Hypothesis testing questions require assigning relative probabilities to possible answers to the question (referred to as ‘hypotheses’). These questions require us to have formulated a list of possible answers already. Sometimes this is straightforward, with the potential answers simply being ‘yes’ or ‘no’. In other situations we may be required to answer a hypothesis generation question first in order to produce a list of possible options. Example hypothesis testing questions include: Did we experience a cyber attack? Or, How much did our main competitor price their contract bid at?
Hypothesis testing questions tend to be closed questions that require a definitive answer. They relate to events which have already occurred (for questions about the future, see scenario generation and scenario evaluation questions) and are best tackled by methods which conduct a logical and systematic post hoc analysis of the available evidence. The mathematical toolkit of data analysis, and 'soft' techniques such as analysis of competing hypotheses, employ Bayesian principles to use the available information to answer these questions as effectively as possible. |