This conversation ties up some of the loose ends of the broader “stacking” issue. Quentin Anstee addresses Roger Green’s earlier concern about whether stacking variables would magnify statistical error by suggesting that if the goal of stacking is to amplify a “ground truth”, this should not imply greater error.
Louise Campbell asks about future prediction of placebo response, which Quentin suggests is beyond the boundaries of this approach, which Quentin suggests is better suited to interpreting results of individual trials where “stacking” can clarify indeterminate or non-significant findings. As the conversation winds down, Jörn Schattenberg suggests adding quality-of-life variables to the analytical set, which the group agrees can serve as part of a “feel, function, survive” analysis.