ArticlePDF Available

Sometimes IT Just Feels Right: The Differential Weighting of Affect-Consistent and Affect-Inconsistent Product Information

June 2001
Journal of Consumer Research 28(1):1-17

June 2001
28(1):1-17

DOI:10.1086/321944

Source
RePEc

Authors:

Rashmi Adaval

University of Cincinnati

An affect-confirmation process is proposed to explain the conditions in which information that is similar in valence (i.e., evaluatively consistent) with a person's mood is weighted more heavily in product judgments. Specifically, the affect that participants experience as a result of a transitory mood state may appear to either confirm or disconfirm their reactions to product information, leading them to give this information more or less weight when evaluating the product as a whole. This affective confirmation typically occurs when hedonic criteria are considered more important in evaluation than utilitarian criteria. Four experiments confirmed implications of this conceptualization. Copyright 2001 by the University of Chicago.

Content uploaded by Rashmi Adaval

Content may be subject to copyright.

䉷2001 by JOURNAL CONSUMER RESEARCH, Inc. ●Vo l. 28 ●June 2001

Sometimes It Just Feels Right: The Differential

Weighting of Affect-Consistent and Affect-

Inconsistent Product Information

RASHMI ADAVAL*

An affect-conﬁrmation process is proposed to explain the conditions in which in-

formation that is similar in valence (i.e., evaluatively consistent) with a person’s

mood is weighted more heavily in product judgments. Speciﬁcally, the affect that

participants experience as a result of a transitory mood state may appear to either

conﬁrm or disconﬁrm their reactions to product information, leading them to give

this information more or less weight when evaluating the product as a whole. This

affective conﬁrmation typically occurs when hedonic criteria are consideredmore

important in evaluation than utilitarian criteria. Four experiments conﬁrmed impli-

cations of this conceptualization.

The affect that people experience at the time they receive

information about an object can inﬂuence the manner

in which they process this information and, therefore, the

judgments that follow. This inﬂuence has been detected in

both consumer judgment and elsewhere (for reviews, see

Clore, Schwarz, and Conway [1994]; Cohen and Areni

[1991]). In some cases, affect has been shown to have a

direct effect on people’s judgments. That is, people judge

a product more favorably when they are feeling happy than

when they are not, regardless of the information theyreceive

about it (Gorn, Goldberg, and Basu 1993; Pham 1998).

However, affect can also have an indirect inﬂuence on judg-

ments through its impact on the way speciﬁc pieces of prod-

uct information are treated. For example, consumers might

give a piece of information more weight when its evaluative

implications are similar in valence to the affect they are

experiencing for other reasons. (That is, happy individuals

may weight favorable pieces of information more heavily

than unfavorable pieces when making a judgment, whereas

unhappy persons may give relatively more emphasis to un-

*Rashmi Adaval is an assistant professor at the Hong Kong University

of Science and Technology, Clear Water Bay, Kowloon, Hong Kong

(mkadaval@ust.hk). Part of the research was performed for a Ph.D. dis-

sertation under the guidance of Kent B. Monroe and was funded by the

J. M. Jones Chair at the University of Illinois. The author wishes to thank

Bob Wyer for his exceptionally insightful comments, Sharon Shavitt for

providing the impetus for this research, and members of her committee

(Gerald L. Clore and Seymour Sudman) for their feedback. Gerald Gorn,

Simona Romani, and Jaideep Sengupta gave comments on previous ver-

sions of the article. Edward R. Hirt provided the ﬁlms for use as stimulus

materials. All the above contributions, as well as suggestions made by the

editor and reviewers, were invaluable in the research process and are grate-

fully acknowledged.

favorable pieces than to favorable ones.) The present article

provides evidence of this differential weighting and circum-

scribes the conditions in which it occurs. In doing so, it

provides an explanation that differs from other, memory-

based explanations that have been previously proposed

(Bower 1981; Forgas 1995; Isen et al. 1978).

To understand more clearly the phenomena of concern in

this article, imagine a typical shopping situation in which

consumers evaluate a product. Several different types of

product information may be available in this situation, each

of which may have different implications for the evaluation

of the product as a whole. For instance, when consumers

are making on-line evaluations of jeans, they could respond

favorably to a brand name like Levis but unfavorably to the

speciﬁc style or color of the jeans. These responses could

include affective reactions. Moreover, some of these reac-

tions might be consistent (in valence) with the extraneous

affect that consumers happen to be experiencing at the time

they receive the product information (e.g., their mood)

whereas other reactions might be inconsistent with it. This

extraneous affect, which could have occurred for reasons

that are unrelated to the product (e.g., bad trafﬁc conditions),

might nevertheless inﬂuence the relative weight that the

consumers attach to product information that is evaluatively

consistent or inconsistent with it.

Although affect-induced differences in weighting have

not been documented in the product domain, at least two

studies suggest that this phenomenon occurs in other judg-

ment domains (see Bower, Gilligan, and Monteiro 1981;

Forgas and Bower 1987). Previous explanations of this phe-

nomenon, however, have relied on the assumption that mood

inﬂuences the accessibility of mood-consistent evaluative

concepts in memory and that these concepts are used to

2JOURNAL OF CONSUMER RESEARCH

FIGURE 1

ALTERNATIVE PROCESSES MEDIATING THE INFLUENCE OF AFFECT AND PRODUCT INFORMATION ON JUDGMENTS

OTE

.—Path Adescribes memory-based processes. Paths Band Cdescribe processes implied by the affect-conﬁrmation hypothesis. Path Ddescribes the direct

use of affect as information in judging the product. Shaded boxes indicate processes in which affect is theoretically involved.

selectively encode and retrieve information (Bower 1981;

Forgas 1995). This assumption has recently been called into

question (Niedenthal, Halberstadt, and Setterlund 1997; Nie-

denthal and Setterlund 1994; see also Wyer,Clore, and Isbell

1999), highlighting the need for alternative explanations. In

the present research, an affect-conﬁrmation process is pro-

posed to account for the phenomenon. Unlike other theories

(e.g., Bower 1981; Forgas 1995), this conceptualization re-

quires no assumptions about the impact of affectivereactions

on the accessibility of previously acquired concepts and

knowledge in memory. The theory therefore applies to con-

ditions in which people base their judgments primarily on

the information they have available to them at the timerather

than previously acquired information that they recall.At the

same time, the conceptualization circumscribes more clearly

than previous formulations the conditions in which people’s

affective reactions are likely to inﬂuence the impact of the

information they receive.

THEORETICAL BACKGROUND

The different processes that underlie the impact of affect

on judgments are outlined in Figure 1. After reviewing some

of these processes and discussing their underlying assump-

tions, an alternative account of the differential weighting of

affect-related information is proposed.

Memory-Based Theories of Affect and Judgment

Theories of affect and cognition (e.g., Bower 1981; For-

gas 1995; Isen et al. 1978) often assume that the affect

people experience at a particular moment in time increases

the accessibility in memory of concepts and knowledge that

are evaluatively consistent with it (Bower 1981). When a

large amount of information is presented, a portion of the

information that can be easily interpreted in terms of mood-

activated concepts is thought about more extensively and

may be selectively encoded into memory (Bower et al. 1981;

AFFECT CONFIRMATION 3

Forgas and Bower 1987). This selectively encoded infor-

mation is more likely to be recalled later and used as a basis

for judgment. The process outlined above is shown in Figure

1, path A. It suggests that the differential weighting of affect-

consistent and affect-inconsistent information at the time of

judgment is the result of differences in the likelihood that

the information is selectively encoded at the time it is re-

ceived and therefore retrieved from memory later on.

However, this interpretation assumes that mood inﬂuences

the accessibility of similarly valenced concepts and knowl-

edge in memory. As Wyer et al. (1999) point out, the as-

sumption that mood inﬂuences the accessibility of similarly

valenced concepts and knowledge in memory is question-

able. Niedenthal and her colleagues (Niedenthal and Set-

terlund 1994; Niedenthal et al. 1997) found that although

the mood that people happen to be experiencing increases

the accessibility in memory of concepts that pertain directly

to this mood, it does not inﬂuence the accessibility of ev-

aluatively congruent concepts in general. Nor does thinking

about mood-congruent information about a referent remind

persons about other characteristics of the referent (Gilligan

and Bower 1983). Thus, a fundamental assumption under-

lying memory-based theories of the impact of mood on

judgment may not be valid.

Affect-Conﬁrmation Hypothesis

The proposed conceptualization recognizes that people

often do not distinguish between the different sources of

their affect at the time they are called upon to make a judg-

ment (Schwarz and Clore 1983). Consequently, the affect

they experience for reasons that have little to do with a

product they are judging (e.g., a transitory mood state) can

be confused with feelings elicited by the product informa-

tion. When extraneous affect is similar in valence to one’s

affective reactions to this information, it can make these

reactions appear more appropriate or valid and, therefore,

can increase the perception that one’s feelings about the

information have been assessed correctly. As a result, it can

increase the weight or importance that people give to this

information in the course of computing a judgment.

For example, suppose consumers wish to purchase ice

cream. They may think about the attribute “made from full

creamy milk” in terms of its implications for how good the

ice cream would taste. These thoughts might give rise to

affective reactions. However, people may not clearly distin-

guish these reactions from the affect they are experiencing

for other, unrelated reasons. Consequently, the extraneous

affect can inﬂuence their perceptions of how they feel about

the product information. When the affect from the two

sources is similar in valence, consumers may perceive that

they have assessed their feelings about the product infor-

mation correctly. They may therefore weight this informa-

tion more heavily in estimating their liking for the product

and may have greater conﬁdence in this judgment.However,

if the extraneous affect that consumers are experiencing dif-

fers from that elicited by the product information, they may

interpret these conﬂicting feelings as ambivalence about

their reactions to this information, and so they may give

less weight to the information than they otherwise would.

The processes described above assume that people base

their evaluations of the product information on their affec-

tive reactions to it. However, this is not always the case.

That is, some evaluations may be based on affect but others

may not (Zanna and Rempel 1988; see also Pham 1998).

Mano and Oliver (1993) distinguish between two general

criteria that consumers use (see also Batra and Ahtola 1990).

Utilitarian criteria concern the product’s ability to perform

a useful function, whereas hedonic criteria focus on the

feelings that consumers expect to experience when the prod-

uct is used. When persons are not explicitly told what criteria

they should apply in evaluating a product, the criteria they

use may depend on their personal values, goals, or needs

(Babin, Darden, and Grifﬁn 1994). For example, a person

who wants to eat ice cream may evaluate it on the basis of

affective or hedonic criteria (i.e., its taste and the enjoyment

it will give). However, a hospital dietician may evaluate the

product on the basis of utilitarian criteria (e.g., its value to

postsurgery patients on high calorie diets).

Pham (1998) notes that the affect consumers experience

only inﬂuences their judgments of a product when they con-

sider hedonic criteria to be a relevant basis for evaluating

it. Similarly, affect-conﬁrmation processes are also unlikely

to occur unless consumers base their evaluations on affective

criteria. Thus, the dietician in the previous example might

evaluate the attribute “made from full creamy milk” favor-

ably because of its high caloric content. On the other hand,

she might consider its implications for hedonic criteria (e.g.,

taste) to be irrelevant and might actively try to avoid using

them. To this extent, extraneous affect might not inﬂuence

the weight she attaches to the information in evaluating the

product.

The processes assumed by this conceptualization are con-

veyed in paths Band Cof the ﬂow diagram shown in Figure

1. Suppose people receiving information about a product

consider utilitarian criteria to be most relevant in evaluating

it (path B). Then they should construe the implications of

this information without considering their affectivereactions

to it and integrate these implications to arrive at an overall

evaluation. If, however, people consider the hedonic (affec-

tive) criteria to be most relevant, they may assess their af-

fective reactions to each piece of information. In doing so,

they may consider their reactions to a given piece of in-

formation to be relatively more reliable if these reactions

are similar to the feelings they happen to be experiencing

for other reasons. Consequently, they may assign relatively

greater weight to this piece of information when they com-

pute an overall evaluation. These processes are captured by

the following hypotheses.

H1: Participants who base their judgments on hedonic

criteria will give greater weight to attribute infor-

mation when this information is evaluatively con-

sistent with their mood than when it is inconsistent

4JOURNAL OF CONSUMER RESEARCH

with their mood. This differential weighting will not

be evident when participants base their judgments

on utilitarian criteria.

A second hypothesis formalizes an assumption thatunderlies

Hypothesis 1:

H2: Participants will have more conﬁdence in their judg-

ments when the affective reactions to a product’s

attributes and the affect they are experiencing for

other reasons are similar in valence than when they

differ.

Alternative Conceptualizations

The affect-conﬁrmation hypothesis does not preclude

other possible inﬂuences of affect on information processing

and judgment. However, these inﬂuences theoretically occur

independently of the impact of affect on the weighting of

information that is evaluatively consistent or inconsistent

with it. The distinction between these formulations is worth

noting brieﬂy.

Affect as Information. People often base their judg-

ments of an object on the affective reactions they are ex-

periencing at the time of judgment (Schwarz and Clore

1983). Although these affective reactions may have been

caused by external events, they might be interpreted as feel-

ings about the object being evaluated. As a result, persons

might evaluate an object more favorably when they are

happy than when they are not, and this could be true re-

gardless of whether the speciﬁc information about the object

is favorable or unfavorable. The use of affect as information

is quite well established (for reviews, see Clore et al. [1994];

Schwarz and Clore [1996]). There may nevertheless be con-

tingencies in the conditions in which this occurs. Forgas

(1995) suggests that affect is used as a heuristic basis for

judgments when people are unable or unmotivated to think

carefully about other information available (see also Shiv

and Fedorikhin 1999). Pham (1998), for example, found that

people used their feelings as a direct basis for judging a

product when they were told to make a judgment to which

affective reactions were relevant. When participants’ judg-

ments required a consideration of utilitarian criteria, how-

ever, the affect they experienced had no impact on their

judgments.

Pham’s (1998) ﬁndings are consistent with the assumption

that people’s affective reactions enter into information pro-

cessing only if the judgment to be made is one to which

these reactions are relevant. However, Pham’s research was

concerned with the direct inﬂuence of affect as a basis for

judgment (reﬂected in Fig. 1, path D). If this were the only

process by which affect inﬂuences judgments, it would im-

ply that a piece of information has the same impact under

positive and negative affect conditions. The affect-conﬁr-

mation hypothesis, however, suggests that a favorable piece

of information will have more inﬂuence and unfavorable

information will have less inﬂuence when consumers feel

happy than when they do not. In principle, both types of

inﬂuences could occur. That is, the use of affect as infor-

mation is presumably independent of the inﬂuence of affect

on the way speciﬁc pieces of information are evaluated and,

therefore, the weight attached to this information (Fig. 1,

path C).

Martin et al. (1997) propose a somewhat different version

of the affect-as-information hypothesis. They ﬁnd that persons

evaluate an object more favorably if it performs the function

for which it is intended. The affect being experienced is used

as input in determining whether this is true. (Thus, they eval-

uate a funny ﬁlm more favorably if they are happy and a sad

ﬁlm more favorably when they are unhappy.) If one assumes

that products are typically expected to perform well and make

people feel good, then Martin et al.’s (1997) conception pre-

dicts that people will evaluate all products more favorably if

they are happy than if they are not—a prediction similar to

that of Pham (1998). Both these processes are presumably

independent of those investigated in the present article and

do not make any predictions about the weighting of individual

pieces of information.

Motivational Inﬂuences. Motivation can inﬂuence in-

formation processing in at least three ways. First, Chaiken

(1987) hypothesizes that when people are asked to make a

judgment, they ﬁrst apply the criterion that comes to mind

most easily and is simplest to apply. They then assess their

conﬁdence that the judgment based on this criterion is valid,

and if this estimate is above some minimal threshold, they

apply it without engaging in further processing. If their con-

ﬁdence is below this threshold value, however, they identify

and apply additional criteria that may require more extensive

processing. According to the affect-conﬁrmation hypothesis,

people’s conﬁdence in their judgments is more likely to be

above threshold when the affect elicited by a piece of prod-

uct information is consistent with their mood. Consequently,

they may be more inclined to weight this information heavily

in making a judgment without considering carefully other

information available. To this extent, predictions of the af-

fect-conﬁrmation hypothesis are compatible with Chaiken’s

conceptualization. However, Chaiken (1987) does not pre-

dict any contingencies in the occurrence of this effect,

whereas the affect-conﬁrmation hypothesis does.

A second possibility is that the affect being experienced

may itself inﬂuence the motivation to process information

in general (Bless et al. 1990; Bodenhausen 1993). In this

case, it could inﬂuence the extent to which affect enters into

all the three paths shown in Figure 1 (namely, paths A, C,

and D). Finally, Isen (1984) and others (Singer and Salovey

1988) suggest that happy people want to maintain their feel-

ings, whereas unhappy people are motivated to change their

feelings. These motives could lead both happy and unhappy

participants to think more extensively about positive aspects

of the information they receive than about negative aspects.

This type of motivated thinking could produce a generally

greater inﬂuence of favorable information than unfavorable

information. In contrast, the affect-conﬁrmation hypothesis

AFFECT CONFIRMATION 5

suggests that only individuals who experience positive affect

will weight favorable information more heavily. Note, how-

ever, that the effects implied by affect conﬁrmation and the

effects of motivation are not incompatible. That is, the in-

ﬂuence of affect-conﬁrmation processes could occur over

and above the motivational effects suggested by both Isen

(1984) and Bless et al. (1990).

An Assessment of Weighting: Methodological

Considerations

Although Hypothesis 1 can be tested directly by assessing

the impact of the attribute information, it is not an unam-

biguous test of the hypothesis. Anderson (1981) and others

(e.g., Birnbaum 1974) have pointed out that the inﬂuence

of a given type of information (i.e., the difference in judg-

ments when the information is favorable and judgments

when it is unfavorable) is governed by perceptions of the

extremity of its evaluative implications as well as the weight

that is attached to these implications. (That is, individuals

might perceive the implications of favorable attribute in-

formation to be more extremely favorable, and the impli-

cations of unfavorable information to be more extremely

unfavorable, independently of the weight they attached to

the information when computing the judgment.) To this ex-

tent, an inference of the weight attached to attribute infor-

mation from the magnitude of its inﬂuence on judgments

would be equivocal. For example, suppose a product de-

scribed by a favorable attribute and a product described by

an unfavorable attribute are evaluated ⫹5 and ⫺5, respec-

tively, when the attribute information was evaluatively con-

sistent with participants’ mood, but are evaluated ⫹3 and

⫺3, respectively, when the attribute information was in-

consistent with participants’ mood. This could indicate that

participants attached more weight to the attribute informa-

tion when it was consistent with their mood than when it

was not. However, it could also indicate that participants

interpreted the attribute as having more extreme evaluative

implications in the ﬁrst case than in the second.

To eliminate this confound, an indirect procedure was

used to detect differential weighting. Numerous studies both

in the area of consumer judgment (e.g., Bettman, Capon,

and Lutz 1975; Meyer 1981; Troutman and Shanteau 1976)

and elsewhere (for a review of evidence, see Anderson

[1981]; Birnbaum [1974]) have shown that the greater the

weight that participants attach to one piece of information

in making a judgment, the less inﬂuence the information

accompanying it will have. Therefore, the relative weight

attached to one piece of information can be indirectly in-

ferred from the weight attached to the other. This procedure

eliminates confounds produced by the extremity of the ev-

aluative implications of the information itself (see Wyer and

Carlston 1979).

Participants in three of the four experiments to be reported

were given only two pieces of information about each of the

product categories they evaluated. Each product was de-

scribed by a favorable or an unfavorable brand name and by

a favorable or an unfavorable speciﬁc attribute. Under con-

ditions in which participants attach relatively moreimportance

to affective criteria than to utilitarian criteria, they shouldgive

relatively more weight to mood-consistent attributes than

mood-inconsistent ones. (That is, happy persons should attach

greater weight to the favorable attribute information, whereas

unhappy participants should give greater weight to unfavor-

able attribute information.) To this extent, the effect of the

information accompanying it (i.e., brand name) should be less

in the former conditions than the latter. Under conditions in

which participants attach relatively more importance to util-

itarian criteria, however, this difference should not be evident.

Experiment 1 assessed this prediction by varying the cat-

egory of products that participants judged. Experiment 2

replicated the ﬁndings by experimentally manipulating par-

ticipants’ evaluation criteria. Additional experiments sup-

ported the assumption that participants weight affect-con-

ﬁrming information more in judgments but that the inﬂuence

of extraneous affect on the weight attached to information

decreases when the source of this affect is called to one’s

attention.

EXPERIMENT 1

According to Hypothesis 1, information that is evalua-

tively consistent with a person’s mood state will be given

relatively greater weight in judgments if affective reactions

are typically used to evaluate the products. However, this

will not be the case if the products are evaluated primarily

on the basis of utilitarian considerations. To test this hy-

pothesis and to determine if this phenomenon generalized

to the large number of products available in the marketplace,

experiment 1 considered different product categories that

varied in the criteria people used to evaluate them.

Assessment of Judgmental Criteria—Preliminary

Data

Four product categories (jeans, running shoes, sweatshirts,

and backpacks) were selected on the basis of focus group

discussions involving undergraduates. The importance of

hedonic and utilitarian criteria in evaluating the types of

products under consideration was determined on the basis

of additional data from a different group of 80 male and 95

female college participants. These participants were asked

to consider a product in one of the four product categories

selected and to indicate the extent to which they were likely

to think about a number of different features when pur-

chasing it. One feature, “how the product feels when it is

used,” exempliﬁed affective, feelings-related criteria, and a

second, “construction quality,” exempliﬁed utilitarian cri-

teria. These features were embedded in a number of others

in the questionnaire in order to disguise the speciﬁc variables

that were of interest. Judgments were reported along a scale

from 0 (not at all) to 10 (a lot).

Both affective and utilitarian criteria were generally con-

sidered important regardless of the product category in-

volved (in all cases, ). However, the relative im-M16.7

6JOURNAL OF CONSUMER RESEARCH

portance attached to these criteria varied over categories and

in some cases differed between males and females. Specif-

ically, both males and females considered affective criteria

more important than utilitarian criteria in judging jeans

(mean and 1.0 for males and females, re-difference p1.3

spectively). This was also true in the case of running shoes

( and 1.3, for males and females, respectively).Mp0.6

diff

In contrast, although females attached more importance to

affective criteria than to utilitarian criteria in judging sweat-

shirts ( ), males attached relatively less impor-Mp0.5

diff

tance to affective criteria ( ). Finally, both malesMp⫺1.0

diff

and females attached less importance to affective criteria

than to utilitarian criteria in judging backpacks (Mp

diff

vs. ⫺2.1 for males and females, respectively). These⫺1.4

conclusions were conﬁrmed by an analysis of these differ-

ence scores, which yielded an overall effect of product cat-

egory, and an in-

F(1, 167) p13.84, p!.01, hp0.20,

teraction of product category and participant sex,

F(3, 167) p2.33, p!.08, hp0.04.

Predictions

Based on these data, speciﬁc predictions were made for

each product category. That is, both males and females

should weight mood-consistent information more heavily

than mood-inconsistent information when they evaluated

jeans and running shoes. Females but not males should

weight mood-consistent information more heavily than

mood-inconsistent information when they evaluated sweat-

shirts. Finally, neither males nor females should weight

mood-consistent information more heavily when they eval-

uated backpacks.

Method

Overview, Subjects, and Design. Participantswho had

been induced to feel either happy or sad by watching a movie

were asked to estimate how much they liked each of four

products. The products, selected on the basis of pretesting,

varied in terms of the type of criteria (hedonic vs. utilitarian)

that were used to evaluate them. The description of each

product considered by a given subject represented a different

combination of brand favorableness, attributefavorableness,

and product category. The particular combination to which

participants were exposed was varied in a latin-squaredesign

in a manner described later. Eighty-four male and 76 female

undergraduate business majors participated in the study for

extra course credit.

Selection of Brand and Attribute Information. This

information was selected on the basis of a pretest in which

16 male and 15 female undergraduate business majors as-

sessed the favorableness of different brand names and at-

tributes along a scale from ⫺5 (very undesirable) to ⫹5

(very desirable). The desirable and undesirable brands were

as follows: for jeans, Levis versus Rustler (3.54 vs. ⫺3.11,

respectively); for running shoes, Reebok versus Trax (3.90

vs. ⫺1.26); for sweatshirts, Champion versus NuBlends

(4.00 vs. 0.45); and for backpacks, Jansport versus Strike

Force (4.03 vs. 0.32). Although the undesirable brands in

the latter two product categories were not clearly unfavor-

able, they were substantially less favorable than the corre-

sponding desirable brands.

The favorable attributes selected for jeans, running shoes,

sweatshirts and backpacks, respectively, were “preshrunk”

(mean ), “soft and ﬂexible sole” (desirability p3.50 Mp

), “80 percent cotton–20 percent polyester” (3.53 Mp

), and “large number of pockets” ( ). Unde-4.23 Mp3.26

sirable attributes were deﬁned as the absence of the attribute

considered desirable or a lower amount of this attribute.

Thus, the undesirable attributes associated with the four

product categories were not preshrunk, hard, inﬂexible sole,

20 percent cotton–80 percent polyester, and one pocket,

respectively.

Presentation of Stimulus Information. Each partici-

pant evaluated one product from each of the four categories.

To ensure against a confounding of the type of information

presented and presentation order, eight different question-

naire forms were constructed. In four forms, jeans were

evaluated ﬁrst, followed by sweatshirts, backpacks, and run-

ning shoes. In the remaining four forms, the order of the

product categories was reversed. The four product descrip-

tions in each form represented a different combination of

brand favorableness and attribute favorableness. These de-

scriptions varied over forms and product categories in a

latin-square design so that pooled over forms (a) all com-

binations of brand favorableness and attribute favorableness

were represented within each product category, and (b)the

mean serial position of each product category, as well as

the mean serial position of each combination of brand fa-

vorableness and attribute favorableness, was the same. An

approximately equal number of male and female participants

in each mood-induction condition completed each form.

Procedure. Participants were told that the departments

of business administration and psychology had jointly spon-

sored the study. They were informed that the two departments

wished to collect information about a number of different

aspects of students’ lives, and so they would be asked to

provide information on a variety of topics (e.g., the classes

they were taking, the forms of entertainment they enjoyed,

product purchases they made, life experiences they had while

they were students, how they felt on a daily basis, etc.). They

were also told that different groups of students were being

asked different sets of questions and that not everyone was

asked questions on all topics because of time constraints.

They were then given the mood-induction task. The

mood-induction procedure was similar to that employed by

Martin et al. (1993). Participants were told that the ﬁrst task

was being conducted to determine student preferences for

different genres of ﬁlm. On this pretense, they were asked

to watch either (a) two amusing ﬁlm clips from Pretty

Woman and Mrs. Doubtﬁre or (b) two sad ﬁlm clips taken

from Ordinary People and Sophie’s Choice. Each clip was

seven minutes long. Both sets of ﬁlms were preceded by

AFFECT CONFIRMATION 7

another clip from the ﬁlm Bullit, which was included to

mask the emotional tone of the other ﬁlms. To justify the

rationale for showing these clips, participants were askedto

react to each ﬁlm by answering questions about them on a

Pilot Movie Ratings questionnaire. The questions asked

were general and dealt with issues such as knowledge of

the movie’s title, the quality of acting, enjoyment of ﬁlms

of that genre, and whether it was possible to tell the plot of

the movie from the brief clip. After they had evaluated the

three ﬁlms, they were told to move on to the next task.

Participants were told that the next task they would perform

was concerned with product preferences. They were then

given one of the eight versions of the questionnaire described

earlier. In each case, they were asked to read the product

description and then to evaluate the product along threescales,

ranging from ⫺5to⫹5, pertaining to attractiveness (very

unattractive/very attractive), goodness (extremely bad/ex-

tremely good), and desirability (very undesirable/very desir-

able). Responses along the three scales were subsequently

averaged to provide a single estimate of liking for each prod-

uct (Cronbach’s alpha was greater than .91 for each combi-

nation of brand and attribute favorableness).

Participants at the end of the experiment were given a

manipulation check questionnaire. They were asked to report

the extent to which they experienced eight speciﬁc emotions

(happy, angry, pleasant, sad, delighted, glad, unpleasant,and

distressed) while engaging in the mood-inducing activity.

These ratings were made along scales from 0 (not at all) to

10 (extremely) and taken from Penner et al. (1994). Re-

sponses to these scales were averaged, after appropriate re-

verse scoring, to provide a single index of participants’ mood

at the time it was induced (Cronbach’s alpha was .97). In

addition, they were asked to indicate their feelings “at this

moment” along the two scales, one from ⫺5 (extremely

unhappy) to ⫹5 (extremely happy) and the other from ⫺5

(extremely bad) to ⫹5 (extremely good). These responses

were averaged to provide a single index of participants’

mood at the time they were asked to report it (current mood).

After they completed this task, they were thanked for their

participation and debriefed.

Results

Manipulation Check. Participants recalled being in a

happier mood while watching movies that were intended to

induce positive affect than while watching movies intended

to induce negative affect (6.81 vs. 2.41), F(1, 136) p

In addition, participants’ esti-

468.08, p!.00, hp.775.

mates of their mood at the end of the experiment were also

more positive in the former condition than the latter (1.94

vs. 1.36), Thus, the

F(1, 158) p3.78, p!.05, hp.023.

effect of the mood induction persisted throughout the

experiment.

The effectiveness of this technique and the procedure used in experiment

2 was pretested. Participants reported being happier at the time of the mood

induction if the induction was positive than if it were negative (8.20 vs.

2.74), and, at the end of the experi-

F(1, 24) p84.42, p!.01, hp.78,

ment (2.88 vs. 0.18), This difference

F(1, 24) p18.77, p!.01, hp.44.

Overall Analyses. The inﬂuence of mood on the weight

attached to product attribute information was inferred from

the difference between the effect of brand name when the

attribute information was evaluatively consistent with par-

ticipants’ mood and its effect when the attributeinformation

was inconsistent with their mood. If attribute information is

given more weight in the former condition than the latter,

this effect of brand name should be less in the former

conditions.

To obtain an overall indication of the effect of mood on

the weight of the attribute information, data were pooled

over the four product categories and analyzed as a function

of induced mood, attribute favorableness, and brand favor-

ableness, treating the latter two variables as a repeated mea-

sure variable. Results of this analysis revealed no signiﬁcant

main or interactive effects involving mood ( ). Thus,p1.10

participants’ overall evaluations were no more favorable

when they were feeling happy ( ) versus when theyMp0.7

were not ( ). Products were evaluated more favor-Mp0.6

ably when the brand name associated with them was fa-

vorable ( ) than when it was not ( ),Mp2.0 Mp⫺0.6

However, this dif-

F(1, 158) p273.70, p!.01, hp0.64.

ference was virtually identical regardless of whether the

attribute information was evaluatively congruent with par-

ticipants’ mood (1.9 vs. ⫺0.5, for favorable and unfavorable

brands, respectively) or incongruent with it (2.0 vs. –0.7,

respectively),

F!1.

Test of Predictions. Although the overall effect of

mood on the impact of product information was not reliable,

the effect was predicted to depend on the product category

involved. To evaluate these predictions, evaluations of prod-

ucts in each category were analyzed separately as a function

of induced mood, gender, brand favorableness, and attribute

favorableness. In these analyses, brand favorableness and

attribute favorableness were between-subject variables.

Table 1 shows the mean evaluations of products in each

category as a function of gender, brand favorableness, at-

tribute favorableness, and mood. However, more relevant to

the predictions of concern is a comparison of the effects of

brand name under conditions in which the attribute infor-

mation was evaluatively consistent with participants’ mood

(i.e., either when participants were happy and the attribute

did not depend on the particular induction technique employed ( ).p1.10

Moreover, additional pretests on two different groups of subjects revealed

no differences in arousal as a result of the two techniques. Participants

who reported how stimulated, excited, calm, and relaxed they were on 0

(not at all) to 10 (very much) scales showed similar levels of arousal

regardless of whether the movie they watched was happy ( ) orMp4.77

sad ( ), or, whether they wrote about a happy or unhappyMp4.59 F!1

event (2.62 vs. 3.12), F!1.

This difference is evaluated statistically by the three-way interaction of

mood, brand favorableness, and attribute favorableness. The same inter-

action could alternatively be interpreted as indicating the effect of attribute

information when brand name and mood are evaluatively consistent versus

inconsistent. The emphasis placed on the impact of mood-attribute con-

sistency rather than mood-brand consistency was motivated by the need

to compare the data in experiment 1 with those obtained in experiment 2,

where characteristics of the attribute information that might have an impact

on its weight were experimentally manipulated.

8JOURNAL OF CONSUMER RESEARCH

TABLE 1

EFFECTS OF BRAND AND ATTRIBUTE IN POSITIVE AND NEGATIVE MOODS: EXPERIMENT 1

Conditions

Males Females

Favorable brand Unfavorable brand Favorable brand Unfavorable brand

Jeans:

Positive mood:

Favorable attribute 2.8 ⫺1.1 2.9 .2

Unfavorable

attribute 2.0 ⫺2.3 2.3 ⫺2.2

Negative mood:

Favorable attribute 3.5 ⫺.7 3.5 ⫺.3

Unfavorable

attribute 1.8 ⫺1.0 1.6 ⫺1.0

Running shoes:

Positive mood:

Favorable attribute 2.7 ⫺.9 3.7 .5

Unfavorable

attribute ⫺.2 ⫺2.6 ⫺.4 ⫺1.9

Negative mood:

Favorable attribute 3.0 ⫺1.0 1.6 1.5

Unfavorable

attribute ⫺2.0 ⫺3.2 ⫺1.5 ⫺2.5

Sweatshirts:

Positive mood:

Favorable attribute 4.1 .8 2.9 .3

Unfavorable

attribute 1.6 ⫺1.8 2.4 ⫺1.6

Negative mood:

Favorable attribute 3.2 .9 3.0 .0

Unfavorable

attribute 2.2 ⫺1.2 .8 .0

Backpacks:

Positive mood:

Favorable attribute 2.2 1.3 3.1 2.1

Unfavorable

attribute 1.2 ⫺.5 ⫺.1 ⫺1.0

Negative mood:

Favorable attribute 2.4 1.0 2.4 1.3

Unfavorable

attribute 1.3 ⫺1.5 1.3 ⫺1.3

information was favorable or when participants were un-

happy and the attribute information was unfavorable) with

conditions in which it was inconsistent. These data, shown

in Table 2, are discussed for each product category

separately.

Jeans. Both males and females considered hedonic cri-

teria more important than utilitarian criteria in evaluating

jeans. Therefore, both groups of participants should weight

mood-consistent attribute information more heavily when

evaluating products in this category. Results summarized in

Table 2 conﬁrm this prediction. That is, the effect of brand

name was less when mood and attribute information were

evaluatively consistent ( ) than when they were

Mp3.0

diff

inconsistent ( ) and this was true for both malesMp4.2

diff

and females.

This interpretation is conﬁrmed by an inter-

action of mood-attribute consistency and brand favorable-

ness,

that is not

F(1, 144) p3.01, p!.05, hp0.02,

dir

contingent on participants’ sex, F!1.

The data shown in Table 2 can be computed from the means shown in

Table 1. For example, the two mood-congruent conditions occur when (a)

participants are in a positive mood and the attribute information is favor-

able, and (b) participants are in a negative mood and the attribute infor-

mation is unfavorable. Males’ judgment of jeans in these two conditions

were 2.8 and 1.8, respectively, when the brand was favorable ( ),Mp2.3

and were ⫺1.1 and ⫺1.0, respectively, when the brand was unfavorable

( ). Therefore, as shown in Table 2, the mean effect of brandMp⫺1.0

name when the mood and attribute information were evaluatively congruent

is the difference between these two judgments or The2.3 ⫺(⫺1.0) p3.3.

effect of brand when participants’ mood and the attribute information were

evaluatively incongruent was determined analogously.

Here and elsewhere, predicted main effects and interactions were eval-

uated on the basis of a directional F-test (F

dir

). In all cases, these tests,

which involve a comparison of the mean of half the cells of the design

AFFECT CONFIRMATION 9

TABLE 2

EFFECTS OF BRAND WHEN MOOD AND ATTRIBUTES ARE CONSISTENT VERSUS INCONSISTENT: EXPERIMENT 1

Conditions

Males Females Overall

Favorable

brand Unfavorable

brand Difference

diff

)Favorable

brand Unfavorable

brand Difference

diff

)Favorable

brand Unfavorable

brand Difference

diff

)

Jeans:

Mood and attribute:

Congruent 2.3 ⫺1.0 3.3 2.3 ⫺.4 2.7 2.3 ⫺.7 3.0

Incongruent 2.7 ⫺1.5 4.2 2.9 ⫺1.2 4.1 2.8 ⫺1.4 4.2

Running shoes:

Mood and attribute:

Congruent .3 ⫺2.0 2.4 1.1 ⫺1.0 2.1 .7 ⫺1.5 2.2

Incongruent 1.4 ⫺1.8 3.2 .6 ⫺.2 .8 1.0 ⫺1.0 2.0

Sweatshirts:

Mood and attribute:

Congruent 3.1 ⫺.2 3.3 1.8 .1 1.8 2.5 .0 2.5

Incongruent 2.4 ⫺.4 2.9 3.1 ⫺.8 3.9 2.8 ⫺.6 3.4

Backpacks:

Mood and attribute:

Congruent 2.3 ⫺.1 2.4 2.2 .4 1.8 2.2 .2 2.0

Incongruent 1.8 .3 1.5 1.2 .2 1.0 1.5 .2 1.3

Running Shoes. Both males and females also reported

attaching more importance to hedonic criteria in evaluating

running shoes. To this extent, greater weight should be at-

tached to information that is mood-consistent for both

groups of participants. Contrary to expectations, however,

this effect was evident only among males. Speciﬁcally, the

effect of brand name on males’ judgments was less when

attribute information was consistent with their mood

( ) than when it was inconsistent ( ).Mp2.4 Mp3.2

diff diff

In contrast, the effect of brand on females’ judgments was

greater in the former condition ( ) than the latterMp2.1

diff

( ). Although the interaction of mood-attributeMp0.8

diff

consistency and brand favorableness was not signiﬁcant in

analyses for male and female participants separately (ineach

case ), the three-way interaction of these variablesp1.10

and participants’ sex was reliable, F(1, 144) p2.93, p!

dir

.05, hp0.02.

Sweatshirts. Only females attached relatively more im-

portance to feeling-related criteria in evaluating sweatshirts.

Consequently, the effect of mood-attribute consistency on

the weight attached to the information presented should be

evident for females but not for males. Data in Table2 support

this prediction. That is, the effect of brand name on females’

judgments was signiﬁcantly less when the attribute infor-

mation was consistent with mood ( ) than whenMp1.7

diff

it was not ( ),Mp3.9 F(1, 144) p5.14, p!.05,

diff dir

This suggests that the attribute information re-

hp0.07.

ceived more weight in the former conditions. In contrast,

the effect of brand name on males’ judgments was nonsig-

niﬁcantly greater when the attribute information was eval-

uatively consistent with participants’ mood than when it was

not ( ). The different effects of mood-attribute con-p1.10

with the mean of the other half, are equivalent to a one-tailed t-test, where

(for a further discussion, see Keppel 1991, pp. 122–123).

Fpt

sistency on males’ and females’ judgments are conﬁrmed

by an interaction of mood-attribute consistency, brand fa-

vorableness, and participants’ sex, F(1, 144) p4.88,

dir

p!.05, hp0.03.

Backpacks. Consistent with expectations, neither males

nor females evaluated backpacks on the basis of affective

considerations. To this extent, the effects of mood-attribute

consistency on the weight attached to information aboutthis

product should be negligible for both groups of participants.

This was in fact the case. If anything, the impact of brand

on judgments was nonsigniﬁcantly greater when the attribute

information was consistent with participants’ mood

( ) than when it was inconsistent ( ),Mp2.1 Mp1.3

diff diff

and this was true for both males and females. The interaction

of mood-attribute consistency and brand favorableness was

not signiﬁcant, and

F(1, 144) p2.06, p1.05, hp0.01,

dir

was not contingent on participants’ sex, F!1.

Supplementary Data

The only deviation from predictions based on Hypothesis

1 occurred in the case of running shoes. However, this de-

viation might be spurious. To evaluate this possibility, ad-

ditional data were collected for that product category. Fifty-

two male and 53 female participants were induced to feel

either happy or sad by writing about a pleasant or unpleasant

life experience in a procedure similar to that used by

Schwarz and Clore (1983) and described in detail later (see

experiment 2). Then, they were asked to indicate their liking

for a pair of running shoes based on a description repre-

senting one of the four combinations of brand favorableness

and attribute favorableness. (These judgments were made

along the same scales used in the experiment 1.)

Mood was again successfully manipulated. That is, par-

ticipants reported experiencing more positive feelings when

10 JOURNAL OF CONSUMER RESEARCH

they wrote about a happy life event ( ) than whenMp8.4

they wrote about a sad one ( ),Mp2.2 F(1, 86) p

Further, as expected, the dif-

565.14, p!.01, hp0.87.

ference between product evaluations when the brand was

favorable and evaluations when the brand was unfavorable

was appreciably less in those conditions in which the at-

tribute information and participants’ mood were evaluatively

consistent ( ) than when they were not (Mp0.5 Mp

diff diff

), Moreover, the

2.1 F(1, 89) p3.50, p!.05, hp0.04.

dir

effect of mood-attribute congruence was greater for females

( vs. 3.3, respectively) than for males (Mp0.7 Mp

diff diff

vs. 0.9, respectively). Although this gender difference0.3

was not reliable ( ), it is directionally consistent withp1.10

the difference in emphasis that males and females placed

on hedonic criteria for judgment, as indicated by the pretest

data reported earlier. Thus, it seems reasonable to conclude

that the effects of mood-attribute consistency on judgments

of running shoes, like its effects on judgments of products

in other categories, are compatible with implications of the

affect-conﬁrmation hypothesis.

EXPERIMENT 2

Experiment 1 provided preliminary support for Hypoth-

esis 1. That is, the effect of mood on the weight attached

to product information appeared to depend on the extent to

which affect-relevant criteria were spontaneously used as a

basis for judgments. It is conceivable, however, that differ-

ences in the criteria over the four product categories con-

sidered in this experiment are confounded with other factors

that might in some way inﬂuence the differential weighting

of information. Therefore, it seemed desirable to conﬁrm

the implications of the ﬁrst study under conditions in which

the likelihood that participants would attend to the affective

implications of the attribute information was experimentally

manipulated.

Participants in experiment 2 received information about

a pair of jeans that was described by its brand name (Levis

vs. Rustler) and a speciﬁc attribute (preshrunk vs. not pre-

shrunk). However, some participants were explicitly told to

base their evaluations on hedonic criteria (e.g., how it would

feel to wear the jeans), whereas others were told to base

their judgments on utilitarian criteria (e.g., quality of con-

struction). Moreover, the attribute information was described

in some cases as having implications for comfort and in

other cases as having implications for durability. It seemed

likely that both of these factors might inﬂuence the impact

of mood-induced affect on the weight attached to the at-

tribute information in evaluating the products.

The experiment also provided data bearing on Hypothesis

2. Speciﬁcally, participants who base their product evalu-

ations on hedonic criteria should have more conﬁdence in

these evaluations when the mood-induced affect they are

experiencing conﬁrms the evaluative implications of the in-

formation they receive. Data collected in experiment 2 con-

ﬁrmed this assumption as well.

Method

Participants were 325 male and female students from an

introductory marketing class. They were randomly assigned

to cells of a completely randomized between-subject design

consisting of mood (positive/negative), brand favorableness

(Levis/Rustler), attribute favorableness (preshrunk/not pre-

shrunk), processing goal (affective/utilitarian), and attribute

focus (affective/utilitarian).

Stimulus Materials. Jeans was the product category

selected for use in this study since both males and females

evaluated them on the basis of primarily hedonic criteria.

As in experiment 1, participants received information con-

sisting of a favorable or unfavorable brand name (“Levis:

ranked ﬁrst out of 10 different brands of jeans” vs. “Rustler:

ranked ninth out of 10 different brands of jeans”) and either

a favorable or an unfavorable attribute (“preshrunk” vs. “not

preshrunk”). In presenting the attribute information, how-

ever, a phrase was added that focused attention on either

the hedonic implications of the attribute dimension (“pre-

shrunk jeans feel soft against the skin”) or its utilitarian

implications (“preshrunk jeans have tighter stitches and bet-

ter durability”).

Procedure. Participants were given the same cover

story as in experiment 1 concerning the purpose of the re-

search being conducted. With this preamble,they were given

a different mood-induction task. This procedure, used by

Schwarz and Clore (1983), required participants to collab-

orate in the development of a life-event inventory that would

ostensibly be used to assess the types of experiences that

students go through while in school. Participants were then

asked to describe a recent event that “made you feel really

happy (unhappy) and continues to make you feel happy

(unhappy) whenever you think about it.” They were given

20 minutes to complete the task. To ensure that they paid

attention to the emotional aspects of the event, participants

were asked to focus on how the event made them feel and

to write about it in the form of a letter to a very close friend.

After completing this task, participants as part of an osten-

sibly unrelated study were given information about a pair

of jeans representing one of the eight combinations of brand

favorableness, attribute favorableness, and attribute focus

described above. Participants under utilitarian goal condi-

tions were asked to evaluate the jeans on the basis of how

well constructed they were. In contrast, participants under

hedonic goal conditions were asked to evaluate the jeans on

the basis of how it would feel to wear them.

Participants reported their evaluations along two scales

from ⫺5 (dislike extremely/extremely bad product) to ⫹5

(like extremely/extremely good product). After doing so,

they reported how conﬁdent they felt about their judgment

of the product and how certain they were that their evalu-

ation was correct, along scales from 0 (not at all conﬁdent/

certain) to 10 (extremely conﬁdent/certain). Responsesalong

each pair of scales were averaged to provide a single index

of both the favorableness of participants’ evaluation of the

AFFECT CONFIRMATION 11

TABLE 3

EFFECTS OF PROCESSING GOALS, ATTRIBUTE FOCUS, AND MOOD-ATTRIBUTE CONSISTENCY ON THE IMPACT OF BRAND

FAVORABLENESS ON PRODUCT EVALUATIONS AND CONFIDENCE: EXPERIMENT 2

Conditions Affect-focused

attributes Utilitarian-focused

attributes Mean

A. Effect of brand favorableness:

Feeling processing goal:

Mood and attribute evalua-

tively consistent 2.39 2.14 2.26

Mood and attribute evalua-

tively inconsistent 2.50 2.63 2.56

Utilitarian processing goal:

Mood and attribute evalua-

tively consistent 3.05 2.61 2.83

Mood and attribute evalua-

tively inconsistent 1.94 1.82 1.88

B. Reported conﬁdence:

Feeling processing goal:

Mood and attribute evalua-

tively consistent 5.75 5.82 5.78

Mood and attribute evalua-

tively inconsistent 5.05 5.69 5.37

Utilitarian processing goal:

Mood and attribute evalua-

tively consistent 5.50 4.99 5.25

Mood and attribute evalua-

tively inconsistent 4.76 5.23 5.00

Effect of brand favorableness is estimated from the difference in product evaluations when the brand is favorable and evaluations when the brand is

unfavorable.

product and their conﬁdence that this evaluation was correct.

Following the product evaluation task, participants com-

pleted the manipulation check questionnaire described in

experiment 1.

Results

Manipulation Check. Participants reported feeling

happier after writing about a positive life experience than

after writing about a negative one (6.96 vs. 4.04),

In addition, par-

F(1, 296) p323.03, p!.01, hp.522.

ticipants’ estimates of their mood at the end of the exper-

iment was signiﬁcantly more favorable in the former con-

dition ( ) than in the latter ( ),Mp0.70 Mp0.08

Thus, as in exper-

F(1, 309) p10.87, p!.01, hp.023.

iment 1, participants’ mood persisted throughout the

experiment.

Product Evaluations. The differential weighting of at-

tribute information implied by the affect-conﬁrmation hy-

pothesis seemed likely to depend on both (a) the extent to

which the attribute information actually elicited affect and

(b) whether participants’ processing goal was one to which

affective reactions were relevant. The results shed light on

this issue.

The overall impact of mood on judgments was not sig-

niﬁcant ( ). That is, participants evaluated productsp1.10

no more favorably when they were happy ( ) thanMp1.03

when they were not ( ). Nevertheless, mood inﬂu-Mp1.32

enced the weight that participants attached to the attribute

information. As in experiment 1, this weight was inferred

from the impact of the brand information that accompanied

the attribute descriptions in the two conditions in which the

attribute information and participants’ mood were evalua-

tively consistent and the two conditions in which they were

inconsistent. This impact or weight attached to brand name

is shown in the top section of Table 3 as a function of

participants’ processing goal, attribute focus, and attribute-

mood consistency.

The implications of these data are quite clear. When par-

ticipants were told to base their judgments on feelings, the

effect of brand on these judgments was less when their mood

was evaluatively consistent with the implications of the at-

tribute information than when it was not (2.26 vs. 2.56).

When they were told to base their judgments on utilitarian

criteria, however, the effect of brand name wassubstantially

greater in the former case than in the latter (2.83 vs. 1.88).

These results are conﬁrmed by an interaction of mood, brand

favorableness, attribute favorableness, and processing goal,

F(1, 280) p2.86, p!.05, hp.01.

dir

Thus, these data suggest that participants who were told

to base their judgments on their feelings attached greater

weight to attribute information when their mood-induced

affect was consistent with the evaluative implications of the

attribute information. On the other hand, participants who

were told to use utilitarian criteria attached less weight to

the attribute information in similar conditions. Although this

latter reversal was unexpected, it is somewhat interesting.

That is, it suggests that although mood-induced affect con-

12 JOURNAL OF CONSUMER RESEARCH

ﬁrmed participants’ affective reactions to the information

and increased their conﬁdence in the use of this information

as a basis for affect-based judgments, it decreased theircon-

viction that they should use the information to make non-

affect-based judgments. Additional data provide further in-

sight into this possibility.

Conﬁdence. Averaged over judgmental processing goal

and attribute focus, participants reported generally more

conﬁdence in their judgments when their mood was eval-

uatively consistent with the attribute information (Mp

) than when it was not ( ). This difference is5.51 Mp5.18

conﬁrmed by an interaction of mood and attribute favora-

bleness, However,

F(1, 291) p3.49, p!.05, hp.01.

dir

the data pertaining to this effect, summarized in the bottom

half of Table 3, suggest a contingency in this difference.

That is, when participants were told to use utilitarian criteria

and the attribute description also emphasized these criteria,

they reported less conﬁdence when their mood was eval-

uatively consistent with the attribute information than when

it was not (4.99 vs. 5.23). In the other three conditions (i.e.,

when participants were told to use hedonic criteria, the at-

tribute description emphasized hedonic implications, or

both), participants reported greater conﬁdence in their judg-

ments when their mood was evaluatively consistent withthe

attribute information, as expected (averaged over three con-

ditions, 5.69 vs. 5.17, respectively). A comparison of the

effect of mood-attribute consistency on conﬁdence in the

ﬁrst set of conditions with its effect in the last condition

was marginally signiﬁcant, F(1, 291) p2.72, p!.10,

hp.01.

This contingency is consistent with Hypothesis 2. How-

ever, the evidence that mood-attribute consistency increased

participants’ conﬁdence in their judgments when the attrib-

ute descriptions focused on affect but they were told to use

utilitarian criteria is noteworthy in light of the fact that par-

ticipants decreased the weight they attached to the attribute

information in this condition. When the attribute information

that participants received in this condition elicited affect that

was consistent with their mood, they were apparently more

certain that this information should not be used as a basis

for judgments. Consequently, they weighted thisinformation

less heavily than they would otherwise while reporting

greater conﬁdence in their judgments.

Summary

Experiment 2 conﬁrmed several assumptions of the affect-

conﬁrmation hypothesis. First, mood-attribute consistency

increased the weight attached to attribute information when

participants were told to use hedonic criteria as a basis for

evaluating the products but not when they were told to em-

phasize utilitarian criteria. Second, mood-attribute consis-

tency also increased participants’ conﬁdence in their judg-

ments when hedonic criteria were emphasized. Theincrease

in conﬁdence with mood-attribute consistency was also ev-

ident when participants were told to base their judgments

on utilitarian criteria provided the attributes were described

in a way that focused attention on feelings. In this case,

however, participants rejected the affect-eliciting attribute

information as a basis for judgment, giving it less weight

than they normally would.

EXPERIMENT 3

Although the results of experiments 1 and 2 support the

affect-conﬁrmation hypothesis and underlying assumptions,

differential weighting was detected by assessing the impact

of the additional information presented (i.e., brand name)

when affect and attribute information were consistent. It

seemed desirable to demonstrate this phenomenon using a

different and more direct paradigm. Experiment 3 provided

this demonstration using a procedure similar to that em-

ployed by Sanbonmatsu et al. (1999). This procedure infers

the weight that participants attach to an attribute from the

likelihood that they would list this attribute dimension later

on as one they consider to be generally important in judging

products of that type. In experiment 3, participants in dif-

ferent affective states judged a product that was favorable

along two attribute dimensions and unfavorable along two

others. After making these judgments, they listed those at-

tribute dimensions that they personally considered to be

generally relevant in assessing products from that category.

Participants experiencing positive affect were expected to

be more likely to list dimensions along which the product

had been described favorably and less likely to list dimen-

sions along which the product had been described unfavor-

ably, than participants who experienced negative affect.

Method

Overview and Design. Forty students from an intro-

ductory marketing class participated in the experiment for

extra course credit. They were randomly assigned to one of

two affect conditions (positive vs. negative) and presented

with information about a pair of jeans. The information

presented was favorable along two attribute dimensions and

unfavorable along two others. Participants were asked to

indicate how much they liked the product described. They

were then asked to list those attributes that they personally

felt were important in evaluating jeans.

Stimulus Materials. Product information was pre-

sented in a table that summarized how other consumers had

rated the jeans along four attribute-dimensions (comfort,

style, fabric quality, and availability of sizes). The average

ratings by consumers were presented in two columns. One

column indicated how the jeans scored on each attribute

along a 1 (very undesirable) to 9 (very desirable) scale, and

the second indicated how the jeans performed relative to

other brands on each of these attributes. In one set of ques-

tionnaires, comfort and fabric quality were rated favorably

(rating of eight and seven; better than 95 percent and 80

percent of the other brands), while style and availability of

sizes were rated unfavorably (a rating of two and three;

better than only 25 percent and 40 percent of other brands).

AFFECT CONFIRMATION 13

In a second set of questionnaires, these numbers were re-

versed. That is, style and availability of sizes were rated

favorably and comfort and fabric quality were rated unfa-

vorably. The order of presenting the favorable and unfa-

vorable attributes was therefore counterbalanced.

Procedure. The mood-induction procedure employed

was similar to that used in experiment 1. Participants who

had watched either happy or sad ﬁlms were given product

information that ostensibly came from a research report pre-

pared by “Consumer Forum.” This ﬁctitious organization

had surveyed approximately 200 consumers and asked them

to try on and evaluate the jeans. Participants were told that

the researchers were interested in how they felt about the

jeans based on the information provided. Withthis preamble,

participants were shown the product information and eval-

uated the jeans using three scales that ranged from ⫺5to

⫹5 and pertained to liking (dislike extremely/like ex-

tremely), goodness (very bad/very good), and desirability

(very undesirable/ver y desirable). Following this evaluation,

they were told, “Please list those attributes that you per-

sonally feel are important in evaluating jeans. These attrib-

utes may or may not include the attributes that were listed

in the table you saw recently. We would like to know what

attributes you feel are important when you evaluate jeans.

List as many as you like.” Participants were given as much

time as they wanted to list out the attributes and could look

at the table seen earlier if they wished to do so. After par-

ticipants had ﬁnished the product evaluation task, they were

given the mood manipulation check questionnairedescribed

earlier and on completion of that questionnaire they were

dismissed.

Results

Manipulation Check. Participants reported experienc-

ing more positive affect when they had watched happy ﬁlms

( ) than when they had watched sad ﬁlms (Mp7.6 Mp

), and also re-

4.4 F(1, 38) p101.07, p!.01, hp.728,

ported being in a relatively more positive mood at the time

they completed the manipulation check (2.4 vs. ⫺0.5),

F(1, 38) p85.14, p!.01, hp.691.

Thought Listings. The total number of attribute di-

mensions listed was determined for each participant. The

proportion of these dimensions that pertained to (a) attrib-

utes that were described favorably in the information pre-

sented earlier, (b) attributes that were described unfavorably,

and (c) the ratio of these proportions were then computed.

As expected, participants experiencing positive affect

listed a greater proportion of dimensions pertaining to pre-

viously presented, favorable attributes ( ) than par-Mp.40

ticipants experiencing negative affect ( ), and aMp.34

lower proportion of dimensions pertaining to previously pre-

sented, unfavorable attributes (.26 vs. .36, respectively).

This difference is conﬁrmed by an interaction of mood and

listed attribute valence,

F(1, 36) p4.65, p!.05, hp

dir

.114.

Similarly, the ratio of the proportion of listed dimensions

pertaining to previously presented favorable attributes and

the proportion of listed dimensions pertaining to previously

presented unfavorable attributes was greater for participants

who were experiencing positive affect ( ) than forMp1.43

those experiencing negative affect ( ),Mp0.96

Moreover, this ratio

F(1, 30) p7.55, p!.05, hp.201.

dir

was correlated .359 ( ) with participants’self-reportedp!.05

mood state. These results provide independent conﬁrmation

of the assumption that participants placed more emphasis

on aspects of the product information that were evaluatively

consistent with the feelings they were experiencing at the

time they received this information.

EXPERIMENT 4

The affect-conﬁrmation hypothesis implicitly assumes that

participants are typically not motivated to make a deliberative

assessment of the different sources of affect they are expe-

riencing at the time of judgment. Thus, they do not distinguish

between the feelings elicited by a mood state and the affective

reactions to product information. Consequently, when the af-

fect they are experiencing from different sources has similar

evaluative implications, they are more conﬁdent of their re-

actions to the attribute information than when it is inconsis-

tent. If this assumption is correct, calling participants’ atten-

tion to the fact that the affect they are experiencing is partly

due to product-unrelated factors should decrease its inﬂuence

on their conﬁdence in their reactions to the attribute infor-

mation and the weight attached to this information.

Making product-irrelevant sources of affect salient could

have other inﬂuences as well. For example, it could lead

participants to misattribute the affect that is actually elicited

by the product information to these extraneous sources instead

(for evidence of these misattribution tendencies, see Schwarz

and Clore [1983]). If this is so, and if participants typically

base their product evaluations on their affective reactions,this

would reduce the impact of the attribute information on their

judgments independently of other considerations.

These possibilities were examined in experiment 4. Par-

ticipants who had been induced to feel happy or sad by

watching a movie evaluated a pair of jeans described by a

favorable or unfavorable brand name and a favorable or

unfavorable attribute. In no-attribution conditions, these

judgments were made immediately after watching the

movie. In attribution conditions, however, the manipulation-

check questionnaire that was normally completed at the end

of the experiment was administered before product evalu-

ations were made. This questionnaire was expected to draw

participants’ attention to the extraneous source of the feel-

ings they were experiencing at the time they performed the

product-evaluation task. Therefore, it was expected to de-

crease both the impact of the attribute information on their

product evaluations and the weight they attached to this

information under conditions in which its implications were

evaluatively consistent with their mood.

14 JOURNAL OF CONSUMER RESEARCH

Method

Participants were 66 undergraduate business students who

participated to fulﬁll a course requirement. After receiving

the same cover story provided in earlier studies, they saw

either the happy or the sad ﬁlms used in experiment 1 and

completed the movie-rating questionnaires that pertained to

these ﬁlms. Then participants in attribution conditions were

administered the mood-manipulation check questionnaire

used in earlier studies, whereas participants inno-attribution

conditions performed the product-evaluation task before

completing this measure.

The product-evaluation task was similar to that used in

experiment 2 except that no indication was given of the

criteria to use as a basis for judgments and the attribute

descriptions did not specify the implications of these de-

scriptions for comfort or construction quality. After reading

the product description, participants reported their evalua-

tions of the product along the scales employed in experiment

1, and their responses along these scales were later averaged

to provide a single estimate of their liking for the product.

Results

Manipulation Check. Participants reported having

more positive affective reactions to the happy ﬁlms (Mp

) than to the sad ones ( ),7.88 Mp2.73 F(1, 62) p

Although the difference was

235.54, p!.01, hp0.79.

somewhat greater under no-attribution conditions (7.91 vs.

2.26) than under attribution conditions (7.71 vs. 3.20), this

contingency was only marginally reliable, F(1, 62) p

Participants also estimated their

3.20, p!.10, hp0.05.

mood, at the time they completed the manipulation check

questionnaire, to be more positive when they had watched

happy movies ( ) than when they had watched sadMp2.66

ones ( ), This was

Mp0.57 F(1, 62) p19.94, hp.24.

true regardless of whether these estimates were made im-

mediately after the movies (2.70 vs. 0.09) or not until after

the products were evaluated (2.61 vs. 1.06). Thus, as in

earlier studies, the effect of the mood induction persisted

throughout the experiment.

Product Evaluations. Participants presumably based

their evaluations of jeans on primarily hedonic criteria and

thus on their affective reactions to the product. Therefore,

reminding participants of an alternative source of the affect

they were experiencing should lead them to attribute their

feelings to this source rather than to the product information

and, therefore, should decrease the effects of this infor-

mation on product judgments. This was in fact the case.

That is, under no-attribution conditions, participants re-

ported greater liking for the product when the attribute in-

formation was favorable ( ) than when it was notMp1.38

( ). When the extraneous source of their affect wasMp0.18

called to their attention, however, this difference was non-

signiﬁcantly reversed (0.33 vs. 0.79). The interaction of at-

tribute favorableness and attribution of feelings was reliable,

F(1, 50) p4.70, p!.05, hp.09.

Second, reminding participants of the true source of their

affect should decrease the inﬂuence of this affect on the

weight they attach to the attribute information and, therefore,

should increase the impact of the brand information that

accompanies it. This possibility was evaluated under the two

conditions in which the participants’ mood was consistent

with the implications of the attribute information and, there-

fore, potentially conﬁrmed these implications (i.e., condi-

tions in which the participants’ mood and the attribute in-

formation were either both positive or both negative). As

expected, the effect of brand on liking judgments in these

conditions was greater when the source of the product-ir-

relevant affect that participants were experiencing was called

to their attention (2.21 vs. ⫺1.18 when brand was favorable

vs. unfavorable, respectively) than when it was not (1.67

vs. 0.03, respectively). Although the difference in the effect

of brand under these two conditions only approached sig-

niﬁcance, it is consistent withF(1, 27) p2.27, p!.07,

dir

expectations.

DISCUSSION AND CONCLUSIONS

The present research provides the ﬁrst evidence in the

area of consumer judgment to establish that affect can in-

ﬂuence the relative weight that individuals give to different

pieces of product information in making a judgment. This

inﬂuence occurs even when the product information is avail-

able at the time of judgment and when minimal demands

are placed on recipients’ ability to keep the information in

mind.

The affect-conﬁrmation hypothesis used to account for

these ﬁndings provides new insights into the cognitive mech-

anisms that underlie the impact of affect on judgments. In

doing so, it constrains the conditions in which affect is likely

to have an impact. That is, differential weighting of affect-

consistent information occurs only when people base their

product evaluations on hedonic criteria. Experiments 1 and

2 provided converging evidence of this contingency. In the

ﬁrst experiment, no a priori indication was given concerning

the criteria participants should use. In this case, the criteria

that participants spontaneously applied depended on both

the type of product they were asked to judge and their per-

sonal needs and values, and their differential weighting of

mood-consistent product information varied accordingly.

Although differences in weighting over product categories

might be attributed to other factors that were confounded

with the criteria that participants applied in making evalu-

ations, experiment 2 demonstrated similar effects by ex-

perimentally inducing demands to employ hedonic or util-

itarian criteria.

Three assumptions underlying the affect-conﬁrmation

process were also conﬁrmed. Experiment 2 indicated that

when participants were told to base their judgments on feel-

ings, they not only gave more weight to attribute information

that was evaluatively consistent with their mood but reported

greater conﬁdence in the judgments they made. Experiment

3 suggested that people thought more extensively about

product attributes that were evaluatively consistent with their

AFFECT CONFIRMATION 15

mood. Finally, experiment 4 indicated that calling partici-

pants’ attention to product-irrelevant sources of their affect

decreased the impact of mood-attribute consistency on the

weight they attached to the attribute information. In com-

bination, these ﬁndings provide compelling evidence for the

assumptions underlying the theory being proposed.

Theoretical Issues and Assumptions

Support for the affect-conﬁrmation processes identiﬁed

in the present research does not preclude other inﬂuences

that affect might have on product judgments. For example,

memory processes (see Fig. 1, path A) might conceivably

be evident when a large amount of information is presented

and must be retrieved from memory in order to make a

judgment (Forgas 1995; but see Wyer et al. 1999). However,

memory processes seem unlikely to have played a signiﬁcant

role in the present experiments, in which only two pieces

of information were presented and this information was

available at the time judgments were made. In contrast, af-

fect-conﬁrmation processes could occur independently of

the amount of information presented and its availability at

the time of judgment. To this extent, the affect-conﬁrmation

hypothesis has more general applicability than memory-

based theories of differential weighting.

Second, the extraneous affect that people experience can

sometimes have a direct informational impact on judgments

independently of the information presented (see Fig. 1, path

D). The occurrence of this effect is marked by more fa-

vorable evaluations by happy people than unhappy ones.

This direct effect of affect was not apparent in the present

research. This may be somewhat surprising. As speculated

earlier, however, affect may only be used as a heuristicwhen

a relatively large amount of product information is available

and its implications are difﬁcult to assimilate (for a discus-

sion of the use of affect as a heuristic, see Forgas [1995];

Schwarz and Clore [1996]). In the present research, only a

small amount of information was presented whose impli-

cations were fairly easy to integrate. Consequently, the use

of affect as a heuristic basis for judgment may not have

occurred.

Two assumptions should nevertheless be noted in eval-

uating the generality of the ﬁndings reported. First, an im-

plicit assumption underlying the affect-conﬁrmation hy-

pothesis is that individuals consider the implications of each

piece of information independently (see Wyer and Carlston

[1979]; Wyer and Srull [1989], for discussions of this con-

tingency). Second, the indirect procedures for evaluating

differential weighting assume that people subjectively av-

erage the implications of the information rather than sum-

ming them (cf. Anderson 1981). Although summative pro-

cesses might sometimes occur (see Fishbein and Ajzen

1975), these assumptions appear justiﬁed in many studies

of information integration both in the consumer domain

(Bettman et al. 1975; Troutmanand Shanteau 1976; see also

Adaval 1996) and elsewhere (Anderson 1981).

The applicability of the affect-conﬁrmation hypothesis

should also be considered in the context of ﬁndings reported

by Ditto and Lopez (1992). They found that when people

have an a priori preference for a particular outcome, they

devote more thought to information that is inconsistent with

this preference in an attempt to refute its implications. This

suggests that when consumers already have an affect-based

disposition to like (or dislike) a particular product before

receiving information about it, they may devote more

thought to information that is inconsistent with this dispo-

sition in an effort to discredit it. This discrediting of incon-

sistent information would presumably lead them to attach

relatively greater weight to the conﬁrming information, as

the affect-conﬁrmation hypothesis also predicts. The pro-

cesses postulated by Ditto and Lopez (1992), however, pre-

sumably occur only when people have a strong predispo-

sition to evaluate the product favorably or unfavorably and,

therefore, are motivated to refute the implications of infor-

mation that calls this predisposition into question. In the

present studies, this was clearly not the case. When these

conditions exist, however, they could add to the effects of

affect conﬁrmation.

Implications for Consumer Information Processing

and Decisions

To the extent that the research reported is generalizable

to consumer purchasing situations outside the laboratory, it

has several interesting implications. For example, suppose

consumers see a package of Ha¨agen-Dazs chocolate-marsh-

mallow ice cream that is priced somewhat high. Moreover,

suppose they consider “Ha¨agen-Dazs” and “chocolate” to

be favorable but the “high price” and “marshmallows” to

be unfavorable. If the consumers have a hedonic goal when

they evaluate the product, they should weight “chocolate”

and the brand name “Ha¨agen-Dazs” more heavily in their

purchase decision when they are feeling happy than when

they are not. At the same time, other product attributes (e.g.,

marshmallows) should be weighted less heavily and have

less inﬂuence on their decision than they would otherwise.

Although increased thought and counterarguing could

change the weight assigned to different attributes, this is

more likely to occur when there is a delay between the on-

line-evaluation and purchase decisions. This is because ad-

ditional cognitive activity might result in the change of the

criteria that underlie the actual purchase decision. Millarand

Tesser (1989) note that attitudes only predict behavior when

the criteria that underlie the behavioral decision match the

criteria on which the original attitude was based. This means

that the impact of affect on the weight attached to product

information may inﬂuence decisions to purchase the product

only when the decision is also affectively driven. It may be

possible of course to inﬂuence the criteria that persons bring

to bear on a product evaluation by presenting in-store ad-

vertising material that gets people to focus on feelings versus

functionality.

In some situations the notion that favorable product in-

formation will have greater inﬂuence, and unfavorable in-

formation will have less inﬂuence, when consumers are

16 JOURNAL OF CONSUMER RESEARCH

happy than when they are not may have undesirable con-

sequences. For example, cigarette advertisements contain

warnings as well as positive descriptions of the products.

The affect-conﬁrmation hypothesis suggests that consumers

will weight the warnings less heavily when they are feeling

happy than when they are not. It is important to note, how-

ever, that these possible inﬂuences of affect on the weight

attached to product information are more likely to occur

when consumers base their judgments and decisions on af-

fective rather than utilitarian criteria.

In a related vein, many product advertisements attempt

to induce positive affect in recipients for reasons that are

objectively unrelated to the characteristics of the product

being advertised (e.g., through the use of unrelated albeit

beautiful visuals, music, humor, celebrities,etc.). This strat-

egy is often assumed to increase the inﬂuence of product

features that also elicit positive affect. However, the present

research suggests that this will occur only when recipients

consider affective criteria to be a primary basis for judg-

ments. If the product being advertised is evaluated on the

basis of utilitarian considerations (e.g., a cough syrup), affect

that is elicited either by the product features or by the context

in which it is described should have no impact.

[Received August 1999. Revised November 2000. David

Glen Mick served as editor, and John Deighton served as

associate editor for this article.]

REFERENCES

Adaval, Rashmi (1996), “The Role of Affective (Mood) States in

Perceptions of Quality, Sacriﬁce and Value,” unpublished dis-

sertation, Department of Business Administration, College of

Commerce and Business Administration, University of Illi-

nois, Champaign, IL 61820.

Anderson, Norman H. (1981), Foundations of Information Inte-

gration Theory, New York: Academic Press.

Babin, Barry J., William R. Darden, and Mitch Grifﬁn (1994),

“Work and/or Fun: Measuring Hedonic and Utilitarian Shop-

ping Value,” Journal of Consumer Research, 20 (March),

644–656.

Batra, Rajeev and Olli T. Ahtola (1990), “Measuring the Hedonic

and Utilitarian Sources of Consumer Attitudes,” Marketing

Letters, 2 (April), 159–170.

Bettman, James R., Noel Capon, and Richard J. Lutz (1975), “Mul-

tiattribute Measurement Models and Multiattribute Attitude

Theory: A Test of Construct Validity,” Journal of Consumer

Research, 1 (March), 1–15.

Birnbaum, Michael F. (1974), “The Nonadditivity of Personality

Impression,” Journal of Experimental Psychology, 103

(March), 554–561.

Bless, Herbert, Gerd Bohner, Norbert Schwarz, and Fritz Strack

(1990), “Mood and Persuasion: A Cognitive Response Anal-

ysis,” Personality and Social Psychology Bulletin, 16 (June),

331–345.

Bodenhausen, Galen V. (1993), “Emotions, Arousal, and Stereo-

typic Judgments: A Heuristic Model of Affect and Stereo-

typing,” in Affect, Cognition and Stereotyping: Interactive

Processes in Group Perception, ed. Diane M. Mackie and

David L. Hamilton, San Diego, CA: Academic Press, 13–37.

Bower, Gordon H. (1981), “Mood and Memory,” American Psy-

chologist, 36 (2), 129–148.

———, Stephen G. Gilligan, and Kenneth P. Monteiro (1981),

“Selectivity of Learning Caused by Affective States,” Journal

of Experimental Psychology: General, 110 (December),

451–473.

Chaiken, Shelly (1987), “The Heuristic Model of Persuasion,” in

Social Inﬂuence: The Ontario Symposium, Vol. 5, ed. Mark

P. Zanna et al., Hillsdale, NJ: Erlbaum, 3–39.

Clore, Gerald L., Norbert Schwarz, and Michael Conway (1994),

“Affective Causes and Consequences of Social Information

Processing,” in Handbook of Social Cognition, Vol. 1, ed.

Robert S. Wyer and Thomas K. Srull, Hillsdale, NJ: Erlbaum,

323–419.

Cohen, Joel B. and Charles S. Areni (1991), “Affect andConsumer

Behavior,” in Handbook of Consumer Behavior, ed. Thomas

S. Robertson and Harold H. Kassarjian, Englewood Cliffs,

NJ: Prentice Hall, 188–240.

Ditto, Peter H. and David F. Lopez (1992), “MotivatedSkepticism:

Use of Differential Decision Criteria for Preferred and Non-

preferred Conclusions,” Journal of Personality and Social

Psychology, 63 (October), 568–584.

Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, Intention

and Behavior: An Introduction to Theory and Research,Read-

ing, MA: Addison-Wesley.

Forgas, Joseph P. (1995), “Mood and Judgment: The Affect In-

fusion Model (AIM),” Psychological Bulletin, 117 (January),

39–66.

——— and Gordon H. Bower (1987), “Mood Effects on Person

Perception Judgments,” Journal of Personality and Social

Psychology, 53 (July), 56–68.

Gilligan, Stephen G. and Gordon H. Bower (1983), “Reminding

and Mood-Congruent Memory,” Bulletin of the Psychonomic

Society, 21 (November), 431–434.

Gorn, Gerald J., Marvin E. Goldberg, and Kunal Basu (1993),

“Mood, Awareness and Product Evaluation,” Journal of Con-

sumer Psychology, 2 (3), 237–256.

Isen, Alice M. (1984), “Toward Understanding the Role of Affect

in Cognition,” in Handbook of Social Cognition, Vol. 3, ed.

Robert S. Wyer, Jr. and Thomas K. Srull, Hillsdale, NJ: Erl-

baum, 179–236.

———, Thomas E. Shalker, Margaret Clark, and Lynn Karp

(1978), “Affect, Accessibility of Material in Memory, and

Behavior: A Cognitive Loop,” Journal of Personality and

Social Psychology, 36 (January), 1–12.

Keppel, Geoffrey (1991), Design and Analysis: A Researcher’s

Handbook, Englewood Cliffs, NJ: Prentice Hall.

Mano, Haim and Richard L. Oliver (1993), “Assessing the Di-

mensionality and Structure of the Consumption Experience:

Evaluation, Feeling, and Satisfaction,” Journal of Consumer

Research, 20 (December), 451–466.

Martin, Leonard L., Teresa Abend, Constantine Sedikides, and

Jeffrey D. Green (1997), “How Would It Feel If . . . ? Mood

as Input to a Role Fulﬁllment Evaluation Process,” Journal

of Personality and Social Psychology, 73 (August), 242–253.

———, David W. Ward, John W. Achee, and Robert S. Wyer

(1993), “Mood as Input: People Have to Interpret the Moti-

vational Implications of Their Moods,” Journal of Personality

and Social Psychology, 64 (March), 317–326.

Meyer, Robert J. (1981), “A Model of Multiattribute Judgments

under Attribute Uncertainty and Information Constraints,”

Journal of Marketing Research, 18 (November), 428–441.

Millar, Murray G. and Abraham Tesser (1989), “The Effects of

AFFECT CONFIRMATION 17

Affective-Cognitive Consistency and Thought on the Atti-

tude-Behavior Relation,” Journal of Experimental Social Psy-

chology, 25 (March), 189–202.

Niedenthal, Paula M., Jamin B. Halberstadt, and Marc B. Setter-

lund (1997), “Being Happy and Seeing ‘Happy’: Emotional

State Mediates Visual Word Recognition,” Cognition and

Emotion, 11 (July), 403–432.

——— and Marc B. Setterlund (1994), “Emotion Congruence in

Perception,” Personality and Social Psychology Bulletin,20

(August), 401–411.

Penner, Louis A., Saul Shiffman, Jean A. Paty, and Barbara A.

Fritzsche (1994), “Individual Differences in Intraperson Var-

iability in Mood,” Journal of Personality and Social Psy-

chology, 66 (April), 712–721.

Pham, Michel T. (1998), “Representativeness, Relevance and the

Use of Feelings in Decision Making,” Journal of Consumer

Research, 25 (September), 144–159.

Sanbonmatsu, David M., Frank R. Kardes, David C. Houghton,

Edward A. Ho, and Steven S. Posavac (1999), “Overweighing

the Given Information in Multi-attribute Judgment,” working

paper, University of Utah, Salt Lake City, UT 84112.

Schwarz, Norbert and Gerald L. Clore (1983), “Mood, Misattri-

bution, and Judgments of Well-Being: Informative and Di-

rective Functions of Affective States,” Journal of Personality

and Social Psychology, 45 (September), 513–523.

——— and Gerald L. Clore (1996), “Feelings and Phenomenal

Experiences,” in Social Psychology: Handbook of Basic Prin-

ciples, ed. Edward Tory Higgins and Arie W.Kruglanski, New

York: Guilford, 433–465.

Shiv, Baba and Alexander Fedorikhin (1999), “Heart and Mind in

Conﬂict: The Interplay of Affect and Cognition in Consumer

Decision Making,” Journal of Consumer Research, 26 (De-

cember), 278–292.

Singer, Jefferson A. and Peter Salovey (1988), “Mood and Mem-

ory: Evaluating the Network Theory of Affect,” Clinical Psy-

chology Review, 8 (2), 211–251.

Troutman, Michael C. and James A. Shanteau (1976), “Do Con-

sumers Evaluate Products by Adding or Averaging Attribute

Information?” Journal of Consumer Research, 3 (September),

101–106.

Wyer, Robert S., Jr. and Donal E. Carlston (1979), Social Cog-

nition, Inference and Attribution. Hillsdale, NJ: Erlbaum.

———, Gerald L. Clore, and Linda M. Isbell (1999), “Affect and

Information Processing,” in Advances in Experimental Social

Psychology, Vol. 31, ed. Mark P. Zanna, San Diego: Academic

Press, 1–77.

——— and Thomas K. Srull (1989), Memory and Cognition in

Its Social Context. Hillsdale, NJ: Erlbaum.

Zanna, Mark and John K. Rempel (1988), “Attitudes: A New Look

at an Old Concept,” in The Social Psychology of Knowledge,

ed. Daniel Bar-Tal and Arie Kruglanski, Cambridge: Cam-

bridge University Press, 315–334.

Impulsive hotel consumption intention in live streaming E-commerce settings: Moderating role of impulsive consumption tendency using two-stage SEM

Article

Sep 2023
Int J Hospit Manag

The research intends to identify the predictors of impulsive hotel consumption behaviour among tourism for a live-streaming e-commerce context in Malaysia. The research also examines the moderating relationship of impulsive consumption tendency in the same perspective. This research developed the customer affecting predictor model of live streaming e-commerce on the basis of the cognitive emotion theory (CET) and Theory of pleasure-arousal dominance (PAD). In an online cross-sectional research design, the study collected 430 responses selected randomly and analysed applying Structural Equation Modelling (SEM) using AMOS 26. The outcome of the research confirms that live streaming features, representational delight, positive emotion, negative emotion, usage intention, and pleasure, are the factors that influence impulsive hotel consumption significantly except arousal. The study results confirmed impulsive consumption tendency as moderator. The study suggests some implications at the end.

From comparison to purchasing: Effects of online behavior toward associated co-visited products on consumer purchase

Article

Apr 2024
INFORM MANAGE-AMSTER

Research on the Effectiveness of Virtual Endorsers: A Study Based on the Match-Up Hypothesis and Source Credibility Model

Article

Full-text available

Feb 2024

In the digital era, businesses are actively integrating advanced technology and innovative marketing strategies to achieve sustained growth. Notably, virtual endorsers play a key role in driving enterprises towards digital transformation in the field of digital marketing. Understanding consumer attitudes towards the use of virtual endorsers in digital marketing is especially important for enterprises employing digital tools to realize a sustainable business model. To this end, this research adopts the match-up hypothesis and source credibility model as its theoretical framework, delving into the impact of virtual endorser–product fit and credibility features (attractiveness, expertise, and trustworthiness) on product attitudes, as well as the interactive effects of these features with product types. We collected feedback data from 376 participants through an online questionnaire and validated our hypotheses using the PLS-SEM model. The results demonstrate that virtual endorser–product fit positively affects credibility, with higher credibility further enhancing consumers’ attitudes toward products. Additionally, the source credibility model partially mediates the relationship between the match-up hypothesis and product attitudes, with attractiveness exerting the most significant impact. Finally, we observed variations in consumer attitudes toward products endorsed by virtual endorsers based on product types. The findings of this study provide a solid theoretical basis for a deeper understanding of consumer attitudes towards the application of virtual endorsers in marketing and offer practical suggestions for businesses to leverage digital tools for sustainable development.

The Influence of Brand Satisfaction and Luxury Brand Attachment on Customer Advocacy Through Brand Loyalty on Zalora Marketplace

Article

Full-text available

Dec 2023

This study is an exploratory study where researchers try to measure the influence of luxury brand fashion objects during the Covid-19 pandemic. The pursose of this research is to produce an overview of the marketing strategy for the online store Zalora Indonesia which sells luxury fashion brands, in an efffort to face very tight competition in the market in Indonesio and to recover after being affected by the Coved-19 pandemic by measuring the magnitude of the influence of brand satisfaction, luxury brand attachment, and brand loyalty on customer advocacy. The respondents in this study were 150 who were customers of luxury fashion brands in Zalora Indonesia who had made a purchase at least once. Non-probablity sampling is the sampling technique, and purposive sampling is the sampling method. The results show that luxury brand attachments have an influence on brand loyalty, and will increase customer advocacy. The practical implication of this research is that Zalora Indonesia, which sells luxury goods, can increase customer advocacy for online purchases through Zalora Indonesia.

Are you looking for something specific or just looking around? Adaptive selling on the basis of customer shopping goals in retail sales

Article

Mar 2024

A key advantage that brick-and-mortar retailers have over online retailers is their salespeople, who can adaptively interact with customers on a one-on-one basis. When starting an interaction with a customer, a retail salesperson generally first aims to determine their shopping goal (Hall et al., Journal of Marketing, 79(3), 91–109, 2015), often with questions such as, “How can I help you today?” Yet, little is known in the literature or in practice about how salespeople should adapt their sales approach based on customers’ shopping goals. This is unfortunate, because customers clearly want better help from salespeople, and the potential gains of doing so are substantial (Accenture, 2021; Hochstein et al., Journal of the Academy of Marketing Science, 47(1), 118–137, 2019). To address this limitation, this research focuses on the practice of adaptive selling in retail settings, in which salespeople adapt their tactics on the basis of customers’ shopping goals. Using information processing theory, we propose that matching sales influence tactics (SITs) to two aspects of customers’ shopping goals, namely goal specificity and product type (utilitarian vs. hedonic), improves purchase outcomes. Across a series of field and lab experiments, we demonstrate that purchase behavior and purchase intention are higher when salespeople use informational (emotional) SITs with customers who have a high (low) shopping goal specificity level, which we term the match strategy. The match strategy has a direct positive effect on purchase outcomes and an effect mediated by processing fluency on purchase outcomes, with product type serving as a moderator. This research concludes with specific, actionable recommendations for retail salespeople and managers.

Understanding emotions

Chapter

Jan 2024

The role of religiosity on seeking help

Article

Nov 2023

Xuefeng Liu

Previous research shows that religiosity increases prosocial tendencies such as donating and volunteering. Less is known on whether and how religiosity influences individuals' decision to seek help when needed. The current research tackles this question and shows that religiosity, whether manipulated (Study 2) and measured (Study 1 and Study 3), decreases individuals' tendency to seek help from other people or entities. We further propose that religiosity enhances individuals' sense of control, which makes them rely more on themselves and less likely to seek help when encountering difficulties. Three studies across different contexts (i.e., applying government aid, asking for help from other people, and requesting donations from a crowdfunding platform) support our thesis. Utilizing data from the General Social Survey (Study 1), we first demonstrate that individual‐level religiosity is negatively correlated with the behavior of getting help (i.e., receiving government aid such as welfare and unemployment insurance). Next, we prime religion (vs. control) in Study 2 and show that religious prime decreases people's willingness to seek help. Finally, Study 3 shows that chronic religiosity discourages people from seeking financial assistance on crowdfunding platforms such as Indiegogo. Furthermore, this negative effect is mediated by sense of control. We conclude with a discussion of the theoretical and practical contributions of this research as well as its limitations and future research directions.

Do fake reviews promote consumers' purchase intention?

Article

Sep 2023
J BUS RES

Stick to my guns: The impact of crowding on consumers’ responsiveness to sale promotions

Article

Aug 2023
J ACAD MARKET SCI

The current research examines the relationship between crowding and consumers’ responsiveness to sales promotions. Six studies show that the experience and feeling of crowdedness reduce the impact of sales promotions, demonstrating that consumers’ product/service purchase intention changes to a lesser extent in response to such promotions. This effect is found to be driven by consumers shifting their attention from the external environment to their internal feelings and thoughts when experiencing crowdedness. As a result, consumers rely more on their internal feelings and thoughts than on external cues in judgment, and consequently their purchase intention becomes less susceptible to external sales promotion information. In addition, this effect is found to be attenuated in situations where product attitudes are detached from consumers’ own preferences, such as in the context of gift choices, and when the experience of crowding is not aversive (e.g., watching an exciting football game in a bar).

Contextual Targeting in mHealth Apps: Harnessing Weather Information and Message Framing to Increase Physical Activity

Article

Aug 2023

Mobile technologies provide a unique opportunity for practitioners to identify users’ real-time context and provide personalized interventions to influence their behaviors. However, less is known about a way to improve the effectiveness of mobile health intervention by using context information. This study provides design guidelines on how to use weather information with messaging formats to spur exercise. Through a field experiment that each participant experience different weather conditions in two different treatment periods under the gain or loss interventions, we found that the effects of gain or loss interventions under different weather conditions are heterogeneous. Loss intervention leads to higher fulfillment of exercise goals than gain intervention in sunny weather, whereas gain interventions are more effective than loss interventions in cloudy weather. In addition, we found that weather-based intervention can be used repeatedly over time without losing its effectiveness. Furthermore, we reveal that weather-based intervention is effective toward at-risk populations such as inactive individuals or lower income groups, serving as an mhealth solution that closes the health gap between the haves and have nots. Our findings provide useful guidelines for health service providers and health policymakers regarding how to effectively leverage contextual cues into mobile health intervention.

Mood and Memory

Article

Full-text available

Feb 1981

Gordon Bower

Describes experiments in which happy or sad moods were induced in Ss by hypnotic suggestion to investigate the influence of emotions on memory and thinking. Results show that (a) Ss exhibited mood-state-dependent memory in recall of word lists, personal experiences recorded in a daily diary, and childhood experiences; (b) Ss recalled a greater percentage of those experiences that were affectively congruent with the mood they were in during recall; (c) emotion powerfully influenced such cognitive processes as free associations, imaginative fantasies, social perceptions, and snap judgments about others' personalities; (d) when the feeling-tone of a narrative agreed with the reader's emotion, the salience and memorability of events in that narrative were increased. An associative network theory is proposed to account for these results. In this theory, an emotion serves as a memory unit that can enter into associations with coincident events. Activation of this emotion unit aids retrieval of events associated with it; it also primes emotional themata for use in free association, fantasies, and perceptual categorization. (54 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)

Reminding and mood-congruent memory

Article

Full-text available

Jun 2013
Bull Psychonomic Soc

Earlier experiments found that happy or sad subjects learn more about prose materials that are affectively congruent with their feelings. This study asked whether this memory effect was mediated by selective reminding. Subjects induced through hypnosis to feel happy or sad indicated whether or not happy or sad prompting phrases reminded them of a personal experience; they also rated the phrase’s imagery. Later, free recall showed the mood-congruity effect, which related strongly to imagery ratings but only weakly to remindings. In particular, the differential recall of happy versus sad phrases was about the same regardless of whether the phrases provoked reminiscences. Thus, selective reminding is at best a weak contributor to the moodcongruity effect in memory.

Affect and Information Processing

Article

Full-text available

Dec 1999
ADV EXP SOC PSYCHOL

It is noted that the most active area of research and theory of social information processing that emerged in the past two decades concerns the cognitive determinants and consequences of affect and emotion. This chapter illustrates development of conceptualization that incorporates the implications of diverse phenomena's such as creativity, persuasive messages, impression formation, stereotyping, self-evaluations, and political judgment along with cognitive processes that underlie them. The chapter specifies the possible determinants and consequences of the affect that individuals experience in both laboratory and daily life situations. The chapter considers one basic assumption of conceptualization that essentially distinguishes it from other formulations of affect and cognition. Specifically, it states that although affective reactions can be responses to previously acquired concepts and knowledge that are activated in memory and although one can have concepts about their own and other's reactions, but affect per se is not itself part of the cognitive system. This assumption places restrictions on the ways that affect can influence information processing.

Affective causes and consequences of social information processing

Chapter

Full-text available

May 1994

Mood and persuasion: A cognitive response analysis

Article

Full-text available

Feb 1990

The impact of happy and sad moods on the processing of persuasive communications is explored. In Experiment 1, sad subjects were influenced by a counter attitudinal message only if the arguments presented were strong, not if they were weak Happy subjects, however, were equally persuaded by strong and weak arguments, unless explicitly instructed to pay attention to the content of the message. Subjects' cognitive responses revealed a parallel pattern, suggesting that the findings reflect the impact of mood on cognitive elaboration of the message. In Experiment 2, working on a distractor task during message exposure eliminated the advantage of strong over weak arguments under bad-mood conditions. Good-mood subjects were not affected by a distracting task, suggesting that they did not engage in message elaboration to begin with. It is concluded that subjects in a good mood are less likely to engage in message elaboration than subjects in a bad mood.

Design and Analysis: A Researcher's Handbook

Article

Dec 1976

Being Happy and Seeing "Happy": Emotional State Mediates Visual Word Recognition

Article

Jul 1997

Lexical decision and word-naming experiments were conducted to examine influences of emotions in visual word recognition. Emotional states of happiness and sadness were induced with classical music. In the first two experiments, happy and sad participants (and neutral-emotion participants in Experiment 2) made lexical decisions about letter-strings, some of which were words with meanings strongly associated with the emotions happiness, love, sadness, and anger. Emotional state of the perceiver was associated with facilitation of response to words categorically related to that emotion (i.e. happy and sad words). However, such facilitation was not observed for words that were related by valence, but not category, to the induced emotions (i.e. love and anger words). Evidence for categorical influences of emotional state in word recognition was also observed in a third experiment that employed a word-naming task. Together the results support a categorical emotions model of the influences of emotion in information processing (Niedenthal, Setterlund, & Jones, 1994). Moreover, the result of the word-naming experiment suggests that the effects of emotion are evident at very early stages in cognitive processing.

A Model of Multiattribute Judgments Under Attribute Uncertainty and Information Constraint

Article

Nov 1981

Robert J. Meyer

A multiattribute judgment model is advanced which is purported to characterize how consumers form evaluations of products when information about products' attributes is limited and variability or unreliability is present within attributes. The model is based on the hypothesis that the subjective value consumers associate with a product attribute can be represented in terms of a weighted average of the value of the information they have received about that attribute and the perceived mean value of that attribute within the product class. This average is integrated with the consumer's perception of the amount of variability both within and between other alternatives on that attribute-a measure of uncertainty. A series of laboratory experiments testing aspects of the model are described. Results support both the model's predictions and aspects of its structure.

The heuristic model of persuasion

Article

Jan 1987

S. Chaiken

Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research

Article

Mar 1977

Sometimes IT Just Feels Right: The Differential Weighting of Affect-Consistent and Affect-Inconsistent Product Information

Abstract

Recommended publications

Custeio alvo: uma abordagem conceitual e utilitarista

Duygusal Finansın Kavramsal Çerçevesi: Finansal Kriz ve Balonların Anlaşılmasında Yeni Bir Yaklaşım...

Individual differences in affect and personality predict attentional and conceptual breadth

Cognitive Attraction Theory and Moral Judgment