The bias of expectation in experimental set-up

(58) Expectation bias

When we set up experiments to explore the nature of our particular scientific discipline, the actual experiments that we choose to perform are chosen under a great deal of bias. OK! The objectivity of what we assume we are distinguishing, in a particular experiment, is performed under the disciplined expectation that our hypothesis may prove right or it may prove wrong (Popper's falsification); so we try to avoid observer bias. But, this interpretation of observations is carried out under the (potentially immense) bias of "experiments that are not yet imagined – let alone performed"; these could prove to be critical in helping us to clarify the broader structure of the whole system. In all probability we are choosing to examine focal pockets in the fog of the broader unknown; and we do this in a way that follows an "in vogue fashion" that, of course, may well, in the end, help us to clarify the whole picture; but it could also prove, retrospectively, to be very distorting. Later experiments, designed and performed as the general fog is clearing, may even dramatically accelerate the process of discovery – for it is then that the broader structure becomes guessable (eg, Watson and Crick and the structure of DNA). We must never forget how parochial (and thus biased) our experimentation may eventually prove to be.