BASIC AND APPLIED SOCIAL PSYCHOLOGY. 28(1). 1-16
BASIC AND APPLIED SOCIAL PSYCHOLOGY. 28(1). 1-16
Copyright 0 2006. Lawrence Erlbaum Associates. Inc.
Improved Self-Control: The Benefits of a Regular
Program of Academic Study
Megan Oaten and Ken Cheng
Macquarie University, Sydney
Academic examination stress impairs regulatory behavior by consuming self-control strength
(Oaten & Chang, 2005). In this study, we tested whether a study intervention program. a form
of repeated practice of self-control, could improve regulatory strength and dampen the debili-
tating effects of exam stress. We assessed 2 cohorts at baseline and again at the commencement
of exams. Without any intervention, we replicated our previous findings of deteriorations in
regulatory behaviors at exam time. Students receiving the study program, however, showed sig-
nificant improvement in self-regulatory capacity as shown by an enhanced performance on a
visual tracking task following a thought-suppression task. During examinations, these partici-
pants also reported significant decreases in smoking, alcohol. and caffeine consumption and an
increase in healthy eating, emotional control. maintenance of household chores, attendance to
commitments, monitoring of spending, and an improvement in study habits. Hence, the study
program not only overcame deficits caused by exam stress but actually led to improvements in
self-control even during exam time.
Self-regulation or self-control (terms used interchangeably
here) can be defined as the capacity to enact control over
one's behavior. Self-control is needed to override dominant
behaviors that may be self-destructive, irrational, or undesir-
able in the long term. Examples of typical self-control prob-
lems include not exercising enough, engaging in unsafe sex-
ual practices, abusing drugs and alcohol, overspending, and
not sticking to study schedules.
Our goals in this study were to (a) replicate the finding
that real world stress, specifically academic examinations,
consume self-control strength and consequently produce im-
pairments in a number of unrelated regulatory behaviors
(Oaten & Cheng, 2005a), and (b) test whether the repeated
practice of self-control (a study intervention program) could
improve regulatory strength and make students less vulnera-
ble to the debilitating effects of periods of high academic
demand.
RESOURCE MODEL OF SELF-CONTROL
A recent model suggests a lack of self-regulatory resources
as one reason why self-regulation might fail (Baumeister,
Correspondence should be addressed to Megan Oaten. Department of
Psychology. Macquarie University. Sydney. New South Wales. Austra-
lia 2109. E-mail: Heatherton, & Tice, 1994; Muraven, Tice, & Baumeister,
1998). The resource model considers self-control to operate
like a muscle. Any act of self-control tires this muscle, leav-
ing less available strength for subsequent self-control tasks.
This muscle is considered to fatigue easily, as all acts of
self-control have been argued to draw on a common resource
or regulatory strength that is of limited capacity and is there-
fore readily depleted. This aspect of the model is well estab-
lished, with evidence to suggest that in the short term, peo-
ple's capacity for self-control diminishes following exertion
much like muscular action. For example, when individuals
were asked to engage in tasks involving self-regulation, the
ability to self-regulate in subsequent activities significantly
declined (Muraven et al., 1998; Vohs & Heatherton, 2000;
Vohs & Schmeichel, 2003). This effect of depletion has been
reported across a variety of tasks in physical, intellectual, and
emotional domains.
ACADEMIC STRESS
AND SELF-CONTROL FAILURE
Failures of self-control may be related to experienced stress.
A disturbing trend in student health is the reported increase in
student stress internationally (Sax, 1997; Cotton, Dollard, &
de Jonge, 2002). Students report experiencing academic
stress at predictable times each semester, with the greatest
EFTA01113830
2 OATEN AND CHENG
sources of academic stress resulting from studying for and
taking exams, grade competition, and the large amount of
course content to master in a small amount of time (Archer &
Lamnin, 1985; Britton & Tesser, 1991; Kohen & Fraser,
1986). Examination periods have been used to investigate a
number of stress responses. A finding that surfaces in these
studies is that many forms of self-regulation break down
when people are managing stress. For example, West and
Lennox (1992) reported that smoking level among students
was higher immediately preceding exams than at a more neu-
tral period. Cartwright et al. (2003) revealed that greater aca-
demic stress was associated with more fatty food intake, less
fruit and vegetable intake, more snacking, and a reduced like-
lihood of daily breakfast consumption. Recent longitudinal
research has found that academic examination stress was as-
sociated with increases in cigarette smoking and decreases in
physical activity (Steptoe, Wardle, Pollard, Canaan, &
Davies, 19%).
In a previous study (Oaten & Cheng, 2005a), we tested
whether at stressful times (during examination periods) peo-
ple fail at self-regulation in domains in which control has pre-
viously been successful (e.g., diet). We found that students at
exam time reported breakdowns in regulatory behavior that
were not found in a control group. We found this effect in both a
laboratory task (Stroop Test; Stroop, 1935) and on a range of
self-reported day-to-day behaviors. Performance on the
Stroop Test deteriorated following thought suppression, a
form of regulatory activity, during the examination period.
Outside of the examination period, no such effect due to
thought suppression was evident. Exam time also proved det-
rimental to a number of other self-control operations. During
the examination period, students reported an increase in smok-
ing and caffeine consumption; a decrease in healthy dietary
habits, emotional control, frequency and duration of physical
activity, maintenance of household chores and self-care hab-
its, attendance to commitments, and monitoring of spending;
and deterioration of sleep patterns and study habits.
In light of the resource model of self-control, our interpre-
tation of the link between exam stress and self-control failure
is that managing stress requires self-regulation and thus de-
pletes limited regulatory resources. An important part of the
body's defenses for coping with stress is the "fight-or-flight"
response. The fight-or-flight response prepares people for
physical, emotional, and mental action and is considered es-
sential for survival (Selye, 1956). These fight-or-flight re-
sponses, however, can be counterproductive when dealing
with the stresses of modem life such as academic examina-
tions (Zillman, 1983). People therefore require self-regulation
to override these natural responses to substitute other, quite
unnatural responses (such as studying harder) in their place.
Stress regulation may also involve processes that demand
inhibition, such as ignoring sensations, overriding negative
thoughts, and suppressing emotions (Wegner & Pennebaker,
1993) as well as regulating attention (Hockey, 1984). Glass,
Singer, and Friedman (1%9) found that there is a "psychic cost" of controlling stress such that this cost is reflected in a re-
duced capacity to regulate task performance following an
external stressor (unpleasant electric shock or unpredictable
noise). Glass et al.'s (1969) findings that performance is im-
paired following stressors have been replicated many times us-
ing measures of frustration tolerance (Glass & Singer, 1972),
proofreading (Gardner, 1978; Glass & Singer, 1972), and the
Stroop Task (Glass & Singer, 1972). These tasks all required
the individual to override a dominant response, thus requiring
self-control (Muraven & Baumeister, 2000). It seems that the
work required to control stress leaves the individual less able to
regulate behavior successfully. Poorer self-control is a conse-
quence of previous attempts to regulate stress.
SELF-REGULATORY IMPROVEMENT
Thus, artificial laboratory tasks of self-regulation and having
to deal with the stress of examination both lead to poorer
self-control. These findings support one important aspect of
the resource model: depletion. In addition, the resource
model makes a second prediction: Self-control should also
become stronger with repeated practice, and such strengthen-
ing may provide a strategy to counter regulatory failure.
Previous research has found that the repeated practice of
self-control was followed by increments in self-control per-
formance (Muraven, Baumeister, & Tice, 1999; Oaten &
Cheng, 2005b; Oaten, Cheng, & Baumeister, 2003). In the
study with the longest duration, the uptake and maintenance
of an exercise program over a 2-month period produced sig-
nificant improvements in a wide range of regulatory behav-
iors (Oaten & Cheng, 2005b). Improvements were found in a
laboratory task (visual tracking under distraction, which is
used in this study as well) and on many self-reported every-
day behaviors. The laboratory measure and the self-reported
behaviors bore no resemblance to the exercise program other
than that they all involved self-regulation. In particular, indi-
viduals who participated in the exercise program demon-
strated better self-regulation in other spheres: related (e.g.,
they engaged in more healthy behaviors), unrelated (e.g.,
missed fewer appointments), and laboratory based (visual
tracking task WM.
There are two ways in which self-control strength could
be improved. These are consistent with the ways in which
muscular strength can be increased: power (an increase in the
simple baseline capacity) and stamina (a reduction in vulner-
ability to fatigue). Muraven et al. (1999), Oaten et al. (2003),
and Oaten and Cheng (2005b) found evidence for increased
stamina. The self-regulatory training appears to make people
less vulnerable to the effects of resource depletion.
THIS RESEARCH
In this study, we examined how students fared in the exami-
nation period after they had been partaking in a regular study
EFTA01113831
IMPROVED SELF-CONTROL 3
TABLE 1
Timeline for Study Program
Thu Semester I
Baseline Exams Semester Break Semester 2
Control Baseline Control Follow-up Baseline Exams
Cohort
Cohort 2 SP
WL SP
WL C C SP SP
Note. SP = intervention phase (study program): WL = no-intervention phase (waiting list control): C = control phase (non-stressful testing sessions.
program. In the experimental design, two cohorts partici-
pated in the study intervention program (Table () at different
times of the academic year. Cohort 1 entered the study inter-
vention program directly; they were tested twice across Se-
mester 1 (baseline, exams). Cohort 2 was tested across a time
span that included parallel testing sessions to Cohort I during
Semester 1 (waiting-list control). Cohort 2 then entered a
control phase that included two assessments of self-regula-
tory behavior (baseline, follow-up) during the semester
break, which provided a neutral period of academic demand.
The control phase tests whether any obtained findings were
the result of repeated testing and provides measures of retest
reliability. Finally, Cohort 2 entered the study intervention
program in Semester 2.
Cigarette smoking, alcohol consumption, and caffeine
consumption are some of the behaviors included in this
study. Cigarettes, alcohol, and caffeine are the most widely
used psychoactive substances in the world (Nehlig, 1999).
Despite differing levels of social acceptability, these behav-
iors are all considered addictive (Stepney, 1996) and there-
fore require some level of regulatory management
(Mumford, Neill, & Holtzman, 1988). The other regulatory
behaviors of interest are diet, physical activity, self-care hab-
its such as household chores, emotional control, study habits,
spending habits, and time management. If managing the
stress of examinations does deplete regulatory resources, and
the repeated practice of self-control does improve regulatory
capacity, then we would expect (a) maintenance or even im-
provement in regulatory behavior at exam time for those peo-
ple participating in the intervention phase (study program),
(b) impairment in regulatory behavior for those people in the
no-intervention phase (waiting-list control) during exam
time, and (c) no change in regulatory behavior across the
control phase (nonstressful testing sessions).
We were also interested in finding out whether academic
stress affects self-control performance on a standard labora-
tory task. We used visual tracking under distraction, which
requires participants to perform a computerized VTT while
a distracter video is played simultaneously at a loud vol-
ume. The VTT requires participants to track the movement
of multiple independent targets displayed on a computer
monitor (Pylyshyn & Storm, 1988; Scholl, Pylyshyn, &
Feldman, 2001). The participant must ignore the distracter
video content and attend only to the VIT. In a recent set of
studies, VTT performance deteriorated only when follow-ing tasks that required some form of regulatory exertion—
in particular, a thought-regulation task (Oaten & Cheng,
2005b) or emotion regulation (Oaten, Chau, & Cheng,
2005)—but was unaffected when following tasks that did
not require self-control (watching humorous videos; Oaten
et al., 2005). Thus, this task is sensitive to depletion manip-
ulations but not to nondepleting intervening tasks. In this
study, we administered the VTT twice at each session, and
in between V11' testings, participants were told to control
their thoughts by not thinking about a white bear. This is a
standard manipulation of regulatory depletion used in past
research (Muraven et al., 1998). Our (Oaten & Cheng,
2005b) previous research has found that performance on
the VTT is highly sensitive to an intervening thought-sup-
pression task, performance being worse after 5 min of
thought suppression. A program of regular physical exer-
cise, however, alleviated the adverse effect of the
thought-suppression task on the VTT. We were therefore
interested in finding out whether a study intervention pro-
gram would have similar effects. We predicted similar per-
formance on the VTT before thought suppression in all
conditions. After thought suppression, however, perfor-
mance on the V11' should be most impaired in participants
tested at exam time without intervention (waiting-list con-
trol), next most impaired in participants tested during
nonstressful times (control), and least impaired in partici-
pants who had partaken the study intervention program
(study program).
METHOD
Participants
A total of 45 Macquarie University undergraduates (7 men
and 38 women) recruited from introductory psychology
courses participated in return for partial fulfillment of a
course requirement. The age of participants ranged from 18
to 51 years, with a mean age of 23 years.
We randomly assigned participants to one of two cohorts
(Cohorts 1 and 2). Cohort 1 = 28; 4 men and 24 women)
entered the study intervention phase directly and was indi-
vidually tested in 2- to 30-min sessions separated by 8-week
interim periods. Cohort 2 = 17; 3 men and 14 women) first
entered the no-intervention phase (wait-list control) and then
EFTA01113832
4 OATEN AND CHENG
provided general controls (control phase) before proceeding
to the study intervention phase and were individually tested
in 6- to 30-min sessions separated by 8-week interim periods.
Design
Table 1 shows the schedule of testing for the two cohorts. Co-
hort 1 entered the intervention phase (study program) di-
rectly. We obtained baseline measures for Cohort I in Week 5
of Semester 1, the commencement of the study program, and
then again during the exam period for that semester. Cohort 2
entered the no-intervention phase (waiting-list control) in Se-
mester I with no study program. Parallel to Cohort 1, we ob-
tained baseline measures for Cohort 2 in Week 5 and then
again during the exam period. Cohort 2 entered the interven-
tion phase (study program) in Semester 2. We again obtained
baseline measures in Week 5, at the commencement of the
study program, and then during the exam period. Cohort 2
also provided general controls by participating in two testing
sessions (baseline, follow-up) occurring during nonstressful
times. This served as within-subjects and between-subject
control for the effects of the study program. All testing ses-
sions were uniform. Experimental sessions were separated
by 8-week periods.
We tailored study programs to suit each participant's stu-
dent workload and included the provision of a study register
(log of hours spent studying, which was submitted to us in
testing sessions), study diary (which was also submitted in
experimental sessions), artificial early deadlines, and a study
schedule for the examination period. We give more details
following.
We analyzed each experimental phase (intervention,
no-intervention, and control) separately using a more conser-
vative alpha value of .01 for all statistical tests due to re-
peated analysis of the same participants.
Study Program
Participants were instructed to bring both their student time-
table (i.e., a schedule of class contact hours) and assessment
timetable (due dates for coursework assessments) to the ini-
tial testing session. We discussed with the participants any
work commitments that needed to be incorporated into the
study program.
Artificial early deadlines. Self-imposed deadlines are
a popular strategy used by many in attempts to curb procrasti-
nation (Tice & Baumeister, 1997). In fact, recent research
suggested that external deadlines are more effective than
self-imposed deadlines in boosting task performance (Ariely
& Wertenbroch, 2002). We therefore imposed early artificial
deadlines on participants' assessment schedules. The artifi-
cial deadlines required the breaking down of the distant goal
into several proximal, specific, clear, achievable goals, thus making participants aware of their own concrete progress,
which was required to maintain their long-term engagement
with the program (Schunk, 1995; Zimmerman, 1989).
Study schedule. The study schedule provided a tem-
poral plan for studying in the lead up to examinations. The
study schedule specified all of the available dates and times
during that specific semester (taking into consideration uni-
versity contact hours and any specified work commitments),
along with a "suggested" study task designated to a specific
date(s). We administered the study schedule so as to enable
participants to detect and react to any discrepancies resulting
from the comparison of their current level of study and final
study goal state over the course of the semester. Students
were expected to (a) gradually increase awareness to these
suggested versus enacted discrepancies and (b) learn to mod-
ify their behaviors so as to reduce incongruities, thus enhanc-
ing self-regulation and improving performance.
Study register and study diary. These tools provided
opportunities for students to monitor themselves and to gen-
erate the feedback necessary for self-regulation. Self-moni-
toring refers to the activities involved in observing and re-
cording one's own behavior (Mace, Belfiore, & Shea, 1989).
Feedback is generated by a perceived discrepancy between
the outcome state (in this case, the study goal) and the current
state regarding the task. This feedback fosters attempts to re-
duce any disparity by changing plans, tactics, or strategies;
modifying aspects of their goals; or even abandoning the task
(Ruder & Winne, 1995). Participants' utilization of these
tools was expected to reveal their planning process and their
awareness of various cues while monitoring.
Manipulation Checks
We employed the study register and study diaries as manipu-
lation checks to ensure that participants were adhering to the
study program.
Study register. Average study time was assessed by
having participants complete a study register (a log of the
time spent studying) throughout the no-intervention (wait-
ing-list control) and intervention (study program) phases. For
analyses, study time was defined as the total number of hours,
on average, that participants studied per week.
Study diaries. To assess ease of uptake and mainte-
nance of the study program, we employed the use of study di-
aries. As part of their diary logs, participants were asked the
following questions: "What level of difficulty, if any, have
you experienced complying with the program?"; "Do you
feel your study habits are improving with the program'?"; and
"Do you wish to comment on the program generally?". Par-
ticipants were instructed to record their progress in the dia-
EFTA01113833
IMPROVED SELF-CONTROL 5
ries provided and to return them to the experimenter at each
experimental session.
Psychosocial Self-Reports
The General Health Questionnaire (GHQ; Goldberg,
1972). We assessed emotional distress in all sessions using
the 28-item version of the GHQ. This measure assesses
symptoms of emotional distress in four areas: anxiety/insom-
nia, somatic symptoms, social and cognitive dysfunction,
and depression. The questionnaire referred to respondents'
experiences over the past week and was coded using a
method that assigns weights of 0, 1, 2 and 3 to each answer
option. The GHQ has a high degree of internal consistency,
with a reported Cronbach alpha of .87, and retest reliability
was reported as .88 (Goldberg, 1972).
Perceived Stress Scale (ASS; Cohen, Kamarck, &
Mermelstein, 1983). We measured perceived stress in all
sessions using the 10-item version of the PSS. We used the
PSS to assess the degree to which situations in life are ap-
praised as stressful. Each item (e.g., "In the last week, how
often have you felt that things were going your way?") was
assessed on a 5-point scale ranging from 0 (never) to 4 (very
often), with higher scores indicating greater stress. The PSS
has been shown to be very useful to assess perceived stress,
with an overall Cronbach alpha of .87, and retest reliability
was reported as .85 (Cohen et al., 1983). This measure has
also been used in studies of academic examination stress
(Steptoe et al., 1996; Oaten & Cheng, 2005a).
General Self-Efficacy Scale (GSES; Jerusalem &
Schwarzer, 1992). We measured self-efficacy in all ses-
sions using the 10-item version of the GSES. Each item (e.g.,
"It is easy for me to stick to my aims and accomplish my
goals") was assessed on a 5-point scale ranging from 0 (not at
all tnte) to 4 (very true), with higher scores indicating higher
perceived self-efficacy. The scale has been used in numerous
research projects in which it has typically yielded internal
consistencies between a = .76 and .91. Its stability is satisfac-
tory, with retest reliability reported as .75 (Jerusalem &
Schwarzer, 1992).
Behavioral Self-Reports
We designed a questionnaire to assess cigarette smoking, al-
cohol and caffeine consumption, physical activity, dietary
habits, and other regulatory behavior. We administered the
questionnaire in both sessions. The test—retest reliability of
the questionnaire is reported in the Results.
Chemical consumption. We assessed cigarette smok-
ing, caffeine consumption, and alcohol consumption by the
use of open-ended questions presented in a questionnaire for-mat. We estimated current cigarette smoking as the number
of cigarettes smoked over the past 24 hr. We assessed current
alcohol consumption using a 7-day recall procedure in which
quantity of alcoholic beverage was recorded. We also as-
sessed caffeine consumption using a 7-day recall procedure,
with quantity being the measure of interest.
Dietary habits. We assessed dietary habits by ques-
tioning participants about food choice (e.g., "In the last
week, how successfully did you maintain a healthy diet?")
and dietary restraint (e.g., "In the last week, how often did
you eat junk food?) over the past week. Response sets were
recorded on a 5-point scale ranging from 0 (never) to 4 (more
than once per day). We derived 2 measures for analysis: junk
food and healthy eating.
Physical activity. We measured exercise by question-
ing participants about the frequency and duration of physical
activity sessions over the past week. Response sets were re-
corded on a 5-point scale ranging from 0 (never) to 4 (more
than once per day). We derived 2 measures for analysis: the
number of episodes of physical activity and the total duration
of physical activity sessions.
General regulatory behavior. We measured various
everyday behaviors that involve self-control (e.g., "In the last
week, how often did you go out with friends instead of study-
ing?"). We aimed to include those behaviors that do not serve
a stress-relieving function. We recorded response sets on a
5-point scale ranging from 0 (never) to 4 (more than once per
day). We derived nine measures for analysis: self-care habits
(laundry habits, leaving dishes in the sink), time management
(keeping appointments and procrastination), study habits
(spending time with friends instead of studying and watching
television instead of studying), spending habits (spending
without thinking and overspending), and emotional control
(loss of temper).
Visual Tracking Under Distraction
We gave a laboratory task of self-control twice in each test
session. Participants performed a VTI' while a distracter
video played at the same time in the forefront of the partici-
pant. We instructed the participant to ignore the distracter
video content and attend only to the Vff. The VTT requires
participants to visually track the movement of multiple tar-
gets displayed on a computer monitor (see Figure 1). The
distracter video included excerpts from a comedy routine by
Eddie Murphy (Murphy, Tieken, & Wachs, 1983). The use of
the VTT to assess self-regulatory capacity has been validated
in previous research (Oaten and Cheng, 2005b; Oaten, et al.,
2005), and we selected it for that reason.
Stimuli were displayed on an I-Mac* computer equipped
with a 15-in. monitor set to a resolution of 800 x 600 pixels
EFTA01113834
Step 1 6 OATEN AND CHENG
•
• • •
U
Step 2 NI MI IN II NI
Step 3
FIGURE 1 A representation of a visual tracking task experimental sequence. Participants view items on computer monitor. In the target identification
phase (Step I). six cubes appear on the screen. and three of them Hash briefly to indicate that they are the targets: then all squares move randomly (Step 2).
The task of the participant is to select the three targets once they have stopped moving by placing the cursor on them andclicking with the mouse (Step 3).
and a refresh rate of 95 Hz. Participants were seated 54 cm
away from the monitor. We controlled and measured the VT 1'
using Psyscript (Version 4; Bates & D'Oliviero, 2000). Each
V11' consisted of 16 trials. At the beginning of each trial, six
black squares (20 x 20 mm) were presented in a horizontal
line. After 2 sec, three target items were highlighted with
small blinking probes (disappearing and reappearing for five
flashes). Then all items moved in random trajectories for S
sec. After all of the objects stopped moving, the participant
had to indicate the three target items using the mouse. The fi-
nal mouse click caused the display to disappear, and the par-
ticipant initiated the next trial with a key press.
Forty-eight sets of trajectories (along with target selec-
tions) were generated and stored offline. Participants com-
pleted a practice trial for which the data were not collected
and then completed the experimental trials in a randomized
order (different for each participant).
Thought Suppression Task
Following the first assessment of self-regulatory perfor-
mance, we administered a thought suppression task to ma-
nipulate regulatory exertion. The procedure, developed by
Wegner, Schneider, Carter, and White (1987), requires the
participant not to think about a white bear. This task has been
used previously to manipulate self-regulatory depletion
(Muraven et al., 1999, 1998). We told participants that over
the course of the experiment, they would be asked to perform
a cognitive task (thought suppression). We instructed partici-
pants to write down all their thoughts on a piece of paper for S
min, one thought per line, so that we could "see how you use
words in naturally occurring sentences" (Muraven et al.,
1998). We then administered the experimental manipulation.
We instructed participants to list any thoughts that came to
mind with the caution that they should avoid thinking about a
white bear. We told participants that whenever they thought
of a white bear, they were to write that thought down. We em-
phasized that it was critical to change their thoughts immedi-
ately and to try not to think of a white bear again. Following
the thought suppression task, we recorded a follow-up mea-sure of self-regulatory performance by administering a sec-
ond VT1'.
Procedure
Testing procedure was uniform across sessions. Participants
first signed experimental consent forms and we then admin-
istered in order a VTT, the thought suppression task, and then
a second VT!'. We then obtained measures of emotional dis-
tress, perceived stress, perceived self-efficacy, and general
regulatory behaviors. We conducted data collection between
Tuesday and Friday of each week so that all smoking infor-
mation related to a weekday.
RESULTS
Overall, 9 (24%) women and 2 (28%) men smoked at some
point throughout the testing session; 17 (45%) women and 4
(57%) men consumed caffeine; and 21 (55%) women and 4
(57%) men consumed alcohol. The numbers that engaged in
regular physical activity included 32 (84%) women and 7
(100%) men. There was no significant difference between
genders in the proportions carrying out these behaviors and
no baseline differences between the exam-stress and control
groups. We restricted analyses of each behavior to those indi-
viduals who engaged in these activities rather than the entire
sample.
Manipulation Checks
Study register. The study register (log of hours spent
studying) indicated that participants did adhere to the study
program. Figure 2 summarizes the mean hours spent study-
ing. Cohort 2 was the only cohort to participate in the no-in-
tervention phase (waiting-list control) and was therefore the
only cohort included in the following analyses. The reported
average number of hours spent studying were entered into a
session (baseline, exams) repeated measures analysis of vari-
ance (ANOVA). The ANOVA showed no effect of session
EFTA01113835
IMPROVED SELF-CONTROL 7 Hours studying per week 25
20
5
10 -
5 -
0 Study Habits
study time
■baseline: no intervention CI CXIIIIIS: no intervention
lIbaseline: intervention Sextons: intervention
FIGURE 2 Reported average number of hours spent studying per
week (mean ± standard error) across the testing sessions.
across the no•intervention phase. Both cohorts participated
in the intervention phase (study program) and we included
them in the analyses. The reported average number of hours
spent studying were entered into a session (baseline, exams)
repeated measures ANOVA. The ANOVA found a significant
main effect for session, F(1, 44) = 24.58, p< .001. These re-
sults suggest that although on average, participants' spent 11
hr per week studying, study time increased to an average of
22 hr per week during the intervention phase (study pro-
gram).
Study diaries. All study diaries were returned to us as
instructed. An inspection of the diaries indicated that all par-
ticipants recorded progress on the study program as in-
structed. Accordingly, the diary content suggested a roughly
equal expenditure of effort from all participants.
Entries from the study diaries indicate that the study pro-
gram required ongoing regulatory effort. For example, some
participant comments include the following: "My studying is
improving but it is a constant struggle ... especially when ev-
eryone is watching TV ... I want to join them so bad"; "In or-
der to stick to the program I have to get out of bed an hour
earlier so I can get the study hours in ... some mornings it is
so hard to get up ... I'd much prefer to lie in"; and "Studying
at uni isn't so bad as everyone is pretty much doing the same
thing ... but when I get home and my flatmates are heading
out to the pub ... it is so hard not to go with them ... so far
I've managed to stay strong and stick to the planned study-
ing:' The comments suggest that the academic study pro-
gram required self-control.
Study Intervention Phase
V77: Figure 3 summarizes (striped bars) performance
on the VTT across the intervention phase (study program).
Both cohorts participated in the intervention phase and were
included in the analyses. The thought suppression task
caused deterioration in performance at baseline (depletion).
This effect of depletion, however, appeared to attenuate I 40
35
30
25
20
IS
10
S
0 Visual Tracking Task
baseline: no exams: no baseline: mama
intervention intervention intervention intervention
la pre Moil& suppression epees thought inippressien
FIGURE 3 Error rate on the visual tracking task (meant standard
error) measured before and after the thought suppression task across
sessions.
across sessions, with less depletion during the examination
period following participation in the study program. These
impressions were confirmed by a Session (Baseline, Exams)
x Time (before thought suppression vs. after thought sup-
pression) repeated measures ANOVA. With the ANOVA, we
found significant main effects for time, F(I, 44) = 2395.40, p
< .001, indicating a general tendency toward depletion fol-
lowing a previous self-regulatory act; a significant main ef-
fect for session, F(I, 44) = 79.96,p < .001, suggesting that vi-
sual tracking performance improved across sessions; and a
significant Time x Session interaction, F(1, 44) = 359.98, p<
.001. The pattern of results indicates that the study program
improved regulatory stamina, increasing resistance to the de-
bilitating effects of a manipulation of regulatory depletion (a
thought suppression task).
Behavioral sett-reports. Figures 4 through 10 (black
and striped bars) show the reported changes in regulatory be-
haviors across the intervention phase (study program). Both
cohorts participated in the intervention phase and were in-
cluded in the analyses. We entered the data in Figures 4
through 10 into a repeated measures ANOVA, with Session
(Baseline, Exams) as the within-subjects variable. We re-
stricted analyses of each behavior to those individuals who
engaged in these activities rather than the entire sample. Ta-
ble 2 summarizes the main effects of session.
As predicted, people seemed better able to control their
behavior during the exam period following the intervention
phase (study program). In fact, all of the behaviors showed
changes in the predicted direction. Figure 4 shows a reported
decrease in chemical consumption during examinations for
those people in the study program. Smoking decreased by a
mean of 7 cigarettes per day, caffeine consumption decreased
on average by 2 cups per week, and alcohol decreased on av-
erage by 2 drinks per week. Figure 5 shows changes in di-
etary trends across sessions. Dietary patterns improved for
those participants in the study program, with decreased junk
EFTA01113836
2 Consumption Patterns
alcohol cigarettes caffeine
Ohneline: no Lawn...ono° Denims! no tatelvention
Obasebne: intervention intenention
FIGURE 4 Number of cigarettes (over 24 hr). cups of caffeine.
and standard units of alcohol (over 7 days) across sessions (mean ±
standard error). We restricted analyses of each behavior to those indi-
viduals who engaged in these activities rather than the entiresample.
1. 4
.2 2 3
° 2
TiDietary Intake
healthy eating junkfood
bkiehne: DO annum. 0 MANS Intenential
Inbasdinc: intervention Meurer interment
FIGURE 5 Dietary intake across sessions (mean ± standard error).
Frequency of behaviors were coded as follows: 0 = never I = once
per week:2 = 2 to 3 times per week: 3 =daily:4 = more than once per
day.
Physical Activity
exercise duration
Obaselme: no intervention Cl CLAMS: no intervention
• baseline: intervention •cuens: intervention
FIGURE 6 Frequency and duration of physical activity acrass ses-
sions (mean ± standard error). Frequency of behaviors were coded as
follows: 0= never: I = once per week: 2 =2 to 3 times per week: 3 =
daily: 4 = more than once per day. Self-can Habib
leaving dishes leafing laundry
lebaseline no nen ration Oceans no inienen000
•buekne ententotieo •e :Nis': unentonon
FIGURE 7 Self-care habits across sessions (mean ± standard er-
ror). Frequency of behaviors were coded as follows: 0 = never: I =
once per week: 2 = 2 to 3 times per week: 3 = daily: 4 = more than
once per day.
8 Study Habits
tv instead of study friends instead of study
Obaschse. no MIcrveatina Delinens. no intenention
▪ internxilien Monson: intervention
FIGURE 8 Study habits across sessions (mean 3 standard error).
Frequency of behaviors were coded as follows: 0 = never: I = once
per week: 2 =2 to 3 times per week: 3 =daily:4 = more than once per
day.
Impulse Control
impulse over-spending
Welding emotional
ontrol
in basdiiic no interretti00 0 Wens: no inlervential
II bother isavennoti Omani. intervention
FIGURE 9 Impulse control across sessions (mean ± standard er-
ror). Frequency of behaviors were coded as follows: 0 = never: I =
once per week: 2 = 2 to 3 times per week: 3 = daily: 4 = more than
once per day.
8
EFTA01113837
4
Iv 3
2
1
0 time NIonagement
procrastination missing appointments
libawItne no intern:mon Clautor m i;oinnntio• I
IIII bowline IIIICIN(11001. ffilexamg Savant.
FIGURE 10 Time management across sessions (meant tandard
error). Frequency of behaviors were coded as follows: 0 = never. 1=
once per week: 2 = 2 to 3 times per week: 3 = daily: 4 = more than
once per day.
food consumption and an increase in healthy eating habits
during the examination period. Figure 6 shows the same pat-
tern for physical activity. During the exam period, the fre-
quency and duration of physical activity increased for those
participants in the study program.
Figures 7 through 10 show improvements in general regu-
latory habits in the lead up to examinations. Following inter-
vention, participants reported an increase in attendance to
household chores (leaving the dishes in the sink less often
and doing the laundry more often), emotional control, and a
decrease in impulse spending, overspending, watching tele-
vision instead of studying, spending time with friends instead
of studying, failures to attend to commitments, and procrasti-
nation.
No-Intervention Phase (Wading-List Control)
VT7: Figure 3 (black bars) summarizes performance on
the VTT across the no-intervention phase (waiting-list con-
trol). Cohort 2 was the only cohort to participate in the no-in-
tervention phase and was therefore the only cohort included
in the following analyses. The thought suppression task
caused deterioration in performance at baseline (depletion).
This effect of depletion, however, appeared to worsen at
exam time for those not participating in the study program.
These impressions were confirmed by a Session (Baseline,
Exams) x Time (before thought suppression vs. after thought
suppression) repeated measures ANOVA. With the ANOVA,
we found significant main effects for time, F(I, 16) =
3136.52, p < .001 and session, F(l, 16) = 155.82, p < .001,
this time suggesting that visual tracking performance wors-
ened across sessions and importantly, a significant Time x
Session interaction, F(I, 16) = 252.12, p < .001. The pattern
of results indicates that participants not in the study program
were more vulnerable to the debilitating effects of a manipu-
lation of regulatory depletion (a thought suppression task)
during the examination period. IMPROVED SELF-CONTROL
TABLE 2
Regulatory Behavior: Intervention Phase
(Study Program) 9
Behavior df F
Consumption
Cigarettes' 1. 10 135.87 <.001
AlcohoP 1.24 28.47 <.001
Caffeine 1.20 43.33 < .001
Physical activity
Frequency" 1.38 67.86 <.001
Duration' 68.14 <.001
Diet(
Junk food 1.44 103.53 <.001
Healthy habits 78.22 <.001
Self-cam habitst
Leaving dishes in sink 1.44 29.75 <.001
Leaving laundry 29.33 <.001
General regulator,
TV instead of study 1.44 47.43 <.001
Friends instead of study 47.42 <.001
Impulse spending 45.34 < .001
Overspending 70.82 < .001
Emotional control 57.00 < .001
Procrastination 43.90 < .001
Missing appointments 47.42 < .001
Note. Analyses restricted to participants who engaged in these behav-
iors.
art=11.bn=25. 4tt=21."n=39."n=39. 1N=45.
We also compared the no-intervention phase (wait-
ing-list control) with the intervention phase (study pro-
gram) across cohorts. The two cohorts were compared at
the same time of year; they were randomly assigned to con-
ditions (see Table 1). We conducted a mixed analysis with
session and time serving as within-subjects variables and
cohort as the between-subject variable. In the ANOVA, we
compared Cohorts I and 2, with Session (Baseline, Exams)
x lime (before thought suppression vs. after thought sup-
pression) x Cohort (Cohort I [intervention phase] vs. Co-
hort 2 ]no-intervention phase]) as factors. The ANOVA
found a significant main effect for time, F(l, 43) =
5016.22, p < .001. There was also a significant Time x Co-
hort interaction, F(1, 43) = 295.56, p < .00I, indicating that
the rates of depletion differed across the cohorts; a signifi-
cant Session x Cohort interaction, F(1, 43) = 110.30, p <
.001, indicating that overall visual tracking performance
differed across groups; and a significant Session x Time x
Cohort interaction, F(I, 43) = 406.64, p < .001. These find-
ings suggest that during the examination period, partici-
pants in the intervention phase (study program) showed a
pattern of performance consistent with improved stamina,
whereas participants not in the study program appeared
more susceptible to the depleting effects of a prior regula-
tory exertion (a thought suppression task).
Behavioral self-reports. Figures 4 through 10 (grey
and white bars) show the reported changes in regulatory be-
EFTA01113838
10 OATEN AND CHENG
haviors across the no-intervention phase (waiting-list con-
trol). Cohort 2 was the only cohort to participate in the no-in-
tervention phase and was therefore the only cohort included
in the following analyses. We entered the data in Figures 4
through 10 into a repeated measures ANOVA, with session as
the within-subjects factor. Table 3 summarizes the main ef-
fects of session.
As predicted, people not in the study program (no-inter-
vention phase) appeared less able to control their regulatory
behavior during the examination period. In fact, all of the re-
ported behaviors show changes in the predicted direction.
Figure 4 shows a reported increase in cigarette smoking, caf-
feine, and alcohol consumption during the examination pe-
riod for those people not participating in the study program.
Cigarettes increased at a mean rate of 13 cigarettes per day,
caffeine consumption increased at a mean rate of 4 cups per
week, and alcohol increased at a mean rate of 4 drinks per
week.
Figure 5 shows changes in dietary trends across the no-in-
tervention phase, with a reported increase in junk food in-
take, and a decrease in healthy eating habits. Figure 6 shows a
similar pattern for physical activity, with the reported fre-
quency and duration of physical activity of participants not in
the study program decreasing during the examinations.
Figures 7 through 10 show deficits in general regulatory
habits for those not in the study program in the lead up to ex-
aminations. Participants reported a decrease in household
chores (laundry, leaving the dishes in the sink) and emotional
control and an increase in spending without thinking, over-
TABLE 3
Regulatory Behavior: No•Intervention Phase
(Waiting•List Control)
Behavior df F p
Consumption
Cigarettes• 1.5 106.50 <.001
Alcohol') 1. I 0 19.55 <.031
Caffeine•' 1.8 23.10 <.031
Physical activity°
Frequency 1. 16 35.86 <.00I
Duration 12.78 <.031
Diet°
Junk food I. 16 47.80 <.001
Healthy habits 27.20 <.001
Self-care habits,'
Leaving dishes in sink 1.16 13.19 <.031
Leaving laundry 13.18 <.031
General regulatory behavior°
TV instead of study 19.43
Friends instead of study 19.42 <.001
Impulse spending 16.10 <.001
Overspending 12.24 <.001
Emotional control 34.00 <.001
Procrastination 19.43 <.001
Missing appointments 15.61 <.001
Note. Analyses restricted to participants who engaged in these behav-
iors.
an= 6.6n= 11.0n= 9. dN= 17. spending, spending time with friends instead of studying,
watching television instead of study, failures to attend to
commitments, and procrastination.
As with the VTT, we conducted mixed ANOVAs to com-
pare the no-intervention phase (waiting-list control) with the
intervention phase (study program) within a single statistical
test. Again, the two cohorts were randomly assigned and
compared at the same points in the semester (see Table 1).
We entered each dependant variable in Figures 4 through 10
into the following analyses: a Session (Baseline, Exams) x
Cohort (Cohort 1 [intervention phase] vs. Cohort 2 [no-inter-
vention phase]) repeated measures ANOVA. Table 4 summa-
rizes the inferential statistics. Consistent with the within-sub-
jects analyses, significant Cohort x Session interactions
indicate that during the examination period, self-regulation
in all variables improved for those participants in the inter-
TABLE 4
Regulatory Behavior: Cohort 1 (Intervention
Phase) Versus Cohort 2 (No•Intervention Phase)
Behavior df
Consumption
Cigarette0 1.9 16.03 <.001
x Cohort 145.24 <.001
Alcohol') 1.23 5.95 <.001
x Cohort 37.97 <.001
Caffeine, 119 4.23 <.001
x Cohort 45.58 <.001
Physical activity
Frequency° 1.37 6.46 <.001
x Cohort 60.28 <.001
Duration° 12.04 <.001
x Cohort 44.32 <.001
Diet'
Junk food 1.43 7.22 <.001
x Cohort 131.65 <.001
Healthy habits 9.43 <.001
x Cohort 131.64 <.001
Self-care habits'
Leaving dishes in sink 1.43 6.29 <.001
x Cohort 31.01 <.001
Doing laundry 6.32 <.001
x Cohort <.001
General regulatory behavior
TV instead of study 1.43 5.12 .010
x Cohort 4155 <.001
Friends instead of study 11.12 <.001
x Cohort 4155 <.001
Impulse spending 11.47 <.001
x Cohort 47.18 <.001
Overspending 12.82 <.001
x Cohort 41.75 <.001
Emotional control 4.67 .030
x Cohort 4652 <.001
Procrastination 15.25 <.001
x Cohort 15.25 <.001
Missing appointments 12.97 <.001
x Cohort 15.29 <.001
Note. Analyses restricted to participants who engaged in these behav-
iors.
an =II. bpi= 25. to =21. tin= 39. 'N = 45.
EFTA01113839
IMPROVED SELF-CONTROL 11
TABLE 5
Regulatory Behavior Control Phase Mean
and Standard Error
Behavior Baseline Follow-Up
R df SE M SE
Emotional responses'
Perceived stress scale 19.1 0.5 19.2 0.8 .98°
General health questionnaire 18.8 0.4 18.6 0.3 .94.
General self-efficacy scale 193 0.7 19.3 0.8 .96°
Consumptionb
Cigarettes b 3.1 1.0 3.0 10 .96°
Alcohol, 2.1 0.6 2.1 0.6 .94.
Caffeined 6.0 0.5 6.2 05 .91*
Physical activity°
Frequency 21 0.3 2.3 0.2 .97*
Duration 1.2 0.3 1.2 0.2 .97*
Diet°
Junk food 14.2 0.2 14.2 0.2 .97*
Healthy habits 3.2 0.3 3.2 0.2 .97*
Self-care habits'
Leaving dishes in sink 2.9 0.3 2.9 03 .92*
Leaving laundry 2.8 0.3 2.7 0.2 .93*
General regulatory*
TV instead of study 2.7 0.3 2.7 0.2 .96°
Friends instead of study 2.6 0.3 2.7 0.2 .96°
Impulse spending 2.4 0.2 2.5 0.3 .90*
Overspending 2.4 0.3 2.5 03 .80*
Emotional control 2.9 0.2 2.9 0.2 .97.
Procrastination 2.6 0.3 2.7 0.2 .98*
Missing appointments 23 0.3 2.5 0.3 .94•
£N= 17. ha =6. = 11. 4=9.
'p is significant at the .01 level, two-tailed.
vention phase (study program), whereas regulatory behavior
worsened for those participants not in the study program
(no-intervention phase).
Control Phase
Cohort 2 was the only cohort to participate in the control
phase (testing during nonstressful times) and was therefore
the only cohort included in the following analyses. Table 5
reports the regulatory behavior (mean ± standard error) dur-
ing the control phase. There were no significant effects for
any of the regulatory behaviors (laboratory or self-reported)
across the control sessions, indicating that regulatory behav-
ior remained stable during the control phase (Table 5).
Test—retest reliability of the general regulatory question-
naire was calculated using the Pearson correlation coefficient
by correlating Session I (baseline) scores with Session 2
(follow-up) scores from the control phase. Retest reliabilities
(Table 5) were generally high, with all but one at .90 or better.
Relation Between VTT
and Behavioral Self-Reports
We tested whether the degree of change in VTT performance
across the intervention phase (study program) predicted: (a) TABLE 8
Relationship Between VTT
and Behavioral Self-Reports
Behavior Difference VTT Differrnee
Consumption
Cigarettes'
Alcohols
Caffeine'
Physical activity
Frequency'
Duration'
Mete
Junk food
Healthy habits
Self-care habits difference
Leaving dishes in sink
Doing laundry
Regulatoryt
TV instead of study
Friends instead of study
Impulse spending