Content uploaded by Peter M. Gollwitzer
Author content
All content in this area was uploaded by Peter M. Gollwitzer
Content may be subject to copyright.
Journal of Personality and Social Psychology
1997,
Vol. 73. No. 1, 186-199Copyright 1997 by die American Psychological Association, Int.
0022-3514/97/53.00
Implementation Intentions and Effective Goal Pursuit
Peter M. Gollwitzer
University of KonstanzVeronika Brandstatter
University of Munich
The theoretical distinction between goal intentions ("I intend to achieve -c") and implementation
intentions ("I intend to perform goal-directed behavior y when I encounter situation z"; P. M.
Gollwitzer, 1993) is explored by assessing the completion rate of various goal projects. In correla-
tional Study 1, difficult goal intentions were completed about 3 times more often when participants
had furnished them with implementation intentions. In experimental Study 2, all participants were
assigned the same difficult goal intention, and half were instructed to form implementation intentions.
The beneficial effects of implementation intentions paralleled diose of Study 1. In experimental
Study 3, implementation intentions were observed to facilitate the immediate initiation of goal-
directed action when the intended opportunity was encountered. Implementation intentions are inter-
preted to be powerful self-regulatory tools for overcoming the typical obstacles associated with the
initiation of goal-directed actions.
Whether people meet their goals depends on both how goal
content is framed and how people regulate the respective goal-
directed activities (Gollwitzer & Moskowitz, 1996). Content
theories focus on the thematic properties of set goals and how
these affect the regulation of goal pursuit and actual goal
achievement. Such theories attempt to explain differences in
goal-directed behaviors in terms of what is specified as the goal
by the individual, as the content characteristics of the goal are
expected to affect a person's successful goal pursuit. Goal con-
tent has been considered both in terms of the different needs on
which it is based (e.g., autonomy needs vs. materialistic needs;
Deci & Ryan, 1991; Kasser & Ryan, 1994) as well as in terms
of implicit theories (e.g., entity theories vs, incremental theories
of ability; Dweck, 1991, 1996). Numerous other relevant as-
pects of goal content have been suggested, such as specific-
abstract (Emmons, 1992; Locke & Latham, 1990), proximal-
distal (Bandura & Schunk, 1981), and positive versus negative
outcome focus (Higgins, Roney, Crowe, & Hymes, 1994).
Self-regulation theories of goal striving, on the other hand,
focus on the question of how people overcome certain imple-
mentational problems. Having set a goal is considered to be just
a first step toward goal attainment, one that is followed by
a host of implementational problems that need to be solved
successfully. These problems are manifold as they pertain to
initiating goal-directed actions and bringing them to a successful
ending. Various theoretical notions have addressed these issues
in particular, delineated useful self-regulatory strategies, and
We thank Ulf Klebi for his assistance in collecting and analyzing the
data. Thanks are also due to Dieter Frey, Gordon B. Moskowitz, and
Gabriele Oettingen for discussing with us the ideas and findings pre-
sented in this article. ,
Correspondence concerning this article should be addressed to Peter
M. Gollwitzer, Fachgruppe Psychologie, University of Konstanz, Post-
fach 55 60, 78434 Konstanz, Germany, or to Veronika Brandstatter,
Ludwig-Maximilians-Universitat, Institut ftir Psychologie, University of
Munich, Leopoldstr, 13, 80802 MUnchen, Germany. Electronic mail may
be sent via the Internet to gollwitz@soz.psychologie.uni-konstanz.de.
addressed questions of why and how these strategies are effec-
tive.
Typical self-regulatory problems of goal pursuit are, for
instance, warding off distractions (see implemental mindsets,
Gollwitzer, 1990; various action control strategies, Kuhl, 1984;
Kuhl & Beckmann, 1994), flexibly stepping up efforts in the
face of difficulties (see effort mobilization, Wright & Brehm,
1989),
compensating for failures and shortcomings (see
self-
regulation of motivation, Bandura, 1991; discrepancy reduction,
Carver & Scheier, 1981; symbolic self-completion, Wicklund &
Gollwitzer, 1982), and negotiating conflicts between goals (see
intelligent pursuit of life tasks, Cantor & Fleeson, 1994; conflict
resolution in the face of contradictory personal strivings, Em-
mons & King, 1988).
The present article focuses on a further self-regulatory prob-
lem of goal pursuit—the initiation of goal-directed actions.
On the basis of the model of action phases (Gollwitzer, 1990;
Heckhausen, 1989, 1991), we construe people's goal pursuits
as extending from the awakening of wishes and desires to the
evaluative thoughts people have once goal striving has led to
some kind of outcome. The course of wish fulfillment consists
of four action phases (in chronological order: predecisional,
preactional, actional, and postactional), whereby each phase is
associated with a typical task (i.e., setting preferences between
concurring wishes and desires, promoting the initiation of goal-
directed actions, bringing goal-directed actions to a successful
ending, and evaluating what has been achieved as compared
to what was desired, respectively). These action phases are
connected by three crucial transition points. The first relates to
turning the selected wish or desire into a chosen goal and occurs
between the predecisional and the preactional phases. The model
speaks of this transition in terms of forming goal intentions.
Such intentions take the form of ltI intend to achieve -t." The
x specifies a desired end state, which may be defined rather
abstractly (i.e., getting to know a certain person) or concretely
(i.e.,
invite that person to a dinner). The consequence of having
formed a goal intention is a feeling of commitment to achieve
this end state.
However, having formed goal intentions is just one prerequi-
186
IMPLEMENTATION INTENTIONS187
site for making progress toward goal pursuit. The next transition
point to be crossed is located between the preactional and ac-
tional phases and implies the initiation of goal-directed actions.
Getting started with one's goal pursuit is quite simple when the
necessary behaviors are well practiced or routine. More often
than not, however, this is not the case, and goal pursuit comes
to an unnecessary halt or delay. There may be several reasons
for this. First, there may be conflict between various ways of
acting on the goal intention (e.g., one cannot make up one's
mind on how to pursue an interpersonal goal such as getting to
know someone). Second, a given situational context conducive
to one's goal pursuit may be habitually used in the service of
antagonistic behaviors (e.g., when meeting the desired person
at work leads to a competitive response). Moreover, good oppor-
tunities to act on one's goal may sometimes escape our attention
(e.g., when we are absorbed in some ongoing involving activity,
wrapped up in demanding ruminations or intense emotional
experiences), or we may fail to seize them because we did not
respond in time (e.g., when a good opportunity to get to know
somebody presents itself only for a short moment).
Gollwitzer (1993) suggested that forming a certain type of
intention, called implementation intention, is a powerful
self-
regulatory strategy that alleviates such problems and thus pro-
motes the initiation of goal-directed behaviors. Such intentions
take the format of "I intend to do
_y
when situation z is encoun-
tered." In an implementation intention, an anticipated future
situation (opportunity) is linked to a certain goal-directed be-
havior. Holding implementation intentions commits the individ-
ual to perform certain goal-directed behaviors when the critical
situation is actually encountered. Implemention intentions are
formed in the service of goal intentions. Staying within the
example of the goal intention listed above, a possible implemen-
tation intention would link a suitable situation with an appro-
priate behavior, such as "First thing I will do when I run into
him at the office is invite him to my party." The issue is not
only one of translating the desired goal state from a higher level
of abstractness to a lower level (Vallacher & Wegner, 1987), but
also of linking a certain goal-directed behavior to an anticipated
appropriate situational context.
The result of an implementation intention is a commitment to
perform the specified goal-directed behaviors when the critical
situation is actually encountered. In this sense, the control of
one's goal striving is no longer based solely on one's commit-
ment to the goal state (i.e., the goal intention) but is also dele-
gated to the critical situation (Gollwitzer, 1993). The critical
situation is thought to instigate goal-directed behavior in a way
that is similar to the situational elicitation of habitual behavior.
The difference is that habits create strong links between situa-
tions and actions by consistent and frequent performing of the
same behaviors in the same situations (Guthrie, 1959). In the
case of implementation intentions, strong links are thought to
originate from a single mental act. Accordingly, by forming
implementation intentions, people supplement the control of
goal-directed actions that is based on their goal commitments
with environmental control that stems from the specified situa-
tional contexls.
On the basis of these ideas, we formulated two different
hypotheses. The first hypothesis extends to the effects of imple-
mentation intentions on the rate of goal completion. We postu-
lated that goal intentions that are furnished with implementation
intentions will show a higher rate of completion than will bare
goal intentions. To test this hypothesis, one would have to ask
people about their goal intentions and then determine whether
people have also formed respective implementation intentions.
An alternative way would be to experimentally induce goal
intentions and then vary whether implementation intentions are
subsequently formed. In both cases, one would then have to
wait and give people enough time to act on their goals.
We conducted two field studies that were patterned along this
line of thought. The first study took a correlational approach.
University students were contacted prior to Christmas and were
asked to list personal projects (i.e., goal intentions) they in-
tended to complete during Christmas break. The second study
took an experimental approach. Prior to Christmas, participants
were assigned a project to be conducted during the Christmas
holidays. In Study 1 implementation intentions were measured;
in Study 2 they were manipulated. In both studies we assessed
the rate of goal completion when the period of time implied by
their goal intentions had passed.
Our second hypothesis focused on the idea that implementa-
tion intentions delegate the control of goal-directed behaviors
to the situational contexts specified. This assumption implies
that people should readily initiate goal-directed behaviors once
the critical situation is encountered—very much like habitual
behavior is immediately performed when the respective situa-
tional stimuli are presented (Bargh, 1997; Bargh & Gollwitzer,
1994;
Guthrie, 1959). Because our two field studies did not
allow us to reliably check on whether participants responded
immediately to the situations specified in their implementation
intentions (as participants acted on their goals at home), we
conducted a third study. In this laboratory study, we presented
participants a variety of opportunities to act on an assigned goal
and observed whether opportunities that had been specified in
implementation intentions were seized immediately.
Study 1: Getting Things Done Over Christmas Break
From December 8 to December 18, we asked university stu-
dents to list two personal projects that they intended to achieve
during Christmas break (i.e., before mid-January). As Christ-
mas time is characterized by many concurring goals or external
distractions and restrictions (e.g., social obligations), we
thought that many of the problems of initiating goal-directed
actions described above should be particularly pressing. Accord-
ingly, this offered a good test of our notion that implementation
intentions promote a person's goal pursuit. We also wanted to
explore whether the effects of implementation intentions interact
in some meaningful way with special qualities of the respective
goal intentions. For this purpose, we measured numerous fea-
tures of goal intentions (e.g., importance of the goal) and as-
sessed the perceived difficulty of goal completion. We expected
that implementation intentions should promote goal completion
for both easy-to-implement and difficult-to-implement goals. If
anything, this effect should be somewhat weaker with difficult
goals,
because factors related to issues other than successfully
getting started should play a more prominent role with difficult
goals.
In other words, it should be simple to follow through on
an easy-to-implement goal intention once goal-directed actions
188GOLLWITZER
AND
BRANDSTATTER
have been initiated successfully. With goal intentions that are
difficult to implement, however, successfully getting started still
means that there is a long and cumbersome way to go, one that
requires many steps and sustained intentional effort.
We assessed implementation intentions by asking participants
whether they had committed themselves to getting started with
certain goal-directed behaviors at specific points in time. Four
weeks after Christmas break, we probed in a follow-up question-
naire whether participants had completed their projects.
Method
Participants
One hundred eleven students with different academic majors from
the
University
of
Munich were recruited
for
this study.
The
study
was de-
scribed
as an
attempt
to
discover college students' goal pursuits. Partici-
pants were paid
10
Deutsche Marks
(DM) ($6) for
participation.
To
secure comparability
of the
personal projects being listed,
we
invited
only female students
to
participate.
The
average
age of
the sample
was
22.6 years; ages ranged from
19 to 33
years.
Procedure
and
Material
Participants arrived
at the
laboratory
in
groups
of 6 to 8.
They were
seated
at
tables separated
by
partitions, which were arranged such that
participants could easily
see the
female experimenter
but
none
of the
other participants.
It was
explained
to
them that they would have
to
answer
a
questionnaire that inquired about their personal goals and about
many aspects related
to the
formation
and
execution
of
goals.
On the
first page
of
the questionnaire, participants were given
a
detailed descrip-
tion
of
what
was
meant
by a
personal project
or
goal (i.e.,
a
goal
intention).
It
was explained that personal goals can differ in many ways.
They may
be
more
or
less difficult
to
implement; require only
a few or
a complex sequence
of
action steps; represent different areas
of a per-
son's life;
and be
more
or
less time consuming, attractive,
or
urgent.
Participants were asked
to
name
one
personal goal that
was
easy
to
implement and one that was difficult. Each goal
had to
meet two prereq-
uisites, however. First, participants were instructed
to
name only such
goals that they held prior
to
arriving
at the
laboratory. Second, partici-
pants were
to
intend
to
implement these goals during Christmas break.
To assure that participants kept
to
this rule,
we
asked them
to
indicate
the exact week when they wanted
to be
done with their goal project.
Assessment
of
various goal qualities. To control
for
certain features
of participants' intended projects that might directly
or
indirectly
(via
the formation
of
implementation intentions) facilitate completing
the
project,
we
asked participants
to
rate several aspects
of the
goals they
had listed
on
10-point scales anchored
by 0
{not a lot,
a
little,
or
never)
and
9
(very,
a
lot,
or
very often). Participants were asked
the
following
questions:
(a)
How long have
you
held this goal?
(in
number
of
days),
(b) How close
do you
feel
to
goal completion? (participants marked
a
15-cm horizontal line with point
of
completion indicated
at
10 cm),
(c)
Does goal completion require
a
single action
or a
sequence
of
actions?
(3-point answer scale ranging from
1 =
single action
to 3 =
several
different actions with
2 = in
between),
(d) Are you
familiar with
the
necessary actions? (yes-no answer),
(e) How
important
is the
goal?
(f) How much social pressure
do you
feel
to
complete your goal?
(g)
How certain
are you to
achieve your goal?
(h) How
much have
you
invested
for
goal completion
so
far?
(i) How
likely
is it
that
you
will
encounter possible obstacles? (four possible obstacles were listed: com-
peting goal intentions, forgetting
the
goal,
low
self-control, external
hindrances; when
the
items were combined
to an
index, Cronbach's
alpha
was .74 and .71 for
easy
and
difficult goals, respectively),
(j)
How many times have
you
tried unsuccessfully
to
complete your goal?
and
(k) How
many times have
you
failed
to
seize
an
opportunity
to
do
so?
Assessment
of
implementation intentions. Finally,
we
assessed
whether participants
had
formed implementation intentions
on
when
and where
to
initiate goal-directed actions. Participants were told
the
following:
In order
to
achieve their goals people execute behaviors that
can
be described along three dimensions:
One
refers
to the
point
in
time,
the
second
to the
place,
and the
third
to the
type
of
action
that
is
linked
to the
situational context specified
by
time
and
place.
Please indicate whether
you
have
by now
committed yourself
to a
certain situational context
for
initiating specific goal-directed
ac-
tions
in the
sense
of
'
'When
I
encounter
the
situational context
y,
I will perform behavior
z!"
Once participants
had
indicated
the
formation
of
implementation
in-
tentions
by a
single
yes or no
answer
for
each
of
the
two
goal projects
listed,
we
checked whether participants
who
answered
yes had
indeed
committed themselves
to
perform certain goal-directed actions
at se-
lected situational contexts
and
whether participants
who
answered
no
had not. For this purpose, participants were asked
to
describe the chosen
actions
and the
respective situational contexts
in
their own words. As
it
turned out, only the participants who reported
to
have formed implemen-
tation intentions were
in a
position
to
follow
our
request.
Assessment of goal completion. When participants were thanked and
paid
for
their participation, they were asked
to
collaborate
in a
follow-
up study.
The
vast majority
of
participants agreed (89%,
i.e., 99 of 111
participants)
and
left their addresses with
the
experimenter.
To
reduce
experimenter demand (i.e., participants feel obligated
to
behave
ac-
cording
to
their implementation intentions
in the
eyes
of the
experi-
menter), we used a coding scheme that allowed us
to
match participants'
first and second questionnaires without having
to
refer
to
their names
and addresses. Most important,
the
experimenter explained
to
partici-
pants this consequence
of
the
coding scheme. When participants received
die second questionnaire right after
the end of
Christmas break, they
were asked
to
indicate for each
of
the two projects
(by a
yes-no answer)
whether they
had
actually completed them. These items were hidden
among
the
questions described above that assessed various features
of
participants* goal pursuits, such
as
importance
or
proximity
to
goal
completion.
Results
Of the 99 participants who agreed to participate in the follow-
up study, 92 sent back the second questionnaire (i.e., 93% of
the participants). When we matched first and second question-
naires, we discovered that various participants had indicated
goal projects they intended to complete much later than the
end of Christmas break (e.g., during summer break). Seven
participants named long-term projects with respect to the easy-
to-implement goal, whereas 22 participants did so for the diffi-
cult-to-implement goal. It was obviously harder for participants
to name a difficult project to be performed during Christmas
break than to name an appropriate easy-to-implement goal.
Hence we had to run our statistical analyses with 85 participants
when considering easy-to-implement goals and 70 participants
for difficult-to-implement goals.
Differences Between Easy
and
Difficult Goals
Goal content. Of the easy-to-implement goal intentions,
8.2% were career related {e.g., to buy a textbook on civil law),
IMPLEMENTATION INTENTIONS189
55.3%
were lifestyle related (e.g.,
to
finish reading
a
novel),
and 36.5% covered interpersonal issues (e.g.,
to
write
a
letter
to
a
friend).
Of the
difficult-to-implement goal intentions,
52.9%
were career related (e.g.,
to
write
a
term paper), 31.4%
were lifestyle related (e.g., to find
a
new apartment), and 15.7%
were
of
an interpersonal nature (e.g., to settle a conflict with my
boyfriend). Participants obviously named projects
of
different
content when asked
to
list easy-to-implement than when asked
to list difficult-to-implement projects; x2(2,
N
= 155)
=
37.94,
p
<
.001. More than half
of
the difficult goals and less than
10%
of
the easy goals centered on professional issues.
Goal qualities.
On the
measured goal qualities, easy
(e)
goals differed from difficult
(d)
goals
on
those variables that
were indicators
of
difficulty
of
implementation. Easy-to-imple-
ment goals were rated by participants as being closer to accom-
plishment than were difficult-to-implement goals (Me
= -2.81
vs.
Md =
-5.18), ?(65)
=
5.28,
p <
.001;
as
requiring less
complex
and
more familiar action sequences (complexity:
Me
= 1.80 vs. Md 2.66; r[63]
=
6.73,
p <
.001; familiarity: Me
=
0.61
vs. MA =
0.25; r[63]
=
4.59,
p <
.001);
as
associated
with
a
higher certainty
of
goal completion (Me
=
8.03 vs.
Md
= 6.80), Z(65)
=
3.39,
p =
.001; and as requiring fewer unsuc-
cessful trials
to get
started (Me
=
1.60 vs. MA
=
2.66), f(64)
= 2.66,
p
— .01. Apparently, participants complied
to the in-
struction
to
name
two
goal intentions
of
different difficulty.
Moreover, easy goals tended
to be
perceived
as
less important
than difficult goals
(Mc =
6.45 vs. Md
-
7.18),
t(65) = 1.89,
p
=
.06.
Formation
of
Implementation Intentions
Easy versus difficult goals. Before analyzing the effects
of
implementation intentions
on
goal achievement,
we
wanted
to
know whether easy goals differed from difficult goals with
re-
spect to the formation
of
implementation intentions. The rate of
forming implementation intentions was almost identical for both
types
of
goals: 61% (i.e., 51 out
of
83) for easy goals and 66%
(i.e.,
45 out
of
68)
for
difficult goals;
x2(h N =
151)
= .36,
p
>
.50.
In
addition,
we
explored whether people who formed
implementation intentions with one
of
the goals did so with the
other goal
as
well. The correlation was unreliable
(r =
.16,
p
> .20), indicating that there was
no
generalized tendency
to
furnish one's goals with implementation intentions.
Relation to goal qualities. We also looked at the correlations
between goal characteristics and the formation
of
implementa-
tion intentions. For easy and difficult goals, we found significant
positive correlations for perceived closeness to goal completion
(re
=
.24 vs.
rd =
.39), amount
of
investment for goal comple-
tion
(re =
.30 vs.
rd =
.32), and importance
of
goal completion
(/•c
=
.43 vs.
rd ~
.46).
A
significant negative correlation was
found with likelihood
of
potential obstacles
(re =
—.37 vs.
rd
= —.34). It appears then that the same variables were associated
with
the
formation
of
implementation intentions
for
easy
and
difficult goals, respectively. We
do not
know
the
exact causal
direction
of
the observed relations, although
it
seems plausible
to assume that having formed implementation intentions makes
people feel closer to goal completion and anticipate fewer obsta-
cles.
Moreover, implementation intentions possibly make people
feel that they have made an investment, resulting in their judging
the goal
as
comparatively more important.
Dependent Variable: Rate
of
Goal Completion
We expected that implementation intentions should promote
goal completion
for
both easy and difficult goals;
if
anything,
this effect should
be
somewhat weaker with difficult goals be-
cause easy goals are presumably more readily achieved through
a single action
in a
specific situation, whereas difficult goals
take many steps and sustained effort. Looking at the completion
rate for difficult goals, we observed that without an implementa-
tion intention, these goals were completed
in
only 22%
of
the
cases.
However, when furnished with
an
implementation inten-
tion, completion rate increased to 62%
of
the projects; x2(
1>
N
=
68) =
9.99,
p =
.002. The completion rate
of
easy goals
without an implementation intention was more than three times
as high (i.e., 78%) than that
of
difficult goals. Furnishing easy
goals with implementation intentions still increased this comple-
tion rate, albeit minimally (i.e., from 78%
to
84%), x2(l,
N
=
83) =
.51,
p >
.48 (see Figure
1).
Inspection
of
the respective correlations between the forma-
tion
of
implementation intentions
and
completion
of
difficult
versus easy goals revealed that the observed difference between
difficult and easy goals
is
reliable
(rd =
.38,
p =
.001 vs.
re -
.08,
p >
.48;
Z =
1.92,
p =
.03). Although there was
a
highly
significant relationship between
the
formation
of
implementa-
tion intentions and goal completion
for
difficult goals, the
re-
spective relationship
for
easy goals
was
significantly weaker.
Apparently,
in the
present study, participants benefited from
forming implementation intentions with respect
to
difficult but
not
to
easy goals.
Potential Third Variables
The pattern
of
results observed implies that implementation
intentions are an effective means
of
promoting the achievement
of difficult goals. Because
of the
correlational nature
of the
100
90
80
70
60
50
40
30
20
10
i\
u
•
•
•
•
•
•
m
EMy-to-wnplsmant
Implementation Intention
•not formed
•formed
I
•
DttlaJMo-i mptwrant
Type of Goal
Figure
1.
Percentage
of
goal completion
for
easy-to-implement
and
difficult-to-implement goals, Study
I.
190GOLLWITZER AND BRANDSTATTER
present study, however, we cannot be certain whether implemen-
tation intentions actually affected goal achievement. All of the
goal qualities that correlated significantly with forming imple-
mentation intentions (see the above Relation to goal qualities
section) qualify as potential promoters of goal completion:
closeness to goal completion, amount of investment for goal
completion, importance of goal completion, and likelihood of
potential obstacles. However, actual promotion of goal comple-
tion through these variables requires that these variables also
show a substantial correlation to goal completion. Three of the
four named variables fulfilled this requirement and correlated
significantly or marginally significantly with successful goal
completion: perceived closeness to project completion, likeli-
hood of obstacles, and goal importance (r = .25, p < .05; r =
-.28,
p < .05; and r = .19, p = .11, respectively).
To secure that the implementation intention effect on comple-
tion of difficult goals was not attributable to any of these three
variables, we conducted two further analyses. First, we calcu-
lated the correlation between forming implementation intentions
and project completion while partialing out closeness to project
completion (pr
—
.32), likelihood of obstacles (pr = .32), and
goal importance (pr
—
.34), respectively. For all correlations,
the coefficients remained significant at the .01 level, suggesting
that the implementation intention effect was not based on these
variables. Second, we computed correlations between closeness
to goal completion, likelihood of
obstacles,
and goal importance,
on the one hand, and actual project completion on the other,
excluding implementation intentions. As expected, the resulting
partial correlations were no longer significant (p = .40, p =
.17,
and p = .84, respectively), suggesting that the relation
of these variables to goal completion was based on forming
implementation intentions.
In summary, then, implementation intentions made a differ-
ence above and beyond the goal qualities we measured, and this
was true even for those qualities that systematically varied with
forming implementation intentions (i.e., initial investment) or
with both forming implementation intentions and goal comple-
tion (i.e., perceived closeness to goal completion, likelihood of
potential obstacles, and importance of goal completion). Ac-
cordingly, it cannot be argued that our assessment of implemen-
tation intentions might be a mere surrogate assessment of these
goal quality variables or that these goal qualities may have
ultimately produced the observed goal completion effects.
Discussion
When people furnish difficult personal projects (i.e.. goal
intentions) with implementation intentions, the rate of goal com-
pletion increases. Committing oneself to achieving a goal can
apparently be enriched with further commitments (i.e., imple-
mentation intentions) that relate to responding to a given situa-
tion with goal-directed behaviors. Two thirds of our participants
chose to do so, and these additional efforts paid off in lhat the
successful completion of difficult-to-implement projects tripled.
In the present study, the benefits of implementation intentions
affected only difficult goals. The fact that implementation inten-
tions facilitated the completion of difficult goals attests to their
enormous impact on goal completion. Even when the course of
implementation of a given goal poses problems, implementation
intentions still manage to be effective. We interpret this to mean
that once goal-directed behaviors are initiated, people have a
better chance of moving successfully toward reaching their goal.
Although we have no evidence that implementation intentions
facilitated the initiation of the intended behaviors in the antici-
pated situations, it seems plausible to assume that goal comple-
tion was mediated by easing action initiation. Once the intended
behavior was executed, goal pursuit started and continued to-
ward completion. This interpretation is supported by research
conducted by Lcwin (1926) and colleagues (e.g., Mahler, 1933;
Ovsiankina, 1928) who demonstrated that, once goal pursuit is
on its way, disruptions caused by difficulties or barriers do not
lead to retreat but to resumption of goal pursuit. It seems plausi-
ble,
then, that the completion of difficult goals does benefit from
implementation intentions, even though the respective commit-
ments relate to getting started with specific goal-directed actions
only.
With easy goals, we were confronted with a different pattern
of results. We did not observe any beneficial effects of imple-
mentation intentions, which were again formed by close to two
thirds of the participants. In our view, there are at least two
possible explanations for this. From a methodological point of
view, this observation may be a result of the high basic comple-
tion rate of the easy goals participants named. The completion
rate was close to perfect (i.e., 4 out of 5 projects were com-
pleted), therefore a further increase through implementation
intentions could not be achieved. One cannot expect a perfect
completion rate, in particular as the present study was conducted
during Christmas break. This is a time full of external con-
straints (e.g., social obligations) that force people to postpone
the completion of their projects.
A more theoretical explanation of the lack of implementation
intention effects with easy goals is also possible, however. If
one considers the different qualities of easy and difficult goal
projects (remember that participants rated easy goal projects as
implying less complex and more familiar action sequences and
as being associated with a higher subjective probability of suc-
cess and with less unsuccessful trials of getting started), it
appears that the implementation of easy-to-implement goals is
more habitualized than the implementation of difficult-to-imple-
ment goals. We argued above that implementation intentions
achieve their effects by habitualizing the initiation of goal-di-
rected actions. If these actions are habitualized to begin with,
however, implementation intentions should not have much of an
additional effect on goal completion. From this perspective, it
follows that difficult-to-implement goals will benefit more from
forming implementation intentions than will easy-to-implement
goals,
as goal implementation can be assumed to be less habitu-
alized with difficult than with easy goals.
Finally, additional analyses on the measured characteristics
of the goals participants named revealed that none of these
variables produced the observed implementation intention effect
with difficult goals. Still, there is the possibility that we did not
measure these goal qualities reliably or that characteristics other
than those measured were effective as third variables. These
considerations led us to conduct an additional—this time exper-
imental— study in which implementation intentions were ma-
nipulated. In this study, we assigned participants the same
proj-
ect that could be assumed to have a low base rate of completion.
IMPLEMENTATION INTENTIONS191
Moreover, as we did not want to rely on participants'
self-
reports in assessing goal completion, we used more objective
criteria.
Study 2: Writing an Assigned Report
Over Christmas Holidays
We conducted this study 1 week before Christmas. All partici-
pants were asked to form the goal intention of writing during
the Christmas holidays a vivid report on how they spent Christ-
mas Eve. (The 2 days following Christmas Eve are holidays in
Germany.) We instructed participants that written reports were
to be sent back to us immediately afterwards. Half of the partici-
pants were asked to form implementation intentions on when
(e.g., right after church on the first holiday) and where (e.g.,
at my father's desk) they intended to sit down and start writing.
By requesting a performance to be shown to us, we were able
to check on actual goal completion. In addition, we thought
that this assignment was both novel and awkward enough to
guarantee a low base rate of accomplishment.
Method
Participants
Eighty-six students (43 men and 43 women) studying different majors
at
the
University
of
Munich participated. Participants
did not
receive
any financial remuneration
but
volunteered. We used volunteers
to
keep
participants' feelings
of
obligation
to
meet
the
assigned goal
(and
thus
the base rate
of
goal completion) rather low. The average age
of
partici-
pants
was 23
years; ages ranged from
19 to 29
years.
Procedure
One week before Christmas, participants were addressed
at
several
meeting spots around the university. Participants were randomly assigned
to
one of
two experimental conditions, although
we
took care
to
ensure
that male
and
female participants were distributed equally across condi-
tions.
The
experimenter kept himself unaware
of
participants' assign-
ment
to a
condition.
He
distributed
an
equal number
of
envelopes that
carried
a
symbol
for
female
or
male participants
but not for the
experi-
mental condition.
The cover story
was
presented orally
to the
participants.
It was ex-
plained that
a
demographic study
was
being conducted
on how
people
spend their leisure time (i.e., weekends, holidays, and vacations). Partici-
pants were told that,
in
this study,
the
researchers were interested
in
how people celebrate Christmas
Eve.
Assignment
of
the goal. Participants were then handed
a
sheet
of
paper that described details
of
the procedure. The first part
of
the written
instructions
was
identical
for
both conditions. Instructions stressed that
participants would remain anonymous. Anonymity
was
secured
by a
specific coding scheme that required participants
to
create their
own
code number from three letters and one digit (i.e., first letter
of
mother's
first name,
of her
maiden name,
and of
their
own
place
of
birth; first
digit
of
their date
of
birth). Then, participants were told that over
the
Christmas holidays they should write
a
valid report
of
how they spent
the afternoon
and
evening
of
Christmas Eve. More specific,
we
wanted
them
to
describe
how
they felt about
it and
how much
it met
their idea
of pleasant leisure time. Length
and
style
of the
report
was up to the
participants.
It was
explained
to
participants that memories
of
experi-
ences during leisure time fade rather quickly;
to
assure valid reports,
it
was
imperative that participants write them during
the
Christmas
holidays.
Manipulation
of
implementation intentions.
At
this point,
the in-
struction sheet started
to
differ
for
participants
in the
implementation
intention condition
as
compared
to the
control condition. Half
of the
participants were instructed
to
form
an
implementation intention speci-
fying when
and
where during these two holidays they intended
to
write
their report. These participants picked
a
specific point
in
time (e.g.,
right after breakfast
the
next morning)
and a
certain place (e.g.,
in a
quiet corner
in the
living room)
for
starting
to
write. Moreover, they
were asked to visualize the chosen opportunity and to commit themselves
to seize
it by
silently saying
"I
intend
to
write the report
in
die situation
z."
In
addition, participants
had to
record their choices
of
time
and
place
on the
instruction sheet. Control participants
did not
have
to go
through this part
of
the procedure.
Assessment
of
goal completion. When participants submitted
the
instruction sheet to the experimenter, they were supplied with
a
prepared
form
(on
which they were
to
write the report)
and a
stamped envelope.
The envelope carried
our
address,
and the
prepared form prompted
participants
to
write down their personal code
and the
time (hour
and
date) when they wrote
the
report. There
was
also
a
short instruction
that repeated that participants should write
as
valid
a
report
as
possible
using
no
more than two pages
(the
form consisted
of
two pages only).
All participants were asked
to
send
the
finished reports back
to us as
soon
as
possible. Note that,
on the
prepared form, participants were
not
reminded again
to
write
the
report during
the
Christmas holidays.
Finally, participants were asked
to put
their names
and
addresses
on
small index cards. Participants were told that
we
would need these
addresses
to
debrief them
on the
findings
of our
ongoing leisure time
research. The experimenter put the index cards
in a
stack kept separately
from
the
instruction sheets
and
stressed again that participants' data
would be confidential and that there was simply no way
to
identify them.
When
we
received
the
letters sent back
to us, we
checked whether
the
reports they contained were written during the Christmas holidays. Four
weeks after Christmas, when participants
had
stopped sending reports
to
us, we
debriefed them about
the
purpose
of
the study.
Results
Dependent Variable: Rate
of
Goal Completion
Overall, 39 participants (i.e., 45% of the original sample)
sent a report on Christmas Eve to us. Three participants failed
to write their personal code on the answer sheet, thus rendering
a correct identification of their experimental condition impossi-
ble.
We analyzed whether the remaining 36 participants wrote
their reports during the Christmas holidays by looking at various
measures (Table 1). First, when we checked the dates partici-
pants had written on the prepared form, there was clear evidence
that implementation intention participants wrote their reports
earlier than did control participants. The mean for implementa-
tion intention participants was 2.3 days after Christmas Eve,
whereas for control participants the respective mean was 7.7
days,
f(2l.3) = 2.85, p < .01. More important, whereas 71%
of the implementation intention participants (i.e., 12 of 17 parti-
cipants) wrote their report within the critical time specified (i.e.,
within 2 days after Christmas Eve), only 32% of the control
participants (i.e., 6 of 19 participants) did so, X2O> N = 36)
= 5.46, p = .02. There were no main or interaction effects of
gender.
One might argue that the implementation intention partici-
pants systematically deceived us by writing a false date on their
reports. There are good arguments against this assumption.
When we checked the dates of the postmarks on participants'
192GOLLWITZER AND BRANDSTATTER
Table 1
Various Measures Assessing Participants' Success in Writing
a Requested Report During Christmas Holidays
Dependent variable
Implementation
intentions
Formed Not formed
Mean time of writing the report (in days) 2.3a 7.7b
Mean time of mailing the report (in days) 4.9a 12.6b
Reports written during Christmas
holidays (%) 71a 32b
Participants sticking to the intended day
(%) 83 —
Nate. Means within rows not sharing a common subscript are signifi-
cantly different at p ±s .02. Dash indicates that this dependent variable
does not apply to participants who have not formed implementation
intentions.
letters, control (c) participants' letters carried postmarks with
significantly later dates than did those of implementation inten-
tion (ii) participants (Ma = 4.9 vs. Mc - 12.6), f( 18.5) = 2.73,
p — .01. Moreover, the time between the date of writing the
report and mailing it, reflected in the difference between date
of writing and date of the postmark, did not differ between
conditions (p > .35). This result suggests that the dates indi-
cated on the prepared form were equally trustworthy in both
conditions.
How Did Implementation Intentions Increase the Rate
of Goal Completion?
Implementation intentions did bind the execution of goal-
directed behaviors to the specified opportunities. Of the imple-
mentation intention participants who wrote their reports during
Christmas holidays, 83% (i.e., 10 of 12) did so on the day they
had committed themselves to on the instruction sheet, x2(
1»
N
— 12) = 4.80, p — .03. This finding supports our assumption
that implementation intentions promote goal achievement by
means of the facilitation of action initiation when the specified
situational context is present.
In Study 1, we demonstrated that implementation intentions
do not achieve their beneficial effects on goal completion by
means of an increase in goal importance (i.e., enhanced moti-
vation to reach the goal). Was there any evidence in Study 2
that also addresses this issue? Although we did not measure
perceived importance of the assigned goal after the implemen-
tation intention manipulation had occurred, we observed that
instructing participants to form implementation intentions did
not enhance the overall rate of cooperation in the sense of
more reports being sent to us. Implementation intention partici-
pants and control participants did not differ in the number of
mailed reports, F(l, 82) = 2.31, ns, nor was there a gender
main effect (F < 1) or an interaction effect with gender (F
< 1). Moreover, implementation intentions did not produce
more extensive reports. A 2 (experimental condition) X 2 (gen-
der of participant) analysis of variance (ANO\A) on the num-
ber of words written revealed that implementation intention
participants wrote shorter reports than did control participants
(MVl = 183 vs. Mc = 258), F(l, 35) - 5.43, p < .03. It
appears then that implementation intentions did not make it
easier for participants to meet their goal of writing a report
on Christmas Eve during the Christmas holidays by increasing
participants motivation to write and send reports to us; rather,
implementation intentions helped participants meet their goal
by facilitating action initiation when the specified opportunity
to write the report was encountered.
Discussion
All participants had agreed to write a report about Christmas
Eve during the Christmas holidays (i.e., during the 2 days fol-
lowing Christmas Eve). This goal intention led to a completion
base rate (i.e., 32%) that was very similar to that of difficult
goals analyzed in Study 1 (i.e., 22%). When the goal intention
of Study 2 was furnished with implementation intentions, how-
ever, completion rate drastically increased from 32% to 71%.
The amount of this increase parallels that of Study 1 (i.e., from
22%
to 62%). Taken together, the two studies strongly suggest
that the completion of difficult goal intentions is promoted by
implementation intentions. Implementation intentions were
self-
set in Study 1, whereas in Study 2 these intentions were as-
signed. Still, both types of implementation intentions facilitated
goal completion. Apparently, whether implementation intentions
are self-set or induced by others is not a crucial variable.
With respect to our ideas on how implementation intentions
help people meet their goals, the additional data collected in
Study 2 are only suggestive. Our theory (Gollwitzer, 1993)
states that implementation intentions create strong associative
links between the specified situational contexts and intended
behaviors. These links are thought to promote goal completion
because the respective behavior is immediately triggered when
the specified situational context is encountered. When we looked
at the time when successful implementation intention partici-
pants wrote their reports, the vast majority did so on the day
that was marked by their implementation intentions. Still, Study
2 was a field experiment, and we could not observe when the
initiation of goal-directed behaviors actually occurred. In addi-
tion, participants specified the situational contexts in their im-
plementation intentions in their own words (e.g., right after
breakfast on the first Christmas holiday). On the reports sent
back to us, participants indicated the hour and date of action
initiation. Accordingly, it was difficult for us to determine how
immediately participants used the specified opportunities to
write their reports.
All of these problems can be easily circumvented if one runs
a controlled laboratory experiment in which good opportunities
to act on an assigned goal are presented to all participants
whereby some participants are allowed to form implementation
intentions specifying these opportunities as the critical situa-
tional stimuli. To test the idea that intended opportunities are
seized immediately once they are encountered, one only needs
to present these opportunities at a later point in time and observe
whether participants who have formed implementation inten-
tions respond to these stimuli more readily than participants
who have not formed implementation intentions. Our third study
followed this line of thought.
IMPLEMENTATION INTENTIONS193
Study
3:
Immediate Seizing
of
Intended Opportunities
We figured that the postulated immediacy of action initiation
with participants who have formed implementation intentions
can be studied best in situations in which seizing a presented
good opportunity is rather difficult. This is true, for instance,
when good opportunities to act are rather short lived and when
the intended behavior is somehow problematic. An example of
a situation that ideally operationalizes these prerequisites is a
controversial and confrontational conversation in which the
other person expresses opinions with which one disagrees and
wants to counterargue. Good opportunities to do so are fleeting
because conversations are in a constant flux (Duncan & Fiske,
1977). Moreover, among the many opportunities, it is often hard
to pick the right occasion when one's own statement really hits
the nail on the head. Finally, people generally follow the norma-
tive rule of not contradicting others and thus are hesitant to
make their point (Tesser & Rosen, 1975).
Following this line of reasoning, participants in the present
study had to take a convincing counterposition towards xeno-
phobic remarks made by a confederate presented on videotape
(goal intention). To increase the importance of the task, we told
participants that communication researchers would judge the
quality of their counterarguments at the end of the experiment.
After participants were made familiar with the statements of
their opponent in a first viewing of the video, they had to mark
in a second run those points they personally considered to be
particularly suitable for commenting on (i.e., a good opportu-
nity).
One group of participants was additionally instructed to
form implementation intentions. They committed themselves to
counterargue at the chosen good opportunities. Finally, in a
modified third run, participants viewed a new videotape that
consisted of all of the confederate's earlier statements mixed
with several new ones of comparable content and length. Partici-
pants' task then was to stop the videotape whenever they liked
and to deliver their opinion on audiotape. We hypothesized that
implementation intention participants would more successfully
use marked good opportunities than control participants who
had not formed implementation intentions.
Method
Participants
Sixty male students enrolled
at the
Technical University
of
Munich
participated
in the
study
for a
remuneration
of 15 DM
(approximately
$10).
Participants were recruited through leaflets.
The
study
was de-
scribed
as
part
of a
line
of
research
on
verbal behavior
in
controversial
and confrontational discussions. Participants were between
20 and 29
years
of
age, with
a
median
age of 22
years.
Materials
and
Apparatus
Apparatus. Participants were seated
at a
table
in
front
of a
color
video monitor,
a
panel with
two
push buttons,
and a
microphone.
The
button panel
as
well
as the
video recorder (Sony U-Matic)
and
audio
recorder were connected
to a
microcomputer placed behind
a
screen.
The right button
was
used
to
mark
the
points
on the
videotape that
seemed suitable
for
comment.
At the
beginning
of
each statement,
a
magnetic signal
on
the second sound track
of
the videotape
set in
motion
a computer-controlled clock. When participants pressed the right button,
this clock
was
stopped
and the
time was recorded
by the
computer.
The
left button was used
to
stop
and
start
the
videotape
in
the third viewing
of
the
video when participants were supposed
to
interfere. Again,
the
time when this button was pressed was recorded
by the
computer. This
procedure allowed
us to
check later
on how
immediately participants
responded
to the
marked opportunities.
Videotapes. The videotapes showed
a
confederate sitting
in
front
of
a microphone. On the first tape, which was shown
in the
first
and
second
run,
the
confederate delivered
8
statements that were very hostile
to
foreigners. Each statement
was
about
30 s
long
and
separated from
others
by a
pause
of 4 s. The
overall duration
of the
videotape
was
approximately
5 min. The
second tape, shown
in the
critical third
run
(i.e.,
where participants were allowed
to
counterargue), displayed
8
additional statements
so
that, altogether,
16
statements were shown.
Procedure
Participants arrived
at the
experimental room individually
and
were
randomly assigned
to one of
three experimental conditions (i.e.,
the
implementation intention condition
and two
control conditions). They
were greeted
by a
female experimenter
and
asked
to
fill
out
three intro-
ductory questionnaires,
two
attitudinal measures;
and a
personality
in-
ventory,
the
Sixteen Personality Adjectives Scales (16-PA),
an
adjective
version
of the
Sixteen Personality Factors Questionnaire (16-PF;
H.
Brandstatter, L988). The first questionnaire recorded participants' inter-
est
in and
attitudes about
the
integration
of
foreigners
in
Germany.
Six
items (e.g.. How attentively do you follow the discussion
on
the integra-
tion
of
foreigners
in the
media?) that were answered
on
10-point scales
anchored
by 0 (not at all) and 9
(very much) were averaged
to
form
a
composite score
of
interest
in
this topic (Cronbach's
a =
.76).
In the
second questionnaire, participants were asked
to
rate
on
5-point scales
ranging from —1 (strongly disagree)
to 2
(strongly agree)
how
much
they agreed with 28 racist statements. These statements touched different
themes, including those
the
confederate
on the
videotape focused
on
(e.g., "Because
of the
high number
of
foreigners many Germans lost
their jobs."). These ratings were also averaged
to
form
a
composite
score
of
attitudes toward these racist statements (Cronbach's
a = .86).
All participants were then given
the
following oral instructions:
The present study aims
at
analyzing
the
different strategies people
employ
in
controversial discussions when they want
to
speak
up
and express their own point
of
view. Hence, you will
be
confronted
with
the
arguments
of a
person
on a
topic that
is
currently being
discussed
all
over, namely the integration
of
foreigners
in
Germany.
For experimental reasons your counterpart
in the
discussion
is re-
corded
on
videotape.
The
arguments presented
by the
person
on
the videotape originate from
a
different study conducted some time
ago.
There
we
interviewed students about their attitudes toward
foreigners
in
Germany. A student researcher
of
the Institute selected
the most terse arguments.
He
will
try to
present these arguments
grouped into eight thematically homogenous statements
in the
most
persuasive manner possible.
Implementation intention condition
(I/C).
At this point, the instruc-
tions
for the
three experimental conditions started
to
differ.
In the IIC,
participants were informed that they were
to
express their own opinion
convincingly
by
stopping
the
videotape
at any
point they liked.
At the
end
of the
experimental session their contributions, recorded
on
audio-
tape,
would
be
evaluated
by
social psychologists. They would then
re-
ceive feedback concerning
the
persuasiveness
of
their arguments
on the
basis
of
timing and content. However, participants were told that before
they started arguing they should first familiarize themselves with
the
persuasive attempts
of
their interlocutor
in a
first viewing
of the
videotape.
After participants watched
the
tape they were told,
194GOLLWITZER
AND
BRANDSTATTER
You will
be
shown
the
same tape
in a
second
run
now.
For
each
statement
of
your interlocutor you may mark that point
in
time that
seems most suitable
to you to
intervene. Press
the
right button
on
the panel
box for
marking. When
you
have marked this particular
point, please commit yourself
to
seize this opportunity
for
express-
ing your opinion
in the
next video run
by
saying:
"I
will speak
up
here!"'
Each
of the
eight statements
may be
marked only once.
When participants
had
gone through this marking procedure, they
were asked
to
perform
a
5-min distractor task (counting
the
planes
of several different geometrical figures drawn
on
paper). Thereafter,
participants
had to
recall (free recall)
as
many
of the
presented argu-
ments
as
possible. This recall test
was
conducted
to
control
for the
possibility that implementation intention participants followed the eight