Version Control

The table describes the changes that have been made to InVivoStat since the first release in 2010. Earlier versions of InVivoStat are available upon request from invivostat@hotmail.co.uk.

Version number Date Description of change
1.1 17/02/2010 Initial version
1.2 16/07/2010 Initial website version
1.3 01/09/2010 Inclusion of Nested Design Analysis module
Inclusion of Paired t-test/within-subject Analysis module
Inclusion of Incomplete Factorial Parametric Analysis module
Bug fix: if the user defines an X-axis title then it is not included in the plot
Bug fix: to stop InVivoStat crashing when running complex operations
Additional functionality: Allow dataset entries to include + and – symbols
Additional functionality: Rank transformation added to list of transformations
Non-parametric Analysis module: Dialogue to include “Mann-Whitney” as well as “Kruskal-Wallis”
Bug fix: Improvements to Excel data import
Non-parametric Analysis module: Warning messages to only include ‘Mann-Whitney’ rather than ‘Wilcoxon’ to tie in with interface
Power Analysis module: Rename ‘MS error’ to ‘MS residual’ to tie in with the rest of InVivoStat
Additional functionality: Inclusion of Normal probability plot in Summary Statistics module
Inclusion of Unpaired t-test Analysis module
Single and Repeated Measures Parametric Analysis modules: QQ plot title changed to Normal probability plot to tie in with interface
Repeated Measures Parametric Analysis Module: ‘Animal’ label on GUI replaced with the more general ‘Subject’ label
1.4 01/03/2011 Bug fix: In the Repeated Measures Parametric Analysis module Overall Effects Table, the Lowest p-value is not formatted properly
Additional functionality: In the Repeated Measures Parametric Analysis module no analysis performed if Animal factor level has replication of 1
Repeated Measures Parametric Analysis Module: Decision to use the within-animal variability when Studentising the residuals that are used in the predicted vs. Residuals plot
Bug fix: Trailing spaces in variable names causes analyses to fall over
For consistency, all p-values formatted to 3 decimal places
Bug fix: Tukey’s test in Single Measures Parametric Analysis module does not give unique results due to a random seed (seed now fixed)
2.0 19/07/2011 Bug fix: Problem with user defined formulas involving brackets
Additional functionality: Ability to print analysed data in the output
2.1 01/08/2011 Bug fix: In the Single and Repeated Measures Parametric Analysis module, if the covariate has a zero value and the covariate is log transformed, the analysis stops.
Bug fix: When plotting two continuous variables (in the Graphics module or as part of the output when fitting a covariate in Single and Repeated Measures modules) if all the x-axis values in one group are the same then the slope is infinite and InVivoStat halts.
2.2 13/10/2011 Bug fix: In the Non-parametric Analysis module, the treatment group labels in the all pairwise comparisons table were not always correct
Inclusion of Fishers Exact and Chi-squared tests module
Additional functionality: all means and differences to be quoted to 3 decimal places
Bug fix: Warning message about missing data incorrectly generated when log transformation applied to the covariate
2.3 13/03/2012 Inclusion of R-runner module
Change to interfaces to make them consistent across modules
Bug fix: Change to Unpaired t-test as the analysis does not proceeded if the response variable is called “Response” in the dataset
2.4 07/08/2012 Inclusion of Correlation module
Inclusion of Survival Analysis module
Additional functionality: Update to Excel reader
Update to Nested Design Analysis interface (Effects now called Factors)
2.5 26/11/2013 Additional functionality to all modules: ability to create black and white plots
Additional functionality to all modules: ability to create pdf versions of the plots
Bug fix: Remove limit of 2 decimal place format of data printout in Output window
Bug fix: In Single Measures Parametric Analysis module, drop Planned Comparisons terminology from LSD tests
Bug fix: Align y-axis on the Predicted vs. residual plots
Bug fix: Align all p-values to 4 decimal places
3.0 14/08/2015 Additional functionality: Update all plots to use the GGPlot2 R graphics package
Additional functionality: Allow user to manipulate appearance of graphics using a new interface
Additional functionality: Allow user to manipulate various global parameters and output options
Additional functionality: Graphics Module – for the scatterplot replace “correlation and p-value” with “Horizontal jitter”. The former now appear automatically.
Bug fix: Align quartile definition Q1 and Q3 across all modules
Additional functionality: User has ability to change plot size
Additional functionality: Repeated measures module, add ability to choose the Selected Effect
Additional functionality: Single and Repeated Measures Parametric Analysis module – ability to choose any model term as the Selected Effect
Additional functionality: Single and Repeated Measures Parametric Analysis module – addition of the LSMeans package so that means are adjusted for blocking factors
Update: For Repeated Measures Parametric Analysis module, use partial (Type III) sums of squares rather than marginal (Type I) to align with other modules
Inclusion of new Multivariate Analysis module (results not verified)
Additional functionality: Ability to import dataset of p-values, generated using the Single and Repeated Measures Parametric Analysis modules, into the P-value adjustment module
Additional functionality: Repeated Measures Parametric Analysis module – following log transformation ability to generate the geometric means and ratios between the means
Additional functionality: Single Measures Parametric Analysis module – following log transformation ability to generate the geometric means and ratios between the means
Additional functionality: Single Measures Parametric Analysis module – If making comparisons back to control, ability to plot the differences from control + 95% CI
Additional functionality: Single and Repeated Measures Parametric Analysis module – Ability to test the treatment by covariate interaction formally rather than just via a plot
Additional functionality: Single and Repeated Measures Parametric Analysis module – Ability to generate the covariate regression coefficient
Update to R V3
Bug fix: Rename the Predicted vs. residuals plot as the Residuals vs. predicted plot
Bug fix: Update to css file so that background output is always white
Additional functionality: Fishers and Chi-squared test module – Include the Barnard’s test (alternative to Fishers)
3.1 23/10/2015 Bug fix: Change to model parameterisation to correct p-values in the ‘Table of overall tests of model effects’ table in Repeated Measures Parametric Analysis module.
3.2 26/11/2015 Bug fix: Case selection conditions with re-ordered data corrected
Additional functionality: Regression lines extend to ends of plot window
Additional functionality: Random seed fixed so that random jitter on plots is consistent across repeated runs
Additional functionality: Difference with 95% confidence interval included in Paired and Unpaired t-test modules
Additional functionality: Output Options GUI re-organised
3.3 07/01/2016 Update to installer
3.4 19/01/2016 Bug fix: to allow spaces in % changes list in the Power Analysis module and user-defined equation start values in the Dose-Response module
3.5 05/09/2016 Clarification to Summary stats output: QQplot not adjusted for categorical factors
Bug fix: Issue with variable length in Dose response module
Output update: Covariate descriptive text amended
Additional functionality: Ability to have continental Windows setup
Interface update: Graphics module Y-axis title relabelled as Response variable
3.6 02/02/2017 Output update: P-value adjustment module to include contrast labels
3.7 14/07/2017 Update to improve reliability of installation process
4.0 13/02/2020 Non-parametric Analysis module: Inclusion of Freidman’s test and the Wilcoxon Rank Sum test
Graphics module: Ability to include labels on scatterplots
Single Measures Parametric Analysis module: Add coefficients from the model if requested
All modules: Inclusion of analysis options at the end of the output
Single and Repeated Measures Parametric Analysis modules: Change label in ANOVA from ‘Residuals’ to ‘Residual’
One sample t-test module: New module created to perform the one sample t-test
Graphics module: Ability to fit spec line for all plots
Graphics and Non-parametric Analysis modules: Ability to vary transparency of fill of bars/boxes
Power Analysis modules: One-way ANOVA power analysis module added
Correlation Analysis module bug fix: If the dataset contains a ‘y’ variable, then the analysis fails as it internally uses a ‘y’ variable. Update to remove the internal y variable
All modules: Generation of .PNG plots instead of .JPG plots to improve quality
Graphics and correlation modules: Generation of correlation coefficient and p-value which were missing when x-axis contained two distinct values
All modules bug fix: Print of raw data that contains only variables included in the analysis
MVA analysis module: Module validated
Graphics module: Ability to control categorised factor levels on plots
All modules: Ability to have ‘?’ in variable name
Summary Statistics module: %CV now positive when mean is negative
Correlation Analysis module: Update to matrix plot axis range to make them more sensible
All modules: Ability to generate a pdf of the output
Bug fix:  Correction to code to deal with p<0.0001 in output
Non-parametric Analysis module: Information about continuity correction and whether tests are exact or asymptotic
Non-parametric Analysis module: Exact tests used when there are no ties in the data and sample size if less than 50
All parametric modules: Ability to include multiple covariates in the statistical model
All parametric modules: Alignment of statistical models when interactions are selected
Bug fix in T-test modules: Title on Predicted Means plot corrected – Confidence interval title fixed at 95%
Single Measures Parametric Analysis module: Remove line connecting means from Predicted Means plot by default (ability to switch on in options tab)
Paired t-test module: Addition of ability to define control group (to compare back to)
Unpaired t-test module: Addition of ability to define control group (to compare back to)
Paired t-test module: Inclusion of plot comparing one response to the other (paired t-test only)
Graphics module: removal of default line comparing means in line plot of means with SEM
Graphics module: Removal of box around symbol in legend
Graphics module: Ability to fit a spec line in all plots
Graphics module: Ability to change ordering of levels in categorical x-axis and categorical factors
Ability to use InVivoStat on a max and linux platform
Ability to generate a pdf of all plots in output
4.0.1 13/02/2020 Bug fix to allow use on computer with continental European regional settings
4.0.2 18/03/2020 Bug fix to allow control group names to have spaces
Change pdf output to a more flexible HTML format
Update to R version 3.6 so that all packages can be installed (solves issue with Repeated Measures Analysis module)
4.1 13/07/2020 Graphics module: Legend included on the case profiles plot
Interface update: Removal of redundant post-hoc test option in unpaired t-test module
New module to perform logistic regression
Bug fix: When displaying back transformed pairwise comparisons (following log transformation) the effect size column was labelled ‘Ratio’ rather than ‘Difference’
Bug fix: In the P-value adjustment module, when using an existing dataset that contains the unadjusted p-values, the headers were not selected correctly
Bug fix: In the P-value adjustment module, p-values of the form p<0.001 were not analysed correctly
Bug fix: Datasets containing commas were not imported correctly
Mac edition: Update to instructions and R script for library installation
4.2 08/04/2021 New functionality: For repeated measures analyses, the AIC, BIC and log likelihood have been included to allow users to compare different correlation structures
New functionality: A new module created to allow users to transform “short-fat” repeated measures datasets into the “long-thin” format required by InVivoStat’s repeated measures analysis tools
New functionality: Back and Forward navigation arrows included on the InVivoStat interfaces
New functionality: For all parametric modules, when the blocking factor has only one level, an error message is generated
Bug fix: In Graphics module, legend containing Animal ID’s can now be included on the case profiles plot
New functionality: In Graphics module, new functionality to allow users to plot confidence intervals or standard errors in the means with error bar plots
New functionality: A new module to generate the Area Under Curve (AUC) summary measure for repeatedly measured responses. Original data can be in either long-thin or short-fat formats
Bug fix: the incorrect error bars that appear underneath the columns on the Means with Error bar plots have now been removed
New functionality: Plot axis titles reflect any transformation applied to the variables
Bug fix: When data is log transformed, for SMPA the plot of pairwise comparisons y-axis label changed to “Ratio of geometric means” rather than “Difference of geometric means”
Update: When data is log transformed, p-values now included in table of back transformed ratio of the geometric means in Single Measures, Repeated Measures Parametric Analysis modules, Paired and unpaired t-test modules
Update: When the log transformation is used, the mean column in the table of back-transformed means has been renamed to “Geometric Mean”
Update: For Single Measures and Repeated Measures Parametric Analysis modules, the order of output has been changed so that table of predicted means is presented before the corresponding plot
Update to output options: User can choose to display back-transformed geometric means with or without the log transformed results
Update: The user output option to change the angle of the labels is not applied to the x-axis and y-axis labels on the diagnostic plots