You can conclude very little from the poll.

1. You don't know how many people are participating.
2. You don't know if they're telling the truth.
3. As you say, you don't know their media.
4. You don't know how many people are submitting/polling to/for each agency. If you wanted to know which was 'best', everyone would need to be submitting and polling on each of them. If someone only submits to e.g. four agencies, the poll results don't give an accurate comparison.
5. Plus many people report on the forum that they are holding their ports at some agencies, and only actively uploading to certain others. The poll can't reflect this.
6. And, for example, I don't poll my Alamy earnings, because if I did, my iS poll would move from exclusive to indie, which would be inaccurate.
7. You don't know how big the submitters portfolios are, or how long they've been submitting.

Anyway, the answer to your questions is 'it depends on the photos/videos/vectors' (e.g. info I have from within and outwith msg suggests (not proves) that SS does much better for US-based photographers than UK-based ones) and from the forum it seems that Ft does better for Europeans. Forum posts seem to indicate that for several people, the results are fluid.
It's not an exact science.

Poll Results at the right are more accurate for photographers.  I'm a vector artist and I got #1.SS, #2.IS and #3.FT. I never tried to upload to Low Earners

You can conclude very little from the poll.

1. You don't know how many people are participating.
2. You don't know if they're telling the truth.
3. As you say, you don't know their media.
4. You don't know how many people are submitting/polling to/for each agency. If you wanted to know which was 'best', everyone would need to be submitting and polling on each of them. If someone only submits to e.g. four agencies, the poll results don't give an accurate comparison.
5. Plus many people report on the forum that they are holding their ports at some agencies, and only actively uploading to certain others. The poll can't reflect this.
6. And, for example, I don't poll my Alamy earnings, because if I did, my iS poll would move from exclusive to indie, which would be inaccurate.
7. You don't know how big the submitters portfolios are, or how long they've been submitting.

Anyway, the answer to your questions is 'it depends on the photos/videos/vectors' (e.g. info I have from within and outwith msg suggests (not proves) that SS does much better for US-based photographers than UK-based ones) and from the forum it seems that Ft does better for Europeans. Forum posts seem to indicate that for several people, the results are fluid.
It's not an exact science.

No poll is 100% right, all have a margin of error. Yet polls are widely used, in all domains, to identify trends.

A statistician could quantify the error margin and minimize it, by asking more questions about each respondent's background.

But this simple poll is better than no poll.

I too find it fairly accurate, the only exception, for me, being 500px, which ranks higher than what I see here (4th)