課堂上我們因為兩台Windows 10在執行老師的sjPlot製表語法時，同時出現錯誤，作出了「這是Windows10的問題」的推論。不過，這是個錯誤的推論。老師仔細看了看，原來是sjPlot進入了最新版2.0之後，所有的argument的寫法都統一調整了，導致老師去年使用的舊語法在套件更新後無法被正確執行。所以，請同學重新下載更新後的教材，並記住這一課：千萬不要輕易推論，也不要輕易接受別人的推論。事實是：那兩台Windows10的使用者才是最乖乖更新套件的人。其他同學就趁此更新套件及重新研讀一下講義吧。

sjPlot 2.0.0

Major changes

This package update includes a major revision of function arguments and their naming, in order to get a consistent argument pattern across all package functions. This means that your existing code, which uses sjPlot-package-functions, most likely needs adaptions to work again.

• Arguments were harmonized across all package functions. This includes refactoring of many function argument names, to get consistent argument names in functions (e.g. sort.coef now no longer exists, and was renamed to sort.est, which was already used by some other functions).

• Camel cased argument names were replaced by lowercase dot-separated names (e.g. showCI was renamed to show.ci) and harmonized bewteen different functions

• type arguments of sjp.lm, sjp.glm, sjp.lmer and sjp.glmer were harmonized, so that one type does the same in all functions. "pred" and "fe.pred" were renamed to "slope" and "fe.slope", "fe.ri" and "ri.pc" were renamed to "ri.slope", "resp" and "y.pc" were renamed to "pred" and "pred.fe".

• Arguments in functions were re-ordered and bundled according to their functionality (e.g., the variuous show... arguments now should appear on after another in the function and package manual).

General

• Improved label detection for sjp.lm, sjp.glm, sjt.lm, sjt.glm, sjt.lmer and sjt.glmer.
• Improved handling of different link functions for generalized linear (mixed) models (including negative binomial) for effect plots in sjp.glmer and sjp.glm.

Changes to functions

• Effect plots (type = "eff") for (generalized) linear (mixed) models (sjp.lm, sjp.glm, sjp.lmer and sjp.glmer) get a vars and facet.grid argument.
• Effect plots (type = "eff") get a ... argument, to pass down other arguments to the effects-package.
• Predicted values for response (type = "pred" or type = "pred.re") for sjp.glm, sjp.glmer, sjp.lm and sjp.lmer get a vars argument to specify x-axis and optional grouping variables. Furthermore, models from other model families and link functions (including negative binomial) now also work with this plot type.
• Functions sjp.lmer, sjp.glmer, sjt.lmer, sjt.glmer, sjp.lmm and sjp.glmm get a p.kr argument, to decide whether computation of p-values should be based on Kenward-Roger approximation or not (for very large data sets, it's recommended to set this argument to FALSE because it is very time consuming).

Bug fixes

• During code clean-up, argument group.pred did not work for sjt-functions in past update.
• Fixed bug with computation of confidence intervals and relative confidence intervals in sjp.frq.