Visualisation of confidence in two-factor designs where model, replication and star points are varied

Confidence in predictions over an experimental domain can be calculated using the known experimental error and a function that relates to the shape of the confidence bands, often called leverage, In this paper, leverage is calculated for two-factor designs based on the central composite design, changing the nature of the model, the position of the star points and the number of replicates in the centre, The shapes of the graphs of leverage are presented, Several conclusions can be deduced, for example, for many common arrangements of experiments, such as the ease where three replicates are taken in the centre; confidence is not highest in the centre of experimentation.