Version 0.10.1
------------------------------------------------------------------------------
General:

* Revised imports: Labelled data functions from package *sjmisc* have been moved to package *sjlabelled*.

New functions:
* `boot_est()` to return the estimate from bootstrap replicates.

Changes to functions:
* The `print()`-method for `svyglm.nb()`-objects now also prints confidence intervals.

Bug fixes:
* `se()` did not work for `icc()`-objects, when the mixed model had more than one random effect term.

Version 0.10.0
------------------------------------------------------------------------------
New functions:
* `cv_error()` and `cv_compare()` to compute the root mean squared error for test and training data from cross-validation.
* `props()` to calculate proportions in a vector, supporting multiple logical statements.
* `or_to_rr()` to convert odds ratio estimates into risk ratio estimates.
* `mn()`, `md()` and `sm()` to calculate mean, median or sum of a vector, but using `na.rm = TRUE` as default.
* S3-generics for `svyglm.nb`-models: `family()`, `print()`, `formula()`, `model.frame()` and `predict()`.

Bug fixes:
* Fixed error in computation of `mse()`.

Version 0.9.0
------------------------------------------------------------------------------
General:
* Functions `std()` and `center()` were removed and are now in the sjmisc-package (https://cran.r-project.org/package=sjmisc).

New functions:
* `svyglm.nb()` to compute survey-weighted negative binomial regressions.
* `xtab_statistics()` to compute various measures of assiciation for contingency tables.
* Added S3-`model.frame()`-function for `gee`-models.

Changes to functions:
* `se()` gets a `type`-argument, which applies to generalized linear mixed models. You can now choose to compute either standard errors with delta-method approximation for fixed effects only, or standard errors for joint random and fixed effects.

Bug fixes:
* `prop()` did not work for non-labelled data frames when used with grouped data frames.

Version 0.8.0
------------------------------------------------------------------------------
New functions:
* `svy()` to compute robust standard errors for weighted models, adjusting the residual degrees of freedom to simulate sampling weights.
* `zero_count()` to check whether a poisson-model is over- or underfitting zero-counts in the outcome.
* `pred_accuracy()` to calculate accuracy of predictions from model fit.
* `outliers()` to detect outliers in (generalized) linear models.
* `heteroskedastic()` to check linear models for (non-)constant error variance.
* `autocorrelation()` to check linear models for auto-correlated residuals.
* `normality()` to check whether residuals in linear models are normally distributed or not.
* `multicollin()` to check predictors in a model for multicollinearity.
* `check_assumptions()` to run a set of model assumption checks.

Changes to functions:
* `prop()` no longer works within dplyr's `summarise()` function. Instead, when now used with grouped data frames, a summary of proportions is directly returned as tibble.
* `se()` now computes adjusted standard errors for generalized linear (mixed) models, using the Taylor series-based delta method.

Version 0.7.1
------------------------------------------------------------------------------
General:
* Package depends on R-version >= 3.3.

Changes to functions:
* `prop()` gets a `digits`-argument to round the return value to a specific number of decimal places.

Version 0.7.0
------------------------------------------------------------------------------
General:
* Largely revised the documentation.

New functions:
* `prop()` to calculate proportion of values in a vector.
* `mse()` to calculate the mean square error for models.
* `robust()` to calculate robust standard errors and confidence intervals for regression models, returned as tidy data frame.

Version 0.6.0
------------------------------------------------------------------------------
New functions:
* `split_half()` to compute the split-half-reliability of tests or questionnaires.
* `sd_pop()` and `var_pop()` to compute population variance and population standard deviation.

Changes to functions:
* `se()` now also computes the standard error from estimates (regression coefficients) and p-values.

Version 0.5.0
------------------------------------------------------------------------------
New functions:
* Added S3-`print`-method for `mwu()`-function.
* `get_model_pval()` to return a tidy data frame (tibble) of model term names, p-values and standard errors from various regression model types.
* `se_ybar()` to compute standard error of sample mean for mixed models, considering the effect of clustering on the standard error.
* `std()` and `center()` to standardize and center variables, supporting the pipe-operator.

Changes to functions:
* `se()` now also computes the standard error for intraclass correlation coefficients, as returned by the `icc()`-function.
* `std_beta()` now always returns a tidy data frame (tibble) with model term names, standardized estimate, standard error and confidence intervals.
* `r2()` now also computes alternative omega-squared-statistics, if null model is given.

Version 0.4.0
------------------------------------------------------------------------------
New functions:
* `inequ_trend()` to calculate proportional change of absolute and relative inequalities between two status groups for a vector of given prevalence rates.

Changes to functions:
* `bootstrap()` is now much more memory efficient due to use of pointers.
* `boot_ci()`, `boot_se()` and `boot_p()` now accept multiple variables as input.
* `resp_val()` now also applies to models fitted with `nlme::lme()`.

Version 0.3.0
------------------------------------------------------------------------------
General:
* Removed non-necessary checks for package-availability.

New functions:
* `bootstrap()` to generate bootstrap replicates of data frames.
* `boot_ci()` to compute confidence intervals from bootstrapped values.
* `pred_vars()` to get the names of predictor variables from fitted models.
* `resp_var()` to get the name of the response variable from fitted models.
* `resp_val()` to get the values of the response vector from fitted models.

Version 0.2.0
------------------------------------------------------------------------------
New functions:
* Added functions `weight()` and `weight2()` to weight vectors.
* Added functions `wtd_sd()` and `wtd_se()` to compute weighted standard deviations and standard errors.
* Added function `merMod_p()` to compute p-values for merMod-objects.

Changes to functions:
* `r2()` now supports `plm` objects.

Bug fixes:
* Fixed typo in print-method for `icc()`.

Version 0.1.0
------------------------------------------------------------------------------
General:
* Initial release on CRAN.
