## Changes in 'drc' 2016-07-25: na.omit() is now the default in drm(). New functions included: CIcomp(), CIcompX(), and plotFACI() for calculating combination indices based on effective doses and also effects as described in the paper by Martin-Betancor et al. (2015). An accompanying dataset named "metals" has also been included. 2016-07-21: A small bug in printing confidence intervals has been fixed (after a report from Johannes Ranke). The output text has also been modified slightly. 2016-06-03: A small error for binomial data in the function estfun.drc for use with the package "sandwich" has been fixed (after a report from Andrew Kniss). 2015-12-16: The predict method has been updated (after a report from Duncan Mackay). The function SI() has been completely replaced by the function EDcomp(). 2015-09-16: The function isobole() now also propagates graphical arguments to fitted isoboles. 2015-08-19: The print.summary.drc method shows a warning message in case of df<1. Calculation of the log likelihood for binomial data has updated and improved. 2015-08-14: The function backfit() has been updated. 2015-08-10: The function diagnostics() has been removed. 2015-08-07: Help page for EDcomp() updated. 2015-07-30: Robust estimation is working again (ater reports from Kathy Mutambanengwe, Sten Ilmjärv, and Corina Dueńas Roca). 2015-04-14: Minor labelling issue in the plot method resolved (with help from Bert Oosthuyse). 2015-04-13: Help pages for ED() and isobole() has been updated. Some code has been tidied up. 2015-04-12: The plot method extended to provide confidence bands (contribution by Gregory Warnes). 2014-12-17: The argument "pshifts" has been added to drm() to allow weights on parameters (after a comment from Florent Baty). 2014-11-13: The argument "control" is now correctly propagated in case separate = TRUE in drm() (after a report from Andy Liaw). 2014-06-30: The help page for ursa() has been improved (in particular the example section). 2014-06-25: A number of not fully implemented model functions for mixture data have been removed. 2014-04-23: The model function gammadr() and multi2() have been added. 2014-04-06: The argument "vcov." has been added to compParm(). 2014-03-26: The argument "vcov." has been added to EDcomp() and predict(). 2014-03-19: The argument "vcov." has been added to ED() to allow choosing between the standard vcov method and the sandwich function for robust standard errors. SI() has been renamed into EDcomp(). 2014-03-05: The bread and estfun methods have been extended to event-time data. The help page for G.aparine has been improved. 2014-03-03: A number of functions depending on ED() and SI() have been updated. The predict method has been improved. Plot functionality for event-time data has been improved (after a report from Eshagh Keshtkar). 2014-02-24: ED() and SI() return (invisible) a list with 1 components where the second can be used directly with the package multcomp (after a suggestion by Daniel Gerhard). 2014-02-12: Small bug in the plot method (not ordering labels in legend text correctly) has been fixed (after a report by Francois Keck). 2013-10-22: Adding "fixed" value to output from logistic() (after a suggestion from Daniel Gerhard). 2013-10-01: Help page for "germination" has been updated. 2013-09-30: bread.drc() and estfun.drc() have been updated to handle fits for binomial and Poisson data (after a suggestion by Signe M. Jensen). 2013-08-22: Help page for plot.drc() has been improved regarding the explanation on the use of error bars (after an enquiry from Julien Delafontaine). 2013-05-24: Help page for "selenium" has been improved (after a comment from Keith Taulbee). 2013-05-21: An argument for specifying the reference for the normalization (after a suggestion from Sunniva Foerster). 2013-05-14: An argument for showing normalized data and fitted curves has been added to the plot method (after a suggestion from Ludwig A. Hothorn). 2013-03-25: Help page for mselect() has been improved (after a comment from Sona Jesenska). 2013-02-11: cedergreen() has been improved to provide meaningful names (after a suggestion from Dave Smithson). 2013-01-11: plot methods works for fits obtained from drm() using separate = TRUE (after a report from Thomas Kroeber). 2012-12-29: Model functions gaussian() and lgaussian() have been included (after a suggestion from Ismael Rodea). The method for residuals has been extended to provide residuals on the transformed scale in case a Box-cox transformation was applied. 2012-12-14: Help page for maED() has been extended. 2012-11-19: Small bug in drm() related to starting values for event-time models resolved. 2012-10-22: hatvalues and cooks.distance methods have been added (after a question from Sunniva Förster). noEffect() function included for testing the dose-response model against a simpler model with no dose effect (after a suggestion by Ryan Hechinger). 2012-09-08: backfit() function added (after a suggestion by Keld Sorensen). 2012-08-25: Poisson models can be fitted weights (after a suggestion by Marie Laure Delignette-Muller). 2012-08-20: Help page of NEC() has been updated. 2012-07-04: Small error in plot.drc for event times fixed (after report from Christian Andreasen). 2012-06-22: vcov.drc() has been updated. 2012-05-29: ED() now also works for gompertz models (after a question from Calvin Odero). 2012-05-29: The dataset germination has been included. 2012-05-27: The dataset chickweed has been included. 2012-05-23: Help page of lettuce has been revised. 2012-05-16: The dataset selenium has been included. 2012-05-16: Limit in dose scaling has been lifted (after a report from Andreas Wernitznig). 2012-03-13: Help page for the function modelFit() has been revised (removing a typo in the title) (after a comment from John Lynch). 2012-03-01: Help page for the function mr.test() has been updated. 2012-02-20: Summary output for separate = TRUE now shows the original labels (levels in the variable curveid) (thanks a comment from Radu Slobodeanu). 2012-01-23: Small scaling error in fitted method has been removed (after a report from Andreas Betz). 2011-12-28: Plotting for event-time data has been improved. 2011-12-21: Error in calculation of ED values for fplogistic() has been fixed. 2011-12-16: Help page of drm() (type argument) has been updated (after comment from Radu Slobodeanu). 2011-12-07: Error in label ordering and calculation of standard errors in SI() have been fixed (bug report by Andrew Kniss). 2011-11-06: Problem with confidence intervals in the predict method has been solved (was reported on R-help sometimes ago: 2010-11-28). 2011-10-27: Help page of mixture() has been slightly improved. 2011-09-20: Problem with logDose argument in drm() and subsequent plotting has been solved (reported by Ralf Schäfer). 2011-09-02: Constrained estimation uses the actual limits provided (bug report by Andrew Kniss). 2011-08-31: The functions genLoewe(), genLoewe2(), genBliss(), genBliss2(), and ursa() have been improved w.r.t. starting values. The model specification has also been simplified. 2011-08-24: genBliss() and genBliss2() has been updated (after suggestions from Hugo Ceulemans). Functions iceLoewe1() and iceLoewe2.1() have benn added. 2011-08-23: Error in ED calculation for the logistic models (e.g., L.4()) has been fixed (after a report from Daniel Gerhard). The error was caused by an update of the function deltaMethod() in alr3 has been fixed. 2011-07-04: Small changes for the anova method. Adding df for event time and Poisson models. 2011-07-01: Small update in the calculation of confidence intervals (after a suggestion from Scott Ray). 2011-06-24: The model function W2x.4() has been added (after a suggestion by Cécile Cornou). 2011-05-24: Small improvement in the calculation of ED value for the log-normal and Weibull type 2 models. "display" and "type" arguments have been added to maED(). 2011-05-19: The argument "type" can now also take the value "event" for fitting event times. 2011-05-10: The help page for spinach has been improved. 2011-05-02: Simplifying the R lines in print.summary.drc() for printing the summary output of a model fit. 2011-05-02: The self starter function for llogistic models can now handle infinite dose values (after a report from Marc Weimer). 2011-04-09: multdrc() has been completely removed. 2011-04-04: The model function weibull2x() (a model including a sort of "lag time" parameter) has been added (after a suggestion by Cécile Cornou). 2011-03-23: A small bug in ED() has been fixed (after a report from Nathan Pace). 2011-03-09: The BIC method has been removed (after suggestion from Prof. Brian Ripley). 2011-02-04: The functions isobole() and mixture() have been modified in order to fix an error in the plotting of the isoboles (after a report from Andreas Betz). 2010-12-16: The model function genursa2() has been replaced by actimL(). 2010-12-14: The model genursa2() for fitting the generalized URSA model has been included (after an idea by Hugo Ceulemans). 2010-12-10: The model genursa() for fitting the generalized URSA model has been included (after an idea by Hugo Ceulemans). 2010-12-09: The model function genBliss(), genBliss2(), and genLoewe2() for fitting generalized Bliss independence and Loewe additivity with different maxima has been included (after an idea by Hugo Ceulemans). 2010-11-23: The model function genLoewe() for fitting generalized Loewe additivity has been included (after an idea by Hugo Ceulemans). 2010-11-16: Small improvement in the summary output. Now the actual curve names are used. 2010-11-11: A small bug in ED() (mismatch of curve names and parameter estimates) has been fixecd (after a report from Andreas Betz). 2010-11-03: A small bug in the plot produced by isobole() has been fixed (after an inquiry from Andreas Betz). 2010-10-12: A bug in the calculation of standard errors in SI() has been fixed (after a report from Andrea Onofri). 2010-09-23: mselect() has been extended to include an argument for specifying the type of information criterion to use. 2010-09-20: maED() has been extended to include simple linear regression. 2010-09-16: mselect() has been extended to include a few standard polynomial regression models. 2010-08-03: Studentised residuals are now also available for binomial responses (after an inquiry from Stuart Rosen). 2010-06-04: Studentised residuals are now also available (after an inquiry from John). 2010-06-02: ED values are now correct for cedergreen() models (after a bug report by Claire Della Vedova). 2010-05-31: Small bug in plot.drc() related to "xt" and "xtlab" arguments has been fixed (after a comment by Anja Coors). A small bug in comped() has also been fixed (after a question from Jochen Zubrod). 2010-05-06: The argument "clevel" has been added to ED(). The function maED() has been extended and now also handles model fits involving several curves (after a suggestion from Andre Kleensang). 2010-04-29: Dataset 'ecvam' has been removed. The mixdrc() function has been temperarily removed. 2010-04-23: Minor bug in the predict method has been fixed. This means that type="bars" in the plot method now works again (after a bug report by Andy Robinson). 2010-04-16: The help for NEC() has been improved (after a question from Inés González). 2010-04-15: The comped() has been re-introduced (after suggestion from Jochen Zubrod). 2010-03-17: logLik has been extended with a "nobs" attribute (provided by Tobias Verbeke). A S4 BIC method has been added (also provided by Tobias Verbeke). 2010-03-12: The help page of drm() has been improved a bit w.r.t. the use of weights (after a question by Xuesong Yu). 2010-03-10: A slight modification in vcov.drc() to suppress unnecessary error messages (reported by Xuesong Yu). 2010-01-26: A slightly different bisection method has been implemented in ursa(). The corresponding help page has also been extended (more example lines). 2010-01-13: The model function ursa() for describing combination effects has been added (after an idea by Hugo Ceulemans). The help page for MM.2() and MM.3() has been improved. 2010-01-11: The model function cedergreen() has been updated to ensure correct calculation of ED values (reported by Claire Della Vedova). 2009-11-17: The help page of drm() has been updated. 2009-10-13: confint has been improved to automatically use the appropriate reference distribution for the confidence intervals; Internal structure of ED() has also been modified (reported by Marc Weimer). 2009-10-12: The self starter for twophase() has been slightly improved. 2009-09-03: The functions NEC.2(), NEC.3(), NEC.4() for estimation of no effect concentration have been included (after an idea by Ralf Schaefer). 2009-09-03: The function comped() has been removed. 2009-09-02: The argument "extended" has been added to the function maED(). 2009-09-01: The model function twophase() based on log-logistic models has been added (after an idea by Ida Katarina Auf der Maur Hindrichsen). 2009-07-28: Small bug in cedergreen() and ucedergreen() related to calculation of ED values has been fixed (reported by Clare Della Vedova). 2009-07-27: Small bug in the function modelFit() has been fixed (after feedback from Heike Schmitt). 2009-07-27: Help page for CRS.5a has been improved (after comment from Claire Della Vedova). 2009-07-06: The functions lin.test(), mr.test(), and neill.test() have been added. 2009-06-30: The data frame etmotc has been included. 2009-06-08: The argument 'display' (with same functionality as in Ed()) has been added for compParm() (after a suggestion from Scott Ray). 2009-06-02: The help pages for anova.drc and BC.4(), BC.5(), CRS.5.() have been improved. 2009-05-29: The dataset ecvam has been included (migrated from the package 'mrdrc'). 2009-05-28: The function maED() for parametric model averaging has been added. 2009-05-27: The functions fplogistic() and FPL.4() enable fitting dose-response models based on fractional polynomials. The function maED() can be used to do model-average based estimation of ED values. 2009-05-24: Structure of self starter functions has been completely revamped. Four initial value procedure are now available for almost all implemented dose-response models. 2009-05-23: The function getInitial() has been included. 2009-05-20: Help page for ED() has been improved (after comment from Claire Della Vedova). 2009-05-07: Bug in ED() concerning the arguments type="absolute" and reference="upper" has been resolved (reported by Yue Zeng-Li). 2009-04-22: Help page for gompertz() has been improved slightly. 2009-04-08: Help page for earthworms dataset has been improved. Error in likelihood calculation for some binomial models has been fixed. 2009-04-07: mixture() also works for binomial data. Argument "legendCex" in the plot method renamed into "cex.legend" in line with other cex arguments. 2009-04-01: Minor internals changes in drm() and in the plot method. 2009-03-27: The convenience functions b.3(), B.3(), b.4(), B.4(), b.5(), B.5(), and boltzmann have been removed. Use L.3(), L.4(), and L.5() instead. compParm(), ED(), SI() have been restructured. drm() has been extended to allow fitting models separately for each curve. confint, summary, vcov, ED, MAX, SI have a new argument "pool" to allow pooling of separate fits. Argument "ci" in relpot() and SI() has been renamed into "interval". modelFit() now also works for binomial data. 2009-03-25: The dataset "algae" has been included. vcov method has been re-structured. 2009-03-24: The function plotraw() has been removed. Initial plots of the data should be done using R's standard plotting functionality, e.g. plot and xyplot(). 2009-03-19: The argument "conLevel" has a more sensible default (no longer hardcorded at 0.01). The argument "xsty" has been introduced to control the arrangement of tick marks on the dose axis. 2009-03-17: The function yieldLoss() has been included to handle a different parameterization of the Michaelis-Menten model (after a suggestion by Andrew Kniss). 2009-03-11: The argument "fctList" has been removed from drm(). 2009-03-06: mixture() has been completely revised with a lot of changes to the arguments. The model function richards() has been removed as it is simply a different parameterization of the five-parameter log-logistic model. colFct() has also been removed. 2009-03-04: A new function modelFit() has been introduced for assessing the model fit, partly replacing the anova method. 2009-02-27: Bug in mselect() has been fixed (reported by John Lewis). 2009-02-25: Bug in boxcox.drc has been fixed. Moreover, this method has been extended to include functionality previously available through drm(). 2009-02-19: multdrc() and associated mdControl have been taken completely out of use. drm() has been improved w.r.t. handling extremely small or large dose or response values. Arguments "lowerc" and "upperc" have been removed in model functions as they were redundant. 2009-02-13: Argument "ci" renames into "interval" in ED(). 2009-02-12: Bug in predict.drc fixed (thanks to Mario D'Antuono). 2009-02-09: Redundant encoding removed. Argument "fixed" added to BC.4() and BC.5() (thanks to Nina Cedergreen). 2008-12-02: Dataset "H.virescens" added. Asymptotic regression and exponential decay implemented differently. 2008-11-26: "lnormal" function has been added plus three new datasets. 2008-11-18: "gompertz" function has been added. 2008-11-13: Error in level argument in plot method has been fixed. 2008-11-03 Datasets 'lepidium' and 'nasturtium' have been added. 2008-10-31: Bug in 'level' argument in plot method has been fixed. 2008-10-02: New function mrdrm() for model-robust modelling included. Accompanying ED and predict methods also added. 2008-09-30: The lettuce dataset got correct row numbers.