assessment of physical activity behavior

Journal of Physical Activity and Health, 2012, 9(Suppl 1), S11-S18

© 2012 Human Kinetics, Inc.

Framework for Physical Activity

as a Complex and Multidimensional Behavior
Kelley K. Pettee Gabriel, James R. Morrow, Jr., and Anne-Lorraine T. Woolsey

Context: The selection of the most psychometrically appropriate self-report tool(s) to measure specific physi-
cal activity constructs has been a challenge for researchers, public health practitioners, and clinicians, alike.
The lack of a reasonable gold standard measure and inconsistent use of established and evolving terminology
have contributed to these challenges. The variation of self-report measures and quality of the derived summary
estimates could be attributed to the absence of a standardized conceptual framework for physical activity.
Objective: To present a conceptual framework for physical activity as a complex and multidimensional behavior
that differentiates behavioral and physiological constructs of human movement. Process: The development of
a conceptual framework can provide the basic foundation from which to standardize definitions, guide design
and development of self-report measures, and provide consistency during instrument selection. Conclusions:
Based on our proposed conceptual framework for physical activity, we suggest that physical activity is more
clearly defined as the behavior that involves human movement, resulting in physiological attributes including
increased energy expenditure and improved physical fitness. Utilization of the proposed conceptual frame-
work can result in better instrument choices and consistency in methods used to assess physical activity and
sedentary behaviors across research and public health practice.

Keywords: conceptual model, human movement, sedentary, self-report, measurement

The relations between physical activity, energy Cancer Institute (NCI) and American College of Sports
expenditure, and physical fitness with health outcomes are Medicine (ACSM) conducted a conference on objective
well established.1,2 The 2008 Physical Activity Guidelines monitoring of physical activity. Logically, the following
for Americans (PAG) document the variety of physical year, the NCI, ACSM, and other organizations including
activity types (eg, aerobic, moderate, vigorous, lifestyle, the Centers for Disease Control and Prevention (CDC),
resistance/strengthening) that influence health status, National Institutes of Health Office of Disease Preven-
quality of life, and physical functioning.2 As such, it tion, and National Collaborative on Childhood Obesity
becomes necessary to identify, validate, and track physi- Research conducted a similar conference on self-reported
cal activity in individuals and populations. Importantly, measurement of active and sedentary behaviors. The
a common foundation or framework with consistent intent was to “close the gaps” on self-reported method-
interpretation and understanding must be established ologies related to the assessment of physical activity and
when measuring physical activity. Accurate assessment sedentary activities. The workshop objectives were to
of physical activity is critical in efforts utilizing physi- develop a framework, process, procedures, and recom-
cal activity as an outcome-, exposure-, or confounding- mendations to consider when using self-report physical
variable when relating to health outcomes. Consistency activity measures. The goal was to establish informed
in the understanding, usage, and assessment of physical guidelines and recommendations from an expert panel
activity and related terms can result in better interpreta- that could be used by a broader audience interested in
tion of effects, outcomes, health-related decisions, and self-report assessment of active and sedentary behaviors.
verification of effects. This, in turn, would provide a systematic method for
Direct measures of physical activity currently exist; measurement decisions with the goal of optimizing the
however, these technologies are often expensive and precision of self-reported estimates of physical activity
can be burdensome to individuals. In 2009, the National and sedentary behaviors.
The evidence-based relationship between physi-
cal activity and many health outcomes is important
Pettee Gabriel is with the Dept of Epidemiology, Human Genet- enough for surveillance systems, such as the CDC’s
ics, and Environmental Sciences, University of Texas School Behavioral Risk Factor Surveillance System (BRFSS),
of Public Health, Austin, TX. Morrow and Woolsey are with National Health and Nutrition Examination Survey
the Dept of Kinesiology, Health Promotion and Recreation, (NHANES), and Youth Risk Behavioral Surveillance
University of North Texas, Denton, TX. System (YRBSS), to include physical activity questions

S11

Unauthenticated | Downloaded 12/06/21 08:44 PM UTC

S12  Pettee Gabriel, Morrow, and Woolsey

in large-scale national risk factor assessments. Responses public health recommendations for physical activity
to these questions provide estimates of the prevalence were applied.
of individuals accumulating sufficient physical activity Caspersen et al12 attempted to define and standard-
to meet public health guidelines for physical activity2 ize the terminology that researchers, clinicians, and
within nationally representative samples of U.S. adults practitioners use to denote constructs related to physical
(eg, BRFSS and NHANES) and high school students (ie, activity to provide a common foundation for interpreta-
YRBSS). Further, attempts have been made to develop tion and comparison. Interestingly, despite that effort,
generalized physical activity assessments worldwide. The confusion around the meaning of relevant terms contin-
International Physical Activity Questionnaire (IPAQ)3,4 ues 25 years later. Consider the definitions of physical
and Global Physical Activity Questionnaire (GPAQ)5 are activity contained in “Physical Activity and Health: A
illustrative of global surveillance self-report instruments. Report of the Surgeon General.”1 In this report, physi-
The application of self-report physical activity measures cal activity is defined as “bodily movement produced by
in research settings has existed for several decades.6–8 the contraction of skeletal muscle that increases energy
However, confusion and misunderstanding about how expenditure above the basal level” (page 20).1 On the next
to best quantify active and sedentary behaviors using page, physical activity is defined as “bodily movement
self-report methods have permeated into even the most that is produced by the contraction of skeletal muscle
well-designed health promotion efforts, public health and that substantially increases energy expenditure”
research, or clinical studies. The cause of this confusion (page 21).1 The first definition could include standing
may stem from the use of nonstandardized definitions and or light-intensity activities. Alternatively, depending on
individual-level misinterpretation of words or meanings the interpretation of the term “substantially increases,”
that can result from differences in context or cultures. For one may only think to report vigorous-intensity physical
example, in 2010, the annual BRFSS survey question for activity. In this single yet critically important example,
exercise was, “During the past month, other than your the definition of physical activity literally differs from
regular job, did you participate in any physical activities one page to the next. Thus, it is not surprising that con-
or exercises such as running, calisthenics, golf, gardening, fusion still exists today among researchers, health care
or walking for exercise?” Therefore, this survey question and public health practitioners, health promotion experts,
ignores physical activity accumulated in domains other and the general population regarding physical activity.
than leisure-time and provides no context on frequency,
duration, or specific type (eg, walking vs. soccer) of Physical Activity: A Complex
physical activity. Transportation and occupation are
and Multidimensional Behavior
important omissions. Further, the older adult who “walks
down the lane” daily to the mailbox may report this Physical activity is a complex and multidimensional
5-minute walk as “exercise” based on this questioning. behavior that does not stand in isolation from other related
Such interpretive errors can result in imprecise prevalence constructs, including sedentary behavior, energy expen-
estimates of physical activity or inactivity in the popula- diture, and physical fitness. A conceptual framework,
tion. For example, the CDC State Indicator Report on often visual in nature, is used to present a standardized
Physical Activity, 20109 reports that 64.5% of U.S. adults approach to an idea, thought, or concept. This approach
are “physically active” [ie, ≥150 moderate-to-vigorous can be applied to physical activity research to provide the
intensity physical activity (MVPA) min per week], 43.5% basic foundation from which to standardize definitions
are “highly active” (ie, ≥300 MVPA min per week), and and guide instrument development for measurement and
25.4% engage in “no leisure-time physical activity.” Inter- selection of appropriate tool(s) to quantify physical activ-
estingly, Troiano et al, using NHANES accelerometer- ity and related parameters in research, clinical, or com-
derived physical activity data from 2003–2004, report munity settings. Although several conceptual frameworks
that < 5% of the adult population accumulated enough have been proposed, none have been widely adopted and
MVPA to meet public health guidelines recommended used in physical activity research or related applications.
at the time of data collection.10 This lack of congruency To advance the field of physical activity measurement
between self-report and accelerometer-derived prevalence and related research, it becomes imperative to develop,
estimates of meeting recommendations could be due to standardize, and use a framework for physical activity that
many reasons including misinterpretation of the question, is both useful and applicable to all subdisciplines related
reporting biases intrinsic to self-report measures (eg, to physical activity and exercise science.
recall, social desirability), the ability of the accelerometer We are not the first to propose a conceptual frame-
to detect primarily ambulatory-like movement, and the work for physical activity. In describing the relationship
slight differences in public health physical activity recom- between physical activity and health, Shephard and
mendations from 2003–2004 (ie, 30 minutes or more of Bouchard13 proposed a framework centered on health-
MVPA on most days of the week10) and 2009–2010 (ie, related fitness. Their framework implied a reciprocal
150 minutes or more of MVPA per week2). A 2009 report relationship between physical activity and health-related
by Thompson et al11 further illustrates the discrepancies fitness as well as a shared association between health-
in physical activity status (ie, meeting vs. not meeting related fitness and health status. Further, the framework
recommendations) that can occur depending on which proposed by Shephard and Bouchard showed the potential

Unauthenticated | Downloaded 12/06/21 08:44 PM UTC

 Framework for Physical Activity   S13

contributions of heredity, personal attributes, lifestyle operationalize the components and expand on the main
behaviors, and the social and physical environments on constructs of human movement.
the relationships between physical activity, health-related
fitness, and health.13 The Behavior of Human Movement
Later, LaMonte and Ainsworth published a concep-
tual model for quantifying energy expenditure.14 Unlike The behavior of human movement can be conceptualized
the framework by Bouchard and Shephard, they proposed as active (ie, physical activity) and sedentary behavior
a model that could be used to identify appropriate instru- (Figure 2). Both active and sedentary behavior can be
mentation to measure constructs of interest. Within their influenced by upstream factors, including physiologi-
framework, LaMonte and Ainsworth14 identified “move- cal, psychological, social, and environmental correlates.
ment” as the global construct, with physical activity and Within the behavior of physical activity, 4 main domains
energy expenditure listed as 2 related dimensions. The or types of activity have been identified. These activity
authors further differentiate the 2 dimensions of move- domains include leisure-time physical activity; work- or
ment as the behavior (ie, physical activity) and energy school-related activity; household, domestic, or self-care
cost (ie, energy expenditure) of human movement. activities; and activity for transport from place to place.
Examples of both direct and indirect measures of physi- Although exercise shares a number of similar qualities
cal activity and energy expenditure were provided, with with physical activity, the terms are not synonymous
inference on how those dimensions could be extrapolated and should not be used interchangeably. Exercise is
to a common unit of expression (ie, kilocalories) with considered a subcategory of physical activity and is
which to compare with health outcomes.14 often classified under the leisure-time physical activity
Although both conceptual models used for our illus- domain. In the 1985 seminal paper by Caspersen et al,12
trative purposes contain useful and scientifically-sound exercise was defined as “planned, structured, and repeti-
principles, a framework that is applicable to the many tive bodily movement done to improve or maintain 1 or
subdisciplines of physical activity and exercise science more components of physical fitness.” When quantifying
has yet to be identified. Further, previous frameworks total physical activity, or activity within a specific domain
have not illustrated the direct relationship between the or activity type (eg, aerobic, muscle-strengthening, flex-
behavior and the physical and physiological results of that ibility), using self-report methods, data often collected
behavior. Finally, sedentary behavior was not addressed include frequency, intensity, duration, and sometimes
in previous conceptual frameworks. specific activity type (Figure 3). Frequency can be defined
Much like the conceptual framework proposed as the number of times a given activity, or set of activities
by LaMonte and Ainsworth,14 our global construct of within a specific intensity range (eg, moderate-intensity),
interest is “human movement.” Unlike its predecessor, was performed within a predetermined time period.
however, in the current framework a directional relation- This time period can either be usual/typical or explicitly
ship is implied between the behavioral aspect of human defined (eg, previous 7 days). Intensity can be defined
movement, the characteristic of human movement, and as the level of effort or physiological demand needed to
the physiological result or consequence of movement perform a specific activity. Duration is the amount of time
(Figure 1). With this basic framework, we can begin to (ie, minutes or hours) that the activity or activities within

Figure 1 — The directional relationship between the behavior (physical activity) and consequence (physiological attributes) of
human, musculoskeletal movement.

Unauthenticated | Downloaded 12/06/21 08:44 PM UTC

Figure 2 — A conceptual framework for physical activity as a complex and multidimensional behavior.

Figure 3 — Quantifying the behavior of human movement using self-report methods.

S14

Unauthenticated | Downloaded 12/06/21 08:44 PM UTC

 Framework for Physical Activity   S15

a given intensity range was performed. Duration can be contractile stimulation. Standing activities involve little or
used to quantify a single activity episode or reported as no movement; however, these activities can be differenti-
an average amount of time across a specific time period ated from sitting activities in that they require isometric
(ie, past 7 days). contraction of the antigravity (ie, postural) muscles.15
When examining the total behavioral profile of To quantify human movement, often ambulatory-like
human movement, one should also consider sedentary activities are measured using activity monitors including
behavior. Sedentary behavior has emerged as a new field pedometers and accelerometers. Through advancements
of focus in physical activity and public health research.15 in technology, many commonly used accelerometers
Sedentary comes from the Latin word sedere, - which now have the ability to capture activities that occur in 3
means “to sit.” Sedentary behaviors can be categorized planes/axes that characterize human movement. Further,
as nondiscretionary or discretionary. Nondiscretionary some accelerometer models contain inclinometers, which
sedentary pursuits include activities such as sitting during can detect changes in position, making it easier to dif-
work or school hours or while in a car driving, whereas ferentiate sitting from standing or lying down. Therefore,
discretionary sedentary activities include sitting while accelerometers can be quite useful to measure the lack
watching television, reading, playing video games, or of movement characterized by sedentary activities.15–17
computer use during nonwork or school-related hours.
When quantifying sedentary behavior via self-report The Consequences of Human Movement
methods, the frequency and duration of individual sed-
entary behaviors (eg, television watching) or the total or Movement results in physiological attributes including
average time spent in nonspecific sitting activities over increased energy expenditure and improved physical
a specific or usual period of time (ie, sitting time) is fitness. Total energy expenditure is comprised of resting
ascertained. Given that sedentary activities have similar metabolic rate, energy cost associated with thermogenesis
associated metabolic equivalent (MET) values (ie, 1.0– (ie, thermic effect of food), and physical activity-related
1.5 METs), quantifying intensity level is not necessary. energy expenditure. The contribution of basal (or resting)
Since the activities that define active and sedentary metabolic rate and thermogenesis to total energy expen-
behavior are quite different, instruments that measure diture is relatively stable (ie, ~70% and ~10%, respec-
active behaviors should not be used to quantify or infer tively). Physical activity-related energy expenditure is
time spent in sedentary pursuits. For example, if an indi- the most variable component of total energy expenditure
vidual is considered low active when quantifying his/her and typically accounts for a relatively low percentage (ie,
leisure physical activity levels, it does not necessarily ~20%) of total day energy expenditure.18 Given the rela-
mean that she is highly sedentary. Although individu- tive stability of the other components, the measurement
als can be both sedentary and physically inactive, there of physical activity-related energy expenditure is com-
are many individuals with behavioral profiles in which monly used to infer total energy expenditure. Regardless
physical activity and high sedentary time coexist (eg, of whether total or a component of energy expenditure
active couch potato phenomenon).15 Therefore, sedentary is of interest, energy expenditure is typically measured
behavior is not simply the opposite of physical activity directly, often in a laboratory setting.
and should be quantified separately. Further, sedentary In addition to energy expenditure, human movement
behavior should not be ascertained by simply subtract- also results in improvements in the 5 dimensions of
ing the time spent being physically active from total day health-related physical fitness, including cardiorespira-
hours. tory fitness, musculoskeletal fitness (ie, muscular strength
Other than direct observation, there are no gold- and endurance), flexibility, balance and coordination, and
standard or criterion measures of active and sedentary body composition. Moreover, depending on the actual
behavior. Thus, when attempting to quantify these type of movement, benefits can accrue in 1 or more facets
behaviors using self-report methods, it is important to of physical fitness. Much like energy expenditure, physi-
note that the derived summary variables are estimates cal fitness parameters are typically measured directly in
of perceived, not actual, behavior. Traditionally self- a laboratory or clinical setting. However, if self-report
reported physical activity estimates were largely derived measures use specific activities (eg, gymnastics) during
from aerobic activities. However, given that other types recall, then certain fitness-related aspects of that activity
of activities, such as muscle-strengthening and flexibility might be inferred (eg, balance or flexibility).
exercises, are encouraged in public health recommenda- Based on our conceptual framework for physical
tions, more work is needed to either improve existing activity as a complex and multidimensional behavior, we
tools or develop new instruments to complement more propose that physical activity is more clearly defined as
conventional aerobic-centered measures. the behavior that involves human movement, resulting
in physiological attributes including increased energy
The Characteristics of Human Movement expenditure and improved physical fitness. In general,
human movement is beneficially related to health,
Physical activity involves human movement, whereas whereas the consequence, or lack of movement, is con-
prolonged sitting (ie, sedentary behavior) involves skel- sidered health compromising. Physiologically, sleep has
etal muscle unloading characterized by the loss of local different effects on health when compared with sedentary

Unauthenticated | Downloaded 12/06/21 08:44 PM UTC

S16  Pettee Gabriel, Morrow, and Woolsey

behaviors; therefore, it is strongly recommended that including missed associations. That is, the true effect of
time spent sleeping be excluded when examining asso- physical activity on health outcomes may be even greater
ciations between sedentary behaviors and health-related than currently reported or theorized.
outcomes.15 As the role of sedentary behavior on health Broadly defined, the methodological issues poten-
becomes more valued in physical activity and exercise tially limiting assessment of active and sedentary behav-
science research, the question remains whether both ior include 1) study, 2) population, 3) instrument, and
active and sedentary behaviors need to be quantified 4) activity characteristics (Figure 4). Each category is
to generate a profile of active and sedentary behavior described in greater detail in the following sections. It is
or rather, is it satisfactory to ascertain one or the other important to note, however, that the specific characteris-
(ie, physical activity levels in the absence of sedentary tics provided are meant to serve as examples; not as an
behaviors or vice versa). According to 2010 data from the exhaustive list of all items within that particular category.
National Time Use Survey implemented by the National
Bureau of Labor Statistics,19 adults spend an average of Study Characteristics
8.67 hours of the day sleeping. That leaves approximately
15.33 hours left in the day for work, domestic, and Specific study characteristics that may impact instrument
leisure-time activities. Based on these data, adults report selection include study budget, staff resources, study
spending approximately 1% of time in health-enhancing design, population sample size, and study objectives.
(ie, exercise) physical activity, 25% in sedentary pursuits, Further, whether physical activity and/or sedentary
and the remaining 38% at work, which for some might behavior will be used as an outcome, exposure, or con-
be active and for others sedentary, and lower intensity founding variable can also influence instrument selection.
domestic or leisure-time activities. Thus, considerable For example, in applications where researchers want to
time and effort to date have been focused on a relatively simply control for physical activity, a global question-
small, yet important, aspect of daily life. However, there is naire that elicits information on current physical activity
a large proportion of daily life that self-report instruments status (ie, active vs. not active) might be all that is needed.
might not be capturing. This example provides supportive Issues of seasonality might also impact physical activity
evidence to measure both active and sedentary behavior. estimates, particularly when measured over shorter time
periods (eg, past 7 days) in geographical locations that
experience variation in temperature or precipitation. In
Methodological Considerations addition, a large prospective cohort study, with multiple
When Quantifying Physical Activity data collection time points and study sites across the U.S.,
and Sedentary Behaviors might not have the budget or staff resources necessary
to use accelerometers. Further, the outcomes of interest
Regardless of whether direct- or self-report instruments
in this example might develop over several years (eg,
are used to quantify active and sedentary behavior, it is
chronic conditions); therefore, implementing a self-report
important to consider methodological issues that may
measure(s) that uses a historical or a longer time frame
limit the precision of the derived estimates (Figure 4).
might be the most salient solution to measure active and/
Researchers, clinicians, and health care practitioners
or sedentary behavior.
alike must reflect on these issues before implementing a
measurement strategy to reduce the risk of information
bias in their efforts, a type of bias that can occur when Population Characteristics
the methods for obtaining information are inadequate, Individual- or population-level characteristics can influ-
which can lead to erroneous estimates. ence the selection process and the quality of collected
Specific types of information bias include recall, data. Factors including age, gender, race and/or ethnicity,
reporting, and misclassification bias.20 Issues related to cultural norms, primary or native language, socioeco-
recall and reporting biases are most relevant to self-report nomic status, highest educational attainment, cognitive
methods. Recall and reporting biases involve problems ability, and general health status, including functional
with cognitive processes that are necessary to retrieve ability and disability status, can drive the assessment
relative physical activity and sedentary information strategy. For example, complex self-report measures
stored in memory and generate a response in the format quantifying specific active and sedentary behaviors over
required for the answers.21,22 Misclassification bias could the past year are not appropriate for children. Moreover, a
occur in physical activity-related research when a truly self-report measure originally developed in English might
insufficiently-active individual is categorized as meet- not be appropriate for nonnative English speakers without
ing recommended levels of physical activity. Specific to first thoroughly examining the cultural relevancy of the
active and sedentary behavior measurement, nondiffer- specific items and adapting the measure accordingly to
ential misclassification is particularly problematic. Non- fit the needs of the population subgroup.
differential misclassification results from inaccuracies
in the methods used to characterize active and sedentary Instrument Characteristics
behaviors.20 In most instances, nondifferential misclassifi-
cation reduces the overall strength of association between The specific elements of the instrument may also help
the outcome of interest and estimation of active and/ direct researchers, clinicians, and health care practitioners
or sedentary behavior leading to spurious conclusions, decide what measurement tool(s) to use. For example,

Unauthenticated | Downloaded 12/06/21 08:44 PM UTC

 Framework for Physical Activity   S17

Figure 4 — Methodological issues that impact physical activity and sedentary behavior measures.

some self-report instruments use specific activities as cues time saving, low staff and participant burden and cost)
during recall, whereas other tools use broader, pooled cat- solution. However, the “digital divide” (ie, computer liter-
egories based on intensity-level. Differences in the format acy) must be considered in some population subgroups.24
of activities provided as recall cues may impact derived
summary estimates. Further, the psychometric properties Activity Characteristics
of the instrument (ie, reliability, validity, and sensitivity
to behavior change over time), mode of administration As mentioned previously, aerobic, large muscle group
(eg, interviewer- or self-administered), recall time frame activities were often the central (or only) focus of
(eg, usual, specific-time period, or historical), and the physical activity self-report measures. More recently,
suitability of measured constructs to the study, clinical, or the importance of additional physical activity domains
community population and outcome(s) of interest might has been identified. The inclusion of occupational, life-
also impact decisions regarding measurement tool selec- style, transportation, and other types of physical activity
tion. For example, it is important for a study evaluating behaviors is important in addition to the traditionally
the effectiveness of a 12-month walking intervention to assessed leisure-time aerobic physical activities. What
reduce high blood pressure levels in older adults to use types of sedentary behaviors are being reported? For
a questionnaire that specifically measures walking and is example, Figure 4 (bottom right) illustrates that a vari-
sensitive to detect changes in walking over time. Given ety of physical activity types are currently assessed and
the lower levels of computer literacy and cognitive issues tracked in different national-level surveillance systems.
related to the normal aging process, it might be prudent An important example includes resistance training and
to use an interviewer-administered self-report measure muscle strengthening activities related to musculoskel-
that utilizes a shorter-recall time frame, collected multiple etal health. The 2008 PAG recommend adults engage in
times during the 12-month intervention to measure walk- strengthening activities ≥2 days per week and for adoles-
ing and other active and sedentary behaviors in this group cents to do so ≥3 days per week.2 Yet, the interpretation
of older adults. The validity of derived estimates is typi- of “strengthening activities” can vary. Does this include
cally higher when an interviewer- vs. self-administered “resistance training” only or does it include Pilates,
format is used;23 however, interviewer-administered yoga, hand-held weights, exercise bands, calisthenics
surveys are not always practical. Computer-assisted and using body weight as the resistance, and other forms of
online surveys are gaining popularity as an efficient (ie, musculoskeletal activities? Researchers must consider

Unauthenticated | Downloaded 12/06/21 08:44 PM UTC

S18  Pettee Gabriel, Morrow, and Woolsey

what is included and what is omitted when delimiting 4. Hallal PC, Victora CG. Reliability and validity of the
their physical activity questioning. This logic extends to International Physical Activity Questionnaire (IPAQ). Med
sedentary behaviors. Sci Sports Exerc. 2004;36(3):556.
5. Bull FC, Maslin TS, Armstrong T. Global Physical Activity
Questionnaire (GPAQ): nine country reliability and valid-
ity study. J Phys Act Health. 2009;6(6):790–804.
Conclusion 6. Kriska AM, Caspersen CJ. Introduction to a collection of
physical activity questionnaires. Med Sci Sports Exerc.
We propose a framework for basic and applied research- 1997;29(6):S5–S9.
ers, clinicians, behavioral scientists, and public health 7. Paffenbarger RS, Jr, Wing AL, Hyde RT. Physical activity
practitioners to use when quantifying physical activity as an index of heart attack risk in college alumni. Am J
and related constructs. While the original intent was Epidemiol. 1978;108(3):161–175.
related to self-report, the framework is valid for both 8. Sallis JF, Haskell WL, Wood PD, et al. Physical activity
direct and self-report instruments. Consistency of the assessment methodology in the Five-City Project. Am J
interrelationship between related constructs and subse- Epidemiol. 1985;121(1):91–106.
quent terminology is the sine qua non if measures are 9. State Indicator Report on Physical Activity http://www.
to be interpretable, generalizable, accurate, and used cdc.gov/physicalactivity/downloads/PA_State_Indica-
tor_Report_2010.pdf. Accessed August 29, 2010.
appropriately in research or for surveillance and tracking
10. Pate RR, Pratt M, Blair SN, et al. Physical activity and
purposes. We provide study, population/individual, instru- public health. A recommendation from the Centers for
ment, and activity considerations/characteristics that all Disease Control and Prevention and the American College
must consider when assessing physical activity behaviors. of Sports Medicine. JAMA. 1995;273(5):402–407.
The choices made by researchers and practitioners regard- 11. Thompson D, Batterham AM, Markovitch D, Dixon NC,
ing instrument selection have important consequences. Lund AJ, Walhin JP. Confusion and conflict in assessing
Inclusion of some constructs and omission of others the physical activity status of middle-aged men. PLoS
have both intended and unintended consequences to the ONE. 2009;4(2):e4337.
derived estimate. It is important that these consequences 12. Caspersen CJ, Powell KE, Christenson GM. Physical
be fully understood. activity, exercise, and physical fitness: definitions and
distinctions for health-related research. Public Health Rep.
In summary, the key to moving the field of physical
1985;100(2):126–131.
activity assessment forward is to think about the nature 13. Shephard RJ, Bouchard C. Principal components of fitness:
of physical activity as a complex and multidimensional relationship to physical activity and lifestyle. Can J Appl
behavior. When selecting an assessment strategy, it Physiol. 1994;19(2):200–214.
is imperative to think about what parameters will be 14. Lamonte MJ, Ainsworth BE. Quantifying energy expendi-
included and what will be omitted, and then think about ture and physical activity in the context of dose response.
how it will be interpreted by the respondent, researcher, Med Sci Sports Exerc. 2001;33(6 Suppl):S370–378; dis-
or practitioner, and in the literature. Our proposed frame- cussion S419–420.
work lays the foundation for the remaining papers in this 15. Owen N, Healy GN, Matthews CE, Dunstan DW. Too
supplement that guide researchers, clinical and public much sitting: the population health science of sedentary
behavior. Exerc Sport Sci Rev. 2010;38(3):105–113.
health practitioners, health promotion experts, and the
16. Healy GN, Dunstan DW, Salmon J, et al. Breaks in sed-
general population when making decisions about instru- entary time: beneficial associations with metabolic risk.
ment selection and interpretation of self-report physical Diabetes Care. 2008;31(4):661–666.
activity behaviors through self-report. 17. Healy GN, Wijndaele K, Dunstan DW, et al. Objectively
measured sedentary time, physical activity, and metabolic
Acknowledgments risk: the Australian Diabetes, Obesity and Lifestyle Study
(AusDiab). Diabetes Care. 2008;31(2):369–371.
The authors would like to thank Melba S. Morrow, MA, Cooper 18. Ravussin E, Bogardus C. A brief overview of human
Institute, for her thoughtful review of this work. Material pre- energy metabolism and its relationship to essential obesity.
Am J Clin Nutr. 1992;55(1, Suppl):242S–245S.
sented in this paper was originally presented at the Measure-
19. American Time Use Survey Summary. http://www.bls.gov/
ment of Active and Sedentary Behaviors: Closing the Gaps in news.release/atus.nr0.htm. Accessed August 29, 2010.
Self-Report Methods workshop in Bethesda, Maryland on July 20. Gordis L. Epidemiology. Philadelphia, PA: W.B. Saunders
21–23, 2010. Company; 2000.
21. Baranowski T. Validity and reliability of self report
measures of physical activity: an information-processing
References perspective. Res Q Exerc Sport. 1988;59:314–327.
22. Durante R, Ainsworth BE. The recall of physical activity:
1. U.S. Department of Health and Human Services. Physical using a cognitive model of the question-answering process.
Activity and Health: A Report of the Surgeon General. Med Sci Sports Exerc. 1996;28(10):1282–1291.
Atlanta, GA: U.S. Government Printing Office; 1996. 23. Pettee Gabriel K, McClain JJ, Lee CD, et al. The evalu-
2. 2008 Physical Activity Guidelines for Americans. www. ation of physical activity measures used in middle-aged
health.gov/paguidelines. Accessed October 10, 2008. women. Med Sci Sports Exerc. 2009;41(7):1403–1412.
3. Craig CL, Marshall AL, Sjostrom M, et al. International 24. Gracia E, Herrero J. Internet use and self-rated health
physical activity questionnaire: 12-country reliability and among older people: a national survey. J Med Internet
validity. Med Sci Sports Exerc. 2003;35(8):1381–1395. Res. 2009;11(4):e49.

Unauthenticated | Downloaded 12/06/21 08:44 PM UTC