Roy wrote:You are making an assertion based on what what you think the questioner wants to know, rather than following the logic of the questions as they were asked and considering that, maybe, they were asked in that way for a reason.
Source of the post
This way in your statistics the one who thinks the entire galaxy or universe is dead gives you more points than the one who partially believes in life. This way in your statistics the one who thinks the entire galaxy or universe is dead gives you more points than the one who partially believes in life. I hope you understand me now. This test is wrong, it doesn't work.
I've known FFT for many years. If he wanted to study the differences in how confidently people believe in life existing off of Earth now vs. it ever existing, I am confident he would have phrased the questions in a way that reflected that.
Source of the post
You don't need 20 points for this. You need only 3:
I don't believe.
If it's too many options for your liking, you could answer with 20, 10, and 0, respectively, and not have to change anything about the poll. But not everyone has such a limited view of what their confidence level can be as you do, as evident by the responses themselves.
In science, we use confidence levels all the time, and they can be any number between 0% and 100%. I am very confident for example that the expansion rate of the universe is accelerating, but I am not 100%
confident. The data support it at about 5-sigma or 99.99994% confidence. So there is a small chance it could be wrong. But it's much less likely to be wrong than if the confidence level was 50%, or even 99%.
Sure, FFT could have made each question have just 3 options. But that would hide a lot of potentially interesting information from the histogram -- exactly the kind of information that he might be interested in. Suppose one of the questions gave a bi-modal distribution with peaks at 14 and 6, but with very few responses between 8 and 12. That would be interesting, and you would not be able to see it if there were instead just 3 possible responses. You would instead just see one peak at "I don't know".
And, sure, there's a reasonable limit to how finely the scale should be divided. If it's too fine, then you get few duplicate responses. Deciding how to set the scale is both a science and an art, and in this case 0 to 20 makes some good sense.