Hi Shuanghe,

❝ I was interested in the method so several weeks back I played with R using data from EPAR (none of the articles gives sample size of the BE study so without going to the EPAR for this additional information you can not reproduce/validate their results at all) …

Yep. I wrote a package for this stuff. I think they were using Excel cause (a) everybody has it (b) of its flat learning curve and (c) – most important – Luther Gwaza* gave his “template” to the others.

❝ … and the difference in some comparison is 0.01% due to different rounding method used in Excel and R.

If you want to reproduce Excel’s wacky “commercial rounding”:

cround <- function(x, d) ((trunc(abs(x) * 10^d + 0.5)) / 10^d) * sign(x)
x      <- c(79.994, 79.995, 125.004, 125.005)
comp   <- data.frame(x = x, R = round(x, 2), Excel = cround(x, 2))
print(comp, row.names = FALSE)
x      R  Excel
79.994  79.99  79.99
79.995  80.00  80.00
125.004 125.00 125.00
125.005 125.00
125.01

❝ ❝ Semitransparent colors in Excel? I don’t think that’s possible.

❝ Haha, this is where you err... Actually you can do that in Excel (not that you should). At least I can in my 2013. It's ugly, but semi-transparent it is.

Well, my version is Excel 2000. BTW, navigate up to my OP and hit to clear the browser’s cache / reload. Is it possible in Excel to have the color dependent on the value (the lower, the darker)?

• Gwaza L. Adjusted Indirect Treatment Comparisons of Bioequivalence Studies. PhD Thesis: Utrecht. 2016. free resource.

Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes