International Journal of Behavioral Medicine

https://doi.org/10.1007/s12529-019-09772-9

BRIEF REPORT

Moderation of Mood in the Transfer of Self-Regulation From an Exercise

to an Eating Context: Short- and Long-Term Effects on Dietary Change
and Obesity in Women
James J. Annesi 1,2

# International Society of Behavioral Medicine 2019

Abstract
Background Behavioral obesity treatments require an improved understanding of the dynamics of associated psychological
changes. This study aimed to clarify previous research on self-regulatory skills’ transfer from an exercise to eating context,
effects of mood on self-regulatory strength, and related effects on a targeted eating behavior.
Methods Women with obesity participated in a yearlong community-based cognitive-behavioral treatment that first focused on
self-regulatory skills development for exercise maintenance, then use of similar self-regulatory skills and improved mood to
facilitate short- and long-term increases in fruit/vegetable intake and reduction in weight. Groups were based on high (≥ 5% of
baseline weight; n = 51) and low (< 5%; n = 49) weight reduction 2 years post-initiation.
Results Improvements in eating self-regulation and fruit/vegetable intake were greatest in the high weight-reduction group.
Using lagged variable analyses to assess directionality, mood significantly moderated the prediction of eating self-regulation
change by exercise self-regulation change. The effect of increased exercise self-regulation on fruit/vegetable intake change over
6 months was significantly mediated by eating self-regulation change. Participants’ initial weight moderated the effect of eating
self-regulation change on fruit/vegetable intake. Change in eating self-regulation over 6 months predicted self-regulation at
24 months. Short-term change in fruit/vegetable intake predicted weight change over 2 years through its association with
long-term fruit/vegetable consumption.
Conclusions Findings supported the expected carry-over of self-regulation from an exercise to eating context, mood effects on
self-regulatory strength, and associations of exercise with eating and weight changes via effects on psychological variables.
Results have implications for sustained effects associated with behavioral treatments.

Keywords Self-regulation . Mood . Exercise . Obesity . Treatment . Behavioral

Introduction first several months [1]. A poor understanding of, and

accounting for, theory-based psychosocial changes asso-
Behavioral obesity treatments have generally been un- ciated with eating might be a reason [2]. Research be-
successful at maintaining enough weight loss to reduce ginning to address this shortcoming has often been
health risks (e.g., ≥ 5% of original weight) beyond the guided by social cognitive theory [3], emphasizing indi-
viduals’ control over their internal and environmental
Electronic supplementary material The online version of this article barriers to behavioral changes. When driven by social
(https://doi.org/10.1007/s12529-019-09772-9) contains supplementary cognitive theory, the development of self-regulatory
material, which is available to authorized users. skills such as cognitive restructuring and distal and
proximal goal setting, and consideration of emotional
* James J. Annesi states, is of concern within treatments. To define clear
jamesa@ymcaatlanta.org terminal behaviors, behavioral obesity treatments have
1 also sometimes focused on measurable targets such as
YMCA of Metro Atlanta, 101 Marietta Street, Atlanta, GA 30303,
USA a daily caloric limit, fat and/or sugar intake, or fruit and
2 vegetable consumption. Fruit/vegetable intake has alone
Department of Health Promotion and Physical Education, Kennesaw
State University, Kennesaw, GA, USA been associated with the health of the overall diet [4].

Exercise is the strongest correlate of long-term weight loss
[5]. However, this association might be more from its effects
on psychosocial predictors of improved eating than associated
energy expenditures [6]. As few as 2–3 moderate exercise
sessions per week have been associated with improved mood
[6]. Although minimally addressed in applied weight-loss set-
tings, and contrary to laboratory research indicating that self-
regulation applied to one task (e.g., exercise) can deplete self-
regulation applied to second task (e.g., healthy eating), some
research posits a carry-over of self-regulatory skills learned in
an exercise-support context, to an eating context [7, 8]. More
robust support of such a relationship could justify the need for
exercise early in behavioral intervention processes for reasons
well beyond the minor energy expenditures expected in
deconditioned individuals. Other research suggests that mood
affects self-regulation, especially through its influence on the
volition required to apply such means [9]. An individual’s
weight might also influence self-regulatory effects when seek-
ing a healthier diet for weight loss [10]. For example, individ-
uals with a higher degree of excess weight might require great-
er advances in self-regulation to overcome especially poor
habits [10]. Under conditions of enhanced self-regulation,
short-term (e.g., 6-month) improvements in the diet might
lead to longer-term (e.g., 24-month) improvements through
sustained abilities to address barriers.
If individuals successful with weight loss are more likely to
improve on the aforementioned behavioral variables than
those who are not, clarification of their interrelationships
might be particularly useful for improving intervention archi-
tectures. Treatment implications include whether self-
regulation should be focused first on exercise, if mood change
should be targeted to enhance self-regulation transfer from an
exercise to an eating emphasis, and whether short-term eating
changes transfer to long-term changes and weight loss under
conditions of increased self-regulation.
Thus, a yearlong cognitive-behavioral treatment aimed
at increasing self-regulation skills for both exercise (SR-
exercise) and eating (SR-eating), improving mood, and
increasing fruit/vegetable intake was administered in a
field setting to women with obesity. Treatment effects,
ΔSR-eatingb-6
ΔSR-eatingb-6
a b
Moodb-3/mean
a b a b
c c
ΔSR-exerciseb-3 ΔFruit/vegetableb-6 ΔSR-exerciseb-3
Fig. 1 a Mediation model predicting change in fruit/vegetable intake
from baseline to month 6. b Moderated mediation model predicting
change in fruit/vegetable intake from baseline to month 6. c Mediation
model predicting change in weight from baseline to month 24. Δ =
Int.J. Behav. Med.
and relationships among variables, were tested over 6
and 24 months. Hypotheses were as follows: (a) there will
be significantly greater improvements on all measured
variables in participants successful with weight loss, (b)
change in SR-eating will significantly mediate the fruit/
vegetable change–SR-exercise relationship, (c) mood will
significantly moderate the SR-exercise–SR-eating change
relationship, and baseline weight will significantly mod-
erate the self-regulation–eating change relationships, (d)
change in SR-eating from baseline to month 6 will signif-
icantly predict SR-eating at month 24, and (e) change in
fruit/vegetable intake over 24 months will significantly
mediate the prediction of weight change over 24 months
by fruit/vegetable intake over the initial 6 months. Fig. 1a,
b, and c depicts the proposed relationships.
Methods
Participants
Participants were adult females recruited through written and
electronic media into community health-promotion centers in
the southeast USA. Those currently pregnant/planning a preg-
nancy, in another weight-loss treatment, completing an aver-
age of 2 or more exercise sessions per week, or taking psy-
chotropic or weight-loss medications were ineligible.
Volunteers meeting the established criteria were progressively
accepted into the study until the planned sample size of 100
was obtained (which took approximately 10 weeks). This en-
rollment goal was based on the need for a minimum of 97 total
participants to detect a moderate effect size (f2 = 0.15) at the
statistical power of 0.80 (α ≤ 0.05) within the primary analy-
ses [11]. Groups were established after the data-collection
phase based on whether at least 5% of a participant’s original
weight (a marker of health-risk reduction [6]) was lost at
month 24 (high weight-loss group; n = 51), or 5% of weight
was not lost (low weight-loss group; n = 49). There was no
significant mean difference in age (48.18 ± 7.61 years), ethnic
make-up (75% white, 21% black, 4% other), yearly household
ΔFruit/vegetableb-24
c
Weightb
a b
c
ΔFruit/vegetableb-6 ΔFruit/vegetableb-6 ΔWeightb-24
change in score during indicated times. Subscripts b-3 = baseline–
month 3, b-6 = baseline–month 6, b-24 = baseline–month 24. Letters a,
b, and c′ indicate path a, path b, and path c′
Int.J. Behav. Med.
income (US$50,000–$100,000), or educational level (24%
high school, 32% bachelor’s degree, 44% beyond bachelor’s)
between the groups. Institutional review board-approved in-
formed consent was obtained in writing from each participant.
Measures
SR-exercise and SR-eating were separately measured using 10
items each. The previously developed and validated surveys
[12] assessed respondents’ present use of self-regulatory
skills. Sample items for SR-exercise were “I make formal
agreements with myself to be physically active,” “I set phys-
ical activity goals,” and “I keep a record of my physical
activities.”
Sample items for SR-eating were “I say positive things to
myself about eating well,” “I set eating goals,” and “I schedule
my times to eat.” Response options were from 1 (never) to 4
(often) and were summed. Internal consistencies were
Cronbach’s α = 0.81 and 0.79, respectively [12]; and α = 0.79
and 0.77 in the present sample. Test-retest reliabilities were 0.74
and 0.78 [12].
Mood was measured using the 30-item total mood distur-
bance scale of the Profile of Mood States brief form [13]. It
aggregated feelings related to depression, anxiety, fatigue, an-
ger, confusion, and vigor (e.g., “sad,” “tense,” “energetic”) oc-
curring “over the past two weeks, including today.” Response
options were from 0 (not at all) to 4 (extremely) and were
summed after first reversing the scores on the vigor-related
items. Internal consistencies across factors were Cronbach
α = 0.84–0.95 [13] and α = 0.77–0.87 in the present sample.
Test-retest reliabilities averaged 0.69 [13].
Combined servings of fruits and vegetables “consumed in a
typical day” (fruit/vegetable intake) were measured through 2
items corresponding to portion sizes indicated by the U.S.
Department of Agriculture (e.g., 118 mL canned peaches or
carrots) [14]. Adequate validity was indicated through signif-
icant (p values < 0.001) correspondences with the Block Food
Frequency Questionnaire, other comprehensive food frequen-
cy recalls, and energy consumption [15, 16].
Weight was measured in kg using a recently calibrated
digital scale (Health-o-meter model 800KL, Buffalo Grove,
IL, USA). Heavy clothing was removed prior to recording
the mean of 2 consecutive measurements.
Procedure
All treatment processes were administered by existing staff of
community health-promotion centers. They had appropriate
national certifications and 2.5 days of training in the study’s
protocol components. Participants were initially placed into
the 6-session exercise support component [6]. The first 4 of
these 45-min, 1-on-1 sessions were held within the first
3 months, then were spaced by 2 months to conclude at
6.5 months. The primary emphasis was enabling participants’
use of the self-regulatory skills of cognitive restructuring,
stimulus control, relapse prevention (e.g., planning for viola-
tions in agreed-upon behaviors), dissociation from exercise-
induced discomfort, contingency-based self-reward,
recruiting social supports, and behavioral contracting to over-
come their barriers to regular exercise. Although exercise
types were chosen by each participant, durations and intensi-
ties were checked by instructors for palatability using a previ-
ously validated method of assessing pre- to post-exercise
changes in feeling states [6]. Long-term goals were also self-
set by participants, with instructors actively helping to docu-
ment and track proximal (i.e., next session) goals. A minimum
of 3 exercise sessions per week was assigned, and tracked by
both electronic and self-report methods.
Based on study entry time, either 3 or 4 exercise-support
sessions were completed prior to (a) being trained how to log
energy and fruit/vegetable intake each day, (b) learning her
daily caloric intake limit (e.g., 1500 kcal/day for a weight of
80–100 kg), and (c) initiating the group nutrition-change com-
ponent that met for 60 min every 2 weeks. Nearly all of the
nutrition session time was spent adapting the previously
learned self-regulatory skills from an exercise to a controlled
eating context using brief lectures and structured activities.
The primary nutrition target was increasing fruit and vegetable
intake. The combined exercise support and nutrition-change
intervention terminated 56 weeks post-initiation.
Fidelity checks on approximately 20% of sessions indicat-
ed strong protocol compliance, requiring only minor adjust-
ments by study staff. Manipulation checks indicated compli-
ance with the mandated exercise amount (i.e., estimated mean
of 4.4 exercise sessions per week).
Data Analyses
Because there was no systematic bias in the 10% of missing
data [17], the expectation-maximization algorithm [18] was
used for imputation in the intention-to-treat analytic approach.
Tolerance scores ranged from 0.50 to 0.99, which indicated
acceptable collinearity in the data. Because directionality in
the relationships between the variables being tested was pre-
viously indicated [6], statistical significance was set at α ≤
0.05 (1-tailed) for those analyses, and α ≤ 0.05 (2-tailed) for
all others. SPSS version 22.0 (IBM, Armonk, NY), incorpo-
rating the PROCESS macro-instructional software version 2,
models 4 and 28 [11], was used for the statistical analyses. The
bias-corrected bootstrapping procedure incorporated 20,000
resamples of the data.
Dependent t tests assessed within-group, and overall,
changes in study variables over the incorporated temporal
intervals. Mixed-model repeated measures ANOVAs, with
Bonferroni-adjusted follow-up t tests, assessed group differ-
ences in those changes. Effect sizes were calculated as
 Int.J. Behav. Med.

Cohen’s d and partial eta squared (η2partial), respectively,
where small–large effects are, by convention, 0.20–0.80 and
0.01–0.14.
Using aggregated data, regression models incorporated a
lagged variable approach [19] where change in a predictor
from a preceding interval (e.g., baseline–month 3) was entered
to predict longer-term change (e.g., baseline–month 6).
Following recent suggestions [11], total effects between inde-
pendent and dependent variables (c paths) were not part of the
mediation analyses. Within the first mediation analysis
(Fig. 1a), whether change in SR-eating significantly mediated
the prediction of change in fruit/vegetable intake from SR-
exercise change was assessed. Within a moderated mediation
analysis (Fig. 1b), it was next assessed whether mood change
(and mean mood score) significantly moderated the prediction
of SR-eating change by change in SR-exercise and whether
baseline weight significantly moderated the prediction of fruit/
vegetable intake change by SR-exercise and SR-eating chang-
es. Significance of moderation is determined by the indepen-
dent variable × moderator interaction term’s prediction of the
dependent variable. The prediction of SR-eating at month 24
by change in SR-eating from baseline to month 6 was next
evaluated. Finally, it was assessed whether change in fruit/
vegetable intake over 24 months significantly mediated the
prediction of 24-month change in weight by change in fruit/
vegetable intake from baseline to month 6 (Fig. 1c). Results of
the overall model (R2), the unstandardized beta (B) and its
associated standard error (SEB), and the standardized beta
(β) are reported appropriate. Because bootstrap resampling
was incorporated, 95% confidence intervals (CI) were
assessed as the significance of mediation [11].
Results
There was no significant difference at baseline in any study
variable (p values > 0.20). Significant improvements in SR-
exercise, SR-eating, mood, and fruit/vegetable intake were
found in each group, and overall (Table 1). Changes were
significantly greater in the high weight-loss group on changes
in SR-eating (F1, 98 = 6.77, p = 0.010, η2partial = 0.07) and fruit/
vegetable intake (F1, 98 = 7.99, p = 0.006, η2partial = 0.08).
Weight loss was significant not only in groups over 6 months
but also over 24 months in only the high weight-loss group
(Table 1).
Intercorrelations of aggregated data indicated bivariate re-
lationships between changes in self-regulation and mood, and
fruit/vegetable intake and weight, were each in the expected

Table 1 Descriptive statistics and changes in study measures, by group

Baseline Time 2 ΔBaseline-time 2 Month 24 ΔBaseline-month 24

Self-regulation-exercise
High weight loss 21.14 ± 5.97 33.69 ± 4.41 12.55 ± 7.04*(2.10)
Low weight loss 21.45 ± 5.74 32.04 ± 4.23 10.59 ± 6.51*(1.84)
Overall 21.29 ± 5.83 32.88 ± 4.38 11.59 ± 6.82*(1.99)
Self-regulation-eating
High weight loss 22.37 ± 5.93 33.59 ± 3.78 11.22 ± 6.69*(1.89) 31.82 ± 6.00
Low weight loss 23.67 ± 4.91 31.50 ± 4.51 7.83 ± 6.32*(1.59) 28.18 ± 5.16
Overall 23.01 ± 5.47 32.57 ± 4.26 9.56 ± 6.70*(1.76) 30.04 ± 5.87
Mood
High weight loss 25.61 ± 15.66 6.50 ± 12.47 − 19.11 ± 16.58*(1.22)
Low weight loss 25.14 ± 15.51 9.06 ± 14.63 − 16.08 ± 14.62*(1.04)
Overall 25.38 ± 15.51 7.76 ± 13.57 − 17.63 ± 15.64*(1.14)
Fruit/vegetable serving/day
High weight loss 3.85 ± 2.00 6.56 ± 2.04 2.71 ± 2.12*(1.36) 6.22 ± 1.92 2.36 ± 2.09*(1.18)
Low weight loss 3.84 ± 1.63 5.91 ± 2.11 2.07 ± 2.21*(1.27) 5.08 ± 1.69 1.24 ± 1.86*(0.76)
Overall 3.85 ± 1.82 6.24 ± 2.09 2.40 ± 2.17*(1.32) 5.66 ± 1.89 1.82 ± 2.05*(1.00)
Weight (kg)
High weight loss 95.16 ± 12.59 87.09 ± 12.36 − 8.07 ± 4.64*(0.64) 85.15 ± 12.52 − 10.01 ± 7.13*(0.80)
Low weight loss 93.89 ± 11.18 90.35 ± 11.34 − 3.54 ± 3.06*(0.32) 93.78 ± 11.29 − 0.10 ± 3.17(0.09)
Overall 94.54 ± 11.88 88.69 ± 11.92 − 5.85 ± 4.54*(0.49) 89.38 ± 12.64 − 5.16 ± 7.44*(0.43)

High weight loss group n = 51. Low weight loss group n = 49. Overall group N = 100
Time 2 = month 3 for self-regulation-exercise and mood, and month 6 for self-regulation-eating, fruit/vegetable intake, and weight. Values are given as
mean ± standard deviation. Δ = change in the score during indicated times. Cohen’s d values for within-group changes are given in parentheses
*p ≤ 0.001
Int.J. Behav. Med.

Moderation of path c′

baseline-month 6, b-24 = baseline-month 24. Path a = independent variable→mediator, path b = mediator→dependent variable, path c′ = independent variable→dependent variable, controlling for the
Results are based on a bootstrapping method incorporating 20,000 resamples [11]. Δ = change in the score during indicated times. Moodmean = baseline+month 3/2. b-3 = Δbaseline-month 3, b-6 =
direction and significant (see supplementary data). Change in
SR-eating significantly mediated the relationship between

Weightbaseline

Weightbaseline
changes in SR-exercise and fruit/vegetable intake (B = 0.05,

0.01 ± 0.01

0.01 ± 0.01
Beta±SE
SEB = 0.03, 95% CI = 0.01, 0.09) (Table 2). That overall
model was significant (R2 = 0.11, F2, 97 = 6.23, p = 0.003).
When entered into the previous equation, both mood change
(B = − 0.01, SEB = 0.01, p = 0.029) and mean mood score

−0.42 ± 0.42

−0.42 ± 0.42
0.03 ± 0.04

0.34 ± 0.88
(B = −0.01, SEB = 0.01, p = 0.019) significantly moderated

Beta±SE
Path c′
the SR-exercise–SR-eating change relationship; and weight
at baseline significantly moderated the relationship between
changes in SR-eating and fruit/vegetable intake (B = − 0.02,

Moderation of path b
SEB = 0.01, p = 0.043) (Table 2). Those overall models were
significant (R2 = 0.15, F5, 94 = 3.25, p = 0.010). Change

− 0.02 ± 0.01*

− 0.02 ± 0.01*
Weightbaseline

Weightbaseline
in SR-eating from baseline to month 6 significantly predicted
SR-eating at month 24 (B = 0.17, SEB = 0.09, β = .20, p =

Beta±SE
0.025, 95% CI = 0.03, 0.32). Change in fruit/vegetable intake
from baseline to month 24 significantly mediated the relation-
ship between changes in fruit/vegetable intake from baseline

− 3.84 ± 0.93***
to month 6 and change in weight from baseline to month 24

0.09 ± 0.04*

0.88 ± 0.46*

0.88 ± 0.46*
(B = − 2.27, SEB = 1.10, 95% CI = − 4.73, − 0.95) (Table 2).

Beta±SE
That overall model was significant (R2 = 0.21, F2, 97 = 12.53,

Path b
p < 0.001).
Moderation of path a

Discussion

− 0.01 ± 0.01**
− 0.01 ± 0.01*
ΔMoodb-3

Moodmean
The finding that change in SR-eating was significantly
Beta±SE

greater in participants successful with sustained weight

loss confirmed the need for further evaluation of the dy-
namic processes related to this important variable within
0.58 ± 0.08***

0.43 ± 0.12***

0.64 ± 0.09***

0.59 ± 0.07***
treatments. Through the present attention given to that
quest, findings suggested that improved exercise-related
Beta±SE
Path a

self-regulation carries over to better self-regulation related

Results of mediation and moderated mediation analyses (N = 100)

to healthy eating, which is largely sustained over the long

term (here, over 2 years). Although considerable replica-
ΔFruit/vegetableb-6

ΔFruit/vegetableb-6

ΔFruit/vegetableb-6
Dependent variable

tion across sample types is required to increase confidence

in findings, the current results support a strong self-
ΔWeightb-24

regulatory focus within behavioral treatments beginning

with an aim toward increasing exercise behaviors. This
appeared increasingly relevant as initial weights were
higher. The present finding that mood significantly affects
ΔFruit/vegetableb-24

self-regulatory progress, while previously posited [9],

should be further tested in field settings for both its im-
ΔSR-eatingb-6

ΔSR-eatingb-6

ΔSR-eatingb-6

portant theoretical and treatment implications. The finding

*p ≤ 0.05, **p ≤ 0.01, **p ≤ 0.001
Mediator

that short-term improvements in fruit/vegetable intake (a

proxy for the health of the overall diet) predicted long-
term changes suggests the present treatment paradigm can
have unusually favorable long-term effects [1]. The treat-
Independent variable

ΔFruit/vegetableb-6

ment’s foci of self-regulation, and a seemingly manage-

ΔSR-exerciseb-3

ΔSR-exerciseb-3

ΔSR-exerciseb-3

able 3 exercise sessions per week to induce mood im-

provements, also purposefully addressed adherence—a
mediator
Table 2

consistent problem in real-world settings that use primar-

ily educational (typically atheoretical) methods [6].

Beyond the aforementioned need for replication, limita-
tions of this research include incorporation of a volunteer
sample that might have been highly motivated, incomplete
consideration of constructs related to social cognitive theory
[3] (and other accepted theories), and lack of control over
social support and expectation effects. Additionally, carry-
over effects to eating-related self-regulation when only
exercise-focused self-regulation is addressed remain un-
known. A strength of the study, however, was its field setting
and longitudinal design that increased external validity and
rapid generalization of findings to treatments. Further
theory-based research remains needed so that an improved
understanding of the dynamics of psychosocial changes with-
in the weight-loss treatment process can positively affect fu-
ture behavioral interventions—especially those applicable for
large-scale dissemination.
Compliance with Ethical Standards
All study procedures were in accordance with the ethical standards of the
institutional and/or national research committee and the 1964 Helsinki
declaration and its later amendments or comparable ethical standards.
Conflict of Interest The authors declare that they have no conflict of
interest.
Informed Consent Informed consent was obtained from all participants
included in the study.
Publisher’s Note Springer Nature remains neutral with regard to jurisdic-
tional claims in published maps and institutional affiliations.
References
1. Dombrowski SU, Knittle K, Avenell A, Araújo-Soares V, Sniehotta
FF. Long term maintenance of weight loss with non-surgical inter-
ventions in obese adults: systematic review of meta-analyses of
randomised controlled trials. BMJ. 2014;348:g2646.
2. Baranowski T, Lin LS, Wetter DW, Resnicow K, Hearn MD.
Theory as mediating variables: why aren’t community interven-
tions working as desired? Ann Epidemiol. 1997;7(suppl):S89–
S95, 7.
Int.J. Behav. Med.
3. Bandura A. Social foundations of thought and action: a social cog-
nitive theory. Englewood Cliffs, NJ: Prentice Hall; 1986.
4. Aljadani HM, Patterson A, Sibbritt D, Hutchesson MJ, Jensen ME,
Collins CE. Diet quality, measured by fruit and vegetable intake,
predicts weight change in young women. J Obes. 2013:525161.
5. Svetkey LP, Stevens VJ, Brantley PJ, Appel LJ, Hollis JF, Loria
CM, et al. Comparison of strategies for sustaining weight loss.
JAMA. 2008;299:1139–48.
6. Annesi JJ. Supported exercise improves controlled eating and
weight through its effects on psychosocial factors: extending a sys-
tematic research program toward treatment development. Perm J.
2012;16(1):7–18.
7. Oaten M, Cheng K. Longitudinal gains in self-regulation from reg-
ular physical exercise. Br J Health Psychol. 2006;11:717–33.
8. Mata J, Silva MN, Vieira PN, Carraça EV, Andrade AM, Coutinho
SR, et al. Motivational “spill-over” during weight control: increased
self-determination and exercise intrinsic motivation predict eating
self-regulation. Health Psychol. 2009;28:709–16.
9. Gendolla GHE, Brinkman K. The role of mood states in self-regu-
lation: effects on action preferences and resource mobilization. Eur
Psychol. 2005;10:187–98.
10. Wilson GT, Brownell KD. Behavioral treatment for obesity. In:
Fairburn CG, Brownell KD. Eating disorders and obesity. 2nd ed.
New York, NY: Guilford; 2002. pp. 524–528.
11. Hayes AF. Introduction to mediation, moderation, and conditional
process analysis: a regression-based approach. New York, NY:
Guilford; 2013.
12. Annesi JJ, Marti CN. Path analysis of cognitive-behavioral exercise
treatment-induced changes in psychological factors leading to
weight loss. Psychol Health. 2011;26:1081–98.
13. McNair DM, Heuchert JWP. Profile of Mood States technical up-
date. North Tonawanda, NY: Multi-Health Systems; 2009.
14. U.S. Department of Agriculture. MyPlate and historical Food
Pyramid resources. Beltsville, MD: National Agricultural Library;
2017.
15. Block G, Hartman AM, Dresser CM, Carroll MD, Gannon J,
Gardner L. A data-based approach to diet questionnaire design
and testing. Am J Epidemiol. 1986;124:453–69.
16. Mares-Perlman JA, Klein BEK, Klein R, Ritter LL, Fisher MR,
Freudenheim JL. A diet history questionnaire ranks nutrient intakes
in middle-aged and older men and women similarly to multiple
food records. J Nutr. 1993;123:489–501.
17. White IR, Horton NJ, Carpenter J, Pocock SJ. Strategy for intention
to treat data in randomized trials with missing outcome data. BMJ.
2011;342:d40.
18. Schafer JL, Graham JW. Missing data: our view of the state of the
art. Psychol Methods. 2002;7:147–77.
19. Cromwell JB, Hannan MJ, Labys WC, Terraza M. Multivariate
tests for time series models. Thousand Oaks, CA: Sage; 1994.