Skip to main content
. 2019 May 29;10(2):267–286. doi: 10.1002/jrsm.1351

Figure 3.

Figure 3

Median posterior residual deviance contributions over time from univariate fixed treatment effects models with linear, exponential, piecewise linear, and Emax (model 4) time‐course relationships in the pain in osteoarthritis dataset. Residual deviances closer to 0 indicate a better fitting model. Nonlinearity in these plots indicates that the effect of time has not been properly accounted for. The scales for residual deviance vary between the upper and lower panels