landon 1976) have

68 / Journal of Marketing, Winter 1983

challenged this assumption. While estimates of theincidence of complaints in response to dissatisfactionvary, it is generally accepted that the incidence ofcomplaints is lower than the incidence of either of theother responses. In addition, the vast majority of com-plaints are made at the retail level, and when small-ticket items are involved, information about the com-plaint is rarely passed on to manufacturers. Thus, fre-quency of reported complaints, especially at the man-ufacturer's level, may grossly underestimate consumerdissatisfaction and the firm's vulnerability to negativeword-of-mouth; brand switching will be greater thanthe number of registered complaints would lead man-agers to believe.

While little research has studied responses to dis-satisfaction directly, some areas of the consumer be-havior literature appear at least indirectly relevant tothe three possible responses.

Repeat Purchase Behavior

Repeat purchase behavior is most often addressed inbrand loyalty research. Much of this literature hasbeen concerned with conceptualizing, measuring andmodeling brand loyalty. The effects of a dissatisfac-tory experience on repeat purchase behavior haverarely been investigated directly.

One study (Technical Assistance Report Programs1979) reported that those dissatisfied customers whomade a complaint about their dissatisfaction reportedhigher repurchase intentions than those who did notcomplain, even if their complaint was not satisfactor-ily handled. Newman and Werbel (1973) noted thatconsumers not fully satisfied with a brand are lesslikely to repurchase that brand than satisfied cus-tomers. Many researchers believe that brand loyaltyincludes a positive attitude or preference toward abrand as well as simple repeat purchase (Jacoby andChestnut 1978). Beyond this, however, little is knownabout the influence of dissatisfaction on brand loyaltyand switching.

Complaint BehaviorThe second possible response to dissatisfaction—making a complaint in an attempt to remedy the dis-satisfaction—has received more attention in the lit-erature. This research appears to have grown directlyout of the consumer movement, and perhaps the ma-jority of the studies have been based on analysis ofprivate or government agency complaint files. Somegeneralizations that have emerged from research onconsumer complaining include the following:

• Those who complain when dissatisfied tend tobe members of more upscale socioeconomicgroups than those who do not complain (War-land, Herrmann and Willitts 1975).

• Personality characteristics, including dogma-tism, locus of control, and self-confidence, areonly weakly related to complaint behavior, ifat all (Settle and Golden 1974, Zaichkowskyand Liefeld 1977).

• The severity of the dissatisfaction or problemscaused by the dissatisfaction is positively re-lated to complaint behavior (Lawther, Krishnanand Valle 1979; Swan and Longman 1973).

• The greater the blame for the dissatisfactionplaced on someone other than the one dissat-isfied, the greater the likelihood of complaintaction (Lawther, Krisbnan and Valle 1979;Valle and Koeske 1977).

• The more positive the perception of retailer re-sponsiveness to customer complaints, the greaterthe likelihood of complaint action (Grabicke1980; Granbois, Summers and Frazier 1977).

Word-of-Mouth

Word-of-mouth (WOM) behavior, the third responseto dissatisfaction, has received the least attention ofall three areas and is often subsumed under the opin-ion leadership rubric. In applying opinion leadershipfindings to dissatisfaction responses, however, a num-ber of limitations become evident. Most writers haveconsidered only positive and not negative word-of-mouth (though it is mentioned by both Amdt (1968)and Dichter (1966)), and it is usually discussed interms of informing others about new products (dif-fusion of innovations) rather than consumer commu-nications about existing products. Finally, those en-gaging in negative WOM activities may not actuallybe opinion leaders. A review of work investigatingimpacts of negative information on consumers (Wein-berger, Allen and Dillon 1981) listed only one study(Amdt 1968) investigating negative word-of-mouth.Thus this literature sheds little light on dissatisfactionresponse. While a few studies in the dissatisfactionliterature have examined negative WOM activities,these have been limited to reporting the incidence ofthis behavior. No published research has examinedwhy some dissatisfied consumers engage in WOMwhile others do not, nor have correlates of negativeword-of-mouth activity been investigated. This isclearly a gap in understanding consumer behavior,since several researchers (Lutz 1975, Wright 1974)have found that consumers seem to place more weighton negative information in making evaluations (see,however, Scott and Tybout 1981). In addition, it iswell-accepted by marketing scholars and managersthat nonmarketing dominated sources of informationare given substantial weight by consumers in formingopinions and making product decisions.

Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot Study / 69

Scope of the EmpiricalInvestigation

The study reported here investigates WOM commu-nication as a response to dissatisfaction. Its first ob-jective is to determine whether the same variables thataffect complaining also affect WOM. Of the manyvariables showing a relationship to complaining, threewere chosen for this investigation: Severity of the dis-satisfaction or problems caused by the dissatisfaction,attributions of blame for the dissatisfaction, and per-ceptions of retailer responsiveness to complaints.Three hypotheses relevant to this objective were de-veloped.

HI: As the severity of the problem associatedwith a dissatisfaction increases, the ten-dency to engage in negative WOM ac-tivities increases.

H2: The greater the blame for the dissatis-faction placed on members of the distri-bution channel (marketing institutions),rather than on the consumer, the greaterthe tendency to engage in negative WOM.

H3: The more negative the perception of re-tailer responsiveness to complaints aboutdissatisfaction, should they be regis-tered, the greater the likelihood the dis-satisfied consumer will engage in nega-tive WOM. If an individual is dissatisfiedwith a product but believes attempts toachieve remedy through marketing chan-nels will be either unsuccessful or re-quire extensive effort, it is conceivablethat a less risky and less effortful re-sponse, telling others about the dissatis-faction, will be undertaken.

The second objective is to identify variables de-termining which response to dissatisfaction, amongthose available, a consumer will make. Some of theseresponses may be ranked a priori by the level of effortinvolved. Doing nothing, for instance, requires noeffort or resources, while making a complaint ofteninvolves a great deal of effort and inconvenience.Telling others about the dissatisfaction requires a lowto intermediate level of effort expenditure. Brandswitching is more difficult to rank. The effort in-volved here depends in part on the availability of ac-ceptable substitutes at retail stores the consumer reg-ularly patronizes, search required to identify thesesubstitutes, and other variables. For this study, vari-ables influencing the three responses ranked on effortwere examined to determine which response to dis-satisfaction a consumer may make.

One variable that may influence consumer re-

sponse is dissatisfaction severity. If a dissatisfactionis very minor, a consumer may take no action at all.For very serious dissatisfactions, however, a greatdeal of effort might be expended in response. A sec-ond variable is blame attribution. If the consumerblames him/herself for the dissatisfaction, e.g., throughcarelessness in the choice or use of the product, noaction may be taken. Thus:

H4: The more serious the problem associatedwith a dissatisfaction, the greater the ef-fort a consumer is likely to expend inresponse to the dissatisfaction.

H5: The greater the blame for the dissatis-faction placed on marketing institutions,the greater the effort a consumer is likelyto expend in response to the dissatisfac-tion.

For those dissatisfactions serious enough to arousesome action, the choice of action depends on a num-ber of variables. One important variable that may in-fluence this choice is consumer perception of retailerresponsiveness.

H6: For those consumers taking some actionin response to dissatisfaction, the lesspositive the perception of retailer respon-siveness should a complaint be regis-tered, the greater the likelihood the ac-tion will involve WOM but not complaintbehavior.

Data CollectionThe investigation was carried out in two stages. Inexploratory work, depth interviews were conductedwith eight adult consumers to probe their beliefs aboutthe variables included in the hypotheses. Next, ex-ploratory questionnaires containing open-end itemswere administered to 53 adult consumers and 72 col-lege students to identify relevant aspects of unsatis-factory experiences and complaint situations and toidentify attributions consumers make. Based on thiswork and extensive pretesting, a questionnaire wasdeveloped for use in the descriptive phase of research.

The final version of the questionnaire consisted offour major sections: (1) identification of the dissatis-faction, problem severity and response; (2) attribu-tions of blame and responsibility for the dissatisfac-tion; (3) perceptions of the complaint situation and thecosts and benefits associated with it; and (4) demo-graphic items. The questionnaire was administered toa sample of middle to upper middle-class adult maleand female consumers in Austin, Texas. All respond-ents were prescreened, and only those who had ex-perienced a dissatisfaction with either a clothing item

70 / Journal of Marketing, Winter 1983

or a small or large appliance within the prior sixmonths were questioned. These two product cate-gories were chosen because they are relatively dissim-ilar, yet adult consumers of both sexes can be ex-pected to have had experience with them. The use ofdifferent product classes in the analysis also increasesthe generalizability of findings beyond a single prod-uct class if relationships among variables are similarfor the two classes.

Interviewers visited respondents' homes, pre-screened respondents, dropped off the questionnairesand returned the following day to pick up completedforms. Two call-backs were used for initial contactand three call-backs for questionnaire pick-up whennecessary. Of the 261 individuals who reported dis-satisfactions, completed questionnaires were obtainedfrom 214, a response rate of 82%. Of this number 13were eliminated for incomplete responses on one ormore sections of the survey, leaving a sample size of201 cases for analysis. A breakdown of the sampleby sex and relevant product is shown below:

Appliances Clothing Total

MalesFemales

Total

4080

120

19

81

59142201

WOM and complaint rates did not differ significantlyfor the two product groups (WOM x"̂ = --50, df =1, p = .22; complaint x" = 1-14, df = 1, p = .29).

Analysis and FindingsThe primary dependent variables for this study wereconsumer responses to dissatisfaction. The WOMcommunication response was defined as the act oftelling at least one friend or acquaintance about thedissatisfaction, and 57.2% of the sample did so.Expression of dissatisfaction to family members wasnot included in the definition. Complainmg was de-fined to include complaints to retailers, manufacturersor third parties such as the Better Business Bureau orsome government agency; 33.3% of the sample re-ported making at least one such complaint. A signif-icant number of respondents, 32.3%, engaged in nei-ther WOM nor complaining behavior. Tests of thehypotheses are reported below.

WOM and Problem Severity

Data analysis showed that for this sample, as the se-verity of the problem associated with a dissatisfactionincreased, the tendency to engage in negative WOMincreased. Based on exploratory work and results re-ported by Robinson (1979), four variables were usedto measure problem severity: Length of product own-ership before dissatisfaction arose, whether the prod-uct could be used although it was unsatisfactory, howdifficult it would be for the individual to repair theproduct, and product price. The first two variablesdisplay a negative relationship with perceptions of

TABLE 1Correlations between Predictor Variables and Dissatisfaction Responses"

Word-of-Mouth

ComptaintBehavior

Problem severityLength of ownershipUsability "̂Difficulty of repairProduct price

AttributionsPercent of external attributions

Probability of remedyHighestAverageProbability of preferred remedy

Trouble involved—probability of:Special trip to connplainTime and effort filling out formDifficulty finding complaint procedureBeing treated rudelyHaving to hassle someoneBeing blamed for the dissatisfactionBecoming embarrassed while making the complaint

.10

.06

.16

.21

.19

.15

.16

.13

.06

.17

.21

.25

.19

.04

.04

.12

.04

.22

.24

.21

.12

.14

.14

.03

.11

.12

.08

.09

.11

.07

"Coefficients >.16 are significant at p < .01; coefficients >.12 are significant at p < .05."Phi coefficient; all other coefficients are point biserial.

Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot Study / 71

problem severity and thus with WOM, while the oth-ers are positively related. Length of ownership andproduct cost were measured on continuous scales; us-ability was measured as a bivalent dichotomous vari-able, and difficulty of repair was measured on a 5-point scale. Point biserial correlations between thesevariables and WOM are reported in Table 1. For com-parison, correlations with complaint behavior are alsoshown. Since correlations for the two separate productclasses are not significantly different for any of thevariables, only correlations for the combined sampleare reported. While all correlations are of the expectedsign, those calculated for length of ownership andproduct usability are not significant at p < .05.

These analyses tend to support HI, although notstrongly. The more serious the problem associatedwith the dissatisfaction, the more likely consumers areto tell others about it.

WOM and Attributions

H2 proposed that when greater blame for a dissatis-faction is placed on marketing institutions than on theconsumer, there will be more negative WOM. Thishypothesis was also supported. The survey instru-ment, developed from depth interviews and explora-tory questionnaires, contained a list of 26 possibleattributions of blame one might apply to a dissatis-faction. One-half of these placed the blame externally(e.g., the dissatisfaction occurred because the qualityof the materials and/or workmanship was inferior)and one-half were internal attributions (e.g., the dis-satisfaction occurred because the respondent didn'tadequately inspect the product before buying it). Re-spondents checked as many attributions as applied tothe particular dissatisfaction situation.

To control for variability in number of responseschecked, H2 was tested using an index measure. Thetotal number of external attributions indicated by eachrespondent was divided by the total number of attri-butions indicated, yielding a percentage measure. Arespondent marking three attributions, two of themexternal, would thus score .67 on this index. Pointbiserial correlations between this index and WOM is.19, as reported in Table 1. As hypothesized, thoseindividuals marking a higher percentage of externalattributions of blame are more likely to tell othersabout their dissatisfaction. Since an index of percent-age of internal attributions is the exact complementof the external index, it would show the same levelof correlation with WOM but in the opposite direc-tion. Thus, external attributions of blame are posi-tively related to WOM, and internal attributions neg-atively related.

WOM and Retaiier Responsiveness

Retailer responsiveness can be measured on two di-mensions. One dimension is the retailer's willingnessto provide a remedy for the dissatisfaction should aconsumer complain. The second dimension measuresthe extent to which the retailer makes the complainthandling mechanism available, including the numberof barriers a consumer may face in making a com-plaint. Analysis indicated that both dimensions relateto negative WOM in response to a dissatisfaction.

To evaluate the first dimension, respondents wereasked to report on a four-point scale the likelihood ofreceiving each of three possible remedies should theycomplain: repair of the product, replacement or refundof the purchase price. They also reported which ofthese three remedies they most preferred.

The nature of the product involved in the dissat-isfaction undoubtedly affects consumers' expectationsof remedy and the form of remedy. Appliances, es-pecially larger ones, are usually repaired rather thanreplaced if they are faulty. For less expensive itemssuch as clothing or small appliances, replacement orrefund of the purchase price is a more likely remedythan product repair. Because of the diverse nature ofthe product classes involved in this study, measuresindependent of these product influences were neces-sary. Three measures were used: the highest perceivedprobability for any of the three remedies, the averageof the probabilities that each of the three remedieswould occur, and the perceived likelihood of receiv-ing the remedy the respondent indicated he/she wouldmost prefer. The correlations between these measuresand WOM are reported in Table 1. Those individualswho have low confidence in the effectiveness of mak-ing complaints are more likely to tell others abouttheir dissatisfactions than those expecting remedy.

With regard to the inconvenience or barriers de-terring a complaint, respondents were asked a seriesof questions concerning their perceptions of com-plaining and the complaint interaction. Five items inthis series related directly to the trouble involved inmaking a complaint and two concerned psychologicalvariables. Respondents were asked to rate the likeli-hood of each event occurring. Point biserial correla-tions between these perceptions and WOM are pre-sented in Table 1. Neither of the psychologicalvariables showed a relationship with WOM, nor didthe more objective variable of making a special tripto the store to complain. This latter finding is nodoubt due to the extremely low variance on this item,with 82% of total sample respondents indicating thatit was somewhat to very likely that a special tripwould be required. Though the relationships shownin Table 1 are not extremely strong, they do support

72 / Journal of Marketing, Winter 1983

H3: the more negative a consumer's perceptions ofretailer responsiveness to consumer complaints, themore likely that individual is to engage in negativeWOM.

Problem Severity, Attributionsand Consumer ActionThe remaining analyses presented in this report con-cern consumers' choices of action once dissatisfactionoccurs. For the analyses reported in this section, tbesample was divided into three groups depending onresponses to dissatisfaction. Sixty-five respondents inthe sample neither complained nor told others abouttheir dissatisfaction experience, 69 engaged in WOMactivities but did not complain, and 67 actually madecomplaints to the retailer or manufacturer involved.Some individuals in this latter group engaged inWOM in addition to making a complaint. It was as-sumed that members of the first group had made theleast effort in response to the dissatisfaction, membersof the third group had made the most effort, while thesecond group exerted an intermediate level of effort.

H4 proposed that greater problem severity is as-sociated with greater effort expenditure in responseto a dissatisfaction. This hypothesis was supportedusing two-way analysis of variance with group mem-bership according to action taken and product classas independent variables. Responses to the problemseverity variables included in HI were the dependentvariables.

Results by product class are shown in Figure 1.To illustrate, in the first entry in Figure 1 the groupwho made a complaint and were dissatisfied with anappliance item had owned tbe product, on the aver-age, for 2.19 weeks when the dissatisfaction occurred;those complaining about a clothing item had ownedthe product for .19 week (a little more than 1 day)when the dissatisfaction occurred. These numbers canbe compared with those for the WOM group and thegroup who neither complained nor engaged in WOMimmediately following.

In the analyses reported in Figure 1, the same vari-ables that correlated with WOM as tested in HI alsodiffered significantly among the three dissatisfactionresponse groups. Difficulty of repair and productprice were highest for those individuals who com-plained and were lowest for those who neither com-plained nor told others of their dissatisfaction. TheWOM group fell between these groups for these vari-ables. As expected, duration of ownership before thedissatisfaction arose showed the opposite relationship.Usability shows a mixed pattern and differencesamong groups are not statistically significant. Thesefindings support H4: Those consumers who engagedin more energetic responses were those who experi-

FIGURE 1Group Means: Problem Severity and

Attributions^

PROBLEM SEVERITY

LtNGTH Of OWNERSHIP (IN WEEKS) ^'^

USABILITY (PERCENTAGE SIATIMG PRODUCT WAS U5ABLE)^

ComploioedI 51.2

IIHIIIIIlllllllllHHIIIHIimMIUIIIIIM.5

Nci!h»

lltDllllllllllliiiiiiiiiiiiiiiiiiiiiliiimiHIiHmui 79.3

DIFFICULTY OF REPAIB

i l l l l l lMll l l l l t l l l lDll l tDlt l l l l l l lMll l l l l l l l ir i l l l lHHIl 3.33

nil I I I I IHI I I I I I I I I I I

Illllllllllllllllllllllllllllllllll 2 .3B

PRODUCT PRICE (iOO) ^ - ^

Wofd-of-MtJulh

ATTRIBUTIONS

PERCENTAGE EXTERNAL ATTRIBUTIONS

IHIIIIlllMIt IMIIII Itllllllllllllllllll 75. I

IIIIIHMHIIIIII ••IIIIIIHIIIIIIIIIIlin 69.7I APPLIANCES

" " IIHIIIIIIIII1 CLOTHINGHi l l iilliiniiiHiiiiiiiiiiiHHIIIIIH 56.2

' TWO-WAY ANOVAS WERE CONDUCTED FOR EACH VARIABLE EXCEF7 AS NOTED WITHACTION TAKEN AND MODUCT CLASS AS INDEPENDENT VARIABLES. NO INTERACTIONEFFECTS WERE SIGNIFICANT AT P < . 0 5 .

' MAIN EFFECT FOR ACTION TAKEN SIGNIFICANT AT P < . 0 5 .^ MAIN EFFECT FOR ACTION TAKEN SIGNIFICANT AT P < . 0 1 .* MAIN EFFECT FOR PKODUCT CLASS SIGNIFICANT AT P < . 0 5 .^ CHI 5OUARE ANALYSIS, N.S

enced greater problem severity resulting from the dis-satisfaction.

In the analysis reported in Figure 1, differencesbetween product classes were significant for only twoof the variables, length of ownership when the dis-satisfaction occurred and price of the unsatisfactoryproduct. The latter product difference is expected,since appliances as a class are generally more expen-sive than clothing. With respect to the length of own-ership, dissatisfaction with clothing items was notedmuch sooner after purchase than was dissatisfactionwith appliance items. Perhaps because of the com-

Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot Study / 73

FIGURE 2Group Means: Perceptions of Retailer

Responsiveness^

PROBABILITY OF:

REMEDY SHOULD COMPLAINT B£ MADE "^'

Comploined12 .17

Illlllllllllllllllllllllllllll •••••IIIIIIIIIIMIIIIIIIIIIIIIIII 2.

Word-of-MourhI 1.79

SPECIAL TRIP TO COMPLAIN

Comploined13.25

Word-of-MourVi13.43

illlllllllllllllllllllllllllllllllMnnnilllllllllllMIIINMMMIIIIIMIIIIIIIIII 3.69

TIME AND EFFORT FILLING OUT FORMS ̂

Comploined

Word-of-Mouth3.28

•••••IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIMIIIIIIIIIIIMIIIIII 3.28

DIFFICULT/ FINDING COMPLAINT PROCEDURE

Complained

3,4

2.52

lllllllllllllllHllllllllllllllllllllllllllll 1. 83

Word-of-Moulh

Illlllllllllllllllllllllllllllllllllllllllllllllllllllllllllll 2.66

BEING TREATED RUDELY

Comploined12.20

Illllllllllllllllllllllllllllllllllll 1.61

Word-of-MoufhI 2.35

2.52

HAVING TO HASSLE SOMEONE

Comploined12.64

IIIIIIIIHIIIIIIIIIIMIIIMIIIIIIIIIIIIIIIIMIIII 2.04

Word-of-Mourh2.90

IIIIIIIIMIII nil Hill" •••< 2 . 9 0

BEING BLAMED FOR THE DISSATISFACTION^

Complained2.05

iilllliiiiiiiiiilllllliliiiiiiiiiiiiiiiiiiilllll 2.01

Word-of-Mouth2.25

iiiiiiiiiMiiniiiuiiiiiiiMiiiiiiiiiiiiiiiiiiiiiiiiiiiiinniHi 2.72

BECOMING EMBARRASSED WHILE MAKING COMPLAINT

Comploinedl.t

I l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l l 1.91

Word-of-Mouth1.80

Ilillliiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiilllllllll 2.31

' TWO-WAY ANOVAS WERE CONDUCTED FOR EACH VARIABLE WITH ACTION TAKEN AND PRODUCTCLASS AS INDEPENDENT VARIABLES. NO INTERACTION EFFECTS WERE SIGNIFICANT AT P< .05.

^ MAIN EFFECT FOR ACTION TAKEN SIGNIFICANT AT P< .05 .

3 MAIN EFFECT FOR ACTION TAKEN SIGNIFICANT AT P < . 0 1 .

•* MAIN EFFECT FOR PRODUCT CLASS SIGNIFICANT AT P< .05 .

APPLIANCES

•iiiiiiiiiiiiiiiiiiiiill CLOTHING

plexity of appliances, problems may not appear untilthe product has been used for a period of time. Alldifferences between product classes are attributableto main effects, since no interaction effect was sig-nificant at p < .05. Thus, the nature of the relation-ships between the problem severity variables in Figure1 and action taken are the same for the two productclasses.

Attributions data for the three groups are also re-ported in Figure 1 with significant differences amonggroups in the directions predicted by H5. The com-plaint group made the greatest percentage of externalattributions, and the group who neither complainednor engaged in WOM made the lowest. Thus, thegreater the blame for the dissatisfaction placed on

marketing institutions, the greater the effort a con-sumer is likely to expend in response to the dissatis-faction.

Retailer Responsiveness and Consumer ActionTo determine the effect of perceptions of retailer re-sponsiveness on the kinds of action dissatisfied con-sumers make, only responses of those consumers suf-ficiently dissatisfied to take some action—eitherWOM or actual complaint—were examined. Percep-tions of those who complained were compared withperceptions of individuals engaging in WOM, againusing two-way analysis of variance. Results for thetwo product classes are shown in Figure 2. Those in-dividuals who complained had more positive percep-

74 / Journal of Marketing, Winter 1983

tions of retailer responsiveness on all variables excepttwo, relating to the likelihood of having to make aspecial trip to complain and the likelihood of becom-ing embarrassed while complaining.

There were significant product differences for twoof the variables reflecting retailer responsiveness—consumers who experienced dissatisfaction with anappliance had lower expectations of remedies for theirdissatisfactions and greater expectations of experienc-ing difficulty in fmding the appropriate complaint pro-cedure should they complain. Since these product dif-ferences again are due only to main effects, the natureof the relationship between retailer responsivenessand action taken are the same for the two productclasses.

These analyses provide rather strong support forH6. If a consumer is dissatisfied enough to take someaction in response to a dissatisfaction, the less positivethe perception of retailer responsiveness, the greaterthe likelihood the action will involve WOM but notcomplaint behavior. In examining the results for thisand other hypotheses, however, caution is appropriatesince the use of multiple univariate tests increases thechances of finding statistical significance.

Combined Effects

To determine the combined effects of the variables inH4, H5 and H6 on consumers' choices of responsesto dissatisfaction, a three-group discriminant analysisusing the three groups previously identified was per-formed using as predictors the variables included inthe tests of preceding hypotheses. {For probability ofremedy tested in Hypothesis 6, only one variable—average probability of remedy—was included in theanalysis because of extreme multicollinearity amongthe variables.) To incorporate the main effects forproduct class noted in Figures 1 and 2, a dummy vari-able for product class was added to the analysis. Twofunctions significant at p < .01 emerged from thediscriminant analysis. Table 2 lists function coeffi-cients, canonical correlations and group centroids. Todetermine the variance accounted for by these func-tions together, the statistic I ' (Peterson and Mahajan1976) analogous to R' in multiple regression analysis,was calculated (and equaled .37 for these data). Inclassification analysis, the discriminant functions pre-dicted correct group membership in 63% of the cases,nearly twice the frequency anticipated by chance.

TABLE 2Discriminant Analysis Results

StandardizedFunction Coefficients'

Problem severityLength of ownershipUsabilityDifficulty of repairProduct price

Attributions (percent external attributions)

Retailer responsiveness—probability of:Remedy if complaint madeSpecial trip to complainTime and effort filling out formsDifficulty finding complaint procedureBeing treated rudelyHaving to hassle someoneBeing blamed for the dissatisfactionBecoming embarrassed while making the complaint

Product class

.35

.05

.46

.59

.54

.32

.14

.22

.25

.17

.18

.32

.01

.17

.12

.13

.21

.06

.30

.57

.05

.41

.08

.37

.18

.26

.09

.45

EigenvalueCanonical correlation

Group centroidsComplainedWord-of-mouthDid neither

.36

.51

.82

.27

.56

.19

.40

.12

.57

.47

Loadings of .35 or greater are in boldface.

Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot Study / 75

Thus, the three sets of variables used in this studyconcerning consumers' perceptions of the dissatisfac-tion and retailer responsiveness do a relatively goodjob of predicting consumer response to a specific dis-satisfaction.

Earlier it was suggested that severity of the prob-lem situation caused by the dissatisfaction and attri-butions of blame determines whether action is takenin response to dissatisfaction and the effort involvedin this response. Further, perceptions of retailer re-sponsiveness affect which response should be in-voked, if any occur at all. This hypothesis receivesfurther support in an examination of the discriminantfunctions and group centroids shown in Table 2. Theitems loading most highly on Function 1 relate toproblem severity and attributions; those loading highlyon function 2 relate primarily to perceptions of retailerresponsiveness and product class. With respect togroup centroids, the complaining group scored high-est of the three groups on function 1, the WOM groupscored highest on function 2, and the group who nei-ther complained nor told others of their dissatisfactionscored the lowest on both functions. This result fur-ther supports the contention that problem severity isthe crucial determinant of effort of response, whilethe choice between WOM and complaint behavior isinfluenced by perceptions of retailer responsiveness.

Conclusions and ImplicationsThis study has shown that the nature of the dissatis-faction problem, consumers' attributions of blame forthe dissatisfaction and perceptions of the complaintsituation are related to responses to dissatisfaction.Further, these relationships apply to varied productclasses.

These findings have several implications for man-agement. When a minor dissatisfaction is experi-enced, consumers' responses often are minimal. Mostoften consumers neither complain nor spread negativereports of the product involved. When the dissatis-faction is serious enough, consumers tend to com-plain, regardless of other factors in the situation. Itis at moderate levels of dissatisfaction that manage-ment policy may have the most impact. If complaintsare encouraged, the retailer has the chance to remedylegitimate complaints and win back a customer whomay also make positive reports to others, enhancinggoodwill. Even if the complaint is not settled to theconsumer's satisfaction, he/she is more likely to re-purchase than if no complaint is made {TechnicalAssistance Research Programs 1979). If complaintsare discouraged, fewer consumers may indeed com-plain; instead, they may tell others of their unsatis-factory experiences and may not repurchase the prod-uct in the future. Some level of dissatisfaction is

inevitable in the marketplace. Even with conscien-tious quality control, there will be some defects andsome dissatisfied customers. But the way manage-ment deals with these dissatisfactions can have im-portant impacts on brand and store image.

There are a number of steps a marketing institu-tion can take to decrease the impact of dissatisfactionsthat occur and lessen the incidence of negative word-of-mouth. At the least, they need to show their cus-tomers that they are responsive to legitimate com-plaints. Manufacturers may do this through detailedwarranty and complaint procedure information on la-bels or in package inserts. Some companies have evenoffered toll-free telephone numbers to receive cus-tomer comments and complaints. Retailers can showtheir receptivity through store signs, inserts in monthlycustomer billings and by positive employee attitudes.Perhaps the best way to show responsiveness to cus-tomer complaints is by prompt and courteous handlingof legitimate dissatisfactions. Such responsivenessmay avert negative WOM and may even create pos-itive WOM.

This study noted that consumer perceptions of in-stitutional responsiveness depended on which productwas the source of dissatisfaction. Perceived respon-siveness for complaints about appliance items waslower than for clothing items. Consumer perceptionsare usually (though not always) based either on priorpersonal experience or reports of the prior experiencesof others. It is undoubtedly more expensive to remedya dissatisfaction concerning an appliance becausethese items are more expensive. Retailers may be re-luctant to make these expenditures and may commu-nicate this reluctance by making complaint proceduresmore difficult. Also, dissatisfactions with applianceitems tend to occur farther from the point of purchasethan dissatisfactions with clothing items. By the timethe dissatisfaction occurs, the retailer's warranty onthe product may have expired, so the complaint isreferred to the manufacturer who usually provides alonger warranty period. While the manufacturer mayindeed remedy the dissatisfaction, this referral stepadds more paperwork and delay, sometimes creatingthe impression of unresponsiveness.

In dealing with complaints, marketing institutionsmust examine not only the costs of the remedy butthe costs of not settling the complaint as well. Theselatter costs are, of course, more difficult to assess, butmanagers must be aware that they include the poten-tial for negative WOM in addition to the potential lostrepeat business.

This study also has important implications for con-sumer behavior researchers. While it has shown thata number of variables both within and outside man-agement control affect consumer responses to dissat-isfaction, this single study has only scratched the sur-

76 / Journal of Marketing, Winter 1983

face. One area needing further investigation is brandswitching in response to dissatisfaction. Despite re-search on brand loyalty, it is not clear how frequentlybrand switching in response to dissatisfaction occurs,how the nature of the dissatisfaction influencesswitching, what kinds of products are more suscep-tible to this response, and a host of other details. Thephenomenon of negative word-of-mouth also bearsfurther investigation, including determination of the

extent of negative WOM and its effects on recipientsof negative communications concerning products andretailers. Additional factors that may influence neg-ative WOM such as personality variables (includingsociability), situational variables (e.g., proximity ofothers when the dissatisfaction occurs) and attitudinalvariables (e.g., attitudes toward business), also needinvestigation.

REFERENCESAndreasen, Alan R. (1977), "A Taxonomy of Consumer Sat-

isfaction/Dissatisfaction Measures," Journal of ConsumerAffairs, 11 (Winter). 11-24.

Amdt, Johan (1968). "Word-of-Mouth Advertising and Per-ceived Risk." in Perspectives in Consumer Behavior, H.Kassaijian and T. Robertson, eds., Glenview, IL: Scott,Fores man.

Best. Arthur and Alan R. Andreasen (1977), "Consumer Re-sponse to Unsatisfactory Purchases: A Survey of PerceivingDefects, Voicing Complaints and Obtaining Redress," Lawand Society Review, 11 (Spring), 701-742.

Day, Ralph L.. ed. (1977). Consumer Satisfaction, Dissatis-faction and Complaining Behavior. Bloomington, IN: In-diana University Press.

and E. Laird Landon. Jr. (1976), "Collecting Com-prehensive Consumer Complaint Data by Survey Re-search," in Advances in Consumer Research. 3. B. B.Anderson, ed., Atlanta: Association for Consumer Re-search. 263-268.

Diamond. Steven L., Scott Ward and Ronald Faber (1976),"Consumer Problems and Consumerism: Analysis of Callsto a Consumer Hot Line." Journal of Marketing, 40 (Jan-uary), 58-62.

Dichter, Ernest (1966), "How Word-of-Mouth AdvertisingWorks," Harxard Business Review, 44 (November-De-cember), 147-157.

Diener, Betty J. and Stephen A. Greyser (1978). "ConsumerViews of Redress Needs," Journal of Marketing, 42 (Oc-tober), 21-27.

Grabicke. K. (1980), "Consumer Satisfaction, Dissatisfactionand Complaining Behavior," paper presented at the FifthAnnual Colloquium of the European Association of Eco-nomic Psychologists, Leuven and Brussels.

Granbois, Donald, John O. Summers and Gary L. Frazier(1977). "Correlates of Consumer Expectation and Com-plaining Behavior," in Consumer Satisfaction. Dissatisfac-tion and Complaining Behavior, R. L. Day, ed., Bloom-ington, IN: Indiana University Press, 18-25.

Hunt, H. Keith, ed. (1977), Conceptualization and Measure-ment of Consumer Satisfaction and Dissatisfaction, Cam-bridge, MA: Marketing Science Institute.

Jacoby, Jacob and Robert W. Chestnut (1978), Brand LoyaltyMeasurement and Management, New York: Wiley.

Lawther. Karen, S. Krishnan and Valerie A. Valle (1979)."The Consumer Complaint Process: Directions for Theo-retical Development," in New Dimensions for ConsumerSatisfaction and Complaining Behavior, R. L. Day and H.K. Hunt, eds.. Bloomington. IN: Indiana University Press,10-14.

Lutz, Richard J. (1975), "Changing Brand Attitudes throughModification of Cognitive Structure," Journal of Con-sumer Research, 1 (March), 49-59.

Newman. Joseph W. and Richard A. Werbel (1973), "Mul-tivariate Analysis of Brand Loyalty for Major HouseholdAppliances," Journal of Marketing Research, 10 (Novem-ber), 404-409.

Oliver. Richard L. (1980), "A Cognitive Model of the An-tecedents and Consequences of Satisfaction Decisions."Journal of Marketing Research. 17 (November). 460—469.

Peterson. Robert A. and Vijay Mahajan (1976), "PracticalSignificance and Partitioning Variance in DiscriminateAnalysis," Decision Sciences, 1 (October) 649-658.

Robinson, Larry M. (1979). "Consumer Complaint Behavior:A Review with Implications for Further Research," '\nNewDimensions of Consumer Satisfaction and ComplainingBehavior, R. L. Day and H. K. Hunt. eds.. Bloomington,IN: Indiana University Press, 41-50.

Scott. Carol A. and Alice M. Tybout (1981), "TheoreticalPerspectives on the Impact of Negative Information: DoesValence Matter?," in Advances in Consumer Research, 8,K. B. Monroe, ed.. Ann Arbor, MI: Association for Con-sumer Research, 408-409.

Settle, Robert B. and Linda L. Golden (1974). "Attributionand Action When a Product Fails," paper presented to theFifth Annual Conference of the Association for ConsumerResearch, Chicago.

Swan, John E. and Linda Jones Combs (1976). "Product Per-formance and Consumer Satisfaction: A New Concept,"Journal of Marketing, 40 (April), 25-33.

and Douglas S. Longman (1973), "Consumer Sat-isfaction with Automobile Repair Performance: AttitudesToward the Industry and Governmental Control." in Com-bined Proceedings. B. W. Becker and H. Becker, eds.,Chicago: American Marketing Association. 249-255.

Technical Assistance Research Programs (1979). ConsumerComplaint Handling in America: Summary of Findings andRecommendations, Washington, DC: U.S. Office of Con-sumer Affairs.

Valle, Valene A. and Randi Koeske (1977). "Elderly Con-sumer Problems: Actions. Sources of Infonnation and At-tributions of Blame," paper presented at the annual meet-ing of the American Psychological Association, SanFrancisco.

Warland. Rex H., Robert O. Herrmann and Jane Willitts(1975). "Dissatisfied Consumers: Who Gets Upset andWho Takes Action," Journal of Consumer Affairs, 9 (Win-ter), 148-163.

Weinberger, Marc G., Chris T. Allen and William R. Dillon

Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot Study / 77

(1981), "Negative Information: Perspectives and ResearchDirections," in Advances in Consumer Research, 8, K. B.Monroe, ed,, Ann Arbor, MI: Association for ConsumerResearch. 398-404.

Westbrook, Robert A. (1980), "A Rating Scale for MeasuringProduct/Service Satisfaction," Journal of Marketing, 44(Fall). 68-72.

Wright, Peter (1974), "The Harrassed Decision Maker: Time

Pressures, Distractions, and the Use of Evidence," Journalof Applied Psychology, 59 (October), 555-561.

Zaichkowsky, Judy and John Liefeld (1977), "PersonalityProfiles of Consumer Complaint Letter Writers," in Con-sumer Satisfaction. Dissatisfaction and Complaining Be-havior, R. L. Day, ed., Bloomington, IN: Indiana Uni-versity Press, 124-129.

What'snew in MARKETING for the 8O's?POLITICALMARKETINGAN APPROACH TOCAMPAIGNSTRATEGYGary A. Manser, SimonFraser University, CanadaHow can marketing principlesand research techniques be ex-tended into the arena ofpolitical campaigning?In this important new book,Gary Mauser explains how—byproviding political campaignerswith a practical method fordetermining campaign strategy;illustrating how this method hasbeen used to identify campaignstrategy for political candidatesin both the U.S. and France;and by providing an empiricaJexamination of the validity andreliability of this method.Focuses on the 1968 U.S. Pres-idential election, the 1970 con-test for Governor of California,the 1973 French Legislativeelections, and the 1974 FrenchPresidential contestca, 304 pp. 1983 ca. $21.95(W)3-052591-8

Available through your local bookseller or order directly from:

MARKETINGIN THETHIRDWORLDErdener Kaynak, MountSaint Vincent UniversityAnswers an outstanding needfor a timely overview of marketresearch and market activity inthe developing nations of theThird World. Gathering and in-terpreting data from numerouscountries. Kaynak illuminatessuch areas as product planningthe use of available marketingchannels, and effective marketresearch techniques. In addi-tion, he describes the trendsthat will shape the future ofmarketing in the developingnations.240 pp. 1982 $29.950-03-062179-8

MARKETINGTO CHANGINGCONSUMERENVIRON-M E N T S ENVIRON-MENTAL SCANNINGRonald D. Michmao,Shippensburg State College.PennsylvaniaProvides marketing executives witha starting point for understandingthe socioeconomic changes that af-fect spending patterns and impacton the organization. This dynamicframework for future planningfocuses on six interacting variables;Demography; Economic Condi-tions; Social and Cultural Forces;Political and Legal Forces; Tech-nology; and competition. Includesan analysis of demographic trendsto the year 2000 and techniques foranalyzing future marketopportunities,ca. 192 pp. 1983 ca. $19.950-03-0594294

And these titles in the PraegerSeries in Public and NonprofitSector Marketing...

THE MARKETING OFIDEAS AND SOCIALISSUES

H . F i n e Rutgers Univenity240 pp. 1981 $26.95 (M)3-059277-l

GOVERNMEIVTMARKETING THEORYAND PRACTICEedited by H i c h a e l M o k w a , AnzonaState University, and S t e V e n E .P e r m n t t Yale university400 pp. 1981 $33-95 0-03-058316^

MARKETINGTHE ARTSedited by Michael P. Hokwa,University of Wisconsin, Wl l l ian i M*Dawson, AcucAA,andE. ArthurP r i e V e ^ university of Wisconsin304 pp. 1980 $29.95 0^3-052141-6

PRAEGERPUBLISHERS ,

521 Fifth AvenueAttn: Ms. Sonya HarrooNew York, N.Y. 10175

78 / Journal of Marketing, Winter 1983