Journal Logo

Epidemiology

Validation of interviewer- and self- administered physical activity checklists for fifth grade students

SALLIS, JAMES F.; STRIKMILLER, PATRICIA K.; HARSHA, DAVID W.; FELDMAN, HENRY A.; EHLINGER, SALLY; STONE, ELAINE J.; WILLISTON, JEAN; WOODS, SHERRY

Author Information
Medicine & Science in Sports & Exercise: July 1996 - Volume 28 - Issue 7 - p 840-851
  • Free

Abstract

Children's physical activity can be measured by self-report(11), direct observation (9), electronic motion sensors and heart rate (7), and doubly labeled water (4). Largely due to considerations of expense and convenience, self-reports are most often used in large field studies. Although the validity of children's self-reports is not well researched (11), some tentative conclusions may be drawn from the available data.

There appears to be a rough inverse association between ease of administration of the measure and its performance. Two studies of activity diaries yielded encouraging results, but compliance issues were not addressed, and diaries are unlikely to be useful with preadolescents(11). The most evidence for validity has accumulated for structured interviews. Although their reliabilities are generally acceptable, few have been validated. Parent or teacher reports of child physical activity have not been supported in the literature, possibly because adults do not observe much of children's physical activity (11). Based on the existing data, the two most promising types of self-report instruments appear to be interviewer-administered and self-administered.

Validity of existing measures appears to be highly related to age of the child. Preadolescent children are expected to have difficulty with the complex cognitive task of recalling physical activity (3). No study was located that showed a significant correlation between self-reported activity and a criterion variable in children below the age of nine(11). One study with children from three different grades found a standardized interview was more reliable and valid with eighth and eleventh graders than with fifth graders (14). The validity correlations for fourth and fifth grade students are typically in the 0.3-0.4 range (14,15), indicating children of this age do not provide highly accurate estimates of physical activity.

The time span of the recall is also related to its validity. Recalls vary from 1 d to 3 months, and many elicit reports of “usual” activity(11). Higher reliabilities were found when the delay between recalls was 2-3 d (r = 0.79; P < 0.001) than when the delay was 4-6 d (r = 0.45; P < 0.001) (13). It is expected that younger children are more sensitive to the recall period. The most promising self-reports have used 1 -d recalls with children as young as third grade (5,13-16).

The specificity of data collected from children's physical activity self-reports is often limited. The ideal measure should provide type, frequency, intensity, and duration, but only one(18,19) of 18 self- or proxy-report measures included an assessment of all four dimensions of physical activity(11). Quantitative estimates of physical activity are needed for descriptive descriptive epidemiology and are preferred for evaluating the effects of interventions.

The present investigation evaluated new self-report instruments that were developed to measure physical activity outcomes in the CATCH multicenter school-based health promotion study (10). Because no existing measure met the specifications of the CATCH study, it was necessary to develop and validate a new self-report instrument. First, children in CATCH were fifth graders at the time of post-test, so the new measures were evaluated only with fifth grade students. Second, the CATCH physical activity outcome variable was time engaged in moderate-to-vigorous physical activity, so the new measure must yield estimates of minutes. Third, the measure must be practical for testing large numbers of children with limited resources. Fourth, the measure must be appropriate for use by boys and girls in a variety of ethnic groups in different regions of the United States. Fifth, the measure must have evidence of validity.

The present study design includes a comparison of the validities of alternate forms of the self-report instrument. Although interviewer- and self-administered surveys vary greatly in the cost of administration, they have not been directly compared previously. Such data have important implications for the cost-effectiveness of different measurement approaches. Because of the limitations of all objective criteria commonly used to validate physical activity self-reports (4), both heart rate monitors and accelerometers were used as validation criteria.

METHOD

CATCH Study Description

The Child and Adolescent Trial for Cardiovascular Health (CATCH) is a multicenter study investigating the efficacy of school-based cardiovascular health promotion interventions for third through fifth grade children(10). The four experimental centers are based in San Diego, CA; New Orleans, LA; Minneapolis, MN; and Houston, TX. The coordinating center is in Boston, MA. The goals of the CATCH study are to administer and evaluate food service, physical education, classroom curricula, and family education interventions to improve cardiovascular risk profiles in school children. The CATCH trial involves 96 elementary schools. The present substudy was conducted in nonparticipating schools in the same geographic locations.

Subjects

The study was explained to children in fifth grade classes in public schools at each site. Two hundred twenty-five children took home information sheets and consent forms to their parents. One hundred fifty-three children returned signed consent forms and participated in some portion of the study, so the recruitment rate was 68%. Subjects were eligible for the analyses if they (a) had valid heart rate data and both self-reports; (b) had valid heart rate data, valid accelerometer data, and one of the two self-reports; or (c) had valid accelerometer data with both self-reports. Students were eliminated for the following reasons: six due to absences or they were inadvertently interviewed when their heart rate monitoring data were invalid, seven because of equipment malfunction, 14 because they did not have sufficient heart rate data, and one because of extreme values. Twenty-eight of the 153 participants(18%) did not meet final eligibility requirements. All children were in apparently good health and did not have physical disabilities.

One hundred twenty-five fifth grade children at the four sites met entry criteria: 15 in California, 60 in Louisiana, 26 in Minnesota, and 24 in Texas. PACI data were available on 125, SAPAC data were available on 120, accelerometer data were available on 119, and heart rate data were available on 115 children. Subject characteristics are summarized inTable 1.

Measures

Survey development. Two self-report measures of children's physical activity were developed by representatives of all CATCH sites who jointly conducted extensive formative research during a meeting in New Orleans. Several different formats, including open-ended interviews and checklists, were tested with children not involved in the main CATCH study. Methods of aiding children's memory and improving their time estimation were also pilot tested. After the draft measures and administration protocols were developed, they were further pilot tested at each site, and the list of activities and procedures were refined. Representatives from each site were trained to follow the final protocols, and they participated in the first data collection phase for the present study at another joint meeting in New Orleans. These representatives then trained measurement staff and supervised further data collection at their respective sites.

Physical activity self-reports. Both physical activity self-reports required a 1-d recall using a checklist format. They differed in the form of administration. The Physical Activity Checklist Interview (PACI) was administered in an individual interview, and the Self-Administered Physical Activity Checklist (SAPAC) was administered to an entire class simultaneously. Both measures were modifications and extensions of earlier physical activity checklists (15).

Both forms consisted of a list of 21 physical activities, space for listing up to four other activities, and an additional section for reporting TV/video viewing and video/computer game playing (see Appendix 1). Children reported the minutes they spent in each activity during three time periods of the previous day; before school, during school, and after school. The subject was instructed to report engaging in an activity only if they did so for 5 min or more “at one time.” For every activity reported, children reported whether it caused them to “breathe hard or feel tired none, some, or most of the time.” This rating provided a subjective index of intensity. In summary, for each activity children engaged in the previous day, they reported the segment of the day, the number of minutes, and a subjective intensity rating.

Prior to completing either format of the physical activity checklist, children were prepared with a 5-min introductory presentation. They were oriented to the purpose of the task which was to think about physical activities they did yesterday. “Physical activity” was defined as“bodily movement in which you move your arms and legs.” Children reported the starting and ending of the previous school day, along with the amount of time spent in physical education class and/or recess. To provide some guidelines for estimating time, a clock face was used to demonstrate the length of various common activities such as brushing teeth, having recess, watching commercials on TV, and watching half-hour TV shows. Children's abilities to discriminate activities that were shorter and longer than 5 min were demonstrated by asking open-ended questions. Children were also instructed to report the time they were “actually” active and not to include time they were resting or waiting to play. The preparatory tasks were conducted individually before the PACI and for the entire class before the SAPAC. (Copies of the protocols for administering and scoring the PACI and SAPAC may be requested from Patricia K. Strikmiller, Tulane University School of Public Health and Tropical Medicine, 1501 Canal Street, Room 1207, New Orleans, LA 70112.)

The forms for the PACI and the SAPAC were identical, and the primary difference was the interviewer or self-administration. The PACI also included an interviewer rating of the validity of the interview. During the administration of the SAPAC, one staff member stood in front of the class while giving the preparatory instructions and leading them step-by-step through the administration. One or two other staff members circulated throughout the classroom making sure students were understanding the instructions and following along correctly. In both formats, the research staff read through the checklist three times so the students could recall and record their activities before school, during school, and after school. Students recorded all their own data on the SAPAC, but the interviewer recorded the data on the PACI.

Physical activity monitoring. Two objective measures of physical activity were collected on the day prior to the recalls. Telemetric heart rate monitoring was conducted using the Heart Watch (Computer Instruments Corp., Hempstead, NY). The Heart Watch consisted of a sensing/transmitter unit and a receiver/memory unit. The transmitter unit was secured to the chest with an elastic bandage and adhesive tape to prevent slipping during exercise. The receiver unit was designed to be worn on the wrist, but in the present study it was placed in a “fanny pack” to prevent tampering. The Heart Watch recorded heart rate for each minute up to 16 h. These data were downloaded to a computer via a specialized interface. This instrument is a well-validated measure of physical activity in children(17).

The second physical activity monitor was the Caltrac accelerometer(Hemokinetics, Inc., Madison, WI). This motion detector was about the size and weight of a small pocket calculator and provided a cumulative measure of the quantity and quality of movement in the vertical plane. The accelerometer was programmed to provide an index of motion or physical activity (rather than caloric expenditure) so standard values were entered for subject characteristics (sex = 0, age = 99, weight = 25, height = 36). The accelerometer was placed in the same pouch as the Heart Watch receiver and sealed with a plastic electrical tie. There is extensive evidence that the accelerometer is a highly reliable and moderately valid measure of children's physical activity (13).

Procedure

Children participated in the study for two consecutive days. Immediately after the start of school on day 1, the heart rate monitor and accelerometer were secured to the children. They were instructed to keep the equipment dry and to leave the instruments on until they were ready to go to bed. The monitors were worn for a full school day and most of the evening.

On the morning of day 2, the monitors were collected. Children reported what time they took off the monitors. The accelerometer score was recorded and considered valid if the score was between 20 and 200. The heart rate data were downloaded to computers on-site and were considered usable if there were at least 8 h of valid data. Criteria for acceptance of heart rate data were that at least 90% of heart rates were between 40 and 220 bpm. If the monitoring data were invalid, the child was scheduled for another day of monitoring and did not complete the surveys on that day. The mean minutes of heart rate monitoring for students in the study was over 11.5 h (mean = 699 min; SD = 89 min).

On the morning of day 2, all children completed both the PACI and the SAPAC. After both surveys were collected, participation in the study was completed, and children were provided with their choice of incentives. Choices included colorful shoestrings, pencil sharpeners in novel shapes, and caps with the CATCH logo.

Data Reduction

One of the aims of the present study was to examine the effects of different methods of summarizing self-reported data on validity. Four types of data were available on both the PACI and SAPAC. (a) The simplest measure was number of activities with reported participation of 5 min or more. (b) Students reported minutes in activities, so total minutes of physical activity was the second summary measure. (c) The intensity of activities should be considered in computing the overall volume of physical activity, so time in each activity was weighted with MET values taken from the recently published compendium (1). “MET” is a multiple of resting metabolic rate, so MET values reflect the energy cost of activities.(d) The published MET values represent the midpoint of a broad range of possible intensity values. That is, a given child could play basketball vigorously or with a minimum of exertion. The child's rating of intensity, based on symptoms of exertion (i.e., breathe hard or feel tired), may reflect individual differences in intensity. Including these ratings in the physical activity summary score may enhance validity of the measures.

The use of the intensity ratings was not straightforward. For example, there were no guidelines for adjusting standard MET values for individual differences in intensity, so weightings were arbitrary. In the current study, different weighting rules were applied to activities of different intensities. If an activity was light to moderate, heavy breathing was not expected, so activities in the 1.1-5.9 MET range were multiplied by 1.1 if they reported being out of breath “some” of the time and by 1.25 if they reported being out of breath “most” of the time. For activities that were hard (i.e., 6 METs or higher, based on the categorization of Jacobs et al. (8)), heavy breathing was expected at least some of the time. If they reported being out of breath “most” of the time, the MET score was multiplied by 1.25. However, if they reported being out of breath “none” of the time, the activity was less vigorous than expected, and the MET value was multiplied by 0.75.

Self-reported data were summarized with the following five variables: (a) minutes in sedentary pursuits; (b) number of activities reported; (c) minutes of moderate to vigorous physical activity (min of MVPA); (d) physical activity MET score (MVPA METs) (min of activity × MET value); (e) weighted activity MET score (weighted MVPA METs) (min of activity × MET value× intensity rating).

These variables were computed for the entire day and for the whole day minus the “before school” period. Before school activities were deleted for the validity analyses because monitors were not attached during this period.

Statistical Analysis

Mean values for each activity measure were tabulated overall and by gender. Comparison between the genders was made by analysis of covariance (ANCOVA), controlling for site. Corresponding activity measures from PACI and SAPAC were compared pairwise by ANCOVA, covarying for site and order of administration.

Validity correlations relating PACI and SAPAC to heart rate and accelerometer readings, as well as correlations between PACI and SAPAC, and correlations among the alternative heart rate measures, were computed using the standard Pearson formula. Partial correlations, controlling for order of administration, were also computed, but they were very similar to the simple correlations and are not reported. Validity correlations were calculated overall and by ethnic group. Sample size was 125 for the overall correlation; 70 for girls, 55 for boys; 28 for PACI administered first; 97 for SAPAC administered first; and between 10 and 52 for the ethnic subgroups, making comparisons among the smaller subgroups imprecise.

The accuracy of self-reported time in physical activity was assessed by paired Student t-tests, comparing PACI and SAPAC measures with the minutes of heart rate exceeding resting rate plus 60 bpm.

RESULTS

Administration Time

The mean administration time of the PACI was 17.5 min, with a standard deviation of 5.0 min. The mean administration time of the SAPAC was about 35 min, with a range of 27-40 min.

Ratings of Interview Quality

After each PACI the interviewer rated six items to assess the quality of the interview. Responses ranged from poor to excellent on a 5-point scale. The percent of interviews rated “very good” or “excellent” was as follows: paying attention during interview-67%; accuracy of recall-54%; accuracy of time estimates-41%; co-operativeness-83%; believability of reports-53%; overall quality of the interview-56%. Most of the remaining ratings were “good,” but these interviewer impressions suggest the difficulty of the recall task for children of this age.

Effects of Order of Administration

The intended design was to randomly assign counter-balanced orders of administration to sites, to control for order effects. However, this plan could not be fully implemented, resulting in an imbalance. PACI was administered first in 28 cases, and SAPAC was administered first in 97 cases. Two sets of analyses were conducted to test for order effects. Number of activities, minutes of MVPA, and MVPA METs from the PACI and SAPAC (for the entire day) were compared in repeated-measures ANOVAs, with an additional factor for order of administration. The order of administration had no statistically significant influence on either the PACI or SAPAC values for these three variables. The P-values did not even approach significance.

The second set of analyses examined effects of order of administration on correlations with accelerometer and heart rate monitoring values for PACI and SAPAC variables (based on during and after school). These results are shown inTable 2. Order of administration had no influence on correlations of the accelerometer with either PACI or SAPAC variables. However, there were large differences in PACI and SAPAC variable correlations with heart rate data. The heart rate data suggest that the sample taking the SAPAC first were more accurate reporters than the sample taking the PACI first. However, the accelerometer results are not consistent with this interpretation. Inspection of scatterplots for the small PACI-first sample indicated a restricted range of self-reported MVPA likely reduced the correlation with heart rate.

Even though the order of administration was not fully counterbalanced in the present study, there was limited evidence that the order influenced the ability to compare the performance of the PACI and SAPAC measures. In addition, the self-administered SAPAC was administered first most of the time. There would have been concern about overestimating the validity of the SAPAC if it had followed the individually administered PACI most of the time. To evaluate the effect of order of presentation on validity coefficients, all correlations in Tables 5, 6, and 7 were computed as standard Pearsons and as partial correlations, controlling for order of administration. The typical difference between pairs of correlations was 0.00-0.03, with standard correlations being higher sometimes, and partial correlations being higher sometimes. Because of the lack of practical effect on the results, all correlations in Tables 5-7 are standard Pearsons.

Comparisons of Physical Activity by Gender

Descriptive data for heart rate monitoring, accelerometer, PACI, and SAPAC scores are shown in Table 3 for the entire group, for girls, and for boys. Boys were significantly more active than girls on accelerometer counts and some PACI measures.

Comparisons of PACI and SAPAC Data

It is of interest to determine whether the two forms elicited similar reports from the children. Means and standard deviations of all five activity summary variables from both survey forms are reported inTable 4. On all variables, including sedentary minutes, scores were higher on the SAPAC, and all the differences were significant. For example, children reported 22 more minutes of MVPA on the SAPAC than on the PACI, which is a 25% difference. Because children reported more physical activity and more sedentary behavior on the self-administered form, it is possible that the interview process inhibited reporting or that self-administration promoted exaggeration. The intraclass correlations demonstrated substantial similarity between the two forms, with coefficients ranging from 0.64 (minutes in MVPA) to 0.79 (number of activities).

Validity

Assessment of methods of scoring heart rate monitoring data. There is no consensus on the most appropriate method of summarizing data from heart rate monitoring. Therefore, analyses were conducted to compare various procedures for scoring heart rate data. To select the most appropriate heart rate score to use as a criterion, heart rate and self-report variables were correlated. This approach is not intended to “validate” a heart rate scoring procedure with self-reports. Rather, the aim is to select a heart rate score that appears to assess similar dimensions of physical activity that are being reported.

The most common heart rate scoring method is to tally the minutes spent in heart rates greater than 140 bpm (2,7). This summary reflects physical activity at a vigorous (or moderately vigorous) level, which would be approximately 70% of age-predicted maximum heart rate in this sample. This index has been used to validate self-reports of vigorous physical activity (14), but it has the disadvantage that it does not adjust for substantial individual differences in resting heart rate (7).

Several methods have been suggested to adjust for resting heart rate. In the present sample, the mean resting heart rate (based upon the mean of the five lowest valid heart rates) was 72.7 (SD = 8.3), and the range was 50.0-93.4. DuRant and colleagues (6) recommended adding 25% or 50% to resting heart rate, and these two measures were examined in the present study. The thresholds for resting heart rate plus 25% and 50% for the average child were 90.8 (45% of estimated maximal heart rate) and 109.1 (54% of estimated maximal heart rate), respectively.

Adding a standard amount to resting heart rate may be preferable to adding a variable amount. Therefore, a fourth method was resting heart rate plus 60 bpm (suggested by Patty S. Freedson, personal communication, March 1993). On average, the threshold for defining physical activity with this method was 132.7 (66% of estimated maximal heart rate).

The four methods of scoring heart rate data were compared in two ways. First, the minutes of physical activity were calculated and compared to self-reported minutes from the PACI and SAPAC. The estimated minutes of physical activity differed greatly across the four heart rate summaries: 469 min (SD = 111) for resting plus 25%, 179 min (SD = 95) for resting plus 50%, 29 min (SD = 29) for greater than 140 bpm, and 43 min (SD = 34) for resting plus 60 bpm. Measures of self-reported minutes of MVPA and minutes in vigorous MVPA (i.e., breathing hard or tired “most of the time”) from PACI and SAPAC ranged from 21 to 94 min. Thus, the discrepancies between self-reported and objectively measured minutes of physical activity differed greatly for the resting plus 25% (range of differences = -380 to -450 min) and resting plus 50% (range of differences = -89 to -149 min). The differences were much smaller for heart rate greater than 140 bpm (range of differences =-10 to 61 min) and resting plus 60 bpm (range of differences = -12 to 48 min).

The second method of evaluating the heart rate summaries involved correlating them with the accelerometer measure and the total MVPA minutes and MVPA MET scores from the PACI and SAPAC. These results are presented inTable 5. The correlations were consistently lower for the resting plus 25% and resting plus 50% measures. The correlations were similar for the heart rate greater than 140 bpm and the resting plus 60 bpm measures. Because there appeared to be a slight advantage for the resting plus 60 bpm measure, it was used in all subsequent validity analyses.

Comparison of PACI and SAPAC with objective measures. Correlations of PACI and SAPAC summary variables with the two objective physical activity measures provided evidence to support the concurrent criterion-related validity of the self-reports. No objective measure of sedentary behavior was available, so no validity data are presented for sedentary behavior. Table 6 presents these results for the entire sample, for girls, and for boys. Correlations were consistently higher for heart rate than for the accelerometer measure. Correlations with heart rate tended to be higher for boys, and correlations with the accelerometer were generally higher for girls.

Differences in validity coefficients between the PACI and SAPAC were not large, and they depended upon the validation measure. The SAPAC had slightly higher correlations with heart rate, and the PACI had slightly higher correlations with the accelerometer. Therefore, the data did not strongly favor the validity of one self-report format over the other.

Four self-report scoring methods were compared. The simplest self-report measure, number of activities, always had the lowest correlations with the objective measures. Minutes of MVPA always had higher correlations than number of activities. There were consistently very small increases in correlations when minutes were weighted by MET scores, but there were no further improvements when the perceived intensity ratings were added.

It is important to establish that the self-report measures are valid in subgroups of children. Data in Table 6 indicate that both self-reports were moderately and significantly correlated with two objective measures in both boys and girls. Table 7 shows correlations between self-report and objective measures in four racial/ethnic groups. There were sufficient numbers of African American and European American children to allow confident interpretations of correlations, but the correlations for the small Latino and Asian/Pacific Islander groups are presented for completeness and are considered unreliable. There were significant correlations in all subgroups, supporting the validity of the self-reports in various ethnic groups. In general, the European Americans and Latinos tended to have the lowest correlations, and African Americans and Asian/Pacific Islanders had the highest validity correlations.

Correlations demonstrated that the relative levels of physical activity assessed by self-report and objective measures were significantly related. However, the correlations did not indicate how accurate the children were in estimating the minutes of physical activity. Table 8 shows the results of paired t-tests that compared self-reported minutes of physical activity with minutes based on heart rate monitoring. These analyses excluded reports of before-school activity, because there was no activity monitoring during this time. Two scoring methods were compared for the self-reports. The first score was total minutes of MVPA. The second score was the total minutes in which children indicated on the intensity ratings they were out of breath or tired “most of the time,” termed“minutes of vigorous MVPA.”

Every t-test was significant, indicating that children were not accurate in their estimates of minutes of physical activity. However, reported total minutes of MVPA always overestimated monitored physical activity, while reported vigorous minutes always underestimated time spent in physical activity, as assessed by heart rate monitoring. SAPAC total MVPA minutes overestimated objectively determined activity minutes more than PACI total MVPA minutes, but SAPAC vigorous MVPA minutes were substantially more accurate than PACI vigorous MVPA minutes.

DISCUSSION

A new self-report measure of children's physical activity was developed specifically to meet the needs of the CATCH school-based intervention study(10), but it is likely to be useful for a variety of other research needs as well. The development phase involved a preliminary assessment of several self-report formats, and a checklist format was chosen for further study. The study was designed to compare the validity of two versions of the physical activity checklist; self-administered and interviewer-administered. In addition, different scoring procedures were evaluated, and two objective validation measures were employed. The results indicated the self-administered and interviewer-administered versions provided data with comparable levels of validity.

It is well-documented that boys are more physically active than girls at all ages tested (12), and this pattern was confirmed with both self-reported and objective measures in the current study. Boys were 38% more active on the heart rate measure and 16% more active according to the accelerometer. Boys reported 14% to 17% more minutes of MVPA than girls, but when intensity of the activity (i.e., METs) was considered boys reported 26% to 28% more physical activity than girls. Estimates of the gender differential in physical activity varied by the physical activity measure used.

The intraclass correlations between the PACI and SAPAC instruments indicated that the two measures elicited reliable reports. The correlations, ranging from 0.64 to 0.79, were within the range of test-retest reliabilities for other physical activity self-report instruments(11). This degree of similarity is not surprising, given that the two forms were completed at roughly the same time. However, a comparison of means showed that children reported more minutes of both physical activity and sedentary behavior on the self-administered SAPAC. This difference was not a function of the order of administration.

Comparison of self-reported minutes of activity to heart rate recordings suggests that children were not accurate reporters of the absolute minutes of their physical activity. The average overestimate was 29 min on the PACI and 48 min on the SAPAC, with some variables overestimating activity by more than 100%. These figures are very sensitive to changes in the scoring of heart rate data. Nevertheless, it appears from the present data that self-administered recalls of physical activity produced exaggerated reports. Although children's absolute estimates of minutes of physical activity should not be regarded as accurate, their reports appear to be useful as measures of relative amount of physical activity.

The most general finding was that both objective measures supported the validity of both versions of the children's physical activity self-report measures. All validity correlations in Table 6 were significant, and it may be most instructive to consider the correlations for the total group, because the range of physical activity was maximized. The apparent differences in validity correlations between the PACI and SAPAC were neither large nor consistent across the two objective measures. The SAPAC had slightly higher correlations with the heart rate measure, but the PACI had slightly higher correlations with the accelerometer. None of these differences between the SAPAC and PACI were statistically or practically significant. The present results indicated both instruments were equally valid as measures of relative amounts of physical activity.

There were benefits of including two validation criteria in the study, even though each has limitations. The primary benefit is the ability to replicate validity correlations within the same study. In the present study, validity of self-reports on the PACI and SAPAC was supported by both objective measures.

The validity correlations with the heart rate monitor were substantially higher than the correlations with the accelerometer. This pattern was not found in another study with youth that evaluated a 1-d recall with the same two objective measures (13). However, the method of scoring the heart rate data in the present study was different, and it appeared the resting heart rate plus 60 bpm formula for defining minutes of MVPA more accurately reflected children's engagement in physical activity than several other methods of summarizing heart rate monitoring data. The accelerometer data did not provide the flexibility to evaluate different scoring methods.

The level of the correlations between the self-reports and the objective measures must be considered. The observed correlations ranged from 0.30 to 0.60, so none of them could be considered high. However, the heart rate correlations, ranging from 0.50 to 0.60, should be considered of moderate strength, and they were higher than any other reported correlations with the physical activity self-reports of children of this age(11). The accelerometer correlations were low, but they were very similar to other studies of short-term recalls of children's physical activity (13,15). The shortcomings of the accelerometer as a physical activity measure in children have been identified (7), so it was expected that the heart rate would be a more accurate validation criterion. For example, the accelerometer does not accurately assess the energy expenditure of riding a bicycle which is a very common activity among this age group. The correlation between heart rate and accelerometer monitoring in the present study was 0.47, making clear that one or both of the objective measures was less than desirable. Despite these limitations, the validity of the PACI and SAPAC as measures of relative amounts of physical activity in fifth grade children was supported by three findings: (a) the validity correlations were significant for both objective measures; (b) the validity correlations were higher with the heart rate measure, which was believed to be a more accurate criterion; (c) the validity correlations were higher than those of other published physical activity self-reports with children of this age.

The validities of the SAPAC and the PACI were supported in all subgroups of children examined. Correlations with objective measures were significant for girls, boys, and all ethnic subgroups. These findings are of considerable importance given the racial and ethnic diversity of many schools. The geographic, gender, and ethnic heterogeneity of the sample in the present study supports the generalizability of the results to other groups of fifth grade children. However, if the instruments are used with highly dissimilar samples, validity should be established in those samples.

Though all validity correlations were significant for both girls and boys, an interesting pattern emerged. For the heart rate criterion, boys tended to have higher correlations with self-reports than girls. For the accelerometer criterion, the findings were reversed, and girls tended to have the higher correlations. A possible explanation is that heart rate correlations were higher for boys because they engaged in more vigorous activities that produced elevated heart rates. This surprising finding is worthy of replication and further study. It may be necessary to develop different heart rate thresholds for defining physical activity for boys and girls.

The scoring of the PACI and SAPAC had a substantial influence on the validity correlations. The number of activities reported was clearly an inferior measure. The sum of minutes of MVPA performed nearly as well as two more complex measures. Approximating an estimate of energy expenditure by using MET weightings may have improved the correlations slightly, so the use of MET weightings can be recommended. However, additional adjustment by using perceived intensity weightings was clearly not effective. Because the intensity weightings add an element of complexity to the administration of the self-reports, it is recommended that they not be collected.

The PACI and SAPAC yielded evidence of similar validity, but they were not equivalent in the cost of administration. The individual interviews of the PACI were much more costly than the classroom administration of the SAPAC. Therefore, SAPAC appears to be the more cost-effective choice.

The strengths of the present study included a gender-balanced, ethnically diverse sample that was drawn from four sites around the United States. The design of the study allowed for two formats of the same physical activity assessment instrument to be evaluated and compared. The present study was the first to directly compare self-administered and interviewer-administered versions of a physical activity recall, and further comparisons of different self-report measures are encouraged. A variety of methods of scoring the recalls were examined. Comprehensive analyses compared the reliability of reporting across the two formats, documented validity for subgroups, and compared minutes of physical activity as assessed by self-report and heart rate recording. One of the major limitations of the study was the age-specific nature of the findings. All subjects were in the fifth grade, and due to the strong relation between age and validity of self-reported physical activity(14), it is not possible to generalize these results to other age groups. Both physical activity checklists required only a 1-d recall, so several administrations will be required to adequately characterize habitual activity levels of individuals.

The Physical Activity Checklist Interview (PACI) and Self-Administered Physical Activity Checklist (SAPAC) were carefully developed measures that were shown to have acceptable levels of validity for fifth grade girls and boys. These measures can be considered to be moderately valid measures of relative levels of physical activity, but they produced inaccurate estimates of absolute minutes of physical activity. Both versions were practical for use in studies of children's physical activity, but the SAPAC was found to be equally valid but less costly than the PACI.

REFERENCES

1. Ainsworth, B. E., W. L. Haskell, A. S. Leon, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med. Sci. Sports Exerc. 25:71-80, 1993.
2. Armstrong, N. and Bray, S. Physical activity monitoring patterns defined by continuous heart rate monitoring. Arch. Dis. Child. 66:245-247, 1991.
3. Baranowski, T. Validity and reliability of self report measures of physical activity: an information-processing perspective.Res. Q. Exerc. Sport 59:314-327, 1988.
4. Baranowski, T., C. Bouchard, O. Bar-Or, et al. Assessment, prevalence, and cardiovascular benefits of physical activity and fitness in youth. Med. Sci. Sports Exerc. 24:S237-S247, 1992.
5. Baranowski, T., R. Dworkin, C. J. Cieslik, et al. Reliability and validity of self-report of aerobic activity: Family Health Project. Res. Q. Exerc. Sport 55:308-317, 1984.
6. DuRant, R. H., T. Baranowski, H. Davis, et al. Reliability and variability of indicators of heart-rate monitoring in children. Med. Sci. Sports Exerc. 25:389-395, 1993.
7. Freedson, P. S. Electronic motion sensors and heart rate as measures of physical activity in children. J. Sch. Health 61:220-223, 1991.
8. Jacobs, D. R., B. E. Ainsworth, T. J. Hartman, and A. S. Leon. A simultaneous evaluation of 10 commonly used physical activity questionnaires. Med. Sci. Sports Exerc. 25:81-91, 1993.
9. McKenzie, T. L. Observational measures of children's physical activity. J. Sch. Health 61:224-227, 1991.
10. Perry, C. L., E. J. Stone, G. S. Parcel, et al. School-based cardiovascular health promotion: Child and Adolescent Trial for Cardiovascular Health. J. Sch. Health 60:406-413, 1990.
11. Sallis, J. F. Self-report measures of children's physical activity. J. Sch. Health 61:215-219, 1991.
12. Sallis, J. F. Epidemiology of physical activity and fitness in children and adolescents. Crit. Rev. Food Sci. Nutr. 33:403-408, 1993.
13. Sallis, J. F., M. J. Buono, J. A. Roby, D. Carlson, and J. A. Nelson. The Caltrac accelerometer as a physical activity monitor for school-age children. Med. Sci. Sports Exerc. 22:698-703, 1990.
14. Sallis, J. F., M. J. Buono, J. J. Roby, F. G. Micale, and J. A. Nelson. Seven-day recall and other physical activity self-reports in children and adolescents. Med. Sci. Sports Exerc. 25:99-108, 1993.
15. Sallis, J. F., S. A. Condon, K. J. Goggin, J. J. Roby, B. Kolody, and J. E. Alcaraz. The development of self-adminstered physical activity surveys for 4th grade students. Res. Q. Exerc. Sport 64:25-31, 1993.
16. Simons-Morton, B. G., T. Baranowski, N. M. O'Hara, I. W. Huang, and B. Wilson. Children's frequency of participation in moderate to vigorous physical activity. Res. Q. Exerc. Sport 61:307-314, 1990.
17. Treiber, F. A., L. Musante, S. Hartdagan, M. Levy, and W. B. Strong. Validation of a heart rate monitor with children in laboratory and field settings. Med. Sci. Sports Exerc. 21:338-342, 1989.
18. Verschuur, R. and H. C. G. Kemper. Habitual physical activity. Med. Sports Sci. 20:56-65, 1985.
19. Verschuur, R. and H. C. G. Kemper. The pattern of daily physical activity. Med. Sports Sci. 20:169-186, 1985.

Appendix 1

Table

Appendix 2

Table

Keywords:

MEASUREMENT; SELF-REPORT; VALIDITY; CHILDREN; PHYSICAL ACTIVITY; EPIDEMIOLOGY

©1996The American College of Sports Medicine