NEWS

version 2.3.0
- Updated the FAQ
- Fixed an issue regarding incomplete covariates
- Added scaled index in addition to "formal"
- Added 'long' output in index() and totals()
- Added Trendlines() to process overall() output
- Fixed a bug in autodelete()
- Removed spurious option "imputed" in plot.trim.overall() *)
- Fixed a bug that resulted in negative lower bounds for confidence intervals
  in extreme cases (huge overdispersion) *)
- Fixed issue reporting wrong package version *)
- Fixed documentation of plot.trim.overall which erroneously mentioned that for
  each time point confidence intervals are plotted rather than standard errors *)

*) Thanks for Tomas Telensky for pointing at these issues.

version 2.1.1
- Fixed incompatibility with R 4.0

version 2.0.7
- Improved error messages for invalid beta values
- Improved the use of actual year numbers as change points
- Improved documentation for overall() function

version 2.0.6
- Removed spurious output from index() function
- Fixed bug in totals() which prevented both obs=TRUE and level!=NULL

version 2.0.5
- Fixed bug causing non-zero standard errors for the base year index if this isn't the first year (thanks to Tomas Telensky)

version 2.0
- Many major updates (monthly data; stratified trim; advanced visualizations); See the vignette "trim 2 extensions".

version 1.3.1
- Fixed bug concerning covariates
- Fixed typos and minor edits in vignettes (thanks to Martin Poot)

version 1.3.0
- Integration of experimental monthly version

version 1.2.0
- Introduced time-window-based indexing
- Automatically converts non-factor covariates to factors
- Fixes a bug occuring when empty sites are combined with covariates (thanks to Marnix de Zeeuw)
- Speed improvements
- Minor bug fixes

version 1.1.5
- Now generates an error when the computed overdispersion is 0, and a warning when overdispersion < 1 (thanks to Oscar Ramírez for proving a data set where this happened)

version 1.1.4
- Fixed a bug caused by auto-removal of empty factorial sites (thanks to Tomás Murray)

version 1.1.3
- Fixed a bug resulting in wrong p-values output by the overall() function (thanks to Anna Marie Holand for reporting).

version 1.1.2
- Fixed problem when to many years have no or 0 counts, such that the expected value becomes effectively 0 as well. This case now generates an error message “Zero expected value” (thanks to Arco van Strien for proving this case)

version 1.1.1
- Fixed problem during stepwise regression when the algorithm wants to remove the first changepoint (thanks to Asaf Tsoar)
- Fixed computation of standard errors when model==1 (thanks to Asaf Tsoar)

version 1.1.0
- Added backward compatibility option in overall() to match erroneous overall trend interpretation in original TRIM (thans for Lars Petterson to pointing out the differences in trend interpretation)

version 1.0.2
- Limit on regression parameters now equal to original TRIM (thanks to Lars Petterson)

version 1.0.1
- initial CRAN release