Regarding the 3rd vs 4th gear debate, it can go on and on because we are comparing cars of different gearboxes, different power/drivetrain losses and so forth.
Rule of thumb is to never compare 2 cars of different drivetrains on the dyno; instead use 0-100, quartermile or top speed to compare their real world performance. Dynos can use complex calculations to compare both cars against a constant index on the computer, but this is always suspect to miscalibrations, temperature fluctuations, inaccurate gearbox data and etc.
IMO a dyno is better suited to comparing your car before-after modification and comparing between 2 cars of similar drivetrain configuration. The key here is to never take the values seriously, except when using one of the charts as a base index. What's more important are the changes in the HP curve after your modification, and perhaps compare it to a similar car which has also done mods.