Sometimes some authors leave the units blank, especially on the bar charts and tables. How exactly is the best approach seeing that the measure of the reported information has changed due to data transformation.
If you use statistical test like F and T or other statistics based on Normal distributions, first you should check the normality status of your data (For instance use Shapiro Test). If your collective data wasn't normal then you must use data transformation. Based on my experience in such case, data transformation brings clutter. So i suggest you use the non-parametric statistic in order to analysis your data then you don't need data transformation . Anyway, if you had to use data transformation the best way to represent you result is anti-transformation way. Meaning you must represent Anti-transformation form of your data (Like anti-LOG, anti-arcSine and so) .