[R-meta] Facing some issues on Meta - regression

[Dr. Guido Schwarzer]

Hi Gerta,

As I wrote, in bubble.metareg() the transformed y-axis with appropriate tick marks and back-transformed labels only works for relative effect measures, i.e., OR, RR, HR, DOR (and since the last update with summary measures PLN, MLN and IRLN). In these cases, the scatter plot is generated with the argument 'log = "y"' and R generates appropriate tick marks and labels. See, for example,
plot(1:10, 1:10, log = "y")

I do not know how to automatically generate 'reasonable' tick mark positions for other non-linear transformation (let's say the arcsine or logit transformation). Maybe somebody has a suggestion?

It is now possible to suppress the generation of the x- or y-axis using argument 'axes', e.g.,
bubble(metareg(m, year), backtransf = FALSE, ylab = "", axes = "x")

Afterwards, the y-axis can be manually added, for example,
props <- c(0.05, 0.1, 0.2, 0.3, 0.4)
axis(2, at = p2logit(props), labels = props)
mtext("Proportion", 2, line = 2.5)

I could think of adding an argument 'labels', or similar, to bubble.metareg() to specify the labels for the tick marks on the y-axis. In the spirit of the previous example, bubble(mreg, labels = c(0.05, 0.1, 0.2, 0.3, 0.4)). 

Best,
Guido