>>104805930>A 0.1% sampling rate survey of a >1M population size has a reasonable MoE of less than 3%.Only (and ONLY IF) the sample is homogeneous and representative of what is trying to measure.
If you have a 1000 people sample taken from the line of NijiFes you will have a completely different result from taking a sample taken from the line of HoloFes.
It is not only the size of the sample that matters but the quality of the sample.
The great majority of the audience or vtubers is male and by a huge margin.
Any poll on vtubers that gets a 50-50 male/female ratio will skew towards Nijisanji and away from reality because, in reality, there are a whole lot more men in the audience.
