What is the relative potency statistic in Probit Analysis?

Probit Analysis lets you provide a continuous stress variable and a categorical factor. When you provide a categorical factor, Minitab calculates the relative potency value. This value compares the potency of the stress variable across different levels of the categorical factor.

The relative potency value is only relevant when the factor variable is statistically significant. The factor is significant when the p-value in the Regression table is less than your chosen alpha level, usually 0.05. If the factor is not significant, then the confidence interval for relative potency will contain 1.

When a p-value > α for the factor and a confidence interval containing 1 for relative potency are in agreement, they both indicate that the stress variable has the same potency regardless of what value the factor takes.

Conversely, when the factor is significant, the confidence interval for the true relative potency will not contain 1. In such cases, the potency of the stress variable differs across factor levels, and the relative potency indicates how many times more potent the stress variable is when factor level = A compared to when factor level = B.

Example of interpreting a relative potency value

Suppose the relative potency = 10 for factor level A vs factor level B. Also suppose that the 95% confidence interval for relative potency spans from 8 to 12.

In this case, the p-value < α for factor, and you would conclude the factor level significantly impacts the potency of the stress. Specifically, you would conclude with 95% confidence that the stress is 8 to 12 times more potent with factor = A than with factor = B.