Current Events > PEDs can induce a "red shift" in male democrats

(+) Yeah! (+)

https://www.openicpsr.org/openicpsr/project/155441/version/V1/view

But it lets you use Hyper Mode anytime and utilize the Phazon Beam tho

this has got to be the dumbest study ive ever seen

Why does the search bar automatically suggest search terms without even typing anything and why are they all political

The fuck kind of website is this

NeonOctopus posted...

But it lets you use Hyper Mode anytime and utilize the Phazon Beam tho

Glad I wasn't the only one

And soy turns repubs to democrats?

On T. More left than ever.

Data deposited on a website with no corresponding publication, authored by a known hack. Not sure what I expected, but.

[LFAQs-redacted-quote]


It looks like a study/experiment/project from Claremont University that suggests giving Democrat-identifying males testosterone treatments increased their inclination toward Republicans by 45%.

I've no idea how credible it is though.

It's incredible that anyone read that abstract and thought "yep, seems legit"

here's my study: red hat folk are dumb, QED

lmao this kinda reminds me of that one time Joe Rogan gave Bill Burr some of his elk meat and then asked him "Did it make you more aggressive after you ate it?" and Burr clowned on him for asking that

It's from 2011 so cannot remotely reflect opinions on the current political landscape

Raikuro posted...

It's from 2011 so cannot remotely reflect opinions on the current political landscape

can you imagine it though

some dudes pumped them selves full of steroids and then went "Mitt Romney, that's my guy!"

Raikuro posted...

It's from 2011 so cannot remotely reflect opinions on the current political landscape

It's from 2021.

COVxy posted...

It's from 2021.

Literally says it was done in 2011

Ah whoops, yeah.

99.99% confidence is... very surprising results. I don't think I've literally ever seen a statement made with 99.99% confidence, especially when there is no difference in 3 of the 4 groups looked at.

For context, p value for significance is usually 95% (or p<0.5), so finding a p value less than 0.001 is a really big deal.

This might be interesting if it was an actual science journal, but this is a self-publishing website that afaik has no peer review process.

Proto_Spark posted...

For context, p value for significance is usually 95% (or p<0.5), so finding a p value less than 0.001 is a really big deal.

None of this is accurate.

But I have incredibly high T and am not conservative at all

Mezcla posted...

can you imagine it though

some dudes pumped them selves full of steroids and then went "Mitt Romney, that's my guy!"

And then imagine you have to now simp for some mush brained dumpasack ego turd like Donnie. lol

COVxy posted...

None of this is accurate.

I think my comment is mostly just worded poorly.

95% is p<0.05 not 0.5, I missed a zero there. However, 95% significance is still the statistical significance many studies do try to reach, though strictly speaking there isn't anything special about it. You can show significance at 90% or 99% or 99.99% or whatever, but when places show off statistical significance, its usually at least 95%. I don't think I've seen something below 95% confidence since Stats class, though it could be being in different fields.

a p value less than 0.001 would very strongly refute the null hypothesis and very decisively say that weakly-aligned male democrats have a clear relationship regarding their political views and higher levels of increased testosterone - which is interesting considering there were no changes in any of the other groups, and I wouldn't have considered "political affiliation" that rigid of a group that only one specific group would have these factors statistically related.

"is a really big deal" is exaggerating. a P-value that low is basically as definitively as you can state that these two factors are related. I think that's a big deal.

So, a p-value is less like confidence and more like an incompatability index. How unlikely is it that your sample was drawn from a population distribution that represents the null hypothesis. What that null hypothesis is, and how it fits what was actually done in the study, gives the p-value meaning. You cannot just judge a p-value alone.

And if you are reading studies that all seem to hover at values just under 0.05, that's a bad thing. That suggests the literature you're reading is p-hacked. A p value less than 0.001 is nothing special.

[LFAQs-redacted-quote]


Why not?

In the context of null hypothesis signficance testing, sample size justification takes the form of the question: "given the size of the effect I hope to find (expected size, or minimum size that is interesting), how many samples do I need to collect to find that effect X% of the times I run this experiment?" Power, in statistical jargon.

By saying "sample size is too small", you are saying they cannot detect the effect, because you expect the effect size, or their effective sample size, to be smaller.

There are no simple rules like "100 or more".

It also depends heavily on experimental design. I can design an appropriate study with 2 subjects.

What about female democrats?

COVxy posted...

How unlikely is it that your sample was drawn from a population distribution that represents the null hypothesis. What that null hypothesis is, and how it fits what was actually done in the study, gives the p-value meaning. You cannot just judge a p-value alone.

But the null hypothesis would effectively be that the two factors aren't related? There are issues with say the methodology or some strategic wording to suggest a different null hypothesis, but a p-value that low would strongly reject that null hypothesis and indicate whatever factors being looked at are related? (in this case, being that whatever "warmth towards republican" means and testosterone increases when weakly leaning Democrat). If the p-value in this paper isn't capable of saying something from the abstract, then wouldn't that be an issue with the paper, and not the p-value?

COVxy posted...

And if you are reading studies that all seem to hover at values just under 0.05, that's a bad thing. That suggests the literature you're reading is p-hacked. A p value less than 0.001 is nothing special.

Its not that p-values tend to hover at just below 0.05, but that P<0.05 is generally the lowest acceptable relationship, and therefore its what's commonly used. 0.05 is what will be used for any p value between 0.05 and 0.01. p-values themselves can be affected by things like how strong the relationship is or additional variables that you may not be able to necessarily account for, or sometimes something like a confidence interval is a better way to analyze the dataset, in which case you might set the p-value at 0.05 to see if there is significance there.

ITT a bunch of soy boy beta cucks

Proto_Spark posted...

But the null hypothesis would effectively be that the two factors aren't related?

Null hypotheses need to be carefully constructed. A null distribution from one study could look very different than the null of another. But this is all impossible to judge without a methods section.

All I'm saying is that a p-value in isolation or vague description is impossible to evaluate properly.