Dude... seriously... I know what a plurality is. I also know its meaningless. The percentages tell the story. This isn't voting in a first-past-the-post election, where it would matter!
I agree with everything except the last sentence, bolded, which has no apparent basis for the "probable". At this point Beyond's audience is too large for claims like that to be inherently plausible, probable. If this was 5 years ago, and Beyond was new and only had 500k people or 1m or whatever, I'd agree, most people using it would be more "advanced" players. In which case I'd agree with "probable".
But this is 2022. Beyond has 10m+ users (compared to a worldwide total of either 30m or 50m D&D players, depending on how WotC is measuring today), which is frankly staggeringly, insanely high market penetration. Can you imagine? Even if it's only 10m that's utterly wild, to have 20% of millions of people who do a hobby using the same specific bit of entirely optional software. So we're now far beyond (npi) the "advance users" phase. I think this is obvious from experience too - I've met extremely casual adults who've barely heard of D&D, nine-year-old kids, and many others, who manage to find and use Beyond. So I don't think they "kind of people" argument holds any water at all, at this point, and I think you should admit that.
But I do offer you an olive branch here! Because whilst we must, I suggest, dismiss the "kind of people" argument, there is a valid argument you've skipped over. The "structural" argument, which does support your contention. If people come to Beyond before being introduced to rolling, then they're not really going to have much of a reason to pick manual input. And likewise, for people who want to create characters out-of-session, which I think is an increasing number of people, rolling doesn't really make sense. So these two things together create a structural situation which is pushing the popularity of Stat Array and 27-point Point Buy (the only kind allowed on Beyond) up, artificially. And it may mean that Beyond's users use them more than people who don't use online platforms at all. The same pressures would apple to Roll 20, Foundry, etc. etc. so anyone using online tools or a VTT, or just wanting a character builder is somewhat more likely to use them than roll, due to this. Whether that is statistically significant, and what the long-term impact will be remains to be seen, but I strongly suspect rolling is on the decline. I mean, indeed, all the evidence we have suggests that. The question is how steep, really.
Also, let's be real re: the history of "variant" methods. D&D has a long history of variant methods actually being the dominant method, going back to at least 2E. I mean, who actually used Method I for a significant part of AD&D 2E's life? I literally never came across or even heard of anyone using it in the time 2E was "live". Much later, when people started doing OSR stuff, suddenly there was a lot of talk about 3d6 in order, but not at the time. Even the 1E players I knew didn't use it. There's no question that 4d6DtL, arrange to taste was the dominant method by, say, the very very early '90s. That's Method V. It's not even the first alternative! But it was dominant.