Because those are means comparisons (statistics jargon) above and the relationship between mean serum TT level and weekly dose is pretty well known (and predictable). See my Dose-Response plots. Hence, weekly dose above is just proxy for a mean weekly TT level.
Summary: Tools to estimate your individual dose response for injectable testosterone ester in comparison to a normal distribution (95 percentile confidence estimate) of adult males (Ref). Plot below shows a summary of data from (1) clinical trials gathered from the literature + (2)...
forums.t-nation.com
The variability of TT level above vs mean dose is partly due to SHBG variation as well as the elimination of FT from the body (which is a function of clearance and apparent volume of distribution, etc...). This variation is then propagated into the means comparison way above. Hence you see the typical "error" or standard deviation bars on the bar charts. But if the means are really different you can still pick up "measureable differences" using standard statistics methods...mean's comparison, t-test, blah blah.