Monday, 5 February 2018

What are Digital Experiments?

Have you heard of digital experiments? I had a weird idea that these were computer hardware experiments, but not so. Professor Matthew Salganik, presenting at the 2016 International Conference on Computational Social Science (IC2S2), suggests that our past research “experiments were analog experiments, and now more and more can take advantage of digital infrastructure", which he calls 'digital' experiments. "What are digital experiments?" he asks. "I would say an experiment is more digital to the extent that you use digital infrastructure in the four main steps of experiments: in recruiting participants, randomization, delivering treatment and measuring outcomes”.

This is a good point: that by using digital tools, we are more able to randomise and anonymise data than we ever have been - with a couple of provisos. Our catchment pool needs to be large enough; and our call to action needs to baited well enough to access our intended audience, often through digital media platforms such as LinkedIn and Facebook, to access widely-diverse networks.

However, Matthew is not talking about using methods to contact anonymous participants. He is talking about medical and social research, where the more we know about participants, the clearer the effects of a particular treatment or intervention may be. Digital experiments allow for big data, and for cross-tabbing of multiple factors to isolate likely effects and, therefore, results. This ends up being very big stuff indeed. He suggests four approaches for setting up research: industry or NGO partnership to share cost while giving up some control; go it alone, using existing systems, which will also be a control compromise; go it alone, building your own experiment, which is costly and probably slow to get numbers; build your own product to get control, but high cost and slow to get numbers and the system can be used again and again by others. This latter is not very common yet, but with apps, may well be a way to get access to data, if there is a hook for use. View eight minutes of his conference presentation here:

This presentation is a tiny slice of a book he wrote, Bit by Bit: Social Research in the Digital Age, which explores the issues of ethics, analysis, recruitment of participants, reliability and research relevance. 

Worth a read!


No comments :

Post a Comment

Thanks for your feedback. The elves will post it shortly.