Skip Navigation LinksHome > November 2008 - Volume 40 - Issue 11 > Tracking of Physical Activity and Inactivity in Middle Schoo...
Medicine & Science in Sports & Exercise:
doi: 10.1249/MSS.0b013e318180c390
BASIC SCIENCES: Epidemiology

Tracking of Physical Activity and Inactivity in Middle School Girls


Free Access
Article Outline
Collapse Box

Author Information

1Department of Epidemiology, University of North Carolina at Chapel Hill, NC; 2Department of Nutrition, University of North Carolina at Chapel Hill, NC; 3Department of Exercise and Sport Science, University of North Carolina at Chapel Hill, NC; 4Division of Epidemiology, The Ohio State University, Columbus, OH; and 5Department of Biostatistics, University of North Carolina at Chapel Hill, NC

Address for correspondence: Chris D. Baggett, Ph.D., Department of Epidemiology, CB 7435, University of North Carolina at Chapel Hill, NC 27599; Email:

Submitted for publication February 2008.

Accepted for publication May 2008.

Collapse Box


Purpose: The purpose of this study was to describe and compare the levels of tracking of physical activity and inactivity as assessed by self-report and accelerometry in middle school girls during a 2-yr period.

Methods: Participants (n = 951) were from the Trial of Activity for Adolescent Girls (TAAG). The TAAG intervention had minimal effect on physical activity; therefore, both intervention and control participants were included. Inactivity and physical activity were measured by accelerometry (MTI ActiGraph) and self-report (3-d physical activity recall).

Results: Weighted kappa statistics ranged from 0.14 to 0.17 across inactivity, moderate-to-vigorous physical activity (MVPA), and vigorous physical activity (VPA) for self-report, from 0.13 to 0.20 for 3-d accelerometry, and from 0.22 to 0.29 for a 6-d accelerometry. Intraclass correlations ranged from 0.17 to 0.22 for self-report, 0.06 to 0.23 for 3-d accelerometry, and 0.16 to 0.33 for a 6-d accelerometry. In general, the estimates from the 6-d accelerometry tended to be higher than those from self-report, whereas few differences were observed between 3-d accelerometry and self-report. Odds ratios (OR) for being in the highest quintile at eighth grade for those in the highest quintile at sixth grade compared with those in any other quintile at sixth grade were 3.26 (95% confidence interval = 2.28-4.67), 3.64 (2.55-5.20), and 3.45 (2.42-4.93) for the 6-d accelerometry-measured inactivity, MVPA, and VPA. Corresponding OR from self-report were 2.44 (1.66-3.58) for inactivity, 2.63 (1.83-3.79) for MVPA, and 2.23 (1.54-3.23) for VPA.

Conclusion: Tracking of inactivity and physical activity in middle school girls was fair to moderate. Our results suggest that physical activity and inactivity habits are dynamic for most girls during early adolescence. Population-based efforts should be made in this age group to promote physical activity and offer alternatives to inactivity for all girls.

Many studies have examined the stability, or tracking, of physical activity levels over time in children and youth. Tracking refers to the stability of relative rank or position within a group over time and is related to the ability to predict a measurement later in life knowing the value of the same variable earlier in life (8). Tracking of physical activity in children and youth is low to moderate at best; most tracking studies of children and youth have reported correlations of 0.30-0.60 depending on measurement method, age at baseline, and number of years between measurement (2,9,10,13,16,19,21,31). These findings indicate that there is considerable within-person variability in physical activity levels during childhood and adolescence and that the ability to predict an individual's current or future physical activity levels on the basis of past physical activity levels is modest.

Several studies have assessed tracking in childhood (9,19,31) and late adolescence (2,8,21); however, few have focused on early adolescence (11-14 yr) (10). Given the dramatic physiologic and psychosocial changes associated with puberty that occur during this time, it is feasible that tracking during early adolescence may be different than during other points in the life span. This may be particularly true in girls who begin sexual maturation earlier and move through the stages of maturation at a greater rate than boys during this time (26). At least one study has found that early-maturing girls have lower activity levels than later-maturing girls of the same age (3). Studies focusing on physical activity habits during this dynamic time of a girl's life may help identify adverse health behaviors that can be successfully targeted for intervention.

Self-reported physical activity is not usually precise, so it is plausible that the modest tracking of physical activity observed in studies that use self-reported measures was at least partly due to error inherent in the method. Objective measures of physical activity, such as accelerometry, may provide more accurate measures of physical activity habits than those obtained by self-report (23). To date, only a few studies have used an objective measure of physical activity to assess tracking of physical activity in youth (9,11,19,31), none of which have focused on the early pubescent years (10). Even fewer studies have assessed the tracking of inactivity using either objective or subjective methods (9-11) during childhood and adolescence. Additional studies using objective measurements, such as accelerometry, are needed to improve our understanding of the development and stability of physical activity habits in adolescents, which may translate into the development of more successful interventions. In addition, no study has compared tracking measures as assessed by accelerometry and self-report methods. A comparison of this kind would provide information on the ability of self-report to measure this phenomenon.

The purpose of this study was to investigate the tracking of physical activity and inactivity as assessed by 3 d of self-report and 3 and 6 d of accelerometry in a racially diverse group of middle school girls. We also compared tracking measures between self-report and accelerometry.

Back to Top | Article Outline


Back to Top | Article Outline
Study design.

Data were collected as part of the Trial of Activity for Adolescent Girls (TAAG). TAAG is a multicenter, group-randomized trial designed to test an intervention to reduce the usual decline in moderate-to-vigorous physical activity in middle school girls (25). TAAG has six field centers (at the Universities of Arizona, Maryland, Minnesota, and South Carolina; San Diego State University; and Tulane University). The project was coordinated by the University of North Carolina at Chapel Hill, and the project office at the National Heart Lung and Blood Institute collaborated on the work. Girls were recruited from six middle schools within each field center, for a total of 36 schools. The parent or guardian of each participant provided written informed consent, and girls provided assent. The study was approved by each participating university's Human Subjects Review Board.

Back to Top | Article Outline

The TAAG design included two cross-sectional samples of girls, one drawn from sixth graders at the beginning of the study in the spring of 2003, and a second drawn from eighth graders in the spring of 2005 after the implementation of the 2-yr intervention. In addition, we recruited all eighth grade girls who had been measured in sixth grade in 2003 and who attended a TAAG school in the spring of 2005 regardless of whether they were identified as part of the eighth grade cross-sectional random sample. A total of 951 girls who were measured at both sixth grade and at eighth grade were included in the current analyses. Given the modest effect of the TAAG intervention (29) on physical activity, participants from the intervention and control schools were pooled for these analyses.

Back to Top | Article Outline

Objectively measured physical activity and inactivity were assessed using the MTI ActiGraph accelerometer model 7164 (Manufacturing Technologies, Inc, Fort Walton Beach, FL). Six complete days of ActiGraph data were collected. Girls were instructed to wear the accelerometer on a belt around their waist over their right hip and were asked not to remove the ActiGraph except when sleeping, bathing, or swimming. Activity counts were accumulated over 30-s epochs during the 6 d. ActiGraph data were processed using methods described by Treuth et al. (27). Missing ActiGraph data were imputed using the Expectation Maximization (EM) algorithm (4). Previous work in the TAAG cohort determined that MET threshold ranges of 0-2.09, 2.1-4.59, 4.6-6.49, and ≥6.5 METs best discriminated between activities classified as inactive, light, moderate, and vigorous activity, respectively (27). These MET ranges corresponded to accelerometer count ranges of 0-50, 51-1499, 1500-2600, and >2600 counts·30 s−1 for inactivity, light, moderate, and vigorous activity, respectively (27).

Self-report of physical activity and inactivity was obtained using the 3-d physical activity recall (3DPAR) (20). The 3DPAR is a modification of the previous day physical activity recall, which has been validated in youth (30). Weston et al. (20) reported that 30-min blocks of moderate-to-vigorous physical activity (MVPA) and vigorous physical activity (VPA) from the 3DPAR were significantly correlated with both 3 d (r = 0.27-0.41) and 7 d (r = 0.35-0.45) of ActiGraph measurements. Girls recalled their past physical activity behavior for each of the three previous days. Each day was segmented into 36 30-min time blocks from 6 a.m. to midnight. A list of commonly performed activities was provided. Girls recorded the one activity that they performed for the longest period during each 30-min time block. There was no lower limit set on the period that the activity identified was performed.

For nonsedentary activities, participants rated the intensity of the activity as light, moderate, hard, or very hard. Illustrations of individuals performing an activity at each of the four intensities were provided to help participants select the proper intensity. All activities classified as "inactivity" were assumed to be performed only at a single intensity. MET values were assigned to each block using standard published values (1). MET values used as cut points for classifying the activity in blocks as inactive, light, moderate, or vigorous were the same as those for accelerometry.

From the 3DPAR, the numbers of blocks classified in each intensity category each day were totaled, and daily averages were calculated for each girl. The 3-d average was used in the analyses of self-reported activity. For objectively measured activity, minutes spent in each intensity category were summed over the course of a day. Two averages were obtained, one using all the 6 d measured and another limited to the same 3 d that were assessed using the 3DPAR to assess any differences between measurement methods that were a result of differing number of days and type of days (weekend vs weekdays).

At the completion of the 6-d ActiGraph monitoring period, participants completed the 3DPAR. This protocol resulted in having data from both self-report and ActiGraph on three corresponding days. The 3DPAR was most frequently completed on a Wednesday (33%) or a Tuesday (28%) and was never completed on a Saturday or Sunday. Therefore, weekend days were overrepresented in the 3-d sample with 47% weekdays and 53% weekend days assessed, as opposed to the 71% weekdays and 29% weekend days that would be expected had the sampling of the days been uniform. The 6-d accelerometry data were less skewed, with 67% of the data from weekdays and 33% of the data from weekend days. Previous work from TAAG has shown that physical activity levels differ by the day of the week (18).

Body mass was measured while wearing light clothing by use of an electronic scale (Model 770; Seca, Hamburg, Germany). Height was assessed without shoes using a portable stadiometer (Shorr Height Measuring Board; Shorr Productions, Olney, MD). Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Percent body fat was estimated from anthropometric measures using an equation that was developed by TAAG investigators for use in middle school girls (12).

Race/ethnicity was indicated via self-report on a six-item checklist that included 1) white (non-Hispanic), 2) African-American, 3) Hispanic, 4) Asian/Pacific Islander, 5) American Indian, and 6) other. A proxy measure of socioeconomic status (SES) was assessed at the school level by obtaining the proportion of students who received free or reduced cost lunch during the 2002-2003 school year.

Back to Top | Article Outline

In the sixth grade sample of girls, consent was obtained from 1721 girls. Of these girls, 118 had incomplete accelerometry data, 54 had fewer than 3 d of physical activity recall, 19 had incomplete body composition measures, 7 were missing age, and 1 was missing race/ethnicity. Of 1522 girls with complete measures in the sixth grade, 1243 were also assessed in the eighth grade sample. Among these girls, 290 had incomplete or missing accelerometry data and 2 had fewer than 3 d of physical activity recall. Thus, 951 girls provided data for the current analyses. Body composition and MVPA did not differ between those girls measured at both sixth and eighth grades compared with those measured only at sixth grade.

Back to Top | Article Outline
Statistical analysis.

Statistical analyses were conducted using SAS version 8.02 (SAS Institute, Cary, NC). There were only trivial differences in the tracking measures in the girls assigned to intervention and control schools, and therefore, the data were combined for all analyses. Weighted kappa statistics (6) were calculated to assess the agreement in inactivity and different intensities of activity between baseline and follow-up (tracking). To calculate the kappa statistics, physical activity and inactivity data from both self-report and accelerometry were categorized into quintiles. The kappa statistics were interpreted according to the recommendations of Munoz and Bangdiwala (17), whereby a kappa value of <0.00 equates to poor strength of agreement, 0.00-0.20 is fair agreement, 0.21-0.45 is moderate agreement, 0.46-0.75 is substantial agreement, and 0.76-1.0 is almost perfect agreement. Differences in the kappa statistics were assessed using the method described by McKenzie et al. (15) to compare kappa values drawn from the same sample.

Intraclass correlations (ICC) between sixth and eighth grades were calculated using PROC MIXED and methods described by Shrout and Fleiss (22). Differences between the ICC were assessed using the method described by Donner and Zou (7) to compare correlations drawn from dependent samples. The odds ratios of being in a specific quintile of MVPA, VPA, and inactivity at eighth grade given assignment in the same quintile at sixth grade, relative to those in any other quintile at sixth grade, were assessed using a logistic mixed model. For the ICC and logistic models, field center and school within field center were included as random effects in all models. Covariates tested included race/ethnicity, BMI, percent body fat, SES, age, and intervention assignment (treatment or control). Inclusion of these variables in the models did not improve the fit (likelihood ratio test, P > 0.05), and none of the estimates changed by more than 8% when the variables were included. Therefore, these potential covariates were not included in the analyses shown here.

Back to Top | Article Outline


Table 1 shows descriptive characteristics of the girls measured in the sixth and eighth grades. The sample was racially diverse (48.2% nonwhite). The mean time between measurements was 2 yr (SD = 0.14). On average, body weight, BMI, and percent body fat were higher in eighth grade than in sixth grade. There was an increase in accelerometer-measured inactivity, although self-reported inactivity was relatively stable. MVPA and VPA tended to decline from sixth to eighth grade for both accelerometer-measured and self-reported activity.

Table 1
Table 1
Image Tools

Table 2 presents the sample sizes and ranges of values across quintiles of inactivity, MVPA, and VPA for the 6-d accelerometry and self-report. For objectively measured physical activity, the sample sizes within the quintiles were almost identical. The ranges of the measured minutes were much larger in the extreme quintiles (quintiles 1 and 5) than those in quintile 2, 3, or 4. For example, in the sixth grade, the spans were 124 and 178 min, respectively, in the first and fifth quintiles, whereas the span was only 32 min in the middle quintile. The same trends were found in the 3-d means from accelerometry (data not shown).

Table 2
Table 2
Image Tools

The distribution of self-reported blocks of inactivity also showed larger spans in the extremes and relatively even numbers of participants within the five quintiles. However, for MVPA and VPA ties resulted in an uneven number of girls within quintiles. This was most evident for VPA in eighth grade because 421 girls (44% of the sample) reported no blocks. For this reason, self-reported VPA was analyzed in four rather than five categories.

Weighted kappa statistics and intraclass correlations (ICC) for self-report and 3- and 6-d accelerometry are presented in Table 3. For self-report and 3-d accelerometry measures, the kappa statistics suggested fair agreement between sixth andeighth grades, although the kappa statistic from 6 d of accelerometry indicated moderate agreement. Kappa values ranged from 0.14 to 0.20 across inactivity, MVPA, and VPA for both self-report and 3-d accelerometry, and from 0.22 to 0.29 for 6-d accelerometry. Significant differences (P < 0.05) were observed between kappa statistics produced from self-report and 6-d accelerometry, with the kappa values from 6-d accelerometry being higher for inactivity and MVPA. No differences were observed between kappa values from self-report and 3-d accelerometry.

Table 3
Table 3
Image Tools

For the self-report and the 3- and 6-d accelerometry, the highest ICC were for MVPA (ICC = 0.22, 0.23, and 0.33, respectively), whereas the lowest ICC were for inactivity (ICC = 0.17, 0.06, 0.16, respectively). The ICC of inactivity from the 3-d accelerometry was significantly lower (P < 0.05) than that from self-report. ICC for MVPA and VPA were significantly higher (P < 0.05) from the 6-d accelerometry compared with those from self-report.

To calculate the odds ratios shown in Table 4, dichotomous variables were created to indicate whether a girl's inactivity or activity was in a given quintile in sixth and eighth grades. These analyses showed that for accelerometry-measured activity and inactivity, the odds of being in the same quintile in both sixth and eighth grades were greatest for those in the extreme quintiles, whereas the odds of remaining in the middle three quintiles were smaller and generally not significantly different. For example, the odds of being in the highest quintile of inactivity (the most inactive) at eighth grade were greatest for those in quintile 5 at sixth grade (OR [95% CI] = 3.26 [2.28-4.67]) compared with those in any other quintile at sixth grade (quintiles 1-4), whereas the odds of being in the third quintile of inactivity at eighth grade were not different for those in the third quintile at sixth grade (OR [95% CI] = 1.32 [0.90-1.93]) compared with those in any other quintile at sixth grade (quintiles 1, 2, 4, and 5).

Table 4
Table 4
Image Tools

The odds ratios for being in a given quintile of self-reported inactivity, MVPA, and VPA in eighth grade given assignment in the same quintile at sixth grade are also presented in Table 4. In general, these odds ratios were smaller than those from accelerometry but displayed a similar trend. The odds ratios of remaining in the same quintile in both sixth and eighth grades were greatest for those in the extreme quintiles, whereas the odds ratios of remaining in any of the middle three quintiles were not significantly different from those in any other quintile at sixth grade.

Back to Top | Article Outline


This study showed that over a 2-yr period, tracking of physical activity and inactivity in middle school girls was fair to moderate depending on the assessment method and the days measured. Six days of accelerometry tended to provide higher tracking measures than 3 d. When the 3 d measured were matched, tracking assessments were similar whether measured by self-report or accelerometry.

To our knowledge, this is the first study to assess the tracking of physical activity and inactivity in middle school girls using accelerometry. Accelerometry generally provides a more accurate and reliable measure of physical activity than self-report (12,24). Earlier results from the TAAG study suggested that the reliability and validity of the 3DPAR decreased with each day of recall compared with accelerometry (27). It might be expected that the greater precision and repeatability associated with accelerometry would lead to higher tracking measures compared with self-report. Nevertheless, our results suggested little difference between tracking measures of MVPA or VPA compared with those from self-report when the measurement days used were matched. Only when physical activity was measured over 6 d did accelerometry produce higher tracking measures than 3DPAR.

The discrepancies in the comparison of 3DPAR to accelerometry from 3 d versus 6 d may be due to the percentage of days measured that were weekend days. It is known that there is greater variability (lower ICC) in physical activity measurement on weekends than on weekdays (14,27,28) when assessed by accelerometry. Thus, the overrepresentation of weekend days in our 3-d matched analysis may have resulted in lower tracking. It was interesting that the measured tracking was so similar between self-report and accelerometry in the 3-d matched analysis. This does not mean that self-report and accelerometry quantify physical activity and inactivity levels equally well, but that that the two assessment methods may have similar abilities to evaluate changes in physical activity and inactivity over time. Self-report may provide a good measure of tracking that requires less monetary resources and subject burden compared with accelerometry. Additional work is needed in comparing tracking measures of physical activity and inactivity from self-report to accelerometry with varying number of days.

We know of only three studies that have used accelerometry to assess tracking of physical activity in youth. In the Iowa Bone Development Study (9), accelerometers were used to assess tracking of physical activity over a 3-yr period in a group of elementary school-aged boys and girls. Spearman rank-order correlation coefficients between baseline and follow-up MVPA and VPA were modest (r = 0.32-0.40). Wilkin et al. (31) assessed tracking of physical activity using accelerometry in elementary school-aged children over a 1-yr period. Correlations for daily activity between years were moderate (r = 0.49 total, r = 0.36 girls, r = 0.55 boys, all P < 0.001). Kelly et al. (11) observed Spearman rank correlations of 0.35-0.37 for total PA, inactivity, and MVPA in a small group of young children (mean age at baseline = 3.8 yr) over a 2-yr period. Raw kappa statistics were 0.17, 0.013, and 0.21 for total PA, sedentary behavior, and MVPA, respectively. All of these studies used the same accelerometer model that we used, the data were reduced using alternate methods, and the length of time the monitors were worn varied; however, the conclusions from these studies are similar to ours in that fair-to-moderate tracking of accelerometer-measured physical activity and inactivity was found.

Tracking in girls similar in age to those studied here has been examined using self-report measures. Janz et al. (10) assessed tracking of vigorous activity and television viewing/video game playing over a 5-yr period in a small group of early adolescent girls (n = 62, mean age at baseline = 10.3 yr). Vigorous activity was assessed using the 3-d sweat recall and television viewing (TV)/video game playing using an interviewer-administered previous-day recall. They found strong tracking of vigorous activity with Spearman correlations between year 5 and each of the four preceding years ranging from 0.43 to 0.65. Low tracking was found in TV viewing/video game recall except between years 5 and 4 (Spearman correlation = 0.59); Spearman correlations between year 5 and years 1-3 ranged from 0.16 to 0.26. Thus, similar to our study, Janz et al. found some degree of tracking in self-reported activity and inactivity in adolescent girls. Discrepancies in the estimated level of tracking between our study and that of Janz et al. may have been influenced by differences in the measurement intervals and the recall methods used.

Intuitively, tracking should increase as the time interval between measures decreases. The 2-yr interval examined here is relatively brief. However, early adolescence is a stage in life of particularly great biological as well as psychosocial changes that may result in substantial changes in physical activity and inactivity levels. Indeed, Baker et al. (3) have demonstrated that 11-yr-old girls who experienced puberty early relative to their peers had lower objectively measured physical activity at age 13 yr than later-maturing girls, even after controlling for differences in physical activity and body composition at age 11 yr. Although we do not have a measure of pubertal status, our mean age at baseline was 11.9 yr and at follow-up was 13.9 yr. Nationally representative data from NHANES III indicated that at age 11.7 yr, approximately 25% of girls had experienced their first menstruation, while 2 yr later, 90% of girls had achieved menarche (5). Differences in the timing and tempo of sexual maturation throughout our 2-yr measurement period and changes in activity associated with maturation may have reduced tracking. Additional research is needed focusing on the physiologic as well as psychosocial and environmental determinants of tracking of physical activity habits in this age group.

The odds ratios of remaining in the same quintile of accelerometry-assessed inactivity, MVPA, and VPA over 2 yr were higher than being in a different quintile, and this effect was larger and statistically significant at the extremes of the distribution. Although the size of the odds ratios from self-report tended to be smaller than those from 6-d accelerometry, the trends were similar. This effect in the extreme quintiles may be due to the large range of the measured variables in the extreme quintiles compared with the range in the middle three quintiles. On average, a much greater change in minutes or blocks would be needed to move from either the lowest or the highest quintile at sixth grade into another quintile at eighth grade, although moving from one of the middle quintiles would require a smaller change. Conversely, these findings may suggest that these behaviors are already firmly rooted by early adolescence in a large portion of our population, and modification of these behaviors may be difficult in these girls.

Our study had several limitations. The 3DPAR has been validated for the measurement of total physical activity, MVPA, and VPA but not for inactivity. It should be noted that our use of a 3-d accelerometry measure is not consistent with the current recommended guidelines for accelerometry used in adolescents of 4-9 d to obtain a reliable measure of physical activity (28). Another limitation of this study is that comparisons of the 3- and 6-d accelerometry measures were complicated by the mixture of weekdays versus weekend days assessed.

The primary strength of this study was the use of both accelerometry and self-report to quantify the tracking of physical activity and inactivity. In addition, we used population-specific intensity cut points to determine time spent in inactivity, MVPA, and VPA. We used three different statistical methods to quantify tracking, and this allowed us to understand better the complexities of the issue. Kappa statistics and ICC provided global measures of agreement and described agreement over the entire distribution, whereas the odds ratios allowed us to look within specific portions of the distribution to identify where agreement occurred. The ICC analyzed the activity and inactivity data as continuous variables, whereas the kappa values and odds ratios examined data in categories. The ICC and odds ratio analyses tested for influences of demographic characteristics of the participants on tracking. Finally, this study included a larger, more ethnically diverse sample than has been previously reported for accelerometry-measured tracking of physical activity and inactivity.

Our results suggest that physical activity and inactivity habits are dynamic for most girls during early adolescence. Efforts should be made in this age group to promote physical activity and offer alternatives to inactivity to all girls, not just those presently inactive, to maintain or increase activity levels in those already active as well as increase activity in those currently inactive. Intervention before reaching early adolescence may be necessary to achieve recommended levels of activity in the least active girls, who already demonstrate a propensity to maintain unhealthy physical activity behaviors at this age.

The authors thank the faculty and staff of the 36 schools that participated in the trial. The authors thank all the investigators and support staff at the various study sites for their tireless efforts in conducting this trial. Finally, the authors thank all of the girls who participated in the intervention and measurements.

TAAG was funded by grants from the NIH/NHLBI (#U01HL-066845, HL-66852, HL-066853, HL-066855, HL-066856, HL-066857, and HL-066858).

Back to Top | Article Outline


1. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc. 2000;32(9 Suppl):S498-516.

2. Andersen LB, Haraldsdottir J. Tracking of cardiovascular disease risk factors including maximal oxygen uptake and physical activity from late teenage to adulthood. An 8-year follow-up study. J Intern Med. 1993;234(3):309-15.

3. Baker BL, Birch LL, Trost SG, Davison KK. Advanced pubertal status at age 11 and lower physical activity in adolescent girls. J Pediatr. 2007;151(5):488-93.

4. Catellier DJ, Hannan PJ, Murray DM, et al. Imputation of missing data when measuring physical activity by accelerometry. Med Sci Sports Exerc. 2005;37(11 Suppl):S555-62.

5. Chumlea WC, Schubert CM, Roche AF, et al. Age at menarche and racial comparisons in US girls. Pediatrics. 2003;111(1):110-3.

6. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20(1):37-46.

7. Donner A, Zou G. Testing the equality of dependent intraclass correlation coefficients. Statistician. 2002;51(3):367-79.

8. Fortier MD, Katzmarzyk PT, Malina RM, Bouchard C. Seven-year stability of physical activity and musculoskeletal fitness in the Canadian population. Med Sci Sports Exerc. 2001;33(11):1905-11.

9. Janz KF, Burns TL, Levy SM. Tracking of activity and sedentary behaviors in childhood: the Iowa Bone Development Study. Am J Prev Med. 2005;29(3):171-8.

10. Janz KF, Dawson JD, Mahoney LT. Tracking physical fitness and physical activity from childhood to adolescence: the Muscatine Study. Med Sci Sports Exerc. 2000;32(7):1250-7.

11. Kelly LA, Reilly JJ, Jackson DM, Montgomery C, Grant S, Paton JY. Tracking physical activity and sedentary behavior in young children. Pediatr Exerc Sci. 2007;19(1):51-60.

12. Loftin M, Nichols J, Going S, et al. Comparison of the validity of anthropometric and bioelectric impedance equations to assess body composition in adolescent girls. Int J Body Compos Res. 2007;5(1):1-8.

13. Malina RM. Tracking of physical activity and physical fitness across the lifespan. Res Q Exerc Sport. 1996;67(3 Suppl):S48-57.

14. Mattocks C, Leary S, Ness A, et al. Intraindividual variation of objectively measured physical activity in children. Med Sci Sports Exerc. 2007;39(4):622-9.

15. McKenzie DP, Mackinnon AJ, Peladeau N, et al. Comparing correlated kappas by resampling: is one level of agreement significantly different from another? J Psychiatr Res. 1996;30(6):483-92.

16. McMurray RG, Harrell JS, Bangdiwala SI, Hu J. Tracking of physical activity and aerobic power from childhood through adolescence. Med Sci Sports Exerc. 2003;35(11):1914-22.

17. Munoz S, Bangdiwala S. Interpretation of kappa, and B-statistics measures of agreement. J Appl Stat. 1997;24(1):105-11.

18. Murray DM, Stevens J, Hannan PJ, et al. School-level intraclass correlation for physical activity in sixth grade girls. Med Sci Sports Exerc. 2006;38(5):926-36.

19. Pate RR, Baranowski T, Dowda M, Trost SG. Tracking of physical activity in young children. Med Sci Sports Exerc. 1996;28(1):92-6.

20. Pate RR, Ross R, Dowda M, Trost SG, Sirard JR. Validation of a 3-day physical activity recall instrument in female youth. Pediatr Exerc Sci. 2003;15(3):257-65.

21. Raitakari OT, Porkka KV, Taimela S, Telama R, Rasanen L, Viikari JS. Effects of persistent physical activity and inactivity on coronary risk factors in children and young adults. The Cardiovascular Risk in Young Finns Study. Am J Epidemiol. 1994;140(3):195-205.

22. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychological Bulletin. 1979;86(2):420-8.

23. Sirard JR, Pate RR. Physical activity assessment in children and adolescents. Sports Med. 2001;31(6):439-54.

24. Sirard JR, Riner WF Jr, McIver KL, Pate RR. Physical activity and active commuting to elementary school. Med Sci Sports Exerc. 2005;37(12):2062-9.

25. Stevens J, Murray DM, Catellier DJ, et al. Design of the Trial of Activity in Adolescent Girls (TAAG). Contemp Clin Trials. 2005;26(2):223-33.

26. Sun SS, Schubert CM, Chumlea WC, et al. National estimates of the timing of sexual maturation and racial differences among US children. Pediatrics. 2002;110(5):911-9.

27. Treuth MS, Schmitz K, Catellier DJ, et al. Defining accelerometer thresholds for activity intensities in adolescent girls. Med Sci Sports Exerc. 2004;36(7):1259-66.

28. Trost SG, McIver KL, Pate RR. Conducting accelerometer-based activity assessments in field-based research. Med Sci Sports Exerc. 2005;37(11 Suppl):S531-43.

29. Webber LS, Catellier DJ, Lytle LA, et al. Promoting physical activity in middle school girls: trial of activity for adolescent girls. Am J Prev Med. 2008;34(3):173-84.

30. Weston AT, Petosa R, Pate RR. Validation of an instrument for measurement of physical activity in youth. Med Sci Sports Exerc. 1997;29(1):138-43.

31. Wilkin TJ, Mallam KM, Metcalf BS, Jeffery AN, Voss LD. Variation in physical activity lies with the child, not his environment: evidence for an 'activitystat' in young children (EarlyBird 16). Int J Obes (Lond). 2006;30(7):1050-5.

Cited By:

This article has been cited 4 time(s).

Medicina Dello Sport
Evaluation of motor activity, energy expenditure and physical fitness in children aged 8-11
Astegiano, P; Bartolomei, R; Ganzit, GP; Micheletti, G; Milanese, C; Scarzella, F
Medicina Dello Sport, 65(4): 437-457.

Obesity Facts
Tracking of Physical Activity from Childhood to Adulthood: A Review
Telama, R
Obesity Facts, 2(3): 187-195.
Pediatric Exercise Science
Tracking of Accelerometer-Measured Physical Activity in Early Childhood
Edwards, NM; Khoury, PR; Kalkwarf, HJ; Woo, JG; Claytor, RP; Daniels, SR
Pediatric Exercise Science, 25(3): 487-501.

European Journal of Public Health
Low vigorous physical activity at ages 15, 19 and 27: childhood socio-economic position modifies the tracking pattern
Suppli, CH; Due, P; Henriksen, PW; Rayce, SLB; Holstein, BE; Rasmussen, M
European Journal of Public Health, 23(1): 19-24.
Back to Top | Article Outline


©2008The American College of Sports Medicine


Article Tools



Search for Similar Articles
You may search for similar articles that contain these same keywords or you may modify the keyword list to augment your search.

Connect With Us