Closed nabury closed 7 months ago
I think the solution is a dynamic number of digits in the round function here:
scale_y_continuous(labels = function(x) round(plogis(x), digits = 2)).
Actually signif(plogis(x), digits = 2))
will do the job.
In the text underneath I would prefer this: The plotted baseline risk value is the same for all treatment arms across a study and is the outcome in the reference treatment arm.
In the binary case I prefer this text for the error bars: Error bars: logit(outcome) +/- 1.96 * se(logit(outcome))
In both continuous and binary we can delete "it is assumed that the data are normally distributed", because anyone interested will be able to tell from the +/- 1.96.
Changes made as suggested
I don't suppose the study names on the x-axis can be better aligned? They're too far to the right.
Will look at this in the separate BUGSnet task
All my requested changes look good, but the download doesn't work for me.
All my requested changes look good, but the download doesn't work for me.
Good spot. Should be fixed now.
Note the change in path for BUGSnet when running locally
When I loaded a dataset with a covariate, the text underneath the covariate graph was wrong, because it said the value displayed was the outcome in the reference arm, which is the baseline risk text. The dataset I loaded was "Cont_long_continuous_cov".
For both baseline risk and covariate, the reference arm only is plotted, so the caption text should be the same (except for covariate & baseline being changed as appropriate)
Also the text underneath the baseline risk graph was going off the page to the left, although it was fine in the download.
This is a known issue to be fixed in the plot rotation
Includes:
Known issues: