What are the best measures to take when measuring sedentary behaviour and physical activity measurement? Do we have a fundamental problem in our approach to measuring sedentary behaviour and physical activity? Is this negatively impacting our attempts to understand these behaviours and find effective interventions?
On Wednesday 2nd May 2018 Dr Paul Kelly presented a webinar for us titled Should we re-frame how we think about physical activity and sedentary behaviour measurement? Which you can watch now on the Human Kinetics website. Following on from his webinar Paul has kindly put some of these considerations into this article for Human Kinetics.
We focus too much on “best measure” instead of the research aim
There is a prevailing ‘hierarchy of measures’ in sedentary behaviour and physical activity research that places self-report at the bottom, devices above and laboratory measures at the top. This is closely aligned to a ranking of measures based on their correlation with assumed gold standards such as Doubly Labelled Water. Hierarchies of measures and quantified values for validity are attractive for researchers and seem to help in selecting measures, but Paul challenged this in his 2016 in his article – Should we reframe how we think about physical activity and sedentary behaviour measurement? Validity and reliability reconsidered.
If the measure at the top of your pyramid is only best at assessing energy expenditure, then your hierarchy is for energy expenditure, not physical activity or sedentary behaviour; it tells you nothing about the usual pace of walking, or workplace screen-time for example, which are both potentially important determinants of health.
So, we argue that the question we should ask is ‘What is my research aim?’ and ‘Which measure best addresses my aim?’ not ‘Which measure is closest to the top of the hierarchy?’ Unless your aim is to assess energy expenditure or some closely related concept, the prevailing hierarchy may be leading you astray.
Sedentary behaviour and physical activity are not single, simple things
We all know this; physical activity is a broad set of behaviours including gentle walks, perhaps with friends or your dogs, cycling to work, weekend sports, training for a 5k, gardening, housework and for some occupational activity. Similarly, sedentary behaviour may be sitting in a café with a friend, staring at a computer screen at our desks at work (as the writer currently is!), lying on a bed reading or studying, or reclining on the sofa enjoying our favourite TV show (or some terrible mind numbing show – insert your own option).
Each of these behaviours may have different impacts on physical, mental and social health and require different approaches to promote or reduce them. What we often forget is that they are likely to need different approaches to measure them as well. How well a measure agrees with a certain gold standard is only relevant if that gold standard measures the behavioural construct of interest.
For example, let’s say we wanted to assess the way people commute to work (e.g. car, bus, walk or cycle). Self-report of mode and usual distance is likely to outperform a device based measure of movement or energy expenditure over a 7-day wear-time period. Conversely, if we want to know when in a 24 hour period most activity or sedentary time is accumulated a continuous device-based assessment will likely out-perform memory or recall.
Start with face and content validity in the context of your research aim
When it comes to selecting a measure, we recommend starting with face and content validity. In essence, this means; will the measure assess the construct (thing) of interest and how much of that thing will it cover? If you want to know about walking, laptop use, sport or other, use that to start the decision-making process (factoring in relevant considerations around ethics, feasibility and cost). If you are running an intervention, build a logic model; this will often tell you what you should be measuring and it will rarely be total energy expenditure!
Recommendations for sedentary behaviour and physical activity measurement
Here are Paul’s top tips:
- Forget existing hierarchies – self-report may be your best option;
- State your research aim and use that to explore options (a logic model may help);
- Specify which component(s) of physical activity or sedentary behaviour are of interest
- Consider face and content validity before statistical comparisons (e.g. correlations) to other methods;
- Factor in ethical, feasibility and cost considerations.
For more information please check out Paul’s webinar it’s free to watch and contains his fantastic research. The 45-minute presentation is followed by a 15-minute question and answer session.