Content uploaded by Lynn Darby
Author content
All content in this area was uploaded by Lynn Darby
Content may be subject to copyright.
The Relationship Between Self-Monitoring, Outcome Expectancies, Difficulties With Eating
and Exercise, and Physical Activity and Weight Loss Treatment Outcomes
Robert A. Carels, Ph.D., M.B.A., Lynn A. Darby, Ph.D., Sofia Rydin, M.A., Olivia M. Douglass, M.A.,
Holly M. Cacciapaglia, M.A., and William H. O’Brien, Ph.D.
Bowling Green State University
ABSTRACT
Background/Purpose: During a behavioral weight loss
program (BWLP), weight loss and exercise can vary consider
-
ably from week to week. Weekly fluctuations in outcome expec
-
tancies and perceived difficulties with eating and exercise may
be associated with weekly variability in weight loss and exer
-
cise. Also, inconsistent self-monitoring of exercise may be asso
-
ciated with poor weight loss and physical activity treatment out
-
comes. Methods: Forty obese, sedentary participants completed
a 6-month BWLP. Body weight, outcome expectancies, and diffi
-
culties with eating and exercise were assessed weekly. Weekly
self-monitoring of exercise was computed from physical activity
diaries. Physical activity, VO
2
max, and caloric intake were as
-
sessed pre- and posttreatment. Results: Within-subjects analy-
ses indicated that participants exercised less during weeks that
participants reported greater difficulties with exercise, relative
to weeks participants reported fewer difficulties. Participants
lost significantly more weight during weeks that participants re-
ported more positive outcome expectancies and greater difficul-
ties with exercise, compared to weeks participants reported less
positive outcome expectancies and fewer difficulties with exer-
cise. Consistent self-monitoring of exercise was associated with
fewer difficulties with exercise and greater exercise and weight
loss. Conclusions: Interventions that are targeted to increase
self-monitoring and to improve transient difficulties with exer
-
cise and diminished outcome expectancies may improve BWLP
treatment outcomes.
(Ann Behav Med 2005, 30(3):182–190)
INTRODUCTION
In the U.S. population, nearly 60% of adults are overweight
(body mass index [BMI] ≥ 25 kg/m
2
), and 25% are obese
(BMI ≥ 30 kg/m
2
) (1). Obesity is a major cause of illness and
death in the United States and may soon replace smoking as the
primary cause of preventable death (2,3). Despite the many po
-
tential health benefits of weight loss, people have difficulty los
-
ing weight and maintaining weight loss (4,5). In addition, peo
-
ple have difficulties initiating and maintaining regular exercise
(5), a factor known to contribute to successful weight loss and
maintenance (6–8).
Although participants in behavioral weight loss programs
(BWLPs) lose, on average, 9% of their total body weight by the
end of treatment (9), their weight loss throughout treatment is
rarely uniform (10). Weight plateaus and occasional weight
gains are common. Similarly, the amount of time an individual
exercises from week to week can be quite variable. Even the
most successful BWLP participants can demonstrate variable
weight loss and exercise from week to week throughout treat
-
ment. It is plausible that transient psychosocial factors, such as
difficulties with eating and exercise (e.g., resulting in excess cal
-
orie consumption and diminished exercising), may contribute to
the commonly observed fluctuations in exercise and weight loss
from week to week.
In light of the difficulties people commonly experience
while attempting to exercise regularly and lose weight, re-
searchers continue to search for factors that are associated with
treatment outcome. For example, Bandura (11) suggested that
personal change would be trivially easy if there were no impedi-
ments or difficulties to surmount. Similarly, judgments about
whether significant improvements in health will occur in re-
sponse to exercise and regular weight loss are likely to influence
whether an individual is motivated to exercise or restrict calories
(i.e., outcome expectancies) (12). Outcome expectancies and
perceived difficulties are prominent factors in several models of
health behavior change, such as social-cognitive theory (12),
and have been significantly associated with regular exercise and
weight loss (13–18).
Although social-cognitive factors (e.g., outcome expectan
-
cies, perceived difficulties) are commonly assessed at the begin
-
ning or the end of an intervention (13–18), it is likely that these
factors will increase and decrease throughout treatment (i.e.,
vary from week to week). As such, participants may feel more
confident that they will achieve their desired weight loss goals,
or they may experience greater difficulties with eating or ex
-
ercise on some weeks more than other weeks. In turn, this vari
-
ability in outcome expectancies and perceived difficulties with
eating and exercise should be significantly associated with fluc
-
tuations in weekly weight loss and exercise levels. To our
knowledge, no research has examined the relationship between
transient psychosocial factors and weekly exercise and weight
loss during a BWLP.
In light of the aforementioned discussion, this investigation
was designed to examine whether weekly changes in outcome
expectancies and difficulties with eating and exercise through
-
out treatment were associated with weekly changes in weight
and exercise (i.e., within-subjects relationships). It was hypoth
-
esized that weeks characterized by greater weight loss would be
associated with positive outcome expectancies, fewer difficul
-
182
Reprint Address: R. Carels, Ph.D., M.B.A., Department of Psychology,
Bowling Green State University, Bowling Green, OH 43403. E-mail:
rcarels@bgnet.bgsu.edu
© 2005 by The Society of Behavioral Medicine.
ties with eating and exercise, and higher levels of weekly exer
-
cise. In addition, it was hypothesized that weeks characterized
by greater exercise would be associated with fewer difficulties
with exercise and more positive outcome expectancies.
As noted earlier, within-subjects weekly variation in ex
-
pectancies, difficulties with eating and exercise, exercise, and
weight loss may be significantly related. However, it is also im
-
portant to compare whether an effect is similar at the aggregate
(i.e., between-subject) level. Therefore, consistent with prior re
-
search on exercise and weight loss (6–8), it was hypothesized
that greater exercise and cardiorespiratory fitness would be as
-
sociated with greater weight loss by the end of treatment. It also
was hypothesized that greater self-reported difficulties with eat
-
ing and exercise and negative outcome expectancies would be
associated with lower levels of physical activity and cardio
-
respiratory fitness and less weight loss by the end of treatment.
Another behavior known to promote weight loss and main
-
tenance and is strongly encouraged in BWLPs is self-monitor
-
ing (7). Although the mechanism through which self-monitor
-
ing affects weight loss is unclear, self-monitoring is likely to
increase participant self-awareness of their targeted behaviors.
For example, an individual who regularly records and monitors
exercise is likely to (a) become aware of weeks when exercise is
diminishing, (b) implement strategies to counteract compliance
problems, and (c) grow in their understanding of the association
between regular exercise and weight loss. Despite the potential
importance of regular self-monitoring of exercise in weight loss
treatment outcomes (7), little research has examined the factors
that might contribute to poor compliance with exercise self-
monitoring. It was hypothesized that greater difficulties with ex-
ercise and lower outcome expectancies would be associated
with inconsistent self-monitoring of exercise. In turn, inconsis-
tent self-monitoring of exercise was hypothesized to be associ
-
ated with lower exercise levels and less weight loss during and
following treatment.
METHODS
Participants
Participants were 53 obese, sedentary men and women ran
-
domly assigned to receive a BWLP or BWLP plus glycemic in
-
dex (GI) education. Forty participants completed the interven
-
tion. Participants’ mean age was 43.4 (SD = 9.4). Eighty-three
percent were women, 75.5% worked full time, 77.4% had an an
-
nual income that exceeded $30,000 per year, and 41.5% had at
least a baccalaureate college degree. Participants were recruited
through local advertisements (e.g., newspapers) and e-mail so
-
licitation to university employees. Individuals were included in
the investigation if they were (a) obese (BMI ≥ 30 kg/m²), (b)
sedentary (i.e., not participating in a program of physical condi
-
tioning two or more times per week for at least 20 min per ses
-
sion), (c) willing to accept random assignment, (d) nonsmokers,
(e) able to provide informed consent, and (f) cleared for partici
-
pation by their primary physician. Participants were excluded
from participation if they had (a) past or current cardiovascular
disease (e.g., myocardial infarction, stroke) determined from
medical history; (b) musculoskeletal problems that would pre
-
vent participation in moderate levels of physical activity (e.g.,
self-reported osteoporosis); (c) a history of insulin-dependent
diabetes (self-reported); (d) resting blood pressure greater than
or equal to 160/100 mmHg (assessed during screening); and (e)
a life-limiting or complicated illness including cancer, renal
dysfunction, hepatic dysfunction, or dementia. All procedures
received human subjects review board approval and were fol
-
lowed without deviation.
Study Design
At baseline, participants completed assessments of body
weight, leisure time physical activity, cardiorespiratory fitness,
and caloric intake. All assessments were obtained again at the
conclusion of the intervention. At the beginning of class each
week, outcome expectancies for achieving weight loss, eating
and physical activity goals, and difficulties with eating and exer
-
cise during the prior week were assessed. Participants were
weighed weekly at the end of each class. Participants also re
-
corded the duration of daily planned exercise in physical activity
diaries throughout the intervention.
Participants underwent a 6-month weight loss and physi
-
cal activity intervention based on the LEARN program (19).
The LEARN program is designed to encourage gradual weight
loss, increased physical activity, and a progressive decrease in
energy and fat intake through permanent lifestyle changes. Ad-
ditional information on the LEARN program can be found at
http://www.learneducation.com. Half of the participants were
randomly assigned to receive the weight loss and physical ac-
tivity intervention alone, whereas the other half received the
weight loss and physical activity program augmented by educa-
tion on the GI of foods. All participants in the GI condition re-
ceived a copy of the book The New Glucose Revolution (20) and
weekly instruction on the GI. The GI is a physiologically based
measure of how quickly carbohydrates are broken down during
digestion (21). It has been reported that a low GI diet may con
-
tribute to weight loss by influencing insulin levels, appetite, and
metabolism (21).
A clinical health psychologist and graduate students in
clinical psychology administered the weekly sessions in small
groups (i.e., 8–12 participants). Participants in the BWLP met
weekly for 60- to 75-min sessions, whereas participants in the
BWLP plus GI education met for 90 to 120 min.
Measures
Cardiorespiratory fitness. To determine functional capac
-
ity or maximal oxygen uptake (VO
2
max), each participant com
-
pleted a submaximal graded exercise test (22) using the modi
-
fied Balke (23) protocol walking on a treadmill. Heart rate via a
12-lead EKG was recorded at the end of each stage. The test was
discontinued if any test termination criteria, as described by the
American College of Sports Medicine, were present during the
test (22). Functional capacity was predicted from the regression
equation for the relationship between submaximal VO
2
and
heart rate at two or more submaximal workloads (22). Although
Volume 30, Number 3, 2005 Self-Monitoring and Weight Loss 183
some error is introduced because this is an extrapolation of
submaximal data, the utility of this measure is found in its use
for monitoring training changes over time. Participants were
tested at pre- and postintervention. Data were unavailable for 3
participants. Also, as a safety precaution, exercise testing was
not performed on 2 men who presented at baseline with im
-
paired fasting blood glucose (> 110 mg/dl) (22).
Physical activity questionnaire. To assess leisure time physi
-
cal activity, participants completed the Paffenbarger Physical
Activity Questionnaire (PPAQ) (24). The PPAQ asked partici
-
pants about stairs climbed, blocks walked, and recreation and
sports played through open-ended questions concerning fre
-
quency and duration of activity. The PPAQ provided an estimate
of energy expenditure per week in kilocalories and has been
used in numerous research investigations (25).
Physical activity logs and self-monitoring. Exercise partic
-
ipation rates were determined for all participants. Participants
were instructed to complete a daily exercise diary describing the
exercise type and duration. Exercise diaries were collected ev
-
ery 4 to 5 weeks during the intervention. This method of daily
physical activity assessment has been used successfully in pre-
vious weight loss research (26). Weekly estimates of the total
time spent in planned exercise were calculated. Participants
were encouraged to complete physical activity diaries through-
out the program (i.e., 21 weeks of physical activity diaries were
collected). Self-monitoring of exercise was defined as the num-
ber of weeks that physical activity diaries were completed by
each participant.
Dietary assessment. Participants recorded food intake over
4 days (2 weekends and 2 weekdays) at pre- and posttreatment.
Oral and written instructions on food measurement estimation
were provided to participants. Estimates for total calories were
calculated using NutriBase 2001 Professional Nutrition soft
-
ware (CyberSoft, Phoenix, AZ).
Body weight. Body weight was measured using a digital
scale (BF–350e; Tanita, Arlington Heights, IL) to the closest 0.1
lb, and height was measured in inches to the closest 0.5 in. BMI
was calculated as kg/m².
Outcome expectancies and difficulties with eating and exer
-
cise. Outcome expectancies and difficulties with eating and ex
-
ercise were assessed weekly throughout the intervention. Each
week participants completed a 13-item questionnaire designed
to assess outcome expectancies and difficulties in achieving eat
-
ing and physical activity goals (27).
Eleven items were assessed on a 9-point scale ranging from
1(strongly disagree)to9(strongly agree). Three items assess
-
ing self-confidence asked participants to respond to the state
-
ment “I feel confident that I can … ” (a) achieve my weight loss
goals, (b) achieve my eating goals, and (c) achieve my exercise
goals. Two items assessing self-control asked participants to re
-
spond to the statement “I have self-control over … ” (a) my eat
-
ing habits and (b) my physical activity habits. Four items asked
participants to respond to the statement “I have difficulty … ”
(a) modifying my eating habits, (b) following my physical activ
-
ity plan, (c) keeping track of my eating habits, and (d) keeping
track of my physical activity habits. Two items asked partici
-
pants to respond to the statement “I was unable to … ” (a) eat in
a manner that I would have liked and (b) be physically active in a
manner that I would have liked. These were rated on a 6-point
scale ranging 1 (no problems), 2 (1–2 times), 3 (3–4 times), 4
(5–6 times), 5 (7–8 times), and 6 (> 8 times).
A factor analysis was conducted to evaluate the adequacy of
the questionnaire and to identify the underlying structure of the
inventory.
1
A principle-components approach with varimax ro
-
tation was conducted to accomplish these ends. The eigenvalue
for factor retention was set at 1. An item was judged to be ade
-
quate and was retained if two conditions were met. First, the
item must have attained a factor loading of at least .40 on a sin
-
gle factor. Second, the difference between any two-factor load
-
ings for an item must have exceeded .30.
A three-factor solution was determined. The first factor as
-
sessed outcome expectancies (i.e., self-confidence in meeting
weight loss and eating outcomes; two items, α = .91). The sec
-
ond factor assessed difficulties with eating habits (four items, α
= .71). The third factor assessed difficulties with physical activ-
ity habits (five items, α = .85). Two items (I feel confident that I
can achieve my exercise goals; I have self-control over my phys-
ical activity habits) did not meet the conditions for item ade-
quacy indicated earlier and were not retained for data analyses.
Data analyses. There were no significant differences be-
tween the BWLP and BWLP plus GI education on any treatment
outcomes (28). Therefore, the treatment groups were combined
for data analyses. Pre- and posttreatment change in cardio
-
respiratory fitness and leisure time physical activity were evalu
-
ated using repeated measures analysis of variance (ANOVA).
Average weekly exercise (from the physical activity logs) was
also evaluated using separate repeated measures ANOVA across
three time periods (i.e., exercise during the first 7 weeks, middle
7 weeks, and final 7 weeks of the intervention were compared).
Pearson correlations and t tests were used to examine the
relationships between the demographic variables (i.e., age, in
-
come, education, race, gender) and each of the variables at base
-
line: outcome expectancies, difficulties with eating, difficulties
with exercise, self-monitoring of exercise, leisure time physical
activity, cardiorespiratory fitness, and body weight. Treatment
dropouts were excluded from data analyses. Pearson correla
-
tion coefficients also were used to examine the between-sub
-
ject associations among outcome expectancies, difficulties with
eating, difficulties with exercise, caloric intake, weight loss,
weekly exercise, leisure time physical activity, and predicted
VO
2
max.
184 Carels et al. Annals of Behavioral Medicine
1
Factor analysis was conducted with questionnaires from 144 par
-
ticipants collected across three BWLPs. Fifty-three participants were
included from this investigation. Additional participants (n = 91) were
selected from similar BWLPs (26,28). Participants across programs
were obese, sedentary, and participating in 6-month BWLPs.
Generalized estimating equations (GEEs; SAS Institute,
Cary, NC) were used to examine the within-subjects association
between the dependent variable (weekly changes in weight) and
independent variables (outcome expectancies, difficulties with
eating, difficulties with exercise, and level of exercise during the
prior week). The change in caloric intake from pre- to post
-
treatment was included as a covariate. Change in weight was
computed by subtracting each participant’s current weight from
the participant’s weight from the preceding week. GEEs were
also used to examine the association between the dependent
variable (level of exercise) and the independent variables (out
-
come expectancies, difficulties with eating, and difficulties with
exercise).
GEEs allow for multiple observations per participant while
controlling for autocorrelation (29). GEEs use a multiple step
maximum likelihood approach to estimation and testing. An es
-
sential feature of this technique is that it is possible to recognize
that repeated measures data have two random components: One
is due to the sampling of persons, and the other is due to the sam
-
pling of repeated measurements within persons. To make cor
-
rect inferences from repeated measures data, both sources of
variance need to be considered.
The within-subjects variables (i.e., outcome expectancies,
difficulties with exercise, difficulties with eating, duration of
weekly exercise, and weekly change in weight) represent the
within-subjects variability in the reported constructs. This is
useful for examining the weekly relationships among variables
of interest, such as weekly outcome expectancies, difficulties
with exercise, difficulties with eating, changes in weight, and
level of exercise. It is often interesting to inquire whether an ef-
fect is similar at the weekly (i.e., within-subjects) and aggregate
(i.e., between-subject) levels, because the relationship between
variables may be different at different levels of analysis (30).
Self-monitoring of exercise was defined as the number of
weeks that physical activity diaries were completed by each par
-
ticipant. Pearson correlation coefficients were used to examine
the between-subject association between self-monitoring and
outcome expectancies, difficulties with eating, difficulties with
exercise, caloric intake, weight loss, weekly exercise, leisure
time physical activity, and predicted VO
2
max. An ANOVA was
used to compare consistent (missed no more than 1 week of
self-monitoring) versus inconsistent self-monitoring on weight
loss, exercise, leisure time physical activity, predicted VO
2
max,
outcome expectancies, difficulties with exercise, difficulties
with eating, and caloric intake.
Some participants failed to complete physical activity dia
-
ries. Eight participants (15.1%) who dropped out of treatment
and 5 participants (9.4%) who remained in treatment failed to
complete physical activity diaries. Participants (N = 40) pro
-
vided 626 weekly physical activity diary entries (M = 15.6 per
participant, SD = 6.2), 469 weekly assessments of outcome ex
-
pectancies and difficulties with eating and exercise (M = 11.7
per participant, SD = 4.8), and 436 changes in body weight en
-
tries (M = 10.9 per participant, SD = 5.1). Mean ratings were
11.8 (SD = 1.9) for outcome expectancies, 18.6 (SD = 3.4) for
difficulties with eating, and 25.7 (SD = 7.7) for difficulties with
exercise. There were fewer body weight entries than physical
activity diary entries, outcome expectancies, and difficulties
with eating and exercise entries because body weight entries
represent a difference score. Therefore, no body weight data
were available from the 1st week of the program, and change
scores could not be computed on weeks following a participant
absence. Other than GEEs, additional statistical procedures
were conducted using SPSS for Windows
®
Version 11.0 using
alpha set at .05.
RESULTS
Background Characteristics and Attrition
Age, income, and education were not significantly asso
-
ciated with outcome expectancies, difficulties with eating, diffi
-
culties with exercise, self-monitoring, caloric intake, or any
physical activity and cardiorespiratory fitness variables. Men
weighed significantly more than women at baseline, F(1, 40) =
19.30, p < .01, and consumed significantly more daily calories at
baseline, F(1, 52) = 9.07, p < .01. Of the 53 participants, 40
(75.5%) completed the study. Participants who completed the
program (N = 40) had significantly greater education, F(1, 52) =
5.88, p < .05, and had significantly greater baseline leisure time
physical activity, F(1, 52) = 3.99, p ≤ .05, than individuals who
did not complete the program. There were no other significant
differences on demographics or baseline physical activity char-
acteristics between program completers and noncompleters.
Changes in Cardiorespiratory Fitness, Physical
Activity, Weekly Exercise, and Caloric Intake
After BWLP
There were significant pre- and posttreatment differences
in maximal oxygen consumption and treadmill time (Table 1).
Average increase in maximal oxygen consumption was 5.74
ml
–1.
min
–1.
kg, F(1, 34) = 10.70, p < .01; average increase in
treadmill time was 3 min 4 sec, F(1, 34) = 37.5, p < .05. There
were significant pre- and posttreatment differences in weekly
leisure time physical activity (i.e., PPAQ; Table 1). Average
weekly kilocalorie expenditure from exercise increased by 950
kilocalories per week by the end of the BWLP, F(1, 39) = 28.45,
p < .05.
As reported elsewhere, from pre- to posttreatment, there
was a significant decrease in total calories, F(1, 39) = 105.2, p <
.01 (28). Weight loss, caloric intake, leisure time physical activ
-
ity, and cardiorespiratory fitness were also analyzed employing
an intent-to-treatment approach. For treatment dropouts, base
-
line data were carried forward to posttreatment. Change in pre-
to posttreatment weight loss, caloric intake, leisure time physi
-
cal activity, and cardiorespiratory fitness remained significantly
different (see Table 1).
Data from the daily physical activity diaries were computed
to indicate weekly duration of time spent in weekly planned ex
-
ercise. For the time spent in planned exercise, there were signifi
-
cant differences among the three time periods, F(2, 25) = 7.4, p
< .01. Paired t tests showed that the time spent in weekly
planned exercise increased significantly, t(31) = 3.28, p < .05,
between the first 7 weeks and the middle 7 weeks of the pro
-
Volume 30, Number 3, 2005 Self-Monitoring and Weight Loss 185
gram, and then showed a nonsignificant change between the
middle 7 weeks and final 7 weeks of the program. The mean
(standard error of mean) time spent in planned exercise was
130.2 (24.7) min per week during the first 7 weeks, 189.1 (18.5)
min during the middle 7 weeks, and 208.4 (24.5) min during the
final 7 weeks.
Outcome Expectancies, Difficulties With Eating
and Exercise, Physical Activity, Cardiorespiratory
Fitness, Caloric Intake, and Weight Loss
(Between-Subject)
Outcome expectancies, difficulties with eating and ex
-
ercise, average level of weekly exercise over the course of treat
-
ment, improvement in leisure time physical activity (i.e.,
PPAQ), cardiorespiratory fitness, and change in caloric intake
were correlated with body weight lost by the end of treatment.
Greater weight loss was significantly associated with fewer dif
-
ficulties with exercise (r = –.45, p < .01; Table 2). In addition,
greater weight loss was significantly associated with greater cal
-
ories expended during leisure time physical activity (r = .71, p <
.01) and a greater duration of weekly exercise (r = .61, p < .01).
Greater weight loss was not significantly associated with out
-
come expectancies, difficulties with eating, and cardiorespira
-
tory fitness.
In addition, outcome expectancies and difficulties with eat
-
ing and exercise were correlated with average weekly exercise
over the course of treatment. Greater weekly exercise was sig-
nificantly associated with fewer difficulties with exercise (r =
–.65, p < .01; Table 2). Additional analyses revealed that indi
-
viduals reporting greater positive outcome expectancies were
also reporting fewer difficulties with exercise (r = –.32, p < .05)
and fewer difficulties with eating (r = –.33, p < .05).
Weekly Fluctuations in Exercise and Its
Association With Outcome Expectancies
and Difficulties in Eating and Exercise
(Within-Subjects)
Lower levels of weekly exercise were significantly associ
-
ated with greater within-subjects difficulties with exercise: Z(1,
40) = 3.53, p < .05 (Table 3). Duration of weekly exercise was
not associated with outcome expectancies or difficulties with
eating.
Weekly Fluctuations in Weight and Its
Association With Exercise, Outcome
Expectancies, and Difficulties in Eating and
Exercise (Within-Subjects)
Within-subjects greater weekly weight loss was signifi
-
cantly associated with greater positive outcome expectancies,
Z(1, 40) = 4.54, p < .01, and greater difficulties with exercise,
Z(1, 40) = 2.81, p < .01 (Table 3). Level of weekly exercise
186 Carels et al. Annals of Behavioral Medicine
TABLE 1
Pretreatment–Postreatment Weight Loss, Physical Activity, and Cardiorespiratory Fitness
40 Participants 53 Participants (Intent-to-Treat)
Pretreatment Posttreatment Pretreatment Posttreatment
Variable M SD M SD Diff. M SD M SD Diff.
PA and fitness
Max VO
2
(ml/kg) 32.3 7.1 38.0 10.8 +5.7* 32.8 7.5 36.8 10.3 +4.0*
Treadmill (min) 8.4 3.1 11.4 3.5 +3.0* 8.3 3.0 10.5 3.5 +2.2*
Leisure PA (kcal/wk) 393 417 1,343 1,001 +950* 307 369 934 1,020 +627*
Caloric intake
Daily calories 2,443 559 1,667 516 –776* 2,413 511 1,812 553 –601*
Weight loss
Weight (kg) 102.8 18.5 95.2 15.3 –7.6* 105.3 20.1 99.4 19.1 –5.9*
Note. Diff. = difference; PA= physical activity.
*p < .05.
TABLE 2
Pearson Correlations of Outcome Expectancies, Difficulties With Eating and Exercise, Caloric Intake,
Weight Loss, Physical Activity, Exercise, and Cardiorespiratory Fitness
Variable Weight Loss Leisure Time PA Exercise Time VO
2
Max Caloric Intake
Outcome expectancies .15 .05 .12 .01 –.01
Difficulties: Exercise –.45** –.32 –.65** .02 –.05
Difficulties: Eating .03 –.05 –.08 .24 .19
*p ≤ .05. **p ≤ .01.
and difficulties with eating were not associated with weekly
weight loss.
The Association Between Self-Monitoring
and Outcome Expectancies, Difficulties
With Eating and Exercise, Physical Activity,
and Weight Loss (Between-Subject)
Participants completed, on average, 15.8 (SD = 6.2) weeks
of self-monitoring of exercise. Greater self-monitoring of ex-
ercise was significantly associated with fewer difficulties with
exercise (r = –.48, p < .01), greater weight loss (r = .44, p < .05),
and greater weekly exercise (r = .52, p < .01). Greater self-moni
-
toring was not associated with leisure time physical activity,
fitness, difficulties with eating, caloric intake, and outcome
expectancies
A comparison of consistent (missed ≤ 1 week of self-moni
-
toring) versus inconsistent (missed > 1 week of self-monitoring)
self-monitors revealed that consistent self-monitors lost signifi
-
cantly more weight by the end of the program, F(2, 40) = 5.28, p
< .05; participated in greater weekly exercise throughout the
program, F(2, 40) = 8.29, p < .01; and reported fewer difficulties
with exercise, F(2, 40) = 7.54, p < .01 (see Figure 1). Only 17 of
40 (43%) participants were consistent self-monitors (missed ≤
1 week of self-monitoring). As shown in Figure 1, consistent
self-monitors lost on average 10.5 kg (SD = 5.1), whereas incon
-
sistent self-monitors lost on average 5.5 kg (SD = 8.6). Simi
-
larly, consistent self-monitors reported exercising 174 min per
week (SD = 102.3) versus 89 min (SD = 85.4) for inconsistent
self-monitors. There were no significant differences between
consistent and inconsistent self-monitors on leisure time physi
-
cal activity, F(2, 40) = 3.52, p = .07; cardiorespiratory fitness,
F(2, 40) = 1.34, p = .26; outcome expectancies, F(2, 40) = 3.00,
p = .09; difficulties with eating, F(2, 40) = 0.12, p = .73; and ca
-
loric intake, F(2, 40) = 0.98, p = .33.
DISCUSSION
The primary aim of this investigation was to examine the re-
lationships among self-monitoring of exercise, outcome ex-
pectancies, difficulties with eating and exercise, physical activ-
ity, cardiorespiratory fitness, and weight loss in a BWLP. On
average, the participants in this investigation lost 7.6 kg (28), in-
creased their cardiorespiratory fitness by 17.6%, increased
weekly exercise by 78 min per week, increased kilocalorie ex-
penditure from leisure time physical activity by 950 calories per
week, and decreased self-reported caloric intake by 776 calories
per day.
Volume 30, Number 3, 2005 Self-Monitoring and Weight Loss 187
TABLE 3
Generalized Estimating Equations for the Relationship
Between Outcome Expectancies, Difficulties with Eating,
Difficulties With Exercise, Weekly Exercise, Caloric Intake,
and Weight Loss
Predictors Estimate SE Z
Weight loss
Outcome expectancies .16 .05 2.98**
Difficulties with eating –.03 .02 0.86
Difficulties with exercise .06 .02 2.81**
Weekly exercise .00 .00 0.88
Caloric intake .00 .00 0.00
Weekly exercise
Outcome expectancies –.65 2.23 0.77
Difficulties with eating –.54 1.23 0.44
Difficulties with exercise –5.20 1.17 4.42**
Note. Z = Z score.
*p < .05. **p < .01.
FIGURE 1 Graphic illustration of the differences between consistent (missed ≤ 1 week of self-monitoring) versus inconsistent (missed > 1 week of
self-monitoring) self-monitoring participants and their mean weight loss, level of weekly exercise, and difficulties with exercise. SM = self-monitor
-
ing. *p < .05.
Data from this investigation examined both within-subjects
and between-subject relationships among outcome expectan
-
cies, difficulties with eating and exercise, and weekly weight
loss and exercise. Because the relationship between two vari
-
ables may be different at different levels of analyses (30), it is in
-
teresting and important to inquire whether an effect is similar at
the weekly (i.e., within-subjects) and aggregate (i.e., between-
subject) levels. For example, research consistently shows that
patients treated with a conventional calorie-reducing diet and
exercise program lose significantly more weight than those who
receive the identical treatment without exercise (e.g., 8,31,32).
All participants in this investigation were encouraged to in
-
crease physical activity. Participants who evidenced the greatest
improvements in leisure time physical activity and higher levels
of weekly exercise lost significantly more weight. It is interest
-
ing to note that the within-subjects association between weight
loss and exercise was not significant. Generally speaking, al
-
though greater levels of exercise were associated with greater
weight loss, the relationship between greater (or lower) levels of
weekly exercise and weight loss was not observed. These find
-
ings suggest that regular exercise may have only a modest abil
-
ity to influence weight on any given week. For example, an indi
-
vidual who weighs 75 kg and engages in 30 min of moderate
intensity physical activity on 5 or more days each week (i.e., the
American College of Sports Medicine’s minimum recommen-
dation for physical activity) is likely to expend the equivalent of
approximately 1,000 kilocalories per week. This level of exer-
cise is approximately equivalent to one third of a pound or less
(e.g., 1 lb = 3,500 calories). There is a limit to the amount of
weight an individual can gain in a given week because he or she
did not exercise. The converse is true for increasing exercise. In
the earlier example, doubling one’s exercise would result in an
additional weight loss of one third pound. Nevertheless, al
-
though exercise is not likely to result in dramatic short-term in
-
creases or decreases in weight, the cumulative effects of regular
physical activity and its association with successful weight loss
is well documented (8,31,32). Therefore, participants might
benefit from the reassurance that although regular exercise
might not produce detectable weekly changes in weight, it
will contribute to successful weight loss and maintenance in the
long run.
Between-subject analyses indicated that greater positive
outcome expectancies were significantly associated with fewer
difficulties with eating and exercise, but outcome expectancies
were not associated with level of exercise or weight loss. In con
-
trast, within-subjects analyses indicated that weeks character
-
ized by more positive outcome expectancies were also charac
-
terized by greater weight loss. Because of the correlational
nature of this investigation, it is not entirely clear whether
weekly outcome expectancies influenced weight change or
weight change influenced outcome expectancies. Although it is
conceivable that participants who possess positive expectancies
for weight loss outcomes will be more motivated to engage in
-
consistent, positive lifestyle changes, we cannot rule out
the possibility that the converse was occurring. Although par
-
ticipants were weighed after completing an assessment of out
-
come expectancies, many participants may have been weighing
themselves regularly at home. Therefore, knowledge of in
-
creases or decreases in weight prior to the weekly assessment of
outcome expectancies may have influenced self-reports of these
constructs.
Greater between-subject difficulties with exercise (not dif
-
ficulties with eating) were associated with less weekly exercise
and weight loss. Surprisingly, greater difficulties with exercise
during any given week (i.e., within-subjects) were associated
with greater weight loss during that week. It is difficult to ex
-
plain why participants tended to lose more weight during weeks
that they report greater difficulties with exercise compared to
weeks that they reported fewer difficulties. It is conceivable that
during weeks that participants were not experiencing difficulties
with exercise they may have been less strict with their regulation
of energy intake. Anecdotally, in the weight loss program, we
often heard participants justify their consumption of additional
calories by stating that they increased their exercise in a given
week. Alternatively, it is plausible that during weeks that a par
-
ticipant was experiencing difficulties with exercise, he or she
may have compensated for the diminished calorie expenditure
through reducing energy intake. During the BWLP, participants
were instructed to closely monitor physical activity and eating
and to make compensatory adjustments in energy expenditure
or energy intake when imbalances were observed. The com-
pensatory reductions in energy intake may have more than offset
the reductions in energy expenditure. Future studies would ben-
efit from collecting data on weekly energy intake to test this
hypothesis.
Self-monitoring of dietary intake is often considered to be
one of the cornerstones of behavioral weight loss treatment, and
is commonly associated with successful weight loss and mainte-
nance (7,33,34). Some research suggests that beyond dietary
self-monitoring, self-monitoring of exercise also contributes to
successful weight loss (35–37). Greater self-monitoring of exer
-
cise was significantly associated with fewer difficulties with ex
-
ercise, higher levels of weekly exercise, and greater weight loss.
For example, consistent self-monitors of exercise lost approxi
-
mately twice as much weight (23 vs. 12 lb), exercised nearly
twice as long each week (174 vs. 89 min), and reported fewer
difficulties with exercise than inconsistent self-monitors.
A number of limitations in this investigation should be
noted. The sample size was modest and included predominantly
White Americans of European descent from the Midwest. These
findings may not be generalizable to urban or ethnically diverse
participants. Also, the well-known metabolic adjustments to
calorie restriction were not assessed in this investigation (38).
Therefore, the influence of metabolic adjustments on the ob
-
served fluctuations in weight is unknown. Because of the
correlational nature of this investigation, it is not clear whether
outcome expectancies or difficulties with eating or exercise
influenced weight change or weight change influenced out
-
come expectancies and difficulties with eating or exercise. An
-
ecdotally, we observed that prior to the weekly “weigh in,” par
-
ticipants often predicted with “reasonable” accuracy whether
they had gained or lost weight during the prior week. It is possi
-
188 Carels et al. Annals of Behavioral Medicine
ble that through the process of self-monitoring, participants be
-
came aware of the impact that difficulties with eating and exer
-
cise had on their weight loss progress. Because participants
dropped out of treatment at various times throughout the inter
-
vention, dropouts (n = 13; 24%) were excluded from data analy
-
ses. The 24% attrition rate in this investigation is higher but con
-
sistent with prior attrition rates for published weight loss
treatment outcome studies (17%) (9). The findings in this inves
-
tigation are not generalizable to treatment dropouts. Finally,
only pre- and posttreatment caloric intake data were collected in
this investigation. The absence of weekly dietary data is a limi
-
tation in this investigation.
The findings in this investigation suggest that although reg
-
ular exercise is likely to contribute to successful weight loss, ex
-
ercise levels on any given week may not be associated with de
-
tectable changes in weight during that week. Participants might
benefit from the reassurance that regular exercise is likely to
contribute to successful weight loss, even if the impact of exer
-
cise on weight loss is not obvious from week to week. Given that
participants lost more weight on weeks that they reported
greater difficulties with exercise, it is possible that difficulties
with exercise resulted in participants being more vigilant in their
regulation of energy intake. Although our findings are quite pre-
liminary, participants might be cautioned against being less vig-
ilant in regulating energy intake on weeks in which they are not
experiencing difficulties with exercise. In addition, this investi-
gation suggests that maintaining a positive attitude toward the
attainment of weight loss goals may motivate participants to
consistently engage in positive lifestyle changes that promote
weight loss. Finally, consistent self-monitoring of exercise is
important for maintaining regular exercise and obtaining favor-
able weight loss treatment outcomes. Consistent self-monitors
lost twice as much weight, exercised twice as long each week,
and reported fewer difficulties with exercise than individuals
who inconsistently self-monitored their exercise.
Despite significant advancements in the treatment of obe
-
sity over the last several decades, the treatment outcomes (i.e.,
percentage of weight loss) for BWLPs have reached a plateau
(9). Helping participants to maintain positive outcome expec
-
tancies and self-monitoring habits, and to resolve eating-related
difficulties throughout treatment, may not only help to reduce
weight loss fluctuations during treatment but also improve be
-
havioral weight loss treatment outcomes.
REFERENCES
(1) Mokdad AH, Ford ES, Bowman BA, et al.: Prevalence of
obesity, diabetes, and obesity-related health risk factors, 2001.
Journal of the American Medical Association. 2003,
289:76–79.
(2) NHLBI Obesity Education Initiative Task Force Members:
Clinical Guidelines on the Identification, Evaluation, and
Treatment of Overweight and Obesity in Adults: The Evidence
Report, NIH Publication No. 98–4083. Washington, DC: U.S.
Department of Health and Human Services, 1998.
(3) Peeters A, Barendregt JJ, Mackenbach JP, Al Mamun A,
Bonneux L: Obesity in adulthood and its consequences for life
expectancy: A life-table analysis. Annals of Internal Medicine.
2003, 138:24–32.
(4) Jeffery RW, Drewnowski A, Epstein LH, et al.: Long-term
maintenance of weight loss: Current status. Health Psychology.
2000, 19(Suppl.):5–16.
(5) Marcus BH, Dubbert PM, Forsyth LH, et al.: Physical activity
behavior change: Issues in adoption and maintenance. Health
Psychology. 2000, 19:s32–s41.
(6) Perri MG, Corsica JA: Improving the maintenance of weight
lost in behavioral treatment of obesity. In Wadden TA, Stunkard
AJ (eds), Handbook of Obesity Treatment. New York: Guilford,
2002, 357–379.
(7) Wadden T, Letizia K: Predictors of attrition and weight loss in
patients treated by moderate and severe caloric restriction. In
Wadden T, Vanitallie T (eds), Treatment of the Seriously Obese
Patient. New York: Guilford, 1992, 383–410.
(8) Pavlou KN, Krey S, Steffee WP: Exercise as an adjunct
to weight loss and maintanence in moderately obese subjects.
American Journal of Clinical Nutrition. 1989, 49:1115–1123.
(9) Perri MG, Fuller PR: Success and failure in the treatment of
obesity: Where do we go from here? Medicine, Exercise, Nutri
-
tion, and Health. 1995, 4:255–272.
(10) Jeffery RW, Wing RR, Stunkard A: Behavioral treatment of
obesity: The state of the art 1976. Behavior Therapy. 1978,
9:189–199.
(11) Bandura A: Self-Efficacy: The Exercise of Self-Control. New
York: Freeman, 1997.
(12) Bandura A: Social Foundations of Thought and Action: A
Social-Cognitive Framework. Englewood Cliffs, NJ: Prentice
Hall, 1986.
(13) Ali NS, Twibell RK: Health promotion and osteoporosis pre-
vention among postmenopausal women. Preventive Medicine.
1995, 24:528–534.
(14) Dishman R, Sallis J: Determinants and interventions for physi-
cal activity and exercise. In Bouchard C, Shephard R, Stephens
T (eds), Physical Activity, Fitness, and Health: International
Proceedings and Consensus Statement. Champaign, IL: Hu
-
man Kinetics, 1990, 214–233.
(15) Dishman R, Steinhardt M: Health locus of control pre
-
dicts free-living, but not supervised, physical activity: A test of
exercise-specific control and outcome-expectancy hypotheses.
ResearchQuarterly for Exerciseand Sport.1990, 61:383–394.
(16) Godin G, Valois P, Jobin J, Ross A: Prediction of intention to
exercise of individuals who have suffered from coronary heart
disease. Journal of Clinical Psychology. 1991, 47:762–771.
(17) Hofstetter C, Hovell MF, Macera C, et al.: Illness, injury, and
correlates of aerobic exercise and walking: A community study.
Research Quarterly for Exercise and Sport. 1991, 62:1–9.
(18) Edell BH, Edington S, Herd B, O’Brien RM, Witkin G: Self-ef
-
ficacy and self-motivation as predictors of weight loss. Addic
-
tive Behaviors. 1987, 12:63–66.
(19) Brownell KD: The LEARN Program for Weight Management
2000. Dallas, TX: American Health, 2000.
(20) Brand-Miller JC, Wolever TMS, Foster-Powell K, Colagiuri S:
The New Glucose Revolution. New York: Marlowe & Com
-
pany, 2003.
(21) Ludwig DS: The glycemic index: Physiological mechanisms
relating obesity, diabetes, and cardiovascular disease. Journal
of the American Medical Association. 2002, 287:2414–2423.
(22) American College of Sports Medicine: ACSM’s Guidelines for
Exercise Testing and Prescription. Baltimore: Williams &
Wilkins, 2000.
Volume 30, Number 3, 2005 Self-Monitoring and Weight Loss 189
(23) Balke B: Biodynamics: Humans. In Glasser O (ed), Medical
Physics. Chicago: Year Book, 1960, 50–51.
(24) Paffenbarger R, Wing RR, Hyde R: Physical activity as an index
of heart attack risk in college alumni. American Journal of Epi
-
demiology. 1978, 108:161–175.
(25) Pereira MA, FitzGerald SJ, Gregg EW, et al.: A collection of
physical activity questionnaires for health-related research
[Supplement]. Medicine & Science in Sports & Exercise. 1997
(June), 29.
(26) Carels RA, Darby LA, Cacciapaglia HM, Douglass OM: Re
-
ducing cardiovascular risk factors in postmenopausal women
through a lifestyle change intervention. Journal of Women’s
Health. 2004. 13:412–426.
(27) Carels RA, Cacciapaglia HM, Douglass OM, Rydin S, O’Brien
WH: The early identification of poor treatment outcome in
a women’s weight loss program. Eating Behaviors. 2003,
4:265–282.
(28) Carels RA, Darby LA, Douglass OM, Cacciapaglia HM, Rydin
S: Education on the glycemic index of foods fails to improve
treatment outcomes in a behavioral weight loss program.
Eating Behaviors. 2005, 6:145–150.
(29) Zeger SL, Liang K, Albert PS: Models for longitudinal data: A
generalized estimating equation approach. Biometrics. 1988,
44:1049–1060.
(30) Schwartz JE, Stone AA: Strategies for analyzing ecological mo
-
mentary assessment data. Health Psychology. 1998, 17:6–16.
(31) Perri MG, McAdoo WG, McAllister DA, Lauer JB, Yancey
DZ: Enhancing the efficacy of behavior therapy for obesity: Ef-
fects of aerobic exercise and a multicomponent maintenance
program. Journal of Consulting and Clinical Psychology. 1986,
54:670–675.
(32) Hill JO, Schlundt DG, Sbrocco T, et al.: Evaluation of an al
-
ternating-calorie diet with and without exercise in the treatment
of obesity. American Journal of Clinical Nutrition. 1989,
50:248–254.
(33) Boutelle KN, Kirschenbaum DS, Baker RC, Mitchell ME: How
can obese weight controllers minimize weight gain during the
high risk holiday season? By self-monitoring very consistently.
Health Psychology. 1999, 18:364–368.
(34) Baker RC, Kirschenbaum DS: Weight control during the holi
-
days: Highly consistent self-monitoring as a potentially useful
coping mechanism. Health Psychology. 1998, 17:367–370.
(35) Dubbert PM, Wilson GT: Treatment failure in behavior ther
-
apy for obesity. In Foa E, Emmelkamp PMG (eds), Treatment
Failure in Behavior Therapy. New York: Wiley, 1983,
127–153.
(36) Dubbert P, Wilson G: Goal-setting and spouse involvement in
the treatment of obesity. Behavioral Research and Therapy.
1984, 22:227–242.
(37) Perri MG, Nezu AM, Patti ET, McCann KL: Effect of length of
treatment on weight loss. Journal of Consulting and Clinical
Psychology. 1989, 57:450–452.
(38) Donahoe CP, Lin DH, Kirschenbaum DS, Keesey RE: Meta
-
bolic consequences of dieting and exercise in the treatment of
obesity. Journal of Consulting and Clinical Psychology. 1984,
52:827–836.
190 Carels et al. Annals of Behavioral Medicine