Sample Suitability Criteria
Goodness of fit of the model
, or ‘linearity’
-
If there is more than one replicate at each dose: lack of fit error can be compared with pure error via an F test.
This test will result in many failures if the replicates are not independent. -
The level (critical p value) can be chosen by the user.
-
-
A quadratic model can be fitted to the data and the significance of the quadratic term tested via a t test.
-
The level (critical p value) can be chosen by the user.
-
-
The R2 for the regression can be calculated.
-
The minimum value must be chosen by the user (no default setting).
Statistical significance of the slope
, BTest via a one-sample t test comparing the slope with zero.
-
The level (critical p value) can be chosen by the user
-
-
The sign of the slope must be stated (positive or negative).
Parallelism with reference
-
Ranges for combinations of test and reference slope parameters can be chosen.
-
Difference between slopes: 
-
Ratio of slopes: 
-
Normalised slope difference: 
-
Equivalence tests for combinations of test and reference slope parameters.
-
Ranges for the confidence intervals for
-
Difference between slopes:
-
Ratio of slopes:
-
Normalised slope difference:
-
The confidence level for the each interval can be chosen by the user (no default setting).
-
Significance tests for lack of parallelism can be used:
-
F test comparing the model with separate slopes with the parallel model.
-
The level (critical P value) can be chosen by the user.
-
-
χ2 test comparing the model with separate slopes with the parallel model.
-
The level (critical P value) can be chosen by the user (no default setting)
