Version 1.6
------------------------------------------------------------------------------
New functions:
* `rec_pattern` to generate recode patterns for the `rec` function.
* `drop_labels` to drop labels from values with zero-counts.
* `str_contains` to check whether a string contains another string pattern.
* `r2` to compute R-squared values for linear (mixed) models.
* `lbl_df` to create a labelled data frame, and related S3-generic print method for `lbl_df` objects.

Changes to functions:
* `cv` now accepts multiple arguments.
* `icc` now accepts multiple arguments.
* `weight` now also weights character vectors.
* `overdisp` now wraps `AER::dispersiontest` to also support simple glm's.
* Removed deprecated functions.

Bug fixes:
* Fixed bug in `ref_lvl`, where value labels were not correctly re-ordered for factors that had a `0` as level.
* Fixed bug in `rec`, where value labels were not automatically re-ordered when `x` was a numeric factor.


Version 1.5
------------------------------------------------------------------------------
New functions:
* `to_long` to gather (transform from wide to long format) multiple column groups at once.
* S3-method `model.matrix` for `nlme::gls` class.

Changes to functions:
* `std_beta` now supports `nlme::gls` models.

Bug fixes:
* `word_wrap` now removes `NA` values from `labels` before wrapping strings.
* `set_na` now drops unused factor levels that have been replaced with `NA` values.


Version 1.4
------------------------------------------------------------------------------
New functions:
* `to_dummy` to split vectors with more than two categories into 0/1-coded dummy variables.
* `overdisp` (based on [DRAFT r-sig-mixed-models FAQ](http://glmm.wikidot.com/faq)) to compute approximate overdispersion parameters for `glmer` models.

Changes to functions:
* `to_value` can now also deal with character vectors.
* `is_empty` now also checks non-character vectors for zero-length.

Bug fixes:
* Fixed bug with `std_beta`, which computed wrong standardized beta coefficients for factor variables in `lm`.


Version 1.3
------------------------------------------------------------------------------
General:
* `set_labels` and `add_labels` now make consistent use of values as `labels` attributes and value labels as `names` attribute of `labels` attribute (being in line with `labelled` class now).
* `remove_labels` was renamed into `remove_all_labels`.
* Removed deprecated functions and arguments.

New functions:
* `split_var` to split a vector into equal sized groups.
* `ref_lvl` to change the reference level of (numeric) factors.
* `remove_labels` to remove selected label attributes from a vector.

Changes to functions:
* `to_factor` gets a `ref.lvl` argument, to automatically change the reference level when converting numeric vectors into factors.
* `replace_na` gets a `na.label` argument to optionally set a value label for the NA value.
* `get_label` gets a `def.value` argument, which will be returned when vectors have no label attribute.


Version 1.2
------------------------------------------------------------------------------
New functions:
* `converge_ok` to test convergence for merMod-objects (mixed effects models).

Changes to functions:
* `weight` gets a `digits` argument.

Bug fixes:
* Fixed bug in `to_na`, where in some cases `is_na` attribute was not properly set.
* Fixed bug in `set_na`, where in some cases `is_na` attribute was not properly removed for character vectors.

Version 1.1
------------------------------------------------------------------------------
General:
* Updated namespaces to meet new CRAN namespace requirements.
* Renamed `add_labels` to `copy_labels`.

New functions:
* `get_na` to get value codes of missing values from labelled vectors (that have an `is_na` attribute).
* `to_na` to convert value codes of missing values from labelled vectors (that have an `is_na` attribute) into NA.
* `fill_labels` to add missing labels to non-labelled values of partially labelled vectors.
* `as_labelled` to convert vectors to labelled.
* `add_labels` to add additional value labels to a labelled vector.
* `zap_labels` and `zap_unlabelled` to convert (non-)labelled values into `NA`.
* `frq` to print summary and frequency tables for labelled-class objects.
* Added S3-methods `is.na` and `mean` for labelled-class objects.

Changes to functions:
* `std_beta` gets a `type` argument to compute standardized estimates following Gelman's approach by dividing estimates by two standard deviations.
* `se` now accepts fitted linear mixed model (from lme4) to compute standard errors for joint random and fixed effects.
* Added argument `attr.only` to `get_labels` to get value labels also of factor levels, if variable has no value label attributes.
* Argument `include.values` of `get_labels` has now two options for returning includes values.
* `get_labels` now supports vectors with string label attributes.
* `to_factor` and `to_value` better deal with vectors that don't need to be converted.
* `get_labels` gets a `include.non.labelled` argument to also return non-labelled values as label.
* `to_label` and `to_factor` get a `add.non.labelled` argument to also convert non-labelled values to labels.
* `set_labels` gets a `force.values` argument to add values without associated labels as labels, too.
* `set_na` gets a `as.attr` argument, so values are not converted to NA, but rather missing codes are added as `is_na` attribute.
* `get_values`, `to_label` and `to_factor` get a `drop.na` argument, to exclude values of missing codes from return value.
* Renamed argument `autoAttachVarLabels` of `read_spss` into `attach.var.labels`.
* Functions now better deal with mix of labelled and non-labelled values.

Deprecated:
* `set_val_labels` is deprecated, use `set_labels`.
* `set_var_labels` is deprecated, use `set_label`.
* `get_val_labels` is deprecated, use `get_labels`.
* `get_var_labels` is deprecated, use `get_label`.
* `to_fac` is deprecated, use `to_factor`.
* `std_e` is deprecated, use `se`.


Version 1.0.3
------------------------------------------------------------------------------
New function:
* `add_labels` to set back labels from subsetted data frame, or remove any label attributes from data frames.
* `remove_labels` to remove any label attributes from data frames or vectors.
* `get_values` to return values associated with value labels from labelled vectors.
* `replace_na` to replace `NA`'s with specific value (counterpart to `set_na`).
* `is_even` and `is_odd` to check whether values are odd or even.
* `is_empty` to check whether a string is empty or not.

Changes to functions:
* `dicho` now also dichotomizes non-numeric values.
* `rec` now can keep (copy) not yet recoded values with `else=copy`.
* `get_val_labels` gains a `include.values` argument to also return values associated with the value labels.
* `get_val_labels`, `get_var_labels`, `set_val_labels` and `set_var_labels` now also accept `list`-objects with variables.
* `dicho`, `rec`, `set_na` and `recode_to` now also accept `list`-objects with variables.
* `set_val_labels` gets argument `force.labels` to force using all labels, even if length of labels if longer than unique values of vector.
* `rec` gets a `asFac` argument, to return recoded variable as factor.
* `dicho`, `group_var` and `group_labels` keep variable label attributes.
* `rec` and `dicho` get `varLabel` and `valLabels` arguments, to optionally add value and variable label attributes for recoded or dichotomized variables.

Bug fixes:
* `get_var_labels` returned NULL if first variable in data.frame had no variable label (but other variables had) - fixed.
* Fixed code with non-exact matching of `haven`-attributes `label` and `labels`, which in certain situation may return wrong vector attributes.
* `rec` did not recode `NA`s into values, if followed by `else`-token.
* `to_value` no longer drops unused factor levels.

Version 1.0.2
------------------------------------------------------------------------------
New functions:
* `icc` to compute intraclass-correlations for random-intercepts of mixed models.
* `hoslem_gof` to perform a Hosmer-Lemeshow-Goodness-of-Fit-test for logistic regression models.
* `cod` to compute the Coefficient of Discrimination, aka Tjur's Pseudo-R2, for logistic regression models.
* `pseudo_r2` to compute the Nagelkerke's and Cox-Snell's Pseudo-R2 for logistic regression models.

Changes to functions:
* Functions `std_beta` and `cv` now support `merModLmerTest` objects (fitted by `lmerTest` package).
* `mean_n` has a `digit` argument to round returned mean values.
* `rec`, `recode_to` and `std_e` now also accept data frames as argument.
* `chisq_gof` now accepts `glm`-objects, however, computing the Chi-squared-Goodness-of-Fit-test for logistic regression models sometime may fail.
* `set_val_labels` and `set_var_labels` can now remove label-attributes.
* `set_na` removes label-attribute, if removed `NA` values were the last labelled values.

Bug fixes:
* `set_na` did not work with logical vectors - fixed.
* `recode_to` did not work when `var` had value-label-attributes - fixed.

Version 1.0.1
------------------------------------------------------------------------------
New functions:
* `is_crossed` to check whether two factors are crossed.
* `is_nested` to check whether two factors are nested.
* `is_num_fac` to check whether a factor has only numeric levels.

Changes to functions:
* `std_beta` now accepts plm-objects.
* `to_value` now auto-detects lowest minimum value of numeric factors, instead of always setting minimum value to 1.
* `set_na`, `dicho`, `to_fac`, `to_value` and `to_label` now also accept data frames as argument.

Bug fixes:
* `std_beta` did not work in some cases - fixed.
* `mwu` did not accept factors as grouping levels - fixed.
* `mwu` did not work when value range of `grp` vector was not continuously - fixed.
* `cv` did not work with lme-objects (from nlme-package) - fixed.
* `cramer`, `phi` and `table_values` did not work with tables of class `xtabs` - fixed.
* `to_value` (resp. `set_val_labels`) did not work in certain situations when factor had not continuously numeric levels - fixed.


Version 1.0.0
------------------------------------------------------------------------------
General:
* First release of this package - it contains all utility, recode and statistical test functions from the `sjPlot` package.

New functions:
* `rec` to recode variables.
* `cv` to compute coefficient of variance.
* `rmse` to compute root-mean-square error.

Changes to former sjPlot-functions:
* Improved internal management of imported data via `haven` and `foreign` packages, so users don't need to care of structure and classed of data read with either haven, foreign or sjPlot's read-functions.
