After literally 2 years, I’ve finally finished making notes on Paul Meehl’s philosophy of science lectures. This is the portal to all the different lectures. I’ve also included a single sentence summary of them. They are excellent and I strongly recommend you watch them. However, his papers are the best scientific articles I have ever read of any topic. I sincerely believe they should be required reading for anyone considering working in psychology. Consequently, I’ve included a few recommendations (along with a short summary) below.
Lecture 1: Philosophy of science in the 19th and 20th century
Lecture 2: Popper, Bayes theorem, & Lakatos
Lecture 3: Theories of truth
Lecture 4: Different kinds of theories
Lecture 5: The Lakatosian defence and research programs
Lecture 6: Lakatosian retreat and the 10 obfuscating factors
Lecture 7: 10 obfuscating factors
Lecture 8: How to test your theory
Lecture 9: Theories of probability
Lecture 10: Clinical versus statistical prediction
Lecture 11: How scientific is psychoanalysis?
Lecture 12: The weakness of significance tests and the corroboration index
Theory-testing in psychology and physics: a methodological paradox (1967). A wonderfully written paper demonstrating how the way statistics is used in soft sciences (with point-null hypotheses and vague directional hypotheses) results in weaker evidence for a theory given greater measurement accuracy. A genuinely eye-opening read.
Theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the slow progress of soft psychology (1978)2. An in-depth look into why psychology struggles with stronger tests and how it can be improved.
What social scientists don’t understand (1986). How weak tests of theory lead us to believe it has “money in the bank” and how a research program is stronger than it really is while publication bias further muddies the water.
Why summaries of research on psychological theories are often uninterpretable (1990). The 10 factors described mean you can’t judge how true a psychological theory is based on looking at how many corroborating significant results it has.
1 They can be read in any order but I would recommend the 1967 paper first as it is the most accessible and sets the stage for his way of reasoning.
2 If you’re not familiar with implicative syllogism, I would recommend reading my notes on Meehl’s second lecture prior to this as it briefly explains it.
3 Meehl makes a mistake when discussing pilot studies: he talks about using them to gain a rough estimate of the effect size of the phenomenon you’re studying. But you shouldn’t do this as the effect size from the pilot study is almost certainly inflated and you shouldn’t power your study based on what has come before; you should power it based on what effect size you want to capture (I explain in more detail here). There’s also a typo on p. 199: it reads “The prior probability p(O1|O2), absent theory, should be small” but it should be “p(O2|O1)”.