Journal Logo


Determinants of physical activity in obese children assessed by accelerometer and self-report


Author Information
Medicine & Science in Sports & Exercise: September 1996 - Volume 28 - Issue 9 - p 1157-1164
  • Free


Television watching, a major component of sedentary behavior, is cross-sectionally (18,46,51) and prospectively (18) related to obesity in childhood and adolescence. Lean infants (41) and young children(30) at high risk for obesity are less active than children at low risk for becoming obese. Obese children may also have lowered activity in comparison with nonobese children. A recent study using the doubly labeled water method to assess energy expenditure (17) showed similar total daily expenditure for obese and nonobese children, which, given the higher weight of the obese children, indicated reduced activity in the obese children.

Consistent with the influence of low activity level on the development of obesity, increasing activity is important for the treatment of childhood obesity. A number of studies have documented that exercise alone, or exercise plus diet (22), are useful for the short-term treatment of childhood obesity. In addition, we have shown that children randomized to either lifestyle or aerobic exercise interventions had better percent overweight changes over a 10-yr period than children randomized to low-expenditure calisthenic exercises (25).

Research is needed to identify determinants of activity relevant to special populations, such as the obese, which can be modified in intervention studies. However, research on the determinants of activity is limited by problems in measurement of activity (19). The majority of research on determinants of activity has used self-report methods to measure activity. Self-report methods, which include diary and recall, generally provide less accurate indications of activity than more objective methods, such as motion sensors, heart rate monitoring, and doubly labeled water(35,43,52). The degree of relationship between objective and self-report measures of activity suggests that there may be a substantial amount of variance not shared among self-report and objective measures of activity, which may lead to differences in the relationships between specific determinants and activity level as a function of the method of measurement. In an important test of this hypothesis, Dishman and colleagues (20) contrasted psychological predictors of activity in adults when activity was measured by self-report or objective measures, and concluded that the determinants depended upon the type of measurement employed.

This provocative finding requires replication and extension using sedentary clinical populations, such as obese children, and different sets of determinants of activity. There has been a wealth of research on determinants of activity in children (45) that may be used to test whether the relationships between activity level and predictors differ as a function of measurement methodology. These variables include fitness and body composition (49), and psychological and behavioral predictors, such as activity hedonics (32), decisional balance (37), and interfamilial activity patterns(29,42).

Activity levels in obese children may also be related to child and parent psychological problems. We have previously documented that a sizable minority of obese children experience psychopathology(23,24), which may be relevant to assessment of activity level, since changes in activity level can occur in certain psychopathological states, such as depression, eating disorders, or attention deficit disorder (4). In a recent test of the relationship between activity level and depression in a population database, Brown (15) found that subjects with moderate and low levels of activity were two and three times more likely to be depressed than very active subjects. Subjects who were very active at baseline were half as likely to develop depression over the 7-9 yr of follow-up as less active subjects. It is particularly important to control for psychological problems when assessing activity in obese children, since obesity may lead to increased prevalence of social withdrawal and depression(23,24), resulting in decreased activity.

The primary goal of this study was to evaluate similarities or differences in determinants of activity when activity was measured using self-report and objective methods, controlling for child and parent psychological problems. The self-report method we chose is the activity diary based on the research by Baranowski (6), in which activities are recorded in discrete time intervals. Activity diaries provide the most detailed analyses available using self-report methods (43). The objective measure we used was the TriTrac accelerometer, which researchers have found to have adequate reliability and validity (35,52). Hierarchical regression methods were used to establish the incremental variance explained by behavioral and psychological variables when child and parent psychological problems, child age, gender and socioeconomic status, and body composition and fitness were already in the regression model.



The subjects for this study were children and parents from 59 families accepted for entry into the Childhood Weight Control Program at the University at Buffalo. The children were entered in two cohorts in consecutive years at the same time of the year, with 33 families from the first cohort and 26 families from the second cohort. The sample of children consisted of 20 males and 39 females. Children were 10.5 ± 1.2 (mean ± SD) yr of age, 62.1 ± 17.6% overweight, had 34.1 ± 3.8% body fat, a body mass index (BMI) of 28.0 ± 3.2, sagittal diameter of 19.7 ± 2.3, and an estimated maximal oxygen uptake of 27.6 ± 5.8 ml·kg-1·min-1. Mothers were 39.9 ± 4.9 yr of age, 31.4 ± 28.5% overweight, had 35.4 ± 7.7% fat, and a BMI of 28.6 ± 6.0. Fathers were 43.4 ± 4.4 yr of age, 43.2 ± 17.8% overweight, had 28.5 ± 7.2% fat, and a BMI of 33.1 ± 4.2. Each family had one participating parent who completed the baseline assessment. This included 49 mothers and 10 fathers. Families were 96.6% white, 1.7% black, and 1.7% Hispanic, with socioeconomic status scores (SES) of 47.8 ± 10.0 using the Hollingshead's Four-Factor Index of Social Status (33).

This research was approved by the University at Buffalo Human Subjects Review Board, and informed consent was obtained from parents and the child.


During the baseline assessment, parents and children from families seeking entry to the childhood weight control program completed body composition measurement, a laboratory-constructed activity hedonics questionnaire, a decisional balance questionnaire to assess the perception of the advantages and disadvantages (Pros and Cons) of activity, and parent and child measures of psychological problems. The majority of the children and parents scored within normal limits on the psychological scales, with 5.1 and 6.8 of the children meeting clinical criteria on the Child Behavior Checklist Total Behavior Problem and Total Competence scores, respectively, and 11.9% of the parents meeting clinical criteria for the Cornell Medical Index. No parents had scores in the clinical range on the Beck Depression Inventory, the Bulimia Test, or the Interpersonal Inventory of Problems.

Children were also given submaximal fitness tests. During the next week, children and their participating parent recorded activity, and children wore the TriTrac accelerometer. The participating parent in each family also completed the battery of questionnaires to be described. The descriptive characteristics for the predictor variables are shown inTable 1 for children and participating parents.


Body composition. Subjects' height was measured using a stadiometer (Seca, Columbia, MD) calibrated to 1/8 inch (0.32 cm), and weight was measured on a balance beam scale (Healthometer, Bridgeview, IL). BMI was calculated as kg·m-2. Percent overweight was calculated by referencing the observed BMI to the 50th percentile BMI for the specific child and parent age and gender (39). Percent body fat was determined by bio-electrical impedance (Model BIA-101A, RJL Systems, Clinton Township, MI). Sagittal diameter was obtained using a Holtain Kahn abdominal caliper (Seritex, Carlstadt, NJ) with subjects measured in the supine position according to manufacturer instruction. Three measures were recorded to the nearest 0.1 cm, and the measures averaged.

Activity monitoring. Children's activity was recorded using a TriTrac accelerometer (TriTrac-R3D Research Ergometer, Hemokinetics, Inc., Madison, WI). Measured activity is stored in the monitor on a minute-by-minute basis and is downloaded to a computer through an interface. Children were instructed in procedures and precautions for using the TriTrac accelerometer that they wore on two weekdays after school, and for one full weekend day(usually Thursday, Friday, and Saturday). The children recorded the times that they wore the accelerometer, and any times they removed it (e.g., if they went swimming).

The TriTrac accelerometer collected minute-by-minute data in three planes of movement: medio-lateral (x), antero-posterior (y), and vertical (z), as well as a composite movement score for all three directions, called the“vector magnitude” ([x2 + y2 + z2]0.5). The accelerometer used the vector magnitude movement count to calculate estimated energy expenditure (kcal) for each minute of data collection. This calculation is the same as that used to convert Caltrac® activity counts to kilocalories (personal communication, Hemokinetics, Inc.). In addition, the accelerometer provided an estimate of resting metabolic rate using formulas provided by Hemokinetics that were developed with healthy adults. Bray and colleagues (12) have shown that these equations overestimate RMR by 7% in children. Each minute of accelerometer data was converted to METs (multiples of resting metabolic rate in kcal·kg·h-1(2)) by dividing the activity kilocalories computed for that minute by the child's calculated resting metabolic rate.

The TriTrac accelerometer succeeds the Caltrac® activity monitor, which was extensively tested for reliability, validity, and measurement(36,40). Comparison of the Caltrac® accelerometer with 24-h total energy expenditure assessed by indirect calorimetry in a wholeroom calorimeter showed significant correlations (r = 0.80) between Caltrac® estimated energy expenditure and calorimeter values, though the Caltrac underestimated total daily energy expenditure by about 13% (13). Initial validation of the TriTrac and related triaxial accelerometers suggests the methodology provides reliable and valid assessment of activity level and caloric expenditure(11,35,52). Bouten and colleagues(11) found a correlation of 0.95 between integrated triaxial accelerometer values and total energy expenditure assessed by oxygen consumption during rest and standardized activities.

Self-report of activity. Children and participating parents were instructed to keep detailed self-reports on the types of behaviors they engaged in, and the times these active behaviors began and ended during the same two weekdays and weekend day period that the children wore the accelerometer. The self-report recording sheets were divided into before school or work, and after school or work; they included general active and sedentary activities including watching television, walking, running, watching videotapes and other behaviors, as well as blank spaces to record specific behaviors that did not fit the sample categories. This approach to designing self-report sheets has been shown to produce more accurate indications of activity patterns in children than data gathered without division of the day into discrete time periods (6). Parents were instructed to review the recording sheets each evening, and assist the children as needed with recording. Upon return of the self-report sheets they were reviewed by a trained staff member and discrepancies were resolved by review with the parent and child. Only activities recorded for a duration of 5 min or more were considered in the analyses. Activity was converted to METs using values from the Compendium of Physical Activities (2).

Data Reduction For Activity Measures

Data for the TriTrac and self-report was reduced to 10-min blocks of time for all 3 d of recording from the hours of 630 and 2300 (16.5 h). The average activity value for each day was calculated, and the mean for the 3 d was then calculated. The mean and standard deviation of the number of accelerometer observation intervals for the two weekdays and the weekend day were 47.9± 14.4, 47.5 ± 14.6, and 76.1 ± 18.6 (8.0, 7.9, and 12.7 h, respectively), with 98.3% of the children having measures on 3 d, and 1.7% having measures on only 2 d.

Fitness. Children were given a submaximal fitness assessment using a Monark cycle ergometer (Monark 868, Monark-Crescent AB, Varberg, Sweden). Heart rate was monitored by a Polar Vantage XL heart rate monitor(Polar CIC Inc., Port Washington, NY) that provided a readout of bpm every 15 s. Children were seated for a 5-min rest prior to recording resting heart rate. They began the test at 150 kpm, and the workload was increased by 75-150 kpm when the subject had completed 3 min at a workload and their heart rate was stable (heart rate within 8 bpm in the last minute of the workload). Fitness testing was complete when the subject had achieved at least three workloads. The regression line between heart rate and workload was established, and the workload needed to achieve a heart rate of 200 bpm was estimated. Fitness was expressed as ml·kg-1·min-1 based on the oxygen consumption required for that workload(5).

Socioeconomic status. The Hollingshead's Four-Factor Index of Social Status (33) was used to assess family socioeconomic status. The four factors used to determine status were gender, marital status, education, and occupation. Education level and occupation status are scored on a predetermined scale, and these values are multiplied by a weight of three or five, respectively. The resulting values are then summed for the head(s) of households, and if there is more than one head of household, their individual scores are averaged.

Decisional balance. This 16-item questionnaire is designed to measure the perceived costs and benefits for exercise(37). This questionnaire was initially designed to predict factors that would be associated with movement across stages of change in exercise (37), and the concept of making decisions to exercise based on the perceived utility of exercising is relevant to predicting activity level. Persons who perceive the benefits of exercising exceeding the costs are more likely to be active than those who perceive the costs outweighing the benefits. The benefits and costs of exercise were standardized separately using a mean of 50 and standard deviation of 10, and the questionnaire scored by subtracting the con t-scores from the pro t-scores to yield one decisional balance score.

Activity hedonics. Activity hedonics, or liking for activity, was measured using a laboratory-constructed questionnaire that asked subjects to rate their liking for 39 activities using a five-point Likert scale, anchored by “don't like” (1), “O. K.”(3), and “like” (5). The questionnaire was scored by developing average hedonic ratings for the activity categories of sedentary (1.0-2.9 METs, 14 items), moderate intensity(3.0-4.9 METs, 8 items), high intensity (5.0-6.9, 7 items), and very high intensity (≥7 METs, 10 items) exercise.

Psychological tests. Parents completed a series of questionnaires designed to assess parent and child psychological distress and psychological problems. These measures were collected as part of ongoing research on the relationship between childhood obesity, child psychopathology, and parent psychopathology (23,24). The Child Behavior Checklist (CBCL) (1), a well-validated questionnaire, was used to determine childhood psychopathology. This questionnaire yields eight behavior problem scale scores for both boys and girls (Withdrawn, Somatic Complaints, Anxious/Depressed, Social Problems, Thought Problems, Attention Problems, Delinquent Behavior, Aggressive Behavior). In addition, a Total Problem score, an Internalizing Behavior Problem (Withdrawal + Somatic Complaints + Anxious/Depressed) and Externalizing Behavior Problem (Delinquent+ Aggressive Behavior) score were calculated. There are three competence scores: Social, School, and Activity Competence, and a Total Competence Score. The Total Behavior Problems and Total Competence standardized T-scores were used for the regression analysis. Clinical criteria for the CBCL are ≥67 for the Behavior Problems scales and ≤33 for the competence scales.

The Cornell Medical Index (CMI) (14) is a 195-item self-report questionnaire that assesses adult physical and psychiatric complaints. The psychiatric items comprise six scales (scales M-R) that assess inadequacy, depression, anxiety, sensitivity, anger, and tension. The total CMI score for all psychiatric scales has been used as a screening device to detect emotional disturbances (16) and has been shown to discriminate emotionally malad-justed from normal individuals(28), and was used as an independent variable. The Beck Depression Inventory (BDI) (9) is a measure of depressive symptomatology widely used in both psychiatric and nonpsychiatric populations. The psychometric properties of the BDI have been well documented, and the BDI has demonstrated high internal consistency and concurrent validity(8). The Bulimia Test (BULIT) (47) is a questionnaire designed to assess the symptoms of bulimia as described by the Diagnostic and Statistical Manual of Mental Disorders-III(3), including bingeing behavior, feelings after binges, weight fluctuations, and vomiting. The BULIT has demonstrated adequate reliability and predictive ability in both clinical and nonclinical populations. The Interpersonal Inventory of Problems (IIP)(34) is a 127-item scale used to assess interpersonal problems. The IIP has been shown to discriminate persons with interpersonal versus noninterpersonal problems, has demonstrated criterion validity with other indices of therapeutic improvement, and has adequate test-retest reliability (34). Based on the literature, the cutoffs we used to establish clinical criteria were eight for mothers and seven for fathers on the CMI (48), 17 for the BDI(10), and 102 for the BULIT (47). The IIP does not have clinical criteria, and a score two standard deviations from the mean, 2.1, was used to establish the cutoff(34).

Analytic Plan

The first stage of analysis was to compute basic descriptive statistics on each measure, and to run analyses of variance comparing the measures on each cohort to ensure they were similar. Second, the Pearson product moment correlation between accelerometer and self-report data was calculated, and a one-way analysis of variance compared the mean activity levels using these two methods. Third, separate hierarchical regression models were constructed to examine variables that would predict child self-reported activity and activity measured with the TriTrac. Hierarchical models were used to establish the predictors that added a significant amount of incremental variance in activity controlling for the predictors that were in the model. The child and adult psychological measures (CBCL, Beck, BULIT, IIP, CMI), were forced into the model in the first step to control for their effects. Predictor variables were added to the model in the following order: child age, gender, and family SES(second step); percent overweight, percent body fat, sagittal abdominal diameter and fitness (third step); and self-reported parent activity, child and parent activity hedonics, and child and parent decisional balance (fourth step). The variables were entered in this sequence to ensure that the theoretically derived psychological predictors of activity added variance beyond child or parent psychological problems, sociodemographic variables, and body composition. To reduce variables included in each step, individual models comprising just the variables within each conceptual step were constructed to identify the best set of predictors, and only predictor variables that had a significance level of at least (P < 0.10) were tested in the final hierarchical model.


Analysis of variance revealed no differences between cohorts for any of the predictors or children's self-reported and TriTrac activity data. The correlation for average MET values between accelerometer and self-report was significant (r = 0.46, df = 58, P < 0.001). The mean self-reported activity (2.26 ± 0.64) was significantly (41.2%) higher than accelerometer (1.60 ± 0.18) measured values (F(1,58) = 77.61, P < 0.001).

The hierarchical regression models are shown for prediction of average TriTrac or self-reported METs in Table 2. The beta coefficients (B), standard error of the coefficients (SE B), and the standardized beta coefficients (B) are presented for each of the psychological variables included in the final model, with the incremental R2 for each step and F and P values for the total model presented at the bottom of each portion of the table. In the accelerometer model, the variables introduced in Step 1, child and parent psychological problems, accounted for 8.3% of the variance, with a multiple R of 0.421 (P = 0.063). Socioeconomic status, introduced in Step 2, made a significant contribution to the model, increasing the variance accounted for by 6.8% (F(1,51) = 4.00, P = 0.05) for a total multiple R = 0.503 and an R2 of 0.151 (P = 0.029). Parent self-reported activity level, introduced in Step 3, increased the variance accounted for by 8.0% (F(1,50) = 5.33, P < 0.025) for a total multiple R = 0.581 and an R2 of 0.231 (P = 0.005).

In the self-report model, the variables introduced in Step 1 accounted for 3.4% of the variance, with a multiple R of 0.336 (P = 0.255). Variables introduced in Step 2 did not make an additional significant contribution to the model, but fitness introduced in Step 3 increased the variance accounted for by 23.5% (F(1,51) = 16.79, P < 0.01), for a total multiple R = 0.598 and an R2 of 0.269 (P = 0.001).

The beta weights and significance levels for each of the psychological variables that were entered together as Step 1 in the hierarchical model are also shown in Table 2. In the model that included all the psychological predictors the child total competence score on the CBCL(P = 0.034), and parent score on the interpersonal inventory of problems (P = 0.023), were significant predictors of accelerometer-measured activity, while child competence scores on the CBCL(P = 0.018) was a significant predictor of self-reported activity. If the variables that were not significant were removed from the equation using a backward regression procedure to produce the best-fitting model, the objective activity model would have included the total competence score on the CBCL (P = 0.038), parental distress on the CMI (P = 0.032), and parental interpersonal problems on the IIP (P = 0.005) for a multiple R of 0.41, and adjusted r2 = 0.124 (P = 0.008). The significant incremental variance added by SES in Step 2 was 0.056, and by parent activity in Step 4 was 0.076, for a total amount of variance accounted for of 0.256 (P = 0.001). There were no significant independent predictors based on psychological problems for self-reported activity.


The major finding in this study was that different sets of predictor variables were related to self-reported and objectively measured activity in obese children. The objectively measured activity was related to two previously studied factors, socioeconomic status (29) and parent activity levels (29,42), which accounted for 14.8% of the variance in activity. On the other hand, child self-reported activity was related to fitness, which is commonly related to activity (7,21,49), and accounted for 23.5% of the variance in activity. Dishman and colleagues also found with college students (20) that a different set of factors was related to diary-collected activity compared with accelerometer-measured activity. These two studies taken together suggest that the determinants of activity depend, in part, on the method of activity measurement. Given that the majority of determinants have been identified using self-reported activity, our current understanding of factors that influence activity may need to be modified as more research using objective measures of activity accumulate.

The correlation between self-report and objective measurement of activity was significant, but the magnitude of the relationship (r = 0.46) suggested the measures are not interchangeable. Self-report estimates of energy expenditure were almost 43% higher than those provided by the accelerometer. This is a very large discrepancy, even when potential underestimation of total energy expenditure over a day by accelerometers is considered(11). The overreporting of activity compared with objective measurement is consistent with other investigators who have shown greater activity levels for self-reported versus objective measures(38). The fact that the two measures of activity do not share the same predictors, that there are significant differences in activity data collected using self-report and TriTrac, and that these measures are only moderately correlated, suggest that the self-report and objective methods used in this study are not measuring the same activity construct.

The second important finding was that child and parent psychological problems were related to child accelerometer activity. In the model to predict accelerometer-measured activity, psychological variables entered in the first step accounted for 8.3% of the variance in activity. Adding significant predictors of activity accounted for an additional 14.8% of the variance in TriTrac activity. It should not be surprising that activity was related to child psychological problems, since activity levels may be a symptom of specific syndromes, such as depression or attention deficit disorder(4,15). We have observed psychopathology in a large minority of the obese children we treat (23), and the low activity levels that are observed in obese children may be due, in part, to these psychological problems. For example, a low level of activity in obese children could be attributed to their body weight, while it also might be due in part to depression or social withdrawal experienced by many obese children (23). Parent psychological problems are perhaps less directly related to child activity, but there are several plausible pathways. Parents who are experiencing psychopathology such as depression may show a disruption in their own activity, which would reduce the degree to which they could serve as a model for their children's activity, and parent psychopathology can interfere with parenting behaviors(26,27), which would decrease the support and reinforcement they were able to provide for their children to be active. In any case, the large degree of variance accounted for by child and parent psychological problems suggest that these are factors that should be controlled for in subsequent research.

There are several limitations that must be considered in interpreting these data. The study population was only obese children and parents who had presented for treatment of their obesity. This is relevant to understanding determinants of this portion of the population of children, but the findings may not generalize to other samples of children. The use of only obese children may explain the failure to find adiposity measures were related to activity, which is often observed in samples that include both obese and nonobese children (21,31,50). In addition, the measure of adult activity level was self-reported, which may influence the true relationship between adult and child activity. Finally, only a sampling of variables that can influence activity were studied. As reviewed by Sallis and colleagues (45), there are a wide variety of variables that are consistently related to activity. The variables that were related, parental activity (29,42), socioeconomic status (29), and fitness(21,44,49) were chosen because they have been previously studied and shown to be related to child activity. It is possible that different models may have been observed if a different set of variables were studied. On the basis of these results, and those of Dishman and colleagues (20), even if a different set of predictors was used, the models for self-report and objectively measured activity would have been different.

In summary, this study provides additional support for the need to use objective measures of activity when the goal is to assess determinants of activity. In research in which both the predictors of activity and activity level are measured by self-report, biases inherent in self-report may be responsible for a portion of the shared variance and the observed relationships between subjective measures of the independent and dependent variables. Future research is needed to assess factors that influence activity in other clinical populations using objective measures for the determinants as well as activity level.


1. Achenbach, T. M. Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT: University of Vermont Department of Psychiatry, 1991.
2. Ainsworth, B. E., W. L. Haskell, A. S. Leon, et al. Compendium of physical activities: Classification of energy costs of human physical activities. Med. Sci. Sports Exerc. 25:71-80, 1993.
3. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 3rd Ed. Washington, DC: American Psychiatric Association, 1980.
4. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, 4th Ed. Washington, DC: American Psychiatric Association, 1994.
5. Astrand, P. and K. Rodahl. Textbook of Work Physiology: Physiological Bases of Exercise, 2nd Ed. New York: McGraw-Hill, 1977.
6. Baranowski, T. Validity and reliability of self report measures of physical activity: an information-processing perspective.Res. Q. Exerc. Sport 59:314-327, 1988.
7. Baranowski, T., C. Bouchard, O. Bar-Or, et al. Assessment, prevalence, and cardiovascular benefits of physical activity and fitness in youth. Med. Sci. Sports Exerc. 24:S237-S247, 1992.
8. Beck, A. T., R. A. Steer, and M. G. Garbin. Psychometric properties of the Beck Depression Inventory: twenty-five years of evaluation.Clin. Psychol. Rev. 8:77-100, 1988.
9. Beck, A. T., C. H. Ward, M. Mendelson, J. Mock, and J. Erbaugh. An inventory for measuring depression. Arch. Gen. Psychiatry 4:561-571, 1961.
10. Beckman, E. E. and W. R. Leber. The Handbook of Depression: Treatment, Assessment, and Research. Homewood, IL: Dorsey Press, 1985, pp. 984-986.
11. Bouten, C. V., K. A. Westerterp, M. Verduin, and J. D. Janssen. Assessment of energy expenditure for physical activity using a triaxial accelerometer. Med. Sci. Sports Exerc. 26:1516-1523, 1994.
12. Bray, M. S., J. R. Morrow, J. M. Pivarnik, and J. T. Bricker. Caltrac validity for estimating caloric expenditure with children.Pediatr. Exerc. Sci. 4:166-179, 1992.
13. Bray, M. S., W. W. Wong, J. R. Morrow, N. F. Butte, and J. M. Pivarnik. Caltrac versus calorimeter determination of 24-h energy expenditure in female children and adolescents. Med. Sci. Sports Exerc. 26:1524-1530, 1994.
14. Brodman, K., A. J. Erdmann, and H. G. Wolf.Cornell Medical Index Health Questionnaire Manual. New York: Cornell University Medical College, 1956.
15. Brown, D. R., J. B. Croft, R. F. Anda, D. H. Barrett, and L. G. Escobedo. Evaluation of smoking on the physical activity and depressive symptoms relationship. Med. Sci. Sports Exerc. 28:233-240, 1996.
16. Costra, P. T. and R. R. Mccrae. Psychiatric symptom dimensions in the Cornell Medical Index among normal adult males. J. Clin. Psychol. 33:941-946, 1977.
17. Delany, J. P., D. W. Harsha, J. C. Kime, J. Kumler, L. Melancon, and G. A. Bray. Energy expenditure in lean and obese prepubertal children. Obes. Res. 3:67-72, 1995.
18. Dietz, W. H. and S. L. Gortmaker. Do we fatten our children at the television set? Obesity and television viewing in children and adolescents. Pediatrics 75:807-812, 1985.
19. Dishman, R. K. The measurement conundrum in exercise adherence research. Med. Sci. Sports Exerc. 26:1382-1390, 1994.
20. Dishman, R. K., C. R. Darracott, and L. T. Lambert. Failure to generalize determinants of self-reported physical activity to a motion sensor. Med. Sci. Sports Exerc. 24:904-910, 1992.
21. Durant, R. H., T. Baranowski, T. Rhodes, et al. Association among serum lipid and lipoprotein concentrations and physical activity, physical fitness, and body composition in young children. J. Pediatr. 123:185-192, 1993.
22. Epstein, L. H., K. J. Coleman, and M. D. Myers. Exercise in treating obesity in children and adolescents. Med. Sci. Sports Exerc. 28:428-435, 1996.
23. Epstein, L. H., K. R. Klein, and L. Wisniewski. Child and parent factors that influence psychological problems in obese children.Int. J. Eating Disord. 15:151-158, 1994.
24. Epstein, L. H., M. D. Myers, and K. Anderson. The association of maternal psychopathology and family socioeconomic status with psychological problems in obese children. Obes. Res. 4:65-74, 1996.
25. Epstein, L. H., A. M. Valoski, R. R. Wing, and J. McCurley. Ten year outcomes of behavioral family-based treatment for childhood obesity. Health Psychol. 13:373-383, 1994.
26. Forehand, R., G. J. Lautenschlager, J. Faust, and W. G. Graziano. Parent perceptions and parent-child interactions in clinic-referred children: a preliminary investigation of the effects of maternal depressive moods. Behav. Res. Ther. 247:73-75, 1986.
27. Forehand, R., M. Wierson, A. McCombs, G. Brody, and R. Fauber. Interparental conflict and adolescent problem behavior: An examination of mechanisms. Behav. Res. Ther. 27:365-371, 1989.
28. Gibson, H. B., R. Hanson, and D. T. West. A questionnaire measure of neuroticism using a shortened scale derived from the Cornell Medical Index. Br. J. Soc. Clin. Psychol. 6:129-136, 1967.
29. Gottlieb, N. H. and M. Chen. Sociocultural correlates of childhood sporting activities: their implications for heart health.Soc. Sci. Med. 21:533-539, 1985.
30. Griffiths, M. and P. R. Payne. Energy expenditure in small children of obese and non-obese parents. Nature 260:698-700, 1976.
31. Gutin, B., C. Basch, S. Shea., et al. Blood pressure, fitness, and fatness in 5-and 6-year-old children. J.A.M.A. 264:1123-1127, 1990.
32. Hay, J. A. Adequacy in and predilection for physical activity in children. Clin. J. Sport Med. 2:192-201, 1992.
33. Hollingshead, A. B. Four Factor Index of Social Status. New Haven, CT: Yale University, 1975.
34. Horowitz, L. M., S. E. Rosenberg, B. A. Baer, G. Ureno, and V. S. Villasenor. Inventory of interpersonal problems: psychometric properties and clinical applications. J. Consult. Clin. Psychol. 56:885-892, 1988.
35. Janz, K. F., J. Witt, and L. T. Mahoney. The stability of children's physical activity as measured by accelerometry and self-report.Med. Sci. Sports Exerc. 27:1326-1332, 1995.
36. Klesges, R. C., L. M. Klesges, A. M. Swenson, and A. M. Pheley. A validation of two motion sensors in the prediction of child and adult physical activity levels. Am. J. Epidemiol. 122:400-410, 1985.
37. Marcus, B. H., W. Rakowski, and J. S. Rossi. Assessing motivational readiness and decision making for exercise. Health Psychol. 11:257-261, 1992.
38. Matthews, C. E., and P. S. Freedson. Field trial of a three-dimensional activity monitor: comparison with self-report. Med. Sci. Sports Exerc. 27:1071-1078, 1995.
39. Must, A., G. E. Dallal, and W. H. Dietz. Reference data for obesity: 85th and 95th percentiles of body mass index (wt/ht 2) and triceps skinfold thickness. Am. J. Clin. Nutr. 53:839-846, 1991.
40. Noland, M., F. Danner, K. Dewalt, M. McFadden, and J. M. Kotchen. The measurement of physical activity in young children.Res. Q. Exerc. Sport 61:146-153, 1990.
41. Roberts, S. B., J. Savage, W. A. Coward, B. Chew, and A. Lucas. Energy expenditure and intake in infants born to lean and overweight mothers. N. Engl. J. Med. 318:461-466, 1988.
42. Ross, J. G. and R. R. Pate. The National Children and Youth Fitness Study II: a summary of findings. J. Phys. Educ. Recreat. Dance 56:35-39, 1985.
43. Sallis, J. F. Self-report measures of children's physical activity. J. Sch. Health 61:215-219, 1991.
44. Sallis, J. F., T. L. McKenzie, and J. E. Alcaraz. Habitual physical activity and health-related physical fitness in fourth-grade children. Am. J. Dis. Child. 147:890-896, 1993.
45. Sallis, J. F., B. G. Simons-Morton, E. J. Stone, et al. Determinants of physical activity and interventions in youth. Med. Sci. Sports Exerc. 24:S248-S257, 1992.
46. Shannon, B., J. Peacock, and M. J. Brown. Body fatness, television viewing and calorie-intake of a sample of Pennsylvania sixth grade school children. J. Nutr. Educ. 23:262-268, 1991.
47. Smith, M. C. and M. H. Thelen. Development and validation of a test for bulimia. J. Consult. Clin. Psychol. 52:863-872, 1984.
48. Stefansson, J. G. and I. Kristjansson. Comparison of the General Health Questionnaire and the Cornell Medical Index health questionnaire. Acta Psychiatr. Scand. 72:482-487, 1985.
49. Sunnegardh, J. and L. E. Bratteby. Maximal oxygen uptake, anthropometry and physical activity in a randomly selected sample of 8 and 13 year old children in Sweden. Eur. J. Appl. Physiol. Occup. Physiol. 56:266-272, 1987.
50. Sunnegardh, J., L. E. Bratteby, U. Hagman, G. Samuelson, and S. Sjolin. Physical activity in relation to energy intake and body fat in 8- and 13-year-old children in Sweden. Acta Paediatr. Scand. 75:955-963, 1986.
51. Tucker, L. A. The relationship of television viewing to physical fitness and obesity. Adolescence 21:797-806, 1986.
52. Welk, G. J. and C. B. Corbin. The validity of the TriTrac-R3D Activity Monitor for the assessment of physical activity in children. Res. Q. Exerc. Sport 66:1-8, 1995.


©1996The American College of Sports Medicine