Setting up experiments for socionics

[Jimbean]

I can think of two ways to set up experiments to support/invalidate socionics

1.) Look for language patterns across a sample of people and predict their results on a socionics test.

2.) Ask a sample of people about their psychological comforts/discomforts and then predict their results on a socionics test.

To anyone's knowledge, has this been done before?

[misanthropex6]

If you want to type someone with precision (it's complex) you need to check this person's childhood. Every detail about it, memories, the exact context and what the individual did in each occasion, taking it into consideration each category of what each function would probably fit, then you will see every piece falling into place, nothing seems out of place because it all connects perfectly and smoothly, because it's fail-safe if analysed correctly. This happens because as children our cognitive functions are way more present in our personality, we don't hold much back like when we're growing up and learning how society and people works and all that, in childhood your cognitive functions are as raw and clear as they can be, even those functions you suck at that as an adult maybe you developed more because you learn how to deal with it (especially true for Fi and Fe polrs)
If you want to type it based on common sense, yeah, you would just have to observe what kind of "vibe" that person is emanating, how it conducts actions, and all that. It's effective too but obviously not as precise as a complete analysis. And it's very prone to error, but it's just a tool anyway, even having a purpose you use it in the way it works for you, like any tool you could have in life; I see mistyped folks out there who didn't understand the system properly, but if it works for them that way and they don't want to learn the right way, what can you do?

[squishycans]

This may be obvious, but

Psychology itself is pseudo-scientific because we are not automatons, but change, adapt, and learn to the best of our abilities. And trying to find an experiment that somehow validates pseudo-science, doesn't make much sense, unless you'd be cherry-picking the results, which isn't very convincing. So you can't invalidate/validate pseudo-science, but yes, you could support it. For example, The Big Five uses statistics to give itself some kind of pseudo-scientific validity, despite being based on abstract dimensions and pseudo-scientific. You could probably do the same for Socionics.

So then how would you do that? Since Socionics concepts and types are simply abstractions or abstract language that are not well-understood or easy-to-understand, you need to bridge the gap between abstraction and context to make them more concrete. But because context and the interpretation of context can change from person to person (and depending on information metabolism), there needs to be a lot of supporting material and cross-analysis. This is where contention often arises. But if you can successfully bridge that gap, then Socionics can be properly conveyed so that there is little or no confusion about its distinctions. And with contextual consistency and more theoretical concreteness, you could simply apply statistical methods with Socionics ideas on Cognitive-Behavioral tests and find commonalities that can be properly defined and explained.

tldr; In theory, make Socionics concrete and easy to understand. Then apply Socionics to Cognitive-Behavioral testing.