Couple of things, right? Firstly how much importance do you give to performance against the best when you rate players. For example Lara has pretty pathetic stats against India (and quite honestly gets away with it). If his stats against India and Australia were swapped, and his career stats were still same, would you still rate him as high? I certainly wouldn’t.
Secondly, ultimately it’s the individuals that make up an attack. For example - do you think Aussie attacks from mid-90s to mid-2000s were at the same level with & without McGrath? If you think they were different, how different were their levels?
If you think presence of just one individual (McGrath in this case) is not enough to make an overall difference to the ENTIRE attack, then “average vs bowler in the team” is trash for you.
If, in your mind, absence of just one player (McGrath) significantly reduces the overall strength of the ENTIRE attack from, for example, “Australia of 2000s” level to “Australia of 80s” level then the “average vs bowler in team” could be important to you.
So second question is how much weight you give to the impact of just one individual bowler on the ENTIRE attack of the team.
Another example is if Marshall missed a series or two in 80s West Indies, it probably would NOT have made a significant difference (from the match context as well as difficulty of batting context) because West Indies of the 80s had such a phenomenal overall attack
Obviously the stat is useless to draw any conclusions unless the sample size is large enough (at least >= 15 innings) for both cases, AND all in the same era. In India vs Australia case from mid-90s to mid-2000s, this particular factor should not be an issue because you have decent sample sizes, with and without McGrath
Again, while certainly accepting of the opinion that Lara > Sachin, the problem I have with this very particular methodology is that it is one very specific bowler (and thus, for Sachin, a much smaller sample size than for Lara), and again, ignoring the other supposedly world class ATG bowler that the very same team had!
While there is merit to the idea (testing vs the best), it's a very limited test instead of an overall performance of how the player did if all we include is just the one bowler, as good as he was (Steyn was arguably better anyway, and Ambrose and Donald weren't far behind if at all).
Conversely, I showed in an earlier post that if you take the top 10 bowlers of Sachin's time (a MUCH larger and varied sample size) and see his average vs them, it's certainly lower than his overall average, but not by much. And in his 1993-2002 peak period, it's around about his overall average (though a dip from his 63 or so peak around that period).
The point being, Lara's average is around about the same as Sachin's when you apply that exact filter for him (removing Steyn, because he wasn't playing when Lara was, and obviously his own countrymen, and adding in Kumble instead - therefore an easy argument to be made that this is already therefore considerably weaker...).
So in short, against the very best bowling Sachin faced, his average is the same as Lara's is.
Against a very specific bowler (or two?), it's lower. Which, without going into too much detail, this also suggests its higher vs another bowler or two than Lara's is.