Professional Documents
Culture Documents
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/24099239
CITATIONS READS
138 1,472
2 authors:
All content following this page was uploaded by Bobby J Calder on 14 October 2015.
t (2)
~
tions are typically measured by asking observers to rate
the extent to which an actor possesses a disposition,
e.g., rating the actor's honesty on a semantic dif-
ferential scale. Confidence ratings are sometimes used,
observations of the actor's behavior. Such consistency more in keeping with the notion of correspondence,
eliminates possible confounds with nonpersonal to assess the overall strength of internal attribution.
factors. Finally, the plausibility of an internal Attribution theories, as we shall see, are by no
attribution increases if there is a consensus among means fully developed. Even so, our position is that
other people that a behavior reflects a disposition. these ideas suggest a general approach to interpersonal
Thus, to the extent that an actor's behavior exhibits influence. In purchasing and using products, the con-
distinctiveness and consistency over time, place, and sumer is a social actor whose behavior is largely open
the reactions of others, it is accepted as evidence of a to observation by others. The consumer's behavior is
personal disposition. Otherwise, the behavior is dis- informational input for the attribution processes of ob-
counted and attributed to external factors. servers. Observers infer the consumer's personal dis-
Jones and Davis' perspective is formulated some- positions from his or her behavior. Our argument is
what differently. They pose the question of how an that attributions underlie interpersonal influence. At-
observer can be sure that the language he uses to de- tributions amount to judgments about the consumer.
scribe an actor's behavior is also descriptive ofthe per- These judgments shape the observer's actions with re-
sonal dispositions of the actor. Their term for the match spect to the consumer. The observer's actions may
between observed behaviors and inferred dispositions directly affect the consumer's behavior. Attributions
is "correspondence." The problem for the observer is thus provide a psychological basis, or reason, for the
whether the actor's dispositions correspond to the ob- actions of influencers, something that is missing from
server's description of the actor's behavior. Jones and "group pressure" studies.
Davis explicitly include the "effects" produced by a Influence of a more indirect kind is accomplished if
behavior as well as the behavior itself in their analysis. the consumer is sensitive to the attributions others
Correspondence depends on the number of "non- make, or to those he expects them to make, or to those
common effects" produced by a choice act. "generalized others" might make, and he acts so as to
The theory postulates that observers construe be- produce the attributions he desires. That is, the con-
havior as a choice between either explicit or im- sumer engages in behaviors which he believes will lead
plicit alternatives. A chosen alternative is associated observers to make attributions which he considers
with a set of observed effects; each of the unchosen desirable. Conceptualizing such indirect influence as
alternatives is associated with a set of hypothetical operating through perceived product attributes is po-
consequences which would have been the effects had tentially misleading. The influence does not stem from
it been chosen. The chosen alternative mayor may not the product itself but from the consumer's beliefs about
have effects in common with the unchosen alternatives. the attributions others make from observing his be-
N oncommon effects indicate the basis of choice more havior. This is a far more dynamic process than pro-
clearly than do common effects. Common effects re- duct-attribute social-influence conceptualizations im-
flect external, situational constraints. To the extent that . ply.
a choice results in noncommon rather than common We propose a paradigm for research rather than a
effects, it yields a stronger inference of correspond- full-blown model. This paridigm is diagramed in Figure
ence. Moreover, the inference is stronger the fewer A. The research reported in this paper deals with a
the number of noncommon effects, for a few non- fundamental aspect of this paradigm, shown as Rela-
common effects indicate the actor's intentions more tionship (1) in Figure A. This is the functioning of the
precisely than a larger number. attribution process itself: how an observer infers per-
Jones and Davis also hold that correspondent in- sonal dispositions from a consumer's behavior. In
ference depends on the "assumed social desir- order to pursue the full paradigm, it is necessary, as a
ability" of effects. If most people would not have pre- first step, to investigate this relationship, to explore the
ferred a chosen alternative, the effects of that choice applicability of existing attribution theory to observa-
should be more descriptive of the uniquely personal tions of consumer behavior.
INTERPERSONAL INFLUENCE AND ATTRIBUTIONS 31
There are, however, two other aspects of the full is to generate a consumer franchise for a brand, to
paradigm (see Figure A). One is the accuracy of the create the impression that a brand is prestigious and
consumer's knowledge of the attributions observers widely purchased. With the cosmetic products used
make from his behavior. We refer to this as the con- in this study, a heavily advertised manufacturer's brand
sumer's attributional sensitivity.l The other aspect of (Rev Ion) should be perceived as more widely desired
the full paradigm is how the consumer's attributional than a private-label manufacturer's brand (Wal-
sensitivity affects his own subsequent behavior. greens).3 Observing the purchase of the Walgreen
Attribution theory thus provides an integrative ap- brand may thus be expected to lead to stronger internal
proach to the study of interpersonal influence. This attributions than the purchase of the Revlon brand.
approach encompasses the concerns of previous in- The second variable operationalizes the discounting
fluence studies. Relationship (2) in Figure A, between principle in terms of the structure of the choice situa-
the observer's attributions and the actor's sensitivity tion observed. A high-choice situation is one in which
to these attributions, can involve overt pressure by the an actor is observed to select from a variety of dif-
observer or the mere presumption of an attribution by ferent alternatives; a low-choice situation is one in
the actor. Relationship (3), between the actor's attribu- which he selects from fewer, more similar alternatives.
tional sensitivity and his subsequent behavior, neces- In the high-choice situation, the actor is less con-
sarily involves the question of how this sensitivity is strained by the alternatives present. The chosen al-
represented psychologically (e.g., normative beliefs?) ternative is likely to be associated with noncommon
and how it affects other determinants of behavior such effects. In the low-choice situation, the alternatives
as attitudes. Though directed more at normative in- are likely to have common effects. The high-choice
fluence, the approach can be generalized to informa- situation implicitly indicates distinctiveness. This logic
tional influence as well: In assessing the adequacy of implies that internal attributions are stronger under
his behavior, the consumer may be sensitive to the high choice than low choice. 4
attributions others make about his competence. 2 It may be argued, however, that an actor's considera-
Let us further consider how people make internal tion of only a few alternatives in a low-choice situa-
attributions from observing consumer behavior-the tion indicates a prior choice, a choice of these
fundamental aspect of the proposed influence para- alternatives from a wider consideration class or evoked
digm. The basic principle emerging from Kelley's and set. Accordingly, both the prior choice of the alterna-
Jones and Davis' work is that a person is more likely tive set and the terminal choice act may provide a basis
to attribute an internal disposition or personal charac- for attributions. This argument is particularly applicable
teristic to an actor when there are no plausible ex- to consumer choices which frequently entail shopping
ternal, alternative explanations for an act. This "dis- decisions as well as purchasing decisions. An actor
counting principle" may be further operationalized in who shops at a store providing a low-choice situation,
terms of two variables which we believe are especially when there are different, alternative stores available,
significant for the consumer context. The first variable has indeed made two choices, in which case an ob-
is suggested by Jones and Davis' hypothesis that the server might not construe the overall situation as one of
strength of an internal attribution varies inversely with low choice. An observer might take into account both
the assumed social desirability of a choice. If most choices in making attributions about the actor.
other people would not have made the choice, the It should be noted that Jones and Davis explicitly
actor is less likely to have been forced into the choice contend that prior choices have no effect on attribu-
by external constraints. In Kelley's terms, the choice tions.
is highly distinctive if most people would not have If we observe that a man leaves his chair, crosses
made it. Applied to the context of consumer behavior, the room, closes the door, and the room becomes
assumed social desirability is closely linked to brand less noisy, a correspondent infe~ence would be that
advertising. One of the major purposes of advertising he intended to cut down the noise. One might ask
whether the inference that the man intended to reach the
1 It is of interest to note that recent work in the social psychology
literature has examined how accurate observers are in inferring the 3 The greater social desirability of the Revlon over the Wal-
dispositions of actors (e.g., Calder, Ross, and Insko 1973), but there green brand name is an assumption of this research. While this
has been no work on the problem posed here-how sensitive actors assumption was suggested by preliminary interviews before the
are in their knowledge of observer attributions. study, we have no pilot data to support it.
2 Note that our approach applies to circumstances which might 4 This logic assumes that the low-choice situation results in no
at first seem removed from the actor-observer attribution paradigm. noncommon effects. According to Jones and Davis' theory, low
Situational advertisements, for example, portray an actor-observer choice would result in a stronger attribution than high choice if
scenario for the consumer. For many products, attributional sensi- there were some noncommon effects but fewer than under high
tivity may stem from vicarious exposure to the attributions mani- choice. While this does not seem the more likely possibility in this
fested by these fictitious observers. Also, attribution theory deals case, and for this reason the high-choice prediction is stated here,
with the actor's observation of his own behavior. Thus, attributional it illustrates the difficulty of developing attribution theory pre-
sensitivity need not depend on observers, real, generalized, or fic- dictions. Note that the high-choice prediction is bolstered by Kel-
titious, but on the actor's own "self-perception" (Bern 1972). ley's (1967) distinctiveness hypothesis.
32 THE JOURNAL OF CONSUMER RESEARCH
door is not also a correspondent inference since "reach- tion of use variable, it is of interest to determine
ing the door" is an effect of crossing the room. But the whether attributions are made in a parallel manner for
subordinate parts of a meaningful action sequence do objectively different products.
not have to be confused with the effects of an action.
In this case, the perceiver is likely to "organize"
the action in his mind as beginning with the decision to METHOD
leave the chair and ending with the closing of the door.
It is the effects of the terminal act in a meaningful se- Subjects
quence, then, that provide the grist for our theory. (Jones One hundred twenty-four female subjects partici-
and Davis 1965, p. 225)
pated in this study. They were students at the U ni-
While the effects of prior choices have not been versity of Illinois at Urbana-Champaign enrolled in
empirically ruled out, most attribution theory research home economics courses. The study was conducted
does support the prediction of stronger internal attribu- during the scheduled sessions of five classes. No com-
tions under high rather than low choice (cf. Calder, munication was allowed between subjects, and sub-
Ross, and Insko 1973). This study tests this prediction jects in each class received all the experimental treat-
in a consumer context in which prior choices may be ments.
more salient than in previous research. If prior choice
is salient, low choice may well yield stronger attribu- Independent Variables
tions than high choice.
The present study thus seeks to demonstrate that The independent variables were manipulated by ask-
people make attributions from observing consumer be- ing subjects to take the role of observers while read-
havior in accordance with the discounting principle. In ing written scenarios portraying a consumer's be-
addition to whether the product chosen was a Walgreen havior. The choice situation varied according to
or Revlon brand and whether the situation was one of whether the consumer chose a brand of a given product
high or low choice, two other variables were included from two similar brands (low choice) or from four,
to explore the generality of the discounting principle. more dissimilar brands (high choice). Also varied was
One variable was the product usage situation. Belk's whether the consumer chose a heavily advertised
(1974) research suggests the importance of different manufacturer's brand (Revlon) or a private-label brand
situations for consumer behavior. He conceives of the (Walgreens). In the low-choice condition, the consumer
situation as everything which exists at a point in time chose Revlon after considering Revlon and Max Factor
that is not a property of the product or the consumer. or chose Walgreens after considering Walgreens and
He defines the situation in terms of observable ag- K-Mart. In the high-choice condition, the consumer
gregate effects that are susceptible to external verifi- chose Revlon after considering Revlon, Max Factor,
cation without reference to any psychological state. Walgreens, and K-Mart or chose Walgreens after con-
This objective approach is employed here to examine sidering all four brands.
how different situations modify attributions made The two variables included for generality were ma-
about a consumer. The two situations investigated dif- nipulated by changing the situation of use and the pro-
fer, at a minimum, in the extent to which the use of duct described in the scenarios. The situation of use
the chosen product is public and involving, or more was either to wear on an evening out with people the
private and less involving. consumer considered important (public and involving
This situational difference seems particularly rele- use) or to keep in the lounge at the consumer's
vant to attribution theory, though existing theories do place of work (more private and less involving use).
not make clear predictions. Certainly, public involve- The product was either mascara or deodorant.
ment is more likely to reflect external constraints,
thereby hindering attributions. Sometimes, however, Procedure
a public ally involving situation of use also conveys These variables were manipulated in a 2 x 2 x 2 x 2
stronger behavioral evidence of commitment to the between-subjects factorial design. Each subject re-
choice, which might lead to a stronger attribution than ceived a description of a typical consumer situation
would private usage. In any event, the two situations
that could well face a young woman similar to herself.
are less important for theoretical prediction than for
The specific description read as follows:
assessing the cross-situational generality of the two dis-
counting variables. For this, it is only required that A young woman about 20 years old is going to
the situations differ in a way that might be expected to college and working part-time in a medium-sized
affect attributions. midwestern city. She works in a small office with
The other variable included for generality was the one other female employee. The woman shops
product itself. Two cosmetic products, mascara and fairly regularly at one of the larger local shopping
deodorant, were employed. They were selected be- centers. There are two women's specialty shops,
cause usage of mascara is more observable and con- a Walgreens, and a K-Mart that she goes to. On
spicuous than usage of deodorant. As with the situa- one such occasion, one of the items she intends
INTERPERSONAL INFLUENCE AND ATIRIBUTIONS 33
to buy is deodorant [mascara]. As she has noted dicted to affect all traits which are perceived as po-
on her shopping list, she wants the deodorant [mas- tential causes of the consumer's behavior. In the ab-
cara] for everyday use. In fact, she would like to sence of any theory about which traits might be rele-
keep it in her compartment in the women's lounge vant as potential causes, it was necessary to select
at work to freshen up occasionally [for a special traits on a representative basis. Null-hypothesis re-
occasion. In fact, she has been invited out to din- sults, as usual, are uninformative: the absence of ex-
ner with several people whom she likes and for perimental effects could be due to inadequacy of the
whom she is especially anxious to be at her best]. discounting principle or to failure to include appropri-
On this particular trip, the woman looks at the ate traits. The existence of the predicted effects for
following brands of deodorant [mascara] displayed the selected traits, however, provides evidence for the
at the cosmetic counters in these stores: Revlon discounting principle.
deodorant [mascara], Max Factor deodorant [mas-
cara], Walgreen deodorant [mascara], and K- RESULTS
Mart deodorant [mascara]. These are the only
brands she considers buying on this trip. Since the personality traits were selected on a repre-
[In the low-choice condition, only two brands were sentative basis, it would be capitalizing on chance to
provided: either Revlon and Max Factor or Wal- analyze the effects of the independent variables for
greens and K-Mart.] each trait separately. It was thus necessary to deter-
After looking at these brands, the woman mine the interrelationships among traits. To do this, a
chooses the Walgreen [Revlon] deodorant [mas- within-cells correlation matrix was computed for the
cara] to keep in the women's lounge at work for twenty-seven personality trait ratings. The correlation
her private use [to wear out to dinner with her
friends]. TABLE 1
Subjects were contacted in the normal classroom FACTOR MATRIX FOR PERSONALITY RATINGSa
environment. After being introduced by the instructor, Orthogonally
the researchers informed the subjects that they would rotated facto rs b
be given a brief questionnaire in which a typical con-
sumer situation would be followed by a series of ques- Variable 2
tions about the person described in that situation.
Rugged/delicate c (.59) -.06
Each questionnaire contained one of the sixteen pos- Unattractive/attractive (.77) .18
sible descriptions. The questionnaires were distributed Low status/high status (.73) -.07
to subjects in each classroom so that descriptions Inferior/superior (.66) .18
were assigned at random. After reading the brief de- Insincere/sincere .22 .41
Unsociable/sociable (.70) .18
scription of the consumer situation, subjects were asked Insecure/secure .41 .37
to evaluate the shopper's personality. They were in- Introvert/extrovert (.68) .14
structed to "think back to the person described on Masculine/feminine (.73) .16
the previous page and try to determine the personality Foolish/wise .45 (.58)
traits she might have." Worthless/valuable .48 (.64)
Unhappy/happy (.69) .50
Awkward/graceful (.73) .39
Dependent Variables Submissive/dominating .37 .49
Unpopular/popular (.70) .50
Consistent with previous attribution research, two Extravagant/economical -.20 (.77)
Immature/mature .30 (.74)
types of dependent variables were assessed-semantic Unsuccessful/successful (.55) (.58)
differential ratings of the consumer on twenty-seven Uninformed/informed .32 (.72)
personality traits and a rating on a ten-point scale of Dull/interesting .50 (.66)
confidence in these personality ratings. An internal at- Conformist/nonconformist -.13 (.65)
tribution, the extent to which the consumer is seen to Cautious/impulsive .41 .04
Critical/tolerant .30 .37
have a given disposition, is indicated by the extremity Frivolous/serious -.15 (.63)
of the trait ratings. Both positive and negative extremes Quiet/talkative .43 .42
on the bipolar adjectives indicate strong internal at- Sloppy/neat (.65) .35
tributions. The confidence ratings measure the overall Ungenerous/generous .49 .48
extent to which observers believe that their attribu- Percent Total Variance 55.7 44.3
tions correspond to the actor's actual internal
dispositions. Higher confidence ratings indicate that , Loadings greater than an absolute value of .50 are shown in parentheses.
b Varimax rotation.
observers are more sure that they learned something e The order of the adjectives represents the order in which they were scored from
about the consumer. one to seven (for some the order was reversed on the questionnaire).
NOTE: These results are based on a principal components/principal axis factor analysis
The personality traits were selected to represent a of the within·cells correlation matrix. The factors are not affected by the experimental
broad spectrum. The two discounting variables are pre- treatment conditions.
34 THE JOURNAL OF CONSUMER RESEARCH
TABLE 2
MEANS FOR PERSONALITY FACTORS, SEPARATE TRAITS, AND CONFIDENCE
Mascara Deodorant
Private Public Private Public Private Public Private Public Private Public Private Public Private Public Private Public
Social Evaluation b 40.89 49.29 59.89 58.13 48.71 56.00 58.13 53.57 51.50 48.12 60.14 58.89 52.63 43.43 58.12 59.57
Personal effectiveness' 31.89 37.43 38.00 10.67 38.86 40.43 34.88 35.57 41.38 38.50 35.71 34.56 44.25 35.14 30.62 34.28
InSincere/sincere 5.44 3.86 4.33 3.00 4.57 5.71 5.25 3.57 6.88 6.00 7.43 5.56 5.12 6.14 2.88 5.29
Cautious/impulsive 3.33 4.14 4.78 4.25 4.57 4.00 4.50 4.29 5.00 3.75 3.86 3.89 5.75 4.86 3.75 4.43
Critical/tolerant 2.89 2.57 4.00 3.38 2.57 2.71 3.38 3.57 2.75 2.25 4.00 3.58 2.25 2.71 3.88 4.14
QuiteJtalkative 3.00 3.57 4.56 4.50 3.14 3.86 4.00 4.14 4.00 2.13 4.29 3.89 4.00 4.14 3.00 4.00
Ungenerous/generous 3.00 3.71 4.22 4.00 3.57 5.00 4.00 5.43 4.25 4.00 4.14 4.44 3.25 3.43 3.88 4.43
Confidence 3.11 4.00 4.44 4.50 3.86 3.14 4.88 4.14 4.12 4.00 4.14 4.67 4.37 3.57 3.12 4.71
n 9 7 9 8 7 7 8 7 8 8 7 9 8 7 8 7
between each pair of elements in such a matrix is ad- "personal effectiveness." Scores on these variables
justed to remove treatment effects. The within-cells were summed to yield two derived dependent vari-
correlation matrix thus reflects the general structure of ables, one for each factor. The remaining analyses were
the subjects' attributions without being affected by the conducted for the summed (raw score) social evalua-
independent variables. To uncover this structure, the tion and personal effectiveness variables and the seven
matrix was submitted to a principal-components analy- separate personality traits not included in either
sis. factor. Table 2 presents the means for all but two of
Two factors clearly emerged from this analysis these variables.
(Table 1). Limiting a factor pattern to those variables An exact least-squares analysis of variance was per-
with more than 25 percent of their variation involved formed on each of the variables in Table 2. As shown
in a pattern (a loading greater than .50), reveals in in Table 3, there was a highly significant main effect
Table 1 that the variables composing Factor 1 might be for Walgreens versus Revlon and a significant product
labeled "social evaluation" and those for Factor 2 by Walgreens-Revlon by pUblic-private use interaction
for the social evaluation dependent variable. In general,
subjects' social evaluation attributions were more posi-
TABLE 3
tive with Revlon than Walgreens. (Note that 44 is the
ANALYSES OF VARIANCE FOR PERSONALITY FACTORS objective midpoint ofthis scale.) The triple interaction,
AND CONFIDENCE however, adds further information and is displayed in
Figure B. For the mascara product, there is a sharper
Social Personal increase in social evaluation from Walgreens to Revlon
Source a evaluation effectiveness Confidence
F's F's F's with private use than with public use. On the other
hand, for the deodorant product, there is a sharper in-
Product (P) <1 <1 4.95 b crease with public use. Subjects infer a more positive
Choice (C) <1 <1 <1 social evaluation from the choice of Revlon, especially
Walgreens-Revlon
(WR) 22.60< 2.84 2.65 for the private use of mascara and the public use
Public-private use of deodorant.
(U) <1 <1 <1 The analysis of variance for the personal effective-
PxC <1 <1 4.65 b ness variable revealed a significant product by Wal-
PxWR <1 4.46 b <1 greens-Revlon interaction. The form of this interaction
PxU 2.28 2.52 1.08
C xWR 1.26 2.56 <1 is quite simple (see Figure C): There is no difference
CxU <1 <1 4.13 b between Walgreens and Revlon for mascara but a large
WR xU <1 <1 <1 difference for deodorant. Attributions of personal ef-
P x C x WR 2.35 <1 <1 fectiveness lie at the objective midpoint of the scale for
PXCxU <1 <1 <1
PxWRxU 4.12b 2.61 <1 mascara. For Walgreens deodorant, however, personal
CxWRxU <1 <1 <1 effectiveness is rated positively. For the Revlon de-
P x C x WR xU <1 <1 1.75 odorant, it is rated negatively. Apparently, subjects felt
the consumer to be wiser, more informed, etc., for
• Each source has one degree of freedom.
b Significant at .05 level.
buying the Walgreen brand deodorant and not so wise
, Significant at .01 level. for buying the Revlon brand deodorant, while the brand
INTERPERSONAL INFLUENCE AND ATTRIBUTIONS 35
FIGURE B
MEAN SOCIAL-EVALUATION SCORES FOR THE PRODUCT BY 32
WALGREENS-REVLON BY PRIVATE-PUBLIC SITUATION OF
USE INTERACTION
Social Evaluation
60
MASCARA DEODORANT
30~------~------------------~------~
58 Walgreens Revfon
56
three significant effects for the quietltalkative variable,
a Walgreens versus Revlon main effect (F = 4.92,
p < .029), a private-public use main effect (F = 4.38,p
54 < .039), and a product by choice interaction (F = 5.86,
p < .017). Subjects inferred that the consumer was
52 more talkative about the Revlon brand than the Wal-
green brand and also more talkative about public than
50 private use. The former result fits the greater social
evaluation ratings under Revlon and, in fact, may be
implied by this evaluation. The latter effect probably
48 reflects the import of the woman going out to din-
ner. The interaction is more intriguing. Under low
46 choice, the deodorant product revealed a stronger at-
tribution of talkativeness but under high choice the
44 mascara was stronger.
There was a main effect for Walgreens versus Revlon
Wolgreens Revlon Walgreens Revlon on the cautious/impulsive variable (F = 17.61, p
36 THE JOURNAL OF CONSUMER RESEARCH
results, the choice of a Revlon brand may have indi- brands engender stronger internal attributions, the
cated to the observers that the purchase was important marketers of such brands would want to increase their
to the consumer. The greater social desirability of the attention to interpersonal influence in their advertising
Revlon brand implies that the consumer choosing it strategies. They would particularly want to counteract
must be more concerned about the product than one negative internal attributions such as the personal ef-
who buys the Walgreen brand. The significance of the fectiveness attribution associated in this study with the
Revlon choice would explain the stronger internal at- use of Revlon deodorant.
tributions with Revlon than Walgreens. The present results are also useful in suggesting the
Further evidence regarding the discounting principle need for extending existing attribution theory. Al-
comes from the choice variable. Choice affected the though attribution theory is presently one of the most
global confidence ratings rather than, with one excep- active areas of social psychological thinking, almost
tion, the specific trait attributions. Most interesting is all of this work is guided by the discounting
that the effects of choice depended on both the situa- principle. Our results, however, indicate that the dis-
tion of use and the product. Confidence was not always counting principle is not sufficient to explain attribu-
higher in the high-choice situation. This result may tions made in a consumer context. It appears neces-
reflect the prior choice problem discussed earlier. With sary, at a minimum, to consider the importance
the private-use situation and the deodorant product, the observers believe consumers attach to a choice (the sig-
purchase might have seemed more routinized to the nificance of the choice) and the assumptions observers
observers under low than high choice, indicating a his- make about prior choices.
tory of prior choices. Observers may have interpreted Attribution theorists have, in general, neglected
the low choice-private use and low choice-deodorant people's expectations about the meaning of observed
conditions as actually reflecting a higher degree of behaviors. In addition to discounting, it is necessary
choice than the corresponding high-choice conditions. to consider the role of "typicality" in making attribu-
This would explain the stronger internal attributions tions (Calder 1974a, 1974b, 1977). Typicality refers to
under low choice. the observer's intuitive beliefs about the internal char-
The results involving the situation of use and product acteristics that are usually associated with a given be-
variables confirm the desirability of investigating at- havior. To make an internal attribution, apart from
tributions in different settings. Both of these variables discounting situational causes of the behavior, an ob-
affected the degree to which Revlon yielded stronger server must believe that certain characteristics are
attributions than Walgreens. Evidently, the situation of typical of the behavior. The choice of Revlon must
use can have either of the effects postulated: with mean something to an observer. We have suggested
mascara, public use may have indicated external con- that it means that the particular product is important
straints while, with deodorant, public use may have to the consumer. However, the choice of Revlon may
indicated a stronger behavioral commitment. As dis- imply other characteristics which are thought to be
cussed previously, these variables may also have al- typical of Revlon users as well.
tered observers' interpretation of the choice variable. Just as it is plausible that choosing Walgreens may
The present results are useful in two ways. For one, mean that the particular product is not important to
they attest to the need for an attribution approach to the consumer, it may be that the choice of Walgreens
interpersonal influence. Most interesting in this regard means less in general. Observers may have weaker
is the finding that the Revlon brand implied a socially beliefs about the characteristics typical of choosing
positive but not very competent consumer. Suppose Walgreens than about those typical of choosing Revlon.
that our observers were linked to our hypothetical con- Although nonpersonal influences can be discounted
sumer through actual role relationships. Their influence more easily with the choice of Walgreens, this may be
would depend on the consumer's attributional sensi- outweighed by the stronger beliefs about the char-
tivity (Figure A). Assuming that the observers' behavior acteristics typical of the Revlon choice. Thus, the pres-
reflected their attributions, if the consumer were most ent findings of stronger attributions with Revlon than
sensitive to social evaluation attributions, the ob- Walgreens may depend more on typicality than dis-
servers' influence would be to support the Revlon counting.
choice. But, if the consumer were most sensitive to In developing an attribution approach to inter-
personal effectiveness attributions, the observers' personal influence, it will be necessary to clarify the
influence would be to undermine the Revlon deodorant roles of discounting and typicality. While both are prob-
choice. Interpersonal influence is only to be accounted ably important in the attribution process, for consumer
for in terms of the psychological processes which behavior, the question boils down to which kind of
underlie it. choices yield stronger attributions: choices for which
This study illustrates the importance of under- nonpersonal causes can be discounted or choices for
standing the variables which affect observers' internal which people have strong beliefs about the internal
attributions and the consumer's sensitivity to them. If characteristics typical of the choice?
subsequent research confirms that heavily advertised [Received January 1976. Revised March 1977.]
38 THE JOURNAL OF CONSUMER RESEARCH