In short, from a scientific perspective, this calculation is nothing more than a confirmation of a normal pattern under certain rules. In this tournament, the rule was to play with different races!
In general, such research is important to science, and I've always respected it.
The most interesting part here is the summarized results on game duration – this is very telling. The faster the game, the more dynamic the experience, and the more engaging it becomes.
The statistics on hero levels depending on the game duration are also interesting.
For tournaments, I would suggest revising the format of the game summary to include the hero levels in the final game! This is essentially important information.
For example:
73 Smerti_PO (Red, Undead, Kaspar - level 27) lost to Adv (Blue, Orcs, Haggeesh - level 33) - 1-3-6 - (the hero levels here are given as an example)