ArticlePDF Available

The unpacking effect in evaluative judgments: When the whole is less than the sum of its parts

Authors:

Abstract and Figures

Any category or event can be described in more or less detail. Although these different descriptions can reflect the same event objectively, they may not reflect the same event subjectively. Research on Support Theory led us to predict that more detailed descriptions would produce more extreme evaluations of categories or events than less detailed descriptions. Four experiments demonstrated this unpacking effect when people were presented with (Experiments 1 and 4), generated (Experiment 2), or were primed with (Experiment 3) more rather than less detailed descriptions of events. This effect was diminished when the details were less personally relevant (Experiment 4). We discuss several psychological mechanisms, moderators, and extensions of the unpacking effect.
Content may be subject to copyright.
The unpacking effect in allocations of responsibility for group tasks
q
Kenneth Savitsky
a,*
, Leaf Van Boven
b
, Nicholas Epley
c
, Wayne M. Wight
a
a
Department of Psychology, Williams College, Bronfman Science Center, Williamstown, MA 01267, United States
b
University of Colorado, Boulder, United States
c
Harvard University, United States
Received 14 July 2003; revised 5 August 2004
Available online 18 November 2004
Abstract
Individuals tend to overestimate their relative contributions to collaborative endeavors. Thus, the sum of group membersÕesti-
mates of the percentage they each contributed to a joint task typically exceeds the logically allowable 100%. We suggest that this
tendency stems partly from individualsÕinclination to regard their fellow group members as a collective rather than as individuals,
and that leading people to think about their collaborators as individuals should therefore reduce the perceived relative magnitude of
their own contributions. Consistent with this thesis, four experiments demonstrate that peopleÕs tendency to claim more than their
fair share of the credit for a group task is attenuated when they ‘‘unpack’’ their collaborators, conceptualizing them as separate
individuals, rather than as ‘‘the rest of the group.’’
2004 Elsevier Inc. All rights reserved.
William Maxwell, who served as an editor at The New
Yorker for more than 40 years, believed that a light
touch was best when it came to editing the works that
crossed his desk—manuscripts from the likes of John
Updike, John Cheever, and J. D. Salinger. ‘‘I tried to
work so slightly,’’ Maxwell said, ‘‘that...the writer would
read his story and not be aware that anybody was in-
volved but him’’ (Gates, 2000). Without detracting from
the skill with which Maxwell practiced his craft, a large
body of research suggests that little editorial restraint
was necessary for MaxwellÕs authors to reach the con-
clusion that they alone produced the final product. Peo-
ple are notoriously prone to take more than their fair
share of the credit for collaborative endeavors, even
when others have made important and sizable contribu-
tions (Ross, 1981; Ross & Sicoly, 1979). It would thus
not be surprising for authors to overestimate the role
they played in the development of a work and underes-
timate othersÕcontributions.
This tendency to claim more than oneÕs fair share of
the credit for a collaborative endeavor may be especially
pronounced when an individual works with several oth-
ers—a writer, say, whose final product reflects not only
her own efforts, but the work of publishers, editors,
proof-readers, fact-checkers, and so on. In order for
an individual to assess the relative magnitude of his or
her own contributions in such cases, it is necessary to
consider not only his or her own contributions but also
those of each of his or her collaborators. Any tendency
to give insufficient attention to the contributions of one
or more of his or her collaborators will lead the individ-
ual to overestimate his or her own inputs. Accordingly,
we suggest that peopleÕs overestimation of their contri-
butions to collaborative endeavors stems partly from
their tendency to regard at least some of their collabora-
tors collectively, as the ‘‘rest of the group’’—and that
0022-1031/$ - see front matter 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jesp.2004.08.008
q
Lest we take more than our fair share of the credit for this
research, we gratefully acknowledge Steve Fein, Kris Kirby, Derek
Koehler, and Justin Kruger for their helpful comments and advice; and
John Finkbeiner, Jeff Manning, Gabriela Pereira, Carole Savitsky,
Matt Shafeek, and Kevin Van Aelst for their help in collecting these
data. This research originated as an undergraduate honors thesis
conducted by Wayne Wight. Portions of this research were supported
by NSF Grant SES0241544 awarded to Nicholas Epley.
*
Corresponding author. Fax: +1 413 597 2085.
E-mail address: ksavitsk@williams.edu (K. Savitsky).
www.elsevier.com/locate/jesp
Journal of Experimental Social Psychology 41 (2005) 447–457
inducing individuals to consider their collaborators indi-
vidually will therefore reduce their tendency to overesti-
mate their own relative contributions. A person who is
led to consider his or her collaborators as individuals,
each making individual contributions, is likely to see
them as relatively more productive than he or she would
have otherwise, and will therefore decrease estimates of
his or her own relative productivity.
We base this prediction on two bodies of research.
First, research on support theory demonstrates that con-
sidering the constituent elements of a category sepa-
rately, rather than as a whole, makes them seem more
probable and frequent (e.g., Tversky & Koehler, 1994).
This is because ‘‘unpacking’’ the constituent elements
of a category increases their cognitive accessibility and
leads people to consider elements they might not have
considered otherwise. Second, research on egocentrism
in allocations of responsibility shows that the accessibil-
ity of oneÕs own contributions—as opposed to those of
oneÕs collaborators—is a key determinant of individualsÕ
tendency to overestimate their own relative contribu-
tions to collaborative endeavors (e.g., Ross & Sicoly,
1979).
Egocentric allocations of responsibility
Individuals who work with others on collaborative
endeavors—whether married couples who share the
housework, members of a task force who work together
to implement a new policy, or academic colleagues who
co-author a paper—tend to overestimate the magnitude
and importance of their own contributions, and under-
estimate the magnitude and importance of othersÕcon-
tributions. As a consequence, the sum of each
collaboratorÕs self-assessed contributions typically ex-
ceeds 100%. Logically, of course, this cannot be; if three
collaborators each believe they have done 50% of the
work, then at least one of them is wrong. This phenom-
enon was first documented by Ross and Sicoly (1979)
and has been replicated many times since (e.g., Brawley,
1984; Burger & Rodman, 1983; Christensen, Sullaway,
& King, 1983; Deutsch, Lozy, & Saxon, 1993; Gilovich,
Medvec, & Savitsky, 2000; Kruger & Gilovich, 1999;
Kruger & Savitsky, 2004; Thompson & Kelly, 1981;
for a review, see Leary & Forsyth, 1987).
To be sure, this tendency is partly produced by
individualsÕdesire to think well of themselves and to
present themselves in a positive light. Individuals some-
times engage in a motivated ‘‘grab for credit’’ in which
they claim to have contributed an inflated proportion
of the work in order to reap a correspondingly inflated
share of the self- and social-rewards that can be
expected to follow (Miller, Goldman, & Schlenker,
1985; Ross & Sicoly, 1979, Experiment 2; Schlenker
& Miller, 1977).
But there is more to egocentric allocations of respon-
sibility than self-serving motives (Leary & Forsyth,
1987). The tendency to claim more than oneÕs fair share
of the credit for collaborative tasks also occurs because
oneÕs own contributions tend to be more cognitively
accessible than othersÕcontributions (Ross & Sicoly,
1979). Because both the amount of information re-
trieved and the ease with which it can be brought to
mind are used as heuristics for estimating overall fre-
quency (Schwarz et al., 1991; Tversky & Kahneman,
1973), individuals conclude that their own contributions
were more substantial, on average, than they actually
were (Greenwald, 1980; Neisser, 1981). Consistent with
this interpretation, research demonstrates that individu-
als overestimate their contributions not only to activities
that reflect positively on them, but also to activities that
reflect negatively (Brawley, 1984; Kruger & Gilovich,
1999; Ross & Sicoly, 1979; Thompson & Kelly, 1981).
Spouses, for example, not only overestimate the propor-
tion of the housework they have done and the affection
they have demonstrated, but also the proportion of
things they have broken and arguments they have
started—a finding that is inconsistent with an explana-
tion based solely on self-aggrandizement (Kruger &
Gilovich, 1999; Ross & Sicoly, 1979).
In sum, it is clear that individualsÕtendency to over-
estimate their contributions to collaborative tasks oc-
curs partly because their own inputs tend to be more
accessible than those of their collaborators. We suggest
that when individuals collaborate on a project with
more than one other person, this tendency is aided
and abetted by a failure to consider each collaborator
as a separate individual, regarding at least some of them
collectively as ‘‘the rest of the group.’’ A tendency to
aggregate oneÕs collaborators may stem partly from at-
tempts to simplify an otherwise complex assessment of
relative responsibility, or from an inclination to catego-
rize the world egocentrically as ‘‘me’’ vs. ‘‘not me’’
(James, 1892). Regardless, if individuals mentally
‘‘pack’’ their collaborators, then encouraging them to
consider their collaborators as separate individuals
(i.e., to ‘‘unpack’’ them) should increase the accessibility
of their collaboratorsÕcontributions, thereby decreasing
their estimates of the relative magnitude of their own
contributions.
Support theory
This prediction follows from research on support the-
ory, which shows, among other things, that leading peo-
ple to consider the constituent elements of a category
separately rather than holistically increases the per-
ceived probability and frequency of that category (Rot-
tenstreich & Tversky, 1997; Tversky & Koehler, 1994;
see also Brenner & Koehler, 1999;Fischhoff, Slovic, &
448 K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457
Lichtenstein, 1978; Johnson, Hershey, Meszaros, &
Kunreuther, 1993; Kruger & Evans, 2004; Macchi, Osh-
erson, & Krantz, 1999; Mulford & Dawes, 1999; Red-
elmeier, Koehler, Liberman, & Tversky, 1995; Russo
& Kolzow, 1994; Tversky & Fox, 1995; Tversky &
Kahneman, 1973; Van Boven & Epley, 2003; see Bren-
ner, Koehler, & Rottenstreich, 2002, for a review). For
instance, participants in one study indicated that a per-
son was more likely to die from ‘‘heart disease, cancer,
or other natural causes’’ than simply from ‘‘natural
causes,’’ even though the category natural causes in-
cludes heart disease, cancer, and a host of other condi-
tions as well (Tversky & Koehler, 1994).
One reason for this effect is that unpacking the con-
stituent elements of a category—that is, considering
each element separately—renders those elements easier
to think about and leads people to think of elements
they would not have considered otherwise (Rottenstr-
eich & Tversky, 1997; Tversky & Koehler, 1994).
Accordingly, we propose that if individuals overestimate
their contributions to collaborative tasks in part because
of the greater relative accessibility of their own contribu-
tions, then leading them to think about their collabora-
tors individually rather than collectively should diminish
this tendency. We report four experiments that test this
prediction—that is, whether instructions to unpack their
collaborators significantly reduce individualsÕtendency
to overestimate the magnitude of their contributions to
a joint task.
Study 1
Elementary school children participating in a team-
based extracurricular program on creative problem solv-
ing estimated their own contributions to their teamÕsfi-
nal product (a written document). Those in the control
condition simply indicated the proportion of the work
they contributed with no mention of the other members
of their team. Those in the unpacked condition indicated
their contribution after estimating the proportion of
work that each of their collaborators contributed, with
the provision that the four allocations had to sum to
100%. The four members of each team were always as-
signed to the same condition so that their four self-allo-
cations of responsibility could be summed and
compared to a baseline of 100%.
We expected that instructing participants to appor-
tion responsibility to each of their teammates
separately would increase allocations of responsibility
to participantsÕteammates, and thus decrease the
proportion of responsibility claimed by participants
themselves. Thus, we expected participants in the un-
packed condition to overestimate their contributions
to the group task less than participants in the control
condition.
Method
Participants
One hundred thirty-two fourth-grade students work-
ing in one of 33 teams of four participated in the
study (63 males, 69 females). All participants were
enrolled in the ‘‘Future Problem Solving Program’’
(FPSP; see www.fpsp.org), a national program in which
elementary and high school students work in teams to
explore various real-world problems (e.g., depletion of
the rainforests, homelessness, violence in schools). After
researching their problem for several weeks, each team
spends two hours producing a written document pro-
posing solutions for their assigned problem. Judges at
a central office then evaluate these documents.
Procedure
We contacted the coaches of several FPSP teams at
an elementary school in central Indiana and invited their
teams to participate in our research. Questionnaires
were distributed to the students by their coaches several
weeks after they had completed a project on the topic of
organ donation. Thus, prior to receiving our question-
naires, all participants had collaborated with three other
students for several weeks and had spent two hours pre-
paring a written document. They had not yet received
any official feedback on their performance.
Participants were asked to think back to their work
on the problem they had just completed:
Working on a problem involves lots of things: reading
and remembering background research, being creative,
writing down problems and solutions, understanding
the problem-solving process, keeping the team on track,
keeping track of time, and so on. When it comes to
group projects like this, the work isnÕt always divided
evenly. For a variety of reasons, people often do more
or less than an even share.
Participants were then asked to allocate responsibility
for the groupÕs final product. Those randomly assigned
to the control condition (n= 16 groups) were asked to
indicate the proportion of the overall work for which
they were responsible. They did so by indicating a per-
centage between 0% (did none of it) and 100% (did all
of it). Those assigned to the unpacked condition
(n= 17 groups) were asked to record the initials of each
of their teammates and to check off each set of initials
after they had taken a moment to consider ‘‘that personÕs
participation in and contributions to your work on this
problem.’’ These participants were then asked to appor-
tion responsibility to each of their teammates, one by
one, and then to themselves. They were reminded that
their four allocations must sum to 100%. All members
of each team were assigned to the same condition so that
self-allocations could be summed across the members
of each team and compared with a baseline of 100%.
K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457 449
Results and discussion
Gender did not influence the results in this or any of
the following experiments and is therefore not discussed
further.
Self-allocations of responsibility, summed across the
members of each team, exceeded 100% in both the con-
trol condition (M= 154.6%), t(15) = 6.13, p< .0001,
and the unpacked condition (M= 106.8%),
t(16) = 2.54, p< .05. As expected, however, teams as-
signed to the unpacked condition claimed significantly
less responsibility than did teams assigned to the control
condition, t(31) = 5.28, p< .0001. Instructing partici-
pants to consider each of their teammates separately
substantially diminished their tendency to overestimate
the magnitude of their own contributions, but did not
eliminate it entirely.
1
Study 2
Study 2 was designed to replicate the results of
Study 1 in a different domain with an older (and pre-
sumably more mature) sample of participants. We
again made use of naturally occurring groups by exam-
ining undergraduate business students who had been
working in groups on a class project. Two weeks after
completing their projects, students were asked to indi-
cate the proportion of the work they had personally
contributed to each of four activities. Once again, stu-
dents in groups assigned to the control condition sim-
ply indicated the proportion for which they were
responsible, whereas students in groups assigned to
the unpacked condition allocated responsibility to
themselves and to each of their other group members.
We again expected unpacking to diminish participantsÕ
tendency to overestimate the magnitude of their
contributions.
Method
Participants
Participants included the complete enrollment
(N= 104) of two introductory undergraduate marketing
classes at the University of British Columbia.
Procedure
Early in the academic term, participants were ran-
domly assigned to groups of four to work on a pa-
per—a case analysis of a business situation in
which students were to recommend a marketing strat-
egy for a novel household appliance. Groups met
regularly over four weeks, both in class and outside
of class, to discuss their project. Fifteen days after
they handed in their group paper, participants were
asked to complete a questionnaire in exchange for
course credit regarding the experience of working in
their groups. Participants were assured that their re-
sponses were confidential and would not be seen by
their instructor. They had not yet learned of their
grade for their project when they completed the
questionnaire.
Participants completed questionnaires similar to
those used in Study 1. This time, however, they were
asked to indicate the proportion of the work—from
0% (none of it)to100%(all of it)—they had contrib-
uted to each of four activities: writing,creativity and
idea generation,scheduling and administrative work,
and overall work. Participants in groups randomly as-
signed to the control condition (n= 13 groups) did so
without any mention of the other members of their
group. Participants in groups assigned to the unpacked
condition (n= 13 groups), in contrast, did so after
indicating the proportion of work performed by each
of the three other members of their group. Specifically,
participants in the unpacked condition were asked to
record the initials of each of their group members
and then to take a moment to think about each indi-
vidualÕs participation in, and contributions to, the
group. After considering each individualÕs contribu-
tions, they were asked to place a check mark next to
that personÕs initials and to move to the next person
until they had considered the contributions of each
group member. Participants were then asked to indi-
cate the proportion of work performed by each group
member, including themselves. They were reminded
that their four allocations must sum to 100%.
Results and discussion
Once again, we expected participants in the control
condition, who simply allocated responsibility to them-
selves, to overestimate the magnitude of their contribu-
tions to the group task more than participants in the
packed condition, who allocated responsibility to them-
selves and to each of the other members of their group.
ParticipantsÕresponses yielded strong support for this
prediction. As can be seen in Table 1, the summed re-
sponses of groups assigned to the control condition ex-
ceeded 100% by a larger margin, for each of the four
activities, than did the summed responses of groups as-
signed to the unpacked condition. To analyze these re-
1
One member of each of two teams failed to complete the
dependent measures so we used the average of his or her teammatesÕ
self-allocations as a proxy for the missing responses. In addition, in five
groups assigned to the unpacked condition, one or more team
members provided allocations of responsibility for the team that did
not sum to 100%, as instructed. Because we could detect this error only
among participants in the unpacked condition, we included these
participantsÕunaltered responses in our analyses. Re-computing the
analyses excluding all seven of these teams, however, does not affect
the results reported above.
450 K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457
sults statistically, we averaged across participantsÕre-
sponses to the four items to create an overall index
of responsibility allocations (a= .92). As in Study 1,
summed self-allocations exceeded 100% for groups in
both the control and unpacked conditions
(Ms = 141.6 and 110.3%, respectively), ts(12) > 3.64,
ps < .005. As predicted, however, responsibility alloca-
tions were significantly reduced in the unpacked condi-
tion compared to the control condition, t(24) = 4.14,
p< .001.
Study 2 suggest that university business students bear
remarkable similarity to the elementary school students
from Study 1 when it comes to allocating responsibility
for a group project. In both cases, inducing participants
to unpack their group and consider their collaborators
separately substantially diminished (but did not elimi-
nate) their tendency to overestimate the magnitude of
their contributions.
Study 3
The results of Studies 1 and 2 demonstrate that
instructing participants to unpack their collaborators re-
duces their tendency to overestimate the magnitude of
their contributions to a collaborative task. We maintain
that unpacking has this effect because it leads individu-
als to think of othersÕcontributions that they may not
have otherwise considered, and therefore renders othersÕ
contributions more cognitively accessible (cf. Tversky &
Koehler, 1994). This, in turn, allows individuals to rec-
ognize that their own contributions account for a smal-
ler ‘‘slice of the pie’’ than they would otherwise believe.
There are, however, a number of possible alternative
interpretations of these results. First, recall that partici-
pants in the unpacked condition (but not those in the
control condition) were informed that their responsibil-
ity allocations had to sum to 100%. It is conceivable that
this requirement served as a ‘‘reality check,’’ reminding
participants that there was only so much responsibility
to go around (specifically, 100% of it) and alerting them
that their allocations of responsibility should be con-
strained accordingly. The mere mention of the 100%
constraint, then, may have led participants in the un-
packed condition to consider their allocations of respon-
sibility more carefully, perhaps resulting in more
conservative self-allocations. Alternatively, to the extent
that participants converted the 100% constraint into
‘‘about 25% each,’’ self-allocations could have been de-
creased in the unpacked condition as a result of an
anchoring effect (i.e., a tendency to cling to and/or ad-
just insufficiently from the ‘‘anchor’’ of 25%; Chapman
& Johnson, 2002; Epley, 2004). It is instructive to recall
that Brenner & Koehler (1999) demonstrated unpacking
effects even when a set of judgments was required to sum
to 100%, a finding that casts doubt on the possible alter-
natives just mentioned. Nevertheless, we address this is-
sue empirically in the next study.
Finally, note that participants in the unpacked con-
dition were required to make multiple allocations of
responsibility (i.e., to all group members), whereas
those in the control condition made only one allocation
(i.e., to themselves). If participants in the unpacked
condition were disinclined to allocate a small amount
of responsibility to one or more of their collaborators
(e.g., out of politeness, group loyalty, or a reluctance
to use extreme ends of the scale; Fiedler & Armbruster,
1994), then the amount allocated to ‘‘others’’ may have
been increased artifactually, resulting in a decrease in
self-allocations.
To address these issues, we added two new condi-
tions to our design for Study 3: a packed condition
and an implicitly unpacked condition. Participants as-
signed to the packed condition were asked to divide
responsibility for a collaborative endeavor between
themselves and ‘‘the rest of the group,’’ with the stipu-
lation that these two allocations sum to 100%. Because
this condition did not require participants to unpack
their collaborators, we did not expect a reduction in
the degree to which they overestimated the magnitude
of their own contributions—despite the fact that the
instructions did mention the 100% ‘‘reality’’ constraint.
Participants assigned to the implicitly unpacked condi-
tion (cf. Tversky & Koehler, 1994), in contrast, were
asked to consider each of their collaborators individu-
ally, as in the unpacked condition from Studies 1 and
2, but were not asked to make separate allocations of
responsibility for each of them. Because this condition
required participants to unpack their collaborators
(mentally, at least), we expected to observe a reduction
in the degree to which they overestimated the magni-
tude of their own contributions—despite the fact that
they did not make multiple allocations of
responsibility.
Method
Participants
Eighty-one Cornell University students (53 women,
28 men) participated in exchange for extra credit in their
psychology or human development courses.
Table 1
Claimed percentage responsibility for four activities by participants in
the control and unpacked conditions, summed across the members of
each group, Study 2
Activity Condition
Control Unpacked
Writing 135.5 104.9
Idea generation 147.8 110.4
Administrative work 137.3 114.9
Overall work 145.9 111.0
K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457 451
Procedure
Participants were asked to recall a time when they
had worked in a group of three to six people to complete
a task. They were told that they were free to select a
group from any domain that they wished—e.g., a group
from a class they had taken in high school or college or
from a job they had held—but were instructed to have a
specific group in mind. Participants described their
group in a few sentences and indicated how many people
(including themselves) had been in the group. They then
rated their group experience on a number of bipolar
scales ranging from 3 to +3: unpleasant /pleasant,bor-
ing /interesting,useless /useful,inefficient /efficient,and
failure /success. Participants also indicated how inter-
ested they would be in working with the group they
listed in the future, on a scale ranging from 0 (not at
all)to10(very much).
Finally, participants were asked to indicate the pro-
portion of the work—from 0% (none of it) to 100%
(all of it)—they contributed to each of the same four
activities examined in Study 2: writing,creativity and
idea generation,scheduling and administrative work,
and overall work. Participants were reminded that even
if their group had done very little of some activity
(e.g., only a small amount of writing), some person or
group of people was still responsible for 100% of it. If,
however, their group had not engaged in any of some
activity, they were instructed to mark the item as ‘‘not
applicable.’’
Participants were randomly assigned to one of four
conditions, two of which were identical to the conditions
in Studies 1 and 2. Those assigned to the control condi-
tion (n= 20) indicated the proportion of each activity
for which they themselves were responsible without
any mention of their collaborators. Participants as-
signed to the unpacked condition (n= 21) recorded the
initials of each of their collaborators, took a moment
to think about each individualÕs contributions, and then
allocated responsibility to each member of their group
(including themselves), one by one. As before, we re-
minded participants in the unpacked condition that their
allocations must sum to 100%.
We also added two new conditions not included in
Studies 1 and 2. Participants assigned to both the
packed (n= 20) and implicitly unpacked (n= 20) condi-
tions allocated responsibility two ways—to ‘‘the other
group members’’ (i.e., the rest of their group, as a single,
holistic entity), and to themselves; in both conditions,
they were reminded that these two allocations must
sum to 100%. Participants in the packed condition re-
ceived no additional instructions, whereas participants
assigned to the implicitly unpacked condition were
asked to record the initials of their collaborators and
think about each individualÕs contributions, just as par-
ticipants did in the unpacked condition. In short, the
packed condition resembled the control condition in
that neither included instructions for participants to re-
gard their collaborators as individuals, but differed from
the control condition in that participants in the packed
condition were asked to allocate responsibility to their
collaborators (albeit holistically), and were reminded
of the 100% constraint. Likewise, the implicitly un-
packed condition resembled the unpacked condition in
that both included instructions for participants to re-
gard their collaborators as individuals (i.e., list their ini-
tials and so on), but differed from the unpacked
condition in that participants in the implicitly unpacked
condition were not asked to make allocations of respon-
sibility for each of their collaborators separately.
Results and discussion
Participants selected groups with which they were
quite satisfied. They rated their groups on average as
having been pleasant (M= .95), interesting (M= .88),
useful (M= .88), efficient (M= .46), and successful
(M= 1.75), and indicated a willingness to work with
their groups again (M= 5.69). All of these means were
significantly above the midpoint of each scale, all
ts(80) > 2.11, all ps < .05. Because they were randomly
assigned to conditions, we did not expect any between-
condition differences in the groups participants selected.
And indeed, separate one-way analyses of variance indi-
cated no significant differences in any of participantsÕ
ratings of their groups. There were also no between-con-
dition differences in group size (M= 4.32 group mem-
bers, including participants themselves).
We next examined participantsÕallocations of respon-
sibility in each of the four conditions. We expected that
instructing participants to consider their collaborators
as individuals—whether participants were asked to allo-
cate responsibility to them each separately (unpacked
condition) or as a group (implicitly unpacked condi-
tion)—would render their collaboratorsÕcontributions
more cognitively accessible relative to the packed and
control conditions. As a result, we expected participants
in the unpacked and implicitly unpacked conditions to
allocate relatively less responsibility to themselves than
would participants in the packed and control conditions.
To examine between-condition differences across
groups of different size, we multiplied each participantÕs
estimate of his or her own contribution by the reported
size of his or her group (e.g., an individual who worked
in a group of four and claimed to have done 25% of the
work would receive a value of 100%). Table 2 presents
the means of these transformed values across the four
conditions for each of the four dependent measures.
Although the responses of participants in all conditions
appeared to exceed 100%, care should be taken in inter-
preting values in excess of 100% as evidence of ‘‘overes-
timation.’’ Because participants were free to select any
group they wished, it is possible that they called to mind
452 K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457
and reported groups in which they themselves played an
especially large role (i.e., groups in which they really did
do more than their fair share of the work), meaning that
one can hardly fault them for making self-allocations
that, when multiplied by group size, exceed 100%. Thus,
because we did not obtain responses from all members
of participantsÕgroups, as we did in Studies 1 and 2,
we cannot address the extent to which participants over-
estimated their own contributions.
Importantly, however, because this feature of our de-
sign was equally true for participants in all four of our
conditions, we can address between-condition differ-
ences. To do so, we averaged across participantsÕre-
sponses to the four items to create an overall index of
responsibility allocations (a= .81).
2
Analysis of this in-
dex indicated that allocations of responsibility did in-
deed vary by condition, F(3, 77) = 3.65, p< .025.
Planned contrasts revealed that participantsÕself-alloca-
tions
3
in the control and packed conditions did not differ
from one another (Ms = 180.8 and 183.9%, respec-
tively), nor did the self-allocations of participants in
the unpacked and implicitly unpacked conditions
(Ms = 135.8 and 128.1%, respectively), both ts < 1. As
expected, however, participants in the control and
packed conditions claimed more responsibility for their
groupÕs output than did participants in either the un-
packed condition, ts(77) = 2.09 and 2.24, respectively,
ps < .05, or the implicitly unpacked condition,
ts(77) = 2.42 and 2.56, respectively, ps < .025.
These findings are consistent with Studies 1 and 2,
and again demonstrate that inducing individuals to un-
pack their collaborators attenuates the amount of
responsibility they allocate to themselves for collabora-
tive tasks. Moreover, the results from the packed and
implicitly unpacked conditions rule out a number of
possible alternative interpretations for the findings from
Studies 1 and 2—namely, that they stemmed from con-
servative self-allocations inspired by the 100% ‘‘reality
check’’ constraint, that they stemmed from an anchoring
effect, or that they stemmed from an artifact of making
multiple allocations of responsibility. Study 3 thus lends
further support to our position that unpacking influ-
ences responsibility allocations because it renders the
contributions of oneÕs collaborators more accessible
than they would otherwise be.
Study 4
The first three studies indicate that leading individu-
als to consider their collaborators separately, rather
than as a group, diminishes the extent to which they
overestimate their contribution to collaborative tasks.
In Study 4, we wished to replicate this result once again,
this time using a somewhat different technique to
encourage participants to unpack their collaborators,
thereby extending the breadth of the unpacking effect
in this context. Doing so also allowed us to address an
additional possible alternative interpretation of our re-
sults from the first three studies. It could be argued that
the unpacking manipulations in our previous studies
constituted a subtle suggestion to participants that they
should allocate less responsibility to themselves than
they might otherwise. Instructions to list the initials of
their collaborators, think about each personÕs contribu-
tions, and (in some conditions) allocate a portion of the
total responsibility to each of them, may therefore have
served as something of a demand characteristic, alerting
participants to what responses were expected and/or
normative.
We took steps in Study 4 to eliminate this possibility
by disguising the unpacking manipulation, making it
significantly more subtle than in the previous studies.
Specifically, participants in Study 4 were asked to think
about a group of which they had been a member, as in
Study 3, but this time were then asked to draw a picture
related to their group. Those in the control condition
were asked to draw a picture of the physical location
in which their group worked. Those in the unpacked
condition, in contrast, were asked to draw a picture of
themselves and each of the other members in their
group—thereby unpacking their group into its constitu-
ent members. All participants then indicated only the
proportion of the work they themselves contributed.
By not asking participants in the unpacked condition
explicitly to stop and consider each member of their
group, or think about other group membersÕcontribu-
tions in any way, this procedure constitutes a relatively
subtle manipulation of unpacking—one that, if success-
ful, minimizes concerns about possible demand charac-
teristics in Studies 1, 2, and 3.
2
Eight participants (including participants in each of the four
conditions) left blank a total of nine self-allocations of responsibility,
five for writing, two for creativity and idea generation, and three for
scheduling and administrative work.
3
Strictly speaking, these means are not themselves self-allocations
of responsibility because they have been transformed (multiplied by
group size). To remain consistent with our earlier studies, however, we
use the term ‘‘self-allocations’’ throughout the remainder of this paper.
Table 2
Claimed percentage responsibility for four activities by participants in
the control, packed, implicitly unpacked, and unpacked conditions,
multiplied by group size, Study 3
Activity Condition
Control Packed Implicitly
unpacked
Unpacked
Writing 180.0 185.8 112.6 128.0
Idea generation 168.0 170.0 120.5 128.1
Administrative work 200.5 222.6 144.2 149.5
Overall work 169.5 172.5 138.2 141.3
K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457 453
Despite various procedural changes, we expected our
results to be consistent with those from our earlier stud-
ies. Specifically, we expected participants in the un-
packed condition to allocate less responsibility to
themselves than did participants in the control condition.
Method
Participants
Eighty-nine individuals participated in exchange for
extra credit in their psychology class or on a volunteer
basis. Participants were students at Harvard University
(n= 52), the University of Colorado, Boulder (n= 26),
or Williams College (n= 11).
Procedure
Participants completed a questionnaire similar to the
one used in Study 3. Specifically, participants were asked
to recall a time when they had worked with a group of
three to six people to complete a task. They described
their group in a few sentences and indicated how many
people (including themselves) had been in the group.
Next, participants were asked to spend a few minutes
drawing a picture related to their group. Participants in
the control condition were given a page with a blank sec-
tion (approximately 15 cm ·15cm) and were asked to
draw a representation of the physical location in which
their group had worked, including furniture, doorways,
and other contextual elements of the space. They were
told that their drawings would later be analyzed for in-
sights into how they conceptualized the context in which
they had worked.
Participants in the unpacked condition, in contrast,
were given a page on which the blank section was di-
vided into six areas, labeled ‘‘Me,’’ ‘‘Group member
A,’’ ‘‘Group member B,’’ and so on. These participants
were asked to spend a few minutes drawing a picture of
each individual that had been a member of their group,
including themselves. They were told that their drawings
should capture the essential features of each individual,
as he or she appeared during the project. Importantly,
no mention was made at this point of contributions or
allocating responsibility, nor were these participants
asked explicitly to think about their collaborators as
individuals. Participants in the unpacked condition were
simply told that their drawings would later be analyzed
for insights into how they conceptualized themselves
and the other members of their group.
Finally, participants were asked to indicate the pro-
portion of the work they had contributed for each of
the same four activities examined in Studies 2 and 3.
Results and discussion
There was no between-condition difference in the size
of the groups participants recalled (M= 4.62 group
members, including participants themselves). As in
Study 3, in order to examine between-condition differ-
ences on the dependent measures across groups of differ-
ent size, we multiplied each participantÕs estimate of his
or her contribution by the reported size of his or her
group. Table 3 presents the means of these transformed
values for both conditions for each of the four depen-
dent measures, adjusted for participantsÕinstitution
(Harvard, Colorado, or Williams). As before, the re-
sponses of participants in both conditions appeared to
exceed 100%, but we again caution that such values do
not necessarily indicate overestimation because partici-
pants recalled their own groups and we did not obtain
responses from all group members.
More important, further analyses showed that the
unpacking manipulation reduced the proportion of
responsibility participants allocated to themselves. We
averaged across participantsÕresponses to the four items
to create an overall index of responsibility allocations
(a= .78)
4
and subjected this index to an analysis of
covariance, controlling for participantsÕinstitution with
two dummy variables. This analysis revealed a signifi-
cant effect for condition, F(1, 85) = 4.95, p< .05.
5
As ex-
pected, participants assigned to the unpacked condition
allocated less responsibility to themselves (adjusted
M= 131.4%) than did participants assigned to the con-
trol condition (adjusted M= 172.3%).
The results of this study offer a replication of the find-
ings from Studies 1, 2, and 3. Our relatively subtle
manipulation of unpacking was enough to lower the
magnitude of participantsÕself-allocations of responsi-
bility. At the same time, these results cast doubt on
the possibility that our findings in the previous studies
reflected the operation of a demand characteristic.
Finally, note that the results of Study 4 also allow us
to address another question left unanswered in our pre-
vious studies—namely, whether the 100% constraint is
necessary for reducing self-allocations of responsibility.
Recall that our packed condition in Study 3, in which
the 100% constraint was made salient, gave rise to self-
allocations of responsibility that were just as high as
those of participants in the control group. Thus, the
requirement that allocations sum to 100% is not suffi-
cient, by itself, to lower individualsÕself-allocations of
responsibility. In addition, the results of Study 4 show
that the 100% constraint is not a necessary condition
4
Twelve participants (including participants in both conditions)
left blank a total of 16 self-allocations of responsibility, nine for
writing, two for creativity and idea generation, three for administrative
work, and two for overall work.
5
Additional analyses revealed some differences in the amount of
responsibility claimed by participants from the three schools, with
participants from Harvard claiming the most responsibility and
participants from Williams claiming the least—which is why institution
was included as a covariate. Importantly, however, there was no
significant interaction between participantsÕinstitution and condition.
454 K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457
for unpacking effects to emerge in responsibility alloca-
tion. We observed a reduction in self-allocations of
responsibility among participants in the unpacked con-
dition despite the fact that those participants were not
required to ensure that their allocations summed to
100%—or even give the 100% constraint any consider-
ation at all.
General discussion
The tendency to claim more than oneÕs fair share of
the credit for collaborative tasks is ubiquitous in social
life—among co-authors, co-workers, and co-habitators
alike (Ross, 1981). Previous research has indicated that
this phenomenon stems partly from individualsÕten-
dency to be more aware of their own contributions than
those of others and bring them to mind with relative ease
(Ross & Sicoly, 1979). The present research suggests that
this tendency is exacerbated by individualsÕpropensity
to regard their fellow group members collectively, rather
than individually. In four studies, we showed that induc-
ing participants to unpack the members of their group
and think about them as individuals caused them to allo-
cate more responsibility to their collaborators—and cor-
respondingly less to themselves. The results of Study 3
helped us rule out a number of possible alternative inter-
pretations of our results—i.e., that they stemmed from
conservative self-allocations inspired by the 100% ‘‘real-
ity check’’ constraint, that they stemmed from an
anchoring effect, or that they stemmed from an artifact
of making multiple allocations of responsibility. Study
4 cast doubt on the possibility that our earlier findings
stemmed from demand characteristics and showed that
consideration of the 100% constraint was not necessary
for unpacking to bring about a reduction in participantsÕ
self-allocations of responsibility.
Taken together, these studies reinforce our hypothe-
sis that unpacking reduces the degree to which individu-
als overestimate the magnitude of their contributions by
increasing the accessibility of their collaboratorsÕcontri-
butions. Note that this interpretation underscores a
recurrent message from the research literature that cog-
nitive accessibility is a key mechanism in judgments of
relative responsibility (Brawley, 1984; Burger & Rod-
man, 1983; Kruger & Gilovich, 1999; Kruger & Savit-
sky, 2004; Ross & Sicoly, 1979; Thompson & Kelly,
1981). Our findings thus dovetail nicely with existing re-
search on responsibility allocations.
Our findings also extend research on support theory.
In particular, the present results show that assessments
of responsibility and group productivity, like probability
judgments (Rottenstreich & Tversky, 1997; Tversky &
Koehler, 1994), assessments of frequency (Dawes &
Mulford, 1993; Mulford & Dawes, 1999), affective fore-
casts (Van Boven & Epley, 2003), and judgments of
numerosity (Pelham, Sumarta, & Myaskovsky, 1994),
are subject to unpacking effects (for additional examples
of unpacking effects in other domains, see Fiedler &
Armbruster, 1994; Kruger & Evans, 2004; and van der
Plight, Eiser, & Spears, 1987). Thus, the findings we re-
port suggest that support theory describes the way in
which people make a broad range of intuitive judg-
ments. In this regard, our findings were foreshadowed
by Tversky & Koehler (1994), who noted that ‘‘although
unpacking plays an important role in probability judg-
ment, the cognitive mechanism underlying this effect is
considerably more general,’’ reflecting ‘‘a general char-
acteristic of human judgment’’ (p. 562).
In addition to these theoretical contributions, our
findings also have applied value. PeopleÕs tendency to
overestimate the magnitude and importance of their
own contributions to collaborative endeavors can have
serious implications for group functioning and perfor-
mance (Gilovich, Kruger, & Savitsky, 1999). When the
relatively meager public accolades an individual receives
for his or her contributions to a project fail to match his
or her inflated private assessments of what he or she
‘‘deserves,’’ the individual can feel underappreciated
and even taken advantage of—and may be likely to
underperform on future projects as a result. Moreover,
if the individual makes his or her private assessments
public, others may be likely to interpret these inflated
self-assessments as a calculated and unscrupulous grab
for credit, rather than as a logical consequence of the
heightened accessibility of his or her own contributions
(Kruger & Gilovich, 1999). Here too, group functioning
may suffer, as the individualÕs collaborators may harbor
negative assessments of him or her, leading to animosity
and interpersonal tension. This possibility is illustrated
in the results of a study in which participants engaged
in a group exercise and received feedback indicating that
some of their collaborators appeared to overestimate the
magnitude of their own contributions. Predictably, par-
ticipants rated these individuals as harder to get along
with, less likeable, and less desirable as future work part-
ners than they rated those with more modest and realis-
tic self-assessments (Forsyth, Berger, & Mitchell, 1981;
see also Burrus, Kruger, & Savitsky, 2004).
In some cases, individualsÕtendency to overestimate
their relative contributions to collaborative endeavors
Table 3
Claimed percentage responsibility for four activities by participants in
the control and unpacked conditions, multiplied by group size and
adjusted for participantsÕinstitution, Study 4
Activity Condition
Control Unpacked
Writing 143.0 106.9
Idea generation 193.5 159.0
Administrative work 155.0 117.6
Overall work 194.1 146.7
K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457 455
can even lead to the dissolution of a group (Leary &
Forsyth, 1987). We suspect, for example, that many in-
stances of so-called ‘‘creative differences’’ cited to ex-
plain the breakups of collaborations in the music and
entertainment industries may amount to little more
than divergent allocations of responsibility for the
products of the collaboration—and that a tendency
to regard oneÕs collaborators collectively can play a
role in this tendency. Indeed, we cannot help but see
evidence of a failure to unpack a group of collabora-
tors in a recent remark by Yoko Ono. When the music
group The Beatles was honored at the 2004 Grammy
Awards for the 40th anniversary of their original
appearance on the Ed Sullivan Show, Ono noted that
‘‘if John were here, he would have been very happy
that his efforts with the other three were acknowledged
in this way.’’ One wonders if this failure to consider
each collaborator separately is related to the notorious
tension that has existed between Ono and ‘‘the other
three.’’
Given that peopleÕs tendency to overestimate their
contribution to collaborative tasks can lead to such neg-
ative consequences, it is useful to remind readers that
simply asking our participants to consider their collabo-
rators separately, rather than collectively, significantly
reduced self-allocations of responsibility in our studies.
Unpacking the members of the groups one works in
may thus be a useful prescription for avoiding conflict.
The same may be true, moreover, for the groups one
supervises or evaluates. In other research, we have
shown that asking participants to unpack the members
of a group of which they themselves were not a mem-
ber—specifically, a five-person discussion of the 2000
U.S. Presidential election from the political talk show
‘‘The McLaughlin Group’’—resulted in increased
assessments of the amount and quality of the groupÕs
output. Seeing the group as comprised of individuals,
rather than as a holistic entity, caused participants to
see the group as more productive than they would have
otherwise (Savitsky, Wight, Van Boven, & Epley, 2004).
Unfortunately, because an individual can engage in
unpacking only when he or she has multiple collabora-
tors or evaluates the productivity of a group of peo-
ple—as opposed to those times in which an individual
works with or evaluates only one person—the applica-
tion of our research may appear to be confined to only
multi-person groups. Then again, inasmuch as it is pos-
sible to unpack oneÕs collaborators, it is also possible to
unpack their contributions (cf. Van Boven & Epley,
2003). And once again, a failure to do so may lead indi-
viduals to overestimate their own contributions and the
various deleterious consequences that follow. A hus-
band and wife may each believe they have performed a
majority of the housework, for example, because the
husband naturally unpacks his own contribution (‘‘I
wash the dishes, sweep the floor, wipe the countertops,
and store the leftovers’’), but fails to unpack his wifeÕs
(‘‘All she does is the laundry!’’)—at the same time as
the wife commits the complementary error (‘‘I sort,
wash, dry, fold, and iron all of the clothes; all he does
is clean the kitchen!’’). Such a predicament appears to
render tension virtually inevitable. Encouraging individ-
uals who collaborate, whether in pairs or in larger
groups, to unpack not only their groups, but also the
contributions of each of their collaborators, may help
alleviate some of this tension.
References
Brawley, L. R. (1984). Unintentional egocentric biases in attributions.
Journal of Sport Psychology, 6, 264–278.
Brenner, L. A., & Koehler, D. (1999). Subjective probability of
disjunctive hypotheses: Local-weight models for decomposition of
evidential support. Cognitive Psychology, 38, 16–47.
Brenner, L. A., Koehler, D. J., & Rottenstreich, Y. (2002). Remarks on
support theory: Recent advances and future directions. In T.
Gilovich, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases:
The psychology of intuitive judgment. Cambridge, England: Cam-
bridge University Press.
Burger, J. M., & Rodman, J. L. (1983). Attributions of responsibility
for group tasks: The egocentric bias and the actor–observer
difference. Journal of Personality and Social Psychology, 45,
1232–1242.
Burrus, J., Kruger, J., & Savitsky, K. (2004). The interpersonal
consequences of stealing the glory (submitted).
Chapman, G. B., & Johnson, E. J. (2002). Incorporating the irrelevant:
Anchors in judgments of belief and value. In T. Gilovich, D.
Griffin, & D. Kahneman (Eds.), Heuristics and biases: The
psychology of intuitive judgment (pp. 120–138). Cambridge,
England: Cambridge University Press.
Christensen, A., Sullaway, M., & King, C. E. (1983). Systematic error
in behavioral reports of dyadic interaction: Egocentric bias and
content effects. Behavioral Assessment, 5, 129–140.
Dawes, R. M., & Mulford, M. (1993). Diagnoses of alien kidnappings
that result from conjunction effects in memory. Skeptical Inquirer,
18, 50–51.
Deutsch, F. M., Lozy, J. L., & Saxon, S. (1993). Taking credit:
CouplesÕreports of contributions to child care. Journal of Family
Issues, 14, 421–437.
Epley, N. (2004). A tale of tuned decks? Anchoring as adjustment and
anchoring as activation. In D. J. Koehler & N. Harvey (Eds.), The
Blackwell Handbook of Judgment and Decision Making (pp. 240–
256). Oxford, U.K.: Blackwell Publishers.
Fiedler, K., & Armbruster, R. (1994). Two halves may be more than
one whole. Journal of Personality and Social Psychology, 66,
633–645.
Fischhoff, B., Slovic, P., & Lichtenstein, S. (1978). Fault trees:
Sensitivity of estimates failure probabilities to problem represen-
tation. Journal of Experimental Psychology: Human Perception and
Performance, 4, 330–344.
Forsyth, D. R., Berger, R. E., & Mitchell, T. (1981). The effects of self-
serving vs. other-serving claims of responsibility on attraction and
attribution in groups. Social Psychology Quarterly, 44, 56–64.
Gates, D. (2000, August 14). An editorÕs love. Newsweek,p.5.
Gilovich, T., Kruger, J., & Savitsky, K. (1999). Everyday egocentrism
and everyday interpersonal problems. In R. M. Kowalski & M. R.
Leary (Eds.), The social psychology of emotional and behavioral
problems: Interfaces of social and clinical psychology (pp. 69–95).
Washington, DC: American Psychological Association.
456 K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457
Gilovich, T., Medvec, V. H., & Savitsky, K. (2000). The spotlight effect
in social judgment: An egocentric bias in estimates of the salience
of oneÕs own actions and appearance. Journal of Personality and
Social Psychology, 78, 211–222.
Greenwald, A. G. (1980). The totalitarian ego: Fabrication and
revision of personal history. American Psychologist, 35, 603–618.
James, W. (1892). The self. In Psychology: The briefer course
(pp. 41–49). New York: Henry Holt.
Johnson, E. J., Hershey, J., Meszaros, J., & Kunreuther, H. (1993).
Framing, probability distortions, and insurance decisions. Journal
of Risk and Uncertainty, 7, 35–51.
Kruger, J., & Evans, M. (2004). If you donÕt want to be late,
enumerate: Unpacking reduces the planning fallacy. Journal of
Experimental Social Psychology,40, 586–598.
Kruger, J., & Gilovich, T. (1999). ‘‘Naı
¨ve cynicism’’ in everyday
theories of responsibility assessment: On biased assumptions of
bias. Journal of Personality and Social Psychology, 76, 743–753.
Kruger, J., & Savitsky, K. (2004). On the genesis of inflated (and
deflated) judgments of responsibility: Egocentrism revisited.
(submitted).
Leary, M. R., & Forsyth, D. R. (1987). Attributions of responsibility
for collective endeavors. In C. Hendrick (Ed.), Review of person-
ality and social psychology: Vol. 8 Group processes (pp. 167–188).
Newbury Park, CA: Sage.
Macchi, L., Osherson, D., & Krantz, D. H. (1999). A note on
superadditive probability judgment. Psychological Review, 106,
210–214.
Miller, R. S., Goldman, H. J., & Schlenker, B. R. (1985). Egotism in
group members: Public and private attributions of responsibility
for group performance. Social Psychology Quarterly, 48, 85–89.
Mulford, M., & Dawes, R. M. (1999). Subadditivity in memory for
personal events. Psychological Science, 10, 47–51.
Neisser, U. (1981). John DeanÕs memory: A case study. Cognition, 9,
1–22.
Pelham, B. W., Sumarta, T. T., & Myaskovsky, L. (1994). The easy
path from many to much: The numerosity heuristic. Cognitive
Psychology, 26, 103–133.
Redelmeier, D. A., Koehler, D. J., Liberman, V., & Tversky, A. (1995).
Probability judgments in medicine: Discounting unspecified possi-
bilities. Medical Decision Making, 15, 227–230.
Ross, M. (1981). Self-centered biases in attributional of responsibility:
Antecedents and consequences. In E. T. Higgins, C. P. Herman, &
M. P. Zanna (Eds.), Social cognition: The Ontario symposium
(pp. 305–321). Hillsdale, NJ: Erlbaum.
Ross, M., & Sicoly, F. (1979). Egocentric biases in availability and
attribution. Journal of Personality and Social Psychology, 37,
322–336.
Rottenstreich, Y., & Tversky, A. (1997). Unpacking, repacking, and
anchoring: Advances in support theory. Psychological Review, 104,
406–415.
Russo, J. E., & Kolzow, K. J. (1994). Where is the fault in fault trees?.
Journal of Experimental Psychology: Human Perception and Per-
formance, 20, 17–32.
Savitsky, K., Wight, W., Van Boven, L., & Epley, N. (2004). The
unpacking effect in evaluations of group productivity. Unpublished
raw data.
Schlenker, B. R., & Miller, R. S. (1977). Egocentrism in groups: Self-
serving biases or logical information processing?. Journal of
Personality and Social Psychology, 35, 755–764.
Schwarz, N., Bless, H., Strack, F., Klumpp, G., Rittenauer-Schatka,
H., & Simons, A. (1991). Ease of retrieval as information: Another
look at the availability heuristic. Journal of Personality and Social
Psychology, 61, 195–202.
Thompson, S. C., & Kelly, H. H. (1981). Judgments of responsibility
for activities in close relationships. Journal of Personality and Social
Psychology, 41, 469–477.
Tversky, A., & Fox, C. R. (1995). Weighing risk and uncertainty.
Psychological Review, 102, 269–283.
Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for
judging frequency and probability. Cognitive Psychology, 5,
207–232.
Tversky, A., & Koehler, D. J. (1994). Support theory: A non-
extensional representation of subjective probability. Psychological
Review, 101, 547–567.
Van Boven, L., & Epley, N. (2003). The unpacking effect in evaluative
judgments: When the whole is less than the sum of its parts. Journal
of Experimental Social Psychology, 39, 263–269.
van der Plight, J., Eiser, J. R., & Spears, R. (1987). Comparative
judgments and preferences: The influence of the number of response
alternatives. British Journal of Social Psychology, 26, 269–280.
K. Savitsky et al. / Journal of Experimental Social Psychology 41 (2005) 447–457 457
... One way to change time-interval perception is to frame it in an unpacking manner, which has been called the time unpacking effect (Kruger and Evans, 2004;Liu and Sun, 2016). To put it in a colloquial language, the unpacking effect means that the whole is less than the sum of its parts (Van Boven and Epley, 2003). It has been broadly investigated under the framework of Support Theory (Tversky and Koehler, 1994). ...
... The perceived probability of an event (e.g., death from an unnatural death) increases when the event is descriptively unpacked by giving more examples or details (e.g., death from car accidents, homicide, suicide, and fires). Similar effects have been detected for other quantitative judgments, such as the severity of an event's consequence (Van Boven and Epley, 2003). ...
... The findings of some other studies (Van Boven and Epley, 2003;Tsai and Zhao, 2011) also hint at the existence of the time unpacking effect. In daily life, people often use distinctive, memorable events to mark a point in time or segment a period of time (e.g., "the day when we first met," "from the day I graduated till the day I get a tenure-track job"). ...
Article
Full-text available
People often feel that a period of time becomes longer when it is described in more detail or cut into more segments, which is known as the time unpacking effect. The current study aims to unveil how time unpacking manipulation impacts intertemporal decision making and whether the gain-loss valence of choices moderates such impacts. We recruited 87 college students (54 female) and randomly assigned them to the experimental conditions to complete a series of intertemporal choice tasks. The subjective values of the delayed choices were calculated for each participant and then analyzed. The results showed that participants perceived longer time delays and higher subjective values on the delayed gains (but not losses) in the time unpacking conditions than in the time packing conditions. These results suggest that time unpacking manipulation not only impacts time perception but also other factors, which in turn, influence the valuation of delayed outcomes and thereby intertemporal choices. The results are discussed in comparison to previous studies to highlight the complexity of the mechanism underlying the effect of time unpacking on intertemporal decision making.
... Namely, the method that we used to segregate the volunteer event involved breaking the event into smaller pieces across multiple days. In contrast, studies using a Support Theory framework have unpacked an event by adding additional details into the description of the event (e.g., "an accidental plane crash caused by human error" rather than just "a plane crash"; see Rottenstreich & Tversky, 1997;Van Boven & Epley, 2003). associated with the activity. ...
Article
Full-text available
Does the way individuals think about the burdens and benefits of giving time to charity depend on how that time is distributed? Prior work shows that the assessment of a whole is different than the assessment of the sum of its parts. Drawing on this work, we demonstrate that when a volunteer experience is segregated into multiple parts (vs. aggregated into a whole)-for example, volunteering four hours a day for two days versus eight hours in a single day-individuals expect the process to feel less effortful but also the outcome to feel less morally right. In addition, we find that contribution size has a moderating effect on anticipated effort (but not on anticipated morality). Lastly, we find that individuals who focus on minimizing the effort involved in volunteering (vs. maximizing the impact of volunteering) show a greater likelihood of volunteering in a segregated (vs. aggregated) manner. We discuss theoretical and practical implications of these findings. Expectations about Volunteer Schedules 3
... For instance, people are willing to pay more for flight insurance protecting them against losses caused by "terrorism" than for flight insurance protecting them against all losses, regardless of their cause, because terrorist attacks are easier to imagine than "any cause of loss" (Johnson, Hershey, Meszaros, & Kunreuther, 1993). The unpacking effect, which occurs in likelihood estimation, also has its equivalent in expected subjective utility, with instance-based descriptions of events leading to more extreme judgments than general event labels (Van Boven & Epley, 2003). In one study, for instance, people read about an oil refinery that released toxins in the atmosphere and thereby caused an increase in health problems in the surrounding community. ...
Article
Full-text available
Risk perception or people's judgment of future outcomes that may occur if they or other people follow a given course of action may come about either through an analytical or through an experiential process. The analytical mode involves deliberatively and systematically taking into the various dimensions of the risk, such as the likelihood of events, the utility of these events, and the number of individuals affected. The experiential mode involves an affective and intuitive reaction to the risk at stake. Both modes are useful in some circumstances, but both may also give rise to predictable departures from the objective magnitude of the risk.
... Pertinent to the current investigation are the phenomena of 'unpacking' and subadditivity. When people provide more detail about the constituent elements within a category it affects the way they think about that category as a whole (Rottenstreich & Tversky, 1997;Van Boven & Epley, 2003). For example, participants estimate that the likelihood of dying from unnatural causes is greater when this category is unpacked (as accidents, homicides, or other causes) relative to packed (as unnatural causes; Tversky & Koehler, 1994). ...
Preprint
Full-text available
Amid a sea of requests for aid, what factors affect decisions to donate? Researchers and practitioners continue to debate whether we ought to prioritize leveraging passion (e.g., appeal to people’s empathy) or reason (e.g., appeal to utilitarian drives). Here we circumvent the challenges of emotions and obligation as levers by investigating the efficacy of a novel manipulation of choice architecture to affect giving: increasing the number of action options available to donors to increase how much they give. Across six experiments (N = 6,321), participants responded to single or multiple-bid donation requests. Viewing multiple bids for aid increased both intention to donate and actual donation (controlling for affective states that have been previously associated with costly helping). Notably, increasing the number of bids did not decrease the proportion of people who donated at all. We rule out previously-documented heuristics (i.e., 1/n, fairness) and agency as explanations for our effect, and find positive evidence for the role of merely increasing action options (i.e., bids). Finally, we replicate our effect in a natural experiment of charitable donations (N = 10,000): presenting donors with multiple bids increased the average donation by $8.77 (a 19.7% increase). Our findings have theoretical implications for invigorating behavior and offer practical suggestions for how charitable organizations can better engineer solicitations for aid.
Article
Previous research has shown that the judged probability of an event depends on whether its description mentions examples (“What is the probability that a randomly chosen Italian businessman will travel during the next month to Warsaw, Budapest, Prague or some other European city?”) or does not mention examples (“What is the probability that a randomly chosen Italian businessman will travel during the next month to a European city?”). Here, we examined descriptions that mention examples and manipulated whether these are relatively similar (e.g., Warsaw, Budapest, Prague) or diverse (e.g., Warsaw, Marseilles, Helsinki). Four experiments (N = 1112) revealed a diversity effect: Overall, descriptions with diverse examples received higher probability judgments than descriptions with similar examples. We discuss several possible mechanisms for this effect, such as that descriptions with diverse examples prompt fuller representations of the target category or that the effect is driven by a representativeness or proximity heuristic.
Article
In this paper, we ask: how should an agent who has incoherent credences update when they learn new evidence? The standard Bayesian answer for coherent agents is that they should conditionalize; however, this updating rule is not defined for incoherent starting credences. We show how one of the main arguments for conditionalization, the Dutch strategy argument, can be extended to devise a target property for updating plans that can apply to them regardless of whether the agent starts out with coherent or incoherent credences. The main idea behind this extension is that the agent should avoid updating plans that increase the possible sure loss from Dutch strategies. This happens to be equivalent to avoiding updating plans that increase incoherence according to a distance‐based incoherence measure.
Article
Full-text available
Does the way individuals think about the burdens and benefits of giving time to charity depend on how that time is distributed? Prior work shows that the assessment of a whole is different than the assessment of the sum of its parts. Drawing on this work, we demonstrate that when a volunteer experience is segregated into multiple parts (vs. aggregated into a whole)—for example, volunteering 4 h a day for 2 days versus 8 h in a single day—individuals expect the process to feel less effortful but also the outcome to feel less morally right. In addition, we find that contribution size has a moderating effect on anticipated effort (but not on anticipated morality). Lastly, we find that individuals who focus on minimizing the effort involved in volunteering (vs. maximizing the impact of volunteering) show a greater likelihood of volunteering in a segregated (vs. aggregated) manner. We discuss theoretical and practical implications of these findings.
Article
We investigate whether the format of internal control weakness (ICW) disclosures required by the Sarbanes-Oxley Act of 2002 influences perceptions of nonprofessional investors. Using a 2 x 2 between-participants experiment, we examine two facets of ICW disclosure formats: ICW presentation salience (high versus low) and ICW disaggregation type (disaggregated versus aggregated). We hypothesize and find evidence of an interactive effect between presentation salience and disaggregation type, such that investors perceive ICWs as less negative when they are saliently disclosed and this effect is enhanced when the material weakness is disaggregated into its individual control deficiencies. We also find evidence of moderated mediation such that high salience of ICW disclosures has a positive indirect effect on investing judgments through management trust, but only when the ICW is disaggregated into its individual control deficiencies.
Article
Full-text available
We report results of a survey experiment aimed at testing whether eliciting taxpayer preferences on how to allocate the collected taxes over national public goods as well as providing information about the composition of the public expenditure influence the tax rate that taxpayers consider adequate to pay. We find that information exerts no effects on the level of the adequate tax rate. However, taxpayers are willing to accept a higher tax burden when they express their preferences on how to use tax revenues to finance public goods and services.
Article
Full-text available
A series of studies examines whether certain biases in probability assessments and perceptions of loss, previously found in experimental studies, affect consumers' decisions about insurance. Framing manipulations lead the consumers studied here to make hypothetical insurance-purchase choices that violate basic laws of probability and value. Subjects exhibit distortions in their perception of risk and framing effects in evaluating premiums and benefits. Illustrations from insurance markets suggest that the same effects occur when consumers make actual insurance purchases.
Article
Prior research has found that people tend to overestimate their relative contributions to joint tasks (e.g., Ross & Sicoly, 1979). In the present research we investigate one of the causes of this bias, and in doing so, identify an important moderator of the effect. In three studies we demonstrate that when people estimate their relative contributions to collective endeavors they focus on their own contributions and give less consideration to the contributions of their collaborators. This can cause overestimation of relative contributions when absolute contributions are numerous, but underestimation of relative contributions when absolute contributions are few. These results extend Ross & Sicoly's (1979) original egocentrism analysis of bias in responsibility attributions, but also suggest that the tendency to overestimate one's relative contributions to collaborations is not as ubiquitous as once thought.
Article
The perceived self-relevance of a health issue determined whether participants relied on recalled content or experienced ease of recall in assessing risk. Participants recalled either three or eight behaviors that increase or decrease risk of heart disease. Although recalling three risk factors was relatively easy, people had difficulty recalling eight risk factors. When heart disease was not considered self-relevant, participants used a heuristic judgment strategy and relied on their ease of recall. They reported greater vulnerability after having recalled three rather than eight risk-increasing behaviors and lower vulnerability after having recalled three rather than eight risk-decreasing behaviors. When heart disease was considered self-relevant, people used a systematic processing strategy and relied on the content of the information recalled They reported greater (lower) vulnerability after having recalled eight rather than three risk-increasing (decreasing) behaviors. Theoretical implications concerning the inter-play Of recalled content and ease of recall in judgment and applied implications for risk perception are discussed.
Article
Reactions to other's claims of resposibility were investigated by assessing group members' evaluations of a fellow group member who took high, moderate, or low personal responsibility for a positive or negative outcome. As predicted, individuals whose attribution were self-serving (blaming others for failure or claiming credit for success) were liked less than (1) group members who allocated responsibility equally, and (2) members whose "other-serving" attributions indicated they took the blame for failure or credited others for success. These results suggest that attributions-when exchanged among group members-significantly influence social perceptions and group relations.
Article
The attributional egotism of individuals may be particularly important when they serve as members of cooperative groups. Within a group one's fellow members may be granted or denied credit for a group performance in order to manipulate one's own perceived responsibility for the outcome. In this study, group members privately or publicly reported their assessments of their own and others' responsibility for group successes and failures. Subjects privately claimed more responsibility for success than for failure but did not do so (in public) when the other members were expected to see their reports. Moreover, under public conditions, subjects claimed less responsibility for a group success than they gave to the other members, an effect which disappeared in private. Subjects were clearly sensitive to the interpersonal implications of their attributions, displaying less egotism under public conditions.
Article
Following announcements of plans for a new nuclear power station in south-west England, local residents were sent questionnaires in which they were asked to give estimates, in percentage terms, of the present, future (2000 AD) and ideal contributions to the UK electricity supply of different energy technologies. In Expt 1, 184 subjects rated five technologies (plus an ‘other’ category), and 10 months later they were split into five groups, each of which had to rate only one of the technologies previously presented. In Expt 2, three versions of the questionnaire were used, requiring ratings of (a) nuclear only, (b) nuclear, coal, plus ‘other’ and (c) eight technologies (including nuclear and coal) plus ‘other’. These were first completed by 220 subjects; six months later a subsample of 134 were each sent one of the two versions other than the one they had previously completed. In Expt 3, 164 subjects gave ratings, in four conditions, of (a) nuclear only, (b) coal only, (c) four technologies plus ‘other’ and (d) seven technologies plus ‘other’; the order of ratings was also changed with preferences first, present estimates second, and future predictions third. Broadly, all three experiments showed that the more technologies subjects had to rate, the smaller were the percentage ratings they gave to any single technology. These findings are considered in relation to context effects in judgement, and to the availability heuristic. Implications for interpretation of opinion poll findings are discussed.
Article
Decision theory distinguishes between risky prospects, where the probabilities associated with the possible outcomes are assumed to be known, and uncertain prospects, where these probabilities are not assumed to be known. Studies of choice between risky prospects have suggested a nonlinear transformation of the probability scale that overweights low probabilities and underweights moderate and high probabilities. The present article extends this notion from risk to uncertainty by invoking the principle of bounded subadditivity: An event has greater impact when it turns impossibility into possibility, or possibility into certainty, than when it merely makes a possibility more or less likely. A series of studies provides support for this principle in decision under both risk and uncertainty and shows that people are less sensitive to uncertainty than to risk. Finally, the article discusses the relationship between probability judgments and decision weights and distinguishes relative sensitivity from ambiguity aversion. (PsycINFO Database Record (c) 2012 APA, all rights reserved)