Now imagine a publication that only presents the results of the other 3%.
Also, it is somewhat ironic that you chose doctors as an example, as this is the area where it is common practice to get a second opinion when presented with a material conclusion.
A second opinion doesn't require a 20 person sample size to have a a reasonable chance of landing on someone who shares your views.
There is a reason it is to get a second opinion, not 20 opinions.
The basic problem here isn't the irony you perceive but the fact literally _nothing_ in modern medicine functions if you require 97% consensus.
So your argument is scientists are equivalent to journalists who solely exist to manipulate you?
You are free to stop using a computer, taking medication, and driving a car in that case.