Natasha is a third-year dissertation student in the lab. She works on a project investigating how humans perceive and predict the amount of exercise they do on a daily basis. We’re particularly interested in the question whether feedback affects our ability to predict our behaviour and the confidence of our predictions. Natasha is interested in motivating behavioural change and this project works on the basis that being able to judge your own behaviour is a precondition for the ability to change it. Read more about the project, below!
I’ll just walk it off – or will I?
Most of us aren’t active enough – not by a long way, if we follow the recommendations of public health campaigns suggesting 150 minutes of daily physical activity. But do we actually know by how much we’re falling short of what is apparently key to health and longevity?
Current public health tactics focus on increasing knowledge of how much exercise we should be doing, but awareness of what we are doing is just as important. A simple lack of awareness may be a barrier to behaviour change and make us less receptive to public health messages. Being aware of (a lack of) activity may not directly guarantee increased physical activity levels, but an absence of awareness means many people will not even be motivated to change their behaviour. A potential way forward may be the increasingly fashionable use of activity trackers and Smartphone apps that measure and report our behaviour. But are they?
In this study, we were interested in how good people across a range of body types are at estimating how much they move. We were particularly interested in whether feedback would make them learn something about their activity levels and make them better at judging their own behaviour – and whether they would have increased confidence in their own ability to know how active they are.
To address this psychological question, we provided participants with wrist-worn activity trackers (AX3 Axivity Accelerometer) and a phone app (‘Moves’), provided by our collaborators at Imperial College London. The AX3 Axivity Accelerometer gave us a very precise measure of the number of steps participants walked over the course of two weeks. ‘Moves’ provided us with GPS-based daily estimate on how many steps participants walked. All participants in this two-week study were asked every evening to report how many minutes they thought they had walked that day. We also asked participants how confident they were in their estimates.
In the second week of the study, participants kept reporting their estimates, followed by daily feedback on the number of steps they had actually taken and the accuracy of their previous estimate. We assessed body fat using a combined standardised Body Mass Index (BMI) and Waist Circumference (WC) score at the beginning and end of the study.
As a dissertation project, this study was limited to testing 40 participants, each monitored daily over the course of two weeks. And our body-fat measure, whilst hopefully being more accurate than BMI, is clearly just an approximation. We can thus make no bold claims about behaviour outside our sample; however, some of our results may be interesting pieces of evidence providing motivation for future research, especially those running contrary to our expectations. What we take away from the study is that:
Perhaps unsurprisingly, people seem to be not very good at judging their own activity levels; but contrary to our expectations (prejudices!) we found no indication that more overweight people do any worse than others in this task!
Use of technology clearly helps people to get a feeling for how much they move, which we consider an important first step to possible changes in behaviour. This is promising, as it could point towards use of this form of technology to train people and establish skills they maintain even after going ‘unplugged’.
This hope is slightly dampened by the finding that participants’ trust in their own abilities did not seem to improve greatly. However, while we do think that participants would need to trust their estimates to use them as a basis for decision-making, it’s quite possible that a week of receiving feedback a may simply have been to short to build that confidence.
Acknowledgments: We are very grateful to David Taylor & Aliza Abeles at Imperial College London for their support of the project. Matlab code creating the violin plots is based on the Robust Statistical Toolbox. (RST Copyright (C) RST Toolbox Team 2015).