Positioning Computational Research as an Ongoing Process

The next presentation in this ICA 2018 session is by Drew Margolin, who highlights the growing use of computational methods in communication, and therefore the need to further scrutinise the methods that are popular here. Truth is revealed and reviewed through a succession of studies.

Success therefore depends on collective efficiency at testing and corroborating ideas, and replacing discarded ideas with new work. Existing theory must be tested for its relevance, by applying it to explain observable patterns in the available data; observations must also be used to generate hitherto unimaginable hypotheses. This should also encourage multi-causal explanations through causal inventories, accepting neither correlation nor uni-causal certainty. This also means positioning new studies not as final results, but as intermediary results on the way to greater certainty.

Further, we must encourage systematic observation, and aim for representativeness across studies. This also means not downplaying, for example, Twitter research as unrepresentative, but rather outlining means of replicating such research in other platform contexts.