Univariate and multivariate methods to analyze randomized response
(RR) survey designs (e.g., Warner, S. L. (1965). Randomized response: A
survey technique for eliminating evasive answer bias. Journal of the
American Statistical Association, 60, 63–69,

- Bugfix for likelihood ratio test (LRT) in RRlog: refitting nested model if difference between Wald's-Chi^2 and LRT-Chi^2 is larger than one.
- Improved documentation of summary output for RRlog

- Bugfixes and consistent behavior of predict.RRlog() for the argument: type=c("link", "response", "attribute")
- Bugfix for plot.RRlog and possibility to use type=c("link", "response", "attribute")

- Citation updated: Heck, D. W., & Moshagen, M. (2018). RRreg: An R package for correlation and regression analyses of randomized response data. Journal of Statistical Software, 85 (2), 1-29. doi: 10.18637/jss.v085.i02

- Bugfixes in predict.RRlog()

- Improved stability of RRmixed(): Better starting values and estimation options

- Fixed estimated standard error of univariate prevalence estimates (deviding by n-1 instead of n)
- Fixed univariate estimates for Kuk's method

- New RR Model: Triangular Model
- Fixed image links in vignette
- Selective import of package functions
- Updated email address
- Updated vignette

- Bug fixes for RRsimu and RRuni

- New function RRmixed() to fit logistic RR regressions with mixed effects (random slope/intercept in hierarchical models) using RR data based on lme4
- New function getPW() to get misclassification matrices for the implemented RR designs
- New data set 'minarets' for demonstration (type: data(minarets) )
- Proper maximization in RRuni() to get ML estimates (instead of moment estimates)
- Possibility to use a "custom"" RR misclassification matrix in RRuni, RRgen, and RRlog

- New function plot.RRlog to plot predictions and confidence interval for a logistic RR regression
- Bugfix in predict.RRlog
- Random starting values in RRlog for fit.n=1
- Updated dependencies

- Logistic regression in RRlog now uses a combination of EM algorithm and gradient-based optimization (previously, only optim was used)
- Compute predicted values (incl. SE and CI) for logistic RR regression by predict.RRlog
- Bugfixes in RRlin for multiple nonRR predictors and infinite log-likelihood
- More detailed documentation (specification of p in RRuni, available models for RRlin)

- new function powerplot() to plot power of the three implemented multivariate RR methods
- bootstrapped p-values in RRcor and correlations bound to [-1,1]
- random starting values and repeated optimization in RRlin
- RRsimu: increased stability; RRlin included; generating data for RRcor and RRlog separately; estimation of power
- continuous mixture RR models: "mix.norm" and "mix.exp" added to RRgen, RRuni, and RRcor, "mix.unknown" added to RRuni and RRcor

- first stable release of RRreg
- data generation: RRgen, RRsimu
- statistical analysis: RRuni, RRlog, RRcor, RRlin
- one-group RR designs: Warner, Kuk, Mangat, FR, Crosswise UQTknown
- two-group RR designs: UQTunknown, CDM, CDMsym, SLD