Figure 6. Mean square error (MSE) of the generic risk factor (β1) estimated by linear regression with censored observations substituted with values that varied from 0 to 10, at three different levels of gene quantity (approximately 3, 2, and l log₁₀ gene copies per gram).

For comparison, the MSE’s of the intercepts estimated by the hierarchical model at the three concentrations are provided in the legend and are shown as tick marks on the y-axis.