LIBRARY

Michigan State
University

This is to certify that the

thesis entitled

STRENGTH OF PRODUCT ATTRIBUTIONS AND OPERATION
OF THE DISCOUNTING PRINCIPLE FOR SINGLE VERSUS
MULTIPLE EXPERIENCES WITH A PRODUCT

presented by
JAMES CLAYTON ANDERSON

has been accepted towards fulfillment
of the requirements for

Ph .D . degree in Psychology

 

 

afar-s (1/, ram;

Major professor

 

Date ?//0 /7i
/ 7

0-7639

 

STRENGTH OF PRODUCT ATTRIBUTIONS AND OPERATION
OF THE DISCOUNTING PRINCIPLE FOR SINGLE VERSUS
MULTIPLE EXPERIENCES WITH A PRODUCT

By

James Clayton Anderson

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Psychology

1978

ABSTRACT

STRENGTH OF PRODUCT ATTRIBUTIONS AND OPERATION
OF THE DISCOUNTING PRINCIPLE FOR SINGLE VERSUS
MULTIPLE EXPERIENCES WITH A PRODUCT

BY

James Clayton Anderson

Attribution theory seeks to predict and explain how
peeple arrive at explanations for events which occur in
their environment. A theoretical framework by Kelley (1973)
employed in the research deals with two basic attributional
cases: one in which the attributor has information from
multiple experiences with a person or an object, and the
other case in which the attributor has information from
only a single experience. Kelley posits that attributors
employ a covariance principle in the multiple experience
case whereas the discounting principle and causal schemata
are utilized in the single experience case. The present
research investigated differences in the strength and
confidence of attributions made about an object in the two
basic cases, and secondly, investigated the use of discount-
ing by attributors in the multiple experience case. The
attributional object employed was canned slice peaches in
light syrup.

In the present experiment, subjects' attributions were
investigated in a completely crossed design in which the
independent variables were quality of experience (positive

or negative), possible situational explanation (present or

James Clayton Anderson

interaction, subjects in the negative experience condition
who had prior experience with the peaches rated them as
significantly less sour than subjects who had no prior
experience. Given a positive experience, the prior
experience factor had no effect. Cell means and standard
deviations suggested that some subjects may have discounted
the negative experience while others did not. A path
analysis revealed that the independent variables impacted
the dependent attribute measures through the attribute of
sweetness.

A significant effect was also found for quality of.
experience at each level of prior experience.

So, partial support was found for discounting in the
multiple experience case. Lack of results with regard to
interactions involving the possible situational explanation
was likely due to an ineffective manipulation. While
subjects attended to the message, it was not employed in
a situational manner. An explanation for lack of results
for the confidence ratings was that subjects possessed such
a well deve10ped causal schema about the salient attributes
of canned peaches and their ability to evaluate them, that
this schema overshadowed the manipulations.

The research provided a theoretical underpinning for
the observation that consumers who have a negative experi-
ence with a product with which they have had previous
positive experiences will be more likely to purchase the
product again than will consumers whose first experience

is negative.

James Clayton Anderson

absent), and prior experience (several or none). It was
hypothesized that there would be an effect for quality of
experience, a possible situational explanation x quality of
experience interaction, prior experience x quality of
experience interaction, and a three-way interaction among
these factors.

One hundred twenty-eight students enrolled in intro-
ductory courses in industrial and consumer psychology
participated as subjects. For the prior experience factor,
subjects in the prior experience condition had three
experiences with the product on consecutive days prior to
their fourth experience where the other two factors were
varied. For the quality of experience manipulation, subjects
in the negative experience condition received peaches to
which a citric acid solution was added. Subjects in the
positive experience condition received untreated or good
peaches. Half of the subjects received a possible situa-
tional explanation for why the peaches may not have been
as tasty as others while the other half received no explana-
tion.

Subjects gave ratings of sweetness, overall taste, own
purchase probability, others' purchase probability, own
preference, and others' preference. In addition, subjects
gave confidence ratings for each dependent measure.

Separate multivariate analyses of variance revealed a
significant prior experience x quality of experience
interaction for the attribute ratings, and no significant

findings with regard to confidence ratings. For the

ACKNOWLEDGMENTS

I would like to express my gratitude to those who
have helped me in this research. While I could attempt
individual expressions of gratitude to those below, I know
in each case that the words would fall short of the thoughts
and feelings behind them. So, in their place I will use a
series of simple and deeply felt thank yous.

First, I would like to thank my committee members:

Fred Wickert, Bill Crano, Gil Harrell, Jack Hunter, and
Neal Schmitt.

I would like to thank my research assistants: Steve
Burnstein, Jessica Fleisig, Lynn Garfield, Ken Kurtz, Margo
Rosenthal, and Marsha Tomczyk.

I would like to thank Lynn Honaker for her typing and
general support.

Finally, while not directly involved in this research,
I would like to thank my parents for their support throughout

my graduate studies.

ii

TABLE OF CONTENTS

Page

LIST OF TABLES O O O 0 I O O O O O O O O O O O 0 O O 0 iv
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . v
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . v
WIT-10D o o e o o o o o o o o e o o o o e o o e o e o o 1 8
Subjects . . . . . . . . . . . . . . . . . . . . 18
Procedures . . . . . . . . . . . . . . . . . . . 18
Design . . . . . . . . . . . . . . . . . . . . . 22
Analyses . . . . . . . . . . . . . . . . . . . . 24
RESULTS. . . . . . . . . . . . . . . . . . . . . .g. . 26
Preluminary Analyses . . . . . . . . . . 26
Multivariate Analyses of Variance . . . . . . . . 27

Path Analysis . . . . . . . . . . . . . . . . . . 30
DISCUSSION . . . . . . . . . . . . . . . . . . . . . . 35
APPENDIX A . . . . . . . . . . . . . . . . . . . . . . 39
APPENDIX B . . . . . . . . . . . . . . . . . . . . . . 42

LIST OF REFERENCES . . . . . . . . . . . . . . . . . . 45

iii

LIST OF TABLES

Table Page

1. The Attribute of Sweetness as a Function of
Prior Experience and Quality of Experience . . 29
2. Confidence in Attribute Ratings . . . . . . . . 32

3. Correlation Matrix for Path Analysis, and
Residual Matrix of Observed Minus Predicted
Correlations . . . . . . . . . . . . . . . . . 33

iv

LIST OF FIGURES

Figure

1. Information Patterns for the Three Basic
Attr ibut ions 0 O O O O O O O O O O O O

2. Multiple Sufficient and Multiple Necessary
Causal Schemata . .

3. Design for the Research . . . . . . .

4. Path Medel . . . . . . . . . . . .

Page

23
34

INTRODUCTION

Attribution theory seeks to predict and explain how
people arrive at explanations for events which occur in their
environment. An event may be causally explained in several
ways. A person can assign a quality or characteristic to
him/herself, or to another person as a causal explanation
for some particular behavior. As an example, a batter may
attribute striking out to his own inability to hit a curve-
ball, or alternately, he may attribute striking out to the
pitcher possessing an exceptional curveball. A person can
ascribe qualities or characteristics to an object which
serve as a causal explanation for his/her response or
reaction to the object. Continuing with the baseball
example, the batter may explain his striking out by ascribing
the characteristic "has a foreign substance on it" to the
baseball. Alternately, a person may assign causality to some
particular circumstances which were present in the situation
when the effect to be explained occurred. 50, the batter may
attribute striking out to being momentarily blinded by the
sun, or to a distraction in the stands.

A theoretical framework which encompasses the above
attributions has been put forth in several papers by Kelley

(1967, 19723, 1972b, 1973). Kelley's statements of

2
attribution theory deal with two basic attributional cases:
one in which the attributor (that is, the person making the
attribution) has information from multiple observations
or experiences with a person or an object, and the other
case in which the attributor has information from only a
single observation or experience. The present research is
concerned with attributions to objects rather than persons,
and the theoretical statements and empirical research will
be limited largely to this perspective.

In the multiple experience case Kelley posits that a
person utilizes a covariance principle (Kelley, 1972; 1973)
to arrive at an attribution. The covariance principle is
that "an effect is attributed to the one of its possible
causes with which, over time, it covaries" (Kelley, 1972a,
p.3). The attributor applies three covariance criteria of
distinctiveness, consensus, and consistency to his/her
experiences with a given object to decide on an appropriate
explanation. Distinctiveness refers the extent to which a
particular reaction or response is uniquely associated with.
a given object as Opposed to other similar objects. Con-
sensus is the extent to which other persons experience the
same reaction or response to the object, and consistency
is the extent to which the attributor's response to the
object remains the same or nearly so over time. After
examining the experiences with the object with regard to
these criteria, an appropriate attribution is made to the
person him/herself, the object, or the situation (or

possibly some combination of the three).

To illustrate how different information patterns result
in attributions to an object, the person, or to the situation;
consider the event, "George enjoyed the science-fiction
movie." Suppose that we learn that George rarely likes any
science-fiction movie he sees. Thus, his response to the

particular movie is highly distinctive. In addition, we

 

learn that several friends who accompanied George to the

movie also enjoyed it. There is high consensus about the

 

movie. Finally, suppose we learn that George enjoyed the
movie as much when he saw it at home on cable television as
when he saw it at the theater. His response to the movie

would be seen as highlyconsistent. From this pattern of

 

”distinctiveness, consensus, and consistency information, an
appropriate explanation of the event would be an object
attribution such as, "The movie is an exceptional science-
fiction picture."

Instead of this pattern, suppose we learn that George
likes almost every science-fiction movie he sees, and that
none of his friends who accompanied him enjoyed the movie.

Now we have low distinctiveness and low consensus information.

 

 

As before, let's assume that George enjoyed the movie both

times he saw it. So, again there is high consistency. With

 

this pattern of information, an appropriate explanation
would be a person attribution such as, "George is a science-
fiction movie buff.

Finally, let us assume that we know that George rarely
likes any other science fiction movie he sees, and that

none of his friends enjoyed the movie. Here we have an

information pattern of high distinctiveness and low consensus.

 

 

Further, assume that George did not enjoy the movie the

second time he saw it, indicating low consistency. From

 

this pattern of information, an appropriate explanation would
be a situational attribution such as, "The movie gave George
a much-needed break to relax from studying for final exams."
The information patterns with the resultant attributions are

‘summarized in Figure l.

 

Information Pattern

 

 

 

 

 

Distinctiveness Consensus Consistency Attribution
High High High Object
Low Low High Person
High Low Low Situation

 

Figure 1 Information Patterns for the Three Basic Attributions

Kelley (1967) has pr0posed an analysis of variance analog
for the way in which individuals analyze their state of infor-
mation regarding an object. The distinctiveness criterion
corresponds to the numerator or between condition term in the
F-ratio while the consensus and consistency criteria corres-
pond to the denominator or within condition term. This hm-
plicit analysis of variance determines the stability of an

object attribution, or confidence in the validity of the

attribution. To the extent that an object is distinctive
from other objects (thus, large between object variance)
and/or reactions to it are of high consensus and consistency
(small variance within people, or small variance over time),
attributions to the object will be stable. Kelley makes
clear that this attributional analysis is subjective in
nature due to perceptual and cagnitive error.

In the second attributional case where an individual
has just a single experience with an object Kelley (1972b,
1973) puts forth two concepts to account for the attribution
process: the discounting principle and causal schemata
(singular: schema). The discounting principle is simply
stated as, "The role of a given cause in producing a given
effect is discounted if other plausible causes are also
present" (Kelley, 1973, p.113). For example, the role of a
favorable attitude toward a product in producing an individ-
ual's endorsement of it would be discounted if it was thought
that the individual had received money for the endorsement.

A causal schema is a conception about the way in which
two or more possible causes operate in relation to a
particular kind of effect. A causal schema is learned from
past experience with similar cause--effect patterns. Kelley
believes that causal schemata are employed in the more
common, informal attribution situations. Continuing with
his analysis of variance analog Kelley (1972b, p.152) states,
"Given information about a certain effect and two or more
possible causes, the individual tends to assimilate it to

a specific assumed analysis of variance pattern, and from

that to make a causal attribution." To the extent that a
relevant causal schema is well deve10ped, it can be argued
that a causal inference will be made in the single experi-
ence situation with as much stability and confidence as in
the multiple experience case. While Kelley (1972b) presents
a number of causal schemata, for the present purposes, only
the multiple sufficient causes schema and its counterpart,
the multiple necessary causes schema, need be discussed.

'In the multiple sufficient causes schema the presence
of either one or both of two possible causes is adequate to
produce a given effect. An example of this schema is a
brand endorsement by a "person-in-the-street", presented in
Figure 2. Here it can be seen that the presence of either a
monetary payment or a favorable brand attitude would be ade-
quate to induce a brand endorsement, and that the presence
of both possible causes would likewise produce an endorsement.
So for the multiple sufficient causes schema, given knowledge
of the presence of an effect and one possible cause, an
individual would be uncertain as to the presence or absence
of a second possible cause. The discounting principle is
proposed to operate for this schema.

In the multiple necessary causes schema the presence of
both causal factors is required to produce a given effect.
This schema is illustrated in Figure 2 as a brand endorse-
ment by John Wayne. Since John Wayne is known to be stead-
fastly honest and very wealthy, the presence of a monetary
payment alone would not be adequate to produce his endorse-

ment of a brand. Alternately, the presence of a favorable

Multiple Sufficient Causes Schema

 

for "person-in-the-street":

 

 

 

 

 

 

Brand Brand
Mbnetary Present Endorsement Endorsement
Payment
(or perhaps .
desire to be . Brand
on television) Absent Endorsement
Absent Present

Favorable Brand Attitude
Multiple Necessary Causes Schema

for John Wayne:

 

 

 

 

 

.Brand
Menetary Present Endorsement
Payment
Absent
Absent - Present

Favorable Brand Attitude

Figure 2 Multiple Sufficient and Multiple Necessary Causal
Schemata

 

attitude alone would not be adequate to induce an endorse-
ment since it is known that actors of John Wayne's caliber
are highly paid for product endorsements. Rather, a
favorable attitude toward the brand and a large monetary
payment would be necessary to induce John Wayne to give a
brand endorsement. For this schema, given knowledge of the
absence of an effect, an individual would be uncertain if
either cause were singly present. Naturally, these schemata
can be generalized to cases of more than two possible causes,
and to cases of perceived degrees of presence for causes and
effects.

Fishbein and Ajzen (1972) in their review of the
attribution literature observed that "few of the recently
published studies seem to be direct extensions or tests of
the underlying theory" (p.502). Kiesler and Munson (1975)
in a later review reiterate Fishbein and Ajzen's comment. My
research addressed a basic question in attribution theory and
secondly, explored a basic extension to the theory. Specifi-
cally, a comparison was made between the multiple experience
case and the single experience case with regard to the
strength and certainty of attributions made about an object.
Is there a difference in the strength and confidence of
attributions made in the two basic cases? Secondly, the use
of discounting by attributors in the multiple experience
case was investigated. Before specific research hypotheses
are presented, related research will be reviewed.

McArthur (1972) investigated how information data patterns

which varied on the criteria of distinctiveness, consensus,

and consistency affected causal attributions. She presented
subjects with statements about the occurrence of some response
by another person ("John laughs at the comedian.") followed
by three statements which provided information about high

or low degrees of consensus (Almost everyone[Hardly anyone]
who hears the comedian laughs at him."), distinctiveness
("John does not [also] 1augh[es] at almost every other
comedian."), and consistency ("In the past John has almost
always [never] laughed at the same comedian."). For each
set of statements, the subject was asked to decide what
probably caused the event to occur from four alternative
causes; something about the person, the stimulus, the
particular circumstances (situation), or some combination

of the three causes. McArthur found that the frequency of
person attributions was greater for low than for high con-
sensus information, for low than for high distinctiveness
information, and for high than for low consistency informa-
tion. The frequency of stimulus attributions (of which some
were objects) was greater for high than for low consensus
information, for high than for low distinctiveness informa-
tion, and for high than for low consistency information.
Thus, McArthur's data support the predictions for person
attributions and for stimulus attributions which can be
derived from Kelley's theory. MCArthur further found that
the frequency of circumstance (situational) attributions was
greater for high than for low distinctiveness information, and
for low than for high consistency. Consistency information

had the most impact on circumstance attributions followed by

10

distinctiveness and consensus information respectively.

Orvis, Cunningham, and Kelley (1975) provide further
insight into the causal attributions which are made on the
basis of information patterns that vary on the attribution-
al criteria. In addition to the complete information
patterns such as those used by McArthur (1972), these
researchers presented subjects with incomplete patterns
with information on one or two criteria unknown. Half of
the subjects were asked to decide on causal attributions
from several presented factors (following McArthur, 1972),
and the remaining half were asked to judge the value (from
high to low on 7 point scales) of the unknown information.
Results relevant to the present study were that high
distinctiveness and high consistency information yielded a
significantly greater frequency of stimulus attributions
than other attributions (p<.001), and that high distinctive-
ness and low consistency information yielded a significantly
greater frequency of circumstance or situational attribu-
tions than other attributions (again, 23.001). Orvis et al.
also conclude from their data that consistency information
has more effect on attributions than does consensus or
distinctiveness information primarily because of its strong
impact_on situational attributions.

Ruble and Feldman (1976) have found some evidence of a
presentation order bias against consensus information.
However, this finding does not impact the present research

for the following reasons. Subjects in the present research

11

received information from direct experience with the object,
and consensus information was not known. Finally, Ruble and
Feldman (1976) also found consistency information to be most
determinant of situational attributions, regardless of
order.

Turning from the multiple experience case to the single
experience case, a study by Jones, Davis, and Gergen (1961)
demonstrates the discounting principle and the multiple
sufficient causes schema. In this study subjects listened
to a tape-recorded job interview between a psychologist and
a student. While subjects were aware that the student would
be instructed to play the role of a job candidate for either
a submariner or astronaut position, they were unaware that
the recordings were made from carefully constructed scripts.
The psychologist began each interview with a clear role
description of behaviors desirable in a submariner (other-
directed behaviors) or an astronaut (inner-directed behav-
iors). Then to a series of clearly relevant job choice
items, the student either gave answers concordant with the
earlier role description (in-role-behavior) or answers which
were discordant (out-of—role behavior). Subjects were asked
to rate what they thought the student was "really like",
and to indicate how confident they were of their ratings on
a 5-point confidence scale. So, subjects were basically
asked to make an attribution of the student's behavior to
an internal cause (a personality disposition) or to an
external cause (role-constrained behavior). In the case of

the students who enacted in-role behavior, the external cause

12

is known to be present and facilitative of the given be-
havior. Subjects would be uncertain as to the presence of
a personality disposition, and the theory would predict
that subjects would discount it as a possible cause. In
the case of the student who enacted out-of-role behavior,
the external cause is again known to be present, but in-
hibitive of the given behavior. When behavior occurs in
spite of the presence of an inhibitory cause (role demands),
the theory would predict that subjects would be more certain
of a personality disposition (than if the external inhibi-
tory cause were absent). This is known as the augmentation
principle (a reverse of the discounting principle) and can
be stated as follows: "If for a given effect, both a
plausible inhibitory cause and a plausible facilitative
cause are present, the role of the facilitative cause in
producing the effect will be judged greater than if it alone
were present as a plausible cause for the effect" (Kelley,
1972a, p.12). Jones et al.'s results support these predic-
tions. The students who displayed out-of—role behaviors
were rated as significantly more revealing of their actual
dispositions than students who displayed in-role behaviors.
In addition, subjects were significantly more confident in
assigning trait dispositions to the students who displayed'
out-of-role than in-role behavior.

Recent research by Skaklee (1976) suggests a modifi-
cation in the discounting principle. She proposed and found
empirical support for a principle of minimum causation.

"Once an event is sufficiently explained, other possible

13

factors are superfluous" (Shaklee, 1976, p.4764-B). The
relevance of this finding is that to the extent that there
is a sufficient situational explanation for a given effect,
attributors should not seek out or utilize other possible
causal explanations (e.g., an object attribution).

Several studies by Irwin and Smith (1956, 1957; Irwin,
Smith, and Mayfield, 1956) provide some evidence as to which
of the two attributional instances, single versus multiple
experiences, would lead to stronger and more confident
attributions. Two basic procedures were employed. In the
first procedure, subjects were asked to decide if the average
of a pack of cards was greater than or less than zero. For
each card shown, the subject gave his/her judgment and a
confidence rating for the judgment. The results were that
subjects were more confident in their judgments after the
20th card than the 10th card; and for sets of cards with
larger absolute means (high distinctiveness), and smaller
standard deviations (high consistency). In a variant of
the above procedure,.subjects were asked to decide which of
two packs of cards had the higher mean. Subjects were shown
a card from each pack, and gave a judgment with a confidence
rating. Again, the subjects had greater confidence about
their judgments after the 20th pair of cards than the 10th
pair; and for pairs of packs which had larger mean differ-
ences (high distinctiveness) and small standard deviations
(high consistency).

From these results that subjects were more confidant

after more experience, it might be argued that multiple

14

experiences would lead to stronger and more confident
attributions about an object. However, upon further thought
it can be effectively argued that the above research did

not yield an adequate test because the task was biased
against causal schemata. The tasks employed were such that
subjects had no prior experience with them and could not
have possessed a relevant, well deve10ped schema. In fact,
it was a condition of the research that subjects have no
training in statistics to aid them in revising their judg-
ments!

The present research provided a more adequate test of
the multiple versus single experience hypothesis in that the
experience with an object that was employed was one where
participants possess causal schemata. The object employed
was a new product from a well established product category;
canned peaches. Thus, the present research investigated
differences in the strength and confidence of attributions
made by participants who had several experiences with the
new product, and by participants who had only a single
experience with the new product but could draw upon a
relevant causal schema.

While there is no research which directly bears upon
discounting by attributors in the multiple experience case,
research by Beckman (1970) suggests that discounting would
be utilized in the multiple experience case. Beckman
employed a participant condition where subjects instructed
two fictitious students on some mathematical concepts and

symbols for four trials. One student always performed well

15

(high-high) while the second student's performance either
deteriorated (high-low), improved (low-high), or remained
poor (low-low). Following feedback from four sets of
problems that the student "solved", subjects were asked to
make attributions about the student's performance. Subjects
in the high-low condition attributed the student's perfor-
mance to situational factors with greater frequency than any
other causal factor.

Beckman's research suggests that when participants in
the present study had several positive experiences with a
new product followed by a negative experience, they would
attribute the negative experience or change to some situa-
tional factor. Further, if a possible situational explana-
tion for the negative experience were presented to partici-
pants, discounting would be predicted with participants
making more confident situational attributions with resul-
tant less harsh evaluation of the product itself. The pre-
diction that participants who possessed a possible situation-
al explanation would be less likely to make an object
attribution (negative product evaluation) than participants
without an explanation is also supported by the principle
of minimum causation. To test these hypotheses, a quality
of experience factor (positive or negative) was crossed with
a possible situational explanation factor (present or absent).
This is the familiar multiple sufficient causes schema.
These two factors were then crossed with prior experience
factor (several or none) to provide a design in which both

research questions could be addressed. Those were:

16

Which attributional case, single or multiple experiences,
leads to more extreme and confident attributions, and do
attributors employ the discounting principle in the multiple
experience case?

Specifically, from the theoretical statements and
research presented, the following research hypotheses were
made:

First Hypothesis: The quality of experience factor has
a significant effect on the extremity of attribute ratings
for the new product, and on the probability of purchase.
Subjects in the negative experience condition will be harsher
in their evaluations and have a lower probability of purchase.

Second Hypothesis: There is a significant interaction
between the possible situational explanation factor and the
quality of experience factor with regard to extremity and
confidence of attribute ratings, and purchase probability.
Subjects in the negative experience condition who are given
a possible situational explanation will be less harsh and
confident in their ratings with a resultant higher purchase
probability than subjects who receive no explanation. In
the positive experience condition the possible situational
explanation factor should have no significant effect.

Third Hypothesis: There is a significant interaction
between the number of prior experiences factor and the
quality of experience factor with regard to extremity and
confidence of attribute ratings, and purchase probability.
Subjects in the negative experience condition who have been

given several prior experiences will be less harsh and

confident in their ratings with a resultant higher purchase

probability than subjects who have no prior experience.

The prior experience factor should have no significant

effect for subjects in the positive experience condition.
Fourth Hypothesis: There is a significant three-way

interaction with regard to extremity and confidence of

attribute ratings, and purchase probability. Discounting

by subjects who are given a possible situational explanation

(possible situation explanation x quality of experience

interaction) will be significantly greater for subjects who

have prior experience than for subjects with no experience.

METHOD

Subjects

 

One hundred twenty-eight students enrolled in intro-
ductory courses in industrial and consumer psychology at
Michigan State University participated as subjects, and

received extra-credit points for participation.

Procedure

 

The attributional object which was used in the research
was Avondale canned peach slices in light syrup. Subjects
were told that a private company was Sponsoring new product
research for a line of canned fruit in light syrup, and that
they would be trying the new sliced peaches in light syrup.
Using peaches in light syrup as Opposed to the more common
heavy syrup, and the positioning of them as a new product
was done to give a distinctive nature to the attributional
object.

In the prior experience conditions, subjects had three
experiences with the product on consecutive days prior to
their fourth experience where the other two factors were
varied. On their first experience with the product, subjects
were told that in new product research it is necessary to
identify the characteristics or qualities of a product

which are important to consumers. Subjects were asked to

18

19

think of qualities or characteristics of canned sliced
peaches which were important to them. They were then asked
to list the Opposite of the characteristic next to it.
Subjects then tasted the peaches (3 slices), and repeated
the procedure to elicit any attributes they may have over-
looked. From this exposure, bi-polar attributes of canned
sliced peaches which were salient to the subject sample were
elicited. These attributes were used to rate the product on
the fourth and final experience, or the only experience for
the no prior experience condition.

On their second experience with the product, subjects
rated ten possible brand names (e.g., Meadow Grove) and eight
possible slogans (e.g., "The perfect treat, light and sweet.")
for the new product on a five point like very much-~dislike
very much scale, and tasted the peaches again. On their
third experience, subjects rated eight possible marketing
strategies (e.g., a cottage cheese promotional tie-in) on a
five point very effective--very ineffective scale, and again
tasted the peaches. The tasks were chosen to be plausible
but not too involving for the subjects, and served only as
a means for subjects to get experience with the product.
Copies of the three tasks are presented in Appendix A.

In the fourth experience for subjects in the prior
experience condition and the only experience for subjects
in the no prior experience condition, half of the subjects
.in each condition received peaches which were positive

(conSistent with experience) and half received peaches which

20

were negative (inconsistent with past experience). Twenty-
three milliliters of a 39 percent citric acid solution were
added per 29 ounce can of peaches in the negative experi-
ence condition to give the peaches an unpleasant, sour taste.
Research by Peretz (1974) has shown that adding citric acid
is an effective means for varying the liking of canned
peaches. The level employed in the present research was one
which exceeded by 5 grams the level at which there was unani-
mous agreement by the author and six research assistants that
the peaches were bad tasting. Subjects in the positive ex-
perience condition received unadulterated peaches.

For the possible situational explanation manipulation,
half of the subjects received a possible situational explana-
tion for why the peaches may not have been as tasty as others.
while the other half of the subjects received no explanation.
The possible situational explanation appeared on a separate
mimeographed paper and had the appearance of being rather
hastily done. The explanation used was as follows:

These peaches may not be as tasty as others
we've had because the lot which was used may
have been picked too early. However, if the
company decides to go into test market, they're
going to have inspectors go to the groves to
insure the ripeness and quality of the fruit,
even though it may cost the company a little
more.

Subjects in all conditions then rated the product on
the elicited bi-polar attributes, and gave an overall taste
rating. The five bi-polar attributes and overall taste scale
employed were: firm texture--mushy texture, heavy syrup--

light syrup, good "peach" color--discolored, sweet-ssour

ripe--not ripe, and overall tasty--overall not tasty. The

21

direction of scale polarities was randomly assigned for

each attribute, and the scale order was randomly determined
with the overall taste rating appearing last. A confidence
rating was obtained for each of the above ratings on a
confident--not confident scale. Seven point scales with
anchors of extremely, moderately, somewhat, and neither were
used. The rating scales were preceded by the following
instructional statement: "Assuming that the company decides
to go to test market, how do you think the product would be
perceived (rated on the following characteristics)? In
addition, please indicate on the provided scale how confident
you are of each rating." A rating of purchase probability
(own purchase probability) was obtained on a seven point
probable--improbable scale, and was preceded by the following
instructional statement: "Assuming that the company decided
to bring this product on the market and that you needed to
purchase a can of sliced peaches, what would be the proba-
bility that you would purchase this brand?" A confidence
rating was then obtained for the probability ratings. In
addition, subjects were asked 1) what percentage of other
peOple who tried the new peaches would purchase them (others'
purchase probability), 2) their preference on a five point
preference scale for the new peaches in light syrup as
Opposed to peaches in heavy syrup (own preference), and 3)
what percentage of other people who had tried the new peaches
would prefer them to peaches in heavy syrup (others' pre-

ference). The measures of the subject's own preference

22

and his/her estimate of other pe0ple's preference were
included to be parallel measures of the subject's own pur-
chase probability and his/her estimate of other people's
purchase probability. Confidence ratings were also obtained
for each of these three measures. Demographic variables of
sex, age, on- or off-campus residence, and frequency of food
shopping were obtained for use as possible covariates.

A copy of the research instrument appears in Appendix
B. Following its completion, subjects answered open-ended
questions about what led to their purchase probability judg-
ment, what led to their judgment of others' purchase proba-
bility, and their thoughts as to what accounted for the way

the peaches tasted.

Design

Three factors, each with two levels, were manipulated
in the present research. These were: number of prior
(positive) experiences with the product, three or none;
quality of experience with the product, positive or negative;
and possible situational explanation, present or absent. The
three factors are fixed, and completely crossed to provide
a 2x2x2 factorial design. This design is presented in Figure
3.

Sixteen subjects were randomly assigned to each of the
eight conditions, and were considered a random factor. A
statistical power analysis (Cohen, 1977) of this design
for this number of subjects revealed a power of approximately
.80 to detect a main effect or interaction effect of approxi-

mately 6% of the variance, a medium-sized effect.

23

Prior Experience

 

 

 

No Prior Prior
Present
Possible
Positive Situational
Explanation
Absent
Quality
of
Experience
Present
Possible
Negative Situational
Explanation
Absent

 

Figure 3 Design for the Research

24

Analyses
A modified version (Scheifley and Schmidt, 1973) of

MULTIVARIANCE (Finn, 1972) was utilized to perform a prelim-
inary multivariate analysis of covariance to determine if
any of the demographic variables warranted inclusion as
covariates. Following this, two separate multivariate
analyses of variance were carried out. One analysis was
performed on the dependent measures of sweetness, overall
taste, probability of purchase, and preference, and the
second analysis was performed on the confidence ratings for
these dependent measures. A univariate analysis of variance
was also carried out on the dependent measure of ripeness.
This was done as a partial manipulation check on the
possible situational explanation factor in that subjects who
received the explanation should rate the peaches as signifi-
cantly less ripe than the subjects who received no explana-
tion.

A two stage significance-testing approach recommended
by Hummel and Sligo (1971) was employed. If the overall
multivariate null hypothesis was rejected, the univariate
F's were interpreted to determine which dependent measures
were significantly affected. ApprOpriate simple main effects
were computed for significant interactions.

A path analysis (Heise, 1975) was also performed. In-
dependent variables and interactions which were significant
as well as nonsignificant independent variables involved in
significant interactions were dummy coded and treated as

exogenous variables. The dependent measures of sweetness,

25

overall taste, own purchase probability, others' purchase
probability, own preference, and others' preference served

as endogenous variables. The PATHPAC program by Hunter and
Hunter (1977) was utilized for the analysis. With this
program, causal influence indicators are specified which
correspond to hypothesized causal relationships within a
model. From these indicators path coefficients are obtained
using ordinary least-squares estimation. Stated differently,
path coefficients are beta weights derived from a set of
multiple regressions on the posited relationships within a
model. The path coefficients are used to generate a correla-
tion matrix which, in turn, is subtracted from the observed
correlation matrix to provide a residual matrix from which

the goodness of fit of a pr0posed model can be judged.

RESULTS

Preliminary Analyses

 

Correlational analysis indicated that while the dependent
measure of Own Preference was related to the measure of Own
Purchase Probability (r=.4l, p5.01), the measures were not
parallel. The correlation of Own Purchase Probability with
Overall Taste (r-.7l) was significantly greater than the
correlation of Own Preference with Overall Taste (r-.35;
Ediff.=4.42, 25.01), and the correlation of Own Purchase
Probability with Sweetness (r=.60) was significantly greater
than the correlation of Own Preference with Sweetness
(r-.28; Ediffo'3.48,‘2<.01). In addition, it was found that
the correlation of Own Purchase Probability with Others' Pur-
chase Probability (r-.62) was significantly different from the
correlation of Own Preference with Others' Preference (r=.25;
difference-.37, 25.01). Since the analyses indicated that
the measures should not be combined to form a two item cluster
as originally was haped, Own Preference was treated as a
separate measure, and a separate analysis of variance was
performed on it. This analysis revealed that the independent
variables had no significant effect on Own Preference, either
as main effects or interactions.

An analysis of variance was also performed on the depen-

dent measure of Ripeness as a partial check for the possible

26

27

situational explanation manipulation. This analysis revealed
a significant main effect for the possible situational expla-
nation factor, F (1,120) = 20.37, p<.0001. Subjects who
received the explanation rated the peaches as significantly
less ripe. Smaller main effects were found for quality of
experience, 5 (1,120) a 7.10, p<.0088, and prior experience,
.E (1,120) = 4.50, 25.0361 with subjects who received the nega-
tive experience, and subjects who had prior experience rating
the peaches as significantly less ripe. The possible situa-
tional explanation factor accounted for the most variance
in Ripeness ratings (approximately 14%) with quality of
experience (approximately 4%) and prior experience (approxi-
mately 3%) accounting for much less. So, the results suggest
that the subjects attended to the ripeness aspect of the
explanation.

The results of a multivariate analysis of covariance
indicated that none of the demographic variables warranted

inclusion in the analyses of variance as covariates.

Multivariate Analyses of Variance

The results of the multivariate analysis of variance
for the dependent measure set of Sweetness, Overall Taste,
and Own Purchase Probability will be presented first. The
multivariate null hypothesis for the prior experience X
quality of experience X possible situational explanation
interaction was not rejected, §_(3,118) = .68, 25.5677.

The multivariate null hypothesis for the prior experi-

ence X quality of experience interaction was rejected,

Z8

‘5 (3,118) = 3.61, p<.0155. An examination of the univariate
F's revealed that the interaction was significant only for
the dependent measure of Sweetness, F (1,120) = 4.92, p<.0284.
The cell means for the attribute of sweetness and the simple
main effects for the interaction are presented in Table 1.

It can be seen for Table 1 that when given a negative
experience, subjects who had prior (positive) experience
rated the peaches as significantly less sour than did sub-
jects who had no prior experience. Given a positive experi-
ence, there is no significant difference between the no prior
and prior experience conditions. In the negative experience
condition, subjects in the prior experience condition have
a larger standard deviation (5 = 2.24) than subjects in the
no prior experience condition (5 = 1.58). This finding,
along with the pattern of means, is consistent with an
explanation that some subjects in the negative experience--
prior experience condition discounted the negative experience
(and gave a sweetness rating consistent with prior experience)
while other subjects did not (and gave a corresponding low
sweetness rating). At the positive experience condition
the difference in standard deviations can be explained by
sampling error variance in the sweetness of peaches over
four experiences. There is also a significant effect for
quality of experience at each level of prior experience.
Subjects in the negative experience condition rated the
peaches as significantly less sweet.

The multivariate prior experience X possible situational

29

Table l The Attribute of Sweetness as a Function of
Prior Experience and Quality of Experience

 

Table la. Mean sweetness, l=extremely sour, 7=extremely
sweet. Cell standard deviations in parentheses.

Prior Experience

No Prior Prior
P 't‘ 5.50 5.16
031 1V3 ( .92) (1.48)
Quality of
Experience
. 3.03 3.97
Negative (1.53) (2.24)

Table lb. Simple main effects.

No Prior vs. Prior at Positive, F(l,120) =.7l, p>.25

No Prior vs. Prior at Negative, F(1,120) =5.27, p<.025
Positive vs. Negative at No Prior, F(1,120) = 36.56, p<.001

Positive vs. Negative at Prior, FC1.120) = 8.46, p<.005

 

30

explanation interaction was not significant, F (3,118) =
1.88, pfa1367. A significant interaction was not pre-
dicted.

The multivariate null hypothesis for the quality of
experience X possible situational explanation interaction
was not rejected, F (3,118) = .68, 23.5644. So, the second
hypothesis did not receive empirical support.

The multivariate main effect for the possible situational
explanation factor was not significant, F (3,118) = .66, p<
.5811. A significant main effect was not hypothesized.

No significant main.effects or interactions were found
for the analysis of variance of the confidence ratings for
Sweetness, Overall Taste, and Own Purchase Probability. The
cell means and standard deviations for the confidence ratings
are presented in Table 2. Table 2 shows that the cell means
are uniformly high. The flatness of the mean confidence

ratings across cells explains the lack of significant effects.

Path Analysis

 

Prior experience, quality of experience, and the prior
experience X quality of experience interaction were dummy
coded and employed as exogenous variables in the path analysis.
The correlation matrix which was input to PATHPAC appears in
Table 3a. Several models were estimated, and the one which
provided the best fit of the data is presented in Figure 4
with the residual matrix shown in Table 3b.

The exagenous variables impact the endogeneous variables

through the dependent measure of sweetness. Sweetness and

31

 

 

 

 

 

 

.ueoowmcoo muoEoHHXe u s .ucoowmcoo no: xHoEouuxo u H "oaoz
4mm. mmfi.o songwaenona Heq.H coo.o spflfinnmnoea
amazohzmvnzo u omenupzm :30 o
mmo. mom.o mummh Haeuo>o 0 “Nu. mom.o momma Haene>o u
mam. mmo.o mmocnoezm u see. cmm.o mmoeuoozm u
:OMueemmem oz oocowaomxm
:ofiueeeamxm Hanowuesuwm canwmmom o>wuewoz
eNH.H moo.e auwHwnenoum hm~.H mmm.m xufiafinmnoum
. omegowsm :30 u emegousm :3o 0
mac.~ mNm.c memes aaeuo>o u mam. mNH.c mamas Haepo>o o
mmH.a mam.o mmocumozm u muo.H ocm.m mmoeuoozm 0
NBA. wmm.m Regannenoee mNS. mmm.m sonafinmnoua
omenowam :30 u ommaounm :30 o
amo. wmv.e mummy Haeuo>o u mam. moc.o mummy Haeuo>o u
owe. moo.o mmoeuoozm o Ham. mma.o mmoeuoozm o
GOMumcemem oz oonofihonxm
monomeaamxm aaeofluazoflm ofiafimmom o>noumoa
aOH.H mmmfi.e suednnenota Ham. mmm~.o Runawnmnoea
ommnuusm :30 u omesonsm :30 u
awe. mmo.o momma Haeuo>o u Ham. awwH.o memes Haeno>o 0
«mm. moo.o mmoeuoozm 0 «ma. mmmH.o mmonuoozm 0
nm_ x mm x
ooaownomxm HOAHQ cocoanomxm AOAHm oz
mmcfiuem euznwpuu< cw cocoowmcou N canes

32

cc. cc. Hc. ~c.- mc.- cc. mc. Nc.- ma. oocohowowm .mpoguo

cc. cc. NH. ea. «a. me. me. He.- ~c.- monogamocm ego
ac. NH. cc. cc. ma. cc. Nc. cc.- Nc. xuwawnmnonm
ommgohzm .muonuo
~c.- cc. cc. cc. cc. cc. ca. mH.- mc. xufiawnmnonm
ommnouac :30
me.. as. ma. cc. cc. co. mH. SH.- Ac. momma Hflmcm>o
cc. mc. cc. cc. cc. cc. cc. cc. cc. mmocuoozm
mo. me. Na. an. ma. co. cc. cc. co. .m.o x .m.a
Nc.- Hc.- oc.- mH.- SH.- cc. cc. co. co. ooemfieoexm
. mo sondesc
MA. No.- No. me. ha. co. cc. cc. cc. ouconceexm tonne
.Ha.oo .ta.zo .sa.oo .=E.:o .e.o .m .m.ox.m.m .m.c .m.a

meoflumaouuou couowcohm man“: co>uomno mo Rayon: Hescflmom .nm canes

cc.H cm. cm. mm. ca. ca. mc.. ac. ma. cocoaomoum .muoguc
em. oc.~ an. Ne. am. am. .Hc.- HH. Ha. consummate azo
cm. a». cc.H am. mm. m4. mo.- ca. ma. sonfifinenota
ommgonsm .muoauo
aw. Ne. mm. co.~ HR. am. He. MA. ac. saucepanoea
ome:OH=m :30
ma. em. mm. HA, oc.~ be. we. ac. an. momma Hflaeo>o
”H. mN. me. Am. as. cc.H AH.- as. we. mmoeoomzm
me. Ho.- ma.- .:.. we. .5H.- co.~ co. cc. .m.c x .m.a
he. HA. an. ma. be. me. cc. co.H cc. «unannoaxm
. mo sonflaac
ma. Ho. mo. we. AH. mo. as. ea. oc.H muemueoaxm gonna
.ta.oo .pa.zo .aa.oo .sa.2o .e.o .m .m.ox.m.a .m.c .m.a

Hanan: nowueaouuou unacH .en mance

 

mewumHonhoo wouowconm
mesa: co>uomno we xfiwumz szcwmom can .mfimxae:< spam you Kayne: :owuefiouhou m oaseh

 

 

Prior Experience

 

l=no prior 2=prior

33'

 

 

Quality of Experience
P.E. X Q.E.

 

 

l=positive 2=negative

 

 

 

 

 

 

Overall Taste

.49

.07 -.17

.47 .30

 

Own Purchase

 

 

Figure 4 Path Mbdel

Probability
.57

 

 
 
  
   
   
   
   

.59

   

Others' Purchase
Probability

Own
Preference

 

 

  

Others'
Preference

34

Overall Taste, in turn, impact Own Purchase Probability

which predicts both Others' Purchase Probability and Own
Preference. Others' Purchase Probability and to a lesser
extent Own Preference account for Others' Preference. It

was expected that Preference would be causally prior to
Purchase Probability (either Own or Other). The position of
Own Preference in the obtained model suggests that the Prefer-
ence item may not have accurately measured the underlying

construct.

DISCUSSION

The prediction of a prior experience X quality of
experience interaction was supported by the data. Given a
negative exposure, subjects who had prior experience with the
peaches rated them as significantly less sour than subjects
who had no prior experience. Given a positive exposure, the
prior experience factor had no significant effect on subjects'
ratings. An examination of the means in Table 1 alone might
suggest an averaging of experience for subjects in the nega-
tive experience--prior experience condition. However, the
standard deviations indicate that some subjects discounted
the negative experience while others did not. So, partial
support was found for discounting in the multiple experience
case.

The quality of experience X possible situational explana-
tion interaction, and the prior experience X quality of
experience X possible situational explanation interaction
were not supported by the data. The significance of these
interactions was dependent on the effectiveness of the possible
situational explanation. The pattern of results for Ripeness
indicate that while subjects attended to the message, it was
not employed in a situational manner. The large main effect
of the explanation on Ripeness supports this. Two possible
explanations can be offered. Subjects may have doubted the

35

36

the situational nature of the problem, suspecting that the
problem would be recurrent, or they simply may not have
grasped the situational nature of the explanation.

An unexpected finding was the flatness across cells of
mean confidence ratings. Regardless of condition, subjects
were highly confident in all their ratings. A possible after-
the-fact attributional explanation would be that the subjects
possessed such a well developed causal schema about the
salient attributes of a "good" canned peach and their ability
to evaluate them, that this schema overshadowed the experi-
mental manipulations. However, the attributional object was
chosen to be one for which subjects possessed a well-deve10ped
conception. An alternate explanation may be that because
there were no personal consequences to subjects for their
ratings, they may have been more liberal in their confidence
estimates. 1

The path analysis revealed that quality of experience and
the prior experience X quality of experience interaction
impacted the dependent measures through the attribute of
sweetness. Sweetness directly impacted Overall Taste, and
both directly and indirectly impacted Own Purchase Probability.
So, an individual's purchase probability was significantly
predicted by his/her evaluation of sweetness. The path model
also indicates that individuals used their own purchase
probability to estimate others' purchase probability. This
finding is consistent with the egocentric or false consensus
bias discussed by Ross (1977). This bias is one where individ-

uals view their own judgments as relatively common, and that

37

most other individuals would behave as they would. The
position of Own Preference in the path model, and its small
relationship with Others' Preference would suggest that the
preference measure employed in this research did not capture
the underlying construct of preference. Theoretically one
would expect preference to causally preceed purchase prob-
ability.

The present research demonstrated discounting as a more
general attribution phenomenon, consistent with the principle
of minimum causation. A segment of subjects with prior
experience, when given an inconsistent negative experience,
tended to discount it in light of the prior positive experi-
ences; subjects who had no prior experience to draw upon
were more harsh in their evaluation of the product.

Within attribution theory, more thought should be given
to how multiple experiences are translated into causal schema,
and the conditions for employing a causal schema as opposed
to actively seeking further experience. Further empirical
work is needed to define these processes. Following from
this, research is needed to address the questions of when a
decision is made to employ a causal schema or seek further
experience, what are the differences in causal explanations
rendered, and how do they differ in validity?

With regard to consumer behavior, two implications come
to mind. The finding of significant interaction of prior
experience and quality of experience for the sweetness attri-

bute (which significantly impacted purchase probability)

38

provides a theoretical underpinning for the observation
that consumers who have a negative experience with a product
with which they have had previous positive experiences will
be more likely to purchase the product again than will con-
sumers whose first experience with the product is negative.
In addition, if an effective situational explanation were
present, consumers with prior positive experience would be
more likely to actively seek out and use that explanation
to reduce the incongruency. Consumers with no past experi-
ence, however, would have no such motivation.

A marketing implication of this research would be to
have multiple-unit new trial promotions so that if a single
negative experience occurred it would be more likely attri-
buted to situational factors than to the product itself.
Where multiple-unit new trials are not possible, more explicit
preparation instructions/serving suggestions could be included

with the product to facilitate a positive first experience.

APPENDICES

APPEEUIEX.A

In use product research it is necessary to identity the characteristics
or qualities of a product which are important to consueers. Today we
would like you to think of characteristics or attributes of canned sliced
peaches rhich ars.uxxunmntԤggngg. For steeple. cadunuuflon (ouflxnuuudp
flat) light be an ieportant attribute for a soft drink. 0n the lines below,
please list the attributes of canned sliced peaches which are important to you.

 

 

 

 

 

 

 

 

Student pusher
(for‘hhnuulicacunipunwuuusonly.
yonr*responses‘wixltunsmgictlzquuudantial)

39

Bateeachotthesepotentialbrandn-esaccordinsto
placncthenuiaer whichcorrespoads not closely toyonropinion
spacebesidethene-e.

40
New Product Research

scalahelowby
inthe

l. Likeverymch
2. Likesoaewhet

3.
A.
5.

i
U
0 a N O u b u N H
e e e o 0 e e 0 e

5

N O u ' U N P
0 a O a O a s

a
s

Ioopinion
Malibu-user
Disliheverymch

1100‘!qu Peaches.
Sweet Breeae Peach--
Grace Lake Psadaes
Gold "n" lips Peaches
Sun Ripe Peaches
bll-o-Sweet Peaches
Tasty lie-Lita Peaches
Orchard Presh Peaches
bellow Grove Peachee
topical Treat Peaches

theeeslomsnsinathesanscale.

”3o fresh they belou at the produce counter".

”Now there's good taste without the extra calories".

"n:- psrtect treat, light and sweet".

"mute-ta. ”apnea".-

”b you want that scathing sweet. here's a healthy treat".
"Solidtt and treehyoecsneat"-" nytin".

"Psregrsedinlightsyroptoflvethetbetterthutreshpicked
m.

"Iow there's that light trash taste myths of year”.

Student nude:

Reta

Al
New Product Research

of these possible marketing strategies as to how effective you think

each
thsywillbeinhelpingprodoctsalesbynsingthescalebelow.andplacing

ennberwhdchcorrespondsnost closelytoyonropinioninthespacabeside

the strategies.

1.
2e
3.
t.
5.

Very effective
Fairly effective
lo opinion

Fairly ineffective
Very ineffective

E

. busy-oft coupons in daily nuspaper.
. Pecipesserving sometionsonside of can.
. hey-back guarantee.

. contest—"In twenty-firewordsorlesawhy_____lpeachesinlight
syrnparemthebest"

. Cottage cheese—promtional tie-in.
. Prounsnt tront-of-eiele displays in addition to shelf locations.

. EneaJoAnnWorleyty-pespohespersontoessociatawith thebrend
naee.

. A catchy jingle.

Student cuter

Student Nutter

(for identification purposes only
APPENDIX 3 your responses will be agrictlz

confidatial)

mmnm
Asst-ingthstthecoepnydecidestogototsstnrket.howdoyouthinh

the product would be perceived (rated on the following characteristics)?
Inadditia: please indicateontheprovidsd scalehowconridnt yousre

oteachrating.

here-1y Hoderately Soeswhat Neither Soeewhst Moderately Extra-sly

Pita fishy
lot
bevy Light
Not
Confident __ __ _ __ _ __ __ Confident
Good
”P . - Ch I!
Discolored Color
lot
Wt __ __ _ __ __ __ __ Confident
5'0“ __ __ __ __ __ __ __ Sour
Bot
Wt __ __ __ __ __ __ __ Contidnt
lot
Confident __ __ __ _ __ __ __ Confident
Overall Overall

lot
Confident Confident

 

42

43

mmtthe-coapanydecidedtobringthisproductonthenarhet
andthatyouneededtopurcheseacanofslicedpeaches.whatwould.

be the probability that you would purchase this brand? In addition

pleaseindicateonthsprovidedscalehowconlidntyouareotyour

probabilityrating.

Intro-sly Moderately Souwhet Neither Son-axes bderstely ktrnly

‘ lot

Assdngthatthecoqsnydecidedtobriugthisp ductonthenerhet.
whetpercsntageofotherpeoplewhohavetriedthesepeaclnsdoyou
thinkwouldpnrchesethubrend‘!

 

Pleasaindicste'how catident youare or. the above percentage.
ktrseely lbderately Soeswhst neither Soul-hat Ibderately Barr-sly

 

let
be youpreterthesen-peachesinlightsyruptopeachesinheavy
syrup?
Definitely Soeewhat Soeswhat Definitely
Prater Prefer No Preference Prefer Prefer
.ight Syrup __ 3|"! 97m
Please Indicate how confident you are of the above rating.
latrsuely hoderately Soeewhet neither Sonshat lbdarately Entrees”
llot

thatperceetageototherpeoplewhohssetriedthesepeachssdoyou
thinhwouldpreterthutopeachesinhsesysyrwr

2

 

44

Please indicate how confident you are or the above percentage.

Eur-sly lbderatsly Souwhat Neither Somhat Moderately Extreeely
lot

handset _ __ __ __ __ __ __ Confident
D-ographics.
Sex: Hale___ Panels—
Age: __
Doyoulive:
onceapus— outsma—

Bowottendoyouusuallygofoodshopping?
eorethentwiceaweek
twiceawoeh

onceaweek

onceeverytwoweeha

onceauonth

lessthenonceanonth

lessthenonceeveryaixaonths

LIST OF REFERENCES

LIST OF REFERENCES

Beckman, L. Effects of students' performance on teachers'
and observers' attributions of causality. Journal Of
Educational Psychology, 1970, 61(1), 76-82.

 

 

Cohen, J. Statistical power analysis for the behavioral
sciences. New York: *ACadefiic Pfess, 1977.

 

 

Finn, J.D. MULTIVARIANCE: Univariate and multivariate
analysis or variance, covafianfie, and regre551on. Ann
ArBOr: National Educational Resources, Inc., 1972.

 

Fishbein, M., 8 Ajzen, I. Attitudes and Opinions. Annual
Review of Psychology, 1972, 23) 487-544.

 

Heise, D.R. Causal analysis. New York: Wiley 8 Sons, 1975.

 

Hummel, T.J., 8 Sligo, J.R. Empirical comparison of univariate
and multivariate analysis Of variance procedures. Psycho-
logical Bulletin, 1971, 76(1), 49-57.

 

Irwin, F.W., & Smith, W.A.S. Further tests Of theories of
decision in an "expanded judgment" situation. Journal
Of Experimental Psychology, 1956, 52(6), 345-348.

 

Irwin, F.W., & Smith, W.A.S. Value, cost, and information as
determiners of decision. Journal Of Experimental Psycho-
logy, 1957, 54(3), 229-232.

 

Irwin, F.W., Smith, W.A.S., 8 Mayfield, J.F. Tests of two
theories of decision in an "expanded judgment" situation.
Journal Of Experimental PsychOlOgy, 1956, 51(4), 261-268.

 

Jones, E.E., Davis, K.E. 8 Gergen, K.J. Role playing varia-
tions and their informational value for person perception.
Journal of Abnormal and Social Psychology, 1961, 63(2), 302-310.

 

Kelley, H.H. Attribution in social interaction. in E.E.
Jones, D.E. Kanouse, R.E. Nisbett, S. Valins, 8 B. Weiner
(Eds.), Attribution: Perceiving the causes of behavior.
MOrristown (NJ): General Learfiing Press, 1972a.

 

Kelley, H.H. Attribution theory in social psychology. in
D. Levine (Ed.), Nebraska Symposium on MOtivation, (Vol.
15). Lincoln: University Of Nébraska Press, 1967.

 

45

46

Kelley, H.H. Causal schemata and the attribution process.
in E.E. Jones, D.E. Kanouse, K.E. Nisbett, S. Valins,
& B. Weiner (Eds.), Attribution: Perceiving the causes
of behavior. Morristown (NJ): General Learning Press,
1972b.

Kelley, H.H. The process of causal attribution. American
Psychologist, 1973, 28(2), 1072128.

Kiesler, C.A., G Munson, P.A. Attitudes and Opinions. Annual
Review of Psychology, 1975, 26, 415-455.

 

McArthur, L.A. The how and what Of why: Some determinants
and consequences Of causal attribution. Journal of
Personality and Social Psychology, 1972, 22(2), I7I-193.

Orvis, B.R., Cunningham, J.D., 8 Kelley, H.H. A closer exam-
ination of causal inference: The roles Of consensus,
distinctiveness, and consistency information. Journal
Of Personality and Social PSychOlOgy, 1975, 32(4),
605-616.

Peretz, M.D. Studies on the measurement of attitude intensity.
Unpublished master‘s thesis, Michigan State University,
1974. '

Ross, L. The intuitive psychologist and his shortcomings:
Distortions in the attribution process. in L. Berkowitz
(Ed.), Advances in Experimental Social Psychology, (Vol.
10). New York: Academic Press,T1977.

Ruble, D.N., 8 Feldman, N.S. Order of consensus, distinctive-
ness, and consistency information and causal attributions.
Journal Of Personality and Social Psychology, 1976,

371(5), 930-937.

Scheifley, V.M., E Schmidt, W.H. Jeremy D. Finn's MULTIVARI-
ANCE - multivariate analysis Of variance, covariance,
and regression - modified and adapted for use on the
CDC 6500. Occasional Paper NO. 22, Office Of Research
Consultation, School for Advanced Studies, College of
Education, Michigan State University, 1973.

Shaklee, H.L. Limited minds and multiple causes: Discounting
in multicausal attribution (Doctoral dissertation,
University Of Oregon, 1976). Dissertation Abstracts
International, 1977, 37(9), 4764B.

MICHIGAN STATE UN

"(m llll ll I‘lluljlllljl“

129310046