Amid increasing trends for overweight and obesity of epidemic proportions, self-reported participation in leisure-time physical activity (PA) has remained relatively stable (5). Societal transitions in sedentary behaviors (e.g., increased television viewing and passive commuting modes) likely contribute to the current obesity epidemic, although these types of lifestyle behaviors are challenging to assess (28) and their impact on weight gain is therefore only speculative at this time.
A recurring issue in PA assessment has been researchers’ difficulty in quantifying trends in overall PA (e.g. leisure-time, transportation, occupation-related, household) using currently available data (15). To date, self-report methods including diary/records/logs and survey/questionnaires have been the preferred approach to quantification, primarily due to their practicality (i.e., ease of administration). Such methods sometimes portray physical activity data categorically (e.g., insufficiently or sufficiently active) and produce relatively imprecise results. Even when data are collected as continuous variables (e.g., minutes in an activity), rounding and digit preference errors may make the variable perform more like a categorical one (7). It has also become increasingly apparent that, in addition to the well-known limitations of such methods (e.g., recall bias and floor effects), results vary with specific instruments and scoring procedures used (19). Further, some activities, particularly walking, are unreliably recalled (3,18). Because walking is both the most popular leisure-time PA in the United States (16) and a feature of utilitarian daily PA, any assessment of PA should be sensitive to walking in all its forms.
Recent technological advances have spurred a tremendous interest in objective monitoring of PA. Motion sensors (accelerometers and pedometers) are increasingly being used as alternatives and/or adjuncts to self-report methods, primarily due to their sensitivity to walking behaviors and their capacity to objectively quantify PA as a continuous variable (29). Accelerometers can be used to assess volume of PA (activity counts), but they can also capture activity counts in very small units of time (e.g., 30 s, 1 min). They can therefore be used to ascertain time spent in bouts of specific intensity categories (e.g., light, moderate, vigorous). Accelerometers have become important activity assessment tools despite their high cost (e.g., $450 per unit for MTI, formerly CSA, accelerometers), the need for supporting hardware and software, and the oftentimes protracted data management demands (28). Pedometers are less expensive ($15–$30) motion sensors that produce a total output (steps taken, typically reported as steps per day) that correlate strongly (median r = 0.86) with different accelerometers (30). Pedometers typically operate using a horizontal, spring-suspended lever arm that bounces with vertical motion during ambulation (e.g., walking, running). Each movement detected above a critical threshold (e.g., 0.35 g) is counted as a step taken (24). Although pedometers are not designed to capture intensity of PA as accelerometers do, they detect steps taken with acceptable accuracy (2). Aggregated validity evidence of pedometers compared with accelerometers, direct observation, measures of energy expenditure, and self-report measures provides ample support for using pedometers to assess PA (30).
Expected values (e.g., constant values for different populations) for pedometer-determined steps per day have been compiled for various populations, and patterns are discernible (29). For example, males appear to consistently take more steps per day than females (29). Further, steps per day is inversely associated with body mass index (BMI) (25). A closer examination of pedometer data in relation to sex and other important demographic variables such as race, age, education, and income is warranted. In addition, although day-to-day, or intra-individual, variability is now generally accepted as a feature of objective PA measurement (22), few pedometer data exist to inform this line of inquiry (26). Therefore, the dual purposes of this study were to: 1) provide preliminary descriptive epidemiologic data representing pedometer-determined PA in a South Carolina adult sample and 2) explore sources of intra-individual variability of steps per day.
METHODS
Details concerning study participants and procedures summarized here were previously reported in a paper focused on the process of collecting self-monitored pedometer data by mail (27). The Institutional Review Board for Research at the University of South Carolina approved the study as outlined.
Participants.
Participants were recruited through a randomized, population-based telephone survey conducted by the Survey Research Laboratory (SRL) at the University of South Carolina during January and February of 2001. All contacted participants resided in Sumter County, SC, which the U.S. Office of Budget and Management classifies as a metropolitan county. It lies just 45 miles southeast of Columbia, the state’s capital, and about 90 miles from the Atlantic coast. Sumter County includes 18 cities and towns and one military installation (Shaw Air Force Base), although military personnel were excluded from the study sample. The 2000 U.S. Census indicates that the population of Sumter County was 104,646 people. More information about Sumter County is located at http://www.sciway.net/cnty/sumter.html.
This study was supplementary to a community telephone survey to assess respondents’ perceptions of environmental support for PA such as sidewalks and street lighting (9). At the end of that 18-min survey (including physical activity questions from the 2001 Behavioral Risk Factor Surveillance Survey [BRFSS]), 1200 respondents were invited to participate in this ancillary study of pedometer-determined PA. Those who accepted (and provided verbal consent) were mailed a pedometer data self-collection kit (described below) for self-monitoring purposes (one complete week) before returning the kit in a prestamped, preaddressed envelope. Participants were asked to give their age, height, weight, race, education, and annual household income during the telephone interview. This analysis is based on data from 209 participants (representing 56% of 375 survey respondents who originally agreed to the ancillary study) who returned completed kits. The sample consisted of 76 males (age = 48.4 ± 16.3 yr; BMI = 27.1 ± 5.1 kg·m−2) and 133 females (age = 47.4 ± 17.5 yr; BMI = 26.9 ± 5.7 kg·m−2). Sixty-two percent were white, 40% reported having a high school education or less, and the average household income was nearly $30,000. The 209 participants who returned completed kits were more likely to be white, to have a higher education, and to have a higher household income than the original sample of 1200 (27). Additional participant characteristics and process details are available in another article focused on the acceptability and feasibility of this survey method (27).
Procedures.
The mailed kit included: 1) a pedometer (Yamax model SW-200, Yamax Corporation, Tokyo, Japan); 2) instructions for wearing and using the pedometer; 3) an activity calendar for recording one complete week of day-end steps taken, sports, exercise, and work activities; 4) a one-page informed consent form (originally read aloud during the telephone interview) stamped as approved by the University of South Carolina Institutional Review Board; and 5) a self-addressed and stamped envelope for return mailing. The pedometer brand used herein records within 1% of all steps taken under controlled conditions (2). Each pedometer was checked before mailing for accuracy during walking as previously recommended (29).
Participants were instructed to wear the pedometer attached to their waistband during waking hours for seven consecutive days and to reset the pedometer to zero each morning, recording the time that it was attached. Participants were asked to continue with their typical activities and to remove the pedometer only while bathing, showering, or swimming. At the end of each day, they were to record the following on their activity calendars: 1) the number of steps taken; 2) the time the pedometer was removed; 3) answers to closed-ended questions (e.g., Did you work today? Were you sick or injured? Did you participate in any sport? Did you participate in any exercise?); and 4) the type and duration of any sport or exercise performed (e.g., weight lifting, 25 min).
Data treatment and statistical analysis.
Self-reported height and weight were used to calculate BMI. The number of person-days of data was computed by multiplying the number of persons providing pedometer data by the number of days they did so. Additional descriptive analyses included the distribution of time per day the pedometer was worn (no minimal time worn was required), sickness or injury status, numbers of days with reported sport/exercise involvement, and the three most frequently reported sport/exercise activities. Mean steps per day were computed for the whole week for each participant. Missing days (i.e., unrecorded data) were excluded from the computation.
Because the original telephone sampling rates varied by census tract and race, analysis weights were constructed (using SUDAAN) and used to ensure that statistical analyses were generalizable to the population. The weights had two components (following the protocol of the Behavioral Risk Factor Surveillance System; BRFSS): 1) an adjustment for the number of adults and the number of voice phone lines in the households and 2) an adjustment for the census population by age-race-sex group to account for the differential sampling and response rates. The poststratification factor was constructed within each census tract, although the sample was treated as a single stratum for analysis.
The weighted proportion of individuals in Sumter County averaging ≤5000 d−1 was computed to ascertain those falling below a sedentary lifestyle index suggested by Tudor-Locke et al. (25). In addition, the weighted proportion of individuals in Sumter County averaging ≥9,000 and ≥10,000 d−1 was calculated in order to estimate those who might be considered sufficiently active according to recommended standards put forward by Tudor-Locke et al. (25) and Hatano (8), respectively.
Distributions of mean steps per day were evaluated for normalcy, and means, SD, median, and the 25th and 75th percentiles were computed for the whole sample and for groups defined by: 1) sex; 2) race (white vs nonwhite); 3) age (18–29, 30–45, 46–64, and 65+ yr); 4) household income (< $20,000, $20,000–$44,999, ≥$45,000); 5) education level attained (≤ high school or less vs at least some college); and 6) BMI (normal weight = BMI < 25 kg·m−2, overweight = 25 ≥ BMI < 30 kg·m−2, obese = BMI ≥ 30 kg·m−2) (13). Only seven individuals were underweight (<18.5 kg·m−2), and therefore they were grouped with those classified as normal weight for analyses purposes. Student’s t-tests or the Tukey-Kramer post hoc test was used to evaluate differences among means for stratified groups. Two-way ANOVA models were used to compare mean steps per day taken by the participants in the various demographic categories, while adjusting for mean daily duration of pedometer wear. Nonparametric tests were used to confirm parametric results if there was any question about the normality of variable distributions.
Means, SD, and Student’s t-tests were computed for mean steps per day for four pairs of day types only in participants reporting both: 1) day of week (weekdays, weekend days); 2) occupational activity (self-defined workdays, nonwork-days); 3) sickness and injury status (sick or injured, not sick or injured); and 4) sport/exercise (participation or not). For example, using all the data available, a person mean would be computed for weekday steps per day. Throughout, an alpha level of P < 0.05 was used to evaluate the significance of findings. Analyses were conducted using SAS 8.02 (Cary, NC, 2001).
RESULTS
Data were available for 1396 out of a possible 1463 person-days (209 participants × 7 d) of pedometer wear. Participants averaged 13.4 ± 2.9 h·d−1 of wear. Engagement in sport/exercise was reported on 519 person-days. The three most frequently reported sport/exercise categories were walking (64.9% of person-days with activity), strength training (5.2%), and aerobic dancing/fitness class (4.8%).
Figure 1 shows the distribution in the mean steps per day (in 1000-step increments). In total, 44.0% of the population took <5000 d−1, 19.6% took ≥ 9000 d−1, and 13.9% took ≥ 10,000 d−1. Pedometer data distributions differed by demographic subgroups. Figure 2 (A–D) shows the distribution of mean steps per day for each of the age groups. The distributions for males, those who were between 18 and 45 yr of age, those who were white, those with at least some college education, those in the highest income group, and those of normal weight (classified by BMI) were approximately normal distributions and similar to those shown for age groups 18–29 and 30–45 in Figure 2a and b. In contrast, the distributions for females, those who were >65 yr of age, those who were nonwhite, those with high school or less education, those in the lowest income group, and those who were obese had skewed distributions of mean steps per day, much like that shown in Figure 2d, and were characterized by >40% of days < 3000 d−1.
FIGURE 1: Distribution of mean steps per day (in 1000-step increments).
FIGURE 2: Distribution of mean steps per day (in 1000-step increments) by age group.
Details of mean steps per day are presented in Table 1. Mean steps per day varied significantly between males and females (t = 7.88, P < 0.0001), between whites and non-whites (t = 6.76, P < 0.0001), and across age groups (F = 7.64, P < 0.0001). Although there was no significant difference in mean steps per day among the three younger age groups, all three were significantly higher than the oldest age group. There were also significant differences (F = 5.23, P = 0.006) in mean steps per day across income groups, although post hoc analyses indicated that this difference was significant only between the lowest and highest income groups. There was also a significant difference in steps per day by education group (high school or less = 5063 ± 3728 vs at least some college = 6480 ± 3529 steps·d−1, t = −5.75, P < 0.0001) and by BMI category (F = 6.35, P = 0.002), although once again post hoc analyses revealed a significant difference only between the normal and obese weight groups. After adjusting for mean time per day of pedometer wear, significant differences between demographic groups persisted for sex, race, age, and BMI category but not for income and education (not shown).
TABLE 1: Pedometer-determined physical activity (steps per day).
The remaining descriptive data (based on the number of participants reporting both data pairs examined) are as follows: weekdays were higher than weekend days (6355 ± 3975 vs 5445 ± 3648, t = −2.32, P = 0.02, N = 189); workdays were higher than nonwork days (7583 ± 4173 vs 5117 ± 2987 vs. t = −5.11, P < 0.0001, N = 113); and sport/exercise days were higher than nonsport/exercise days (7043 ± 4083 vs 5205 ± 3365, t = −3.77, P = 0.0002, N = 118). Days of relative health were higher than those with reported illness or injury (5982 ± 3770 vs 3778 ± 2866, t = 2.63, P = 0.01, N = 32). Nonparametric tests run on select comparisons (i.e., if the normality of a distribution was questionable) confirmed the findings of the parametric tests.
DISCUSSION
To date, this study represents the single largest survey of adult pedometer-assessed PA conducted in the United States. The entire sample took approximately 6000 d−1 with significant differences noted by sex, race, age, education, income, and BMI. The sample mean is somewhat less than expected for healthy adults who generally take between 7,000 and 13,000 d−1 (29), suggesting that this Sumter sample is, in large part, relatively sedentary. In fact, 44% of study participants took fewer than 5000 d−1, a cut point that has been proposed as a sedentary lifestyle index (25). Although the objectively monitored data from this study are not directly comparable with self-reported data from previous studies, they do at least suggest higher levels of inactivity: a 1999 report (based on responses to BRFSS questions) indicated that only 29% of the Sumter County residents were inactive (i.e., reported no nonoccupational activity during previous 30 d) (14).
During an initial face-to-face meeting, Sequiera et al. (21) collected similar data from 493 Swiss men and women (aged 25–74) who also wore a pedometer for a week and returned data by mail. Consistent with our own findings, the men in that study took more steps per day than the women and there was an inverse relationship between steps per day and age. The authors did not explore differences related to other demographic variables, however. Regardless of sex though, the Swiss participants took approximately 4000 more per day than these American counterparts (21). Although pedometers may detect some nonstep movements as steps taken (e.g., agitation during motor vehicle travel), this potential error is considered small (10). In 2001 the prevalence of obesity (based on self-reported BMI ≥ 30 kg·m−2) in the United States was 20.9% (12). Comparable data from Switzerland in 2000 indicate that the prevalence of obesity was 9–11% (20).
To date, we have collected data in three independent South Carolina samples using the same pedometer brand. One study recruited a convenience sample of 109 adult men and women from a university campus and the greater metropolitan area of the state’s capital to evaluate physical activity questionnaires (25). Participants in that study took approximately 1300 more per day than the Sumter County sample herein. In contrast, although the individuals who agreed to participate in the Sumter study are a self-selected sample, they were initially drawn from a smaller, more rural community in South Carolina that is also geographically distant from the academic center represented by the university. Another convenience sample of 52 adult men and women was recruited from within and immediately surrounding the same university to evaluate a series of physical activity questionnaires. These subjects took 3600 d−1 more and wore the pedometer for about 1 h·d−1 more than the adults in Sumter (16). This 1-h difference is well within accepted self-monitoring practices (24) and is not likely to explain such a large difference in steps per day. Indeed, our adjustment for time worn only altered conclusions about the significance of differences between income and education subgroups. Without an independent measure of PA, however, it is difficult to conclude that these samples actually have different PA levels. In addition, the recruitment strategies used in these earlier studies are not directly comparable to the present one and conclusions about differences between samples must be drawn cautiously.
Close examination of pedometer data distributions in this study revealed interesting patterns. Those subgroups who were more active (i.e., those who had a higher mean and median steps per day) were fairly normally distributed compared with those subgroups who were less active; these latter distributions were skewed by some individuals with extremely low values of steps per day. Because our purpose to was present the descriptive pedometer-data as truly as possible, we made no attempt in the present analyses to identify or reduce outliers from the data set. However, future researchers might consider “trimming” unusually extreme values from their data (e.g., deleting those values above the 99th percentile or below the 1st percentile) in order to present more conservative estimates of physical activity.
Examination of apparent day-to-day variability of steps per day revealed that weekend days elicited significantly lower steps per day compared with weekdays. A similar cyclical appearance (with a drop on the weekend) of week-long objective data has previously been reported using a pedometer (21) and an accelerometer (11). Self-reported data from another study also showed lower PA levels on weekends than on weekdays (1), notwithstanding a perceived increase in leisure time on the weekends (4). Consistent with these findings, we found that workdays elicited ≈2400 more per day than nonworkdays (i.e., for those who reported both). Despite the technology-driven trend toward more sedentary occupations, it appears that the purposeful-ness or “busy-ness” of working and/or commuting behaviors still cause people to be more physically active on workdays. Pedometers have been effectively used to selectively identify occupations characterized by sitting, standing, and moderate effort; individuals who reported little or no occupational activity had the lowest steps per day (21). Although similar data are not currently available for adults’ commute to work, youth who actively commute to school were found to be more active (as determined by objective motion sensor) overall than those who did not (23). Further, results of an intervention study promoting active commuting to work among adults showed that such active commuting was associated with improved health and fitness outcomes (31).
The three sport/exercise most frequently reported in this study (i.e., walking, strength training, and participating in an aerobic dance/fitness class) are also among the most frequently reported leisure time activities in the American population (6). These results also emphasize the popularity of walking as a primary form of exercise (16). Despite its popularity, however, analyses of BRFSS data show that people do not walk for exercise frequently enough to meet public health recommendations for health-related PA (17). Although we found that 89% of participants reported engaging in some form of sport/exercise (including walking for exercise) on at least 1 d of the monitoring period, sport/exercise days only accounted for 37% of all person-days monitored; the majority of respondents participated fewer than 3 d, if at all. Regardless, steps per day were significantly higher on sport/exercise participation days than on nonparticipation days, underscoring the importance of this strategy (i.e., participating in sport/exercise) in elevating PA levels. It appears, however, that increasing the frequency of sport/exercise participation is key to increasing overall PA levels.
This study builds on a line of research into the utility of pedometers for assessing and motivating PA behaviors. We previously determined mean steps per day taken for each of the same BMI-defined categories of body composition used herein (25). Combining those results with the results from this study (both based on cross-sectional data) suggests that obese individuals take 4600–6000 d−1, overweight individuals take 5800–7000 d−1, and normal-weight individuals take 7000–8300 d−1. Although there is some overlap in steps per day ranges between BMI-defined categories, relatively few individuals in either study exceeded 9000 d−1, a value associated with reduced obesity (25). In another study in which we used both pedometer and accelerometers to measure PA, we ascertained that taking approximately 8000 pedometer-determined steps per day roughly corresponded with accumulating about 30 min of moderate-intensity activity (defined using a uniaxial accelerometer also most sensitive to ambulatory activity) taken over and above usual daily activity of lighter intensity (16). That is, participants in that study were more likely to attain public health guidelines if they took at least 8000 d−1 or more (24). These cumulated pedometer data are beginning to shape candidate PA indices (e.g., benchmarks, cut points, threshold values) that might be useful for surveillance, screening, intervention, and program evaluation. It is becoming apparent also that any steps per day indices may be more useful if represented as a range (e.g., defined by 1 SD) rather than a single cut point. We must emphasize that any recommended indices must emerge from empirical data and need to be confirmed in different populations before being accepted and used widely by researchers and practitioners.
As stated throughout, the cross-sectional nature of these descriptive data limit their value in making causal inferences. Further, and as also stated previously, the final sample was self-selected. However, one would expect any selection bias to result in an overestimate of PA levels because relatively active people might be more interested in using a pedometer to assess their activity level. If so, we would expect that the population from which our sample was drawn takes fewer than 6000 d−1. Against a backdrop of increasing prevalence of overweight and obesity, such low levels of PA (regardless of the precision of measurement) are cause for concern.
In summary, the objective data collected in this study extend previously collected self-reported data and show that this South Carolina adult sample is predominantly sedentary. Pedometer-determined steps per day were related to sex, race, age, education, income, and BMI. Because of their descriptive nature, these data could serve as a foundation for comparison with other populations and potentially for longitudinal analyses of the relationship between physical activity and important health-related outcomes, including weight gain. Intra-individual variability is a feature of pedometer-determined PA data that is important enough to deserve rigorous attention and study.
This project was supported under a cooperative agreement from the Centers for Disease Control and Prevention (CDC) through the Association of Schools of Public Health, Grant Number U36/CCU300430-20. This ancillary study was attached to an environmental correlates of physical activity study that was funded by the CDC (U48/CCU409664; SIP-#4-99). We acknowledge the University of South Carolina Prevention Research Center, Arnold School of Public Health, University of South Carolina for their support of this project.
Conflict of Interest: The results of the present study do not constitute endorsement of any products used by the authors or ACSM.
REFERENCES
1. Baranowski, T., M. Smith, W. O. Thompsons, J. Baranowski, D. Hebert, and C. de Moor. Intraindividual variability and reliability in a 7-day exercise record.
Med. Sci. Sports Exerc. 31:1619–1622, 1999.
2. Bassett, Jr., 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.
3. Bassett, Jr., 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.
4. Brail, R. K., and F. S. Chapin. Activity patterns of urban residents.
Environ, Behav. 5:163–190, 1973.
5. Centers for Disease Control and Prevention. Physical activity trends: United States, 1990–1998.
MMWR 50:166–169, 2001.
6. Crespo, C. J., S. J. Keteyian, G. W. Heath, and C. T. Sempos. Leisure-time physical activity among US adults: results from the Third National Health and Nutrition Examination Survey.
Arch. Intern. Med. 156:93–98, 1996.
7. DuBose, K. D., R. S. Cooper, C. Tudor-Locke, and B. E. Ainsworth. Rounding and digit preference: sources of bias in self-reported physical activity.
Med. Sci. Sports Exerc. 34:S265, 2002.
8. Hatano, Y. Use of the pedometer for promoting daily walking exercise.
ICHPER 29:4–8, 1993.
9. Kirtland, K. K., D. E. Porter, C. L. Addy, et al. Environmental measures of physical activity supports: perception versus reality.
Am. J. Prev. Med. 24:323–331, 2003.
10. Le Masurier, G., and C. Tudor-Locke. Comparison of pedometer and accelerometer accuracy under controlled conditions.
Med. Sci. Sports Exerc. 35:867–871, 2003.
11. Matthews, C. E., B. E. Ainsworth, R. W. Thompson, and D. R. Bassett, Jr. Sources of variance in daily physical activity levels as measured by an accelerometer.
Med. Sci. Sports Exerc. 34:1376–1381, 2002.
12. Mokdad, A. H., E. S. Ford, B. A. Bowman, et al. Prevalence of obesity, diabetes, and obesity-related health risk factors, 2001. JAMA 289:76–79, 2003.
13. National Institutes of Health. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: the evidence report.
Obes. Res. 6:51S–209S, 1998.
14. Powell, K., M. Greaney, Y. Huang, and M. Whitt.
Good Health, It’s Your Move: Physical Activity in South Carolina. Report prepared fo the South Carolina Department of Health and Environmental Control by the Prevention Research Center. Columbia, SC: University of South Carolina, School of Public Health, 1999, p. 32.
15. Pratt, M., C. A. Macera, and C. Blanton. Levels of physical activity and inactivity in children and adults in the United States: Current evidence and research issues.
Med. Sci. Sports Exerc. 31:S526–S533, 1999.
16. Rafferty, A. P., M. J. Reeves, and H. B. McGee. Compliance with physical activity recommendations by walking for exercise: Michigan, 1996 and 1998.
MMWR 49:560–565, 2000.
17. Rafferty, A. P., M. J. Reeves, H. B. McGee, and J. M. Pivarnik. Physical activity patterns among walkers and compliance with public health recommendations.
Med. Sci. Sports Exerc. 34:1255–1261, 2002.
18. Richardson, M. T., A. S. Leon, D. R. Jacobs, B. E. Ainsworth, and R. Serfass. Comprehensive evaluation of the Minnesota Leisure Time Physical Activity Questionnaire.
J. Clin. Epidemiol. 47:271–281, 1994.
19. Sarkin, J. A., J. F. Nichols, J. F. Sallis, and K. J. Calfas. Self-report measures and scoring protocols affect prevalence of estimates of meeting physical activity guidelines.
Med. Sci. Sports Exerc. 32:149–156, 2000.
20. Schutz, Y., and V. Woringer. Obesity in Switzerland: a critical assessment of prevalence in children and adults.
Int. J. Obes. Relat. Metab. Disord. 2( Suppl. 2): S3–S11, 2002.
21. Sequeira, M. M., M. Rickenbach, V. Wietlisbach, B. Tullen, and Y. Schutz. Physical activity assessment using a pedometer and its comparison with a questionnaire in a large population survey.
Am. J. Epidemiol. 142:989–999, 1995.
22. Trost, S. G., R. R. Pate, P. S. Freedson, J. F. Sallis, and W. C. Taylor. Using objective physical activity measures with youth: How many days of monitoring are needed?
Med. Sci. Sports Exerc. 32:426–431, 2000.
23. Tudor-Locke, C., B. E. Ainsworth, L. S. Adair, and B. M. Popkin. Objective physical activity of Filipino youth stratified for commuting mode to school.
Med. Sci. Sports Exerc. 35:465–471, 2003.
24. Tudor-Locke, C., B. E. Ainsworth, R. W. Thompson, and C. E. Matthews. Comparison of pedometer and accelerometer measures of free-living physical activity.
Med. Sci. Sports Exerc. 34:2045–2051, 2002.
25. Tudor-Locke, C., B. E. Ainsworth, M. C. Whitt, R. Thompson, C. L. Addy, and D. A. Jones. The relationship between pedometer-determined ambulatory activity and body composition variables.
Int. J. Obes. 25:1571–1578, 2001.
26. Tudor-Locke, C., G. R. Jones, A. M. Myers, D. H. Paterson, and N. A. Ecclestone. Contribution of structured exercise class participation and informal walking for exercise to daily physical activity in community-dwelling older adults.
Res. Q. Exerc. Sport 73:350–356, 2002.
27. Tudor-Locke, C., K. A. Lind, J. P. Reis, B. E. Ainsworth, and C. A. Macera. Using a pedometer in a self-monitoring survey to assess physical activity.
Field Methods (in press).
28. Tudor-Locke, C., and A. M. Myers. Challenges and opportunities for measuring physical activity in sedentary adults.
Sports Med. 31:91–100, 2001.
29. Tudor-Locke, C., and A. M. Myers. Methodological considerations for researchers and practitioners using pedometers to measure physical (ambulatory) activity.
Res. Q. Exerc. Sport. 72:1–12, 2001.
30. Tudor-Locke, C., J. E. Williams, J. P. Reis, and D. Pluto. Utility of pedometers for assessing physical activity: convergent validity.
Sports Med. 32:795–808, 2002.
31. Vuori, I. M., P. Oja, and O. Paronen. Physically active commuting to work: testing its potential for exercise promotion.
Med. Sci. Sports Exerc. 26:844–850, 1994.
