/sci/ - Science & Math

Sent this to my scientist sister and she told me it's a garbage article. I asked her why and she just told me it sucks. What does /sci/ think?
https://www.commonsense.news/p/us-public-health-agencies-arent-following

>>12416617
According to your OP, your null hypothesis is that they are not rigged. There are so many ways to rig the software that whatever parameter space your analysis is using will be very high dimensional. In particular, once the number of dimensions exceeds your sample size ( = 300k), freshman-tier hypothesis testing becomes completely uninformative.
Now this by itself doesn't necessarily invalidate your analysis, but without more specific details of the model used, it's reasonable to be skeptical of your allegation.

>>12163689
>For example it's implied the NTP study's methodology (see below) must be of low quality, and its results useless, because male rats had a longer lifespan, which therefore inherently contradicts DNA damage and higher cancer incidence in certain animals.
Well no, that was just an aside. The more important points were that the study mischaracterized the current state of research in their introduction, that other complaints were levied immediately even before publication, that it contradicts the results of earlier studies, and that RF photons do not have enough energy to do what was claimed. These are essentially my points, too.

If you want to get into more specific problems with the NTP study, the issue is mostly in the interpretation. When you analyze a large number of variables, the probability is very high that at least one will show an effect that on its own is statistically significant. Note that in this study, while one result was significant, the effect size was still very small. Note also that their control group was anomalous. This alone is more than sufficient to explain the result. The study did not find an effect in female rats. It did not find an effect in male+female rats. It did not find effect at low intensities. It did not find an effect on lifespan (or indeed an inverse effect for males). It did not find an effect for glioma. It ONLY found one effect, for one sex, at one intensity. That is hardly a convincing result.

Also, p-hacking

>>9543957
I'm not sure you know what a correlation fishing trip is. It means that you look at enough variables that it's inevitable that you'll find a "statistically significant" correlation.

Twin studies show the lack of a strong genetic or conditions-of-pregnancy basis for homosexuality (identical twins of homosexuals aren't usually also homosexual). They weren't "born that way".

>>9497576
>your example makes no sense. If you want to test whether two populations differ, there are significance tests
you need to take into account that you are running the significance tests multiple times, something that has been proven over and over not to be done in social sciences. I hate to use xkcd, but maybe going down to this level is needed for you to understand.

Is most scientific research bullshit?

http://www.wired.com/2014/08/what-makes-us-fat/

>>7051622
There's an entire literature of this stuff. Literally journals of it. If you can't be bothered to go through a citation index, find a review article. It's not that hard.

> I am ready to read any evidence presented
Then read the article posted 11 hours ago >>7050815 and the references in that. It has what you need.

> Since I have not seen 'the evidence' proving a link between autism and vaccines, I can conclude that their is no link based upon the current evidence.
That is extremely faulty logic (and bad grammar, but that's a lesser sin).

The conclusion is likely correct: a lot of studies have been done, mechanisms have been proposed and rejected, and reasonable alternative explanation for observed effects have been made; the bulk of evidence is that there is no link. But you're a gullible fool if you accept that conclusion simply because you haven't heard anything different, and not because you are confident in the vast bulk of evidence supporting it.

>>7006571
>I've seen studies where adding flouride made absolutely no difference in the incidence of cavities
> this was a comparison between two cities in canada I believe
So you saw one study. And that, somehow, invalidates the entire literature on the subject?

>>6790668

>>6718367
>Read it again anon it's an increase for boys period, blacks get it roughly twice as much.
Nope
>Table 3 shows results for the entire cohort excluding African American children. As can be observed, there is no statistically significant effect for any of the subclasses in either gender or age cut-off for MMR uptake.
general rule is, if you have to go subclass hunting, there's probably nothing there, even if you find something in the end.

>>5660165
Thank you very much kind anon!!!

Guys, the man behind the study is Gilles-Eric Seralini. That is like citing Kent Hovind on Evolution - only a full-blown retard with no knowledge about the issue would do that.

There is not a single person with scientific standards who wants to come close to that guy. Any website or paper that publishes him is pseudoscientific.

In this case, they started the "study" without any plan for the analysis. In other words, they conducted it in a way that *some* anti-Monsanto result would turn out simply by pure chance, even without faking the numbers directly.

Pic related. Every student learns this in his first statistics semester at university. As I said, it is pseudoscientific shit.

>>4758790

Hi^guys

How about a Statistic Humor Thread?

I hope all of you ITT know what "statistically significant" does and doesn't mean.

Oh, who the hell am I kidding. You could probably count the number of people ITT who understand it on one hand.

>>3587260
forgot pic

Guys, this is the single most studied issue in public health. The bulk of the evidence is that fluoridation is both effective and cheap. With thousands upon thousands of published papers, of course you're going to find some that disagree. It's just a matter of statistics. The occasional paper which suggests a correlation with cancer, low IQ, or whatever has always been refuted in the long run, and the massive quantity of evidence showing that fluoridation is generally safe requires extraordinary proof to overturn it.

>>3428266
> I don't know what the concentrations are, so I can't really say whether or not they pose a poisoning danger.
Chronic ingestion (a few decades maybe) at 5 ppm can cause skeletal fluorosis. You only see this in places like India, where they're just grateful to have any potable water, and don't have the resources to control levels. In Canada, water is fluoridated at 0.7 ppm, and it may be as high as 1.0 ppm.

By comparison, if you took in 5 times as much vitamin D as the recommended allowance for a year, you'd die a horrible death. If you took in 5 times as much fluoride, you'd have a few spots on your bones.