The Effect of Reinforcement or Stimulus Control to Reduce Sedentary
Behavior in the Treatment of Pediatric Obesity
Leonard H. Epstein, Rocco A. Paluch, Colleen K. Kilanowski, and Hollie A. Raynor
State University of New York at Buffalo
Obese children were randomly assigned to a family-based behavioral treatment that included either
stimulus control or reinforcement to reduce sedentary behaviors. Significant and equivalent decreases in
sedentary behavior and high energy density foods, increases in physical activity and fruits and vegetables,
and decreases in standardized body mass index (z-BMI) were observed. Children who substituted active
for sedentary behaviors had significantly greater z-BMI changes at 6 (–1.21 vs. –0.76) and 12 (–1.05 vs.
–0.51) months, respectively. Substitution of physically active for sedentary behaviors and changes in
activity level predicted 6- and 12-month z-BMI changes. Results suggest stimulus control and reinforcing
reduced sedentary behaviors are equivalent ways to decrease sedentary behaviors, and behavioral
economic relationships in eating and activity may mediate the effects of treatment.
Key words: pediatric obesity, sedentary behaviors, physical activity, behavioral economics,
Obesity is a prevalent child health problem with increasing
pervasiveness (Troiano, Flegal, Kuczmarski, Campbell, & John-
son, 1995). Considerable progress in the treatment of childhood
obesity has been made (Epstein, Myers, Raynor, & Saelens, 1998),
and reviews have suggested that behavioral treatment of childhood
obesity should be considered an empirically validated treatment
(e.g., Jelalian & Saelens, 1999). One of the newest components to
behavioral interventions for obesity is reducing sedentary behav-
ior. This component has been added in part because obese children
find sedentary activities more reinforcing than physically active
alternatives (Epstein, Smith, Vara, & Rodefer, 1991), and free
access to preferred sedentary behaviors may make it very chal-
lenging to increase physical activity (Epstein & Roemmich, 2001;
Epstein & Saelens, 2000). Decreasing sedentary behavior in obese
children can reduce energy intake associated with sedentary be-
haviors, and increasing sedentary behavior results in a decrease in
physical activity (Epstein, Paluch, Consalvi, Riordan, & Scholl,
2002). Intervention studies have shown that reducing sedentary
behavior in obese children is associated with resulting changes in
obesity and fitness that are as effective as those resulting when
increased physical activity is targeted (Epstein, Paluch, Gordy, &
Dorn, 2000; Epstein et al., 1995).
There are two possible ways in which reducing targeted seden-
tary behaviors could influence energy balance and alter weight loss
in obese children. First, sedentary time can displace time to be
physically active, and a number of studies have shown that there is
a significant inverse relationship between sedentary behaviors,
such as television watching, and physical activity (Dietz & Gort-
maker, 1985; Epstein, Paluch, et al., 2002; Gortmaker et al., 1996).
Thus, one benefit of reducing sedentary behavior may be to free up
time that can be reallocated to being more physically active. The
other benefit of reducing sedentary behavior may be a reduction in
eating (Epstein, Paluch, et al., 2002). Many children eat in asso-
ciation with watching television (Coon, Goldberg, Rogers, &
Tucker, 2001), and for these children a reduction in television
watching may be associated with reducing energy intake.
Behavioral economic theory provides a conceptual framework
to understand how children make decisions about the allocation of
time among available alternatives. Behavioral economics has been
used to understand how children make choices about being sed-
entary or active, and thus about eating or engaging in alternative
behaviors (Epstein & Roemmich, 2001; Epstein & Saelens, 2000).
Children who reduce specific sedentary behaviors such as televi-
sion watching and reallocate some of this time to physically active
behaviors would be considered to substitute physically active for
sedentary behaviors. There would be a negative relationship be-
tween sedentary behaviors and physical activity. If children reduce
a specific sedentary behavior such as television watching they may
substitute some of the time with physical activity and some of the
time on other sedentary behaviors, such as reading or listening to
the radio. Reducing sedentary behaviors could also influence eat-
ing behavior. In this case it is expected that as television watching
is reduced, eating would be reduced. Changes in behaviors in the
same direction are considered complements. If eating behaviors
are reduced when specific sedentary behaviors are reduced, eating
changes complement sedentary behavior changes.
The primary method that has been used in behavioral studies to
reduce sedentary behaviors is positive reinforcement for reduc-
tions in sedentary behavior. Surveys of parent behavior suggest
Leonard H. Epstein, Rocco A. Paluch, Colleen K. Kilanowski, and
Hollie A. Raynor, Department of Pediatrics, State University of New York
This research was supported in part by Grant HD 25997 awarded to
Leonard H. Epstein. We thank James N. Roemmich and Richard Stein for
comments on an earlier version of this article.
Correspondence concerning this article should be addressed to Leonard
H. Epstein, Division of Behavioral Medicine, Department of Pediatrics,
State University of New York at Buffalo, Room G56, Farber Hall, 3435
Main Street, Buffalo, NY 14214. E-mail: email@example.com
2004, Vol. 23, No. 4, 371–380
Copyright 2004 by the American Psychological Association
that most parents would use restriction or instructions to reduce
sedentary behavior (Bybee, Robinson, & Turow, 1982; Sarlo,
Jason, & Lonak, 1988; van der Voort, Nikken, & van Lil, 1992).
In an earlier laboratory study, Epstein, Saelens, Myers, and Vito
(1997) compared positive reinforcement for reducing targeted sed-
entary behavior, punishment for sedentary behavior, and restric-
tion of sedentary behavior versus a control and showed that both
reinforcement and punishment increased activity more than restric-
tion. Research on obesity prevention has used rules and a device
that controls access to sedentary behavior to successfully reduce
sedentary behavior and prevent increases in obesity (Robinson,
1999), but to our knowledge, there is no clinical research compar-
ing alternative methods of reducing sedentary behavior.
One goal of the present study is to determine whether different
methods of reducing targeted sedentary behaviors (watching tele-
vision or VCR/DVDs, playing video games, or using the computer
for non-school-related activities) are associated with differences in
the pattern of change in sedentary and active behaviors and in
percentage overweight change. On the basis of Epstein et al.’s
(1997) laboratory research, we predicted that children positively
reinforced for reducing sedentary behaviors would have greater
changes in these sedentary behaviors than would those provided
stimulus control by reengineering the home environment for sed-
entary behaviors. A second goal is to examine whether individual
differences in substitution of physical activity for sedentary be-
haviors or complements of eating to sedentary behaviors are re-
lated to treatment efficacy. One effective mechanism for the in-
fluence of reducing sedentary behaviors may be for children to
increase their physical activity when sedentary behaviors are re-
duced. If this hypothesis is correct, then children who substitute
active behaviors for sedentary behaviors are likely to be more
successful in obesity treatment than children who do not reduce
their targeted sedentary behaviors or those who reduce sedentary
behaviors but do not increase their physical activity. An alternative
mechanism by which treatment may influence body weight is by
reducing energy intake as sedentary behaviors are reduced. Chil-
dren who show complementary relationships between eating be-
haviors and sedentary behaviors may be more successful than
those children who do not decrease eating in association with
reductions in sedentary behaviors.
Obese 8–12-year-old children and a parent were recruited through
physician referral, brochures, fliers, and newspaper ads (see Figure 1). One
hundred twenty-eight families were interested in participating and were
screened in two cohorts, starting 16 months apart. Seventy-two families
met the following criteria: child being over the 85th body mass index
(BMI) percentile (Must, Dallal, & Dietz, 1991), one parent willing to
attend treatment meetings, no family members participating in another
weight control program, no child or parent with current psychiatric prob-
lems or dietary or exercise restrictions, and child able to read the manuals
and complete the self-monitoring of food and activity. During the child
interview children were given a reading task to use to complete a portion
of a self-monitoring task. This required children to read part of a manual
they would be given during the program and understand the instructions on
how to record foods, amounts, and where to look up calories from a food
reference guide we provided and served to assess their math skills. After
meeting study criteria, 9 of the 72 families withdrew before treatment was
implemented, leaving 63 families. Families were not aware of the group
they were randomized to prior to dropping out. The Institutional Review
Board of the State University of New York at Buffalo approved this study,
children provided written assent, and parents provided written consent to
Families in each cohort were randomized into one of two treatment
groups: reinforced reduced sedentary behavior or stimulus control of
sedentary behaviors. Data were collected at baseline and at 6 and 12
months after randomization.
Common Treatment Components
The treatment program included 16 weekly meetings, followed by two
biweekly meetings and two monthly meetings during the 6-month intensive
treatment. Families received parent and child family-based weight control
workbooks, which included four main sections, (a) introduction to weight
control and self-monitoring, (b) the Traffic Light Diet (described below),
(c) behavior change techniques, and (d) maintenance of behavior change.
Families were seen at 12 months for follow-up treatment and data collec-
tion. At treatment meetings, participating family members were weighed,
met with an individual therapist for 15–30 min, and attended separate
30-min child and parent group meetings. At the weigh-in, weights of each
parent and child were graphed individually on group graphs to provide
immediate feedback on progress.
Families were instructed to record daily food intake (food description;
amount; calories; and number of high-calorie high-fat, “Red,” foods, de-
scribed below) and targeted sedentary behavior times (TV/VCR/DVD,
video games, and nonschool use of a computer) in habit books. Parents
were taught to review habit books during a daily meeting the parent had
with the child. Families were also told to record and graph home weights
at least once per week.
Flow chart of the study.
EPSTEIN, PALUCH, KILANOWSKI, AND RAYNOR
The Traffic Light Diet (Epstein et al., 1998) was used to decrease energy
intake, increase nutrient density, and balance nutrients in the participants’
diet. This diet categorizes foods on the basis of their macro- and micro-
nutrient content into the colors of a traffic light with similar meaning.
Green foods are low in calories and fat (? 2 g of fat per serving), Yellow
foods are medium-calorie and fat foods (2–5 g of fat per serving), and Red
foods are high in calories and fat (? 5 g of fat per serving). Red, Yellow,
and Green coded foods are distributed throughout the U.S. Department of
Agriculture’s Food Guide Pyramid (1992), with the exception of top of the
pyramid foods, which are all designated as Red foods. Children and their
parents were provided a food reference guide, which categorizes most
foods into the traffic light colors and provides calorie values for servings
of each food. The diet starts at 800–1,200 calories per day, with children
encouraged not to go below 800 calories per day. The weight loss goal in
this study was for children to lose 0.5 lbs. (0.2 kg) per week. If children are
losing more than 2 lbs. (0.9 kg) per week for 3 or more weeks, the calorie
range is adjusted upward. Food recording generally underestimates intake
(Bandini, Schoeller, Cyr, & Dietz, 1990), and many children who report
consuming 800–1,200 calories per day are actually consuming more cal-
ories. The goal is for children to gradually reduce the number of Red foods
per week to 15 or fewer. Overweight family members also had a calorie
range of 800–1,200 calories per day adjusted if needed per person’s weight
change. Families who achieved non-overweight status were taught how to
identify energy values that would facilitate maintenance of normal growth.
Information regarding food labels and shopping were included in the diet
Families in both groups were taught to praise children for meeting goals
specific to their group. In addition, children in both groups were provided
a contract reinforcement system to motivate children for behavior change.
In the beginning of the program children and parents completed a rein-
forcer survey to select appropriate reinforcers and to assign point values for
the reinforcers. Small, medium, and large reinforcers could be earned for
meeting behavior change goals for 1, 2, or 4 weeks, respectively. Common
goals for all children were meeting calorie, Red food, and weight goals.
Parents were encouraged to follow through on providing the reward after
all requirements of the contract were met.
Preplanning was taught to facilitate eating and exercise control when
difficult eating and activity situations could be anticipated, such as parties,
holiday gatherings, and school or work functions. There was no specific
activity program provided for the participants. They were provided general
information on lifestyle and aerobic activities, but there were no specific
goals for any participant. Participants did not self-monitor their activity,
and they were not positively reinforced for increasing their activity.
During the individual meeting the therapist reviewed weekly weight
change and discussed the participant’s perceptions of what behavior
change was related to the weight change. Several days of each habit book
were reviewed, followed by review of progress toward contract and pro-
vision of positive reinforcers. Habit books were expected to be completed
prior to each session, and sessions were sometimes delayed or reduced in
time to provide participants the opportunity to complete habit books.
Unique Components of Treatment
All participants were instructed to reduce hours of targeted sedentary
activity to 15 or fewer per week. Children in the reinforcement group were
provided points for reducing their sedentary behaviors to no more than 15
hr per week. The reinforcement group had shaping steps of 25, 20, and 15
hr per week to reduce their sedentary time and were rewarded for meeting
their goals. Goals were set on the basis of the children’s baseline values.
Praise and contract goals specific to decreasing targeted sedentary behav-
iors were used in this group. Children in the stimulus control group were
positively reinforced for recording their sedentary behaviors but not for
behavior change. The stimulus control group were also instructed to change
their environment to prevent them from engaging in the targeted behaviors
and to establish rules regarding the sedentary behaviors. For example, a
family might have the rule that television could only be watched during
specific times, that homework had to be completed before television could
be watched, or that watching television at specific times led to a loss of
privileges. The stimulus control group received additional instructions to
aid sedentary behavior change, which involved posting signs indicating the
sedentary limit and unplugging targeted sedentary activities such as tele-
visions or computers. Thus, the amount of positive reinforcement was
equated across groups, but the positive reinforcement was contingent on
different behaviors. In the reinforced reduction group positive reinforce-
ment was contingent on reducing targeted sedentary behaviors, whereas in
the stimulus control group positive reinforcement was contingent on re-
cording targeted sedentary behaviors.
(Seca, Columbia, MD), and weight was measured in 1/4 lbs. using a
medical balance beam scale (Healthometer, Bridgeview, IL), which was
calibrated daily using a 50-lb. calibration weight (Troemner, Philadelphia,
PA). Standardized body mass index (z-BMI) scores were calculated based
on the value for the 50th BMI percentile and the standard deviation of the
age and sex appropriate sample from the Centers for Disease Control
growth charts (Must et al., 1991).
Although the majority of heights and weights were measured in the
laboratory, self-reported heights and weights were used when families were
unable to attend assessments. Because of underestimation of weight and
overestimation of height (Stewart, Jackson, Ford, & Beaglehole, 1987),
self-reported data was adjusted for self-report based on a data set of more
than 1,000 cases in which adult and child heights and weights were
self-reported and then measured. For children, none of the heights and
weights was self-reported at 6 months, whereas nine of the heights and
weights were self-reported at 12 months (14%). For participating parents,
3 (5%) observations at 6 months and 12 observations (19%) at 12 months
were self-report. Percentage overweight data at 6 months were missing
from 2 participants who provided data at 12 months. These data points
were estimated by a regression model (N ? 58) that included baseline and
1-year percentage overweight, which had a multiple correlation of .95.
Participating parents and children recorded any sedentary or
physical activity that took 10 min or longer in duration on index cards
(structured with columns for start and stop times and the activity descrip-
tion) over 4 days (2 weekdays and 2 weekend days). The index cards were
reviewed in person within a few days of recording to address ambiguous
activities or large time gaps. We used Ainsworth et al.’s (2000) compen-
dium of physical activities to note the values of the multiples of resting
metabolic rate (METS) for each activity. Average METS per day were
calculated. In addition, amounts of time that were at least 3 METS were
coded as moderate to vigorous physical activity (MVPA; Pate et al., 1995).
The average number of minutes of all activities recorded at baseline, 6
months, and 12 months were 5,561.8; 5,611.9; and 5,512.9, respectively.
The total number of possible minutes over the 4 days of recording was
5,760 min, suggesting that the self-recording was able to represent the
majority of the 4 days of recording (96.5%, 97.4%, and 95.7% of the time).
The number of minutes engaged in the targeted sedentary behaviors was
coded, and nontargeted sedentary behavior time was calculated by the
following equation: Nontargeted sedentary behavior time ? total time ?
time spent in targeted sedentary behaviors ? time ? 2 METS.
To validate the physical activity self-report, 41 children from the first
cohort wore a Tritrac (Stayhealthy, Inc., Monrovia, CA) triaxial acceler-
ometer for 4 days, and energy expenditure values were compared for the
time intervals that had corresponding accelerometer and self-report. The
Tritrac has been validated against indirect calorimetry and heart rate (Chen
& Sun, 1997; Welk & Corbin, 1995). Average MET values calculated from
the Tritrac and self-report correlated at .63 (p ? .001), whereas percentage
Height was measured in 1/8 in. using a stadiometer
REDUCING SEDENTARY BEHAVIOR
of self-reported time engaged in MVPA correlated at .60 (p ? .001) with
Tritrac-measured percentage of time engaged in MVPA.
Dietary intake was measured using a food intake ques-
tionnaire (Epstein, Gordy, et al., 2001) designed to assess servings of Red
foods and fruits and vegetables that has been validated against previous day
24-hr recalls (Epstein, Gordy, et al., 2001). Previous research has shown
that changes in Red food intake are related to weight loss (Epstein et al.,
1998; Epstein, Wing, Koeske, Andrasik, & Ossip, 1981).
The Hollingshead Four Factor Index of Social
Status (Hollingshead, 1975) was used to measure socioeconomic status
prior to treatment.
Determination of behavioral economic substitution and complementary
In behavioral economic theory, commodities can substitute
for each other if when consumption of one decreases, consumption of
another commodity increases. For example, someone may prefer coffee as
a morning beverage to “wake up,” but the price of coffee increases, and the
person shifts to tea as the morning beverage. Both beverages provide a
warm, caffeinated drink, and tea is a substitute for coffee for many people.
Commodities are complements when consumption of one commodity
increases consumption of another. For example, increasing aerobic exer-
cise will result in an increase in the consumption of sports beverages.
Likewise, if a person who used to drink a sports beverage after exercise
stopped exercising, consumption of the sports beverage would decrease.
Hypothetical relationships between changes in targeted sedentary behav-
ior and physical activity are shown in the top panel of Figure 2. Changes
in targeted sedentary behavior are shown on the x-axis, with a decrease in
targeted sedentary behaviors to the left of the vertical line, and an increase
in targeted sedentary behaviors to the right of the vertical line. The two
physical activity lines (A and B) represent ways in which physical activity
can change in relationship to sedentary behavior. Line A is a negatively
sloped line in which decreases in sedentary behavior are associated with
the highest levels of physical activity, and increases in sedentary behavior
are associated with reductions in physical activity. Line B is positively
sloping, such that decreases in sedentary behavior are associated with the
decreases in physical activity, and increases in sedentary behavior are
associated with increases in sedentary behavior. The bottom panel shows
complementary relationships between changes in sedentary behavior and
eating. Line A in this case has a positive slope and shows decreases in
eating being associated with decreases in sedentary behavior.
One way to quantify these behavioral economic relationships is by
calculating the slope of the lines that describe the relationship between the
two commodities. For example, in the case that sedentary behaviors are
decreased, a greater negative slope would mean that more physical activity
is substituted for sedentary behavior then a lower slope. Likewise, for
eating behaviors, a greater positive slope would mean that there is a greater
reduction in eating behaviors when sedentary behaviors are reduced.
Slopes have a different meaning based on the direction of change in
sedentary behavior. For example, the solid Line A at the top of Figure 2
indicates an increase in physical activity when targeted sedentary behaviors
are decreased, with the degree of slope indicating the degree of substitut-
ability. However, the dotted line component of Line A indicates a decrease
in physical activity when targeted sedentary behaviors are increased. Two
participants may have the same slope, but on the basis of the slope alone
it would not be possible to tell whether the participant was making a
healthy behavior change consistent with the study goals or whether the
participant was making the opposite change by being more sedentary and
even less active.
In addition, the use of the slope to assess the degree of substitutability is
based on the assumption that the relationships between increases and
decreases in sedentary behavior to changes in physical activity and dietary
intake are symmetrical across the range of values of sedentary behaviors.
Behavioral economic research indicates that relationships between com-
modities may be symmetrical or asymmetrical. For example, there is
generally a complementary relationship between smoking and coffee
drinking. When the behavioral requirements to obtain cigarettes are de-
creased, coffee intake is reduced, but this relationship is not symmetrical
because when the price of coffee intake is increased, smoking is not
reduced (Bickel, Hughes, DeGrandpre, Higgins, & Rizzuto, 1992). Like-
tary behaviors and physical activity (top panel) and high energy density
foods (bottom panel). The solid Line A in the top panel, Quadrant 1
represents a child who substitutes a decrease in targeted sedentary behavior
with an increase in physical activity. The solid Line A in the bottom panel,
Quadrant 3 represents a child who complements a decrease in targeted
sedentary behavior with a reduction in high energy density food. The
x-axes represent changes in sedentary behavior, which could refer to
increases or decreases in time spent watching television and/or playing
computer games. The y-axes represent changes in physical activity, such as
time being active, miles run or biked (top panel), or calories or servings
consumed of high energy density foods (bottom panel).
Theoretical relationships between changes in targeted seden-
EPSTEIN, PALUCH, KILANOWSKI, AND RAYNOR
wise, when the cost of obtaining heroin is increased, heroin consumption
decreases and valium consumption increases, as valium is a substitute for
heroin. When the price for valium is increased, valium consumption
decreases, but consumption of heroin is stable, once again demonstrating
that relationships between commodities are not necessarily symmetrical
across the range of values (Bickel, Madden, & Petry, 1998). Similar
asymmetrical relationships have been observed when sedentary behaviors
are changed. Epstein, Paluch, et al. (2002) observed that increases in
sedentary behavior had significantly greater influences on eating and
physical activity than decreases in sedentary behavior in nonobese youth.
Similarly, decreases in sedentary behavior have a greater influence on
physical activity than increases in sedentary behavior in obese adolescents
(Epstein, Raynor, Trivikram, Paluch, & Roemmich, in press). Thus, in this
study, slopes cannot be used as a continuous variable to quantify the
relationships between variables when behavior may be increased or de-
creased and if the relationships are not symmetrical across the range of
values of sedentary behaviors.
An alternative to the use of slopes to quantify the substitution or
complementary relationship is to dichotomize participants into those who
substitute physical activity for sedentary behavior when sedentary behav-
iors are reduced and those who complement a reduction in sedentary
behavior with a reduction in intake of high energy density foods. The use
of the dichotomous measure instead of a continuous variable may result in
a loss of some power in determining relationships between variables, but
the classification of participants as substituters or complementers does
provide an individual-difference variable consistent with the theory that is
being tested and the asymmetrical nature of the relationship between
sedentary behaviors and physical activity or dietary intake. In this study,
participants in Quadrant 1 in the top panel of Figure 2 who decreased
sedentary behavior and increased physical activity were considered sub-
Child Descriptive Data at Baseline for Each Treatment Group
Stimulus control Reinforce reduced sedentary
M SDM SD
Height in inches (cm)
Weight in pounds (kg)
Targeted sedentary behavior time (%)
Nontargeted sedentary behavior (%)
MVPA (? 3 METS) time (%)
Activity level (METS)
Fruits and vegetables per day
Red foods per day
index; MVPA ? moderate to vigorous physical activity; METS ? multiples of resting metabolic rate; Red
food ? high energy density food.
All values were nonsignificant at p ? .05. M ? male; F ? female; z-BMI ? standardized body mass
Child changes in standardized body mass index (z-BMI) over time by randomized group.
REDUCING SEDENTARY BEHAVIOR
stituters, and participants in Quadrant 4 in the bottom panel were consid-
ered to be complementers.
Between-groups differences at baseline were assessed using two-tailed t
tests. We used mixed analyses of variance to evaluate changes over time,
with group (stimulus control and reinforcement) and cohort as between-
subjects variables and time as the within-subject factor (0, 6, and 12
months). Dependent variables included z-BMI, sedentary behavior, eating,
and activity change. In addition, participants were grouped into those who
substituted active for targeted sedentary behaviors if they reduced their
targeted sedentary behavior below baseline levels while increasing their
physical activity above baseline levels (substituters, n ? 26), and all others
were coded as those who did not substitute (nonsubstituters, n ? 30) or
those who showed a complementary reduction in intake of high energy
density foods when sedentary behaviors were reduced (complementers,
n ? 32) versus those participants who did not decrease intake of high
energy density foods (noncomplementers, n ? 23). Complete 6-month data
for targeted sedentary and active behaviors were not available for 6
participants, and complete data were not available for the relationship
between targeted sedentary behaviors and high energy density foods for 7
participants. These participants were dropped from their respective analy-
ses. The main effects of time or the interaction of Group ? Substituters/
Nonsubstituters were probed using linear contrasts. T tests were used to
identify variables that differentiated groups of children who were substi-
tuters from nonsubstituters or complementers from noncomplementers.
Regression models were used to identify variables associated with
z-BMI change at 6 and 12 months. Age, sex, socioeconomic status, group,
substitution or complementing as dichotomous variables (coded as 0 for
did not substitute or complement and 1 for substituted active for sedentary
behavior or complemented eating for sedentary behavior) and 0–6-month
changes in targeted and nontargeted sedentary behaviors, physical activity,
Red foods, and fruits and vegetables were included as predictors.
One outlier (Studentized residuals ? 3.0) was identified in the analysis
of change in targeted sedentary behaviors (Studentized residual ? 4.4).
This participant was removed from all analyses of targeted sedentary
Baseline anthropometric, body composition, and behavioral data
for participants in each treatment group who completed at least 6
months of treatment (62 of 63) are presented in Table 1. The
sample of children consisted of 23 boys (37.1%) and 39 girls
(62.9%), who were 90.3% White, 6.5% Black, 1.6% Hispanic, and
1.6% other racial/ethnic groups. Characteristics of the children
were as follows (M ? SD): 9.8 ? 1.3 years of age, BMI (BMI ?
kg/m2) of 27.7 ? 2.6, z-BMI of 3.2 ? 1.0; percentage overweight
at entry into treatment of 64.9 ? 14.1. The mean (? SD) Holl-
ingshead Four Factor Index of Social Status score for these fam-
ilies was 45.6 ? 10.2.
Seventy-three percent (45 of 62) of the participating parents
were obese, with 38 obese mothers (70.4%) and 7 obese fathers
(87.5%). Mothers were 41.4 ? 5.5 years of age, 39.6% ?
27.3% overweight, with BMI of 30.0 ? 5.9. Fathers were
43.4 ? 6.0 years of age, 42.8% ? 15.3% overweight, with BMI
of 32.9 ? 3.5.
The average child spent 11.4% ? 6.0% of time engaged in
television watching and computer game play, 11.9% ? 4.1% of
time engaged in other sedentary behaviors, and 4.8% ? 3.5% of
time engaged in physical activity at the MVPA intensity. There
were no differences in any baseline measures by group.
There was a main effect of time for changes in z-BMI, F(2,
112) ? 59.91, p ? .001, with significant treatment effects
observed at both 6 months, F(2, 56) ? 54.94, p ? .001, and 12
months, F(2, 56) ? 22.90, p ? .001, but there were no signif-
icant differences in the rate of change between groups. Z-BMI
values for the stimulus control group were 3.3 ? 1.0, 2.3 ? 1.0,
time for participants who substituted physical activity for targeted seden-
tary behaviors (top panel) or complemented a reduction in targeted seden-
tary behavior with a reduction in high energy density foods (bottom panel).
Child changes in standardized body mass index (z-BMI) over
EPSTEIN, PALUCH, KILANOWSKI, AND RAYNOR
and 2.4 ? 1.0, at 0, 6, and 12 months, respectively, whereas the
values for the reinforced reduction group at the same time
points were 3.2 ? 1.0, 2.2 ? 1.1, and 2.6 ? 1.0 (see Figure 3).
Similar effects were observed for percentage overweight at 6
months, F(2, 56) ? 111.82, p ? .001, and 12 months, F(2,
56) ? 31.88, p ? .001.
Significant decreases in high energy density (Red) foods
(?2.6 ? 2.2) were observed from 0–6 months, F(1, 55) ? 82.63,
p ? .001, along with significant increases in servings of fruits and
vegetables (0.6 ? 2.3), F(1, 56) ? 4.10, p ? .05; increases in
percentage of time above 3 METS (2.9 ? 4.0), F(1, 54) ? 28.08,
p ? .001; and average METS (0.11 ? 0.20), F(1, 54) ? 17.20, p ?
.001. There was also a significant decrease (?2.2 ? 7.4) in
percentage of time in targeted sedentary behaviors, F(1, 53) ?
4.73, p ? .05. There were no significant changes as a function of
Behavioral Economic Relationships With Relative Weight
Analysis of changes in z-BMI over time as a function of sub-
stituters and nonsubstituters (Figure 4, top panel) showed signifi-
cant differences in the rate of change between these two groups,
F(2, 100) ? 5.30, p ? .01, with those who substituted showing
greater change at 6 months (?1.21 vs. ?0.76 z-BMI), F(1, 50) ?
5.84, p ? .02, and 12 months (?1.05 vs. –0.51 z-BMI), F(1, 50) ?
6.63, p ? .02. In addition, those who showed complementary
relationships between high energy density foods and changes in
sedentary behaviors (Figure 4, bottom panel) also showed stronger
z-BMI changes over time, F(2, 98) ? 3.68, p ? .03, with those
who complemented showing greater change at 6 months (?1.17
vs. ?0.69 z-BMI), F(1, 49) ? 6.58, p ? .02, but not 12 months
(?0.93 vs. –0.51 z-BMI), F(1, 49) ? 3.24, p ? .078. There were
no interactions of Group ? Substitution or Group ? Complement.
It would be interesting to observe how substituting physical activ-
ity and complementing eating would interact, but there were so
few participants who substituted and did not complement (N ? 2)
that this analysis could not be run.
Differences between children who substituted physically active
for sedentary behaviors and those who did not, or those who
complemented high energy density foods when targeted sedentary
behaviors were changed are shown in Table 2. The substitutability
groups differed by sex, with a higher percentage of boys (14 of
22 ? 64%) substituting physically active for sedentary behaviors
than girls (12 of 34 ? 33%), ?2(1, N ? 56) ? 4.31, p ? .05. The
groups did not differ by initial levels of sedentary behavior or
activity but did differ in program adherence at 12 months, t(42) ?
2.17, p ? .05. The groups that complemented engaged in more
MVPA, t(53) ? 2.23, p ? .05, and greater general activity level,
t(53) ? 2.04, p ?.05, than did noncomplementers. The groups did
not differ in high energy density foods at baseline.
Changes in targeted sedentary behaviors, MVPA, and Red food
intake as a function of who substituted during treatment (Months
0–6) are presented in Figure 5. There was an effect of substituting
physically active for targeted sedentary behaviors on targeted
sedentary behaviors, F(1, 51) ? 13.82, p ? .001, with greater
decreases at 6 months for those who substituted (?5.8% ? 4.6%
vs. 1.0% ? 8.0%). Greater increases in MVPA were observed at 6
months for those who substituted versus those who did not sub-
stitute (5.5% ? 3.4% vs. 0.5% ? 3.0%), F(1, 52) ? 32.86, p ?
.001. Greater increases in activity level for those who substituted
were also observed (0.20 ? 0.20 vs. 0.03 ? 0.16), F(1, 52) ?
11.96, p ? .001. Red food intake was not differentially influenced
by who substituted or did not substitute physical activity for
targeted sedentary behaviors.
When participants were grouped according to whether they
complemented high energy density food intake there was no sig-
nificant difference in change over time for either percentage of
time in MVPA or average activity level, but targeted sedentary
behaviors differentially changed as a function of who comple-
mented, F(1, 50) ? 30.47, p ? .001, with greater decreases for
those who complemented observed at 6 months (?6.1 ? 5.4 vs.
2.5 ? 5.8). Red food intake was differentially influenced by who
complemented Red food intake when targeted sedentary behaviors
were changed, F(1, 51) ? 4.47, p ? .05, with greater decreases for
Child Descriptive Data at Baseline for Children Who Substituted Physical Activity or Complement High Energy Density Foods for
Reductions in Sedentary Behaviors
pMSD MSD MSDM SD
Targeted sedentary behavior time (%)
Nontargeted sedentary behavior (%)
MVPA (? 3 METS) time (%)
Activity level (METS)
Red food servings per day
Fruit and vegetable servings per day
metabolic rate; Red food ? high energy density food.
M ? male; F ? female; z-BMI ? standardized body mass index; MVPA ? moderate to vigorous physical activity; METS ? multiples of resting
REDUCING SEDENTARY BEHAVIOR
those who complemented versus those who did not complement at
6 months (?3.2 ? 1.8 vs. ?2.0 ? 2.4).
Predictors of Change in z-BMI
The best fitting regression models at 6 months and 12 months
are presented in Table 3. Both models included substitution group
and change in activity level as predictors. The change at 6 months
was also predicted by child age and by change in high energy
density foods. The change at 12 months was also predicted by
child sex. The model for 6 months had a multiple correlation of .62
and a multiple correlation squared of .38. The best fitting model at
12 months had a multiple correlation of .52, and a multiple
correlation squared of .27.
Previous studies from our laboratory have shown that a com-
prehensive family-based behavioral program reducing sedentary
behavior is associated with reductions in percent overweight (Ep-
stein et al., 1995, 2000). This study showed that significant de-
creases in percent overweight were observed for obese children in
both stimulus control and reinforced reduction of sedentary behav-
ior groups. There were similar reductions in sedentary behaviors
and high energy density foods, and similar increases in moderate
to vigorous physical activity across groups. These results suggest
that there are no differences in weight control or behaviors asso-
ciated with weight control for two different methods for modifying
The goal for reducing sedentary behaviors is to provide the
opportunity to be more physically active and to reduce energy
intake. Although the between-groups effects of the interventions
were similar in their influence on sedentary behavior, there were
individual differences across groups in whether children substi-
tuted active for sedentary behavior or complemented energy intake
when sedentary behaviors were reduced. Children who substituted
active for sedentary behaviors showed a twofold advantage in
z-BMI change at 12 months compared with those children who did
not substitute (?1.05 vs. ?0.51), whereas those who comple-
mented a reduction in high energy density foods with reductions in
targeted sedentary behavior showed a 1.8 advantage (?0.93 vs.
Children who substituted active for sedentary behaviors were
similar in baseline levels of activity or targeted sedentary behav-
iors, but sex was related to who substituted. Boys were twice as
likely to substitute physically active for sedentary behaviors as
girls (54% vs. 27%), which is consistent with other research from
our laboratory (Epstein, Paluch, & Raynor, 2001). Developing an
understanding of why sex may be related to the substitutability of
physically active for sedentary behaviors may be important to
understand how to increase activity in girls, who show larger
decreases in physical activity as they grow older (Caspersen,
Pereira, & Curran, 2000). The process of substitution was associ-
ated with a greater reduction in targeted sedentary behaviors and
larger increase in physical activity, but it was not associated with
changes in intake of high energy density foods.
Children who complemented changes in high energy density
foods when targeted sedentary behaviors were reduced were sim-
ilar to baseline levels of activity or targeted sedentary behaviors
and intake of high energy density foods but were more active than
children who did not complement changes in sedentary behavior
with changes in intake of high energy density foods. The process
of complementing high energy density foods was associated with
greater decreases in targeted sedentary behavior and greater de-
crease in high energy density foods than those who did not com-
plement. Children who complemented high energy density foods
did not differ in their changes in moderate to vigorous physical
One of the benefits of reducing targeted sedentary behavior for
weight control may be to increase the opportunity of participants to
erate to vigorous physical activity (MVPA) and high energy density (Red)
foods for those who substituted physical activity for targeted sedentary
behaviors (top panel) or complemented a reduction in targeted sedentary
behaviors with a reduction in Red foods (bottom panel).
Child changes in targeted sedentary behavior (Sed Beh), mod-
EPSTEIN, PALUCH, KILANOWSKI, AND RAYNOR
be more active. This study provides strong support for this hy-
pothesis. A second benefit of reducing targeted sedentary behavior
is to reduce the opportunities to consume high energy density
foods. To our knowledge, this is the first study to demonstrate that
individual differences in substitution of physically active for sed-
entary behaviors or the complementary changes in high energy
density foods when targeted sedentary behaviors are changed are
related to weight control.
The regression models suggest that substitution of physical
activity when sedentary behaviors are reduced is more important
for weight control than complementing decreases in high energy
density foods when sedentary behaviors were reduced. Substitu-
tion and changes in activity levels were both predictors of both 6-
and 12-month z-BMI change in multivariate models. Although
complementing was not a predictor of z-BMI change, changes in
Red foods did predict 6-month change. It may be that the absolute
changes in high energy density foods, rather than the relationship
between changes in high energy density foods and targeted sed-
entary behaviors, are important predictors of changes in dietary
intake that influence weight control when sedentary behaviors are
There are several limitations to this study. First, the determina-
tion of whether children substituted or complemented was based
on individual differences in behavior change and was not an
experimenter controlled variable. It is possible that there was a
third variable that was influencing whether children substituted
physically active for sedentary behaviors and weight loss or
whether children complemented high energy density foods when
sedentary behaviors were changed and weight loss. Thus, it is not
possible to attribute causal influence for substituting active for
sedentary behaviors the same way that would be possible if dif-
ferential treatments had produced differences in substitution of
physically active for sedentary behaviors. Second, the measures of
child time allocation and eating behaviors were self-report. The
physical activity component of self-report could be validated
against accelerometer measures, but the sedentary and eating be-
haviors are more challenging to validate. Although it is reasonable
to generalize from valid recording of activity to valid recording of
sedentary behaviors, it is possible that these classes of behavior are
recorded differently. It would be useful in future studies to provide
objective measures of sedentary behaviors to reduce reliance on
self-report. Third, participants were defined as substituters on the
basis of a reduction in sedentary behavior and an increase in
physical activity. It is assumed that some if not all of the increased
activity was engaged in at times when the participants had been
sedentary, but it was not possible to know that there was a direct
substitution of active for sedentary behaviors. Fourth, the dietary
intake data focused on high energy density foods and fruits and
vegetables, and total energy intake was not measured. Measure-
ment of energy intake, in addition to high energy density foods
would have provided a more complete picture of dietary changes
and perhaps a better assessment of how dietary changes related to
changes in targeted sedentary behaviors. Fifth, in a clinical study
there is less control over changing sedentary behavior than in a
laboratory study. It is possible that implementing the two inter-
ventions in a more intensive, controlled manner may have pro-
duced between-groups differences more consistent with the previ-
ous laboratory research (Epstein et al., 1997).
In summary, the results of this study support the continued use
of treatment components that target reducing sedentary behaviors
such as television watching or playing computer games, but the
effects of this intervention are enhanced when participants engage
in physical activity instead of allocating time to nontargeted sed-
entary behaviors or when participants reduce their intake of high
energy density foods when targeted sedentary behaviors are re-
duced. One challenge for future studies is to develop interventions
that influence children to substitute active for sedentary behaviors
rather than rely on individual differences in substituting active for
sedentary behavior. Another avenue for future research is to pro-
vide a better understanding of why boys are more likely than girls
to substitute physical activity for sedentary behaviors when sed-
entary behaviors are reduced. Finally, future research should study
benefits associated with stimulus control interventions that alter
behavior by restructuring the environment. There have been very
limited but successful previous attempts to study stimulus control
interventions (Loro, Fisher, & Levenkron, 1979). Altering the
Multiple Regression Models Predicting Changes in Standardized Body Mass Index (z-BMI) at 6
and 12 Months
Change in z-BMI at 6 months
Change in Red foods at 6 months
Change in activity level at 6 months
R ? .62, R2? .38, F(4, 50) ? 7.76, p ? .001
Change in z-BMI at 12 months
Change in activity level at 6 months
R ? .52, R2? .27, F(3, 50) ? 6.23, p ? .002
Red foods ? high energy density foods.
REDUCING SEDENTARY BEHAVIOR
environment to reduce opportunities to be sedentary provides a
method that alters the opportunity to be active and does not require
providing consequences for individual behaviors, reducing the
work of parents in implementing the treatment program.
Ainsworth, B. E., Haskell, W. L., Whitt, M. C., Irwin, M. L., Swartz,
A. M., Strath, S. J., et al. (2000). Compendium of physical activities: An
update of activity codes and MET intensities. Medicine & Science in
Sports & Exercise, 32(Suppl. 9), S498–S504.
Bandini, L. G., Schoeller, D. A., Cyr, H. N., & Dietz, W. H. (1990).
Validity of reported energy intake in obese and non-obese adolescents.
American Journal of Clinical Nutrition, 52, 421–425.
Bickel, W. K., Hughes, J. R., DeGrandpre, R. J., Higgins, S. T., & Rizzuto,
P. (1992). Behavioral economics of drug self-administration: IV. The
effects of response requirement on the consumption of and interaction
between concurrently available coffee and cigarettes. Psychopharmacol-
ogy (Berlin), 107, 211–216.
Bickel, W. K., Madden, G. J., & Petry, N. M. (1998). The price of change:
The behavioral economics of drug dependence. Behavior Therapy, 29,
Bybee, C., Robinson, D., & Turow, J. (1982). Determinants of parental
guidance of children’s television viewing for a special subgroup: Mass
media scholars. Journal of Broadcasting, 26, 697–710.
Caspersen, C. J., Pereira, M. A., & Curran, K. M. (2000). Changes in
physical activity patterns in the United States, by sex and cross-sectional
age. Medicine & Science in Sports & Exercise, 32, 1601–1609.
Chen, K. Y., & Sun, M. (1997). Improving energy expenditure estimation
by using a triaxial accelerometer. Journal of Applied Physiology, 83,
Coon, K. A., Goldberg, J., Rogers, B. L., & Tucker, K. L. (2001).
Relationships between use of television during meals and children’s
food consumption patterns. Pediatrics, 107, E7.
Dietz, W. H., & Gortmaker, S. L. (1985). Do we fatten our children at the
television set? Obesity and television viewing in children and adoles-
cents. Pediatrics, 75, 807–812.
Epstein, L. H., Gordy, C. C., Raynor, H. A., Beddome, M., Kilanowski,
C. K., & Paluch, R. A. (2001). Increasing fruit and vegetable and
decreasing fat and sugar intake in families at risk for childhood obesity.
Obesity Research, 9, 171–179.
Epstein, L. H., Myers, M. D., Raynor, H. A., & Saelens, B. E. (1998).
Treatment of pediatric obesity. Pediatrics, 101, 554–570.
Epstein, L. H., Paluch, R. A., Consalvi, A., Riordan, K., & Scholl, T.
(2002). Effects of manipulating sedentary behavior on physical activity
and food intake. Journal of Pediatrics, 140, 334–339.
Epstein, L. H., Paluch, R. A., Gordy, C. C., & Dorn, J. (2000). Decreasing
sedentary behaviors in treating pediatric obesity. Archives of Pediatrics
and Adolescent Medicine, 154, 220–226.
Epstein, L. H., Paluch, R. A., & Raynor, H. A. (2001). Sex differences in
obese children and siblings for family-based obesity treatment. Obesity
Research, 9, 746–753.
Epstein, L. H., Raynor, H. A., Trivikram, R., Paluch, R. A., & Roemmich,
J. N. (in press). Effects of changing sedentary behavior on physical
activity in obese and non-obese adolescents. Annals of Behavioral
Epstein, L. H., & Roemmich, J. N. (2001). Reducing sedentary behavior:
Role in modifying physical activity. Exercise and Sport Science Re-
views, 29, 103–108.
Epstein, L. H., & Saelens, B. E. (2000). Behavioral economics of obesity:
Food intake and energy expenditure. In W. K. Bickel & R. E. Vuchinich
(Eds.), Reframing health behavior change with behavioral economics
(pp. 293–311). Mahwah, NJ: Erlbaum.
Epstein, L. H., Saelens, B. E., Myers, M. D., & Vito, D. (1997). Effects of
decreasing sedentary behaviors on activity choice in obese children.
Health Psychology, 16, 107–113.
Epstein, L. H., Smith, J. A., Vara, L. S., & Rodefer, J. S. (1991). Behav-
ioral economic analysis of activity choice in obese children. Health
Psychology, 10, 311–316.
Epstein, L. H., Valoski, A. M., Vara, L. S., McCurley, J., Wisniewski, L.,
Kalarchian, M. A., et al. (1995). Effects of decreasing sedentary behav-
ior and increasing activity on weight change in obese children. Health
Psychology, 14, 109–115.
Epstein, L. H., Wing, R. R., Koeske, R., Andrasik, F., & Ossip, D. J.
(1981). Child and parent weight loss in family-based behavior modifi-
cation programs. Journal of Consulting and Clinical Psychology, 49,
Gortmaker, S. L., Must, A., Sobol, A. M., Peterson, K., Colditz, G. A., &
Dietz, W. H. (1996). Television watching as a cause of increasing
obesity among children in the United States, 1986–1990. Archives of
Pediatrics and Adolescent Medicine, 150, 356–362.
Hollingshead, A. B. (1975). Four factor index of social status. New Haven,
CT: Yale University.
Jelalian, E., & Saelens, B. E. (1999). Empirically supported treatments in
pediatric psychology: Pediatric obesity. Journal of Pediatric Psychol-
ogy, 24, 223–248.
Loro, A. D., Jr., Fisher, E. B., & Levenkron, J. C. (1979). Comparison of
established and innovative weight-reduction treatment procedures. Jour-
nal of Applied Behavior Analysis, 12, 141–155.
Must, A., Dallal, G. E., & Dietz, W. H. (1991). Reference data for obesity:
85th and 95th percentiles of body mass index (wt/ht2) and triceps
skinfold thickness. American Journal of Clinical Nutrition, 53, 839–
Pate, R. R., Pratt, M., Blair, S. N., Haskell, W. L., Macera, C. A.,
Bouchard, C., et al. (1995). Physical activity and public health: A
recommendation from the Centers for Disease Control and Prevention
and the American College of Sports Medicine. Journal of the American
Medical Association, 273, 402–407.
Robinson, T. N. (1999). Reducing children’s television viewing to prevent
obesity: A randomized controlled trial. Journal of the American Medical
Association, 282, 1561–1567.
Sarlo, G., Jason, L. A., & Lonak, C. (1988). Parents’ strategies for limiting
children’s television watching. Psychological Reports, 63, 433–438.
Stewart, A. W., Jackson, R. T., Ford, M. A., & Beaglehole, R. (1987).
Underestimation of relative weight by use of self-reported height and
weight. American Journal of Epidemiology, 125, 122–126.
Troiano, R. P., Flegal, K. M., Kuczmarski, R. J., Campbell, S. M., &
Johnson, C. L. (1995). Overweight prevalence and trends for children
and adolescents: The National Health and Nutrition Examination Sur-
veys, 1963 to 1991. Archives of Pediatrics and Adolescent Medicine,
U.S. Department of Agriculture. (1992). The food guide pyramid. Wash-
ington, DC: U.S. Government Printing Office.
van der Voort, T. H. A., Nikken, P., & van Lil, J. E. (1992). Determinants
of parental guidance of children’s television viewing: A Dutch replica-
tion study. Journal of Broadcasting and Electronic Media, 36, 61–74.
Welk, G. J., & Corbin, C. B. (1995). The validity of the Tritrac-R3D
Activity Monitor for the assessment of physical activity in children.
Research Quarterly for Exercise and Sport, 66, 1–8.
EPSTEIN, PALUCH, KILANOWSKI, AND RAYNOR