Validity of Four Short Physical Activity

Questionnaires in Middle-Aged Persons

URS MADER1, BRIAN W. MARTIN1, YVES SCHUTZ 2, and BERNARD MARTI1
1
Federal Institute of Sports, Magglingen, SWITZERLAND; and 2Institute of Physiology, University of Lausanne,
Lausanne, SWITZERLAND

ABSTRACT
MADER, U., B. W. MARTIN, Y. SCHUTZ, and B. MARTI. Validity of Four Short Physical Activity Questionnaires in Middle-Aged
Persons. Med. Sci. Sports Exerc.,Vol. 38, No. 7, pp. 12551266, 2006. Purpose: Self-administered questionnaires continue to be the
most widely used type of physical activity assessment in epidemiological studies. However, testretest reliability and validity of
physical activity questionnaires have to be determined. In this study, three short physical activity questionnaires already used in
Switzerland and the International Physical Activity Questionnaire (IPAQ) were validated. Methods: Test retest reliability was
assessed by repeated administration of all questionnaires within 3 wk in 178 volunteers (77 women, 46.1 T 14.8 yr; 101 men 46.8 T
13.2 yr). Validity of categorical and continuous data was studied in a subsample of 35 persons in relation to 7-d accelerometer
readings, percent body fat, and cardiorespiratory fitness. Results: Reliability was fair to good with a Spearman correlation coefficient
range of 0.430.68 for measures of continuous data and moderate to fair with kappa values between 0.32 and 0.46 for dichotomous
measures active/inactive. Total physical activity reported in the IPAQ and the Office in Motion Questionnaire (OIMQ) correlated with
accelerometry readings (r = 0.39 and 0.44, respectively). In contrast, correlations of self-reported physical data with percent body fat and
cardiorespiratory fitness were low (r = j0.26 to 0.29). Participants categorized as active by the Swiss HEPA Survey 1999 instrument
(HEPA99) accumulated significantly more days of the recommended physical activities than their inactive counterparts (4.4 and
2.7 dIwkj1, respectively, P G 0.05). However, compared with accelerometer data, vigorous physical activities were overreported in
investigated questionnaires. Conclusion: Collecting valid data on physical activity remains a challenging issue for questionnaire
surveys. The IPAQ and the three other questionnaires are characterized to inform decisions about their appropriate use. Key Words:
RELIABILITY, MTI, ACCELEROMETER, OVERREPORTING, SURVEILLANCE

T
he extraordinary importance of regular physical
activity and exercise for disease prevention and
health promotion is widely accepted (16,25). In
epidemiological and intervention studies, self-completed
or interviewer-administered questionnaires are widely used
to assess physical activity. However, there is a broad diver-
sity of questionnaires varying in complexity and types of
physical activities included. Questionnaire length ranges
from a few questions to detailed inventories of adapted ques-
tionnaires, which are often developed for specific socio-
cultural settings.
In general, longer and more detailed questionnaires are
expected to be more valid than their shorter counterparts.
Address for correspondence: Urs Mader, Federal Institute of Sports, 2532
Magglingen, Switzerland; E-mail: urs.maeder@baspo.admin.ch.
Submitted for publication August 2004.
Accepted for publication February 2006.
0915-9131/06/3807-1255/0
MEDICINE & SCIENCE IN SPORTS & EXERCISE
Copyright 2006 by the American College of Sports Medicine
DOI: 10.1249/01.mss.0000227310.18902.28
1255
However, there are indications that the validity of shorter
questionnaires can be acceptable (2,5), and short physical
activity questionnaires are more likely to be included in
extensive surveys covering different aspects of lifestyle
and to improve compliance of respondents in interven-
tion studies.
In Switzerland, several self- or interviewer-administered
short questionnaires have been applied to measure physical
activity (13,14). In the first representative assessment
among the Swiss population (Swiss Health Survey 1997,
SHS97), questions on physical activities inducing sweat
episodes during leisure time were used (18,26). Washburn
et al. (26) showed that the reported number of sweat
episodes correlated with triceps skinfold (r = j0.17) and
resting heart rate in young college students (r = j0.29). To
assess physical activity in a smaller, but representative
sample, a short questionnaire was developed according to
the stages of changes described by Marcus and Simkin (12)
for Swiss HEPA Survey in 1999 (14), the first of a series of
representative population surveys on health-enhancing
physical activity (HEPA) with reference to current recom-
mendations in Switzerland (HEPA99; see Appendix A).
This instrument has not yet been validated. For monitoring
the effects of physical activity interventions in occupa-
tional settings, a shortened version of the physical activity

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
frequency questionnaire developed by Bernstein et al. (4) following day and expected back within 1 wk. As soon as it
was applied (Office In Motion Questionnaire, OIMQ; see was returned to the study center, the procedure was first
Appendix B). The total energy expenditure estimated by repeated with the telephone and then with the written
the longer and original version of the latter questionnaire questionnaires in the same order. The reliability study was
correlated well with estimates using heart rate (r = 0.79) (4). conducted over a 14- to 21-d period. The time lag between
At the international level, the World Health Organization the two telephone interviews was chosen to reduce contam-
(WHO), the Centers for Disease Control and Prevention ination by recall of answers from the first occasion.
(CDC), and other partners have developed the International Validation procedure. At the end of the first tele-
Short Physical Activity Questionnaire (IPAQ; see www. phone interview, the volunteers were asked to participate in
ipaq.ki.se), which has already been validated in different the validity part of the study. After a positive answer, they
sociocultural settings (5,21). Reported activities in the were informed about the validation procedure in an
short telephone interview, aimed at a usual week, were additional letter and received the self-administered OIMQ,
reliable (r = 0.83) and correlated with accelerometer the Physical Activity Readiness Questionnaire (PAR-Q)
measurements (r = 0.52) in the U.S. population (5). This (25), and an invitation for the laboratory tests (Fig. 1).
is the first time these four short versions of questionnaires Within 1 wk after the questionnaires and the PAR-Q had
have been validated in a comparable middle-aged popula- been returned to the study center, the volunteers of the
tion sample. This study investigates which of the short validity part visited the laboratory at 8:15 a.m. in a fasted
questionnaires yields the most significant data on physical state. The PAR-Q was used to identify potential health
activity behavior in larger population groups and whether
the IPAQ is a reliable instrument compared with Swiss
national questionnaires.
Various objective methods to monitor physical activity
are available and may be used to validate questionnaire
data on physical activity (7). Registration of body movements
by accelerometry has already shown to provide valid data
about physical activity. Thus, it can be an objective cri-
terion by which questionnaires can be compared (7,10,15).
The purpose of this study was to determine 1) the repro-
ducibility and 2) the relative and absolute validity of three
short physical activity questionnaires as well as the short
version of IPAQ, by comparing them with uniaxial accele-
rometry measures.

METHODS
Volunteers. Subjects were selected randomly from the
list of residents in the city of Biel/Bienne and 14 neighbor
villages in the Seeland region of Switzerland. The volun-
teers had to be German-speaking and aged 1575 yr. Data of
individuals who reported not being able to walk more than
200 ms on their own were excluded from further analyses.
Reliability procedure. In a letter, selected individuals
were informed about the study and a telephone contact was
announced. Up to 10 attempts were made to reach each
person by telephone at different times and days. Individuals
were invited to participate in the reliability study, and vol-
unteers answered three telephone interview questionnaires
in three different consecutive orders assigned randomly. At
the end the telephone interview, the volunteers were asked
to participate in the validity part of the study. If they refused,
they remained in the reliability part. Interested volunteers
followed the validity procedure (Fig. 1). A selected person
who had not been reached or refused to participate on gen-
eral grounds was not replaced.
Physical education students carried out all telephone
interviews, using a standardized computer-aided approach.
FIGURE 1Schematic of the study design. SHS97, Swiss Health Survey
OIMQ, the only written self-administered questionnaire, 1997; HEPA99, Swiss HEPA Survey 1999; IPAQ, International Physical
was sent to the volunteers of the reliability study on the Activity Questionnaire; IOMQ, Office In Motion Questionnaire.

1256 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
hazards, and if necessary, a physician contacted volunteers
for risk management (24). Before laboratory tests, volunteers
had to give their written informed consent and be approved
by the institutional review board. Body mass index and
body composition were determined for every subject. They
answered the three telephone questionnaires on a phone in
the laboratory to assure that the conditions were similar to
the previous interviews and then completed the self-
administered OIMQ (Fig. 1). The data were used for the
testretest reliability part of the study by comparing them
with the first physical activity reports. To determine the
fitness level, expressed as estimated maximal oxygen
consumption (VO2max), subjects performed the 2000-m
walking test (17). Finally, the subjects were carefully
instructed on how to use the accelerometer during the
measurement week. They were told to put the accelerom-
eter on in the morning after getting up and dressing and to
wear it until they prepared themselves to go to bed. As the
accelerometer was not waterproof, it had to be removed
before coming in contact with water. Subjects were asked
to note time of sleeping, swimming, and bicycling in a
small diary. After the measurement period, they answered
all questionnaires in the same order again. Those data were
compared with accelerometry measures to determine
criterion validity of the questionnaires.
Short physical activity questionnaires. Table 1
describes the questionnaires used in this study, specifying
mode, activity type and domains, indicator type, and time
periods covered. The diversity of questions resulted in various
variables that all Bquantify[ performed physical activities. In
the SHS97, only the number of vigorous physical activity
episodes per week could be determined because additional
items referring to more recent recommendations were not
yet included. A dichotomous categorization activeinactive
was generated by using categorical data of HEPA99 and
continuous records of IPAQ, based on the current recom-
mendations for health-enhancing physical activity and
cardiorespiratory fitness (19). Volunteers were classified
as active if they reported either at least 30 min of moderate-
intensity activities 5 dIwkj1 or at least 20 min of vigorous
exercise three times per week. All others were classified as
inactive. The IPAQ was more restricting, with the items
aiming on physical activity bouts with a minimal duration of
10 min at both intensity levels. This limitation is based on
the criteria for health-enhancing physical activity (8). Total
physical activity assessed by questionnaires recording
continuous data was expressed as MET-minutes for IPAQ
and OIMQ.
Percent body fat. Percent body fat of volunteers
attending the validation study was measured in the fasted
state with an empty bladder. A body impedance analyzer
measuring at a frequency of 50 kHz, and the man-
ufacturers software for data analyses (BIA 2000-S, Data
Input, Frankfurt, Germany) were used.
Accelerometers. The Actigraph (Model AM7164,
formerly Computer Science and Applications (CSA), now
Manufacturing Technology Inc. (MTI), Fort Walton Beach,
FL) measures changes in body position. In earlier studies,
MTI accelerometers have been validated and calibrated
(6,7,15). The MTI accelerometer was used continuously
during waking hours of a 7-d measurement period. It was
rigidly fixed to the hip with a belt. The accelerometer takes
40 measurements per second and integrates acceleration
signals continuously minute by minute. MTI recordings
were analyzed with a specifically developed computer
program. To compensate for the accelerometer not being
worn during swimming and for its inherent limitations to
record bicycling activities, MTI recordings were adjusted
for moderate activity for the duration of swimming and
bicycling reported in the diary.
Heart rate monitoring. The Polar Vantage NV device
(Polar Electro, Kempele, Finland) was used to obtain minute-
by-minute heart rate readings, which were downloaded to a
computer for analysis. This system has been shown to be

TABLE 1. Description of physical activity questionnaire measures.

Time
Questionnaire Mode Activity Type and Domains Indicator Type Units Periods
Swiss Health Survey 1997 (SHS97) (26) Telephone interview Sweat episodes, recreational Frequency dIwkj1 Usual week
Health Enhancing Physical Telephone interview Moderate intensity, number of Inactive/active* Combined variable Usual week
Activity Survey 1999 days with 3  10 min
(HEPA99); (14; Appendix A)
Vigorous intensity, number of
days with 20 min
International Physical Activity Telephone interview Vigorous intensity in general, Frequency duration METIminIwkj1 Usual week
Questionnaire, short version, Q 10 min
draft 7 (IPAQ), see Craig et al.
(5; www.ipaq.ki.se)
Moderate intensity in general,
Q 10 min
Walking, Q 10 min
Sitting, Q 10 min
Moderate and vigorous intensity Inactive/active* Combined variable
Office In Motion Questionnaire Self-administered List of 7 household, 8 transport, Frequency duration METIminIwkj1 Last week
(OIMQ), short version of 18 recreational activities,
Bernstein et al. (4) (Appendix B) and table for 3 additional
unlisted activities
Number of items were 2, 8 13, 9, and 33, for SHS97, HEPA99 (variable number of items due to questionnaire filter), IPAQ, and OIMQ, respectively.
* Thresholds of at least 30 min of moderate activity on 5 dIwkj1 or 20 min of vigorous activity on 3 dIwkj1 were used to assign respondents to the two categories of inactive
and active.

VALIDITY OF PHYSICAL ACTIVITY QUESTIONNAIRES Medicine & Science in Sports & Exercised 1257

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
TABLE 2. Descriptive characteristics of the validation study participants.
Subjects Women
N 13
Age (yr) 49.7 T 14.0
Body fat (%) 24.3 T 6.2
Estimated VO2max (mLIkgj1Iminj1) 29.0 T 7.1
VO2max, maximal oxygen consumption.
valid in both laboratory and field settings relative to
electrocardiographic measurements of heart rate (11).
The heart rate monitor consists of a transmitter connected
with electrodes to the volunteer_s chest and a lightweight
receiver worn at the wrist. These monitors were worn
during the 2000-m walking test, as the heart rate was
needed to estimate VO2max (17).
Calculations and statistical analysis. Reliability of
categorical data was tested with kappa statistics, whereas
continuous data were tested by the Spearman correlation
method. Kappa values and Spearman correlations smaller
than 0.40 were interpreted as poor, values between 0.41
and 0.75 as fair to good, and values above 0.75 as excel-
lent agreement.
The validity of questionnaires recording categorical data
was evaluated by comparing the measured, more objective
MTI data for two categories with the MannWhitney U-Test.
Spearman correlations were used to evaluate the criterion
validity of the questionnaires, yielding continuous data by
comparing them with MTI measurements. Therefore, total
physical activity (METIminIwkj1) that was calculated from
self-reported activities in the questionnaires was compared
with MTI data expressed as total counts per registered time
(countsIminj1). The compendium of physical activities was
used to assign METs for each of the reported activities (1).
The multiplication of duration, frequency per week, and
MET intensity produced a weighted estimate of total physical
activity per week (METIminIwkj1). To validate the amount of
moderate and vigorous activities recorded by questionnaires
(min), cut-points of Swartz et al. (23) were used to identify
time spent in moderate (>574 and G4945 countsIminj1) and
vigorous (>4945 countsIminj1) physical activities regis-
tered by the accelerometer. An accelerometer cut-point of
G 100 countsIminj1 was used to determine the time spent
sitting (5).
Absolute amounts of reported activities in minutes per
Men week such as sitting (only IPAQ) and moderate (including
22 walking reported in the IPAQ) and vigorous activities of
57.6 T 14.4
21.9 T 5.3
IPAQ and OIMQ were compared with registered time at
30.2 T 10.3 the respective MTI levels, using the nonparametric analysis
of Wilcoxon. Accelerometry recoding was scanned for
bouts of moderate and vigorous physical activity (MVPA)
with an average counts per minute between the respective
cut-points. Due to the low participation in vigorous activities
among the subjects (4.3 T 4.7 minIdj1), moderate and
vigorous activities were combined. No 1-min average value
below the cut-points was accepted in bouts up to 3 min. One
average value below the cut-points was tolerated from 4 to
9 min, and two for bouts of 10 min or longer.
Finally, Spearman correlations between continuous ques-
tionnaire data were calculated to assess interquestionnaire
validity. All data were analyzed using SPSS 13.0 (SPSS Inc,
Chicago, IL).
RESULTS
Volunteers. Eight hundred eighty-six subjects were
selected from the list of residents; 636 persons were reached
by phone calls; 323 agreed to participate in the reliability part
of the study; and 210 persons answered four questionnaires
twice. Data of 178 (77 women, 46.1 T 14.8 yr; 101 men, 46.8 T
13.2 yr) completely answered questionnaires were used to
assess reliability; 35 (13 women and 22 men) of these 178
volunteers participated in the validation study (Table 2).
Assessment of reliability. Table 3 provides the
kappa coefficients measuring agreement of categorical
data recorded with HEPA99 and IPAQ during the
reliability study for all volunteers in the nonvalidation
and the validation samples. Reproducibility of the dichot-
omous categorization active/inactive of all participants was
fair to moderate. Kappa correlations for these categories
were higher in HEPA99, where the time thresholds of the
physical activity recommendations were included in the
questions. To determine reliability of continuous data,
Spearman correlations were used for SHS97 (leisure time
sweat episodes), IPAQ, and OIMQ (Table 3). Moderate
agreement between the two repeated measurements in all

TABLE 3. Reliability of short physical activity questionnaires among all participants in the nonvalidation and validation samples.

Questionnaire Component Total (kappa) NVS (kappa) VS (kappa) Total (r) NVS (r) VS (r)
j1
SHS97 Sweat episodes, leisure time (dIwk ) 0.63 (N = 178) 0.65 (N = 143) 0.49 (N = 35)
HEPA99 Dichotomous measure active/inactive 0.46 (N = 178) 0.47 (N = 143) 0.42 (N = 35)
IPAQ Sitting 0.60 (N = 178) 0.59 (N = 143) 0.69 (N = 35)
Moderate activities 0.50 (N = 178) 0.43 (N = 143) 0.58 (N = 35)
Walking 0.48 (N = 178) 0.48 (N = 143) 0.54 (N = 35)
Vigorous activities 0.43 (N = 178) 0.43 (N = 143) 0.39 (N = 35)
Total activities (no sitting) (METIminIwkj1) 0.54 (N = 178) 0.56 (N = 143) 0.62 (N = 35)
Dichotomous measure active/inactive 0.32 (N = 178) 0.34 (N = 143) 0.20# (N = 35)
OIMQ Moderate activities 0.70 (N = 188) 0.66 (N = 153) 0.71 (N = 35)
Vigorous activities 0.69 (N = 188) 0.61 (N = 153) 0.69 (N = 35)
Total activities (METIminIwkj1) 0.68 (N = 188) 0.65 (N = 153) 0.83 (N = 35)
r, Spearman correlation coefficient; NVS, nonvalidation sample; NS, validation sample; SHS 97, Swiss Health Survey 1997; HEPA99, Swiss Health Enhancing Physical Activity Survey 1999;
IPAQ, International Physical Activity Questionnaire; OIMQ, Office In Motion Questionnaire. All values were significant (P G 0.05), except the kappa value for dichotomous measure in the
IPAQ (#).

1258 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
respondents was observed for the questions of IPAQ.
Respondents had to report daily average duration for
walking and moderate and vigorous activities including
work, transport, and leisure time (r = 0.50, 0.48, and 0.43,
respectively). Repeated recall of sweat episodes (SHS97),
sitting (IPAQ), and several listed activities (OIMQ)
resulted in clearly higher reliability coefficients (r = 0.60
0.70). Reliability coefficients of total physical activities
summarized as MET-minutes per week reported in IPAQ
and OIMQ were 0.54 and 0.68, respectively. All reported
coefficients were statistically significant.
The repeatability of reported activities among the non-
validation and validation samples was similar except for
moderate activities (r = 0.43 vs 0.58), sitting in the IPAQ (r =
0.59 vs 0.69), and total activities in OIMQ (r = 0.65 vs 0.83,
respectively). For the sweat episodes, the coefficient for the
nonvalidation group was higher compared with the validation
sample (r = 0.65 vs 0.49, respectively).
Criterion validity for categorical data. The
subjects participating in the validation study wore the
accelerometers during 15.4 T 1 hIdj1 for 1 wk. Ten sub-
jects reported swimming and bicycling activities of 36 T
53 minIwkj1 in the diary. Accelerometry data were ad-
justed during these periods. Table 4 shows the number of
respondents who were classified into active and inactive
groups based on questionnaire data. Validity of collected
dichotomous data was tested by comparing the number of
days with at least 3 MTI recorded activity bouts of 10 min
or longer at moderate and vigorous levels between groups.
The difference was statistically significant for HEPA99,
whereas the difference between the groups generated by
IPAQ data just failed to reach the level of statistical
significance (Table 4); 69% of participants were classified
consistently by HEPA99 and MTI data, and the
corresponding value for IPAQ and MTI recordings was
57%. However, most activity bouts of MVPA recorded
by accelerometry among the whole validation subgroup
lasted from 1 to 3 min (43.2 per day). The numbers of
longer bouts were 21.9 and 2.7 per day for 49 and Q
10 min, respectively.
Interquestionnaire validity of continuous data.
Table 5 shows concurrent validity coefficients between
questionnaires recording continuous data. Interquestionnaire
validity showed minor agreement of OIMQ with continuous
data on sweat episodes (r = 0.34) of SHS97. The correlation
coefficient calculated between OIMQ and IPAQ for total
TABLE 4. Number of active days (NOD) according to MTI data by activity classes
corresponding to HEPA99 and IPAQ. On active days, three bouts of moderate and
vigorous activities of 10 min or more were registered.
HEPA99 IPAQ
N NOD (MTI) N NOD (MTI)
Activity classes P = 0.02 P = 0.05
Active 11 2.7 T 1.6 13 3.5 T 1.8
Inactive 24 4.4 T 2.0* 22 4.1 T 2.2
HEPA99, Swiss Health Enhancing Physical Activity Survey; IPAQ, International
Physical Activity Questionnaire. P values are for differences between classes.
VALIDITY OF PHYSICAL ACTIVITY QUESTIONNAIRES
Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
TABLE 5. Interquestionnaire validity, determined by comparing short questionnaires
resulting in continuous data.
SHS97 IPAQ
Sweat Episodes Total Activities
Questionnaire N Leisure Time (dIwkj1) (METIminIwkj1)
IPAQ Total activities 35 0.34*
(METIminIwkj1)
OIMQ Total activities 35 0.41* 0.56*
(METIminIwkj1)
SHS 97, Swiss Health Survey 1997; IPAQ, International Physical Activity
Questionnaire. * Correlations were significant (P G 0.05).
activities (METIminIwkj1) was clearly higher (r = 0.56).
Classification of the volunteers as active or inactive was
consistent for HEPA99 and IPAQ in 80% of subjects.
Criterion validity for duration of activities. Validity
of duration of activities reported in the questionnaires was
determined by comparing the reported time of physical
activities at different intensities with the amount of time at the
corresponding level registered by MTI. The results for mod-
erate activities, sitting, moderate activities including walking,
as well as vigorous activities are presented in Table 6. All
means of activity measurements by IPAQ and OIMQ differed
significantly from registered MTI data at corresponding
intensity level except the moderate activities reported in IPAQ
and OIMQ. In the SHS97, respondents reported 1.7 T 1.7
sweat episodes per week. In Table 7, correlations of the
questionnaire results with MTI measures, percent body fat,
and estimated maximal oxygen consumption are given. The
correlation between reported amount of sitting in the IPAQ
and estimated time sitting accounted for by the MTI was 0.22
and did not reach the level of statistical significance. Higher
associations at a statistically significant level were observed
between reported moderate activity by IPAQ (walking and
moderate activity with walking included) and by OIMQ and
the time registered at the corresponding level by MTI (r =
0.38, 0.39, and 0.53, respectively). Reported durations of
vigorous physical activity in all questionnaires were not
significantly related to registered time of MTI recordings at
the corresponding intensity level (r = j0.22 to 0.08).
The correlation coefficients between questionnaire
measures and physical activity recorded by MTI as total
counts per minute were at an acceptable level for walking (r =
0.38), moderate activity with walking included (r = 0.43), and
total activities (r = 0.39) in IPAQ, and for vigorous (r = 0.34)
and total activities (r = 0.43) in OIMQ. All other associations
between questionnaire data and MTI measures were not
statistically significant (Table 7).
Associations between reported activities in the question-
naires and the two indirect validation criteria, percent body
fat and estimated VO2max, were low and did not reach the
level of statistical significance (Table 7).
DISCUSSION
This study examined the reliability and validity of four
short physical activity questionnaires. In general, we found
reasonable reliability for continuous questionnaire measures
Medicine & Science in Sports & Exercised 1259
TABLE 6. Comparison of absolute values for duration of activities according to questionnaires (IPAQ, OIMQ) and MTI accelerometer data.

Moderate Activities Sitting Moderate Activities and Vigorous Activities

Source (minIwkj1) P (minIwkj1) P Walking (minIwkj1) P (minIwkj1) P
IPAQ 358 T 499* 0.00 2603 T 1397* 0.00 1289 T 1089 0.68 253 T 479* 0.00
OIMQ 1070 T 769 0.10 1070 T 769 0.10 84 T 117* 0.00
MTI 1212 T 340 3852 T 629 1212 T 340 30 T 33
IPAQ, International Physical Activity Questionnaire; OIMQ, Office In Motion Questionnaire. * P values for comparison between questionnaire and MTI data.

and moderate to fair repeatability for dichotomous data.
The validity, expressed as correlation coefficients between
questionnaire measures and accelerometer 7-d records, was
relatively low, but similar to measurement properties estab-
lished in other self-report physical activity measures (5,9).
It remains difficult to determine whether a clear
difference between two administrations of the same ques-
tionnaire is the result of an effect of recall or of actual
changes in behavior. However, as the recall period was
within 3 wk, volunteer behavior may not have changed
substantially. The variety of different items used in the
questionnaires resulted in varying kappa values for re-
peated categorical measures. In HEPA99, items aimed at
assessing whether respondents met the current health-
enhancing physical activity recommendations and there-
fore resulted in a dichotomous outcome. In contrast,
questions in IPAQ asked about frequency and duration
of moderate and vigorous activity and resulted in more
variability. Therefore, the agreement between the repeated
category measures might have been higher for HEPA99.
A similar effect could have been responsible for good
reliability found for the sweat episodes in SHS97 (Table 3).
Overall, the low kappa values for repeated categorical
measures with HEPA99 and IPAQ, respectively, indi-
cated that these categorical measures should be interpreted
with caution.
Reliability of activity measurement by questionnaire
recording continuous data was tested by calculation of
Spearman coefficients (Table 3). The moderate values (r =
0.50, 0.48, and 0.43, respectively) for moderate activities,
walking, and vigorous activities measured by IPAQ
indicated limited agreement between repeated measure-
ments. Although the reliability of total activities was higher
(r = 0.54), even better results have been found in another
study (5). The longer time period between the repeated
telephone interviews (up to 3 vs 1 wk) in this study could
be responsible for the lower correlation.
Compared with the IPAQ, repeated administration of the
OIMQ resulted in higher correlation coefficients for
moderate, vigorous, and total activities (Table 3). These
findings might be explained by the different structure of
questions. Whereas participants had to report global
estimates of moderate and vigorous activities in the IPAQ,
the OIMQ consisted of a more detailed list of activities
selected from a 24-h recall administered to a sample of
residents of Geneva, Switzerland. Moreover, the IPAQ
version that had been used was a telephone interview,
whereas the OIMQ was a self-administered written
instrument that was completed after answering the other
three questionnaires on the phone. However, in contrast to
IPAQ, OIMQ is less suitable for telephone interviews in
large representative surveys.
The correlation coefficient suggested good agreement
for repeated measurement of sweat episodes during leisure
time in SHS97 similar to another study (20), although
respondents had to sum up all vigorous activities during a
usual week. Probably due to the limited variety of possible
answers, the correlation was higher compared with the
vigorous-activity items of the IPAQ.
Validity of dichotomous questionnaire data was assessed
by comparing the number of days with at least three 10-
min bouts of moderate activities registered by MTI. During
the measurement week, the participants categorized as
active showed more such days than the individuals

TABLE 7. Criterion validity coefficients comparing total physical activity recorded by questionnaire with total accelerometer counts per minute, body fat, and maximal
oxygen consumption.

MTI measures
Inactivity Moderate Vigorous Total Activity Body VO2max
Questionnaire Component (min) Activity (min) Activity (min) (countsIminj1) Fat (%) (mLIkgj1Iminj1)
SHS97 Sweat episodes, leisure time (dIwkj1) j0.22 0.23 0.20 j0.11
IPAQ Moderate activities (min) 0.27 0.23
Walking (min) 0.38* 0.42* 0.29
Moderate activities and walking (min) 0.39* 0.43*
Vigorous activities (min) j0.03 j0.18
Sitting (min) 0.22
Total activities (no sitting) (METIminIwkj1) 0.39* j0.11 0.24
OIMQ Moderate activities (min) 0.53* 0.31
Vigorous activities (min) 0.08 0.34* 0.09
Total activities (METIminIwkj1) 0.44* j0.26 0.09
Inactivity and moderate and vigorous activities registered by questionnaires were compared with the time during the MTI accelerometer recorded activities at corresponding levels.
SHS 97, Swiss Health Survey 1997; IPAQ, International Physical Activity Questionnaire; OIMQ, Office In Motion Questionnaire. * Correlations were significant (P G 0.05).

1260 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
classified as inactive by the HEPA99. Statistical analysis
just failed to provide a significant difference between two
categories generated by the IPAQ data (Table 4). The
specific purpose of HEPA99, to classify respondents in
groups of different activity levels, may account for the
significant finding for this questionnaire.
Interquestionnaire validity for continuous data showed
different levels of association (Table 5). OIMQ might be
regarded as a reference questionnaire, as it consisted of a
data-based list of moderate and vigorous activities of the
Swiss population. The correlation was particularly good
with IPAQ.
The criterion validity of questionnaires was tested for
different activity intensities and inactivity in terms of sitting.
A lack of a significant association between sitting and the
registered time of inactivity by MTI has also been found in
other studies (5). Comparing absolute values of IPAQ
measurement and MTI data indicated that inactivity was
highly underreported by study subjects (Table 7). Good
agreement between questionnaire data and MTI measure-
ments was determined for moderate activities reported in
OIMQ (r = 0.53). Moderate activities reported in the IPAQ
correlated significantly with MTI measurements when
walking was included (r = 0.39) but not when it was ex-
cluded. The measurement properties of the MTI acceler-
ometer may account for this finding, as it has its inherent
limitations to register moderate-intensity activities other than
walking (3). However, the total amount of moderate activity
reported in IPAQ and OIMQ corresponded to the time of
activity at moderate levels registered by MTI (Table 6).
Comparison of reported amounts of vigorous physical
activity with MTI measurements indicated poor agreement,
as all criterion validity coefficients were not at a statistically
significant level. No statistically significant validity coef-
ficients were found for percent body fat, estimated VO2max,
or reported activities. The low amount of vigorous activity
recorded by MTI indicated (Table 6) that subjects were not
regularly participating in exercise improving their cardio-
vascular fitness. In addition, VO2max was only estimated by
the submaximal 2000-m walking test (17). Both circum-
stances might explain the unexpected findings. Surprisingly,
the OIMQ yielded a smaller amount of reported vigorous
physical activity compared with IPAQ, although several
activities were listed. Compared with MTI, activities at
vigorous level were overreported in both questionnaires
(Table 6). Overreporting of vigorous activities in IPAQ
has already been described elsewhere (21).
Moderate correlations were found for total physical
activities in IPAQ and OIMQ compared with total activity
recorded by the MTI accelerometer (Table 7). Both
correlation coefficients were lower than in other validation
studies (4,5). The OIMQ is a reduced version of the data-
based instrument of Bernstein et al. (4), which has been
validated in its comprehensive form against heart rate
measurements using a calibration procedure to associate
heart rate with energy expenditure on the individual level.
In the present work, no individual calibration mode for
MTI data was performed. Additionally, in the prior study,
VALIDITY OF PHYSICAL ACTIVITY QUESTIONNAIRES
Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
the questionnaire data were adjusted for overreported
activities (4). Both methods might result in an increased
validity coefficient. However, whereas no MTI data were
individually calibrated, the validity coefficient was higher
for IPAQ in the study of Craig et al (5).
This study is subject to several limitations. As only a
subgroup of the reliability sample participated in the validity
part of the study, the accuracy of generalizing the results may
be limited. However, the reliability data of the nonvalidation
and the validation sample were quite similar. Furthermore, 69
and 63% of the nonvalidation and validation participants
were classified as active by the HEPA99 data, respectively
(data not shown).
Due to the simultaneous administration of a variety of
questionnaires, the answers of respondents might have been
influenced by previously completed instruments. The
consecutive orders assigned randomly were limited to three
options. Nevertheless, the OIMQ, the sole self-administered
questionnaire, was completed after all telephone inter-
views. Participants might have observed their own physical
activities more precisely before answering the OIMQ. An
additional limitation concerns the use of the objective
measurement of physical activity chosen for this study. A
hip-mounted accelerometer cannot accurately quantify
activities such as walking up- and downhill, stair climbing,
biking, or any activity of the arms. Therefore, the draw-
backs of accelerometers might be a reason for misclassi-
fication through underestimation or overestimation of
activity levels, even though accelerometry recordings were
adjusted for swimming and bicycling. In addition, the cut-
points used in this study are relatively low compared with
other proposed values. Therefore, analyses might result in
an overestimation of moderate activity. However, higher
cut-points are based on walking and jogging activities that
result in relatively high MTI scores, and may not be
generalized to all activities (22).
In conclusion, interpretation of categorical data on
physical activity recorded with HEPA99 and IPAQ should
be done with caution, as reliability and validity of these
categorical measures were limited. The rather low correla-
tion coefficient between the number of sweat episodes
during leisure time per week and the accelerometer
recordings indicates that this method cannot be recom-
mended to describe overall physical activity. The short
version of the IPAQ seems to be a reliable and valid
questionnaire to assess physical activity, especially for
activities performed at moderate intensities and for the use
in large telephone surveys that have to ask on physical
activity beside a whole range of other issues. The item on
walking and the criteria of 10-min minimal duration for
bouts of physical activity seem to add to the value off this
instrument. The OIMQ, a self-administered instrument that
asks for duration and frequency of 33 selected activities, is
a reliable tool to measure total energy expenditure of phys-
ical activity.
The study was supported by grants of the Swiss Federal Council
of Sports and the Swiss Federal Institute of Statistic.
Medicine & Science in Sports & Exercised 1261
 REFERENCES
1. AINSWORTH, B. E., W. L. HASKELL, M. C. WHITT, et al. Compen-
dium of physical activities: an update of activity codes and MET
intensities. Med. Sci. Sports Exerc. 32:S498S504, 2000.
2. BARANOWSKI, T. Validity and reliability of self report measures of
physical activity: an information-processing perspective. Res. Q.
Exerc. Sport. 59:314332, 1988.
3. BASSETT, Jr, D. R., B. E. AINSWORTH, A. M. SWARTZ, S. J. STRATH,
W. L. OBRIEN, and G. A. KING. Validity of four motion sensors in
measuring moderate intensity physical activity. Med. Sci. Sports
Exerc. 32:S471S480, 2000.
4. B ERNSTEIN , M., D. S LOUTSKIS , S. K UMANYIKA , A. S PARTI ,
Y. SCHUTZ, and A. MORABIA. Data-based approach for develop-
ing a physical activity frequency questionnaire. Am. J. Epide-
miol. 147:147154, 1998.
5. CRAIG, C. L., A. L. MARSHALL, M. SJOSTROM, et al. International
physical activity questionnaire: 12-country reliability and validity.
Med. Sci. Sports Exerc. 35:13811395, 2003.
6. FREEDSON, P. S., E. MELANSON, and J. SIRARD. Calibration of the
Computer Science and Applications, Inc. accelerometer. Med.
Sci. Sports Exerc. 30:777781, 1998.
7. FREEDSON, P. S., and K. MILLER. Objective monitoring of
physical activity using motion sensors and heart rate. Res. Q.
Exerc. Sport 71:S21S29, 2000.
8. HARDMAN, A. E. Issues of fractionization of exercise (short vs.
long bouts). Med. Sci. Sports Exerc. 33:S421S427, 2001.
9. JACOBS Jr, 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. 25:8191, 1993.
10. JANZ, K. F. Validation of the CSA accelerometer for as-
sessing childrens physical activity. Med. Sci. Sports Exerc. 26:
369 375, 1994.
11. KARVONEN, J., J. CHWALBINSKA-MONETA, and S. SAYNAJAKANGAS.
Comparison of heart rates measured by ECG and microcomputer.
Phys. Sportmed. 12:6569, 1984.
12. MARCUS, B. H., C. A. EATON, J. S. ROSSI, and L. L. HARLOW.
Self-efficacy, decision-making, and stages of change: an
integrative model of physical exercise. J. Appl. Physiol. 24:
489508, 1994.
13. MARTIN, B. W. Physical activity related attitudes, knowledge and
behaviour in the Swiss population: comparison of the HEPA
Surveys 2001 and 1999. Schweiz. Ztschr. Sportmed. Sporttraum.
50:164168, 2002.
14. MARTIN, B. W., U. MADER, and R. CALMONTE. Attitudes, knowl-
edge and behaviour on physical activity in the Swiss population:
APPENDIX A
HEPA99
Telephone interview
(validated in German)
G102509
Think about moderate physical activities that make you
breathe somewhat harder and may include continuous
walking, hiking, dancing, gardening or sport activities.
Currently, do you do any physical activities that make
you breathe somewhat harder?
G19 Yes
G29 No
G99 No statement
PRGR.: If 1, GO TO 10400.
If 2/9, GO TO 10300.
1262 Official Journal of the American College of Sports Medicine
Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
Data of the Baseline Survey 1999. Schweiz. Ztschr. Sportmed.
Sporttraum. 47:165169, 1999.
15. MELANSON, Jr, E. L., and P. S. FREEDSON. Validity of the
Computer Science and Applications, Inc. (CSA) activity monitor.
Med. Sci. Sports Exerc. 27:934940, 1995.
16. National Institutes of Health. NIH Consensus Statement: Physical
Activity and Cardiovascualr Health. Hyattsville, MD: National
Institutes of Health. Publication 13, 1995, pp. 133
17. OJA, P., R. LAUKKANEN, M. PASANEN, T. TYRY, and I. VUORI. A
2-km walking test for assessing the cardiorespiratory fitness of
healthy adults. Int. J. Sports Med. 12:356362, 1991.
18. PAFFENBARGER, R. S. Jr, A. L. WING, and R. T. HYDE. Physical
activity as an index of heart attack risk in college alumni. Am. J.
Epidemiol. 108:161175, 1978.
19. PATE, R. R., M. PRATT, S. N. BLAIR, et al. Physical activity and
public health. A recommendation from the Centers for Disease
Control and Prevention and the American College of Sports
Medicine. JAMA. 273:402407, 1995.
20. ROEYKENS, J., R. ROGERS, R. MEEUSEN, L. MAGNUS, J. BORMS, and
K. DE MEIRLEIR. Validity and reliability in a Flemish population of
the WHO-MONICA Optional Study of Physical Activity Question-
naire. Med. Sci. Sports Exerc. 30:10711075, 1998.
21. RZEWNICKI, R., Y. VANDEN AUWEELE, and B. I. DE. Addressing
overreporting on the International Physical Activity Question-
naire (IPAQ) telephone survey with a population sample. Public
Health Nutr. 6:299305, 2003.
22. STRATH, S. J. Comparison of MTI accelerometer cut-points
for predicting time spent in physical activity. Int. J. Sports
Med. 24:298303, 2003.
23. SWARTZ, A. M., S. J. STRATH, D. R. BASSETT, Jr, W. L. OBRIEN,
G. A. KING, and B. E. AINSWORTH. Estimation of energy expen-
diture using CSA accelerometers at hip and wrist sites. Med. Sci.
Sports Exerc. 32:S450S456, 2000.
24. THOMAS, S., J. READING, and R. J. SHEPHARD. Revision of the
Physical Activity Readiness Questionnaire (PAR-Q). Can. J.
Sport Sci. 17:338345, 1992.
25. U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES, CENTER FOR
DISEASE CONTROL AND PREVENTION, NATIONAL CENTER FOR CHRONIC
DISEASE PREVENTION AND HEALTH PROMOTION SERVICES. Physical
activity and health: A report of the Surgeon General. Atlanta,
GA: 1996, pp. 146148.
26. WASHBURN, R. A., L. L. ADAMS, and G. T. HAILE. Physical ac-
tivity assessment for epidemiologic research: the utility of two
simplified approaches. Prev. Med. 16:636646, 1987.
G103009
Do you intend to do physical activities that make you
breathe somewhat harder?
G19 Yes
G29 No
G89 Do not know
G99 No statement
PRGR.: Continue with 10700.
G104009
Currently, do you do moderate physical activities that
make you breathe somewhat harder for 30 minutes or more
on every day or most days of the week?
Include all activities with a length of 10 minutes or more.
G19 Yes
G29 No
http://www.acsm-msse.org
 G89 Do not know G79 7 days per week
G99 No statement G99 No statement
PRGR.: If 1, GO TO 10600. PRGR.: Continue with 11000.
If 2/8/9, GO TO 10500. G108509
G105009 Do you work up a sweat while being physically active off
leisure time at least once a week? (For example: On the job?)
Do you intend to start with physical activities that
G19 Yes
make you breathe somewhat harder for 30 minutes or
G29 No
longer next month?
G99 No statement
G19 Yes
PRGR.: IF 1, GO TO 11000.
G29 No IF 2/9, GO TO 10900.
G89 Do not know G109009
G99 No statement Do you intend to start with sport activities that make you
PRGR.: Continue with 10700. work up a sweat?
G106009 G19 Yes
Have you been doing physical activities that make you G29 No
breathe somewhat harder for 30 minutes or more every day G99 Do not know / No statement
or most days of the week for half a year or longer? PRGR.: GO TO END.
G19 Yes G110009
G29 No Do you do sport activities that make you work up a sweat
G89 Do not know three times per week for 20 minutes or more on a regular basis?
G99 No statement G19 Yes
G29 No
G107009
G99 No statement
Now think about sport or other physical activities that
PRGR.: IF 1, GO TO 11200.
make you sweat. Do you work up a sweat during leisure
IF 2/9, GO TO 11100.
time while being physically active at least once a week?
G111009
For example: Brisk walking, running, bicycling etc.?
Do you intend to start within the next month with sport
G19 Yes activities that make you work up a sweat three times per
G29 No week for 20 minutes or more?
G99 No statement G19 Yes
PRGR.: If 1, GO TO 10800. G29 No
If 2/9, GO TO 10850. G99 Do not know / No statement
G108009 PRGR.: GO TO END.
On how many days during a usual week you work up a G112009
sweat during leisure time while being physically active? Do you do sport activities that make you work up a
G19 1 day per week sweat three times a week for 20 min or more since more
G29 2 days per week than half a year?
G39 3 days per week G19 Yes
G49 4 days per week G29 No
G59 5 days per week G99 No statement
G69 6 days per week GEND9

VALIDITY OF PHYSICAL ACTIVITY QUESTIONNAIRES Medicine & Science in Sports & Exercised 1263

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
APPENDIX B

1264 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
VALIDITY OF PHYSICAL ACTIVITY QUESTIONNAIRES Medicine & Science in Sports & Exercised 1265

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.
1266 Official Journal of the American College of Sports Medicine http://www.acsm-msse.org

Copyright @ 2006 by the American College of Sports Medicine. Unauthorized reproduction of this article is prohibited.