Fig. 4From: Handling an uncertain control group event risk in non-inferiority trials: non-inferiority frontiers and the power-stabilising transformationPower and type I error of procedure 4 (‘Conditionally modify margin with small threshold’), with different significance levels. Only presenting the risk difference case, as type I error of the base-case scenario was below the nominal 2.5% level when working on the risk ratio scaleBack to article page