H Strong Inference (John Platt)
Strong Inference is a step- by- step scheme for carrying out scientific investigations that Platt argued is the most efficient way to make progress.36 He claims that the major difference between rapidly advancing, quantitatively rigorous sciences (e.g., physics and chemistry) and sciences that progress slowly and more haltingly (e.g., social sciences) is that the former followed the principles of Strong Inference.
His paper describing the program attracted enormous attention (it has been cited more than 3,500 times37), not least because of Platt's frank, colloquial, not to say in-your-face style. Platt felt that a firm conceptual basis was essential for working with hypotheses and for progress; he opposed excessive mathematizing in favor of close logical reasoning and advocated applying the following steps to every scientific problem.2. H.1 The Program
1. Conceive of multiple alternative hypotheses to account for the phenomenon you’re studying (Platt credits T. C. Chamberlin [1897]38 for the original proposal of multiple hypotheses).
2. Devise crucial experiments to exclude one or more of the alternatives.
3. Carry out the experiments and interpret the data.
4. Recycle the procedure to develop subhypotheses and sequential hypotheses to refine the results with further testing.
2. H.2 Deduction and Exclusion
Although Platt often refers to “inductive inference,” he acknowledges that the crucial first steps in this program depend on conjecture to generate the hypotheses and on deduction to extract the key predictions for experimental testing. Indeed, his use of “induction” seems more colloquial than technical. Platt repeatedly emphasizes the vital importance of exclusion in Step 2, and he argues forcefully that merely confirmatory testing cannot lead to progress. The ability to eliminate one or more alternative hypotheses is the sign of a good test, and you continue testing and eliminating until only one hypothesis is left standing. A test that cleanly distinguishes between experiments is how Platt defines a “crucial experiment.”39
Platt asks “How many of us write down our alternatives and crucial experiments every day, focusing on the exclusion of a hypothesis?” He is a strong proponent of formal procedures such as keeping a permanent laboratory notebook and consciously analyzing alternative hypotheses and how to test them.
He took it for granted that skill in employing Strong Inference can be learned but that you have to practice consciously if you’re going to become an expert. Platt cites Popper approvingly and equates exclusion with falsification.2. H.3 Multiple Hypotheses
While Platt sees Strong Inference as a reworking of the traditional Scientific Method (observe-hypothesize-predict-test-modify), Strong Inference does emphasize two points that get little attention in the traditional accounts: the need for multiple competing hypotheses for a given problem and the iterative nature of the process. Having multiple hypotheses helps guard against becoming psychologically attached to any one of them and encourages a concentrated analytic approach to a problem. And the Strong Inference procedure is inherently iterative—conclusions drawn at the end of one investigation serve as the starting material for the next round of hypothesis testing. Platt, a scientist, aimed to establish a workable program for carrying out rigorous scientific investigations and didn't trouble himself with philosophical enigmas such as justified confidence in hypotheses.
In summary, for both Popper and Platt, the hypothesis is, at its core, an investigation, and falsification is the driving force for hypothesis testing and elimination For convenience, I'll usually lump their programs together under Popper's phrase “Conjectures and Refutations.” I'd like to round out the concept of the modern hypothesis by emphasizing a few of its other characteristics and introducing other forms of hypothesis.
2.I