plot.see_parameters_model: option to turn off group colours for dot-and-whisker plots

[egouldo]

Can the colours for CI groups (< y_intercept / > y_intercept, ref'd below) be made optional? It's not clear to the reader what the colour represents. For these type of coefficient plots I think colour is typically used to represent the significance level of the estimate, so I had to cross-check against the raw values to realise this wasn't the case.

Also, documentation could be amended to explain this behaviour - I had to toggle off/on the legend and then dig into the function code before I understood how the colours were determined.

see/R/plot.parameters_model.R

Lines 359 to 375 in 622dcb3

	# plot setup for model parameters with multiple CIs
	x$CI <- as.character(x$CI)
	x$group <- factor(x$Coefficient < y_intercept, levels = c(FALSE, TRUE))
	if (all(x$group == "TRUE")) {
	color_scale <- scale_color_material(reverse = TRUE)
	} else {
	color_scale <- scale_color_material()
	}
	p <- ggplot2::ggplot(x, ggplot2::aes(
	y = .data$Parameter, x = .data$Coefficient,
	size = rev(.data$CI), color = .data$group
	)) +
	ggplot2::geom_vline(ggplot2::aes(xintercept = y_intercept), linetype = "dotted") +
	theme_modern(legend.position = "none") +
	color_scale

[nlubock]

Agreed - this is quite confusing for undocumented behavior. Would be great if this was noted or removed!

[egouldo]

Awesome, thanks for your work on this @strengejacke !