1.1. New Procedures and Features
Additional features and improvements in Release 10.11 are as follows:
A new output option Run with Linear Regression was added to Stepwise Regression for extended output.
Logit / Probit / Gompit Regression
Extended Weighted ANOVA table and weighted R-squared test were added to Logit / Probit / Gompit.
90% confidence interval for regression coefficients was made default for binary and aggregate dependent variable models, but that can be changed by the user.
Case (Diagnostic) Statistics: Deviance residuals are now displayed for binary and aggregate dependent variable models.
Heterogeneity factor (or dispersion parameter) can now be estimated for binary and aggregate dependent variable models, based on Deviance or Pearson statistics.
ED50 option (and its confidence limits), when there is only one independent variable, has been removed. This is already fully covered in Quantal Response Method procedure.
Analysis of Bioassays (all procedures)
Dose prediction: This is a new feature in Case (Diagnostic) Statistics output of all bioassays analysis procedures supported here. A response prediction facility has already been available in all previous versions of UNISTAT’s Bioassay Module. In a row with no other missing values, if only the dose value is missing, then the dose value will be calculated using the estimated model parameters. When a dose value is predicted, its row label will be prefixed by an ampersand (&). See section 10.0.7. Prediction, Interpolation, Extrapolation.
Display of any Pass/Fail test can now be suppressed by entering a threshold of -1 under the [Bioassay] section of Unistat.ini file. See Display test results Pass/Fail and Display test details under section 10.0.5.1. Compliance Output.
Potency confidence limits testing: The new Pass/Fail column on Percent Confidence Interval output tests whether the lower limit is greater than or equal to 75% AND the upper limit is less than or equal to 125% by default. If one of these two results is Fail, then the overall result will also be Fail. These default percentage values can be changed by the user. It is also possible to test only the lower limit against a given lower threshold or only the upper limit against another upper threshold. See Display test results Pass/Fail and Display test details under section 10.0.5.1. Compliance Output.
Individual non-linearity statistics for preparations: All bioassay procedures will report non-linearity tests for individual preparations, as well as the overall non-linearity for all preparations. It is possible not to display the individual non-linearity values. See section 10.1.2.5. Validity of Assay.
Treatments in ANOVA tables: A Pass/Fail test was added.
Procedure headers: Valid and omitted number of cases are reported. If a column has been selected by the Select Row function, deselected cases are also reported separately.
settings: All settings concerning bioassay procedures have now been made
subject to Compliance rules and are stored under the [Bioassay] section of
the following file:
Compliance: A fixed UNISTAT version line added to headers and footers. User-defined Main Title and Sub Title are now under the header, replacing the procedure name.
Parallel Line and Slope Ratio Methods
An optional [Plate] variable has been added to Parallel Line Method and Slope Ratio Method Variable Selection Dialogues. This usually helps to isolate the plate effect or the effect of replicates. If all dose/treatment groups (or cells) have an equal number of replicates, or if the design is balanced, then the plate effect is equivalent to Randomised Block Design.
Plots for both methods have been improved as Dose-Response Plot with choice of Means with Error Bars or Data Points plot options and trend line with confidence limits.
Quantal Response Method
Extended Weighted ANOVA table and weighted R-squared test was added to Quantal Response Method.
Four-Parameter Logistic Model
EP (European Pharmacopoeia) option:
An optional [Plate] variable was added, enabling testing of plate and replicate effects and Randomised Block Design.
Extended Weighted ANOVA table and weighted R-squared test was added.
The ANOVA of Regression table, including an unweighted R-squared test, was added.
Reduced Model USP: Goodness of Fit output options were re-arranged and parallelism tests replaced by a full ANOVA table as described by Gottschalk, P. G. and Dunn, J. R. (2005).
A USP Summary option was added, combining Full and Reduced USP models.
Combination of Assays
The Combination of Assays procedure formerly required confidence limits and degrees of freedom data of all potencies to be combined. Now, a variable selection option Only Potencies are Given is added, enabling the user to combine potencies without any accompanying information.
The new optional variable [Preparation] enables combining potencies belonging to different test samples separately.
Plot of Potencies has now options to draw mean and confidence interval lines of the calculated mean potency as well as individual potencies with error bars.