Power and Time Predictions 1
In Press: Journal of Experimental Social Psychology
How Long Will It Take?
Power Biases Time Predictions
University of Kent
University College London
Keywords: Power, control, time, planning fallacy, forecasting
Please address correspondence to Mario Weick, Department of Psychology,
University of Kent at Canterbury, Canterbury, Kent, CT2 7NP, UK. Email:
Power and Time Predictions 2
People tend to underestimate the time it takes to accomplish tasks. This bias known
as the planning fallacy derives from the tendency to focus attention too narrowly on
the envisaged goal and to ignore additional information that could make predictions
more accurate and less biased. Drawing on recent research showing that power
induces attentional focus, four studies tested the hypothesis that power strengthens the
tendency to underestimate future task completion time. Across a range of task
domains, and using multiple operationalizations of power, including actual control
over outcomes (Study 1), priming (Studies 2 and 3), and individual differences (Study
4), power consistently led to more optimistic and less accurate time predictions.
Support was found for the role of attentional focus as an underlying mechanism for
those effects. Differences in optimism, self-efficacy, and mood did not contribute to
the greater bias in powerful individuals’ forecasts. We discuss the implications of
these findings for institutional decision processes and occupational health.
Power and Time Predictions 3
How Long Will It Take?
Power Biases Time Predictions
Time is a crucial factor in people’s everyday lives. Business executives,
policy makers, engineers, nurses, teachers or students routinely plan their work and
estimate the time it will take to accomplish tasks. Psychological research shows that
these estimates are systematically biased, and people tend to underestimate the time it
takes to accomplish tasks. Biased time predictions are a widespread phenomenon that
affects mundane everyday activities as well as large-scale business projects (e.g.,
Buehler, Griffin, & Ross, 1994; Hall, 1980; Schnaars, 1989). Unreliable time
predictions have attracted a great deal of public attention and are commonly known as
the planning fallacy (Kahneman & Tversky, 1979).
Considering the practical relevance and common interest, relatively little is
known about the factors that alter biases in time predictions. Interventions to reduce
the planning fallacy have had limited success overall (e.g., Byram, 1997; see Roy,
Christenfeld, & McKenzie, 2005, for a review). The present research extends past
research by focusing on the way the wider social context affects time predictions. In
particular, we suggest that being in a position of power strengthens the tendency to
underestimate future task completion time. Our proposal derives from recent research
indicating that power promotes a goal-directed attentional focus (e.g., Guinote,
2007a), and from the observation that biases in time predictions originate from a too
narrow focus on the envisaged goal.
Attentional Focus and Biased Time Predictions
Biased estimates of time originate from the ways individuals process
information (e.g., Newby-Clark, Ross, Buehler, Koehler, & Griffin, 2000; Buehler &
Griffin, 2003; Kahneman & Lovallo, 1993). Specifically, individuals tend to focus
Power and Time Predictions 4
attention too narrowly on the event in question and disregard additional information
that could make predictions more accurate and less optimistic. Below, we discuss
three sources of information that are commonly neglected.
Kahneman and Tversky (1979) suggested that when people plan the future
they adopt an ‘inside view’ that overemphasizes the uniqueness of a target event. By
ignoring the distribution of similar events people fail to consider how long similar
tasks usually take (see also Kahneman & Lovallo, 1993). Later research confirmed
that people are reluctant to consider past experiences in their planning, and this makes
them prone to bias in their forecasts (Buehler et al., 1994; Buehler & Griffin, 2003).
Underestimates of time also derive from the failure to take contingencies
sufficiently into account (e.g., Buehler et al., 1994). Future plans often resemble best-
case-scenarios and people tend to focus on the ways they can successfully accomplish
their goals (e.g., Newby-Clark et al., 2000). By ignoring alternative ways how the
future may unfold people are prone to misjudge impediments (e.g., Griffin, Dunning,
& Ross, 1990); a bias which sharply increases with the number of potential setbacks
Finally, people also tend to focus too heavily on a global representation of the
task at hand. As a result, they may not consider all subcomponents that a task affords
(Kruger & Evans, 2004; see also Fischhoff, Slovis, & Lichtenstein, 1987). Task
subcomponents that are less evident are especially at risk of being ignored (cf. Kruger
& Evans, 2004). Consistent with this reasoning, unpacking tasks into subcomponents
can decrease the tendency to underestimate task completion time (Kruger & Evans,
2004), and focusing attention on the intended outcome increases bias (e.g., Taylor,
Pham, Rivkin, & Armor, 1998).
Power and Time Predictions 5
In sum, biases in time predictions operate through attentional mechanisms.
People tend to focus too narrowly on the event in question and do not consider
sufficiently additional information that could make predictions more accurate. The
more people focus on the intended outcome, the more they are prone to bias in their
forecasts. Factors that alter people’s attentional focus can, therefore, strengthen or
alleviate the tendency to underestimate time. For example, enhancing people’s goal-
focus by means of instructions (Buehler & Griffin, 2003), or incentives (Buehler,
Griffin, & MacDonald, 1997; Byram, 1997) renders time predictions less accurate.
One social variable that has a profound impact on goal-directed attentional
focus is social power – the ability to influence and control others’ outcomes and
resources (see Fiske & Berdahl, 2007). As discussed next, individuals in power
display a processing orientation that should make them more prone to bias in
predictions of task completion time.
Power Affects Goal-Directed Attention
At the basic cognitive level, power fosters selective attention, enhancing the
processing of task relevant information and leading to the inhibition of secondary
information (Guinote, 2007b). Consequently, powerful individuals typically display a
more simplified, narrow focus of attention consistent with activated constructs (e.g.,
goals, needs, affordances), whereas powerless individuals pay less attention to
primary constructs and attend more to secondary information (see Guinote, 2007a, for
a review of the Situated Focus Theory of Power).
When goals are activated, powerful individuals pay more attention to
information that pertains to their focal goal, and less attention to secondary
information as compared to powerless individuals (Guinote, 2007b, 2007c; see also
Overbeck & Park, 2001; Slabu & Guinote, in press). For instance, research on social
Power and Time Predictions 6
perception shows that powerful individuals focus more unequivocally on stereotypic,
or on individuating information depending on their current goals (Overbeck & Park,
2001) and influence strategies (Vescio, Snyder, & Butz, 2003). In contrast, powerless
individuals show less clear prioritizations between these different sources of
Power also fosters an approach-related motivational orientation (see Keltner,
Gruenfeld, & Anderson, 2003, for a review of the Approach/Inhibition Theory of
Power). As a result, powerful individuals are more oriented towards achieving gains
and rewards, and less apprehensive about avoiding threats or constraints as compared
to powerless individuals (see also Galinsky, Magee, Gruenfeld, Whitson, &
Liljenquist, 2008). Power thus magnifies people’s focus on desired outcomes and
reduces the tendency to consider potential setbacks and threats to successful goal
The propensity to achieve rewarding outcomes facilitates action (Galinsky,
Gruenfeld, & Magee, 2003; Keltner et al., 2003; see also Guinote, Judd, & Brauer,
2002), and powerful individuals benefit from their greater focus during goal striving
(Guinote, 2007c). However, little is known about how power affects the planning of
future tasks. The same cognitive orientation that benefits goal-striving should
ironically create conditions for stronger biases in time predictions.
Powerful individuals’ greater focus on single sources of information facilitates
an ‘inside view’ (Kahneman & Lovallo, 1993) and reduces the likelihood that
powerful individuals will consider distributional information in their forecasts. As a
result, powerful individuals are more likely to ignore the duration of similar events
that have taken place in the past. By the same token powerful individuals are less
likely to consider multiple alternatives in their future plans, and therefore are less
Power and Time Predictions 7
prone to recognize potential setbacks and interfering events (cf. Buehler, Griffin, &
Ross, 2002; Griffin et al., 1990; Hoch, 1985). Due to their greater focus on primary
sources of information, powerful individuals may also fail to consider task
subcomponents that are not readily apparent. The greater approach-motivation of
powerful individuals further contributes to the neglect of threatening and impeding
information, and it makes powerful individuals more oriented towards task outcomes
in detriment of task subcomponents (cf. Keltner et al., 2003).
Differences in attentional focus may not be the only reason why powerful
individuals are prone to underestimate time. Powerful individuals might dismiss
impediments or threats because they are generally more optimistic about the future
(Anderson & Galinsky, 2006). Powerful individuals might also have greater
confidence in their abilities to overcome impediments and to complete tasks
successfully (Anderson & Galinsky, 2006). Such differences in self-efficacy beliefs
(Bandura, 1982) could lead to greater bias in forecasts (cf. Hinds, 1999). Lastly,
power can promote positive affect (Keltner et al., 2003; Wojciszke & Struzynska-
Kujalowicz, 2007), which could induce more optimistic estimates of future task
The prospect that optimism contributes to greater bias in the time predictions
of powerful individuals is, however, dampened by past research that failed to find an
association between optimism and temporal bias (Buehler & Griffin, 2003; Jørgensen
& Faugli, 2006). Likewise, empirical evidence is lacking that confirms the link
between mood and biased time predictions. Finally, past research on power and
decision making found that elevated self-efficacy beliefs do not necessarily lead to
more optimistic decisions in powerholders (Anderson & Galinsky, 2006). Taken
Power and Time Predictions 8
together, differences in optimism, self-efficacy, and mood might be less influential in
promoting greater bias in the time predictions of powerful individuals.
The Present Research
The present article tests the hypothesis that power increases the tendency to
underestimate task completion time. We suggest this to be the case because power
affects attention in ways that promote biased time predictions. Specifically, power
induces a goal-directed attentional focus and the tendency to disregard additional
information that lies outside the focal goal. Powerful individuals are, therefore, less
likely to consider information that could make predictions more accurate and less
biased (i.e., impediments, task subcomponents, and past experiences).
Previous research has shown that biases in time predictions arise through the
ways people allocate attention in the planning of future events. This led us to posit
that attentional focus is a key mechanism through which power fosters greater bias in
forecasts. At the same time, we also examined the role of optimism, self-efficacy, and
mood as additional pathways through which power may affect time predictions.
We conducted four studies to test the hypothesis that power increases bias in
predictions of task completion time. Studies 1 and 4 compared predicted and actual
completion time of students’ coursework projects, Study 2 examined biased time
predictions in a computerized formatting task, and Study 3 focused on time estimates
for everyday activities. One study examined attentional focus as underlying
mechanism for the effects of power (Study 3), two studies examined self-efficacy
(Studies 2 and 4) and mood (Studies 3 and 4), and one study explored the role of
optimism (Study 4). Power was operationalized using different experimental
manipulations (Studies 1-3) and an individual difference measure (Study 4).
Power and Time Predictions 9
In the present study, students estimated the time it would take them to
complete an important coursework assignment. The study was conducted
longitudinally, and estimates (time 1) were compared against actual completion time
(time 2). Power was manipulated at time 1 by having half of the participants believe
their opinion would have important consequences for others, placing them in a
position of control over others’ outcomes (see Fiske & Dépret, 1996). As part of an
ostensibly unrelated task, participants then predicted the completion time of their
assignment. If powerful participants focused more on the envisaged goal (i.e.,
completion of the coursework) at the expense of considering additional information
such as impediments or past experiences, then they should anticipate an earlier
completion time than participants who did not have power. This should lead to a
greater discrepancy between actual and predicted completion times for powerful
individuals compared to individuals who did not have power.
Participants and Design
Twenty students (11 females and 9 males) from the University of Kent
participated on a voluntary basis. They were randomly assigned to the experimental
conditions (powerful vs. control).
Procedure and Materials
Participants were approached on a university campus and were invited to take
part in a study focusing on student-life perception. Power was manipulated using a
procedure adopted from Goodwin and Fiske (1995; cited in Fiske & Dépret, 1996),
such that half of the participants believed they controlled the outcomes of other
students. Specifically, participants learned about alleged plans of the University to
Power and Time Predictions 10
introduce a university-wide course credit scheme that would make it compulsory for
future students across all disciplines to participate in Psychology projects.
Participants in the powerful condition were informed that their opinions received a
fixed weight of 50% towards the final decision of the University. In contrast,
participants in the control condition were told that the University was interested in
their opinions, but these would not affect the future introduction of the scheme.
As part of a purportedly unrelated task, participants then indicated the deadline
for their next coursework assignment. They also rated how important they considered
their assignment (1=not at all, 9=very much). Subsequently, participants estimated
when they expected to finish (a) a first draft of their assignment, (b) the final version,
and (c) when they expected to submit their assignment. Actual completion time was
assessed two weeks later over the telephone.
Results and Discussion
Predicted and actual completion times were coded in number of days before
the official deadline (see Table 1). Higher values indicate an earlier point in time
compared to the deadline. The completion times for a first draft and a final version
were more strongly associated with each other than with the completion times for
submission (rs(18) ≥ .83, ps < .001 vs. rs(18) ≤ .54, ps ≥ .015), giving rise to
sphericity. Where appropriate, the subsequent results are corrected for these
differential associations using the Greenhouse-Geisser epsilon correction (e.g.,
Stevens, 1986). We submitted participants’ predicted and actual completion times to
a 2 (power: powerful vs. control) x 2 (time: predicted vs. actual completion time) x 3
(task: first draft vs. final version vs. submission) mixed model analysis of variance
with repeated measurement on the last two factors. The analysis revealed a main
effect of time, suggesting that participants’ estimates were biased, F(1, 18) = 100.92,
Power and Time Predictions 11
p < .001, ηp2 = .85. On average, participants took two days longer to complete their
assignments than originally anticipated (Ms = 3.82 vs. 1.88 days). Of particular
importance, a significant interaction between power and time emerged, F(1, 18) =
7.68, p = .013, ηp2 = .30. As can be seen in Table 2, powerful participants showed a
greater discrepancy between their predicted and actual completion times than control
participants (powerful: Ms = 4.93 vs. 2.47 days; control: Ms = 2.70 vs. 1.30 days).
Confirming our predictions, power led to greater underestimates of time. Further tests
of simple effects confirmed that this bias derived from differences in participants’
predictions. While power also led participants to complete their projects earlier (t(18)
= 3.99, p = .001, d = 1.88), the predictions of participants in power were nevertheless
further off and much more optimistic than the predictions of participants who did not
have power, t(18) = 4.78, p < .001, d = 2.25. Overall, powerful participants erred
76% more in their predictions than control participants (Mspredicted vs. actual time = 2.46 vs.
1.4 days), t(18) = 2.77, p = .013, d = 1.31.
A number of additional effects emerged from the analysis. A significant
interaction between task and time indicated that underestimates of task completion
time were more pronounced for earlier tasks (first draft, final draft) than for later tasks
(submission), F(2, 36) = 60.03, p < .001. A main effect of task confirmed that
completion times differed between the three tasks, F(2, 18) = 158.22, p < .001, ε =
.654. There was also a main effect of power (F(1, 18) = 24.93, p < .001), qualified by
an interaction with task (F(2, 18) = 13.08, p = .001, ε = .654), suggesting that
completion times were overall swifter for powerful than for control participants, but
only for earlier tasks (first draft, final draft). The three-way interaction between
power, time, and task was not significant (p = .154), indicating that the effects of
power on biased time predictions did not differ between the three tasks.
Power and Time Predictions 12
An examination of participants’ ratings of task importance revealed that
powerful and control participants considered their assignment equally important (Ms
= 7.00 vs. 6.70), t(18) < 1.
The results of Study 1 provided support for the hypothesis that power induces
biased predictions of time. Powerful participants were less accurate and tended to
underestimate more the time it would take them to complete a future assignment
compared to participants who did not have power.
Consistent with past research on goal pursuit (Guinote, 2007c) and action
facilitation (Galinsky et al., 2003), powerful participants also completed some parts of
their assignments earlier than control participants. It is noteworthy that these effects
emerged several days after the manipulation of power at time 1.
One limitation of the present study is that we did not control for differences in
the content of participants’ assignments. It is possible that powerful participants
chose more complex assignments than control participants, and such differences in
task complexity could potentially account for the observed differences in time
predictions (cf. Roy et al., 2005). Our next study thus controlled for differences in
In the present study, all participants performed the same computerized task,
using a common word editing software (Kruger & Evans, 2004). Participants
received an unformatted text and their task was to amend the formatting of the text to
make it look like a template provided (see Figure 1). This procedure allowed for
predicted and actual completion times to be recorded within the same experimental
Power and Time Predictions 13
We manipulated power using an episodic priming procedure. Following
Galinsky, Gruenfeld, and Magee (2003), participants recalled a past event in which
they were either in a position of power, or in a position of powerlessness. Through
the activation of affective, cognitive, and behavioral response repertoires associated
with states of power or powerlessness, this procedure affords an examination of the
effects of power in situations in which power is not directly experienced (cf. Galinsky
et al., 2003). We hypothesized that participants primed with power would show a
greater bias in their time predictions than participants primed with powerlessness.
The present study explored the role of self-efficacy beliefs in promoting
greater bias in powerful individuals’ forecasts. As noted earlier, power could lead to
less accurate time predictions because powerful individuals have greater confidence in
their abilities to accomplish tasks (cf. Hinds, 1999). To address this conjecture,
participants indicated how capable they were of formatting texts (cf. Bandura, 1982).
Participants and Design
Forty students (27 females and 13 males) from the University of Kent
participated in exchange for chocolate bars. Participants were randomly assigned to
the experimental power conditions (powerful vs. powerless).
Procedure and Materials
Participants were informed that they would take part in a study on episodic
memory, and in a separate study focusing on visual perception. At the beginning,
participants were asked to provide a vivid written report of a past situation where they
had power over another individual, or where someone else had power over them
(Galinsky et al., 2003).
Power and Time Predictions 14
Subsequently, participants were asked to format a document using a standard
word editing software (Kruger & Evans, 2004). Participants received three dictionary
definitions as templates. As shown in Figure 1, the task required putting unformatted
text in bold, italics, uppercase, lowercase, as well as to insert punctuations, symbols,
brackets, and spaces.
Before participants began the formatting task, they filled in a measure of self-
efficacy, asking them to indicate how skilled they were in formatting texts on a scale
ranging from 1 (not at all) to 9 (very much).1 A female experimenter, who was
unaware of the experimental conditions, prompted participants for an estimate of how
long it would take them to make all the formatting changes. Unbeknownst to
participants, estimates were recorded, and actual completion time was measured using
a stopwatch. After completion, participants were thanked, probed for suspicion, and
Results and Discussion
Participants indicated how much they felt in charge in the situation they
described in their essays (1=not at all, 9=very much). Participants in the powerful
condition felt more in charge than did participants in the powerless condition (Ms =
6.50 vs. 4.05), t(38) = 3.47, p = .001, d = 1.12, indicating that the manipulation was
successful in prompting the recall of past episodes in which participants felt powerful
Task Completion Time
Two participants were identified as outliers (>3 x SD) and removed from
subsequent analyses. We subjected the two measures of predicted and actual
completion time to a 2 (power: powerful vs. powerless) x 2 (time: predicted vs. actual
Power and Time Predictions 15
completion time) mixed model analysis of variance with repeated measurement on the
last factor. The analysis yielded the expected main effect of time, F(1, 36) = 107.83,
p < .001, ηp2 = .75. Participants took considerably longer to complete the formatting
task than they had predicted (Ms = 5.13 vs. 9.02 min). Of special importance, a
significant interaction between power and time emerged, F(1, 36) = 8.22, p = .007,
ηp2 = .19 (see Table 2). As expected, participants primed with power showed a
greater discrepancy between their forecasts and their actual completion time than
participants primed with powerlessness (powerful: Ms = 3.95 vs. 8.91 min; powerless:
Ms = 6.32 vs. 9.13 min). An examination of simple effects revealed that, while power
did not alter participants’ actual completion time (t(36) < 1), it led to shorter time
predictions, t(36) = 4.70, p < .001, d = 1.57. Paralleling the findings of Study 1,
power fostered greater bias in predictions of task completion time. Overall,
participants primed with power erred 77% more in their predictions than participants
primed with powerlessness (Mspredicted vs. actual time = 4.96 vs. 2.81 min), t(36) = 2.87, p =
.007, d = .96.
We subsequently examined the contributions of self-efficacy to the effects of
power.2 Priming power did not alter participants’ self-efficacy beliefs (Ms = 5.68 vs.
5.16), F < 1. Inclusion of self-efficacy as a covariate in the mixed model analysis of
variance described above also did not alter the effects of power; the interaction
between power and time remained unchanged (F(1, 35) = 7.40, p = .010), and none of
the effects involving self-efficacy were significant, ps ≥ .271. The effects of power
on biased time predictions appear, therefore, to be unrelated to differences in self-
The results showed that powerful participants did not complete the formatting
task earlier than powerless participants. This contrasts with previous research, which
Power and Time Predictions 16
has found that power facilitates action and can lead to earlier goal-attainment
(Galinsky et al., 2003; Guinote, 2007c). Consequently, we also examined
participants’ performance in terms of number of mistakes made. There was no
difference between the experimental conditions (Ms = 16.20 vs. 13.55; F < 1),
suggesting that in the present study power did not lead to differences in actual
Studies 1 and 2 support the hypothesis that power fosters biases in estimates of
task completion time. Across studies, power was associated with greater error in
forecasts. These effects emerged in different task domains and could not be attributed
to differences in self-efficacy beliefs.
The findings of Studies 1 and 2 are consistent with the notion that powerful
individuals focus too narrowly on the anticipated outcome and do not consider
sufficiently information that can potentially debias predictions. In the next study, we
examined more directly the role of attentional focus as an underlying mechanism for
the effects of power.
The goal of the present study was to provide evidence for the role of
attentional focus in promoting differences in the time estimates of powerful and
powerless individuals. To this end, we employed an experimental approach in which
we manipulated attentional focus. If the greater bias in the time predictions of
powerful individuals derives from a narrow focus of attention on the goal at hand,
then drawing attention to additional information that has a bearing on task completion
time should render the forecasts of powerful individuals less optimistic. Importantly,
no such effect should be observed for powerless individuals, who naturally attend to
information that lies outside the focal goal. In more generic terms, by manipulating
Power and Time Predictions 17
the pathway between power and attention (the proposed mediator), we expected to
observe corresponding changes in the relationship between power and time estimates
(the outcome variable). This would be a first indication for a mediating role of
attentional focus (see Spencer, Zanna, & Fong, 2005).
The present study focused on predictions of completion time for everyday
activities. Attentional focus was manipulated by drawing attention to relevant past
experiences in half of the participants. As noted earlier, taking into account previous
experiences can render time predictions more accurate and less optimistic. We
expected that power would lead to more optimistic predictions, but this effect would
diminish when participants’ attention is drawn to relevant previous experiences. We
also assessed mood to see if differences in participants’ affective states contributed to
the effects of power on time predictions.
Participants and Design
Sixty-four students (30 females, 32 males, and 2 unknown genders) from the
University of Kent participated in exchange for chocolate bars. They were randomly
assigned to the conditions of a 2 (power: powerful vs. powerless) x 2 (attentional
focus: past vs. control) between-subjects factorial design.
Procedure and Materials
Participants were informed that they would take part in a study dealing with
people’s perceptions of social situations, and a study dealing with the planning of
future events. Power was manipulated using the same episodic priming manipulation
employed in Study 2. As part of an alleged separate study, participants were then
instructed to imagine themselves in four hypothetical scenarios that involved the
completion of future tasks. Specifically, participants imagined they had to write a
Power and Time Predictions 18
2000-word essay, to get ready for an evening out, to shop in the supermarket, and to
prepare a three-course meal. For each scenario, participants could choose one of
several response alternatives to indicate their anticipated completion time. For ‘essay
writing’, response alternatives varied from 1 to 14 days in 1-day incremental steps; for
‘getting ready to go out’ and ‘supermarket shopping’, response alternatives ranged
from 15 min to 4 h in steps of 15 min; and for ‘preparing a three-course meal’, the
responses ranged from 30 min to 6 hs in incremental steps of 30 min.3 Participants
were instructed to imagine themselves having to perform the four tasks. To
manipulate attentional focus, half of the participants were instructed to recall how
much time it had taken them in the past to complete the tasks described in the
scenarios. Following Buehler and colleagues (1994), they were also asked to write a
short description of how they were going to perform the tasks in the future
considering their previous experiences. This manipulation was designed to draw
participants’ attention to previous experiences that were relevant for completing their
goals (cf. Buehler et al., 1994). Upon completion, all participants estimated the task
completion times for the four tasks by choosing one of the response alternatives.
Participants also rated their current mood on four 7-point scales, ranging from -3 (very
bad; very sad; very discontent; very tense) to 3 (very good; very happy; very content;
very relaxed). On completion, participants were thanked and debriefed.
Participants responded to the same manipulation check employed in Study 2.
Participants in the powerful condition felt more in charge of the situation than did
participants in the powerless condition (Ms = 7.20 vs. 2.74), t(60) = 11.88, p < .001, d
= 3.07, indicating that the priming manipulation was successful.
Power and Time Predictions 19
Task Completion Time
Attentional Focus. Answers given to the four scenarios were transformed into
continuous measures (i.e. days or min), standardized, and then collapsed into a single
score of predicted task completion time (α = .44).4 This score was subjected to a 2
(power: powerful vs. powerless) x 2 (attentional focus: past vs. control) between-
subjects analysis of variance. The analysis yielded the predicted interaction between
attentional focus and power, F(1, 60) = 4.01, p = .050, ηp2 = .06 (see Table 3). An
examination of simple effects confirmed that, in the control condition, powerful
participants predicted on average shorter completion times than powerless participants
(Ms = -.31 vs. .14), t(60) = 2.21, p = .031, d = .57. Paralleling the findings of the
previous studies, power led to more optimistic time predictions. Importantly, when
participants’ attention was drawn to previous experiences, the predictions of
participants primed with power no longer differed from the predictions of participants
primed with powerlessness (Ms = .14 vs. -.05), t(60) < 1. Planned contrasts
confirmed that drawing attention to relevant past experiences rendered the predictions
of participants primed with power less optimistic (t(60) = -2.05, p = .045, d = .53), but
it did not affect the predictions of participants primed with powerlessness, t(60) < 1.
No other significant effects emerged, Fs < 1. These results lend support to the
assumption that differences in attentional focus underlie the effects of power on time
Mood. There was a marginal tendency for participants primed with power to
report being in a better mood than participants primed with powerlessness (Ms = .65
vs. .16), t(60) = -1.81, p = .075, d = .47. Nevertheless, inclusion of mood as a
covariate in the analysis of variance described above did not alter the effects of
power; the interaction between attentional focus and power remained unchanged (F(1,
Power and Time Predictions 20
57) = 4.61, p = .036), and differences in mood did not account for any variance in
time predictions, F < 1. These results indicate that the effects of power were
unrelated to differences in mood.
Retrospective Time Estimates. If power affected time predictions by altering
the ways participants allocated attention in the planning of the future, then one would
expect no difference in the retrospective time estimates of powerful and powerless
participants. To confirm this supposition, we created a composite index of past
completion time (α = .56), which did not differ between powerful and powerless
participants (Ms = .12 vs. -.16), t(32) = 1.25, p = .220. We also examined the
correlations between past and future completion times. As expected, this correlation
was very high (r(32) = .93, p < .001) and did not differ as a function of power, z < 1.
These results support the conclusion that power affected time estimates through its
effects on future planning behavior.
The findings of Study 3 support the hypothesis that power affects goal-
directed attention in ways that induce optimistic predictions of task completion time.
Similar to the previous studies, participants primed with power predicted shorter
completion times than participants primed with powerlessness. Drawing participants’
attention to relevant information that lies outside the focal goal (here: past
experiences) eliminated the effects of power on time predictions. The manipulation
of attentional focus only affected powerful, but not powerless participants. Moreover,
power only led to differences in future time estimates, but it did not affect time
estimate for the past. These results are consistent with a causal role of attentional
focus in promoting greater bias in powerful individuals’ forecasts.
We conducted a subsequent study to establish the generalizability of our
findings. To this end, we examined individual differences in power within a larger
Power and Time Predictions 21
survey that was conducted over the course of an academic term. The study also
explored further underlying mechanisms of the effects of power. As noted earlier, a
greater bias in the time predictions of powerful individuals could also arise from
differences in optimism. Powerful individuals might dismiss debiasing information
because they are generally more optimistic about the future (Anderson & Galinsky,
2006). Our next study examined the role of optimism in the present findings. We also
included measures of self-efficacy and mood to see if these variables contributed to
the effects of power in the context of the field survey.
In the present study, students estimated the time it will take them to complete
a coursework assignment, and estimates were compared against actual completion
time, paralleling the procedure of Study 1. Unlike Study 1, the present study followed
an individual difference approach and participants indicated how much power they
experienced in their everyday relationships with other people (Anderson, John, &
Keltner, 2005; cited in Anderson & Galinsky, 2006). This measure of sense of power
was embedded in a university-wide survey. The survey was carried out towards the
beginning (time 1) and the end (time 2) of the academic term. We reasoned that if
individual differences in power predict bias in time estimates in this diverse sample,
and over a prolonged period of time, then this would be a strong indicator for the
generalizability of our findings.
Another goal of the present study was to examine different underlying
mechanisms for the effects of power. Specifically, we included a measure of
optimism to see if the greater optimism typically observed in powerful individuals
contributes to the greater bias in powerful individuals’ forecasts. As in Study 3, we
assessed mood to see if power promoted more optimistic estimates because power can
Power and Time Predictions 22
elevate people’s mood (Keltner et al., 2003). Lastly, as in Study 2, we measured self-
efficacy to examine whether power leads to greater bias in time estimates in part
because powerful individuals have greater confidence in their abilities.
Participants and Design
Four-hundred and twenty-one students (270 females and 151 males) from the
University of Kent participated in the first part of this correlational study (time 1).
Lottery tickets for a £50 (~ $70) cash-prize draw were offered in return for
participation. Two-hundred and forty-nine students (165 females and 84 males)
completed the second part of the study (time 2). Participants were enrolled in various
degrees across more than twelve subject areas; 83% studied at undergraduate level
and 17% pursued graduate studies.
Procedure and Materials
Towards the beginning of the academic term (time 1), an email was distributed
university-wide, inviting students to participate in an online study on individual
differences and student-life perception. The web-based survey asked participants to
specify a written coursework assignment that they had to submit by the end of the
academic term, as well as the exact deadline for their assignment (week, day, hours).
Next, participants made a number of time predictions for different tasks required to
complete their assignment. They indicated when they were going to start reading for
their assignment, when they were going to start writing, when they were going to
finish writing, and when they were going to submit their assignment. Participants
were given the opportunity not to respond if a task did not apply to their particular
Power and Time Predictions 23
The survey also employed a number of individual difference measures. Eight
items assessed how much power participants had in their everyday relationships with
other people (Anderson, John, & Keltner, 2005). Example items include ‘In my
relationship with others, I can get people to listen to what I say’, or ‘I can get others
to do what I want’ (1=disagree strongly, 7=agree strongly). Furthermore, a ten-item
scale measured participants’ degree of optimism (LOT-R; Scheier, Carver, & Bridges,
1994). Example items include ‘In uncertain times, I usually expect the best’, or ‘I
rarely count on good things happening to me’ (reverse coded) (1=disagree a lot,
5=agree a lot). A ten-item scale assessed participants’ self-efficacy beliefs
(Schwarzer & Jerusalem, 1995). Example items are ‘I am certain that I can
accomplish my goals’, or ‘I can always manage to solve difficult problems if I try
hard enough’ (1=not at all true, 4=exactly true). Finally, a four-item scale measured
participants’ mood (-3=very bad, very sad, very discontent, very tense; 3=very good,
very happy, very content, very relaxed). Participants also indicated their academic
and demographic background, and at the end they described what they thought were
the aims of the study.
At time 2, an invitation for a follow-up survey was sent to participants’ email-
addresses, typically within 2 weeks after their assignment had been due. Participants
were reminded of the nature of their assignment. They then indicated when they had
completed each of the four tasks (i.e., start reading, start writing, finish writing,
submission) using the same items employed at time 1. At the end, participants were
probed for suspicion and debriefed. Background information was sent to all
participants enrolled in the study.
Power and Time Predictions 24
For unknown reasons, 39 participants made predictions for past assignments,
or assignments that were due shortly after (<2 weeks) time 1. Moreover, four
participants expressed an intention to submit their assignments more than one week
past the deadline. These participants were omitted from subsequent analyses.5 The
final sample consisted of 206 participants.
We only considered time estimates for future tasks. For example, if
participants had already begun reading for their assignment at time 1, then their time
estimates for this task (start reading) were not considered. This procedure resulted in
the omission of 16.8% of all temporal data (i.e., predicted and actual completion
time). In addition, 1.2% of the temporal data was discarded because of missing time 1
or time 2 comparators.
We coded predicted and actual task completion time in number of days
before the deadline, separately for each of the four tasks. Higher values indicate
earlier completion times. Table 4 displays descriptive statistics for these data. On
average, participants completed all four tasks later than predicted. This difference
was significant for all but one task (‘start reading’). Next, we created single indices
for all variables in preparation for subsequent regression analyses.6 We standardized
participants’ time 1 predictions for the four tasks and averaged those scores into a
single index of predicted task completion time (α = .81, M = .04, SD = .87). A single
index of actual task completion time was created in the same way for the time 2 data
(α = .85, M = -.02, SD = .90). We also subtracted the actual from the predicted
completion times, separately for each of the four tasks to create indices of temporal
bias. This was done using the unstandardized raw data. The difference scores were
then standardized and merged into a single index of biased time predictions (α = .86,
M = .05, SD = .89). Higher values indicate greater underestimates of task completion
Power and Time Predictions 25
time. Lastly, we averaged the individual differences measures into single indices of
sense of power (α = .84, M = 4.66, SD = .88), self-efficacy (α = .86, M = 3.00, SD =
.44), optimism (α = .80, M = 3.26, SD = .78), and mood (α = .81, M = .67, SD = 1.16).
Reverse coded items were recoded prior to data aggregation.
Task Completion Time
Sense of Power. Table 5 displays the zero-order correlations between all
indices derived from the present study. As can be seen, sense of power was positively
related to predicted task completion time, r(204) = .20, p = .005. Paralleling the
findings of the previous studies, the greater participants’ sense of power, the earlier
participants anticipated completing their assignments. Differences in power did not
relate to differences in actual completion time, r(204) = -.03, p = .667. Most
importantly, power was associated with an increased bias in participants’ forecasts,
r(204) = .21, p = .003. The greater participants’ sense of power, the more participants
underestimated the time it would take them to complete their assignment.
Mediating Variables. Sense of power was positively related to optimism, self-
efficacy, and mood (see Table 5), rs(204) ≥ .19, ps ≤ .008. To see if any of these
variables contributed to the effects of power, we conducted a series of regression
analyses in which we entered sense of power and the control measures as predictors of
predicted task completion time, and as predictors of bias. Controlling for the effects
of optimism, sense of power exerted a significant effect on predicted task completion
time (β = .17, p = .023), and on bias (β= .22, p = .004). Similarly, controlling for the
effects of self-efficacy, the effects of power remained significant (predicted task
completion time: β = .18, p = .019; bias: β = .15, p = .044). The same picture
emerged when we controlled for differences in mood (predicted task completion time:
β = .17, p = .013; bias: β = .18, p = .009). Lastly, when all control measures were
Power and Time Predictions 26
entered simultaneously in multiple regression, power emerged as the only significant
(p<.05) predictor (predicted task completion time: β = .16, p = .036; bias: β = .16, p =
.035). Additional Sobel tests revealed that there were no indirect effects of sense of
power on predicted task completion time, or on bias via any of the control measures
(optimism: ZSobels ≤ 1.14, ps ≥ .253; self-efficacy: ZSobels ≤ 1.59, ps ≥ .110; mood:
ZSobels ≤ 1.57, ps ≥ .116). Together, these results indicate that power affected time
estimates independently of optimism, self-efficacy, and mood.7
Gender. The larger sample size of Study 4 allowed us to explore the role of
gender differences. Independent sample t-tests revealed that gender was unrelated to
predicted and actual task completion times, and it did not account for any variance in
bias, ts(204) < 1. Inclusion of gender in the above described multiple regression
analyses did not alter the effects of sense of power (predicted task completion time: β
= .20, p = .004; bias: β = .21, p = .003). Additional moderated multiple regression
analyses following the procedure outlined by Aiken and West (1991) confirmed that
gender did not moderate the effects of sense of power, ps ≥ .159. Taken together,
gender did not contribute to the present findings.
Task complexity. To examine whether power affected time predictions
independently of task content, we also explored the role of task complexity. Two
independent coders rated the complexity of the assignments described by participants
at time 1 on a five-point scale (1=not at all, 5=very much). The ratings were highly
correlated and averaged into a single score (α = .81, M = 3.19, SD = .73). We then
conducted two separate moderated regression analyses, using the standardized
complexity ratings, the sense of power scores, as well as the interaction term of those
two variables as predictors of predicted task completion time, and as predictors of bias
Power and Time Predictions 27
(see Aiken & West, 1991). Only sense of power emerged as a significant predictor
from these analyses (predicted task completion time: β = .20, p = .004; bias: β = .21, p
= .003), and none of the effects involving task complexity were significant, ps ≥ .156.
This suggests that power affected time predictions irrespectively of task complexity.
Focusing on individual differences in power, Study 4 provided further
evidence for the hypothesis that power biases time predictions. A university-wide
survey found that the more students experienced a sense of power in their everyday
relationships with others, the earlier they anticipated completing a future assignment,
and the more their time predictions were prone to error.
Study 4 also explored the contributions of a number of factors to the effects of
power. Consistent with past research on time planning (Buehler & Griffin, 2003;
Jørgensen & Faugli, 2006), no support was found for the role of optimism in
promoting greater error in the forecasts of powerful individuals. Corroborating the
findings of Studies 2 and 3, the effects of power were independent from differences in
self-efficacy and from differences in mood. Contributing to the generalizability of the
present findings, the effects of power emerged outside the laboratory in a diverse
sample, over a prolonged period of time, and across a range of tasks.
People tend to underestimate the time it takes them to accomplish future tasks.
This bias in time predictions originates from the way people allocate attention in the
planning of future events. Specifically, people tend to focus too narrowly on the
intended outcome and tend to neglect additional information that could make
predictions more accurate and less biased (i.e., impediments, previous experiences,
and task subcomponents). In the present article, we tested the hypothesis that power
increases biases in predictions of task completion time. This hypothesis derived from
Power and Time Predictions 28
previous research showing that power induces a goal-directed attention focus
(Guinote, 2007a, 2007c).
Four studies tested, and found support for these hypotheses. Across different
domains, involving the planning of academic work (Studies 1 and 4), computerized
tasks (Study 2), and everyday activities (Study 3), powerful participants, more than
control (Study 1) or powerless participants (Studies 2-4), consistently underestimated
the time it would take them to complete future tasks. This pattern was observed using
a variety of operationalizations of power, ranging from actual control over outcomes
(Study 1), to priming power and powerlessness (Studies 2 and 3), and measuring
individual differences in sense of power (Study 4). Together, these studies show for
the first time that power has a fundamental impact on planning behavior, increasing
biases in time estimates. These findings are important because the planning of
powerful individuals has wide-ranging implications for individual, organizational, and
Across the four studies, we examined several factors that are of relevance to
the link between power and underestimates of time. We did not find evidence that
power leads to greater bias in time predictions because powerful people are more
optimistic; consistent with previous research on time planning (see Buehler & Griffin,
2003; Jørgensen & Faugli, 2006). Likewise, the effects of power could not be
attributed to differences in people’s beliefs in their abilities (i.e., self-efficacy), or to
differences in mood.
Consistent with previous research on time predictions (e.g., Buehler & Griffin,
2003; Buehler et al., 2002; Newby-Clark et al., 2000; Kruger & Evans, 2004), Study 3
found support for the role of attentional focus as an underlying mechanism for the
effects of power. In this study, drawing participants’ attention to relevant information
Power and Time Predictions 29
that lies outside the focal goal eliminated the otherwise observed differences in the
time predictions of powerful and powerless individuals. This lends support to the
assumption that the effects of power on time predictions derive from the way power
affects the allocation of attention in the planning of future events.
Some caveats of the present research should be addressed. First, none of our
studies manipulated powerlessness together with a baseline condition. The present
research is, therefore, not conclusive as to whether lacking power reduces the
tendency to underestimate time. Lack of power broadens individuals’ attentional
focus (Guinote, 2007b) and lessens motivational states associated with approach
(Keltner et al., 2003; Smith & Bargh, 2008). By the same token that high power
increases bias in time predictions we believe that low power should reduce the
tendency to underestimate time. Tentative support for this assertion derives from
Study 3, which found a linear increase in bias moving from low levels of experienced
power to high levels of power.
Second, while previous research has shown that power facilitates action
(Galinsky et al., 2003) and leads to swifter goal-attainment (Guinote, 2007c), the
current work provided only mixed evidence for such effects. Power benefited goal
striving in Study 1, but it did not lead to faster goal-attainment in Studies 2 and 4.
Different factors could account for the variability in the findings. The formatting task
in Study 2 was repetitive and not very demanding on executive functions (e.g.,
prioritization, cognitive flexibility), which could have minimized the benefits of
power for goal pursuit. In Study 4, participants had to deal with a multitude of goals
that typically arise throughout the course of an academic term. Power might have led
to greater efficiency in the pursuit of other, more immediate goals that were not
Power and Time Predictions 30
assessed in the study (e.g., other assignments, social goals), which could explain the
discrepancy between past research and the present findings.
Finally, only one study provided direct evidence for the role of attentional
focus in contributing to the effects of power. More research is needed to further
investigate how attentional focus operates to promote differences in the time
predictions of powerful and powerless people. For instance, it would be important to
establish which sources of information are neglected by powerful individuals that
render their time estimates less accurate. This information might vary between tasks
and contexts. For example, the tendency to neglect previous experiences might be
more influential in promoting bias in the predictions of mundane activities (Study 3),
while the tendency to neglect detailed aspects of a task might carry greater weight in
promoting bias in a novel formatting task (Study 2).
Power affects how people conceive information that lies in the focus of
attention. Specifically, power fosters abstraction (Smith & Trope, 2006; see also
Guinote et al., 2002), leading powerful individuals to construe information at a
higher-level of information processing (cf. Semin & Fiedler, 1988; Trope &
Liberman, 2003; Vallacher & Wegner, 1987). One question that arises is what role
abstraction plays in the present findings. When people construe tasks in abstract
terms they tend to focus more on the desirability of end-states and less on the
feasibility of reaching those end-states (Liberman & Trope, 1998). By drawing
individuals’ attention to task outcomes and away from task subcomponents,
abstraction can contribute to increase biases in time predictions (cf. Kruger & Evans,
At the same time, abstraction can also contribute to reduce biases in time
predictions. It has been shown that when people form concrete plans about the future
Power and Time Predictions 31
they are more prone to bias in their forecasts (Buehler & Griffin, 2003). One reason
why this tends to be the case is that events that are construed at a concrete level are
perceived as more atypical and unique (e.g., Liberman, Sagristano, & Trope, 2002;
Smith, 1998). Conversely, events that are construed at a more abstract level are
perceived as more prototypical and common. As a result, abstraction counteracts the
‘inside view’ and increases the likelihood that people will consider past experiences in
their forecasts (cf. Kahneman & Lovallo, 1993). Taken together, abstraction may be a
double-edged sword and may have opposing effects on the tendency of powerful
individuals to underestimate time.
It is important to note that we only examined time predictions for tasks that
were unrelated to the exercise of power. Previous research on time planning has
shown that the greater individuals’ motivation to accomplish a goal, the more people
tend to be biased in their forecasts (Buehler, Griffin, & MacDonald, 1997; Byram,
1997). One might well expect that power fosters even greater bias in forecasts when
the stakes are high. Indeed, some of the historic mistakes in time planning occurred
when those in charge were under pressure to succeed (Hall, 1980; Schnaars, 1989).
Empirical research is required to establish how power affects time estimates for tasks
that are relevant to the maintenance, or to the furtherance of people’s power.
Lovallo and Kahneman (2003) pointed out that the same mechanisms that
make people prone to underestimate time also increase the likelihood that individuals
take unwarranted risks. The present findings thus complement past research that has
shown that power promotes more optimistic risk-assessment (Anderson & Galinsky,
2006). The planning behavior of powerful individuals could be another factor that
contributes to the propensity of powerful individuals to take risks.
Power and Time Predictions 32
The present research contributes to the literature on time predictions by
showing that one’s position in a social hierarchy can alleviate or strengthen biases in
forecasts. Previous research has focused primarily on the role of situational variables
(e.g., incentives, see Buehler et al., 1997) and particularities of the task (e.g., duration,
see Roy, Christenfeld, & McKenzie, 2005) as determinants of temporal biases. As the
present findings show, the extent to which people control their own and other
individuals’ outcomes affects time estimates.
There has been a recent accumulation of evidence suggesting that power leads
to greater efficiency in a number of domains, including decision making (e.g, Smith,
Dijksterhuis, & Wigboldus, 2008), goal pursuit (e.g., Guinote, 2007c), and general
cognitive functioning (e.g., Guinote, 2007b; Smith, Jostmann, Galinsky, & van Dijk,
2008). The present findings are important because they show that power does not
always have beneficial effects; it can be detrimental for planning and lead to greater
errors in forecasts.
The present findings have significant applied value as they point out that those
who are typically in charge of planning (e.g., managers; decision and policy makers)
are more at risk to fall prey to biases in their forecasts. This is not only important
because of the economic costs associated with inaccurate time estimates (e.g., Buehler
et al., 1994); overly ambitious time scales can put employees under pressure, induce
stress, and contribute to overtime work (see Sparks, Cooper, Fried, & Shirom, 1997;
Burchell, Ladipo, & Wilkinson, 2001). In this view, biases in time predictions of
decision makers could have significant adverse effects on the well-being and general
quality of life of employees. Future research should address these broader
implications of the present findings.
Power and Time Predictions 33
Notably, the present research also points out ways to alleviate the biasing
effects of power. Specifically, Study 3 suggests that forecasting accuracy could be
improved by using techniques that draw people’s attention to information that lies
outside the focal goal. Moreover, changing the power structure of planning
committees could also be a way to render forecasts more accurate (e.g., by assigning
greater weight to the predictions of individuals who do not have power). It remains
for future research to explore the applications of the present findings for the
prevention of human biases in forecasts.
Power and Time Predictions 34
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting
interactions. Thousand Oaks, CA: Sage.
Anderson, C., & Galinsky, A. D. (2006). Power, optimism, and risk-taking. European
Journal of Social Psychology, 36, 511-536.
Bandura, A. (1982). Self-efficacy mechanism in human agency. American
Psychologist, 37, 122–147.
Bar-Hillel, M. (1973). On the subjective probability of compound events.
Organizational Behavior and Human Decision Processes, 9, 396-406.
Buehler, R., Griffin, D., & Ross, M. (1994). Exploring the “planning fallacy”: Why
people underestimate their task completion times. Journal of Personality and
Social Psychology, 67, 366-381.
Buehler, R., Griffin, D., & Ross, M. (2002). Inside the planning fallacy: The causes
and consequences of optimistic time predictions. In T. D. Gilovich, D. W.,
Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of
intuitive judgment (pp. 250-270). New York: Cambridge University Press.
Buehler, R., Griffin, D., & MacDonald, H. (1997). The role of motivated reasoning in
optimistic time predictions. Personality and Social Psychology Bulletin, 23,
Buehler, R., & Griffin, D. (2003). Planning, personality, and prediction: The role of
future focus in optimistic time predictions. Orgnizational Behavior and Human
Decision Processes, 92, 80-90.
Burchell, B. J., Ladipo, D., & Wilkinson, F. (2001). Job insecurity and work
intensification. London: Routledge.
Power and Time Predictions 35
Byram, S. J. (1997). Cognitive and motivational factors influencing time predictions.
Journal of Experimental Psychology: Applied, 3, 216-239.
Fischhoff, B., Slovic, P., & Lichtenstein, S. (1978). Fault trees: Sensitivity of
estimated failure probabilities to problem representation. Journal of
Experimental Psychology: Human Perception and Performance, 4, 330-344.
Fiske, S. T., & Berdahl, J. (2007). Social power. In A. Kruglanski & E. T. Higgins
(Eds.), Social psychology: Handbook of basic principles (2nd ed., pp. 678-
692). New York: Guilford.
Fiske, S. T., & Dépret, E. (1996). Control, interdependence and power: Understanding
social cognition in its social context. In W. Stroebe & M. Hewstone (Eds.),
European Review of Social Psychology (Vol. 7, pp. 31-61). Sussex: John
Wiley & Sons Ltd.
Galinsky, A. D., Gruenfeld, D. H., & Magee, J. C. (2003). Power and action. Journal
of Personality and Social Psychology, 85, 453-466.
Galinsky, A. D., Magee, J. C., Gruenfeld, D. H., Whitson, J. A., & Liljenquist, K. A.
(2008). Power reduces the press of the situation: Implications for creativity,
conformity, and dissonance. Journal of Personality and Social Psychology, 95,
Griffin, D., & Buehler, R. (2005). Biases and fallacies, memories and predictions:
Comment on Roy, Christenfeld, and McKenzie (2005). Psychological Bulletin,
Griffin, D., Dunning, D., & Ross, L. (1990). The role of construal processes in
overconfident predictions about the self and others. Journal of Personality and
Social Psychology, 59, 1128-1139.
Power and Time Predictions 36
Guinote, A. (2007a). Behaviour variability and the Situated Focus Theory of Power.
European Review of Social Psychology, 18, 256–295.
Guinote, A. (2007b). Power affects basic cognition: Increased attentional inhibition
and flexibility. Journal of Experimental Social Psychology, 43, 685-697.
Guinote, A. (2007c). Power and goal pursuit. Personality and Social Psychology
Bulletin, 33, 1076-1087.
Guinote, A., Judd, C. M., & Brauer, M. (2002). Effects of power on perceived and
objective group variability: Evidence that more powerful groups are more
variable. Journal of Personality and Social Psychology, 82, 708-721.
Hall, P. (1980). Great planning disasters. London: Weidenfeld & Nicolson.
Hinds, P. J. (1999). The curse of expertise: The effects of expertise and debiasing
methods on predictions of novice performance. Journal of Experimental
Psychology: Applied, 5, 205-221.
Jørgensen, M., & Faugli, B. (2006). Prediction of overoptimistic predictions. Paper
presented at the 10th International Conference on Evaluation and Assessment in
Software Engineering (EASE). Keele University, UK, 10-11 April.
Kahneman, D., & Lovallo, D. (1993). Timid choices and bold forecasts: A cognitive
perspective on risk taking. Management Science, 39, 17-31.
Kahneman, D., & Tversky, A. (1979). Intuitive predictions: Biases and corrective
procedures. TIMS Studies in Mangement Sciences, 12, 313-327.
Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and
inhibition. Psychological Review, 110, 265-284.
Kruger, J., & Evans, M. (2004). If you don’t want to be late, enumerate: Unpacking
reduces the planning fallacy. Journal of Experimental Social Psychology, 40,
Power and Time Predictions 37
Liberman, N., Sagristano, M. D., & Trope, Y. (2002). The effect of temporal distance
on level of mental construal. Journal of Experimental Social Psychology, 38,
Liberman, N., & Trope, Y. (1998). The role of feasibility and desirability
considerations in near and distant future decisions: A test of temporal construal
theory. Journal of Personality and Social Psychology, 75, 5-18.
Lovallo, D., & Kahneman, D. (2003). Delusions of success: How optimism
undermines executives’ decisions. Havard Business Review, 81, 56-63.
Maibach, E., & Murphy, D. A. (1995). Self-efficacy in health promotion research and
practice: conceptualization and measurement. Health Education Research, 10,
Maner, J. K., Gailliot, M. T., Butz, D. A., Peruche, B. M. (2007). Power, risk, and the
status quo: Does power promote riskier or more conservative decision making?
Personality and Social Psychology Bulletin, 33, 451-462.
Newby-Clark, I. R., Ross, M., Buehler, R., Koehler, D. J., & Griffin, D. (2000).
People focus on optimistic scenarios and disregard pessimistic scenarios while
predicting task completion times. Journal of Experimental Psychology: Applied,
Overbeck, J. R., & Park, B. (2001). When power does not corrupt: Superior
individuation processes among powerful perceivers. Journal of Personality and
Social psychology, 81, 549-565.
Roy, M. M., Christenfeld, N. J. S., &McKenzie, C. R. M. (2005). Underestimating the
duration of future events: Memory incorrectly used or memory bias?
Psychological Bulletin, 131, 738-756.
Power and Time Predictions 38
Scheier, M. F., Carver, C. S., & Bridges, M. W. (1994). Distinguishing optimism
from neuroticism (and trait anxiety, self-mastery, and self-esteem): A
reevaluation of the Life Orientation Test. Journal of Personality and Social
Psychology, 67, 1063-1078.
Schnaars, S. P. (1989). Megamistakes: Forecasting and the myth of rapid
technological change. New York: Free press.
Schwarzer, R, & Jerusalem, M. (1995). Generalized Self-Efficacy Scale. In J.
Weinman, S. Wright, & M. Johnston (Eds.), Measures in health psychology: A
user's portfolio. Causal and control beliefs (pp. 35-37). Windsor England:
Semin, G. R., & Fiedler, K. (1988). The cognitive functions of linguistic categories in
describing persons: Social cognition and language. Journal of Personality and
Social Psychology, 54, 558-568.
Slabu, L., & Guinote, A. (in press). Getting what you want: Power increases the
accessibility of active goals. Journal of Experimental Social Psychology.
Sparks, K., Cooper, C., Fried, Y., & Shirom, A. (1997). The effects of hours of work
on health: A meta-analytic review. Journal of Occupational and Organizational
Psychology, 70, 391-408.
Spencer, S. J., Zanna, M. P., & Fong, G. T (2005). Establishing a causal chain: Why
experiments are often more effective than mediational analyses in examining
psychological processes. Journal of Personality and Social Psychology, 89, 845-
Stevens, J. (1986). Applied multivariate statistics for the social sciences. Hillsdale,
Power and Time Predictions 39
Smith, E. R. (1998). Mental representation and memory. In D. T. Gilbert, S. T. Fiske,
& L. Gerdner (Eds.), The handbook of social psychology (4th ed., pp. 391-445).
New York: McGraw-Hill.
Smith, P. K., & Bargh, J. A. (2008). Nonconsious effects of power on basic approach
and avoidance tendencies. Social Cognition, 26, 1-24.
Smith, P. K., Dijksterhuis, A., & Wigboldus, D. H. J. (2008). Powerful people make
good decisions even when they consciously think. Psychological Science, 19,
Smith, P. K., Jostmann, N. B., Galinsky, A. D., & van Dijk, W. W. (2008). Lacking
power impairs executive functions. Psychological Science, 19, 441-447
Smith, P. K., & Trope, Y. (2006). You focus on the forest when you are in charge of
the trees: Power priming and abstract information processing. Journal of
Personality and Social Psychology, 90, 578-596.
Taylor, S. E., Pham, L. B., Rivkin, I. A., & Armor, D. A. (1998). Harnessing the
imagination: Mental simulation, self-regulation, and coping. American
Psychologist, 53, 429-439.
Trope, Y., & Liberman, N. (2003). Temporal construal. Psychological Review, 110,
Vallacher, R. R., & Wegner, D. M. (1987). What do people think they’re doing?
Action identification and human behaviour. Psychological Review, 94, 3-15.
Vescio, T. K., Snyder, M., & Butz, D. A. (2003). Power in stereotypically masculine
domains: A social influence strategy X stereotype match model. Journal of
Personality and Social Psychology, 85, 1062-1078.
Wojciszke, B., & Struzynska-Kujalowicz, A. (2007). Power influences self-esteem.
Social Cognition, 25, 472-494.
Power and Time Predictions 40
Mario Weick, Department of Psychology, University of Kent, and Ana Guinote,
Department of Psychology, University College London
The research reported in this article was supported by an Economic and Social
Research Council (ESRC) doctoral studentship and the Economic and Social
Research Council (ESRC) grant PTA-026-27-1908 to the first author, and the
Nuffield Foundation social sciences grant SGS/35183 to the second author. We thank
Amal Awadhi for her assistance in carrying out Study 2, and Milica Vasiljevic for
proofreading the manuscript.
Power and Time Predictions 41
1 See Maibach & Murphy (1995) for a single judgment approach to measure
2 The self-efficacy scores were normally distributed and centered on the
midpoint of the scale, M = 5.42, SD = 2.20, Z Kolmogorov-Smirnov = .80, p = .541.
3 The response alternatives were chosen based on a pre-test with students (N =
42) who had responded in an open-ended format to the four scenarios.
4 The effects reported in Study 3 did not differ reliably between the four
5 Natural breaks in the data supported the decision on the cut-off points for
excluding participants. Retention of all participants did not significantly alter the
6 In Study 4 we chose multiple regression as the data analytic strategy due to
the continuous nature of the independent variables (e.g., Aiken & West, 1991; Cohen
& Cohen, 1983). All other studies employed the general linear model (GLM)
procedure, which yields equivalent results as multiple regression analyses.
7 Interestingly, the analyses also showed that sense of power mediated the
effects of self-efficacy on temporal bias. When controlling for power, the effects of
self-efficacy were no longer significant (β = .12, p = .106), and there was a reliable
indirect effect of self-efficacy via power, ZSobel = 1.95, p = .051. One interpretation of
the overall pattern of the results is that elevated self-efficacy beliefs were a
consequence, but not an antecedent of the greater goal-focus of powerful individuals.
Power and Time Predictions 42
Predicted and actual completion times of written assignment (Study 1)
Task completion time
Task M SD M SD
6.50a 2.57 3.60a 1.60 9.45*** 19
Final Draft 4.00b 1.69 1.55b 1.00 9.20*** 19
.95c .76 .50c .51 2.13* 19
Note. * = p < .05, *** = p < .001. Means display number of days before a designated
deadline. Means with differing subscripts within columns are significantly different, p
Power and Time Predictions 43
Actual and predicted completion times as a function of condition (Studies 1-2)
Task completion time
M SD M SD
Powerful 4.93 1.33 2.47 .71
Control 2.70 .66 1.30 .60
Powerful 3.95 1.27 8.91 1.28
Powerless 6.32 1.80 9.13 2.51
a Scores indicate number of days before a deadline
b Completion time of formatting task in minutes
Power and Time Predictions 44
Predicted task completion time as a function of experimental condition (Study 3)
Focus M SD M SD
Control condition -.31 .39 .14 .70
Past focus .14 .80 -.05 .57
Note. Scores are z-standardized time estimates for a series of tasks
Power and Time Predictions 45
Predicted and actual completion times of written assignment (Study 4)
Task completion time
Task M SD M SD
31.01a 16.95 27.85a 26.04 1.49 97
Start Writing 26.21b 17.63 17.40b 19.45 6.41*** 159
8.96c 14.07 3.81c 15.98 5.00*** 188
Submission 3.33d 9.16 -.83d 14.54 4.24*** 205
Note. *** = p < .001. Means display number of days before a designated deadline.
Means with differing subscripts within columns are significantly different, p < .001.
Power and Time Predictions 46
Zero-order correlations between individual difference measures and predicted and actual completion time (Study 4)
Individual Difference Measures (t1)
Task completion time
Optimism Self-Efficacy Mood Predicted Actual Difference (Bias)
Sense of Power
.345*** .436*** .185**
.196** -.030 .207**
.484*** . 388***
.142* .070 .050
.114 -.078 .190**
.155* -.022 .158*
Note. * = p < .05, ** = p < .01, *** = p < .001. N = 206 for all analyses.
Power and Time Predictions 47
Figure 1. Example of formatting task used in Study 2.
Power and Time Predictions 48 Download full-text