Not really. VGTech excel sheet also shows a slight advantage. Why choose to focus on 95th percentile and not 99th percentile?
Because 99th percentile doesn't give the complete picture, especially when you look at the worst-case scenarios (50ms vs 33.33ms is a significant difference). The nth percentile value just tells us that n% of frame times fall below a certain frame time, whereas (100 - n)% are equivalent or above it.
The 95th percentile indicates that 5% of frame times on the Pro are equal to or above 16.67ms. The rest are below it, but since the game is capped at 60fps, those frame times are 16.67ms. The 99th percentile indicates that only 1% of frame times on the Pro are equal to or above 33.33ms, while 99% are between 16.67ms and 33.33ms. But since we already know that 95% of the frame times are exactly 16.67ms, it's just 99% - 95% = 4% that this applies to.
Meanwhile, 5% of the frame times on the XBX are equal to or above 33.33ms, meaning that the remaining 95% fall anywhere between 16.67ms and 33.33ms (we don't have the full percentile graph, so we can't tell at what point things begin to get more consistent for the XBX; maybe 90th percentile, maybe 85th, who knows). Either way, as we already know that the target frame time for a 60fps game is 16.67ms, this is inconsistent for gameplay.
Therefore, not a slight advantage. The worst frame times on both platforms alone are telling enough.
Even then, a ~3% difference is minor.
What metric is this 3% difference based on?
Also, VGTech has ought to include all the frame times data and not just the summary if they are going to just put out an excel sheet.
Agree with this, but the data provided is sufficient enough to know that frame times are less consistent on XBX, and not just slightly. Though the term "slightly" itself is kind of subjective.