There’s no such thing as ‘too much data’

By Prof. Corneel Vandelanotte, Professorial Research Fellow (CQUniversity) & Adjunct Professor (Curtin University)

 

As researchers it’s easy to think that there is no such thing as ’too much data’. But is that really true? E- and mHealth applications are able to capture a huge amount of data. Data that users (patients) provide directly into the applications, as well as data that is generated through usage statistics software (e.g. Google Analytics). Those programs are amazing and can provide researchers with a mountain of multi-layered information!

But the vast amount of data can also be challenging to make sense of in a useful way. As researchers we have an ethical duty to ‘use’ the data we generate, especially if it causes a burden on the participants in our studies. If we don’t intend to use the data we capture, we probably should not collect it.

On the other hand, I always get upset when I have to review a paper that examines the efficacy of an e- or mHealth intervention and the paper does not report any usage data, let along examine how usage and efficacy are associated. One used to be able to hide behind the excuse that it was too expensive or complex to implement usage statistics software, but with the availability of free and high-quality programs that is no longer possible. Yet, many studies still don’t report on usage and engagement with the e- or mHealth interventions. I get upset because there is so much we can learn from knowing how people engage with our interventions. And while usage statistical software creates a lot of data, often it comes down to aggregating the data into a couple of key variables, preferably operationalised similarly to what other studies have done so user engagement can be compared across different studies.

As an example of the importance of examining usage data I would like to refer to a study we did using our publicly available 10,000 Steps Australia program (www.10000steps.org.au), and that was published in the Journal of Medical Internet Research (Open Access link: http://www.jmir.org/2015/7/e176/). The 10,000 Steps program is available both as a website and as an app, and we wanted to know whether participant engagement was different for those who used only the website, only the app or both. Our results showed clearly that app users and app-and-web users had a much higher engagement with the program compared to those who only used the website. That’s really powerful information given the strong association between user engagement and behaviour change. But another important finding was that the engagement parameters differed a lot in relation to how long ago people had signed up to our program. With increasing length of membership, the average duration of usage, the number of individual and workplace challenges, and the number of days’ physical activity logged per week decreased.

This may seem like kicking in an open door, so why is this important? Virtually every study that has reported on user engagement has used a different time frame to do so, which makes it impossible to assess what is a ‘good’ or ‘poor’ level of user engagement when comparing with other studies. Thus, in sum, I think collecting and analysing the massive amount of user engagement data is very important, and it’s about time we started doing it in a more systematic way in order to move our research fields forward. What do you think?

 

BIO: Professor Corneel Vandelanotte leads the Physical Activity Research Group at the Central Queensland University in Australia and is Adjunct Professor at Curtin University. Prof Vandelanotte’s research has a population-based approach to health behaviour change and is focussed on the development and evaluation of web- and app-based physical activity interventions; with a particular focus on using computer-tailoring software to improve health behaviours. He is Editor-in-Chief for the Energy-Balance Related Behaviours Division for BMC Public Health journal and is a ‘Member-at-large’ on the Executive Committee of the International Society for Behavioural Nutrition and Physical Activity (ISBNPA). He is also the Founding Chair of a Special Interest Group within ISBNPA on e- & mHealth intervention research.