Choosing the model and the analysis options available following this are generally self explanatory. The tables below list the options available.
Default values are shown in red.
Linear model options
Option
Selection
Notes
Transform Dose
None
Log 2
Log e
Log 10
c1 x dose + c2
Transform Response
None
Log 2
Log e
Log 10
Square root
Squared
c1 x dose + c2
c1 x dose + c2 option allows users to add an offset to the dose (c2) or multiply by a constant factor (c1).
Outlier Identification
No
Grubb's test
Studentised Residuals
This option may take some time to compute
Method for ratios
Delta
Fieller's Theorem
Use Weights
No
Weight A
Weight B
Weights the response values according to the variance. See Statistical Guide for details.
Linear Part Identification
No
Low dose end
High dose end
Both ends
The analysis fits a quadratic curve to the data and examines the significance of the quadratic (x2) term. If it is significant it removes a point from the upper or lower (or both) ends and repeats until the term is not significant. If only 2 points are left, the process is cancelled and the report continues with all points, noting that linear part identification was not possible.
