BillHoyt said:
I am addressing the relevant issues, sir.
Ok, fair enough.
You continue to deflect at every juncture.
You caught me! And, well, since I'm already deflecting at junctures, I'll take the time to deflect extra at this juncture:
- Do you think your choice of analyzing the J counts (and counting them multiple times for the same instance), and your choice of analysis and interpretation of the analysis, for the JE transcripts is completely free of all types of bias?
- Do you claim that it is not possible for you to impart your unconscious bias into it anywhere?
- Do you claim to not have unconscious bias?
- Do you think if you were presented with say 4 transcripts, 2 each from a well-known medium, say 'Margey Madeline' (made up name) and 2 each from a person, say 'Skeeter Jones' (made up name) who doesn't claim to be a medium, but their names were changed from Margey Madeline and Skeeter Jones to 'Person 1' and 'Person 2', and you analyzed the transcripts, and commented on them, do you think your analysis and interpretation would be improved from what you were doing before being blinded?
You have been asked repeatedly and by multiple parties here to substantiate your unconscious bias claim vis-a-vis model and test choice.
Multiple parties, ie. you and Claus.
Moreover, I did present the Fisher's Exact vs Chi Square test as an example, as well as talked about how an evaluator can impart his/her bias in general, and presented sites with information on unconscious bias as well as online tests to detect/measure it.
Your first response was to compare a two-tailed test with a one-tailed test. You then corrected yourself with a Fisher's Exact of.0237, compared to a Chi-Square of .0179.
You may think that it went unnoted that the test results are the same, sir, but it did not.
I don't understand your last sentence above. The test results are most certainly not the same if alpha = .02 as I stated. In one case you reject the null hypothesis (because the p-value is less than .02), and in the other you fail to reject the null hypothesis (because the p-value is greater than .02).
How is that "the same"? It is only the same conclusion (of rejecting the null hypopthesis) if alpha is greater than .02, which it wasn't because I set it to be .02.
The cut-offs chosen a priori are typically .01, or .05 or .001. Not .02. These two tests do not result in different decisions, do they sir?
Alpha can be set to anything in real life. Find me any mathematical theory that says that alpha is always .05 or .01 (or any value for that matter).
While it is true that the commonly used alphas are .05, .01, etc., alpha can be set to anything in practice, depending on what past studies used, the nature of the phenomena at hand, the experimenters' judgement, and the severity of making a Type 1 Error, for example.
Non .05 or .01 alphas are
extremely common in practice. In many instances in the area of 'multiple comparisons', for example, alpha is calculated to be some truly bizarre non-typical number like .00833, etc. Again, it just depends on the situation.
I repeat: I can create the exact same scenario with alpha = .05 if you'd like. Just lemme know.

The
main idea with the test still stands though, not to mention all the ways unconscious bias can creep in elsewhere, like in the presentation of the data and in the interpretation of the analysis, etc., hence the need for triple blinding when feasible.
And you continue to fail to address the point that has been raised repeatedly about your paranormal claim ...
It's not paranormal despite your belief that it is or that I am claiming it is. Unconscious bias is normal.
...that a statistician can, without calculating the results, choose a test to give him a predetermined outcome.
Not predetermined, just an outcome that reflects his bias(es). For example, the evaluator might know the p-value would be lower (because one test is an approximation of another), but he wouldn't know that the p-value would be exactly .013736, for example.

If I was claiming the latter (I'm not) I'd agree that it was paranormal and prognosthewordIcantsay.
That is prognostication, sir.
No, it is being influenced by bias.
If you don't like these conclusions, I'm afraid your claim degenerates to one of your stat brat committing scientific fraud.
Nope, because we are talking about unconscious bias here, not conscious bias.