Theories, Models and the Hypothetico-Deductive Method

Distinguish a scientific theory from a model, see how the classical positivist picture of science grows out of the deductive-nomological model, and follow the hypothetico-deductive method through the classic case of Ignaz Semmelweis.

1 Introduction

The deductive-nomological model tells us what a scientific explanation looks like. But explanations rely on laws, and laws live inside larger structures: theories and models. And once we accept the DN picture, a particular view of how research should proceed follows almost automatically — the hypothetico-deductive method. This page connects those pieces and illustrates them with a famous case from the history of medicine.

2 Theories and Models

These two words are often used loosely, but they pick out different things.

Scientific theories

• are groups of statements that relate to the same area of phenomena and complement one another, so that further statements can be logically derived from them;
• thereby provide an ordering and systematisation of laws.

A theory, in other words, is a coherent web of laws and concepts from which explanations and predictions can be deduced.

Scientific models are not the same as theories:

• the common understanding is that a model is a simplified representation of some target system;
• but there is a genuine debate about their status — are models merely subsidiary to theories (convenient simplifications of an underlying theory), or do they constitute an independent way of reasoning in their own right?

Note

Holding the theory/model distinction clearly in mind matters for reading research. When a paper presents “a model”, ask whether it is offered as an approximation derived from an accepted theory, or as a free-standing tool for reasoning about a phenomenon for which no settled theory exists.

3 From the DN Model to Positivist Research

The classical positivist approach to research can be read straight off the DN model. If explaining is deducing outcomes from laws and conditions, then:

• the task of science becomes the testing of previously derived hypotheses — the exploratory side of science (where new hypotheses come from) is left outside the picture;
• the goal of science becomes the identification of ever more general and true laws, with an ideal end state in which everything can be explained by a few universally valid laws.

To pursue that goal, an idealised procedure was developed which — despite well-known problems — remains prominent: the hypothetico-deductive (H-D) method.

4 The Hypothetico-Deductive Method

The H-D method runs, in essence:

1. Confront a puzzling phenomenon.
2. Formulate one or more hypotheses that might account for it.
3. Deduce testable consequences from each hypothesis.
4. Confront those consequences with observation or experiment.
5. Reject the hypotheses whose predictions fail; retain (provisionally) those that survive.

The asymmetry in step 5 is important: a hypothesis is never conclusively proven, only not yet falsified. This connects to the problem of induction — confirming instances never strictly establish a universal law, but a single genuine counter-instance can refute it.

4.1 The Semmelweis case

The method is beautifully illustrated by Ignaz Semmelweis, a physician in Vienna between 1844 and 1848. He faced a stark puzzle: one maternity division of his hospital had a high mortality rate from childbed (puerperal) fever, while a second division did not.

He proposed several hypotheses and tested each:

• H1: an epidemic — rejected (it would not spare the second division).
• H2: overcrowding in division 1 — rejected (division 2 was in fact more crowded).
• H3: incompetent medical staff — rejected.
• H4: stress caused by priests rushing through — rejected.
• H5: sepsis from a lack of hygiene — not falsified.

Only the hygiene hypothesis survived testing, and a lack of hygiene (doctors moving from the dissection room to deliveries without disinfecting) was identified as a causal factor in the high mortality. Hand-washing dramatically reduced deaths — decades before the germ theory of disease made why it worked fully intelligible.

Note🎓 MA / PhD

The H-D method’s “despite fundamental problems” caveat is doing real work. Beyond the problem of induction, the Duhem–Quine thesis notes that hypotheses are never tested in isolation but only together with a bundle of auxiliary assumptions — so a failed prediction never tells you which assumption to give up. Popper’s falsificationism sharpened the method into a demarcation criterion, while Kuhn, Lakatos and Feyerabend challenged the idea that science actually proceeds by tidy conjecture-and-refutation. Note too that the positivist framing brackets the context of discovery (where hypotheses originate) in favour of the context of justification — a separation that qualitative and abductive approaches to research deliberately reject.

5 Summary

• A theory is a systematised web of laws from which further statements can be derived; a model is a simplified representation whose status (subsidiary to theory vs. independent reasoning tool) is itself debated.
• The classical positivist view of science follows from the DN model: science tests hypotheses in pursuit of general, true laws.
• The hypothetico-deductive method formulates hypotheses, deduces testable consequences, and rejects those that fail — hypotheses are never proven, only not yet falsified.
• The Semmelweis case shows the method at work: five hypotheses proposed, four refuted, one (lack of hygiene) retained.

6 What to Read Next

• What Makes an Explanation Scientific? The Deductive-Nomological Model — the model this approach rests on.
• Qualitative vs Quantitative Research — how the positivist picture contrasts with constructivist, inductive approaches.

7 Further Reading

• Hempel, C. G. (1966). Philosophy of Natural Science. Prentice-Hall. (On Semmelweis and the H-D method.)
• Popper, K. (1959). The Logic of Scientific Discovery. Hutchinson.