Although a large body of epidemiological evidence has established the health benefits of regular physical activity (PA) (36), this evidence is based almost entirely on self-reported PA behavior. In general, the self-reported PA measures used in these studies have documented validity and reliability (5,21,37) and tend to rank people relatively accurately, especially at the extremes of the activity spectrum (regularly active vs inactive) (20). However, self-reported PA suffers from various sources of measurement error, including recall error, social desirability, cognitive challenges, and incomplete ascertainment of exposure (19,30,37,38), all of which reduce the precision of the PA estimate and increase the likelihood of misclassification. Recall error, which tends to increase as the period of recall and the amount of detail requested (e.g., frequency and duration) increase (12,29), may be particularly problematic for correctly ranking individuals in the middle of the activity spectrum. This middle range of the spectrum is often characterized by more variable and sporadic PA behavior (20) but may be where accuracy matters most in terms of establishing dose–response relations.
To address the problem of recall error, investigators have turned to daily PA diaries or logs as a more direct measure of PA (5). Typically, these diaries ask respondents to record either all of their activities, or selected activities, periodically throughout the day and to maintain the diaries over a period of days, often a week, to account for intraindividual variability in PA behavior (8). Because a PA diary theoretically provides a direct and detailed measure of PA, it has frequently been used as a validation criterion for other types of PA questionnaires (4,5).
However, PA diaries, which traditionally have been paper-and-pen–based records, have their own sources of measurement error. The participant burden is high and requires sustained cooperation with a relatively tedious task, and the amount of detail provided may vary from individual to individual. In addition, the paper diary does not allow for any monitoring of how frequently participants actually record their activities; if some individuals wait until the end of the day, or even the end of the entire recording period, to write down their activities, the recall error may still be considerable (15). Finally, processing the detailed data is burdensome for the investigator. Mobile communications technology provides the opportunity to reduce or eliminate many of these sources of error. As part of the On the Move study, a project designed to identify and quantify the measurement error in self-reported PA instruments, we developed and evaluated a cell phone–based PA diary. The purpose of this study was twofold: 1) to compare the cell phone–based diary with a traditional paper diary in terms of estimates of PA, data quality, and user preferences; and 2) to examine the reliability (test–retest repeatability) of the cell phone diary and its validity (intermethod reliability) against accelerometers, two PA questionnaires, physical fitness, and body fat.
Overview of On the Move
On the Move, funded by the National Cancer Institute under the programmatic initiative to improve the assessment of diet and PA (PAR-03-009), consisted of a pilot study and a validation study. The pilot study, conducted in the summer of 2006, included the development of the cell phone application and a comparison of the cell phone PA diary with a paper-and-pen PA diary. For the comparison of the diaries, participants completed the diaries on 4 d·wk−1 (Friday to Monday) for three consecutive weeks; half of the sample were randomly selected to complete the cell phone diaries in weeks 1 and 3 and the paper diary in week 2, whereas the other half did the reverse (i.e., paper diaries in weeks 1 and 3, cell phone diary in week 2). All participants completed a user survey at the end of the third week.
In the validation study, conducted from March 2007 to April 2009, participants were seen for an in-person visit that included completion of two PA questionnaires, a submaximal graded exercise treadmill test, and body size measurements. They were also given an accelerometer and instruction in how to use the cell phone PA diary and asked to record their activities in the diary and wear the accelerometer for the next seven d. Approximately 6 months later, participants repeated the PA questionnaires and another 7 d of cell phone PA recording and accelerometer monitoring.
All protocols for both the pilot study and the validation study were reviewed and approved by the Northern California Kaiser Permanente Institutional Review Board, and all participants provided written informed consent.
The samples for both the pilot study and the validation study were drawn from members of Northern California Kaiser Permanente who were between 45 and 65 yr and living within geographical proximity to the research clinic. After a recruitment letter, potential participants were telephoned and screened for interest and eligibility. To be eligible, participants had to be English speaking and free of any cardiovascular, respiratory, or musculoskeletal contraindications for participation in PA. Having a cell phone was not an eligibility criterion because study cell phones were supplied to all participants. Of the 249 individuals who were actually screened for the pilot study (55.3% of the 450 who were sent recruitment letters), 38 were ineligible, 160 refused, and 3 enrolled in but failed to complete the pilot protocol, leaving 48 participants, 23 men and 25 women, on whom the analysis of the pilot data is based. For the validation study, 16.5% (n = 444) of those sent a recruitment mailing could not be contacted, and 43.7% (n = 1176) refused further screening; of the 1073 individuals who were screened, 100 (9.3%) refused further participation and 350 (32.6%) were ineligible. Of the 623 individuals (327 women and 296 men) who enrolled, 496 provided at least 1 wk of cell phone recordings and 345 provided 2 wk of recordings.
Cell phone PA diary
The cell phone PA diary, used in both the pilot and validation studies, was developed in collaboration with BeWell Mobile Technology (www.bewellmobile.com), a company based in San Francisco, CA, as an application program that operates on a mobile phone through the soft keys and arrow keys. The basic structure is a database of 136 activities, along with the associated activity codes and MET values, taken from the Compendium of Physical Activities (3) and organized into 15 different domains, such as quiet activities (e.g., sleeping, reading, listening to music), work-related activities, walking, sports and exercise, and household activities. Each domain is introduced by a stem question that asks, “Since [time of last entry] did you do any [domain-specific] activities for at least 10 min at a time? A “yes” response brings up the menu and submenu of specific activities in that domain (e.g., team sports → basketball, soccer). For each activity that is selected, the participant also enters the duration and the intensity, according to the Borg RPE (7). Respondents can select as many specific activities in a single domain as necessary before proceeding to the next domain. After responses are provided for all 15 domains, the program tells the respondent how many activities were selected and the total amount of time accounted for, and allows respondents to go back and add, edit, or delete any entry. The respondent is then prompted to upload the data. If, for any reason, a wireless transmission cannot be completed at that time, the entries are saved and must be transmitted before a new set of entries can be made. Once participants became familiar with the program, they only needed about 5 min to move through the program and enter and transmit their data.
Participants in On the Move were trained in the use of the PA diaries by study staff and asked to record their activities on the phone three times a day: at around 9:00 a.m., 6:00 p.m., and before bed. If expected transmissions were not received, the program sent an automatic text message reminding the participant to record their activities
Traditional paper-and-pen PA diary
The paper PA diary, used in the pilot study, was constructed as a calendar with a separate page for each day of recording and separate rows for each hour of the day, with columns for domain, activity, and duration. As many as six different activities could be entered into each hour. At the front of the booklet were written instructions for the participants, along with a list of activity domains, in the same order as the domains appeared on the cell phone diary. Instructions for frequency and timing of recording were identical with those of the cell phone diary. Once diaries were returned by the participants, trained study staff coded each entry with the appropriate activity code and MET value, based on the Compendium (3).
Data cleaning of diary data
Explicit data cleaning rules were applied to both the cell phone and paper diary data. A valid day of recording was defined as one that had at least one entry and accounted for between 864 and 1440 min (60%–100% of 1440, the total number of minutes per day), based on the distribution in the pilot study. To be included in the analysis, participants had to have four or more valid days of diary data, in at least one of the recording periods in the pilot study, and in at least one of the 7-d recording periods in the validation study. In the pilot study, 39 participants provided sufficient days of cell phone diary data, 38 provided sufficient days of paper diary data, and 33 provided sufficient days of both types of diary data. The primary analyses from the pilot study are based on these 33 participants. In the validation study, 496 participants provided the sufficient number of days of valid cell phone data in any 1 wk of recording.
Other PA measures used in validation study
The accelerometers worn by the On the Move participants were from ActiGraph, Inc. (model 7164, Pensacola, FL). Participants were instructed to wear the accelerometers around their waists, attached to an elastic band under their clothing, continuously for 7 d except when sleeping, swimming, or bathing. They were also asked to maintain a daily log, indicating when they put the accelerometer on or took it off. All accelerometers were calibrated according to the manufacturer’s procedures both before and after participant use.
Data cleaning rules were applied to the accelerometer data that set negative and spurious values to missing and defined 60 or more continuous minutes of zero counts as periods of nonwear, which included sleeping (22). A valid day of accelerometer monitoring required at least 12 h of wear, and inclusion in the analysis required at least four valid days of monitoring in any one monitoring period. Ten percent of participants (n = 62) failed to meet this requirement, leaving 561 for inclusion in the validation study analyses. Of those 561, 484 also provided sufficient days of valid cell phone data.
The self-administered PA questionnaires used in On the Move were those used in two ongoing epidemiological cohort studies, Life After Cancer Epidemiology (LACE), which is a study of behavioral and other risk factors for recurrence and mortality in breast cancer survivors, and the California Men’s Health Study (CMHS), which is a prospective investigation of risk factors for prostate cancer and other men’s health outcomes. The CMHS PA questionnaire, based on the CARDIA PA questionnaire (18), asks about participation in nine types of vigorous intensity activities and eight types of moderate intensity, mostly recreational activities during the past 3 months. The version used here inquires specifically about frequency and duration, thus allowing for construction of summary variables in MET-minutes per week. The LACE PA questionnaire, based on the Arizona Activity Frequency Questionnaire (32), asks about participation during the last 12 months in 49 activities organized by domain, including transportation, household/care giving activities, and sedentary behaviors. Both questionnaires in their original versions demonstrated reasonable reliability and validity (17,32). Questionnaires were completed by the participants and edited by staff at in-person visits.
Indirect validation criteria
Physical fitness was assessed in the validation study by a submaximal graded exercise treadmill test (GXT). The GXT protocol was the same protocol used in the SAFE study (17) and consisted of fourteen 2-min stages, beginning at 3.0 mph, 0% grade, and increasing by approximately 1 MET or 2.5% grade with each stage. The final stage was 5 mph at 20% grade. The test was preceded by a warm-up period at 2.0 mph, 0% grade to allow the participant to become comfortable with walking on the treadmill and ended with a 2-min cool down, also at 2.0 mph, 0% grade. HR and ECG were monitored continuously throughout the protocol and recorded, along with blood pressure and RPE, at the end of each stage. The test was terminated once a participant reached 85% of predicted maximal HR, and the duration of the test was used as a measure of aerobic capacity.
Body size measurements included height, weight, and waist circumference obtained according to standardized protocols by trained staff and body composition (% body fat) assessed by bioelectrical impedance (RJL Systems, Clinton Township, MI), using standard criteria for ensuring constant hydration status. Participants were asked to refrain from vigorous exercise for 24 h before the body composition measurement and not to eat for at least 2 h. Body mass index was calculated as weight in kilograms divided by height in meters squared.
User satisfaction survey for pilot study
At the end of the 3-wk pilot study protocol, participants completed a survey that asked for their opinions about the two different types of PA diaries. They were asked specifically about the convenience, difficulty, and enjoyment of each type of diary as well as whether they preferred the cell phone or the paper version.
Creation of summary variables
After application of the data cleaning rules, the PA data in the pilot study from the cell phone and paper diaries were summarized into two variables: 1) total activity (MET·min·d−1), derived by multiplying the MET value of each activity by the duration, summing over all activities over all valid days of recording and dividing by the number of valid days; and 2) minutes per day of moderate–vigorous PA (MVPA), derived by summing the duration of all activities of ≥3 METs and dividing by the number of valid days.
In the validation study, in which the validity of the reporting of activities of varying intensity was of interest, PA data from the cell phone diary were expressed as minutes per day of sedentary behavior (total number of minutes of activities between 1 and 1.4 METs, excluding sleeping, divided by number of valid days), minutes per day of light activity (total number of minutes of activities between 1.5 and 2.9 METs, divided by number of valid days), minutes per week of moderate activity (total number of minutes of activities between 3.0 and 5.9 METs, weighted by number of valid days), and minutes per week of vigorous activity (total number of activities of ≥6 METs, weighted by number of valid days). The same PA variables were derived from the responses to the LACE questionnaire by summing the minutes per day or per week of all activities in the relevant ranges of MET values. Data from the CMHS questionnaire were treated similarly but only provided measures of time per week spent in moderate and vigorous activity.
Similar variables were derived from the accelerometer data based on number of counts per minute rather than MET value of the activities: sedentary behavior = counts per minute between 0 and 100 (23), light activity = counts >100 and <1952, moderate activity = counts between 1952 and 5724, and vigorous activity = ≥5725, according to the cut points of Freedson et al. (14).
In the pilot study, data quality variables were also defined for the cell phone and paper diaries that included minutes per day accounted for, number of different activities participated in per day, and percent valid days, defined as the ratio of number of valid days of recording to number of expected days (which, by design, was equal to either 7 d of paper diary and 14 d of cell phone diary or vice versa).
In both the pilot study and the validation study, the estimates of PA obtained from the different measures (cell phone diary, paper diary, questionnaires, and accelerometer) were described by means and SDs and medians and interquartile ranges. Paired t-tests were used to determine whether the mean within-person differences in the PA estimates from the cell phone diary and those from either the paper diary (pilot study only) or the accelerometer and questionnaires (validation study only) were significantly different from zero, and Spearman correlation coefficients were used to describe the degree of association between the PA estimates from the cell phone diary and those from the other measures. Spearman correlations were also used to examine the magnitude of the association between the cell phone–based PA variables and the indirect validation criteria (treadmill duration, BMI, and percent body fat). Intraclass correlations were used as a measure of reliability between the 2 wk of cell phone recordings. For the pilot study, responses to the user survey were compared using the Stuart–Maxwell test.
The participants in the pilot study ranged in age from 45 to 65 yr, with a mean ± age of 55.6 ± 5.8 yr; the age range for the validation study participants was the same, and the mean ± SD was 56.9 ± 5.7 yr (Table 1). In the pilot study, about a quarter of the sample was African American, and about 10% were other racial/ethnic minority groups. In the validation study, the proportion of African Americans and those of other race/ethnicity were approximately equal (about 20% each). There were two more women than men in the pilot study (25 vs 23), and in the validation study, women constituted 52.5% of the sample. Slightly less than a third of both samples had not completed a 4-yr undergraduate education. In the validation study, men were more likely to provide valid accelerometer data (92.6% vs 87.5%, P = 0.03), and those who provided both valid cell phone diary and accelerometer data tended to have more education than those who did not (71.7% college graduates or more vs 56.8%, P < 0.001) and, for men, to have a lower BMI (mean ± SD = 28.1 ± 5.4 vs 30.4 ± 5.9, P = 0.03). There were, however, no other differences in terms of age or activity level as measured by the questionnaires.
As shown in Table 2, the group means in the pilot study for total PA and minutes per day of MVPA were quite similar when derived from the cell phone or the paper diaries with no significant within-person mean differences. In general, individuals were ranked similarly by the two types of diaries, as evidenced by correlations of 0.61 for the estimates of total activity and 0.79 for the estimates of minutes per day of MVPA (Table 2).
In terms of data quality, there was no significant within-person mean difference in the number of minutes per day accounted for between the cell phone diary and the paper diary. However, participants tended to record more activities each day and to provide more valid days of recordings, relative to expected days, on the paper diary than on the cell phone diary. With the exception of the number of different activities recorded per day (ρ = 0.52), the data quality variables were not highly correlated (0.16 for minutes per day of recording and 0.03% valid days).
According to self-report on the user survey, 58.3% of the participants made entries in the paper diary three or more times a day, as requested, compared with 76.6% on the cell phone diary (Table 3). The electronic date/time stamp automatically attached to each cell phone entry indicated that the actual mean number of entries per day on the cell phone was 2.9 (SD = 0.87), and 75% of the participants completed cell phone entries three or more times a day. In addition, 30% of activities recorded in the paper diary were missing either duration and/or intensity, whereas missing data were not allowed in the cell phone diary.
Other responses to the user survey are also summarized in Table 3. The proportion of participants who found the cell phone diary either very or somewhat convenient to use was almost double the proportion who reported that the paper diary was convenient (66.7% vs 39.6%), whereas 45.8% found the paper diary very or somewhat inconvenient, compared with only 27.1% who said the same for the cell phone diary (P = 0.03) About a third (31.3%) said the cell phone diary was “a lot” of fun to use, while almost nobody (4.2%) felt that way about the paper diary. On the other hand, more participants (25.0%) reported difficulty using the cell phone diary compared with those who reported difficulty with the paper diary (16.7%), but this was not statistically different (P = 0.31). Overall, 59.6% preferred the cell phone over the paper diary.
In the validation study, for both men and women, the group medians for sedentary behavior and light PA estimated from the cell phone were slightly higher than the estimates from the accelerometer, whereas the median for vigorous PA from the cell phone was considerably higher than from the accelerometer (Table 4). For men, the estimates of moderate activity from the cell phone were also considerably higher than those from the accelerometer, but for women, the median for moderate PA was relatively similar (299 and 255 min·wk−1, respectively). The PA estimates from the cell phone were also higher than those from the questionnaires, except for vigorous PA. The mean within-person differences for these estimates were all statistically different from zero and confirmed the patterns observed for the group means. The one exception, for both men and women, was for vigorous PA, where the differences in estimates between the cell phone and each of the questionnaires were roughly equal to zero.
As shown in Table 5, the estimates of reliability provided by the intraclass correlations between two different weeks of the cell phone diary, separated by approximately 6 months, ranged from 0.52 for sedentary behavior in women to 0.64 for vigorous activity in women. In terms of validity, the results, summarized in Table 6, show little or no association between the minutes per day of sedentary behavior and light PA as measured by the cell phone diary and by the accelerometer, although the correlations between minutes per week of moderate PA and vigorous PA were of moderate magnitude (0.27–0.35). The correlations between the estimates of PA from the cell phone diary and from the questionnaires tended to be more substantial, particularly for moderate and vigorous PA (0.39–0.50 for women, 0.25–0.59 for men). Finally, the correlations between the cell phone diary and body mass index, percent body fat, and treadmill duration also tended to be in the expected direction and of moderate magnitude, although for men, the correlations of moderate PA with treadmill duration and body mass index were minimal (Table 6).
In this evaluation of a cell phone–based PA diary, findings suggest that the cell phone diary is a reasonably reliable and valid measure of self-reported PA. More specifically, the results suggest that: (a) the absolute estimates of PA level obtained from a cell phone diary are comparable to those obtained from a paper diary; (b) the quality of data from a cell phone diary is superior to that from a paper diary; (c) users find the cell phone diary convenient and tend to prefer it to a paper diary; (d) the absolute estimates of PA obtained from a cell phone diary tend to be higher than those obtained from either an accelerometer or activity frequency questionnaires with the exception of vigorous activity; (e) individuals are ranked relatively similarly in terms of moderate and vigorous PA by the cell phone diary and both the accelerometer and PA questionnaires but are ranked very differently by the cell phone and accelerometer in terms of sedentary behavior; and (f) the measures of PA from the cell phone diary are related in the expected directions with measures of body fat and physical fitness.
Although the cell phone and paper diaries yielded generally similar estimates of PA, the significant differences in terms of compliance and data quality are notable. Although prior literature has documented the general tendency of participants to greatly overestimate the frequency of diary entries (9,34), the participants in this study, by their own admission, used the paper diary less often than the cell phone diary. The fact that the proportion of participants self-reporting entries three times a day on the cell phone agreed so closely with the proportion captured by the date/time stamp (76.6% vs 75%) suggests a much greater adherence to study protocol with the cell phone diary, reflecting, perhaps, the difference in convenience reported by the participants. In addition, although a slightly higher proportion of valid to expected days of recording was observed on the paper diary, this may be largely the result of the inability to define a valid day accurately for the paper diary. Finally, the cell phone diary did not allow for missing values and the entries on the cell phone diary required no coding or data management because data were transmitted wirelessly to a study database that included preexisting activity codes and MET values for each selected activity. This resulted in substantial savings in terms of time and cost.
The benefits of the cell phone diary over a paper diary noted in this study are similar to those reported in the literature (31,33). The issue of compliance, in particular, has received considerable attention because the advantages of real-time diary data collection over recalled questionnaire data are lost if the diary is completed in retrospect after the activities have occurred (15). In several studies among chronic pain patients, Broderick et al. (9) and Stone et al. (33) found a very large difference in compliance between an instrumented paper diary that monitors entries without the respondent’s knowledge (90% by self-report vs 11% by electronic monitoring) and an electronic diary (94%). This is consistent with the observation in the present study and suggests that actual participant compliance with real-time data collection is substantially improved by the use of electronic technology that is capable of constraining responses within particular date and time windows and that participants know will automatically date and time stamp their entries.
The finding in the validation study that the estimates of moderate and vigorous PA from the cell phone diary were higher than those from the accelerometer is also consistent with reports from other studies that compare self-reported PA with objectively measured activity (6,10,11,27,28). Most notably, in NHANES, the prevalence of individuals meeting recommendations for health-related PA based on self-report was 51% compared with less than 5% based on accelerometry (35). This overreporting may be attributable to social desirability bias (2), although it may also be due, in part, to the cut points used to translate activity counts generated by the accelerometer into PA intensity. In addition, it may also reflect the inability of the accelerometer to measure a wide range of distinct types and patterns of physical activities (27). Our observation that the cell phone diary yielded relatively similar mean estimates of time spent in sedentary behavior and light activity as the accelerometer has not been widely noted. This may indicate that the large amount of time that many individuals spend sitting on their jobs, as well as the time spent in other distinct sedentary behaviors, such as television viewing, is relatively accurately reported. On the other hand, the low, inverse correlations between the cell phone diary and the accelerometer for sedentary and light activity suggest that the accelerometer cut points that define low-intensity behaviors may vary from individual to individual or, more generally, that the self-reports and the accelerometer counts are not measuring the same aspects of these behaviors.
The relative lack of concordance between the level of PA reported on questionnaires and that recorded objectively by accelerometers or other device-based methods has led to the assumption that PA frequency questionnaires yield overestimates, especially in particular population subgroups (1,16). In this study, the fact that the cell phone diary consistently resulted in a higher estimate of MVPA than the questionnaires may indicate that a comprehensive assessment of activity, which allows respondents to select a wide range of specific activities, results in an even greater degree of overestimation than when fewer activities are assessed, as is typical of questionnaires. However, an alternative explanation may be that the cell phone diary was a more accurate reflection of PA level because participation was recorded in more or less real time, and the questionnaires may simply not have accounted for as many activities as those in which the participants actually engaged. This was undoubtedly the explanation for why the amount of time spent in light activities was so much greater according to the cell phone diary than to the questionnaires; the LACE questionnaire asked about only a few light activities and the CMHS asked about none.
The intraclass correlations between the 2 wk of cell phone diaries (a measure of reliability) are comparable to the repeatability of other self-reported PA instruments (13,25,27), particularly given the 5- to 6-month interval between the 2 wk when real change in activity may have occurred. Further, the correlations between activity level estimated from the cell phone diary and fitness and body size observed in this study are of a magnitude that are consistent with studies of other PA self-report instruments (17,24), suggesting that the cell phone diary is a relatively valid measure of PA. However, the magnitude of those correlations is low to moderate, leaving much of the variance in fitness and fatness unaccounted for. Although genetic influences explain much of the interindividual variability in fitness and fatness (26), PA behavior represents one of the primary environmental influences. However, because not all physical activities contribute equally to aerobic fitness, the relatively modest contribution of PA to fitness with all PA instruments may not be a reflection of deficiencies in the measurement of PA behavior as much as it is a physiological reality.
The findings reported here need to be considered within the context of the limitations of the study. Given the rapid change in cell phone technology and software applications, the diary used in this study represents only a first generation design. In addition, the diary still relied to some extent on respondent recall because respondents were instructed to enter data only three times a day, rather than simultaneously with the activity itself. Finally, the study sample included only a limited age range of middle-age adults.
On the other hand, this study had numerous strengths. The diary was evaluated against both objective measures of PA and other self-report questionnaires, it included the full range of daily physical activities from sedentary behaviors to vigorous recreational activities, and the correlations between the questionnaires and the cell phone diary were in the expected direction and generally of higher magnitude than seen in many validation studies. In addition, the study sample included both men and women and diverse race/ethnicities, and compliance with the study protocol was high, with more than two-thirds of the sample providing high-quality data suitable for analysis.
In conclusion, this study suggests that a cell phone diary provides valid measures of PA, is well accepted and convenient for respondents, and achieves a high level of compliance with the recording protocol. Given the rapid pace of development in mobile technology, the feasibility of using PA diaries in epidemiological research, rather than questionnaires, is likely to grow rapidly for the next few years. The result could be a significant decrease in the recall error of self-reported PA.
This study was funded in part by the National Cancer Institute (R01-CA103974).
None of the authors has any conflict to disclose.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Adamo KB, Prince SA, Tricco AC, Connor-Gorber S, Tremblay M. A comparison of indirect versus direct measures for assessing physical activity in the pediatric population: a systematic review. Int J Pediatr Obes. 2009; 4: 2–27.
2. Adams SA, Matthews CE, Ebbeling CB, et al.. The effect of social desirability and social approval on self-reports of physical activity. Am J Epidemiol. 2005; 161: 389–98.
3. 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.
4. Ainsworth BE, Sternfeld B, Richardson MT, Jackson K. Evaluation of the Kaiser Physical Activity Survey in women. Med Sci Sports Exerc. 2000; 32 (7): 1327–38.
5. Ainsworth B, Levy S. Assessment of health-enhancing physical activity. Methodological Issues. In: Oja P, Borms J, editors. Health Enhancing Physical Activity. Oxford (UK): Meyer & Meyer Sport; 2004. p. 239–70.
6. Banda JA, Hutto B, Feeney A, et al.. Comparing physical activity measures in a diverse group of midlife and older adults. Med Sci Sports Exerc. 2010; 42 (12): 2251–7.
7. Borg GA. Psychophysical bases of perceived exertion. Med Sci Sports Exerc. 1982; 14 (5): 377–81.
8. Bouchard C, Tremblay A, Leblanc C, Lortie G, Savard R, Theriault G. A method to assess energy expenditure in children and adults. Am J Clin Nutr. 1983; 37: 461–7.
9. Broderick JE, Schwartz JE, Shiffman S, Hufford MR, Stone AA. Signaling does not adequately improve diary compliance. Ann Behav Med. 2003; 26: 139–48.
10. Craig CL, Marshall AL, Sjostrom M, et al.. International physical activity questionnaire: 12-country reliability
. Med Sci Sports Exerc. 2003; 35 (8): 1381–95.
11. Cust AE, Smith BJ, Chau J, et al.. Validity
and repeatability of the EPIC physical activity questionnaire: a validation study using accelerometers as an objective measure. Int J Behav Nutr Phys Act. 2008; 5: 33.
12. Durante R, Ainsworth BE. The recall of physical activity: using a cognitive model of the question-answering process. Med Sci Sports Exerc. 1996; 28 (10): 1282–91.
13. Fjeldsoe BS, Marshall AL, Miller YD. Measurement properties of the Australian Women’s Activity Survey. Med Sci Sports Exerc. 2009; 41 (5): 1020–33.
14. Freedson PS, Melanson E, Sirard J. Calibration of the Computer Science and Applications, Inc. accelerometer. Med Sci Sports Exerc. 1998; 30 (5): 777–81.
15. Hufford RM. Special methodological challenges and opportunities in ecological momentary assessment
. In: Stone AA, Shiffman S, Atienza AA, Nebeling L, editors. The Science of Real-time Data Capture. Oxford (NY): Oxford University Press, Inc.; 2007. p. 54–75.
16. Irwin ML, Ainsworth BE, Conway JM. Estimation of energy expenditure from physical activity measures: determinants of accuracy. Obes Res. 2002; 9: 517–23.
17. Jacobs DR Jr, Ainsworth BE, Hartman TJ, Leon AS. A simultaneous evaluation of 10 commonly used physical activity questionnaires. Med Sci Sports Exerc. 1993; 25 (1): 81–91.
18. Jacobs DR Jr, Hahn LP, Haskell WL, Pirie P, Sidney S. Validity
of short Physical Activity History: CARDIA and the Minnesota Heart Health Program. J Cardiopulm Rehabil. 1989; 9: 448–59.
19. Lamonte MJ, Ainsworth BE. Quantifying energy expenditure and physical activity in the context of dose response. Med Sci Sports Exerc. 2001; 33 (6 suppl): S370–8; discussion S370–8; S419–20.
20. LaPorte RE, Adams LL, Savage DD, Brenes G, Dearwater S, Cook T. The spectrum of physical activity, cardiovascular disease and health: an epidemiologic perspective. Am J Epidemiol. 1984; 120: 507–17.
21. LaPorte RE, Montoye HJ, Caspersen CJ. Assessment of physical activity in epidemiologic research: problems and prospects. Public Health Rep. 1985; 100: 131–46.
22. Masse LC, Fuemmeler BF, Anderson CB, et al.. Accelerometer data reduction: a comparison of four reduction algorithms on select outcome variables. Med Sci Sports Exerc. 2005; 37 (11 suppl): S544–54.
23. Matthews CE, Chen KY, Freedson PS, et al.. Amount of time spent in sedentary behaviors in the United States, 2003–2004. Am J Epidemiol. 2008; 167: 875–81.
24. Matthews CE, Shu XO, Yang G, et al.. Reproducibility and validity
of the Shanghai Women’s Health Study physical activity questionnaire. Am J Epidemiol. 2003; 158: 1114–22.
25. Meyer AM, Evenson KR, Morimoto L, Siscovick D, White E. Test–retest reliability
of the Women’s Health Initiative physical activity questionnaire. Med Sci Sports Exerc. 2009; 41 (3): 530–8.
26. Perusse L, Gagnon J, Province MA, et al.. Familial aggregation of submaximal aerobic performance in the HERITAGE Family study. Med Sci Sports Exerc. 2001; 33 (4): 597–604.
27. Peters TM, Shu XO, Moore SC, et al.. Validity
of a physical activity questionnaire in Shanghai. Med Sci Sports Exerc. 2010; 42 (12): 2222–30.
28. Pettee GK, McClain JJ, Lee CD, et al.. Evaluation of physical activity measures used in middle-aged women. Med Sci Sports Exerc. 2009; 41 (7): 1403–12.
29. Sallis JF, Saelens BE. Assessment of physical activity by self-report: status, limitations, and future directions. Res Q Exerc Sport. 2000; 71: S1–14.
30. Shephard RJ. Limits to the measurement of habitual physical activity by questionnaires. Br J Sports Med. 2003; 37: 197–206.
31. Shiffman S. Real-time self-report of momentary states in the natural environment: computerized ecological momentary assessment
. In: Stone AA, Turkkan JS, Bachrach CA, Jobe JE, Kurtzman HS, Cain VS, editors. The Science of Self-report: Implications for Research and Practice. Mahwah (NJ): Lawrence Erlbaum Associates Publishers; 2000. p. 277–96.
32. Staten LK, Taren DL, Howell WH, et al.. Validation of the Arizona Activity Frequency Questionnaire using doubly labeled water. Med Sci Sports Exerc. 2001; 33 (11): 1959–67.
33. Stone AA, Broderick JE, Schwartz JE, Shiffman S, Litcher-Kelly L, Calvanese P. Intensive momentary reporting of pain with an electronic diary: reactivity, compliance, and patient satisfaction. Pain. 2003; 104: 343–51.
34. Stone AA, Shiffman S, Schwartz JE, Broderick JE, Hufford MR. Patient compliance with paper and electronic diaries. Control Clin Trials. 2003; 24: 182–99.
35. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc. 2008; 40 (1): 181–88.
36. US Department of Health and Human Services. Physical Activity Guidelines Advisory Committee. Part E. Washington (DC): US Department of Health and Human Services; 2008. p. 1–35.
37. Wareham NJ, Rennie KL. The assessment of physical activity in individuals and populations: why try to be more precise about how physical activity is assessed? Int J Obes Relat Metab Disord. 1998; 22 (2 suppl): S30–8.
38. Westerterp KR. Assessment of physical activity: a critical appraisal. Eur J Appl Physiol. 2009; 105: 823–8.