Daily and Momentary Mood and Stress Are Associated With Binge Eating and Vomiting in Bulimia Nervosa Patients in the Natural Environment

Abstract

The relation of mood and stress to binge eating and vomiting in the natural environments of patients with bulimia nervosa (BN) was examined using real-time data collection. Women (n = 131; mean age = 25.3 years) with BN carried a palmtop computer for 2 weeks and completed ratings of positive affect (PA), negative affect (NA), anger/hostility (AH), and stress (STRS); they also indicated binge or vomit episodes (BN-events) 6 times each day. Mixed models were used to compare mood and STRS between and within days when BN-events occurred. Between-days analyses indicated that binge and vomit days both showed less PA, higher NA, higher AH, and greater STRS than days with no BN-events. Within-day, decreasing PA, and increasing NA and AH, reliably preceded BN-events. Conversely, PA increased, and NA and AH decreased following BN-events. Demonstration of the temporal sequencing of affect, STRS, and BN-events with a large BN sample may help advance theory and clinical practice, and supports the view that binge and purge events hold negatively reinforcing properties for women with BN.

Full text

Daily and Momentary Mood and Stress Are Associated With Binge Eating
and Vomiting in Bulimia Nervosa Patients in the Natural Environment
Joshua M. Smyth
Syracuse University
Stephen A. Wonderlich
Neuropsychiatric Research Institute and
University of North Dakota
Kristin E. Heron and Martin J. Sliwinski
Syracuse University
Ross D. Crosby, James E. Mitchell, and
Scott G. Engel
Neuropsychiatric Research Institute and
University of North Dakota
The relation of mood and stress to binge eating and vomiting in the natural environments of patients with
bulimia nervosa (BN) was examined using real-time data collection. Women (n ! 131; mean age ! 25.3
years) with BN carried a palmtop computer for 2 weeks and completed ratings of positive affect (PA),
negative affect (NA), anger/hostility (AH), and stress (STRS); they also indicated binge or vomit
episodes (BN-events) 6 times each day. Mixed models were used to compare mood and STRS between
and within days when BN-events occurred. Between-days analyses indicated that binge and vomit days
both showed less PA, higher NA, higher AH, and greater STRS than days with no BN-events.
Within-day, decreasing PA, and increasing NA and AH, reliably preceded BN-events. Conversely, PA
increased, and NA and AH decreased following BN-events. Demonstration of the temporal sequencing
of affect, STRS, and BN-events with a large BN sample may help advance theory and clinical practice,
and supports the view that binge and purge events hold negatively reinforcing properties for women with
BN.
Keywords: bulimia nervosa, ecological momentary assessment (EMA), mood, stress, antecedent
Clinicians working with individuals with bulimia nervosa (BN)
have sought to explain bulimic behavior by focusing on antecedent
events of binge eating and purging. Antecedents are cognitive,
behavioral, or emotional events or conditions that precede a be-
havior. Although numerous potential antecedents have been stud-
ied (e.g., environmental stressors, dietary restriction, rules about
food, overconcern with body shape or weight), a growing body of
evidence has suggested that negative mood changes may be sig-
nificant antecedents of disordered eating behaviors.
Negative emotional states have been associated with binge
eating episodes in controlled experiments in laboratory settings,
and this relationship holds for individuals with binge eating dis-
order and women with noneating disorders (Agras & Telch, 1998;
Telch & Agras, 1996). One of the first studies to establish this
relationship between negative mood and disordered eating behav-
iors using an ecologically valid approach was conducted by John-
son and Larson (1982). Fifteen women with BN were compared
with 24 control women, each of whom completed daily diaries for
1 week. Recordings were made 7 times each day in response to
signals from a wrist-watch-style beeper between 8:00 a.m. and
10:00 p.m. They found that both binge and purge events were
preceded by reports of negative mood in a sample of individuals
with BN. More generally, they also found that women with BN
had more dysphoric and more variable mood states.
Additional studies have since found that individuals with BN
report significantly more negative mood in general and prior to a
binge episode than normal controls (e.g., Kjelsas, Borsting, &
Gudde, 2004; Lingswiler, Crowther, & Stephens, 1989; Waters,
Hill, & Waller, 2001). In a recent study, a group of 27 college
women with subclinical levels of binge eating behaviors reported
their mood and binge eating episodes seven times daily using
handheld computers. Results suggest that binge days were marked
by dysphoric mood, but there was no indication that negative mood
was an immediate antecedent to the binge (Wegner et al., 2002).
Interpretation of the lack of an association between negative mood
and binge eating in this study, however, is limited by the small
sample size, college-student sample, and subclinical nature of the
disordered eating. Schlundt, Johnson, and Jarrell (1985) found that
in a small sample of women with BN that (a) mood before eating,
(b) the time of day that the meal occurred, (c) its social and
physical context, and (d) the activities engaged in prior to eating all
predicted the occurrence of self-induced vomiting. Overall, there is
Joshua M. Smyth, Kristin E. Heron, and Martin J. Sliwinski, Department
of Psychology, Syracuse University; Stephen A. Wonderlich, Ross D.
Crosby, James E. Mitchell, and Scott G. Engel, Neuropsychiatric Research
Institute, Fargo, North Dakota, and Department of Neuroscience, Univer-
sity of North Dakota.
This work was supported by National Institutes of Health Grant R01
MH59674.
Correspondence concerning this article should be addressed to Joshua
M. Smyth, Department of Psychology, 430 Huntington Hall, Syracuse
University, Syracuse, NY 13244-2340. E-mail: Jmsmyth@syr.edu
Journal of Consulting and Clinical Psychology Copyright 2007 by the American Psychological Association
2007, Vol. 75, No. 4, 629 – 638 0022-006X/07/$12.00 DOI: 10.1037/0022-006X.75.4.629
629

evidence that negative mood may precede binge eating and purg-
ing behaviors.
There is also some evidence that disordered eating behaviors
may be preceded by stress, with evidence linking both objective
and subjective indicators of stress to disordered eating behavior.
Individuals with disordered eating reported an increased desire to
binge following laboratory stressors than control participants (Cat-
tanach, Malley, & Rodin, 1988). Anecdotal and clinical reports
suggest that a variety of stressors (e.g., death of a family member,
ending of a romantic relationship) frequently precipitate episodes
of disordered eating behavior (e.g., Crowther & Sherwood, 1997).
In a population of binge-eating college women who kept daily
diaries recording affect and stress for 30 days, higher levels of
stress were associated with binge eating, regardless of ratings of
depressed affect (Yacono Freeman & Gil, 2004). Individuals with
BN have also reported significantly more stress prior to a binge
episode than individuals with binge eating disorder or normal
controls (Lingswiler et al., 1989). Finally, emerging evidence
suggests that anger, especially state anger, may be a particularly
relevant affect in eating disorders (Meyer et al., 2005; Milligan &
Waller, 2000; Waller et al., 2003).
Much of the previous data is limited to retrospective reports,
which is problematic as this methodology requires individuals to
recall transient mood states and stressors over extended periods of
times. This leaves the self-report data susceptible to memory
biases (e.g., effort after meaning, retroactive reconstruction; see
Smyth et al., 2001). There is some evidence that retrospective
reporting of binge events may be especially inaccurate (Bardone,
Krahn, Goodman, & Searles, 2000). New, often technology-based,
assessment methodologies allow research designs that are more
ecologically valid and can be successfully utilized to generate
results that more accurately reflect everyday life situations (as
opposed to laboratory or recall-based designs).
In the present study, we sought to test several specific predic-
tions regarding mood and stress reports associated with binge or
vomit episodes (BN-events) for women with clinical levels of BN.
An ecological momentary assessment (EMA) technique was used
to limit retrospective recall and to obtain a more ecologically valid
representation of the relationship between moods, stress, and dis-
ordered eating events. This method of data collection allowed us to
test several different aspects of the association between antecedent
and BN-events. First, we compared mood and stress reports on
days when BN-events occurred versus days when no BN-events
occurred. Extending previous literature by using momentary mood
and stress reports, we expected that participants would report more
negative affect (NA), anger/ hostility (AH), and greater stress
(STRS), but less positive affect (PA), on days when a BN-event
occurred compared with days with no BN-events. Second, on days
when a BN-event was reported, we examined the mood and stress
trajectories prior to and following the first BN-event of the day.
We predicted that negative mood, AH, and STRS reports would
increase immediately prior to a binge or vomit, whereas PA would
decrease over this interval. Conversely, following a binge or
vomit, we expected negative mood, positive mood, and STRS to
each show recovery. Post-BN-event anger was examined in a more
exploratory fashion, as there are reasons to believe it may increase
(self-focused anger, self-blame) as well as decrease (release from
negative state). Third, we realized that significant trajectories in
mood and stress ratings could be due to BN-event concurrent
reports and not (for example) a true steady increase in dysphoria
prior to a binge. We looked at mood and stress ratings within
several hours of the BN-event (pre and post), but we excluded
binge concurrent ratings (within 10 min of the event) to address
this issue. We expected that the NA, PA, AH, and STRS trajec-
tories would remain associated with BN-events even after remov-
ing event-concurrent ratings.
Method
Participants
Female participants (n ! 131) who met Diagnostic and Statis-
tical Manual of Mental Disorders (4th ed.; DSM–IV; American
Psychiatric Association, 1994) criteria for BN were recruited from
community and campus populations to participate in the present
study. The 131 participants were selected from an initial group of
154 women who were invited to complete a more rigorous BN
assessment based on a phone screening process. Doctoral-level
researchers trained on interviewing with the Structured Clinical
Interview for DSM–IV Axis I Disorders (SCID-I/P; First, Spitzer,
Gibbon, & Williams, 1995) screened potentially eligible women
via telephone for inclusion (i.e., met DSM–IV criteria for BN) and
exclusion (e.g., excluded if male, younger than 18 years, having a
current psychotic disorder, or unable to read) criteria. On the basis
of this more thorough assessment, 11 participants failed to meet
inclusion criteria, leaving 143 participants who began the EMA
protocol. Seven participants dropped out of the EMA protocol
before completion, and 3 participants provided incomplete data on
the EMA, resulting in 133 individuals who completed the EMA
protocol. Two of these individuals provided incomplete informa-
tion on additional questionnaires used in this study and were
subsequently excluded from analyses, resulting in a total of 131
participants.
Participants were mostly single/never married (85.0%), Cauca-
sian (96.9%), full-time students (69.0%), relatively young (M !
25.3 years, SD ! 7.6; range ! 18 –55), well educated (all but 1
completed high school; 60% with at least some college), and had
a mean body mass index of 23.2 (SD ! 4.9; range ! 16.0 – 47.2).
Regarding historical and/or comorbid psychiatric diagnoses as-
sessed during the structured interview, 26.7% of participants re-
ported being an inpatient for psychiatric problems, with 60.8%
reporting outpatient treatment for psychiatric problems at some
point during their life. A majority (87%) of participants had a
lifetime mood disorder diagnosis, with 54.2% meeting all criteria
for this diagnosis within the last month. Similarly, 36.6% reported
a lifetime diagnosis of substance dependence (16% met criteria
within the last month), and 59.5% had a lifetime diagnosis of
anxiety disorder (51.1% met criteria within the last month). There
were no differ ences found whe n comparing participants with
women who dropped out or did not complete required study
materials in demographics, EDE subscale scores, or SCID-I/P
comorbid diagnoses with one exception. Noncompleters were mar-
ginally ( p ! .052) less likely to have current mood disorders (20%
noncompleters; 53.3% completers). Women who agreed to partic-
ipate received $200 (for the 2 weeks of EMA) and were given a
$50 bonus for compliance rates of 85% or better on the daily
assessments.
630
SMYTH ET AL.

EMA Assessments
Participants completed daily diaries on palmtop computers that
included questions regarding mood, affect, stress, and behavior.
All data collected on the palm were automatically date and time
stamped at the start and end of every data entry episode. The
following measures were completed at each assessment.
Profile of Mood States (Lorr & McNair, 1971). The four-item
Anger/Hostility subscale of the Profile of Mood States was used to
assess momentary anger/hostility. This measure has been used
previously to measure daily mood (Bolger & Zuckerman, 1995).
The coefficient alpha in the current study was .89.
Positive and Negative Affect Scale (Watson, Clark, & Tellegen,
1988). We selected items for use in momentary reports from the
Positive and Negative Affect Scale on the basis of high factor
loadings to ensure spread across high and low arousal, and those
that were thought to be clinically and/or theoretically fruitful on
the basis of previous work with EMA assessment. The items
included 13 PA items (happy, alert, proud, cheerful, enthusiastic,
confident, concentrating, energetic, calm, strong, determined, at-
tentive, relaxed) and 11 NA items (afraid, lonely, irritable,
ashamed, disgusted, nervous, dissatisfied with self, jittery, sad,
distressed, angry with self). Good reliability has been reported for
this measure with alpha coefficients of .85 for NA and .87 for PA
(Watson et al., 1988). Coefficient alphas in the current study were
.91 for PA and .92 for NA.
Daily Stress Inventory (DSI; Brantley & Jones, 1989). The
original measure includes 60 items that assess stress from daily
activities. This measure contains two subscales measuring inter-
personal stress (e.g., was ignored by others, was embarrassed) and
environmental hassles (e.g., had minor accident, experienced bad
weather). Participants rate each event on the basis of how stressful
it is to him or her and these appraisals are compiled to create an
impact score. The generalizability coefficients indicate a useful
level of homogeneity among DSI items and stability (Brantley,
Waggoner, Jones, & Rappaport, 1987), and there is high conver-
gent validity between DSI scores and endocrine measures of stress
(Brantley, Dietz, McKnight, Jones, & Tulley, 1988). For the EMA
measurement, 26 DSI items were selected for inclusion by con-
sensus between the investigators and a group of ED assessors and
treatment providers on the basis of clinical and/or empirical rele-
vance to women with BN. This subset of items was selected to
shorten the instrument for repeated assessment and reduce the time
to complete EMA reports (i.e., burden) on participants.
Eating Disorder and Self-Destructive Beh avior Ch ecklist.
Items from several scales of eating disorder and self-destructive
behavior (e.g., Rossotto, Yager, & Rorty, 1998; Vanderlinden &
Vandereycken, 1997) were used to create a 19-item checklist of
momentary behaviors. The checklist included eating-related items
(“I binge ate,” “I vomited”), self-mutilation items (“I hit myself,”
“I cut myself”), and other impulsive behaviors (“I drove danger-
ously or recklessly,” “I engaged in risky and unprotected sex”).
Only the eating related items (binge eating and vomiting) were
used in this analysis. All participants were carefully trained in a
standard definition of BN-events by clinical research staff during
the palm pilot training session. For binge, the definition provided
participants was “an amount of food that you consider excessive or
an amount of food that other people would consider excessive,
with an associated loss of control or the feeling of being driven or
compelled to keep eating.” This definition was then discussed to
clarify the objective amount of food (given the circumstance) and
loss of control (if these feelings did not accompany a binge,
participants were trained to not record it on the palm pilot as a
binge). Examples of what objectively large amounts of food con-
sisted of were provided, and these examples were personally
tailored to the participants’ eating habits as reported during the
intake interview process.
SCID-I/P. A doctoral-level psychologist (trained to criterion
on this interview) administered the SCID-I/P. The Eating Disorder
Module of the SCID-I/P was used as the primary measure of eating
disorder psychopathology from which a DSM–IV diagnosis of BN
was made. Also, we used the SCID-I/P screener to obtain infor-
mation on treatment history. The SCID-I/P has been well docu-
mented as a reliable and valid measure (e.g., Beck, Steer, &
Garbin, 1988). Kappa coefficients to determine interrater reliabil-
ity for BN diagnoses, on the basis of 25 randomly selected cases
from the sample used in this study, were 1.00.
Procedure
Participants were recruited through clinical, community, and
campus advertisements. Eligible individuals attended an informa-
tional meeting where they received further information regarding
the study, completed an informed consent, and provided a blood
sample for electrolyte screening to ensure medical stability. Par-
ticipants were then scheduled for two assessment visits that lasted
a total of 3–4 hr. During this time, a doctoral-level psychologist
conducted several structured interviews, including assessments of
eating disorder symptoms, comorbid psychopathology, and per-
sonality. For the purpose of the present study, most of these data
were not used in analyses.
At the end of first assessment, participants were trained on the
use of the palmtop computers to complete their daily assessments.
They also met with the principal investigator of the study, during
which time they were reminded of the goals of the study, what to
expect during the data collection period, and how to deal with any
questions that might arise from the signaling of the palmtop
computer. Participants were instructed to not complete entries at
any times when they felt unable to reply or if safety was a concern
(e.g., while driving). Participants were encouraged instead to delay
(“snooze”) the palm and to complete the entry as soon as possible.
Participants carried the palmtop computer for 2 practice days, at
which point they returned for the second assessment visit and
returned the practice data they had collected (these data were not
used in analyses). This run-in interval also was used to reduce
concerns about immediate reactivity (although there is little evi-
dence of reactivity; e.g., Stein & Corte, 2003). These data were
reviewed, and participants were given feedback regarding their
compliance rates. Participants were then given the palmtop com-
puter with the self-report questionnaires to be completed over the
course of the next 2 weeks. Attempts were made to schedule two
visits for each participant during this two week interval to collect
palmtop data to reduce the amount of data lost in the event that any
technical problems might arise with the palmtop computer. Par-
ticipants were given feedback at each visit with respect to their
compliance rates. All of the women were compensated for their
participation and given treatment referral information at the end of
the study period.
631
MOOD AND STRESS IN BULIMIA NERVOSA

The current EMA assessment implemented three types of daily
self-report methods (Wheeler & Reis, 1991). Signal-contingent
protocols require individuals to report on experiences at various
times throughout the day in response to a signal. In the present
study, participants were signaled at six semirandom times through-
out the day to provide mood, stress, and BN behaviors. The time
of a signal was determined by randomly selecting times around the
following anchor points on the basis of subdividing the day into six
roughly equivalent time blocks: 8:30 a.m., 11:10 a.m., 1:50 p.m.,
4:30 p.m., 7:10 p.m., and 9:50 p.m. The signal times were ran-
domly distributed around these anchor times in a normal distribu-
tion with a mean of 0 min (i.e., the anchor point) and a standard
deviation of 30 to adequately and evenly sample across the waking
hours of the day. When signaled, participants were asked to rate
their mood and report any recent behaviors not yet recorded.
Event-contingent schedules involve completing assessment in re-
sponse to specific events that have occurred. In the current study,
participants were trained to complete ratings of mood, stress, and
behavior following any event listed on a laminated card that was
carried with the palmtop computer. The card contained the 19
items from the Eating Disorder and Self-Destructive Behavior
Checklist. Finally, end-of-day ratings (interval recordings) were
provided to summarize the experiences of the day, although these
data are not utilized in this report.
Results
Over the course of the 2-week period, binge and/or purge
behaviors were moderately frequent. Participants, on average, re-
ported at least one binge on 40% of the days and at least one vomit
event on 46% of days. On 33% of days, both a binge and vomit
event (or multiple events, but encompassing both) were reported.
Women reported an average of 8.65 binges (SD ! 6.68; range !
1–34) and 11.47 vomits (SD ! 9.24; range ! 1– 48) on EMA
reports over the course of their study involvement.
Compliance with EMA reporting was generally quite good, with
overall compliance rates (the percentage of prompted beeps com-
pleted) for the sample at an average of 86% of beeps completed
(Mdn ! 90%). Of the women, 75% complied with 83% or more of
the beeps. Participants’ timeliness was quite good, with "75% of
beeps completed within 20 min. The majority of prompts were
responded to within 5 min (median lag time to response ! 4 min).
Compliance rates were similar across time of day (indexed by beep
number from morning to night; 1 ! 81%, 2 ! 85%, 3 ! 86%, 4 !
88%, 5 ! 88%, 6 ! 87%).
The general goal of this study was to determine how ratings of
affect and stress differed on binge and vomit days and changed in
relation to an upcoming binge or vomit event. We used multilevel
modeling to examine between-days and within-day relationships
between BN behaviors and affect and stress. We ran all multilevel
models using SAS Proc Mixed (SAS Institute, 2004). We assessed
significance for regression coefficients using a t statistic with a
degree of freedom equal to J–p–1, where J is the number of
participants, and p is the number of predictor variables (Rauden-
bush & Bryk, 2002, p. 58). We assessed significance for variance
components using one-tailed likelihood ratio tests (Snijders &
Bosker, 1999). The within-person dependency among observations
(i.e., autocorrelation) was modeled by allowing the correlation
between observations to vary as function of the time interval
between them using the spatial power option in SAS PROC
MIXED. The autocorrelation for the between-days analyses rep-
resents the model estimate of the correlation between two obser-
vations made 1 day apart, and the autocorrelation for the within-
day analyses represents the model estimate of the correlation
between two observations made 2 hr apart. Model fit was im-
proved for all analyses by allowing measurement error in addition
to the autocorrelation.
Between-Days Analyses
This analysis compared mean mood and stress on BN-event
days versus non-BN-event days. Data were aggregated across
within-day assessments so that affect and stress scores reflected
the average values for each person on each day. Binary dummy
codes were assigned to each day to distinguish days when the
relevant BN-events did (coded 1) and did not (coded 0) occur.
Binge and vomit events were analyzed separately. We hypothe-
sized that on days with either a binge or vomit event, individuals
would report more negative mood and more severe stressors. Table
1 shows the results for the analysis of binge versus no-binge days
and Table 2 shows the results for the analysis of vomit versus
no-vomit days. The estimate for binge day in Table 1 and for vomit
Table 1
Between-Days Multilevel Models for Binge Events
Variable
Negative affect Anger/hostility Stress Positive affect
Estimate SE t Estimate SE t Estimate SE t Estimate SE t
Intercept 23.36 0.70 33.24
***
7.10 0.24 30.23
***
4.16 0.37 11.35
***
35.31 0.74 48.33
***
Binge day 2.82 0.21 13.53
***
0.84 0.08 9.65
***
0.82 0.17 4.92
***
#2.32 0.22 #10.66
***
Variance terms
BP variance 53.86
††
6.06
††
16.36
††
61.02
††
WP variance 20.59
††
2.19
††
6.61
††
18.78
††
Autocorrelation 0.86
††
0.83
††
0.13
†
0.86
††
Residual variance 9.88
††
1.89
††
4.33
††
11.48
††
Note. These analyses are based on 133 individuals. For tests of regression coefficients, the degree of freedom ! 131. Binge day is coded 1 for a binge
day and 0 for a no-binge day. BP ! between-persons; WP ! autocorrelated within-person variance.
†
p $ .01, one-tailed.
††
p $ .001, one-tailed.
***
p $ .001, two-tailed.
632
SMYTH ET AL.

day in Table 2 reflects the average difference on each dependent
variable on a day an event occurred relative to an event free day.
As hypothesized, binge days and vomit days both showed signif-
icantly less PA, higher NA, higher AH, and greater reported STRS
than days with no BN-events.
Within-Day Analyses
The next set of analyses compared mood and stress ratings
preceding and following a binge or vomit event. Because trajec-
tories in mood and stress that precede the BN-event might differ
from trajectories following the BN-event, we separately modeled
pre- and postevent trends using piecewise linear and quadratic
functions centered on the BN-event time. Multilevel models in-
cluded four predictor variables: hours prior to event, (hours prior
to event)
2
, hours following event, and (hours following event)
2
.
Hours prior to event and hours following event captured the linear
trends, and the squares of these terms captured the quadratic
trends. When multiple BN-events were reported in a single day,
only the first event was used to avoid confounding the effects of
mood as an antecedent to the second event with the affective
consequences of the first event.
1
Figures 1
and 2 represent the
fitted trends in mood and stress severity ratings in the hours prior
to and following binge events. The vertical dashed line in the
graphs indicates the time of the BN-event.
Results from the multilevel analyses indicate that prior to a
binge event, women reported increasing NA, AH, and STRS, and
they reported decreasing PA (see Table 3). The intercepts reported
in Table 3 reflect the estimated value of the dependent variable at
the time of the binge, the estimate for hours prior to event reflects
the rate of change in the dependent variable at the time of the
binge, and the estimate for (hours prior to event)
2
reflects the
acceleration in rate of change in mood and stress prior to the event.
The estimate for hours following event reflects the rate of change
in the dependent variable immediately following the binge, and the
estimate for (hours following event)
2
reflects the acceleration in
rate of change in mood and stress postbinge. The significant
hours-prior-to-event effect shown in Table 3 indicated that NA,
AH and STRS was increasing and PA was decreasing at the time
of a binge. The significant (hours-prior-to-event)
2
effect indicated
acceleration in the rate of increase in NA, AH, and decrease in PA
as individuals approached the time of a binge. There was also
evidence of significant acceleration in mood following a binge.
Statistical comparison of the pre- and postbinge quadratic coeffi-
cients indicated significantly more acceleration postbinge for NA,
t(121) ! 3.85, p $ .01; AH, t(121) ! 2.91, p $ .01; and PA,
t(121) ! 2.37, p $ .05; but not for STRS, t(121) ! 0.92, ns. (See
Figures 1 and 2.) An identical pattern of results was obtained for
vomit events (see Table 4), with significantly more post- compared
with prevomit acceleration for NA, t(117) ! 4.47, p $ .01; AH,
t(117) ! 2.74, p $ .01; and PA, t(117) ! 2.85, p $ .01; but not
for STRS, t(117) ! 1.46, ns. We conducted several post hoc
component analyses on NA and STRS to determine whether either
specific negative affects (for NA) or stressor types (interpersonal
or environmental) were driving the overall effects observed in
primary analyses (or would show divergent relationships). There
was no evidence that any observed effects were being driven by
specific affect(s) or solely by one stressor type (analyses not
shown).
To ensure that these findings were not solely because of affect
and stress related to the event itself, we conducted the same set of
analyses again, this time excluding ratings that were concurrent
(within 10 min) with the BN-event. We selected the 10-min cut
time after inspecting the distribution of assessments relative to the
time of the BN-event. Results from this reanalysis are displayed in
Table 5. The only change in significant results was for the pre-
1
Approximately 39% of all vomit events occurred either before the first
binge event of the day or on days without a report of a binge event. To
explore whether vomit events not preceded by binges were dissimilar to
those preceded by binges, we dummy coded all vomit events to indicate
whether they occurred after a binge event. Two sets of supplemental
analyses were run. First, all the vomit analyses were rerun, including only
vomit events that did not follow a binge. The direction of the regression
coefficients were the same as when all vomit events were analyzed to-
gether, indicating the same pattern of results. Second, the dummy variable
and its interaction with the linear and quadratic pre- and posttime variables
were analyzed. The effect of the dummy variable was not significant, nor
was its interaction with the pre- and posttime variables. This second
analysis more directly addresses the question of whether the pattern of
results differs for vomit events that occur after a binge versus those that do
not. The result is that there is no detectable difference in these data.
Table 2
Between-Days Multilevel Models for Vomit Events
Variable
Negative affect Anger/hostility Stress Positive affect
Estimate SE t Estimate SE t Estimate SE t Estimate SE t
Intercept 23.16 0.71 32.49
***
7.06 0.24 29.77
***
4.09 0.37 10.98
***
35.74 0.74 48.11
***
Vomit day 2.84 0.21 13.44
***
0.81 0.09 9.13
***
0.844 0.17 5.04
***
#2.26 0.22 #10.18
***
Variance terms
BP variance 54.86
††
6.08
††
16.76
††
61.14
††
WP variance 20.85
††
2.18
††
6.50
††
18.35
††
Autocorrelation 0.86
††
0.85
††
0.15
†
0.85
††
Residual variance 10.07
††
1.95
††
4.43
††
11.57
††
Note. These analyses are based on 133 individuals. For tests of regression coefficients, the degree of freedom ! 131. BP ! between-persons; WP !
autocorrelated within-person variance.
†
p $ .01, one-tailed.
††
p $ .001, one-tailed.
***
p $ .001, two-tailed.
633
MOOD AND STRESS IN BULIMIA NERVOSA

vomit linear trend for stress, which was significant at the .05 level
when analyzing all responses but was no longer significant after
excluding observations within 10 min of the vomit event. All other
results were unchanged, suggesting that the observed trends were
not attributable to the concurrent effects of BN-events on mood.
Discussion
This study aimed to better understand the role of mood and stress
ratings as antecedents to binge and vomit events for a large population
of women with BN. EMAs were used to obtain self-reports over the
course of 2 weeks in the women’s natural environment. Participants
recorded their moods and stress ratings on a palmtop computer at six
semirandom times through the day and following a BN-event. This
data collection method allowed us to compare mood and stress ratings
at both between-days and within-day levels.
The first prediction from these data was that women would
report more NA, more AH, lower PA, and higher STRS severity
ratings (all assessed during daily life) on days when BN-events
15
20
25
30
35
40
45
-6 -4 -2 0 2 4 6
Negative Affect
Positive Affect
Negative Affect or Positive Affect
Hours Relative to Binge
15
20
25
30
35
40
45
-6 -4 -2 0 2 4 6
Negative Affect
Positive Affect
Negative Affect or Positive Affect
Hours Relative to Vomit
Figure 1. Positive and negative moods relative to binge or vomit. Points represent predicted positive or
negative affect relative to time in hours.
0
2
4
6
8
10
12
-6 -4 -2 0 2 4 6
Anger-Hostility
Stress Severity
Anger-Hostility or Stress Severity
Hours Relative to Binge
0
2
4
6
8
10
12
-6 -4 -2 0 2 4 6
Anger-Hostility
Stress Severity
Anger-Hostility or Stress Severity
Hours Relative to Vomit
Figure 2. Profile of Mood States anger/hostility and stress relative to binge or vomit. Points represent predicted
anger/hostility or stress relative to time in hours.
634
SMYTH ET AL.

(defined as the occurrence of at least one binge or vomit) occurred.
An examination of BN-event days and days with no BN-events
suggested that the participants experienced higher NA, AH, and
STRS severity and less PA on BN-event days. These between-days
estimates were generated by collapsing across momentary reports
taken in the participants’ natural environments rather than relying
on daily diaries or retrospective report, and they are convergent
with previous findings that used these other methods. This solid-
ifies evidence that days marked by disordered eating behaviors are
associated with more dysphoric mood (Johnson & Larson, 1982;
Kjelsas et al., 2004; Lingswiler et al., 1989; Schlundt et al., 1985;
Waters et al., 2001; Wegner et al., 2002) and higher levels of stress
(Cattanach et al., 1988; Crowther & Sherwood, 1997; Lingswiler
et al., 1989; Yacono Freeman & Gil, 2004).
Our second and third predictions addressed the dynamic fluctu-
ations of mood and stress preceding and following binge and vomit
behaviors in the natural environment (i.e., in “real life”). To
examine this question, we looked at days when a BN-event oc-
curred and traced the trajectory of NA, PA, AH, and STRS severity
leading up to and following the first binge or vomit of the day. NA,
AH, and STRS severity ratings all increased up to the BN-event,
whereas PA decreased. Following the binge, there was rapid
recovery, with NA and AH decreasing and PA increasing. In fact,
the rate of (affective) recovery following a BN-event was signif-
icantly more rapid than the rate of affective decay preceding a
BN-event. Again, these findings are consistent with previous lit-
erature (Cattanach et al., 1988; Crowther & Sherwood, 1997;
Johnson & Larson, 1982; Kjelsas et al., 2004; Lingswiler et al.,
Table 3
Within-Day Multilevel Models for Binge Events
Variable
Negative affect Anger/hostility Stress Positive affect
Estimate SE t Estimate SE t Estimate SE t Estimate SE t
Intercept 29.89 0.78 38.32
***
8.79 0.28 31.78
***
6.61 0.49 12.54
***
30.89 0.74 41.74
***
Hours prior to event 1.45 0.08 17.81
***
0.34 0.04 9.78
***
0.30 0.10 3.06
***
#1.06 0.08 #12.69
***
(Hours prior to
event)
2
0.08 0.01 13.22
***
0.02 0.01 6.98
***
0.01 0.01 1.41 #0.07 0.01 #10.93
***
Hours following
event #2.08 0.21 #9.76
***
#0.58 0.09 #6.22
**
#0.45 0.24 #1.89 1.49 0.22 6.68
***
(Hours following
event)
2
0.22 0.04 5.97
***
0.06 0.02 3.98
***
0.05 0.04 24.00 #0.15 0.04 #4.07
***
Variance terms
BP variance 67.15
††
8.01
††
21.77
††
59.49
††
WP variance 33.17
††
6.07
††
38.30
††
Autocorrelation 0.62
††
0.47
††
0.45
††
Residual variance 17.91
††
2.65
††
36.63
††
11.54
††
Note. These analyses are based on 126 individuals. For tests of regression coefficients, the degree of freedom ! 121. BP ! between-persons; WP !
autocorrelated within-person variance.
††
p $ .001, one-tailed.
**
p $ .01, two-tailed.
***
p $ .001, two-tailed.
Table 4
Within-Day Multilevel Models for Vomit Events
Variable
Negative affect Anger/hostility Stress Positive affect
Estimate SE t Estimate SE t Estimate SE t Estimate SE t
Intercept 28.89 0.79 38.67
***
8.55 0.28 30.80
***
5.14 0.52 12.38
***
31.55 0.79 40.04
***
Hours prior to event 1.06 0.07 14.72
***
0.24 0.03 7.79
***
0.20 0.08 2.29
*
#0.79 0.07 #11.67
***
(Hours prior to
event)
2
0.05 0.01 10.89
***
0.01 0.01 5.46
***
0.01 0.01 1.14 #0.05 0.01 #10.20
***
Hours following
event #1.85 0.20 #9.22
***
#0.48 0.09 #5.58
***
#0.42 0.21 #2.09
*
1.33 0.21 6.47
***
(Hours following
event)
2
0.21 0.03 6.04
***
0.05 0.01 3.51
***
0.05 0.04 1.17 #0.15 0.03 #4.33
***
Variance terms
BP variance 67.94
††
8.07
††
14.35
††
67.62
††
WP variance 30.48
††
5.65
††
33.29
††
Autocorrelation 0.61
††
0.52
††
0.58
††
Residual variance 18.11
††
2.71
††
36.71
††
17.35
††
Note. These analyses are based on 122 individuals. For tests of regression coefficients, the degree of freedom ! 117. BP ! between-persons; WP !
autocorrelated within-person variance.
††
p $ .001, one-tailed.
*
p $ .05, two-tailed.
***
p $ .001, two-tailed.
635
MOOD AND STRESS IN BULIMIA NERVOSA

1989; Schlundt et al., 1985; Waters et al., 2001; Yacono Freeman
& Gil, 2004) but extend these findings in several important ways.
First, the use of the momentary data allowed examination of
temporal patterns leading up to and following the BN-events. This
provides a unique opportunity to examine the potentially (affec-
tively) reinforcing nature of engaging in binging or vomiting
behaviors for women with BN. On a more global level, BN-events
are quite negative. That is, BN-events commonly occur on days
with significantly “worse” moods and, even at the moment when
mood is best on BN-event days, moods are more negative than on
days when no BN-events occur. In sharp contrast, however, the
proximal (or local) mood trajectories around a BN-event appear to
hold markedly different reinforcing properties (at least for women
with BN; cf. Wegner et al., 2002). Mood significantly worsened
(less PA, more NA and AH) leading up to BN-events but improved
(significantly more rapidly) following the event. This suggests
that, local to the event (i.e., within a few hours), binging or purging
behaviors are strongly negatively reinforcing as they allow escape/
avoidance of a strongly negative affective state. This (local) rein-
forcing pattern provides support for the persistence (and resistance
to change) of binge and purge behaviors, yet is also consistent with
the view that such behaviors are not effective overall (global)
coping efforts.
Second, the present study also conducted analyses that removed
mood assessments that may be temporally consequential to the
binge event (within 10 min). We were concerned that these find-
ings may be attributed to the negative mood state and stress
associated with the beginning of binge or vomit events themselves.
There is conflicting research regarding the effect that BN-events
have on measures of mood. Some reports suggest that negative
mood worsens during binge events (e.g., Johnson & Larson, 1982;
Lynch, Everingham, Dubitzky, H artman, & Kasser, 2000),
whereas others indicate that negative mood may improve during
the event (e.g., Abraham & Beumont, 1982; Kaye, Gwirtsman,
George, Weiss, & Jimerson, 1986). Disregarding the directional
effects, it is evident that the BN-event itself is affect-laden, which
could partially account for the mood trajectories that are demon-
strated prior to these events. In the present study, including anal-
yses that excluded ratings completed within 10 min of the BN-
event suggested that PA and NA, as well as AH, are robust
predictors of a binge and vomit event. Previous findings have
demonstrated that nonspecific negative feelings precede binge and
vomit events (e.g., Alpers & Tuschen-Caffier, 2001), but this study
is the first to indicate that various types of negative emotions may
lead up to, and be directly associated with, BN-events. This may
have important implications for targeting specific affective ante-
cedents and for considering their unique dynamic associations with
BN and other self-destructive behaviors for research and clinical
practice (e.g., Stice, Burton, & Shaw, 2004).
A third way in which this study extends the findings of previous
research is that we assessed the relationship between mood and
BN-events in the natural environment with minimal retrospective
recall effects. By reducing the recall period from days and weeks
(as is typically used in disordered eating and mood assessments) to
minutes and hours, we create a more realistic depiction of what is
occurring in the daily lives of these women (Smyth et al., 2001).
Stein and Corte (2003) have demonstrated that assessments that
require individuals to recall information about BN-events days or
weeks prior to the assessment time may be inconsistent with daily
reports. They suggest that this may be due to insufficiently detailed
information in memory about the event (e.g., a binge) to be able to
accurately evaluate it and indicate that future studies should limit
the recall period when examining disordered eating behaviors.
Several limitations and possible future studies should be noted.
First, although these data can suggest the temporal ordering of
mood, stress, and binge or vomit events, they are still correlational
in nature and rely upon the accurate reporting by participants. That
is to say, we cannot definitively establish that negative mood or
increased stress severity ratings cause the BN-event. Second, it is
unknown to what extent self-monitoring of affect, stress, and
behavior influenced the participants, although we used a 2-day
run-in period, and there were not gross trends in reporting levels of
our primary constructs over time (a crude indicator that there were
not strong reactivity effects). Also, previous EMA studies with BN
participants have revealed minimal effects of reactivity (e.g., Stein
& Corte, 2003). Third, this study focused on short-term antecedent
Table 5
Within-Day Multilevel Models for Binge and Vomit Events Excluding Data Concurrent (%10 min) With Event
Variable
Negative affect Anger/hostility Stress Positive affect
Estimate SE t Estimate SE t Estimate SE t Estimate SE t
Binge events
Intercept 29.62 0.78 37.87
***
8.71 0.28 31.53
***
5.85 0.48 12.10
***
31.13 0.75 41.50
***
Hours prior to event 1.40 0.08 16.65
**
0.33 0.04 9.26
***
0.23 0.10 2.37
*
#1.02 0.09 #11.83
***
(Hours prior to event)
2
0.08 0.01 12.51
***
0.02 0.00 6.61
***
0.01 0.01 0.84 #0.06 0.01 10.39
***
Hours following event #1.95 0.22 #8.95
***
#0.52 0.09 #5.51
***
#0.34 0.24 #1.44 1.33 0.23 5.82
***
(Hours following event)
2
0.20 0.04 5.56
***
0.06 0.02 3.51
***
0.04 0.04 0.98 #0.14 0.04 #3.50
***
Vomit events
Intercept 28.31 0.79 35.89
***
8.40 0.28 30.30
***
5.46 0.51 10.68
***
31.98 0.80 39.74
***
Hours prior to event 0.96 0.08 12.39
***
0.20 0.03 6.13
***
0.16 0.08 1.90 #0.78 0.08 #9.85
***
(Hours prior to event)
2
0.05 0.01 8.63
***
0.01 0.00 3.84
***
0.01 0.01 0.76 #0.05 0.01 #9.51
***
Hours following event #1.47 0.20 #7.26
***
#0.38 0.09 4.29
***
#0.43 0.22 #1.99
*
1.02 0.21 4.96
***
(Hours following event)
2
0.16 0.03 4.79
***
0.04 0.01 2.63
**
0.05 0.04 1.50 #0.12 0.03 #3.37
***
Note. Analyses of binge and vomit events are based on 126 (degree of freedom [df] ! 121) and 122 (df ! 117) individuals, respectively.
*
p $ .05, two-tailed.
**
p $ .01, two-tailed.
***
p $ .001, two-tailed.
636
SMYTH ET AL.

events. Future research may use these data to predict binge eating
or vomiting behaviors on the basis of reported events that extend
over longer periods of time. These studies would be valuable for
establishing predictive factors over various time intervals. Fourth,
there are a number of important contextual moderators that remain
open to investigation. For example, would mood/stress be most
predictive of BN-events on days that are otherwise marked by high
food restriction (i.e., dieting)? More generally, the study of more
macrocharacteristics (e.g., of the day, of the individual) interacting
with more microcharacteristics (e.g., momentary reports of mood,
social interactions) is an important future direction for research.
A major strength of this study was the use of a substantial
sample (n ! 131) of clinically diagnosed women with BN. Pre-
vious research examining mood and stress as antecedents of eating
pathology has generally focused on either subclinical populations
(e.g., Wegner et al., 2002) or used small sample sizes (e.g.,
Johnson & Larson, 1982; Waters et al., 2001). This is the first time
similar findings have been demonstrated in a large clinical sample.
Another strength of this methodology is the use of palmtop com-
puters to record all data over the course of the 2-week assessment
period. This technique is advantageous over a more traditional
daily diary method because the palmtop computer records the date
and time when each assessment is completed. This helps to ensure
participant compliance and lessens the likelihood of obtaining data
that are marred by retrospective recall biases or completed at times
other than those reported.
These data provide strong evidence that mood and stress in daily
life are related to binge eating and vomiting in women with BN.
BN-events are more likely to occur on dysphoric days, and mood
becomes increasingly more negative leading up to a BN-event. In
contrast, mood improves rapidly following a BN-event, suggesting
the event (binge or vomit) has negatively reinforcing properties.
Demonstration of the temporal sequencing of stress, mood, and
BN-events with a large BN sample, and providing empirical evi-
dence of the reinforcing nature of binge and vomit behaviors for
women with BN, holds potential to advance theory and clinical
practice in eating disorders.
Specifically, these EMA findings may inform evidence-based
prevention (see Stice & Shaw, 2004) and treatment programs for
BN, such as cognitive–behavioral therapy (Fairburn, Marcus, &
Wilson, 1993) or dialectical– behavior therapy (Linehan, 1993),
which target cognitive appraisals of potentially stressful situations
and emotional responding in the treatment of BN. For example,
two recently developed treatments for BN include an explicit
emphasis on the potential maintaining role that emotional insta-
bility may have on bulimic behavior (Fairburn, Cooper, & Shafran,
2003; Wonderlich, Mitchell, Peterson, & Crow, 2001). The present
findings offer empirical support for clinical techniques that em-
phasize emotion regulation skills and alternative action choices in
the face of stress and/or NA. Furthermore, one of these treatments
(Wonderlich et al., 2001) includes palmtop computer-based treat-
ment modules that could easily be integrated with an EMA assess-
ment scheme in a manner that combines momentary assessment
with promotion of alternative coping skills.
References
Abraham, S. F., & Beumont, P. J. (1982). How patients describe bulimia or
binge eating. Psychological Medicine, 12, 625– 635.
Agras, W. S., & Telch, C. F. (1998). The effects of caloric deprivation and
negative affect on binge eating in obese binge-eating disordered women.
Behavior Therapy, 29, 491–503.
Alpers, G. W., & Tuschen-Caffier, B. (2001). Negative feelings and the
desire to eat in bulimia nervosa. Eating Behaviors, 2, 339–352.
American Psychiatric Association. (1994). Diagnostic and statistical man-
ual of mental disorders (4th ed.). Washington, DC: Author.
Bardone, A. M., Krahn, D. D., Goodman, B. M., & Searles, J. S. (2000).
Using interactive voice response technology and timeline follow-back
methodology in studying binge eating and drinking behavior: Different
answers to different forms of the same question? Addictive Behaviors,
25, 1–11.
Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties
of the Beck Depression Inventory: Twenty-five years of evaluation.
Clinical Psychology Review, 8, 77–100.
Bolger, N., & Zuckerman, A. (1995). A framework for studying person-
ality in the stress process. Journal of Personality and Social Psychology,
69, 890 –902.
Brantley, P. J., Dietz, L. S., McKnight, G. T., Jones, G. N., & Tulley, R.
(1988). Convergence between the Daily Stress Inventory and endocrine
measures of stress. Journal of Consulting and Clinical Psychology, 56,
549 –551.
Brantley, P. J., & Jones, G. N. (1989). Daily Stress Inventory: Professional
manual. Odessa, FL: Psychological Assessment Resources.
Brantley, P. J., Waggoner, C. D., Jones, G. N., & Rappaport, N. B. (1987).
A Daily Stress Inventory: Development, reliability, and validity. Journal
of Behavioral Medicine, 10, 61–74.
Cattanach, L., Malley, R., & Rodin, J. (1988). Psychologic and physiologic
reactivity to stressors in eating disordered individuals. Psychosomatic
Medicine, 50, 591–599.
Crowther, J. H., & Sherwood, N. E. (1997). Assessment. In D. M. Garner
& P. E. Garfinkel (Eds.), Handbook of treatment for eating disorders
(2nd ed., pp. 34–49). New York: Guilford Press.
Fairburn, C. G., Cooper, Z., & Shafran, R. (2003). Cognitive behaviour
therapy for eating disorders: A “transdiagnostic” theory and treatment.
Behaviour Research and Therapy, 41, 509 –528.
Fairburn, C. G., Marcus, M. D., & Wilson, G. T. (1993). Cognitive–
behavioral therapy for binge eating and bulimia nervosa: A comprehen-
sive treatment manual. In C. G. Fairburn & G. T. Wilson (Eds.), Binge
eating nature, assessment, and treatment (pp. 361– 404). New York:
Guilford Press.
First, M., Spitzer, R., Gibbon, M., & Williams, J. (1995). Structured
Clinical Interview for DSM–IV Axis I Disorders: Patient Edition (SCID-
I/P). New York: Biometrics.
Johnson, C., & Larson, R. (1982). Bulimia: An analysis of moods and
behavior. Psychosomatic Medicine, 44, 341–351.
Kaye, W. H., Gwirtsman, H. E., George, D. T., Weiss, S. R., & Jimerson,
D. C. (1986). Relationship of mood alterations to bingeing behaviour in
bulimia. British Journal of Psychiatry, 149, 479 – 485.
Kjelsas, E., Borsting, I., & Gudde, C. B. (2004). Antecedents and conse-
quences of binge eating episodes in women with an eating disorder.
Eating and Weight Disorders, 9, 7–15.
Linehan, M. (1993). Cognitive–behavioral treatment of borderline person-
ality disorder. New York: Guilford Press.
Lingswiler, V. M., Crowther, J. H., & Stephens, M. A. P. (1989). Affective
and cognitive antecedents to eating episodes in bulimia and binge-eating.
International Journal of Eating Disorders, 5, 533–539.
Lorr, M., & McNair, D. M. (1971). The Profile of Mood States manual. San
Diego, CA: Educational and Industrial Testing Services.
Lynch, W. C., Everingham, A., Dubitzky, J., Hartman, M., & Kasser, T.
(2000). Does binge eating play a role in the self-regulation of moods?
Integrative Physiological and Behavioral Science, 35, 298 –313.
Meyer, C., Newman, L., Waller, G., Perkins, S., Paice, N., & Mitchell, J.
637
MOOD AND STRESS IN BULIMIA NERVOSA

(2005). Anger and bulimic psychopathology: Gender differences in a
nonclinical group. International Journal of Eating Disorders, 37, 69 –71.
Milligan, R. J., & Waller, G. (2000). Anger and bulimic psychopathology
among nonclinical women. International Journal of Eating Disorders,
28, 446 – 450.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models:
Applications and data analysis methods (2nd ed.). Newbury Park, CA:
Sage.
Rossotto, E., Yager, J., & Rorty, M. (1998). The Impulsive Behavior Scale.
Unpublished manuscript.
SAS Institute. (2004). SAS/STAT user’s guide, Version 9.1. Cary, NC:
Author.
Schlundt, D. G., Johnson, W. G., & Jarrell, M. P. (1985). A naturalistic
functional analysis of eating behavior in bulimia and obesity. Advances
in Behavior Research and Therapy, 7, 149 –162.
Smyth, J., Wonderlich, S., Crosby, R., Miltenberger, R., Mitchell, J., &
Rorty, M. (2001). The use of ecological momentary assessment ap-
proaches in eating disorder research. International Journal of Eating
Disorders, 30, 83–95.
Snijders, T., & Bosker, R. J. (1999). Multilevel analysis: An introduction
to basic and advanced multilevel modeling. London: Sage.
Stein, K. F., & Corte, C. M. (2003). Ecological momentary assessment of
eating-disordered behaviors. International Journal of Eating Disorders,
34, 349 –360.
Stice, E., Burton, E. M., & Shaw, H. (2004). Prospective relations between
bulimic pathology, depression, and substance abuse: Unpacking comor-
bidity in adolescent girls. Journal of Consulting and Clinical Psychol-
ogy, 72, 62–71.
Stice, E., & Shaw, H. (2004). Eating disorder prevention programs: A
meta-analytic review. Psychological Bulletin, 130, 206 –227.
Telch, C. F., & Agras, S. W. (1996). Do emotional states influence binge
eating in the obese? International Journal of Eating Disorders, 20,
271–279.
Vanderlinden, J., & Vandereycken, W. (1997). Trauma, dissociation, and
impulse dyscontrol in eating disorders. Bristol, PA: Brunner Mazel.
Waller, G., Babbs, M., Milligan, R., Meyer, C., Ohanian, V., & Leung, N.
(2003). Anger and core beliefs in eating disorders. International Journal
of Eating Disorders, 34, 118–124.
Waters, A., Hill, A., & Waller, G. (2001). Bulimic’s responses to food
cravings: I. Binge-eating a product of hunger or emotional state? Be-
haviour Research and Therapy, 39, 877– 886.
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and vali-
dation of brief measures of positive and negative affect: The PANAS
scales. Journal of Personality and Social Psychology, 54, 1063–1070.
Wegner, K. E., Smyth, J. M., Crosby, R. D., Wittrock, D., Wonderlich,
S. A., & Mitchell, J. E. (2002). An evaluation of the relationship between
mood and binge eating in the natural environment using ecological
momentary assessment. International Journal of Eating Disorders, 32,
352–361.
Wheeler, L., & Reis, H. T. (1991). Self-recording of everyday life events:
Origins, types, and uses. Journal of Personality, 59, 339–354.
Wonderlich, S. A., Mitchell, J. E., Peterson, C. B., & Crow, S. (2001).
Integrative cognitive therapy for bulimic behavior. In R. H. Striegel-
Moore & L. Smolak (Eds.), Eating disorders: Innovative directions in
research and practice (pp. 173–195). Washington, DC: American Psy-
chological Association.
Yacono Freeman, L. M., & Gil, K. M. (2004). Daily stress, coping, and
dietary restraint in binge eating. International Journal of Eating Disor-
ders, 36, 204 –212.
Received September 5, 2006
Revision received April 25, 2007
Accepted April 30, 2007 !
638
SMYTH ET AL.