See corresponding editorial on page 736.
Is it possible to assess free-living physical activity and energy
expenditure in young people by self-report?1–3
Kirsten Corder, Esther MF van Sluijs, Antony Wright, Peter Whincup, Nicholas J Wareham, and Ulf Ekelund
Background: It is unclear whether it is possible to accurately estimate
physical activity energy expenditure (PAEE) by self-report in youth.
Objective: We assessed the validity and reliability of 4 self-reports
to assess PAEE and time spent at moderate and vigorous intensity
physical activity (MVPA) over the previous week in British young
people between 4 and 17 y of age.
Design: PAEE and MVPA were derived from the Children’s Phys-
ical Activity Questionnaire, Youth Physical Activity Questionnaire,
and Swedish Adolescent Physical Activity Questionnaire; a lifestyle
score indicative of habitual activity was derived from the Child
Heart and Health Study in England Questionnaire. These data were
compared with criterion methods, PAEE, and MVPA derived from
simultaneous measurements by doubly labeled water and acceler-
ometry in 3 age groups: 4–5 y (n ¼ 27), 12–13 y (n ¼ 25), and 16–
17 y (n ¼ 24). Validity was assessed by using Spearman correlations
and the Bland-Altman method, and reliability was assessed by using
intraclass correlation coefficients.
Results: The strength of association between questionnaire and
criterion methods varied (r ¼ 0.09 to r ¼ 0.46). Some question-
naires were able to accurately assess group-level PAEE and MVPA
for some age groups, but the error was large for individual-level
estimates throughout. Reliability of the Youth Physical Activity
Questionnaire and Child Heart and Health Study in England Ques-
tionnaire was good (intraclass correlation coefficient: 0.64–0.92).
Conclusions: Absolute PAEE and MVPA estimated from these self-
reports were not valid on an individual level in young people, although
some questionnaires appeared to rank individuals accurately. Age (the
outcome of interest) and whether individual or group-level estimates
are necessary will influence the best choice of self-report method when
assessing physical activity in youth.
Am J Clin Nutr 2009;89:862–
The observed magnitude of the relation between physical
activity and health varies considerably, especially in children (1);
this is largely due to the difficulties of making accurate assess-
mentsofphysicalactivity energyexpenditure (PAEE)and patterns
of physical activity in large populations.
Despite their accuracy, gold-standard methods of energy ex-
(DLW), are prohibitively expensive for use in large studies; how-
ever, they are useful for the validation of other methods potentially
suitable for large-scale use (2, 3). The DLW method provides
precise information on total daily EE (TEE) during daily living
over a relatively long period and can be used to calculate PAEE by
using measured or estimated REE. Accelerometry is probably the
most widely used objective method of assessing physical activity
and provides an assessment of frequency, intensity, and duration
of physical activity. Recent large-scale studies have successfully
used accelerometry to assess physical activity in youth (4, 5).
However, self-report methods may still be the only feasible way
to assess physical activity in many situations and are important
for assessing aspects of physical activity not easily measured
objectively, such as mode and domain (6). Some questionnaires
can accurately determine the mode of activity and can be used to
adequately rank, group,orcategorizephysicalactivity levels (7,8)
and possibly to assess some aspects of moderate and vigorous
less accurate than objective methods for estimating PAEE and
because of the cognitive limitations of those aged ,10 y (14, 15);
often being used to estimate EE, most recent questionnaire vali-
dation studies have used accelerometryasa criterion (6, 9,10, 17);
relatively few have used DLW (7, 8). Because most questionnaires
rarely explored, even though these dimensions may be of differ-
ential importance to a variety of health outcomes. To our knowl-
edge, this is the first study to simultaneously assess the validity of
estimated PAEE and MVPA from physical activity questionnaires
assessment of the strengths and weaknesses of the self-reported
study compares 4 questionnaires in different age groups, which
1From the MRC Epidemiology Unit, Institute of Metabolic Science,
Cambridge, United Kingdom (KC, EMFvS, NJW, and UE); MRC Human
Nutrition Research, Elsie Widdowson Laboratory, Cambridge, United King-
dom (AW); and the Department of Community Health Sciences, St George’s,
University of London, London, United Kingdom (PW).
2Supported by The Medical Research Council, United Kingdom. The
CHASE study is supported by the Wellcome Trust.
3Reprints not available. Address correspondence to U Ekelund, MRC
Epidemiology Unit, Institute of Metabolic Science, Box 285, Addenbrooke’s
Hospital, Hills Road, Cambridge CB2 0QQ, United Kingdom. E-mail:
Received July 24, 2008. Accepted for publication October 13, 2008.
First published online January 14, 2009; doi: 10.3945/ajcn.2008.26739.
Am J Clin Nutr 2009;89:862–70. Printed in USA. ? 2009 American Society for Nutrition
studies, including those that include multiple age groups of youth.
and validity of 4 physical activity questionnaires compared with
measurements of PAEE and MVPA obtained with the DLW tech-
SUBJECTS AND METHODS
This study was carried out in a convenience sample of 82
volunteers aged between 4 and 17 y and recruited from schools in
the study were 2 full classes taught by contact teachers at each of
3 secondary schools, all of those aged 4 or 5 y at 2 primary
infant study during the measurement period (January 2005 to
December 2006). Those who had a medical condition that se-
verely limited their normal physical activity were to be excluded,
although nonewith such conditions volunteered for the study. All
volunteers received an explanation of the study and all provided
assent; a parent provided written informed consent for partici-
pation in the study, which was approved by the Cambridge local
research ethics committee. Of 337 subjects invited, 86 attended
a testing session, DLW data were obtained for 84 volunteers, and
accelerometry data were obtained for 82 volunteers. Volunteers
were recruited in 3 age groups: 4–5 y (n ¼ 29), 12–13 y (n ¼ 27),
and 16–17 y (n ¼ 26). Of these 82 volunteers, 79 completed the
first questionnaire administered. Of these 79 volunteers, accel-
erometry data were not available for 3 because of monitor
malfunction, which resulted in 76 volunteers with complete
validation data: 4–5 y (n ¼ 27), 12–13 y (n ¼ 25), and 16–17 y
(n ¼ 24). An additional 14 volunteers did not complete the
second administration for reliability purposes, which resulted in
the following valid pairs for analysis: 4–5 y (n ¼ 20 pairs), 12–
13 y (n ¼ 21 pairs), and 16–17 y (n ¼ 20 pairs).
Anthropometric measurements were made on the day of DLW
dosing for all volunteers while they were wearing light clothing
and no shoes and socks (Figure 1). Weight was measured to the
Japan), and height was measured to the nearest 0.1 cm with a cali-
brated stadiometer (Chasmors Ltd, London, United Kingdom).
Doubly labeled water (PAEE)
Total EE (TEE) was measured by using the DLW technique
over 11 consecutive days. Each volunteer received a weighed
dose equivalent to 0.174 g H218O and 0.07 g2H2O/kg body wt,
after providing 2 predose urine samples: 1 on the day of dosing
and 1 on the preceding day. Volunteers were instructed to pro-
vide the first postdose urine sample ’24 h after dosing and then
at a similar time, but not the first void of the day, for the next 10 d
(Figure 1). DLW analysis was carried out by using isotope ratio
mass spectrometry, as described previously (18). TEE was cal-
culated by using standard equations (19) and Schoeller’s esti-
mation of carbon dioxide production (20), which normalizes
2H/18O space ratios to 1.04/1.01 ¼ 1.03 (21, 22). TEE was ob-
tained from carbon dioxide production, assuming carbohydrate,
fat, and protein substrate oxidation with a respiratory quotient of
0.85 (23). REE was measured for a minimum of 15 min by using
indirect calorimetry after at least a 2-h fast and 10 min of supine
rest. Oxygen consumption and carbon dioxide production were
measured with an online system (Jaegar Oxycon Pro; Viasys
Health Care, Warwick, United Kingdom). A ventilated hood was
used for this measurement in the oldest age group, and a hollow
facemask was used in the other groups. Data were averaged over
a 15-s epoch; the 2 most extreme values in each interval were
disregarded. EE was then computed by using the de Weir equa-
tion (24). REE was determined by averaging EE over minutes
5–15 of the resting test. PAEE was calculated as PAEE ¼ 0.9 3
TEE 2 REE and PAEE/kg as (0.9 3 TEE 2 REE)/kg body wt.
Accelerometry (physical activity)
Physical activity was objectively assessed by using the Acti-
graph accelerometer (model 7164; Manufacturing Technologies
Inc, Shalimar, FL). The Actigraph has been shown to accurately
assess EE in European children during free-living conditions (17,
25). The Actigraph was worn for 11 d, concurrent with the DLW
measurement, set to record at 60-sepochs and placed centrally on
the hip with the side of placement randomly assigned. Volunteers
were asked to wear the monitors during waking hours and to re-
move them while bathing, showering, and swimming.
Accelerometry datawere analyzed by using a batch processing
and was used as minutes spent in MVPA (MVPAAcc), derived by
using 3000 accelerometry cpm as the lower limit of moderate
activity (26). Analyses were also carried out by using 1952 cpm
as the lower limit of MVPA (27) because of a lack of consensus
on the most suitable cutoffs when translating accelerometric
intensity into physiologic intensity. When 20 min of consecutive
zeros were present in the accelerometry data, they were removed
and it was assumed that the monitor was unworn at that time;
consequently, all days consisting of .600 min of valid data were
included in the analysis.
Youth Physical Activity Questionnaire
The Youth Physical Activity Questionnaire (YPAQ) was ad-
ministered to the 2 oldest groups and was based on the Children’s
Leisure Activities Study Survey (CLASS) (10). The original intent
of the self-reported CLASS questionnaire for 10–12-y-olds was to
assess type, frequency, and intensity of physical activity over
and EE data (10). The YPAQ lists 47 different activities with
FIGURE 1. Timeline of the study protocol. DLW, doubly labeled water.
VALIDITY OF 4 PAQs IN YOUTH
each activity for both week and weekend days over the past 7 d.
Therefore, the YPAQ assesses mode, frequency, and duration of
physical activity and sedentary activities throughout all domains,
including school time and leisure time over the past 7 d.
Children’s Physical Activity Questionnaire
The Children’s Physical Activity Questionnaire (CPAQ) was
administered to the youngest group and is also based on the
CLASS questionnaire, very similar to the YPAQ but parentally
reported and includes activities specific to young children, such
as ‘‘playing in a playhouse.’’ The original intent of the proxy-
reported CLASS questionnaire for 5–6-y-olds was to assess type,
frequency, and intensity of physical activity over a usual week
and was compared with accelerometer-derived MVPA and EE
data (10). In this study, the CPAQ was parentally reported for the
youngest group of volunteers over the past 7 d. The mother of
the child completed all administrations of the questionnaire for
all participants. Therefore, the CPAQ assesses mode, frequency,
and duration of physical activity and sedentary activities
throughout all domains over the past 7 d.
Child Heart and Health Study in England Questionnaire
This questionnaire is used in the Child Heart and Health Study
in England (CHASE) Study, which is examining the health of
’5000 9–10-y-old primary school children living in the United
Kingdom (www.chasestudy.ac.uk) and is currently unpublished.
This questionnaire consists of 25 questions and addresses the
mode and frequency of physical activity and sedentary activities
throughout all domains, including school time and leisure time,
and also includes multiple-choice questions regarding lifestyle
activities. Therefore, the CHASE questionnaire solicits semi-
quantitative estimates of the duration spent in broad categories
of physical activities and weekly frequency of discrete activities.
It was administered to the 12–13-y-old group.
Swedish Adolescent Physical Activity Questionnaire
The Swedish Adolescent Physical Activity Questionnaire
(SWAPAQ) is a translation of the self-reported past 7-d physical
accelerometry in Swedish adolescents (28). The original ques-
tionnaire assessed frequency, duration, and intensity of activity
during school, transport, and leisure time over the past 7 d, and
data were used as the duration of MVPA and MET (metabolic
equivalent)-minutes of physical activity calculated as duration 3
consists of 25 questions and addresses the mode, frequency, and
duration (in broad categories) of physical activity throughout all
domains and was administered to the 16–17-y-old group.
The questionnaires are available as ‘‘Supplemental Data’’ in
the online issue. The questionnaires were either mailed to the
parents of volunteers (4-5-y-olds) or distributed by school
teachers (to 12–13- and 16–17-y-olds), and no instructions were
given over that written on the questionnaires. For all volunteers,
the first administration of the questionnaire was given on day 11
of the DLW measurement, referring to days 4–10 and then 7 d
later referring to the same period, as shown in Figure 1. The first
administration was used to assess validity; the second admin-
istration was used only to assess reliability. When a volunteer
received 2 different questionnaires simultaneously, the YPAQ
and CHASE for the 12–13-y-olds and the YPAQ and SWAPAQ
for the 16–17-y-olds, the order in which the questionnaires were
administered was randomly determined, however, in the same
order for both administrations.
Questionnaire data processing
For the YPAQ and CPAQ, frequency and duration of listed
physical activities were reported; these activities were assigned
a MET value according to published values (29). Because only
summary items were assessed in the SWAPAQ (eg, How much
time on average did you spend doing vigorous physical activi-
ties?), it was not possible to assign individual MET values to
specific activities. Consequently, values of 4 and 6 METs were
assumed for moderate and vigorous physical activity, respec-
tively, and MET-minutes were calculated as follows: duration 3
frequency 3 MET intensity.
Estimates of PAEE were derived from the YPAQ, CPAQ, and
SWAPAQ by using a method similar to that described previously
(2). It was assumed that 1 MET is equivalent to an oxygen con-
sumption rate of 4.00 mL ? kg21? min21for 16–17-y-old ado-
lescents and of4.58 mL? kg21? min21for12–13-y-olds(30). For
the 4–5 y-old children, the published values for the older age
groups (30) were extrapolated down for 5 y-old children, which
resulted in an estimate of 7.0 mL ? kg21? min21as the MET
equivalent. The oxygen energy equivalent was assumed to be
0.0209 kJ/mL, and the formula used to estimate daily PAEE
kg21? d21) ¼ 1440 3 [(0.0209 3 MET) 3 (total MET-min/total
time frame)].Self-reportedminutes of MVPA perweek were also
summed from the YPAQ, CPAQ, and SWAPAQ (MVPAQ) for
direct comparison with MVPAAcc.
The CHASE questionnaire did not include information on
activity duration, so MVPAQand PAEEQcould not be calcu-
lated. Four multiple choice questions regarding active transport,
school break activities, activity outside school, and the amount
and frequency of ‘‘exercise that makes you out of breath’’ were
summed into a lifestyle score to represent habitual physical ac-
tivity. The multiple-choice answers to each question were ranked
by using consecutive integers with the least active choice given
a score of 1 and the most active option given the highest score
(eg, 5). The numbers corresponding to each answer were then
summed for each volunteer; the scores for each question are
available from the corresponding author.
Stata 10.0 (Statacorp, College Station, TX) was used for all
analyses. Differences in activity levels between age groups were
assessed by using linear or logistic regression, depending on the
nature of the variable. Spearman correlations were used to de-
termine the ability of the questionnaires to rank physical activity
and EE summary variables. Those values most directly com-
parable were used for this analysis; therefore, PAEEQwas cor-
related with PAEEDLWand MVPAQcorrelated with MVPAAcc
for the CPAQ, YPAQ, and SWAPAQ. However, the lifestyle
score from the CHASE questionnaire was correlated with both
CORDER ET AL
Student’s t tests were used to determine whether significant
differences were present between PAEEQand PAEEDLWand
then MVPAQand MVPAAcc.A modified Bland-Altman method
(31) was also used to assess the degree of agreement between
methods. The difference (estimation error) between predicted
and criterion values was calculated (predicted 2 criterion) and
plotted against the criterion for each questionnaire and age
group. The mean difference and direction of any systematic bias
were examined, and the extent of any heteroscedasticity was
determined by using the Breusch-Pagan/Cook-Weisburg test. In
case of any heteroscedasticity, the ratio limits of agreement were
calculated on a log scale (32).
The reliability of the CPAQ, YPAQ, and SWAPAQ to assess
PAEEQand MVPAQand the CHASE questionnaire to assess the
lifestyle score was determined by using intraclass correlation
coefficients (ICCs), derived by using one-factor analysis of vari-
ance.Test-retest reliability wasassessed to establish howmuch of
the total variation in questionnaire variables was between subject
variation, therefore determining the ability of one administration
of the questionnaire to reliably assess physical activity.
Descriptive characteristics and objectively measured physical
activity levels of the volunteers are displayed in Table 1; there
was a wide range of weights and BMIs between and within
groups. The sample was predominantly white; only 3 volunteers
were from another ethnic group. On the basis of accelerometry
with ?1952 cpm as the lower threshold of MVPA, the oldest
group was significantly less active than were the other groups
(all P , 0.01). However, when ?3000 cpm was used as the
lower threshold of MVPA, there were no significant differences
between minutes spent at MVPA in each age group. However,
PAEE per kilogram body weight differed significantly between
age groups. Spearman correlations between PAEE and MVPA
based on a threshold of ?1952 cpm were as follows: r ¼ 0.38
(P ¼ 0.05) for 4–5-y-olds, r ¼ 0.29 (P ¼ 0.16) for 12–13-y-olds,
and r ¼ 0.52 (P ¼ 0.01) for 16–17-y-olds. Spearman correla-
tions between PAEE and MVPA based on a threshold of ?3000
cpm were as follows: r ¼ 0.35 (P ¼ 0.07) for 4–5-y-olds, r ¼
0.38 (P ¼ 0.06) for 12–13-y-olds, and r ¼ 0.54 (P ¼ 0.006) for
16–17-y-olds. A summary of questionnaire-derived physical
activity data is shown in Table 2, by age group, as medians and
minimum and maximum values because of the skewed nature of
some of the data.
Spearman correlations of PAEEQ with PAEEDLW and of
MVPAQwith MVPAAcc(derived by using the higher threshold
of MVPA ?3000 cpm) are shown in Table 3. The strength and
significance of these correlations did not meaningfully differ
when the lower MVPA threshold was used (?1952 cpm). Only
MVPAQfrom the CPAQ and YPAQ (12–13-y-old group) was
significantly correlated with MVPAAcc.The lifestyle scores from
the CHASE questionnaire and PAEEQestimated from both ques-
tionnaires administered to the 16–17-y-old group (YPAQ and
SWAPAQ) were significantly correlated with PAEEDLW.
Modified Bland-Altman plots of the criterion (MVPAAcc),
based on the higher threshold of MVPA (?3000 cpm), plotted
against differences between MVPAQand MVPAAccare shown
in Figure 2. Estimation errors between the predicted and crite-
rion values are summarized in Table 4 and indicate both under-
and overreporting of MVPAQ, depending on the questionnaire,
age group, and threshold used to assess MVPA. When the
Participant characteristics by age group1
a: 4–5-y-olds b: 12–13-y-oldsc: 16–17-y-oldsP value for difference2
No. of subjects
Sex (% male)
PAEEDLW(kJ ? kg21? d21)
PAEEDLW(kJ ? kg21? d21)
Accelerometer (total counts/d)
Accelerometer wear time (d)
4.9 6 0.73
110.1 6 8.4
20.2 6 4.1
16.5 6 1.7
6535 6 1114
1.67 6 0.2
97.7 6 28.0
564.4 6 200.8
250.1 6 133.3
547,714 6 130,090
8.0 6 2.0
13.1 6 0.3
161.4 6 7.5
50.6 6 9.8
19.6 6 1.3
11,759 6 2227
1.86 6 0.3
83.0 6 29.6
547.2 6 195.7
272.0 6 148.9
495,900 6 136,806
8.5 6 2.7
17.1 6 0.6
169.5 6 8.8
63.3 6 9.7
22.0 6 2.5
a , b , c (P , 0.001)
a , b , c (P , 0.001)
a , b , c (P , 0.001)
a , b , c (P , 0.001)
a . c (P ¼ 0.02); b . c (P ¼ 0.003)
a , b, c (P , 0.001)
a , b (P ¼ 0.006)
a . b . c (all P , 0.01)
a, b . c (all P , 0.01)
a, b . c (P , 0.001)
12,058 6 2990
1.78 6 0.3
65.6 6 27.7
371.2 6 194.3
232.2 6 165.5
345,823 6 119,886
7.8 6 3.0
1NSD, no significant difference; NA, not applicable; REE, resting energy expenditure; DLW, doubly labeled water; TEEDLW, total energy expenditure
assessed by DLW; MVPAAcc1952, moderate and vigorous physical activity assessed by accelerometry using 1952 counts as the lower threshold of MVPA (27);
MVPAAcc3000, MVPA assessed by accelerometry using 3000 counts as the lower threshold of MVPA (26); PAEEDLW, physical activity energy expenditure
assessed by DLW; PAL, physical activity level.
2Differences between groups were calculated by using linear regression, except for overweight and sex (logistic regression). The Cole et al (33) threshold
was used to define overweight.
3Mean 6 SD (all such values).
4Median; range in parentheses (all such values).
VALIDITY OF 4 PAQs IN YOUTH
lower MVPA threshold (?1952 cpm) was used, the mean dif-
ference between MVPAQand MVPAAccwas nonsignificant for
all questionnaires, as indicated by the 95% CI (Table 4). The
SWAPAQ had the largest CIs, which indicated the least accurate
predictions of MVPAQ. Strong correlations were present in most
Bland-Altman plots (Figure 2), which indicated that the degree
of questionnaire error was dependent on activity level and man-
ifested as underreporting at higher activity levels for the CPAQ
and SWAPAQ. Additionally, MVPAQestimates from the YPAQ
(16–17-y group) showed heteroscedasticity, which indicated an
increased measurement error as activity level increased. The
ratio limits of agreement are also shown in Table 4. These limits
indicate that the mean bias ranged from an underestimation of
49% to an overestimation of 43%, and any individual estimate
may have differed by between 4.3 and 11.6 times the true value,
depending on the questionnaire and age group. When the higher
threshold of MVPA (?3000 cpm) was used, the mean difference
between MVPAQand MVPAAccwas nonsignificant, as indicated
by the 95% CI (Table 4) for the YPAQ (16–17-y-old group) and
the SWAPAQ. However, estimates from the CPAQ and YPAQ
(12–13-y-old group) differed significantly from the criterion.
The YPAQ (12–13-y-old group) had the largest CIs, which in-
dicated the greatest overestimation of MVPAQ. Strong correla-
tions were present in some Bland-Altman plots (Figure 2), which
indicated that the degree of questionnaire error was dependent
on activity level and was manifested as underreporting at higher
activity levels for the YPAQ (16–17-y-old group) and SWAPAQ.
Additionally, MVPAQestimates from the YPAQ (12–13-y-old
group) and the SWAPAQ showed heteroscedasticity, which in-
dicated increased measurement error as activity level increased.
The ratio limits of agreement are also shown in Table 4. These
limits indicate that the mean bias ranged from an overestimation
of 3% to 201%, and any individual estimate may have differed
by between 2.2 and 3.0 times the true value, depending on the
questionnaire and age group.
Modified Bland-Altman plots of the criterion (PAEEDLW) and
differences between PAEEQand PAEEDLWare shown in Figure 3.
Mean differences, correlation of the error plots, and ratio limits
of agreement are shown in Table 4. All questionnaires (PAEEQ)
underestimated PAEEDLW, and estimates from the YPAQ (16–
17-y-old group) differed significantly from PAEEDLW. All other
group-level estimates of PAEEQwere underestimates with non-
between PAEEDLWand the difference (PAEEQ2 PAEEDLW) in all
Bland-Altman plots, which indicated an increasing underestima-
tion of PAEEDLWby PAEEQat higher activity levels. The ratio
limits of agreement werecalculatedona log scale and showed that
these questionnaires underestimated PAEEQand had wide ratio
limits of agreement. Group-level PAEEQestimates from these
questionnaires were likely to underestimate PAEEDLWby between
56% and 24%, with any individual estimate differing by between
3.1 and 8.5 times the true value.
The test-retest reliability of all questionnaires is shown in
Table 5. The summary variables calculated from the YPAQ
showed high reliability between administrations. Reliability was
somewhat lower for MVPAQestimated from the CPAQ. The
CHASE questionnaire showed significant reliability for the
lifestyle score, but the SWAPAQ showed low reliability.
This study assessed the validity and reliability of 4 physical
activity questionnaires at assessing PAEE and MVPA (measured
by DLW and accelerometry, respectively) in young persons of
different ages. All questionnaires were valid at ranking either
PAEE or MVPA, depending on age group and questionnaire.
Some questionnaires were able to accurately assess group-level
PAEE and MVPA, but error was large for individual estimates. A
summary of the validity and reliability of all questionnaires is
presented in Table 6.
The CPAQ and YPAQ (12–13-y-olds) were valid at ranking
MVPA, and the YPAQ (16–17-y-old group) and SWAPAQ were
valid at ranking PAEE. The lifestyle score from the CHASE
Questionnaire summary variables1
Questionnaire summary variableCPAQ (n ¼ 27)
YPAQ, 12–13-y olds (n ¼ 25)
YPAQ, 16–17-y-olds (n ¼ 24)
SWAPAQ (n ¼ 24)
PAEEQ(kJ ? kg21? d21)
1All values are medians; ranges in parentheses. MVPAQ, moderate and vigorous physical activity assessed by questionnaire; PAEEQ, physical activity
energy expenditure assessed by questionnaire; CPAQ, Children’s Physical Activity Questionnaire; YPAQ, Youth Physical Activity Questionnaire; SWAPAQ,
Swedish Adolescent Physical Activity Questionnaire.
Criterion validity of questionnaire summary variables1
criterion variableCPAQ (n ¼ 27)
0.42 (P ¼ 0.04)
0.22 (P ¼ 0.28)
CHASE (n ¼ 25)
0.45 (P ¼ 0.02)
0.12 (P ¼ 0.57)
(n ¼ 25)
0.42 (P ¼ 0.04)
0.09 (P ¼ 0.67)
(n ¼ 24)
0.11 (P ¼ 0.61)
0.46 (P ¼ 0.03)
SWAPAQ (n ¼ 24)
0.23 (P ¼ 0.27)
0.40 (P ¼ 0.04)
Lifestyle score, PAEEDLW
Lifestyle score, MVPAAcc
1Values are Spearman correlation coefficients. MVPAQ, moderate and vigorous physical activity assessed by questionnaire (min/wk); PAEEQ, physical
activity energy expenditure assessed by questionnaire (kJ ? kg21? d21); CPAQ, Children’s Physical Activity Questionnaire; YPAQ, Youth Physical Activity
Questionnaire; SWAPAQ, Swedish Adolescent Physical Activity Questionnaire; CHASE, Child Heart and Health Study in England; MVPAAcc, MVPA
assessed by accelerometry (min/wk) using 3000 counts as the lower threshold (26); PAEEDLW, PAEE assessed by doubly labeled water (kJ ? kg21? d21).
CORDER ET AL
questionnaire was able to rank overall activity level in com-
parison with PAEE.
There was inconsistency in correlations between age groups
and between estimates from the same questionnaire and the 2
different criterion methods; validity of the CPAQ and YPAQ
differed by age group despite being similar questionnaires. For
the CPAQ, this may be explained by the proxy-reported nature of
this questionnaire. The difference in validity between the self-
FIGURE 2. Modified Bland-Altman plots for the sum of moderate and vigorous intensity physical activity assessed by questionnaire (MVPAQ) compared
with MVPA assessed by accelerometry (MVPAAcc) based on the higher threshold of MVPA (?3000 cpm) for each questionnaire and age group. Mean
differences (MVPAQ2 MVPAAcc) plotted against the mean of MVPAQand MVPAAccfor the Children’s Physical Activity Questionnaire (CPAQ) in the 4–5-y-
old group (A; n ¼ 27), the Youth Physical Activity Questionnaire (YPAQ) in the 12–13-y-old group (B; n ¼ 25), the YPAQ questionnaire in the 16–17-y-old
group (C; n ¼ 24), and the Swedish Adolescent Physical Activity Questionnaire (SWAPAQ) in the 16–17-y-old group (D; n ¼ 24).
Comparison of moderate and vigorous physical activity (MVPA) assessed by questionnaire with MVPA assessed by accelerometry and of physical activity
energy expenditure (PAEE) assessed by questionnaire with PAEE assessed by doubly labeled water (PAEEDLW; in kJ ? kg21? d21) from 3 questionnaires1
Mean bias 6 SD2
P value for
CPAQ (n ¼ 27)
YPAQ, 12–13-y-olds (n ¼ 25)
YPAQ, 16–17-y-olds (n ¼ 24)
SWAPAQ (n ¼ 24)
CPAQ (n ¼ 27)
YPAQ, 12–13-y-olds (n ¼ 25)
YPAQ, 16–17-y-olds (n ¼ 24)
SWAPAQ (n ¼ 24)
CPAQ (n ¼ 27)
YPAQ, 12–13-y-olds (n ¼ 25)
YPAQ, 16–17-y-olds (n ¼ 24)
SWAPAQ (n ¼ 24)
1MVPAAcc1952, moderate and vigorous physical activity assessed by accelerometry (min/wk) using 1952 counts as the lower threshold of MVPA;
MVPAAcc3000, MVPA assessed by accelerometry (min/wk) using 3000 counts as the lower threshold of MVPA; RMSE, root mean square error; LOA, limits
2Mean bias ¼ predicted 2 criterion.
3From the Bland-Altman plots in Figures 1 and 2.
4Derived by using the Breusch-Pagan/Cook-Weisberg test.
5Means 6 SDs calculated on a log scale (and then antilogged to give mean difference 3/O limits of agreement).
276.5 6 361.6
22.4 6 431.2
2110.8 6 324.6
400.9 6 1059.0
0.68 3/O 4.3
0.77 3/O 5.2
0.51 3/O 11.6
1.43 3/O 10.5
235.9 6 362.0
370.8 6 467.9
120.2 6 350.3
212.0 6 239.4
1.63 3/O 2.24
2.01 3/O 2.25
1.38 3/O 2.97
1.03 3/O 2.58
214.4 6 52.4
220.9 6 54.1
237.6 6 32.5
215.6 6 68.9
0.76 3/O 3.1
0.59 3/O 6.3
0.32 3/O 4.6
0.46 3/O 8.5
VALIDITY OF 4 PAQs IN YOUTH
reported YPAQ administered to the 2 different age groups is
likely to be due to a combination offactors, including differences
between the criterion methods, which measure different aspects
of physical activity. However, these discrepancies may be partly
due to differences in activity levels and activity profiles between
groups. Depending on the threshold used, the 12–13-y-olds
carried out significantly more MVPA than did the older group
(P , 0.001), and higher-intensity activities were more accurately
reported than were lighter-intensity activities (7). Age is also
likely to influence validity because older children provide more
accurate self-report data (14, 15). One could hypothesize that
this may especially affect light-intensity activities, which are
already more difficult to report accurately, but that may heavily
influence PAEE. The thresholds used to determine MVPA also
influenced the absolute validity of the questionnaires (Table 4).
Because there is no consensus on the most suitable cutoffs for
time spent at different intensity levels, the results based on 2
different MVPA intensity thresholds were included and were
essentially unchanged between intensity thresholds.
As described in Subjects and Methods, various assumptions
had to be made for the estimation of PAEEQbecause there is no
standardized reference for youth to parallel the 1 MET equiva-
lent of 3.5 mL ? kg21? min21for adults. Nonetheless, this error
is much lower than that which would have resulted if adult
standards were used (data not shown). Estimations of EE from
self-reports in youth generally use adult-derived standard energy
costs of specific activities (29). This is because there are no
comprehensive reference values specific to youth, despite pos-
sible differences between MET multiples for the same activities
in adults and children (34). However, it should be possible to use
adult-derived values to estimate EE in youth if adjusted for the
higher REE of children (30). These factors may unavoidably
FIGURE 3. Modified Bland-Altman plots for the sum of physical activity energy expenditure assessed by questionnaire (PAEEQ) compared with PAEE
assessed by doubly labeled water (PAEEDLW) for each questionnaire and age-group. Mean differences (PAEEQ2 PAEEDLW) plotted against the mean of
PAEEQand PAEEDLWfor the Children’s Physical Activity Questionnaire (CPAQ) in the 4–5-y-old group (A; n ¼ 27), the Youth Physical Activity
Questionnaire (YPAQ) in the 12–13-y-old group (B; n ¼ 25), the YPAQ in the 16–17-y-old group (n ¼ 24), and the Swedish Adolescent Physical
Activity Questionnaire (SWAPAQ) in the 16–17-y-old group (n ¼ 24).
Test-retest reliability of summary variables from 4 questionnaires across 3 age groups with the use of intraclass correlation
Questionnaire summary variableCPAQCHASE
No. of pairs
1MVPAQ, moderate and vigorous physical activity assessed by questionnaire (min/wk); PAEEQ, physical activity
energy expenditure assessed by questionnaire (kJ ? kg21? d21); CPAQ, Children’s Physical Activity Questionnaire; YPAQ,
Youth Physical Activity Questionnaire; SWAPAQ, Swedish Adolescent Physical Activity Questionnaire; CHASE, Child
Heart and Health Study in England.
2P , 0.05.
3P , 0.001.
CORDER ET AL
contribute to the limited validity of self-reports for the estima-
tion of EE in children in this and other studies (35).
Previous studies examining the validity of the CLASS ques-
tionnaire, on which the YPAQ and CPAQ were based, suggested
limited validity when compared with accelerometer data when
parentally reported for 5–6 y-olds (r ¼ 20.06–0.05) and when
self-reported for 10–12-y-olds (r ¼ 20.04 to r ¼ 0.06) (10). The
higher correlation coefficients observed in this study may have
been partly due to our older sample of adolescents, because
others have suggested an age threshold of 10 y as the cutoff for
providing acceptable questionnaire data (14). The CLASS
questionnaire assessed a ‘‘usual’’ week of activity, but the YPAQ
and CPAQ assessed the past week. For validity purposes, data
from the questionnaire and criterion are more directly compa-
rable when referring to the same time period (ie, past week),
and it may also be easier to report physical activity from the
previous week than from a ‘‘usual’’ week. The original Swedish
SWAPAQ showed better validity than did the current study: r ¼
0.51 for total accelerometry-assessed physical activity (28). This
difference should not be due to age, because these British ado-
lescents are slightly older than those originally assessed but may
be explained by differences in sample sizes and activity levels
between groups. The Swedish adolescents (20) were ’15%
more active than were those who participated in the present
study. There are no published validation studies of the CHASE
questionnaire with which to compare.
The small sample size and sex differences in each age group
were limitations of this study but were unavoidable, primarily
because of the high cost of the DLW method and difficulties with
recruitment. However, because these questionnaires were ad-
ministered with no verbal instructions from the investigators,
these questionnaires if they were to be used in a large epide-
miologic study in a similar population. It was deemed necessary
to use different questionnaires in the 3 age groups because
cognitive abilities and activity profiles of different age groups
the CPAQ and YPAQ differ by age group, despite being very
similar questionnaires. These differences indicate that data from
directly comparable, the evaluation of 4 age-specific ques-
tionnaires should be useful to those carrying out research on
physical activity in different age groups of youth. Another limi-
tation is that the 3 methods applied in this study—accelerometry,
the simultaneous assessment of questionnairevalidity to measure
both total volume of physical activity (PAEE) and physical acti-
vity intensity and duration (MVPA). Additionally, use of only
the summary variables considered here (MVPAQand PAEEQ)
somewhat limited the scope of these questionnaires, which do
hold more comprehensive information on specific physical ac-
tivities, including sedentary behavior. An additional subjective
method of measurement, such as a detailed interview, diary, or
The YPAQ was the most reliable questionnaire, which showed
significant intraclass correlations for PAEEQand MVPAQ. The
lifestyle score (CHASE) and MVPAQfrom the CPAQ also showed
acceptable reliability. The SWAPAQ had poor reliability, despite
being completed by the oldest group; the reliability of this ques-
tionnairewas not assessed inthe previous studyand thus cannot be
compared (28). The structured nature of the YPAQ, including de-
questionnaire could account for this finding. The proxy-report na-
ture of the CPAQ could explain the lower reliability because recall
of the children’s physical activity is difficult for adults (36). Proxy
reporting may lead to limitations in questionnaire validity, which
monitor a child and may also be in charge of other children.
For reliability purposes, the 2 administrations of all ques-
tionnaires were answered in reference to the same 7 d. The
advantage of this is that the differences between the 2 admin-
istrations should only consist of reporting error, with no variation
due to real differences in physical activity levels over time. The
subsequent limitation of this method is the recall of a period
further back in time; consequently, reporting error is likely to be
greater with the second administration. Additionally, the short
time interval between administrations may mean that the vol-
unteers remembered their answers from the first administration.
Absolute PAEE and MVPA estimated from these self-reports
were not valid on an individual level in young persons, although
some questionnaires may rank individuals accurately for MVPA
(CPAQ and YPAQ in 12–13-y-olds), PAEE (YPAQ in 16–17-y-
oldsand SWAPAQ),and overallphysicalactivity(CHASE).Age,
Summary of the validity and reliability of all 4 questionnaires1
Able to rank
1MVPA, moderate and vigorous physical activity; PAEE, physical activity energy expenditure; CPAQ, Children’s Physical Activity Questionnaire;
YPAQ, Youth Physical Activity Questionnaire; SWAPAQ, Swedish Adolescent Physical Activity Questionnaire; CHASE, Child Heart and Health Study in
England; MVPAAcc1952, assessed by accelerometry (min/wk) using 1952 counts as the lower threshold of MVPA (27); MVPAAcc3000, assessed by accel-
erometry (min/wk) using 3000 counts as the lower threshold of MVPA (26).
2Compared with PAEE (kJ ? kg21? d21).
VALIDITY OF 4 PAQs IN YOUTH
theoutcomevariableofinterest,and whether individualorgroup-
level estimates are necessary will influence the best choice of
measurement method when assessing physical activity in youth.
We thank Alice Tompson, Kate Westgate, Charlotte Ridgway, Paul Rob-
for assistance with testing. We are also grateful to Mark Hennings for devel-
ments on the manuscript.
The authors’ responsibilities were as follows—KC: collected all of the
data, conducted the data analyses, and drafted the manuscript; EMFvS,
PW, and UE: compiled the questionnaires; NJWand UE: supervised the data
collection; and AW: analyzed and interpreted the doubly labeled water data.
All of the authors provided critical input on the data analyses and on all ver-
sions of the manuscript and approved the final version. None of the authors
had any conflicts of interest.
1. Rowlands AV, Ingledew D, Eston R. The effect of type of physical ac-
tivity measure on the relationship between body fatness and habitual
physical activity in children: a meta-analysis. Ann Hum Biol 2000;
2. Mahabir S, Baer D, Giffen C, Clevidence B, Campbell W, Taylor P.
Comparison of energy expenditure estimates from 4 physical activity
questionnaires with doubly labeled water estimates in postmenopausal
women. Am J Clin Nutr 2006;84:230–6.
3. Ainslie P, Reilly T, Westerterp K. Estimating human energy expenditure:
a review of techniques with particular reference to doubly labelled
water. Sports Med 2003;33:683–98.
4. Ness A, Leary S, Mattocks C, et al. Objectively measured physical
activity and fat mass in a large cohort of children. PLoS Med
5. Riddoch CJ, Bo Andersen L, Wedderkopp N, et al. Physical activity
levels and patterns of 9- and 15-yr-old European children. Med Sci
Sports Exerc 2004;36:86–92.
6. Corder K, Ekelund U, Steele RM, Wareham NJ, Brage S. Assessment of
physical activity in youth. J Appl Physiol 2008;105:977–987.
7. Slinde F, Arvidsson D, Sjoberg A, Rossander-Hulthen L. Minnesota
leisure time activity questionnaire and doubly labeled water in adoles-
cents. Med Sci Sports Exerc 2003;35:1923–8.
8. Arvidsson D, Slinde F, Hulthen L. Physical activity questionnaire for
adolescents validated against doubly labelled water. Eur J Clin Nutr
9. Philippaerts RM, Matton L, Wijndaele K, Balduck A, De Bourdeaudhuij
I, Lefevre J. Validity of a physical activity computer questionnaire in 12-
to 18-year-old boys and girls. Int J Sports Med 2006;27:131–6.
10. Telford A, Salmon J, Jolley D, Crawford D. Reliability and validity of
physical activity questionnaires for children: the children’s leisure ac-
tivities study survey (CLASS). Ped Exerc Sci 2004;16(1):64–78.
11. Janz K, Lutuchy E, Wenthe P, Levy S. Measuring activity in children
and adolescents using self-report: PAQ-C and PAQ-A. Med Sci Sports
12. Welk G, Corbin C, Dale D. Measurement issues in the assessment of
physical activity in children. Res Q Exerc Sport 2000;71:S59–73.
13. Corder K, Brage S, Ekelund U. Accelerometers-pedometers: method-
ology and clinical application. Curr Opin Clin Nutr Metab Care 2007;
14. Sallis JF. Self-report measures of children’s physical activity. J Sch
15. Baranowski T, Dworkin R, Cieslik C, et al. Reliability and validity of
self-report of aerobic activity: Family Health Project. Res Q Exerc Sport
16. Oliver M, Schofield G, Kolt G. Physical activity in preschoolers. un-
derstanding prevalence and measurement issues. Sports Med 2007;37:
17. Ekelund U, Sjo ¨stro ¨m M, Yngve A, et al. Physical activity assessed by
activity monitor and doubly labelled water in children. Med Sci Sports
18. Hoffman D, Sawaya A, Coward W, et al. Energy expenditure of stunted
and nonstunted boys and girls living in the shantytowns of Sao Paulo,
Brazil. Am J Clin Nutr 2000;72:1025–31.
19. Cole T, Coward W. Precision and accuracy of doubly labeled water
energy expenditure by multipoint and two-point methods. Am J Physiol
20. Schoeller D, van Santen E. Measurement of energy expenditure in hu-
mans by doubly labeled water method. J Appl Physiol 1982;53:955–9.
21. Schoeller D, Ravussin E, Schutz Y, Acheson K, Baertschi P, Jequier E.
Energy expenditure by doubly labeled water: validation in humans and
proposed calculation. Am J Physiol 1986;250:R823–30.
22. Schoeller D, Leitch C, Brown C. Doubly labeled water method: in vivo
oxygen and hydrogen isotope fractionation. Am J Physiol 1986;251:
23. Elia M, Livesey G. Theory and validity of indirect calorimetry during
net lipid synthesis. Am J Clin Nutr 1988;47:591–607.
24. de Weir JB. New methods for calculating metabolic rate with special ref-
erence to protein metabolism. J Physiol 1949;109:1–9.
25. Ekelund U, A˚man J, Westerterp K. Is the Arte index a valid indicator of
free-living physical activity in adolescents? Obes Res 2003;11:793–801.
26. Treuth M, Schmitz K, Catellier D, et al. Defining thresholds for activity
intensities in adolescent girls. Med Sci Sports Exerc 2004;36:1259–66.
27. Freedson P, Melanson E, Sirard J. Calibration of the Computer Science and
Applications, Inc. accelerometer. Med Sci Sports Exerc 1998;30:777–81.
28. Ekelund U, Neovius M, Linne Y, Rossner S. The criterion validity of
a last 7-day physical activity questionnaire (SAPAQ) for use in ado-
lescents with a wide variation in body fat: the Stockholm Weight De-
velopment Study. Int J Obes 2006;30:1019–21.
29. Ainsworth B, Haskell W, Whitt M, et al. Compendium of physical ac-
tivities: an update of activity codes and MET intensities. Med Sci Sports
30. Harrell J, McMurray R, Baggett C, Pennell M, Pearce P, Bangdiwala S.
Energy costs of physical activities in children and adolescents. Med Sci
Sports Exerc 2005;37:329–36.
31. Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet 1986;8:307–10.
32. Atkinson G, Nevill A. Statistical methods for assessing measurement
error (reliability) in variables relevant to sports medicine. Sports Med
33. Cole T, Freeman J, Preece M. Body mass index reference curves for the
UK, 1990. Arch Dis Child 1995;73:25–9.
34. Spadano J, Must A, Bandini L, Dallal G, Dietz W. Energy cost of
physical activities in 12-y-old girls: MET values and the influence of
body weight. Int J Obes Relat Metab Disord 2003;27:1528–33.
35. Sallis JF, Buono MJ, Freedson PS. Bias in estimating caloric expenditure
from physical activity in children. Implications for epidemiological
studies. Sports Med 1991;11:203–9.
36. Pate R. Physical activity assessment in children and adolescents. Crit
Rev Food Sci Nutr 1993;33:321–6.
CORDER ET AL