Page 1
THE USE OF COVARIATION INFORMATION IN THE
CAUSAL ATTRIBUTION OF OTHERS' EMOTION
Heidi L. Evre
B.Sc., Utah State Universitv, 1998
THESIS SUBkETED IN PARTML FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS
in the
Department of Psychology
O Heidi L. Eyre 2000
SIMON FUSER UNIVERSITY
April2000
Al1 rights reserved. reproduced in whole or
other means, without
This work mav not be in part, by phobcopy or by permission of the author.
Page 2
National Library Biblioth ue nationale 7 du Cana a
Acquisitions and Acquisitions et Bibliographie SeMces services bibliographiques 3QS Wellington Street 395, rue Wellington OnawaON K 1 A W OttawaON KIAON4 cm&a Canada
The author has granted a non- exclusive licence allowing the National Library of Canada to reproduce, loan, distribute or seU copies of this thesis in microform, paper or electronic formats.
The author retains ownership of the copyright in ths thesis. Neither the thesis nor substantial extracts fiom it may be printed or otherwise reproduced without the author's permission.
L'auteur a accordé une Licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/film, de reproduction sur papier ou sur format électronique.
L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thése ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.
Page 3
Attribution of Emotion ..* I l l
Kelley's (1967) covariation theory as operationalized by Orvis, Cunningham. and
Kelley (1975) is considered one of the prevailing models of causal attribution.
With one notable exception (hIcArthur, lY2) . no one has exarnined the
attribution of emotion within this framework. The present study extended the
findings of McArthur by having participants make causal attributions for four
emotions (anger, happiness, depression, and anxiety) embedded within an
interpersonal context. The sarnple consisted of 155 male and female native
English-speaking Canadian undergraduates. Results supported Orvis et al.3
(1975) predictions for the person and stimulus attributions. There was less
support for the circumstance attributions, and the least amount uf support was
found for the person x stimulus attribution. Discussion focused on design and
analysis limitations, applications of the current study to other realms in emotion,
and the need for a more extensive examination of: other emotions.
Page 4
Attribution of Emo tion
DEDICATION
1 would like to dedicate rny thesis to my closest friends and family: Katlin, Mom,
Dad, Tamara, kcky , Jeni fer, and Alana. Without their ceaseless support,
devotion, and love I would not have been able to accomplish this considerable
task. I camot thank them enough for being there for me.
Page 5
Attribution of Emo tion v
ACKNOWLEDGEMENTS
I would like to thank Demis Krebs for being willing to take me on as a student
rnid-stream and Kim Bartholomew for being there by rny side through the
tvhole process. 1 would also like to thank Cathy McFarland and the members of
the social area seminar for their insighthl comments on mv thesis proposal. I
am grateful to Laura MacKav, Tina Niwinska, and Chris Newitt for their help
collecting data. Finally, I would like to thank Tamara Ferguson for her insightful
comments on a drait of my thesis.
Page 6
Attribution of Emotion
TABLE OF CONTENTS
................................................................................................................. introduction 1
.......................................................................... Ovenriew of the Present $tudy 5
Hypotheses ........................................................................................................... 7
......................................................................................................................... Method 9
............................................................................................................ Participants 9
.............................................................................................................. Procedure 10
Results ....................................................................................................................... 11
Data Analyses ....................................................................................................... 11
.............................................................................................. Person Attributions 13
........................................................................................... Stimulus Attributions 13
................................................................................. Circumstance X ttributions 14
.......................................................................... Person + Stimulus Attributions 15
.................................................................................................................... Discussion 16
............................................................................................. Summary of Results 16
..................................................................... Limitations and Future Directions 19
.................................... Appendix A Causal Attribution Questionnaire - Form 1 24
Appendiu B Lndividual Emotion Results Tables .................................................... 29
.................................................................................................................... Re ferences 35
Page 7
Attribution of Emotion
LIST OF TABLES
vii
Table 1 Mean Attribution Ratings for the Covariation hiormation . . 7 3 Conditions ................................................................................................... ,,
Table 2 Percentage of Person, Stimulus, Circumstance, and Person +
Stimulus Attributions Made by Participants When hsked to
......................................................................... Make a Single Attribution 23
Table 3 blean Attribution Ratings for the Four Emotions in Predicted
....................................... Covariation Information - Attribution Pairs 29
Table 4 ivlean Attribution Ratings by Covariation Pattern and Emotion
(Part 1). ........................................................................................................ .30
Table 5 Mean Attribution Ratings by Covariation Pattern and Emotion
.......................................................................................................... (Part 2) 31
Table 6 Percentage of the Person, Stimulus, Circumstance, and Person +
Stimulus Attributions Made by Participants When h k e d to
......................................... blake a Single Attribution for Eaîh Emotion 33
Page 8
Attribution of Emo tion 1
The Use of Covariation information in the Causal
Attribution of Others' Emotion
How do ùbservers arrive at a causal attribution? That is, how do
individuals deçide ruhy a given event has occurred? According to Kellev (1967:
1973), individuals use the principle of covariation when they have access to
multiple instances oi the same or similar events. The principle is looselv based
on an analysis of variance mode1 containing three types of covariation
information: consensus, consistency, and distinctiveness. For example, let's say
John and Bill are in a bar, and John hits Bill. Kellev states that individuals use
covariation information to determine whether the event was caused by
something about John (person), Bill (stimulus), the particular circumstance, or
some combination of the three. Three pieces oi information corne together to
inform this decision of "why." In the majoritv of studies, this information is
presented to subjects as either high or low (cf. PvlcArthur, 1972). One piece of
information that is taken into account is the distinctiveness of the behavior; that
is, how does the person (John) behave toward other stimuli (individuals hi the
bar) in this situation? John's behavior would be considered highly distinctive if
he only hit Bill and not other people. Another piece of information that may be
taken into account is consistencv (i.e., what happens to the behavior when the
same person and stimulus interact across time?). If John hits Bill every time they
go to the bar together, John's behavior would be highly consistent. The third is
consensus, that is, how do other people react to the stimulus? If Steve, Randy,
and Mark (others who frequent the same bar) also hit Bill, John's behavior would
be highly consensual (see Hansen & Donoghue, 1977; Wells & Harvey, 1977 for
discussions on the importance of consensus information).
Page 9
Attribution of Emotion
Depending on the configuration of the covariation information,
attributions can be made to the person, the stimulus, the circumstance, or some
combination thereof (e.g., Hansen, 1980). Orvis, Cunningham, and Kelley (1975)
developed the template matching model of covariation which was based on
Kelley's covariation theory and, in part, on the attributional patterns described
by McArthur (1972). Although several attribution researchers have made
modifications to Omis et al.'s model, the predictions do not radically diverge for
the patterns under study in the present thesis (cf. Forsterling, 1989; Hilton, Smith,
& Kim, 1995; Jaspars, 1983). The template matchhg model states that stimulus
attributions are made in the presence of high distinctiveness (the actor did riot
respond in tlus manner in response to other stimuli), high consistency (the actor
generallv responded to this stimulus in the same fashion across tirne/ modalities),
and high consensus (most people responded in the same fashion toward the
stimulus as the actor) information. A person attribution (often referred to as a
dispositional attribution) is made when one received low distinctiveness (the
actor responded in this m m e r in response to other stimuli), high consistency
(the actor generally responded to this stimulus in the same fashion across
time/modalities), and low consensus (others did not respond in the same manner
toward the stimuli as the person did) information is present. The other type of
attribution made is something about the circumstance, which is hallmarked by
low consistency information (i.e., it only happened this once) and is most clearly
seen in conjunction with high distinctiveness and low consensus information.
As mentioned above, the three types of information can also be combined
to form complex attributions such as person + stimulus, peson + circumstance,
stimulus + circumstance, and penon + stimulus + circumçtance. There has been
Little empirical support for the person + circumstance, stimulus + circumstance,
Page 10
Attribution of Emotion 3
and person + stimulus + circumstance attributions as operationalized by Orvis
and colleagues (e.g., McArthur, 1972; Orvis et al., 1975). Participants who were
presented covariation information hvpothesized to correspond to these complex
attributions tended to make main effect attributions (i.e., person, stimulus, or
circumstance) instead cf the cornplex attributioil predicted. The cne corn plex
attribution which participants did make is the person + stimulus attribution (e.g.,
in Orvis et al., 1975, participants made the predicted person + stimulus attribution
56% of the time). The information pattern in the template matching mode1
which corresponds to a person + stimulus attribution is low distinctiveness, high
consistencv, and high consensus.
EvIcArthur (1972) conducted the first empirical test of Kelley's theory. She
postulated that the causal attribution made rnav be a function of both the type of
covariation information the participant is given and the verb category.' The verb
types that she investigated were emotions, opinions, accornplishments, and
actions. In her experimental design, McArthur provided participants with brief
scenarios depicting four ernotional events (e.g., "Paul is enthralled with the
painting," McArthur, 1972, p. 174). She then paired these four emotion scenarios,
plus four of each other verb type (opinions, accornplishments, and actions), with
the eight differing covariation patterns that can be created bv combining
dis tinctiveness (high/low), consistency (high/low), and consensus (high/ low)
Although McArthur explicitlv labels her emotion words as verbs, they are
actually adjectives (e.g., "Sue is nfinid of the dog."). For purposes of continuity,
the term "verb" will continue to be used in reference to McArthurts study.
However, when explaining the design of the current study, the correct term
"adjective" will be used.
Page 11
Attribution of Emotion
information. In short, every participant completed a 16-item questionnaire
which contained each of the 16 verbs being investigated paired with one of the
eight covariation information patterns. The attributions the participants made
were measured using a multiple choice format. That is, participants could select
rvhether it was (a) something about the person, (b) something about the
stimulus, (c) something about the particular circumstances, or (d) some
combination of a, b, and c (and they were asked to specify the combination).
McArthur found that across verb type, the highest proportion of stimulus
attributions were made in the high distinctiveness, high consistency, and high
consensus condition. The greatest proportion of person attributions were made
in the low distinctiveness, high consistencv, and low consensus condition. The
majoritv of the circumstance attributions were made when there was low
consis tency. However, the high dis tinc tiveness, low consistency, and low
consensus condition showed the greatest number of circumstance attributions
ovemll. Finallv, two separate conditions showed high person + stimulus
attributions. These were the low distinctiveness, high consistency, and high
consensus condition and the high dis tinctiveness, high consis t enq , and low
consensus condition. The first of these two patterns is what was predicted by
Omis and colleagues (1975) and this pattern produced a significantly higher
proportion than the second pattern (P = .35 vs. P = 23, p < .01). In short, the
results were in line with Orvis et ale's template matching predictions.
As for her results with emotion verbs, she found that person and person +
stimulus attributions were made the least often to emotion verbs as compared
with other verbs. On the other hand, stimulus and circumstance attributions
were made the most often in conjunction with emotion verbs.
Page 12
Attribution of Emotion
Qverview of the Present Studv
The primary goals of the present studv were to examine two questions
regarding the application of covariation theory to the causal attribution of
rmotion: (a) Within ernotion scenarios, do participants rate the predicted
attributions significantly higher than the non-predicted attributions? (b) When
participants are given a multiple choice format and told to pick the single best
causal attribution, do they choose the predicted attributions significantly more
often than the non-predicted attributions? A third question, which was
exploratory in nature, was also addressed: (c) Do the results found in (a) and (b)
hoid true for each individual emotion?
Although Mcilrthur began the investigation into causal attribution of
emotion, she left many questions unanswered. First, McArthur only examined
four of the myriad possible ernotions.' She also relied on a single dependent
variable - picking îhe single best causal attribution. Third, she used asocial
scenarios which may not be the most appropriate format for emotional events.
She also presented the covariation information in an order which led participants
to under-utilize consensus information (Kassin, 1979; Ruble & Feldman, 1976).
These issues were addressed in the current study.
The emotions used in this study were anger, anxiety, happiness, and
depression. These emotions were chosen because they are represented within
many models of "prototypical" or "basic" emotions (Averill, 1994; Ekrnan, 1994;
Hupka, Lenton, & Hutchinson, 1999; Russell, 1997; Russell & Barrett, 1999;
Scherer, 1994; Shweder, 1994). Depression was chosen instead of sadness
' McArthur (1972) only named two of the four emotions she analyzed in her
study; they were enthrallment m d fear.
Page 13
Attribution of Emo tion
because within Russell's (1997) circumplex model, it is located farther away from
anger and anxiety, thus allowing for a more broad coverage of the arousal
spectrum (i.e., anger and anxietv have a high degree of arousal whereas
depression has a low degree of arousal).
Recent attributional research hao also illustra ted the use fulness of
including continuous rating scales instead of simply relving on multiple choice
answers (e.p., Hilton, Smith, & Kim, 1995). Xlthough multiple choice ratings give
the most concise representation of an attributional outcome (e.g., it was the
stimulus), they do not make as fine-grained distinctions as do continuous rating
scales (e.g., Hewstone, 1983). Both types of scales were used in the present study
to give a more complete picture of the attributional process.
hlso as mentioned above, McArthur's scenarios were primarily asocial
(Hewstone & Jaspars, 1983). That is, they involved an interaction between a
person and an object (e.g., Paul is enthralled with the painting), rather than two
people (e.g., Paul is enthralled with Emily). Although Hewstone and Jaspars
(1983) found that there was no difference between social and asocial action verbs,
the results for emotions may be different because most emotions occur within
interpersonal contexts (cf. Parkinson, 1995,1996). Accuracy and coordination in
the decoding and communication of emotion is essential in the lacilitation of
interpersonal relationships. Emotions are often used to clarify the meaning and
intent of speech, to persuade, and to draw others ernotionally near (Parkinson,
1995). Thus, in the current study, emotion words were placed in the type of
interpersonal context in which they would be likely to occur.
There has been some dispute within the literature as to the importance of
consensus information (e.g., Kassin, 1979; McArthur, 1976). Even McArthur's
original results indicated that participants did not utilize consensus information
Page 14
Attribution of Ernotion 7
as much as they did distinctiveness and consistency. However, Wells and
Harvey (1977) later pointed out that the statistical method McArthur used
(calculating the percentage of variance accounted for by each covariation piece)
was not sound and thus McArthurfs results would need to be replicated across
several studies before drawing the conclusion that con~ensus was utilized les .
Also, Ruble and Feldman (1976) demonstrated that when consensus information
is placed last, the elfect of consensus is actually strengthened. Their research
demonstrated that individuals make the predicted attribution most often when
information is presented in the following order: distinctiveness, consistency, and
consensus.
Hv DO theses
Al1 hvpotheses in this study for the various information patterns were
derived from Orvis, C u ~ i n g h a m , and Kelleyfs (1973) template matching mode1
and McArthurfs (1972) empirical results. The main hvpotheses are listed below.
Of secondary interest was the exploratory study of the individual emotion
results. No specific predictions were made for these, but the general pattern of
results are presented in brief.
Person Attributions
(1) Person attributions for the emotions expressed would be made to a
greater extent in the low distinctiveness, high consistency, and 10 w
consensus condition than in any of the other information conditions.
(2) Within the low distinctiveness, high consistency, and low consensus
information condition, ratings of person attributions would be
significantly higher than ratings of stimulus or circumstance
attributions.
Page 15
At tribution of Emotion 8
(3) When participants are asked to pick the single best causal
attribution, person attributions would be made more frequently than
person, circumstance, or person + stimulus attributions within the low
distinctiveness, high consistency, and low consensus information
condition.
Stimulus Attributions
(1) Stimulus attributions for the emotions expressed would be made to
a greater extent in the high distinctiveness, high consistency, and high
consensus condition than in any of the other information conditions.
(2) Within the high distinctiveness, high consistency, and high
consensus information condition, ratings of stimulus attributions
rvould be significantlv higher than ratings of person or circumstance
attributions.
(3) When participants are asked to pick the single best causal
attribution, stimulus attributions would be made more frequently than
person, circumstance, or penon + stimulus attributions within the high
dis tinc tiveness, high consis tency, and high consensus information
condition.
Circumstance Attributions
(1) Circumstance attributions for the emotions expressed would be
made to a greater extent in the high distinctiveness, low consistency,
and low consensus condition than in any of the other information
conditions.
(2) Within the high distinctiveness, low consistencv, and low consensus
information condition, ratings of circumstance attributions would be
significantly higher than ratings of stimulus or circumstance
Page 16
Attribution of Emo tion 9
attributions.
(3) When participants are asked to pick the single best causal
attribution, circumstance attributions would be made more frequently
than person, stimulus, or person + stimuius attributions within the
high distinctiveness, loiv consistency, 2nd lorv consensus information
condition.
Person + Stimulus Attributions
(1) Since there was not a person + stimulus rating scale, it was not
appropriate to examine whether these attributions would be made to a
greater extent in the low distinctiveness, high consistencv, and high
consensus infonnation condition.
(2) W ithin the low distinctiveness, high consis tencv, and high
consensus information condition, ratings of person and stimulus
attributions would be significantlv higher than ratings of circurnstance
attributions.
(3) When participants are asked to pick the single best causal
attribution, penon + stimulus attributions would be made more
frequently than person, stimulus, or person + stimulus attributions
within the low distinctiveness, high consistency, and high consensus
information condition.
~Method
Partici~ants
Introductory psychology undergraduate students from Simon Fraser
University were recmited through the subject pool board (where students sign
up for research participation). Students were given course credit for their
participation. One hundred fifty-five native English speakers were culled from
Page 17
Attribution of Emotion
an initial pool of 226 students. Native speakers were chosen for this study
because it involved interpreting subtle grammatical nuances that may not be
picked up by non-native speakers. Of these 153,94.8% had lived in Canada for
more than 10 vears. There were 51 males (32.9%) and 104 females (67.1%). The
age range rras 17 tu 46 vears with ri mem ripe of 70.3 Feus. European
Canadian students comprised 63.2% of the sample, Asian students 20.6%; 7.19'0
were East hdian and 8.49'0 were of other ethnic background.
Procedure
Participants were given one of four questionnaires to complete that began
with demographic information and instructions (see Appendix A for a copy of
Form 1 of the questionnaire). The four forms were based on a Latin square
design similar to that emploved bv McArthur. Kirk (1968) describes the
fractional factorial Latin square design as having three variables. Two are within
subjects and one is between. The first within subjects variable is the covariation
pattern, which has four levels (HHH, LHL, LHH, HLL). The second within
subjects variable is the emotion, which also has four levels ( h g e r , Anxiety,
Depression, and Happiness). The questionnaire form is the between subjects
variable; it is also four levels (Form 1, Form 2, Form 3, and Form 9.' The four
forms with their respective emotion-covariation information (Distinctiveness,
Consistency, and Consensus) painngs were as follows:
Form 1: HHH-Anger; LHL-Dep ression; LHH-Happiness; HLL-hxiety
Forrn 2: LHL-Happiness; HLL-Anger; HHH-Anxiety; LHH-Depression
Form 3: LHH-Anxiety; HHH-Depression; HLL-Happiness; LHL-Anger
' Because each f o m contained unique information, an analysis examining the
differences between forrns was not possible.
Page 18
Attribution of Emotion 11
F o m 4: HLL-Dep ression; LM-Anger; LHL-h i e ty; HHH-Happiness
Participants were given short statements about an emotion-eliciting event.
These statements contained the three pieces of covariation information along
with pieces of contextual information, such as the re!atiomhip between the
person and stimulus and where they were or what they were doing when the
emo tion occurred.
After participants read the brief scenarios, they were asked to make three
scale ratings which assessed the extent to which the emotion was due to the
person, stimulus, and circumstance (e.g., "To what extent do you believe that
Shawna's happiness in this case was due to somrthing nbori t Mnry?") . Ratings
were made on a 9-point scale which were anchored by the terrns "not at all" and
"very much so." Finallv, participants were asked, "If vou had to choose orrr
reason why [the target] felt [the emotion] in this particular circumstance, is it due
to sornething about: (a) [the person], (b) [the stimulus], (c) [the particular
circumstances], or (d) both [the person and the stimulus].
Resul ts
Da ta Analvse~
Because of the complexity of the Latin square design, each of the analyses
used and the table that is associated with those particular results will be
delineated first. The first set of analyses were repeated-measure ANOVAs
comparing the mean ratings of peson, stimulus, and circurnstance attributions
made in each of the covariation information conditions. These analyses are
illustrated in the columns of Table 1 and were done to address the first set of
predictions made for each attribution, namely, whether a given attribution was
made significantly more often in the predicted covariation information condition
Page 19
Attribution of Emotion
than the other three information conditions (the predicted results in al1 of the
tables are italicized). Although the majority of the predictions in this study were
made a priori, pairwise comparisons were made using posthoc tests as a more
conservative estimate of the effects (Tukey's HSD for the between-subjects
analyses and pairwise repeated-measu res analyses for the tvi thin-subjects
analyses).
The second set of analvses, outlined in the roivs oi Table 1, were also
within-subjects ANOVAs designed to test whether there were significant
differences in the mean ratings of person, stimulus, and circumstance attributions
made within each of the covariation patterns (the second set oi predictions made
for each attribution).
Table 2 contains the results of the third set of analvses which were
conducted on the multiple-choice ratings made by participants (the third set of
predictions made for each attribution). Chi-square analyses were used to test
whether a' higher proportion oi person, stimulus, circumstance, or person +
stimulus attributions were made in each of the four information conditions.
Because the results for the individual ernotions were exploratory in nature
and not of primary interest in this thesis, the tables associated with them are
listed in Appendix B. Table 3 contains the predicted covariation - attribution
pairs and addresses the issue of whether the mean attribution rating for the
expected covariation pattern - attribution pair is sigruficantiy higher or lower
for any one of the four emotions examined. That is, it indicates whether the
results presented across emotions hold true for the individual emotions. Table 4
is the complement to Table 1 columns and it contains between-subjects ANOVAS
for eadi ernotion - attribution pair (as mentioned above, the post-hoc
comparisons were done using Tukeyts HSD test). Table 6 is the complement to
Page 20
Attribution of Ernotion
the rows of Table 1 nnd contains repeated-measures analyses; Table 7 i s the
complement to Table 2.
Person Attributions
Based on Kelley's theorv and McArthurf s results, it was predicted that
peson attributions for emotions would be made most strcngly in the low
distinctiveness, high consistency, and low consensus condition when compared
to the three other information patterns. The first column of Table 1 illustrates
that, collapsed across the four emotions, person attributions were made the most
strongly in the LHL information pattern, as predicted, F (3,462) = 46.97, p c .001.
Within the LHL information pattern, respondents rated the person attribution
significantly higher than the stimulus or circumstance attribution, F (2,308) =
82.77, p < .O01 (see the first row of Table 1 for the rneans and standard
deviations). As illustrated in the fi rst row of Table 2, when respondents were
asked to select the single best causal attribution in the LHL information
condition~significantly more of hem made person attributions than stimulus,
circumstance, or person + stimulus attributions (55.48'30 vs. 9.0396, 20.6s0/0,
14.849/0, respectively).
As mentioned above, the analyses conducted on each of the emotions
were exploratory in nature and thus only the general pattern of results will be
delineated here. The tables associated with each of these analyses are listed in
Appendix B. First, for the LHL information condition, there were no significant
differences in the ascription of person attributions among the four emotions, F
(3,151) = 1.55, p > .O5 (see Table 3). As seen in Tables 4 through 6, the findings
denoted above held mie when each emotion was examined separately.
Stimulus Attributions
[t was also predicted that stimulus attributions would be made to a
Page 21
Attribution of Emotion
greater degree in the high distinctiveness, high consistency, and high consensus
condition than in the three other information patterns. Stimulus attributions,
collapsed across emotions, were made to a greater extent in the HHH
information condition, F (3, -162) = 61.38, p < .O01 (see the second column of Table
1 for the means and standard deviations). The second row of Table 1 shows b a t
within the HHH information condition, individuals rated the stimulus attribution
significantly higher than person or circurnstance attributions, F (2, 308) = 82.72, p
c .001. As illustrated in Table 2, participants who were asked to pick the single
best causal attribution in the HHH information condition made stimulus
attributions more often than person, circurnstance, or person + stimulus
attributions (63.23% vs. 8.39%, 13.48%, and 12.90%, respec tively).
M e n examining the results for the individual emotions, it is clear that
depression was significantly less likely to elicit a stimulus attribution in the HHH
information pattern as compared with the other three emotions, F (3, 151) = 3.63,
p < .O5 (see Table 3). When examining Tables 4 and 5, we see that stimulus
ratings in the HHH information condition were consistently higher than any of
the other attributions, even though this difference was not always statistically
signifiant in the post-hoc analyses. Finally, as demonstrated in Table 6, stimulus
attributions were made significantlv more often in the HHH information
condition across the four emotions.
Circumstance Attributions
It was predicted that circumstance attributions would be made to the
greatest extent in the high distinctiveness, low consistency, and low consensus
condition. The third column of Table I illustrates that when the results are
collapsed across emotions, circumstance attributions were made to a greater
degree in the HLL information condition, F (3,462) = 11.38, p c .001. However,
Page 22
Attribution of Emotion 15
within the HLL condition, there was no difference between the person, stimulus,
person + stimulus, and circumstance means, F (2,308) = 1.65, p > .O5 (see the third
row of Table 1). When participants were asked to pick the single best causal
attribution in the HLL information condition, they made circumstance
attributions significantly more often than person, stimulus, or person 4 ~timului
attributions (45.81% vs. 19.35%, 26.45%, and 8.39%, respectively; see Table 2).
The results with the individual ernotions are slightly more mixed. As
demonstrated in Table 3, circumstance attributions were made to a lesser extent
in the depression scenarios and slightly more often in the happiness scenarios,
al though these ratings are not significantly different from the anviety and anger
ratings, F (3,151) = 3.27, p < .05. Tables 4 and 5 indicate the results for happiness
support the predictions. The results for anger and anviety are less strong, in that
the results are in the general direction predicted, although they are not
statistically significant. The emo tion of depression does no t support the
predictions at all.
Person + Stimulus Attributions
It was hypothesized that LHH mformation would give rise to person +
stimulus attributions. Since this study did not contain a person + stimulus rating
scale, it was not appropriate to compare whether peeon + stimulus attributions
were made to a greater degree in the LHH information condition (results which
had been presented in the columns of Table 1). However, when the rows of
Table 1 are examined, it is evident that when individuals were given LHH
information, they did not make significantly higher person and stimulus
attributions, F (2,308) = 1.75, p > .05. When participants were asked to make the
single best causal attribution in the LHH information condition, they did not pick
the person + stimulus attribution significantly more often than the other three
Page 23
Attribution of Emotion 16
attributions (see Table 2). The proportion of person, stimulus, circums tance, and
person + stimulus attributions were nearlv equal across the four categories
(26.459/0,25.16%, 25.169'0, and 23.23%, respectively).
As to the issue of whether there is a difference in the degree to which
person and stimulus attributions are made for the four emotions, significantly
lower stimulus attributions were made to the emotion of depression than to the
other three emotions, F (3,151) = 16.47, p < .O01 (see Table 3). Also, significantly
lower person attributions were made to anger and anxiety than to depression or
happiness, F (3, 151) = 8.31, p < .001. When examining Table 5, it is apparent that
happiness is the only emotion which conforms to the expected pattern of
attributions (i.e., person and stimulus ratings were significantly higher than
circumstance ratings). Finally, as evident in Table 6, the person + stimulus
attribution was not chosen more ot'ten than anv of the other attributions,
regardless of emotion.
Discussion
Summarv of Resulrs
In sum, participants generally made the predicted attribution for the LHL
condition (person attribution) and for the HHH condition (stimulus attribution).
In tems of the HLL condition (circumstance attribution), the results were more
mixed. When participants had to choose the single best causal attribution in the
HLL condition, they usuallv chose the circumstance attribution. However, the
scale rating results were not as supportive as the discrete causal attribution
results. Finally, the results for the LHH condition did not generally support
Orvis et al.3 template matching model. That is, participants did not generally
choose the person + stimulus attribution in the LHH condition. The scale ratings
were only partially supportive, at best. Research within Kelley's paradigm has
Page 24
Attribution of Emo tion
typically shown mixed support, and the presen t resul ts do generaily replicate
other findings (see Forsterling, 1989 for a current review). That is, there was
strong support for the person and stimulus attributions, less support for the
circumstance attributions, and the least amount of support for the combined
attributions (Le., Ferson + stimulus).
Thus, it appears that individuals do, in some cases, use Kelley's covariation
principle effectivelv for the attribution of emotion. hdividuals in this study were
consistent in using the principle for determinhg person and stimulus
attributions. Participants were less consistent in arriving a t the predicted
circums tance
the predicted
information.
attribution in the case of HLL information and they rarely niade
person + stimulus attribution when presented with LHH
Further study is warranted to narrow down whether the
covariation information patterns are incorrect in the case of circumstance and
person + stimulus attributions or whether it is the application of these patterns to
the realm 'of emotions which are producing the contradictory attributions.
To briefly address the latter of these two possible explanations, 1 will
summarize the exploratorv emotion results. It appears that the emotion of
happiness, the oniy positive emotion examined in this study, produced results
that were rnost consistent with Orvis and colleagues' predictions. For example,
even though the results for anxiety, anger, and depression did not show support
for the person + stimulus attribution predicted by the template matching mode1
in the L H H information condition, happiness did show marginal support for this
attribution. Participants' ra tings of happiness also consis tently showed the
clearest support for the person, stimulus, and circumstance attributions. Because
this study only contained one positive emotion, it is not clear whether this
attributional pattern is unique to happiness or is likely to occur with other
Page 25
Attribution of Emotion
positive emotions. However, because emotion research has demonstrated that
individuals discriminate less between various positive emotional states (e.g.,
elation vs. jov) as compared to negative states (e.g., frustration vs. rage; Ekman
& Davidson, 1994)) 1 would predict that the pattern found with happiness would
generalize to other positive emo tions.
Results for depression, on the other hand, rarely conformed to Orvis et
al.3 predictions. For depression, significan tly less stimulus attributions were
made cvithin the corresponding information conditions. This pattern is also
apparent in other conditions, such as the LHH condition. In short, it appears that
there is a person bias (or dispositional bias) present in participants' rating the
emotion of depression. Most likely, this stems from the term used because
"depression" implies more of a long-lasting, global, interna1 state, rather than a
transitory emotion aroused by situational pressures.
The results for anxietv and anger were not consistent across the
attributional patterns. Ratings from these scenarios conformed to the predicted
results for the person and stimulus attributions, but were mixed for the more
complex circums tance and person + stimulus attributions. Further exploration
into these results is needed before one can speculate on the how these particular
emotions fit within this framework.
It is also important to note at this point that the rating scale results
typically showed that participants indica ted al1 three factors (person, stimulus,
and circurnstance) plaved a rok in the emotional situations. For example, al1 of
the means in Table 1 (excluding one) were over five, which was labeled as
"somewhat" due to the person, stimulus, or circumstance. So, although
participants did make the hvpothesized person and stimulus attributions to a
greater extent, they did not rule out the possibility that other factors (such as the
Page 26
Attribution of Emotion 19
circumstance) may have been playing a role. in the case of the circumstance and
person + stimulus scenarios, participants did not single out a solitary cause, but
instead ratecl al1 three causes as equally likely. However, the rating scale results
for person and stimulus attributions did support the multiple choice results; the
only case in which the rating scale results did not match the multiple choice
results was in the case of circumstance attributions. Taking both of these results
into account, it appears that participants do have an idea about what the primarv
cause of a situation is, although they do not report the primary cause as the only
cause.
Limitations and Future Directions
One potential flaw with this studv is that the Latin square design does not
allow the studv of the interaction between the covariation information pattern
and the emotion. Although this design has been used yuite frequently in the
literature, it is cited in statistics textbooks as most usehl for exploratory studies
and studies in which one dors not expect to find an interaction (e.g., Kirk, 1968).
So, it would be prudent to study Kellev's theory within a fully crossed factorial
design. Nso, a person + stimulus rating scale would be wise to include in future
s tudies.
Another potential limitation of this study is that it lacks extemal validity.
As stated at the beginning of this thesis, I was testing whether individuals apply
Kelley's covariation principle to the attribution of emo tion. The next step to be
addressed is whether or not individuals actually do go through this process
when making attributions and what the limiting factors are. Although several
studies have been conducted to test Kelley's principle in more mundanely
realistic settings (cg., Ferguson, 1977; Ferguson & Wells, 1980; Major, 1980;
Smith & Miller, 1983), none has examined the causal attribution of emotion
Page 27
Attribution of Emotion
within this framework. Given the present results, it appears that when you force
individuals into choosing only one cause, they do use the covariation process.
However, the rating scale results indicate the individuals actually make more
complex attributions and are often unwilling to rule out any causes. That is, it
appears that individuals sav. "It is most likelv x, but it could partlv be due to v
and 2." Participants also may be inferring traits or motivations that are not given
to them in the individual scenarios (e.g., hidden agendas, other personalitv traits,
etc.) Thus, there iç some evidence that for these type of emotion scenarios,
participants go through a more complex attributional process than the one
outlined by Kelley and colleagues.
Although no subjects indicated, either through written or verbal
comment, that thev did not understand the questionnaire, there is the possibility
that participants had a difficult tirne understanding the task or were not
rnotivated to focus their undivided attention on the task. Again, since the
majority of the results followed the hypotheses, it is unlikely that this was the
case for the majoritv of participants (or the results would have appeared more
random).
Also important to acknowledge is the confound between emotion and
scenario content. That is, it is difficult to generalize the results for the given
emotions since there was only one scenario per emotion. Future research should
include multiple scenarios for each emotion in order to better tap the emotional
experience.
Ferguson and Stegge (1998) also recently suggested that emotion should
be studied within the covariation framework. They emphasized the necessity of
this approach because the covariation information can be diagnostic of an
emotion's (ma1)adaptivity. That is, dispositional attributions may be indicative of
Page 28
Attribution of Emo tion
an individual's proclivity to experience a given emotion. For example, if Jocelyn
experiences guilt in the presence of many individuals, across multiple situations
over time, and no one else responds with guilt in these situations, one could
conclude the something about Jocelyn is driving these feelings of guilt. There are
manv studies which have used consistencv information to studv an emotion'ç
maladaptivitv, and Ferguson, Stegge, Evre, Vollmer, and Ashbaker (in press)
recently demonstrated a high correlation between children who respond with
guilt in low consensus situations (as opposed io high consensus situations) and
intemalizing symptoms. The current study shows that individuals do use al1
three pieces of covariation information when making attributions and thus
future research should address Ferguson and Stegge's hypothesis that al1 three
pieces of information can be used to assess an emotion's rnaladaptivity.
in sum, it appears that when individuals are given consensus, consistency,
and distinctiveness information in the manner prescribed by Kelley, they do
make sorne of the causal attributions in the direction he predicted. It is
interesting to h d thnt Kelley's predictions hold even w hen considering
emotional States. Future research should examine this process in more depth
and see where people's implicit stereotypes about emotion (e.g., gender-related
stereotypes of ernotion) and logical attribution models, such as Kelley's, diverge.
Subsequent studies done on emotion should also more thoroughly test Kelley's
mode1 against other covariation models (e.g., Jaspars, 1983; Forsterling, 1989).
Page 29
Attribution of Emo tion 22
Table 1,
Mean Attribution Ratings Made for the Covariation Information Conditions
kz!!l Stimulus Pattern Circums tance F-value
LHL 7.0Se (1.941 4.05d (2.41) 5.2Ia (2.20) 69.61"'
H H H 4.4lC (2.17) 7.36e (1.78) S.IOa (2.17) 32.72"'
HLL 5.30ab (2.23) 5.97b (2.31) 6.36b (2.24) 2.65
LHH 6.05tz (3.47) 5.67a (2.40) 5.52a (2.23) 1.75
Predicted Attributions for rach of the Distinctiveness, Consistencv. and Consensus Information
h t t ~ r n s : LHL - Person; HHH - Stimulus; MLL - Circumstance; LHH - Person + Stimulus
Note: F-vdues listed on the horizontal were computed on the columns; F-values listed on the
vertical were computed on the rorvs. Menns not sharkig a common subscript across a row or
column are signihcantly different. The scde range is 1 to 9 and high numbers indicate stronger
attributions.
' p < .OS; *' p < .Ol; "'p < .O01
Page 30
Attribution of Emo tion
Table 2.
Percentage of Person, Stimulus, Circumstance, and Person + Stimulus
Attributions Made by Participants When Asked to Make a Single Attribution
Pattern P e m n Stimulus Circumstance Person .e Stimulus Chi Sq.
LHL 53.48 9.03 20.65 14.84 31 .QOM'
HHH 8.39 63.23 15.48 12.90 122.39"*
HLL 19.35 26.45 45.81 8.39 46.06***
LHH 26.45 25.16 25.16 23.23 0.33
Page 31
Attribution of Emotion
Causal Attribution Questionnaire - Form 1
Your age:
~ o u r srx: laie ~ e r n d c
Plrase place a " 1 " ncxt to the erhnic group you idrntify with the most. If you identify with more thnn one p u p . rrink order the remainint groups you identify with.
hsian Europenn Cyiadiui First Nations E:ist Indian C] Other
Please place an "x" in the box next to the religious group you affiliate with the mas[.
Hindu Catholic Protestant [7 bluslim 0 Buddhist
Mormon (LDS) Athrist No rrligious affiliation Other religion
How many yevs have you l i v d in Canada? 0 less thon 1 0 1-5 5 - 10 more thnn 10
1s English your tirsr language? 0 Ycs [7
if English is POT your first language: How good is your çornmand of rhe Engiish Ianguage'? 0 Excellent G d Fair Poor
Your religion and ethnic group are being ask for gnly to make sure we accurately characterize the people piirticipating in the study.
This questionnaire contnins 3 numbcr of smtements which report the occumnce of somr emotion (which is
undrrlinrd). FoIlowing rach statrmrnr you will îïnd three items of infarmntion, a11 of which ripply to the
emotion reponcd. Your task is to decide, on the bûsis of the informacion given. whor probably caussd the
emotion to occur. You will be ctskrd to nte the extent to which you think the srnotion was criusrd by three
possible causes. Please circlr a number on the ratine scale (Irom 1 io 9). After rhat. you will be sked to
pick the mosr probable cause by çhoosing monp four alternatives. You can indicnte your choice by . .
çircling the lette[ next to the cause which you think is most probable.
Plrase give me your best estimate, but don't spcnd toa long on any one question. Xlso. once you have
completed a page. continue on to the next page and do nor go back to liny previous pages.
Thank you t'or your participation!
Page 32
Attribution of Emo tion 25
Katie and Emily are in the süme chemistry lab. One day, while talking with Emily during Inb, Katie becomes nnwv. You know that on that day Katie only felt angry around Emily and not around other Iab members; in the past when Katie and Emily have been in Iab together, Kntie has almost always felt angry around Emily; and 311 of the peopie in iüb feit angry when taiking with Ernily.
To what extent do you believe thrit Katic's anger in this case W;ZS due to 50rnemsz rrhaur K&'?
1 3 3 4 5 6 7 8 Y not at ri11 sornewhat vrry rnuch so
To whrit extent do you bclieve thrit Kritie's mger in bis case was due to somethirg about Fmily?
1 7 3 4 5 6 7 8 Y
not rit d l sornewhrit ver? much so
Ti, w h r i r extent do y o u btrlieve thrit Karie's ringer in this case was due to s ~ m e t h i n g about & paniculrir circurnst~ncr$?
1 - 7 3 4 5 6 7 8 Y not rit ; i I I somrw h a very rnuch so
If you had to choose one reuon why Kritie got ringry in this prinicular instance, is it due to somr thing about:
(3) Katitt
(b) Emily
(c) the puciculru circumstrinces
(d) both Kritie and Emily
Page 33
Attribution of Emotion 26
A group of students, including Danielle and Kara, often study together. One day when Danielle and Kara are studying together, Danielle becomes de~ressed. You know that on that day Danielle also felt depressed rround al1 of the other members of the group; when they have studied together in the past, Danielle has almost dways felt depressed; and none of the other students who studied with Kara on that day felt depressed.
To whrit extent do you believe that Danielle's drpression in this case wris due to somethinv about Dmir llc'?
1 3 3 4 5 6 7 8 Y not rit dl some what very much 50
To whrit extent do you believe thrit Danielle's depression in this case w3s dur to something about &gg!
1 2 3 4 5 6 7 8 Y not iit 1\11 somewhrit very much 50
To whrit extrnt do you believr thrit Danielle's depression in this sase wris due to somrthing about the niirticulrir -;ince$?
1 t 3 4 5 6 7 8 9 not ; ~ t 311 somew hrit very much so
If you hrid to choose terison why Danidle felt depressed in this particulu instance, is it dur tn something about:
(a) Danielle
(b)
(c) the particulsu circumstrinces
(d) bah Danielle and Krin
Page 34
Attribution of Emotion 27
Shawna and Mary are friends and one time when Shawna and Mary are talking, Shawna becomes hnogy. You know that on that d r y Shawna also felt happy around al1 of her other friends; in the past, Shawna has almost alwnys felt happy around Mary; and d l of their other friends felt happy around Mary on that day.
1 . To whi t extent do you believe thrit Shawnri's hrippinrss in this case wris due io scirnething ribout Shriwn;i9?
1 9 3 4 5 6 7 8 Y not rit rill some w hat very much so
2. T o whrit extent do you believe thtit Shawnri's happiness in this case wris due to somethinr about 31 3p '?
1 i 3 4 5 6 7 8 Y not rit dl sornew hrit very much 5 0
3 . To whrit extent do you belirve thrit Shriwnri's hrippiness i n rhis case was due to something about fie osiniçulrir circurnstance$?
i i 3 4 S 6 7 8 Y not rit 311 some w h3t vcrry much so
4. If you had ro choose reason why Shriwna felr happy in this puticular instance, is it dur to somrthing about:
(a) Shriwna
(b) Mvlan,
(c) the particulv circumstruices
(d) both S hawnû and Mary
Page 35
Attribution of Emotion 28
One day when Erica and Linda they are hanging out together, Erica becomes anxious. You knuw that on thit day Erica only fek anxious around Linda and not around any of her other friends; when Linda and Ericû have hung out together in the past, Erice never felt ûnxious; and on that day none of their other friends felt anxious around Linda.
To what extent do you believe that Erics's rinxiety in this case w u due to sornerhi~u about F r k ?
1 - 7 3 J 5 6 7 8 Y not at riIl somewhctt very much so
To whst extent do you believe t h a Ericri's rinxiety in this m e was due to ~omethinyabout I inda'!
1 a 7 3 4 5 6 7 8 9 not at di some w hst very rnuch so
To what extent do you belirvt: thrit Erica's anxiety in this case wris due to sornething about p;iniculnr c=trcumsr;uices?
1 a 7 3 4 5 6 7 8 9 not rit d I somewhat very rnuch so
if p u had to choose q~lf reaon why Ericri felt anxious in this particular instance. is i t due to somrrthing about:
a ) Erica
(b j Linda
(c) the pmicular circumstances
(d) both Ericri and Linda
Page 36
Attribution of Emotion 29
APPENDE B
Individual Emotion Results Tables
Table 3.
Mean Attribution Ra tings for the Four Emotions in Predicted Covariation
Information - Attribution Pairs
4kkm î-mW ,Anxiety Depression Happiness F-va lu%
LHL. Person
HHH
S timulus
HLL
circurnstance 6.46ae (1.99) 6.59,b (2.45) 5.43a (2.44) 6.97b (1.70)
LHtI Person 5.30a (2.80) 4.Wa (2.15) 7.18t, (2.25) 6.Sb (2.06) 8.31"'
S tirnulus 6.63a (2.28) 6.1da (1.71) 3.64b (2.41) 6.3& (1.96) 16.47"*
. . . . . fornation Predicted At~but ions tor each of the Distinctiveness. Consistent nsensus in
-: LHL - Person; HHH - Stimulus; HLL - Circmtance; LHH - Person + Stimulus
Note: Menns with different subscripts across the row indicate sigruficant ciifferences
p < .Os; ** p c .01; ***p c ,001
Page 37
Attribution of Emotim 30
Table 4.
Pvlean Attribution Ratings by Covariation Pattern and Emotion (Part 1)
Pattern
Person
Anger
Xnxiety
Dep ression
Happiness
skuudlu
h g e r
Xnxie ty
Depression
Happiness
Circumstance
h g e r
Anxie ty
De pression
Ha p p iness
LHL
7.001, (1.75)
6,771, (3.38)
7.59b (1.93)
6.85b (1.73)
4.89b (2.28)
4.97b (2.40)
1.98b (1.23)
4.74b (2.17)
5.27,b (2.04)
4.43a (1.99)
5.3& (2.44)
5.7Aa (2.15)
HHH
4.2Sa (1.98)
4.6Aa (2.70)
4.30a (1.93)
4.40a (2.06)
7.5ja (2.23)
7.67u (1.66)
6 . S L 3 (2.21)
7.63,1 (1.78)
4.98a (2.33)
4+03a (2.17)
5.9Za (1.83)
5.60a (1.80)
HLL -
5 . Z a (2.53)
6.3Zabc (2.39)
6,0OC (1.78)
5.9/b (2.04)
6.2Ic (2.41)
5.6Sb (2.50)
6.1& (2.14)
5.83bc (3.10)
6.64b (2.06)
6.391, (2.45)
5.43a (2.44)
6.97b (1.70)
LHH
5.20a (3.83)
4.94, (2.15)
7.18bc (2.35)
6.64b (2.06)
6.63aC (2.38)
6.Ub (1.71)
3.& (2.41)
6.3& (1.96)
4.57a (2.15)
5.92b (1.99)
5.9Sa (2.55)
5.57a (3.02)
Predicted Attributions for each of the Distinctiveness. Consistency, and Consensus Information
Patterns: LHL - Person; HHH - Stimulus; HLL - Circumstance; LKH - Person + Stimulus
Note: Means with different subscripts across the row indicate significant differences;
* p c -05; " p. c .01; "'p c .O01
Page 38
Attribution of Emotion 31
Table 5 .
Mean Attribution Ratings by Covariation Pattern and Emotion (Part 2 )
Pattern DE
LHI,
Anger (1,361
Anxiety (1,34)
Depression (1,431
Happiness (1,3S)
HHH
Anger ( U 3 )
Anxiety (1,38)
Depression (1,36)
Happiness (1,34)
Hu h g e r (1,381
h x i e ty (1,43)
Depression (1,34)
Happiness (1,35)
LHH h g e r ( 1,341
Anxiety (1,35)
Depression (1.38)
Happiness (1,431
(Table 5. Continues)
Page 39
Attribution of Emo tion 32
Table 5. Continued
. . . . . - . predicted c\ttributions for ,aa~h of the ~ ) ~ ~ t s
F a t t a : LHL - Person; HHH - Stimulus; HLL - Circumstance; LHH - Person i Stimulus
O : Means with different subscripts across the row indicate signiiicmt differences
* p c .OS; " p c .al; "'p c .O01
Page 40
Attribution of Emotion 33
Table 6.
Percentage of the Person, Stimulus, Circumstance, and Person + Stimulus
Attributions Made bv Participants When Asked to Make a Single Attribution for
Each Emotion
I2ul!xn x w
h g er (37)
Anxiety (35)
Depression (44)
Happiness (39)
HHH
.Lnger WU
.Anxiety ( 39 )
Depression (37)
Happiness (35)
HLL
&ger (39)
Anxiety (44)
De pression (35)
Happiness (36)
LHH
Anger (35)
Anxiety (37)
Depression (39)
Happiness (44)
Stimulus
18.92
17.14
0.00
2.56
56.82
71.79
36.76
68.57
35.90
25.00
20.00
25.00
54.29
24.32
5.13
20.45
Circ- Ferson + S t i m b Chi Sq.
(Table 6. Continues)
Page 41
Attribution of Emotion
Table 6. Continued
Page 42
Attribution of Emotion 35
References
Averill, J. R. (1994). In the eyes of the beholder. In P. Ekman cL R. J.
Davidson (Eds.), The nature of emotion: Fundamental uestions (pp. 3-14). New
York: Oxford University Press.
Ekman, P. (1994). Ail emotions are basic. in P. Ekman gE R. J. Davidsan
(Eds.), The nature of emotion: Fundamental uestions (pp. 15-19). Oxford:
Oxford University Press.
Ekman, P., & Davidson, R. J. (Eds.) (1994). The nature of emotion:
Fundamental auestions. Oxford: Oxford Universitv Press.
Ferguson, T. J. (1977). A free resmnse content analvsis of causal
attribution. Unpublished master's thesis, University of Alberta, Edmonton.
Ferguson, T. J., & Stegge, H. (1998). Measuring guilt in children: A rose by
any other name still has thorns. In J. Bybee (Ed.), Guilt and Children (pp. 19-71).
San Diego, CA: Academic Press.
Ferguson, T. J., Stegge, H., Eyre, H. L., Vollmer, R., & Ashbaker, M. (in
press). Context effects and the (ma1)adaptive nature of p i l t and shame in
children. Genetic. Social, and General Psvcholow Monorzrap hs.
Ferguson, T. J., & Wells, G. L. (1980). Priming of mediators in causal
attribution. Journal of Penonalit and Social Psvcholw. 38,461-470.
Fielder, K., Walther, E., & Nickel, S. (1999). Covariation-based attribution:
On the ability to assess multiple covariates of an effect. Personalitv and Social
PsvchoIow Bulletin. 25,607-622.
Forsterling, F. (1989). Models of covariation and attribution: How do they
relate to the analogy of analysis of variance? Joumal of Personality and Social
Psvcholoev. 57,615-625.
Hansen, R. D. (1980). Cornmonsense attribution. Journal of Personality
Page 43
Attribution of Emotion
and Social Psvcholorrv. 39,996-1009.
Hansen, R D., & Donoghue, J. M. (1977). The power of consensus:
Information derived from one's own and others' behavior. Ioumal of
Personalitv and Social Psvcholog. 35.294-302.
Heider, F. (1958). The clsvcholoev of interpersonal relations. New York:
Wiley.
Hewstone, M. (1983). The role of language in attribution processes. In J.
Jaspars, F. D. Fincham, & M. Hewstone (Eds.), Attribution theorv and research:
Concep tual. develoomental and social dimensions (pp. 241-259). London:
Academic Press.
Hewstone, hl., & Jaspars, J. (1983). A re-examination of the roles of
consensus, consistencv, and distinctiveness: Kellev's cube revisited. British
Journal of Social Ps cholop. 22.41-50.
Hilton, D. J., & Jaspars, J. M. F. (1987). The explanation of occurrences and
non-occurrences: A test of the inductive-logic mode1 of causal attribution. British
Journal of Social Psvcholow. 26.189-201.
Hilton, D. J., Smith, R. H., & Kim, S. H. (1995). Processes of causal
explanation and dispositional attribution. Journal of Personalitv and Social
Psvcholow. 68,377-387.
Hupka, R. B., Lenton, A. P., & Hutdiison, K. A. (1999). Universal
development of emotion categories in natural language. Journal of Personality
and Social Psvcholow. 77,247-278.
Jaspars, J. (1983). The process of causal attribution in cornmon sense. in M.
Hewstone (Ed.), Attribution theo : Social and functional extensions ( p p 28-44.
London: Blackwell.
Jaspars, J., Hewstone, M., Fincham, F. D. (1983). Attribution: Theory and
Page 44
Attribution of Emotion 37
research: The state of the art. In J. Jaspars, F. D. Fincham, & M. Hewstone (Eds.),
Attribu lion theory and research: Conceotual. develoomen ta1 and social
dimensions (pp. 3-36). London: Academic Press.
Kassin, S. M. (1979). Consensus information, prediction, and causal
attribution: A revkw of the literature and issues. Journal ~f Pèrsonalitv and
Social Psvchoio~~. 37.1966-1981.
Kellev, H. H. (1967). Attribution theory in social ps
(Ed.), Nebraska y m osium on motivation (Vol. 15, pp. 192-238). Lincoln:
University of Nebraska Press.
Kelley, H. H. (1973). The processes of causal attribution. American
Psvcholoaist. 28.107-128.
Kelley, H. H., & Michela, J. L. (1980). Attribution theory and research.
Annual Review of Psvchology. 31,457-501.
Krk, R. E. (1968). Ex
sciences. Belmont, CA: Brooks/Cole.
Major, B. (1980). Information acquisition and attribution processes. Joumal
of Personali tv and Social Psvcholoey. 39.1010-1023.
McArthur, L. A. (1972). The how and what of why: Some determinants
and consequences of causal attribution. Joumal of Personality and Social
Psycholow. 22.171-193.
Mckthur, L. 2. (1976). The lesser influence of consensus than
distinctiveness information on causal attributions: A test of the person-thing
hypothesis. Joumal of Personali
Orvis, B. R., Cunningham, J. D., & Kelley, H. H. (1975). A closer
examination of causal inference: The roles of consensus, distinctiveness, and
consistency information. Journal of Personalit
Page 45
Attribution of Emotion 38
Parkinson, B. (1995). Ideas and realities of emotion. New York: Routledge.
Parkinson, B. (1996). Emotions are social. British journal of Psycholo~ . 87,
663-683.
Ruble, O. N., t Feldman, N. S. (19%). Order of ccnsensus, di~tinctiveness,
and consistency information and causal attributions. Journal of Personalitv and
Social Psvcholoev. 34.930-937.
Russell, J. A. (1997). How shall an emotion be called? Ln R. Plutchik & H. R.
Conte (Eds.), Circumolex models of oersonalit and emotion (pp. 205-220).
Washington, DC: American Psychological Association.
Russell, J. A., & Barrett, L. F. (1999). Core affect, protot
episodes, and other things called Emotion: Dissecting the elephant. Journal ot
Personalitv and Social Psvcholmv, 7& 805-819.
Scherer, K. R. (1994). Toward a concept of "modal emotions." In P. Ekman
& R. J. Davidson (Eds.), The nature of emotion: Fundamental
31). New York: Oxford University Press.
Shweder, R. A. (1994). "You're not sick, you're just in love": Emotion as
an interpretive system. In P. Ekman & R. J. Davidson (Eds.), The nature of
emotion: Fundamental questions (pp. 32-44). New York: Oxford University
Press.
Smith, E. R., & Miller, F. 0. (1983). Mediation among attributional
inferences and comprehension processes: Initial findings and a general method.
Journal of Personalitv and Social Psvch~l~p,r . 4-4.492-505.
Wells, G. L., & Harvey, J. H. (1977). Do people use consensus information
in making causal attributions? Journal of Personalit and Social PsFholoatt 35.
279-293.