The ability to accurately assess physical activity (PA) has long been an issue of concern for exercise scientists, epidemiologists, and physical education specialists interested in the health-related outcomes associated with involvement in PA, and nowhere has the issue of measurement accuracy been associated with more difficulties than with children (5–12 yr). This difficulty is primarily due to children’s PA behavior characterized by sporadic, unstructured play, of which a large portion is spent in low- to moderate-intensity activities (1). Subsequently, a limited portion of PA is dedicated to those categorized as vigorous in nature (e.g., running) (25). Because of these characteristics, when assessing PA in this population, it is recommended that researchers focus on the total daily volume of PA (e.g., min·24 h−1, or steps per day), instead of categorizing activities based on intensity level (i.e., low, moderate, and vigorous) (7). Current recommendations acknowledge the need to focus on the daily accumulation of age-appropriate PA for a minimum of 60 min on most, if not all, days of the week (6,8), but still recognize the need for some of the activity to be of moderate (e.g., brisk walking) and vigorous (e.g., greater intensity than brisk walking) intensity, with several of the bouts lasting for 15 min in duration (8).
The above characteristics pose a considerable challenge to those attempting to measure or monitor children’s PA (32). For adolescents and adults, self-report methods have been traditionally used to quantify PA. However, it has been recognized that such methods may not be appropriate with children due to recall bias, social desirability, and, more importantly, the inability of children to distinguish between PA and non-PA behaviors (19,24). Additionally, activities that are low to moderate, such as walking, are unreliably recalled via self-report (4). In short, because children spend a large percentage of their time in low to moderate PAs and given that recommendations call for the increase in activities such as walking and free play (6,8), a need arises for the accurate assessment of these types of PAs.
Accurate estimates of children’s PA are most readily obtained from objective monitoring (6). One such method that has garnered increasing support is the pedometer (17,27,30). Pedometers are reported to have moderate to high correlations with various objective measures of PA. These include accelerometers (r = 0.50–0.98) (12), behavioral observation (r = 0.74–0.85) (22), heart rate monitoring (r = 0.62–0.88) (11), and volume of maximal oxygen uptake during treadmill exercise (r = 0.77) (14). These studies have been instrumental in establishing the construct validity of pedometers (28); however, a thorough examination of the accuracy of pedometer steps and time with children has yet to be conducted.
Considerable research assessing the accuracy of pedometers under controlled and self-paced walking (SPW) conditions has been reported with individuals 18 yr and older (9,10,13,21). Crouter et al. (9) examined the accuracy of 10 pedometers at five different walking speeds (54–107 m·min−1), and found pedometers to demonstrate considerable error in actual step counts at walking speeds <80 m·min−1. During SPW around an outdoor athletic track, Schneider et al. (21) observed sizable variations in the percent accuracy depending on the brand/model of the pedometer. Because these accuracy studies have taken place primarily in the adult population, the findings may not directly translate to children. That is, children are not small adults (32); therefore, applying these results to measure PA, without previous examination of their accuracy with children, allows the introduction of considerable error. The only study comparing hand-tallied steps with pedometer step counts in children (7–12 yr) revealed errors in pedometer accuracy during slow (58 m·min−1) walking, whereas moderate (70 m·min−1) and fast (90 m·min−1) speeds were in agreement (16). Unfortunately, this study (16) examined the accuracy of only one pedometer (Digiwalker SW-200), leaving questions as to the accuracy of additional models.
The ability to accurately create, and subsequently achieve, health-related standards for pedometer step counts (26) necessitates precision in measurement, which has yet to be fully examined with children. Therefore, the purpose of this study was to determine the accuracy of pedometer step counts and time in comparison with hand-tallied step counts and actual time with children (5–11 yr) during SPW and treadmill walking.
METHODS
Participants.
Ten boys and 10 girls (5–11 yr) were recruited through a youth summer activity program sponsored by the university. Chronological age (decimal) was computed by subtracting date of birth from date of measurement. Standing height (to the nearest 0.1 cm) and weight (to the nearest 0.1 kg) were taken for each participant. Stature was measured using a wall-mounted anthropometric stadiometer, and mass was determined using a digital scale (Model HD314; Tanita Corporation, Tokyo, Japan). Two measures were conducted for each variable with the average used for the final analysis. Descriptive data for boys and girls can be found in Table 1. Parental written informed consent and child assent were obtained before testing. All measures were conducted during June and July 2004. The study’s procedures were approved by the institutional review board.
;)
TABLE 1: Participant characteristics (mean ± SD).
Stride length.
Stride length was defined as the distance traveled by the same point on the same foot during two successive steps (one right and one left step length), and was assessed by asking the participants to complete 20 strides at their “normal” walking speed in an indoor hallway. The total distance traveled was measured to the nearest 1-inch using a 12-inch measuring wheel (Model RR318N; Keson, Aurora, IL) and divided by 20 to calculate average stride length.
Pedometers.
Two pedometer models, the Walk4Life 2505 (WL) (Plainfield, IL) and Digiwalker SW-200 (DW200) (Lee’s Summit, MO), were examined for accuracy during the SPW trials. Four models, the WL and DW200, along with Sun TrekLINQ (SUN) (Arvada, CO) and Digiwalker SW-701 (DW701), were examined under controlled (i.e., treadmill walking) conditions. Due to constraints placed on the removal of the participants from their regularly scheduled activities, only two pedometers were assessed during SPW, with all four models tested under controlled conditions. The two pedometers assessed during SPW were selected based on the extensive use of the DW200 in the pediatric population (11,16,18,23), and the WL because of the additional feature to record time concurrently with step counts. To ensure the proper calibration of each pedometer model (two pedometers per model, totaling eight pedometers), pre-, mid-, and postpedometer step count accuracy was assessed using the “shake-test” developed by Vincent and Sidman (31). Each model was securely placed in a box and shook 100 times with one edge of the box remaining in contact with a table surface to minimize aberrant counts. The shake-test was performed twice for each pedometer at pre-, mid-, and posttesting. Observed versus actual counts were calculated and expressed as a percentage. Results were consistent with those of Vincent and Sidman (31), with no pedometer exceeding ±1% accuracy (1 step count out of 100).
Part I: SPW.
Pedometer accuracy during SPW was assessed by having each child walk “at their normal pace” for three single-lap trials on the inside lane of an outdoor athletic track (400 m). Pedometer step counts were assessed concurrently during each lap using the DW200 and WL placed on the right and left sides of the body, midline of the thigh. Studies (3,9,16) indicate that placement of pedometer (right vs left) has no affect on step count accuracy; nonetheless, right and left side placement of model was alternated for each SPW test. The same units, one DW200 and one WL, were used for all tests. Hand-tallied step counts (actual) were recorded by two researchers for 11 of the 20 participants using hand counters, along with pedometer steps (DW200 and WL), pedometer time (WL), and actual time to complete each lap. All other observations of actual steps and time were recorded by the lead author. The observers walked behind each participant at a distance no less than 1.8 m. This was performed to minimize the influence on walking speed of the participants (21). Time required to travel the one-lap distance was recorded to the nearest second using a digital handheld stopwatch (Model 226; Sportline, Yonkers, NY). Distance of the inside lane was measured to the nearest inch using a 12-inch measuring wheel.
Part II: Treadmill protocol.
Before all treadmill tests, participants were provided time to become familiarized with walking on a treadmill. Each participant wore two units of the same model (four different models, totaling eight pedometers), marked as right and left and placed in their respective positions on the hip, midline of the thigh. Participants walked for 2 min at each of the following five treadmill speeds: 40, 54, 67, 80, and 94 m·min−1. During the walking trials, two investigators tallied actual step counts for 13 of the 20 participants using hand counters. All other observations of actual steps and time were recorded by the lead author. At the beginning of each trial, participants were instructed to straddle the treadmill while the walking speed was set. During this time, pedometers were set to zero. Participants were then instructed to walk for the duration of 2 min. At the 2-min mark, participants were instructed to straddle the treadmill, and pedometer steps (all models), time (WL and SUN), and actual steps were recorded for analysis. Participants walked for 2 min at each of the five treadmill speeds, beginning with the slowest speed (40 m·min−1) through the fastest (94 m·min−1), after which the pedometer models were switched and the protocol started over until all four models (two units each, right and left) were tested at each of the five speeds for all 20 participants. The treadmill was checked for speed calibration according to the manufacturer’s instructions using a digital tachometer (Model 21C13; Kernco Instruments, El Paso, TX) before testing participant 1 and 10.
Data Treatment
Part I: SPW.
Single-measure intraclass correlation coefficients (ICC) and 95% confidence intervals (95%CI) were calculated to assess the consistency between the following variables: 1) time to complete one lap for the final two SPW trials, 2) pedometer steps (DW200 and WL) with actual steps for the SPW trials, and 3) pedometer time (WL) with actual time to complete each SPW lap. Coefficients for the last two trials of the SPW test indicated moderate to high single-measure reliability, ICC = 0.822 (0.595–0.927), therefore SPW speed (m·min−1) was calculated using the average of the last two SPW trials (Table 1). Absolute value of percent error was calculated for each comparison according to the procedures described by Le Masurier et al. (13). Percent error ((pedometer steps or time − observed)/observed × 100) was computed to determine the over-/underrecording for each model. The average absolute value (regardless of positive or negative direction) was then calculated.
Part II: Treadmill walking.
Single-measure ICC (95%CI) were calculated for the following comparisons: 1) interunit agreement of pedometer steps (DW200, SUN, WL, DW701) and time (SUN and WL) by placement (right vs left) for each treadmill speed and 2) average (right and left) pedometer steps (DW200, SUN, WL, DW701) with actual steps and average pedometer time (SUN and WL) with actual time for each speed during treadmill walking. The absolute value of percent error (see above) was calculated for each comparison. The absolute value of percent errors were used to graphically display the magnitude of error for each pedometer (steps and time) by placement (right and left) for each treadmill speed.
Interobserver agreement.
Single-measure ICC (95%CI) were calculated to assess the consistency between observers for hand-tallied step counts during the SPW and treadmill walking trials. Coefficients between observers on both SPW and treadmill trials indicated high agreement with ICC = 0.973 (0.943–0.987) and ICC = 0.999 (0.999–0.999), respectively. Subsequently, all comparisons of pedometer steps with actual steps in the following analyses were performed using the average of observer step counts for the trials with multiple observers: 11 out of 20 and 13 out of 20 for SPW and treadmill walking, respectively. All additional trials were compared with hand-tallied step counts from one observer. In accordance with previous standards for evaluating ICC (5), the following guidelines were used to determine the level of agreement of the ICC calculated for each comparison: a) ≤0.79 is low agreement, b) 0.80–0.89 is moderate agreement, and c) ≥0.90 is high agreement.
RESULTS
Part I: SPW.
For the SPW test, pedometer steps (DW200 and WL) in comparison with actual steps resulted in high agreement, with ICC ≥0.985 across all trials (Table 2). Similar findings were observed for pedometer time (WL) recorded during SPW in comparison with actual time (ICC ≥ 0.997). Across the three SPW trials, the absolute value of percent error of steps for both pedometers did not vary more than 0.9%, whereas for time the absolute value of percent error was no greater than 1.3%.
TABLE 2: Intraclass correlation coefficients (95% confidence interval) for steps and time during self-paced walking trials.
Part II: Treadmill walking.
For the treadmill walking speeds, interunit agreement between right and left placement revealed low agreement between units at ≤54 m·min−1 (Table 3). A linear trend in agreement between sides was detected for each model, with agreement between units increasing as treadmill speed increased. Apart from one model (SUN), moderate to high interunit agreement (ICC ≥ 0.856) was achieved by 67 m·min−1. Similar to interunit agreement, a linear trend was observed between average (right and left) pedometer steps and actual steps, with increased agreement occurring with increasing speed (Table 4). High agreement to actual steps (ICC ≥ 0.904) occurred for three models (DW200, WL, and DW701) at speeds ≥67 m·min−1. Only one model (SUN) did not exhibit the linear trend in agreement, with agreement lowest (ICC = 0.08) at 54 m·min−1, whereas for the other models (DW200, WL, and DW701), the lowest ICC (≤0.599) were observed at 40 m·min−1. For pedometer time (SUN and WL) recorded during treadmill walking, high interunit agreement (ICC ≥ 0.952) was observed at speeds ≥54 m·min−1 for the WL (Table 5). Consistent with the step comparisons, time recorded by the SUN pedometers did not exhibit a linear trend in agreement, with the lowest agreement (ICC = 0.288) occurring at 54 m·min−1.
TABLE 3: Interunit intraclass correlation coefficients (95% confidence interval) for steps by placement (right vs left) during treadmill walking.
TABLE 4: Intraclass correlation coefficients (95% confidence interval) between average pedometer steps and actual steps recorded during treadmill walking.
TABLE 5: Intraclass correlation coefficients (95% confidence interval) between right and left placement for time during treadmill walking.
The absolute value for percent error for the right and left placement of pedometers across the five treadmill speeds are displayed in Figure 1. Despite the high interunit agreement observed with increasing speed (see above), the influence of placement on step count accuracy was evident across all models. For the SUN and WL models, left pedometer placement consistently resulted in a greater magnitude of the error for step counts in comparison with the right, whereas for the DW200 and DW701 models, placement on the right was associated with the greater discrepancy between actual steps. Although variations between placement were evident at slower speeds (≤54 m·min−1), as indicated with ICC ranging from 0.330 to 0.746, the average of the two sides for each speed was computed for each model and expressed as an absolute value of percent error. These results are displayed in Figure 2. Results indicate that three models (DW200, WL, and DW701) were within 5% of actual steps for speeds ≥67 m·min−1, whereas one model (SUN) achieved this accuracy until ≥80 m·min−1. Figure 3 displays the absolute value of percent error for pedometer time and steps of two models (SUN and WL). For the SUN, the difference between pedometer time and steps were minimal, signifying that the timer and step counter were registering in unison. For the WL, the mean percent difference between pedometer time and steps was greater, indicating that time and steps were not registering concurrently. A further examination of pedometer time revealed that the average WL time was accurate within ±5.3% of actual time across all speeds, regardless of placement and steps counts, whereas the SUN exhibited inaccurate recording of time at slower speeds and did not come within ±5% accuracy until 80 m·min−1.
DISCUSSION
The purpose of this study was to examine the accuracy of pedometer steps and time during SPW and controlled walking with children. The results found that overall pedometer steps from each model (DW200, SUN, WL, and DW701) exhibited high agreement with actual steps (ICC ≥ 0.931) at treadmill speeds ≥80 m·min−1, and two models (DW200 and WL) at 77 m·min−1 (SPW). Moreover, one pedometer (WL) was accurate in recording time during SPW and treadmill walking, regardless of placement (right vs left), speed, and step count accuracy.
The results of the SPW trials revealed almost complete precision of step counts (DW200 and WL) and time (WL) in relation to hand-tallied steps. This suggests that during SPW, values obtained by these two models can be considered to be accurately representing the number of actual steps and time spent in ambulatory activity. In examining the accuracy of steps and time from SPW in relation to the results from the treadmill trials, the high degree of accuracy during SPW mirrors that from treadmill walking at 80 m·min−1 for the two models (DW200 and WL), with only slightly lower ICC observed for the WL on the treadmill (Tables 2 and 4). Individual data for the SPW showed that only one child (girl, 8 yr) walked slower than 67 m·min−1 (SPW speed = 65.7 m·min−1). Thus, it appears that children, when asked to cover the distance of one lap around an athletic track, walk at speeds that concur with high levels of step count agreement observed during controlled laboratory walking.
The consistent variation observed between units (right vs left placement) is suggested to be an artifact of quality control in manufacturing (3). Differences in quality control would cause pedometers of the same brand/model to record step counts with varying accuracy. This would be of considerable concern for researchers assessing large samples, leading to questions as to whether the same model, purchased in quantity, would all record steps with similar precision. In assessing bilateral reliability (right vs left placement), the same pedometer model, yet with varying degrees of step count agreement between units, may result in favoring one placement over the other depending on which side the “lesser” quality unit was attached. Similar interunit disagreement was noticed by Bassett et al. (3) in their examination of five pedometers at various walking speeds, in which they found only one model (Yamax Digiwalker DW-500) to not significantly differ by placement. Our observations found that two models (SUN and WL) consistently demonstrated greater agreement with actual steps on the right, whereas two models (DW200 and DW701), without fail, were more accurate on the left (Fig. 1). In contrast, several studies (2,9,16) indicate that no significant differences exist for pedometer placement. Crouter et al. (9) reported ICC across five treadmill speeds (54–107 m·min−1) between right and left placement to be ≥0.81 in 8 out of 10 different brands. If we compute ICC across the five speeds in this study, similar results are observed between placements (Table 6). However, several of the overall ICC from this study are slightly lower than the ICC reported by Crouter et al. (9), and may be from the reporting of single versus average ICC, of which the latter would be greater due to the inclusion of multiple trials in the calculation (15). Barfield et al. (2) reported single-day interunit consistency for the DW200 worn during elementary (3rd through 5th grades) classroom, recess, and physical education to be 0.94, 0.98, and 0.92, respectively. These findings are slightly above the interunit assessment for the DW200 at the speed of 67 m·min−1 (ICC = 0.856), and suggest that children may be required to move at speeds >67 m·min−1 to demonstrate high levels of interunit agreement. Interestingly, the pre-, mid-, and postpedometer shake-test (31) did not reveal differences between units. Although the vertical force generated during the shake-test was not determined, it is assumed that this force was beyond the threshold for step count registry, thereby explaining why the shake-test was not sensitive enough to detect the differences. It is possible that the systematic differences observed between units may be due to variations in spring tension, with tighter tensions causing consistent underrecording of steps. Greater spring tension would require additional vertical acceleration to “trigger” a step count. This would explain why at slower speeds the difference between units was greatest, yet as speed increased, and with that vertical acceleration, units recorded almost in symmetry. Because the discrepancies were systematic and not random, we conclude that placement does not appear to be cause for variation between units but rather contributes to a combination of differences in spring tension and walking speed. Systematic error may allow a correction factor to be applied to pedometer units known to over-/underrecord actual values. However, this would require testing all units before use and thus may not be feasible to do on a large-scale basis.
TABLE 6: Overall (five treadmill speeds) intraclass correlation coefficients (95% confidence interval) between right and left steps, average (right and left) steps and actual steps for all models, and right and left time (SUN and WL) during treadmill walking.
The more surprising finding was in regard to the accuracy of time from the WL pedometer, independent of step count accuracy (Fig. 3). Across all speeds and between units, the WL time did not vary more than 5.3 s from actual time. The time differences observed were most likely caused by variations in participant “start” and “stop” for each treadmill speed. That is, although the stopwatch was initiated when the participants began walking and terminated at the end the 2-min trial, the inability to precisely synchronize the two would cause a minimal, yet detectable difference between the actual 2-min walking period and time recorded by the pedometer. However, these differences were negligible, and therefore time recorded from this pedometer model can be viewed as providing an accurate representation of time spent in ambulatory activity. The accuracy of time, independent of step counts, also suggests that, in this model (WL), the lever arm is being “triggered” to record steps (i.e., opening the electronic circuit to record), but that the force generated was not great enough to close the arm before the beginning of a successive step. This would explain why time, which should only record during step count registry, was more accurate than step counts.
Apart from manufacturing issues referred to above, the considerable variation observed between placement (interunit agreement) and across treadmill speeds with one unit (SUN) is likely due to the addition of a random-movement filter. This feature prevents step count recording until five consecutive steps are taken, at which time the five steps will be entered into the total. If only four steps are taken, with a pause in between the fourth and fifth steps, the pedometer will register zero. The result would be the occurrence of false negatives, whereby steps are taken (in this case four steps), yet the pedometer fails to record them (13). The consequence would be consistent underrecording of steps and lead to inaccurate estimates of daily activity levels. Further, variations in sensitivity thresholds across pedometer models, and in this case (SUN) detection thresholds (i.e., random-movement filter), would create difficulties when making comparisons across studies using different pedometer models (13). Using the SUN would therefore seem to be appropriate only under conditions in which continuous movement is expected and occurs at speeds ≥80 m·min−1. However, it is unlikely that such conditions would exist in free-living situations, given the intermittent nature of children’s PA (1).
The ability to record ambulatory activity in units of time (i.e., hours, minutes, seconds) is in agreement with current PA recommendations suggesting that children should participate in age-appropriate activities, of moderate intensity (e.g., brisk walking), for at least 60 min·d−1 (6,8). Furthermore, given the unobtrusive nature of the pedometer (i.e., resides on the individual’s waistline), low cost per unit, and the concurrent assessment of steps and time, it appears that the WL model would be suitable for examining various aspects of health-enhancing PA: steps (26) and time (6,8,29). However, it should be noted that the WL pedometer recorded time accurately during SPW and all treadmill speeds (Table 2 and Fig. 3). Although this may seem to be a desirable property of a measurement tool, it indicates that not only would time spent in moderate- and high-intensity activity be recorded, but also time spent at lower levels of intensity. Thus, the total amount of time accrued over the duration of a day is likely a combination of time spent in all activities, regardless of intensity level, thereby limiting the ability to determine the proportion of activity time spent at various levels of intensity and for what durations these occurred. Nonetheless, the use of pedometer time to measure PA reflects a novel approach in determining whether children are meeting activity guidelines and deserves further study of the practicality of large-scale use and surveillance.
Several limitations need to be addressed in light of these findings. First, the authors were unable to examine all pedometer models during SPW. As stated previously, the reason for this was the inability to remove the children for longer periods of time from the camp activities. Given this constraint, we decided to assess pedometers that appeared regularly in the literature (DW200) focusing on the pediatric population (11,16,18,23) and one that records concurrently with step counts (WL), reserving the additional models for treadmill assessment. Despite this drawback, this study is the first to thoroughly examine the accuracy of these pedometers under self-paced and controlled conditions. Further, although three pedometers (DW200, WL, and DW701) were found to be in agreement with observed steps at speeds ≥67 m·min−1, walking is only one of the many forms (e.g., free-play, sports) of PA, and therefore the results may not be directly transferable to assessing such activities.
In conclusion, the findings from the SPW indicate high levels of agreement for steps for the DW200 and WL, and high agreement for time with the WL. Mixed, but not unexpected, results were observed during the five treadmill speeds, with three models (WL, DW200, and DW701) exhibiting a high degree of step count agreement at speeds ≥67 m·min−1. To the authors’ knowledge, this is the first study reporting the accuracy of time derived using a pedometer. Given the increased number of pedometers on the market that have the capability to assess time concurrently with steps counts, this additional feature may play a complementary role in the promotion and assessment of PA by quantifying activity in more familiar units (i.e., hours, minutes, seconds), which coincides with national recommendations (8). However, the selection of a pedometer that records time should not be done haphazardly, and features such as random-movement filters (SUN) may provide results not directly comparable with those with other models. Contrary to studies with adults (9,20,21), interunit agreement was below acceptable levels at slow walking speeds, and therefore typical walking speed should be considered when selecting a pedometer model to use in the pediatric population. Nonetheless, three pedometers (WL, DW200, and DW701) exhibited acceptable or near-acceptable levels of interunit agreement, and high agreement between pedometer and observed steps at ≥67 m·min−1 (ICC ≥ 0.856), and are therefore recommended for use with this age group.
REFERENCES
1. Bailey, R. C., J. Olson, S. L. Pepper, J. Porszaz, T. J. Barstow, and D. M. Cooper. The level and tempo of children’s physical activities: An observational study.
Med. Sci. Sports Exerc. 27:1033–1041, 1995.
2. Barfield, J. P., D. A. Rowe, and T. J. Michael. Interinstrument consistency of the Yamax Digi-Walker pedometer in elementary school-aged children.
Meas. Phys. Ed. Exerc. Sci. 8:109–116, 2004.
3. Bassett, D. R., B. E. Ainsworth, S. R. Leggett, et al. Accuracy of five electronic pedometers for measuring distance walked.
Med. Sci. Sports Exerc. 28:1071–1077, 1996.
4. Bassett, D. R., A. L. Cureton, and B. E. Ainsworth. Measurement of daily walking distance-questionnaire versus pedometer.
Med. Sci. Sports Exerc. 32:1018–1023, 2000.
5. Baumgartner, T. A., A. S. Jackson, M. T. Mahar, and D. A. Rowe.
Measurement for Evaluation in Physical Education and Exercise Science. 7th Ed. New York: McGraw-Hill, 2003, p. 95.
6. Cavill, N., S. J. Biddle, and J. F. Sallis. Health enhancing physical activity for young people: Statement of the United Kingdom Expert Consensus Conference.
Pediatr. Exerc. Sci. 13:12–25, 2001.
7. Corbin, C. B., R. P. Pangrazi, and G. J. Welk. Toward an understanding of appropriate physical activity levels for youth.
PCPFS. 1:1–8, 1994.
8. Council on Physical Education for Children. Physical Activity for Children: A statement of guidelines for children ages 5–12. In:
Strategies. National Association for Sport and Physical Education. Physical Activity for Children: A Statement of guidelines for children ages 5–12. 2nd Ed. Reston, VA: 2004, pp. 3–8.
9. Crouter, S. E., P. L. Schneider, M. Karabulut, and D. R. Bassett. Validity of 10 electronic pedometers for measuring steps, distance, and energy cost.
Med. Sci. Sports Exerc. 35:1455–1460, 2003.
10. Cytarto, E. V., A. M. Myers, and C. Tudor-Locke. Pedometer accuracy in nursing home and community-dwelling older adults.
Med. Sci. Sports Exerc. 36:205–209, 2004.
11. Eston, R. G., A. V. Rowland, and D. K. Ingledew. Validity of heart rate, pedometry, and accelerometry for predicting the energy cost of children’s activities.
J. Appl. Physiol. 84:362–371, 1998.
12. Kilanowski, C. K., A. R. Consalvi, and L. H. Epstein. Validation of an electronic pedometer for measurement of physical activity in children.
Pediatr. Exerc. Sci. 11:63–69, 1999.
13. Le Masurier G. C., S. M. Lee, and C. Tudor-Locke. Motion sensor accuracy under controlled and free-living conditions.
Med. Sci. Sports Exerc. 36:905–910, 2004.
14. Louie, L., R. G. Eston, A. V. Rowlands, K. K. Tong, D. K. Ingledew, and F. H. Fu. Validity of heart rate, pedometry, and accelerometry for estimating the energy cost of activity in Hong Kong Chinese boys.
Pediatr. Exerc. Sci. 11:229–239, 1999.
15. Morrow, J. R., and A. W. Jackson. How “significant” is your reliability?
Res. Q. Exerc. Sport 64:352–355, 1993.
16. Ramirez-Marrero, F., B. A. Smith, T. E. Kirby, N. Leenders, and W. M. Sherman. Evaluation of a step-counter during treadmill walking in 7–12 year old African-American children.
J. Natl. Black Nurses Assoc. 13:1–5, 2002.
17. Rowlands, A. V., R. G. Eston, and D. K. Ingledew. Measurement of physical activity in children and adolescents with particular reference to the use of heart rate and pedometry.
Sports Med. 24:258–272, 1997.
18. Rowlands, A. V., R. G. Eston, and D. K. Ingledew. Relationship between activity levels, aerobic fitness, and body fat in 8- to 10yr-old children.
J. Appl. Physiol. 86:1428–1435, 1999.
19. Sallis, J., and B. Saelens. Assessment of physical activity by self-report: Status, limitations, and future directions.
Res. Q. Exerc. Sport 71(Suppl 2):S1–S14, 2000.
20. Schneider, P. L., S. E. Crouter, and D. R. Bassett. Pedometer measures of free-living physical activity: Comparison of 13 models.
Med. Sci. Sports Exerc. 36:331–335, 2004.
21. Schneider, P. L., S. E. Crouter, O. Lukajic, and D. R. Bassett. Accuracy and reliability of 10 pedometers for measuring steps over a 400-m walk.
Med. Sci. Sports Exerc. 35:1779–1784, 2003.
22. Scruggs, P. W., S. K. Beveridge, P. A. Eisenman, D. L. Watson, B. B. Shultz, and L. B. Ransdell. Quantifying physical activity via pedometry in elementary physical education.
Med. Sci. Sports Exerc. 35:1065–1071, 2003.
23. Treuth, M. S., N. E. Sherwood, and N. F. Butte, et al. Validity and reliability of activity measures in African-American girls for GEMS.
Med. Sci. Sports Exerc. 35:532–539, 2003.
24. Trost, S. G., A. M. Morgan, R. Saunders, G. M. Felton, D. S. Ward, and R. R. Pate. Children’s understanding of the concept of physical activity.
Pediatr. Exerc. Sci. 12:293–299, 2000.
25. Trost, S. G., R. R. Pate, and J. F. Sallis, et al. Age and gender differences in objectively measured physical activity in youth.
Med. Sci. Sports Exerc. 34:350–355, 2002.
26. Tudor-Locke, C., R. P. Pangrazi, C. B. Corbin, et al. BMI-referenced standards for recommended pedometer-determined steps/day in children.
Prev. Med. 38:857–864, 2004.
27. Tudor-Locke, C., J. E. Williams, J. P. Reis, and D. Pluto. Utility of pedometers for assessing physical activity.
Sports Med. 32:795–808, 2002.
28. Tudor-Locke, C., J. E. Williams, J. P. Reis, and D. Pluto. Utility of pedometers for assessing physical activity: Construct validity.
Sports Med. 34:281–291, 2004.
29. U.S. Department of Health and Human Services. Physical activity and fitness. In:
Healthy People 2010, 2nd Ed. Washington, DC: U.S. Government Printing Office, 2000, p. 22.
30. Vincent, S. D., and R. P. Pangrazi. Does reactivity exist in children when measuring activity levels with pedometers?
Pediatr. Exerc. Sci. 14:56–63, 2002.
31. Vincent, S. D., and C. L. Sidman. Determining measurement error in digital pedometers.
Meas. Phys. Ed. Exerc. Sci. 7:19–24, 2003.
32. Welk, G. J., C. B. Corbin, and D. Dale. Measurement issues in the assessment of physical activity in children.
Res. Q. Exerc. Sport 71:59–73, 2000.
