The development of a hypothesis from a set of premises is based on the logical relationship that exists between premises and hypothesis. The relationship is necessarily one of inductive logic, in which the argument proceeds from the specifics (the premises) to the general (the hypothesis). The strength of the relationship depends on the extent of conjecture involved in making the jump from the facts as summarized in the premises and the hypothesis that goes beyond the premises to provide a more useful explanation. More conjecture leads to weaker relationships; less conjecture leads to stronger relationships. The most meaningful way to assess and convey the strength of this logical relationship is to provide a numerical probability estimate of the confidence one can have that the hypothesis or inference is true.
The critical thinking skill is that of assessing the strength of these relationships in a manner that provides a numerical probability of the validity of hypotheses and inferences. Critical thinking is required because the process is a subjective one—subjective conditional probability—calling for a careful and deliberate assessment. The process is necessarily subjective (and consequently requires critical thinking) because the analyst will hardly ever have the type of statistical evidence needed to provide a simple objective calculation of probability (one that does not require critical thinking). In applying subjective conditional probability, the analyst must answer the following question: Given this specific set of premises (the conditions), what is the probability that the hypothesis (or inference) is true?
As stated earlier in this paper, the objective of intelligence analysis is to develop inferences that can be acted on with confidence. For the product of intelligence analysis to be complete, therefore, it must produce an inference that provides the needed explanation and, also, an estimate of the level of confidence that the user can have in that inference. The goal is to provide the greatest level of detail at the highest level of confidence. However, this usually results in a tradeoff—greater detail typically comes at a lower level of confidence. Conversely, the analyst can provide a higher level of confidence but with less detail. Providing confidence assessments enables the analyst to best meet the needs of the user— more detail at lower confidence or less detail at higher confidence. To provide such estimates, the analyst must be capable of generating and communicating subjective conditional probability estimates.