Predicting Placebo Responses with Multiple Regression and Configural Frequency Analysis: A Re-Analysis of Janke's 1963–1966 Experiment

The work reported here grows out of experimental studies on how placebos affect people psychologically and whether those effects can be forecast from a person's personality. The first author carried out these studies in Marburg and Gießen while preparing his habilitation thesis, over the years 1963 through 1966.

Part of the account is personal. The first author recalls the exchanges he had with Gustav A. Lienert in the period leading up to the experiments. Those conversations left a mark on the research: the way the data were first examined already drew on Lienert's thinking about a configural approach, even though he had not yet published his initial account of Configural Frequency Analysis (CFA) when the studies were under way.

Returning to the same data with current tools gives a clear result. A configural method such as CFA proves well suited to picking out the personality patterns that separate people who respond to a placebo (placebo responders, PR) from those who do not (placebo non-responders, PNR).

Running the analysis as a bi-prediction reveals something more specific about the two groups. Responders and non-responders turn out to carry complementary configurations of particular personality traits, each group defined by a pattern that mirrors the other.