The difference seems meaningful if you’re talking about off-label prescriptions, where a particular treatment has already been shown to be safe. It also seems useful to know whether a particular treatment should be considered for larger, more powerful clinical trials or if there’s enough evidence to rule those out.
I mean, I completely understand and agree with this:
I don’t actually think that ivermectin works, and I am making fun of people for taking horse medicine. But I also don’t agree with the (what I perceive to be) over-interpretation of null results in this thread. Edit: If you think that’s masurbatory, ok, but I disagree that it is.
I even cited who told people to shut up with the pro-ivermectin posting specifically to help you out there. I guess it wasn’t enough help. If you want to semantike about the difference between “I wish people would stop” and “shut up” then I guess that shouldn’t be surprising.
Man. Back to the point a bit. Bobman wasn’t even saying anything about ivermectin in particular. He used it as an example to make a general point. If people actually read his posts and weren’t assholes assuming he was pushing ivermectin when he was absolutely not doing that, none of this would have happened.
Bobo has been engaging in really wild semanticking. Like
As an example, this study found that the patients who were treated with ivermectin got better two days earlier,
Seems like an absurd way to describe a paper that concludes
Among adults with mild COVID-19, a 5-day course of ivermectin, compared with placebo, did not significantly improve the time to resolution of symptoms. The findings do not support the use of ivermectin for treatment of mild COVID-19, although larger trials may be needed to understand the effects of ivermectin on other clinically relevant outcomes
Apparently there’s an extremely interesting nuance about the nature of hypothesis testing that I’m missing, but it seems like a silly exercise.
I’m sure I’m going to regret this, but let me see if I’m right about why people are talking past each other.
I walked away with two impressions from ChrisV’s initial post:
The “70% of hospitalizations are ivermectin poisonings” story sucks, because if the media had done the slightest due diligence, they’d know it was in error. Yes. It is neither controversial nor interesting to say that the media shouldn’t print things that aren’t true. I don’t think anyone is arguing with that.
Paraphrasing: “If ivermectin turns out to have some benefit, we’re all going to look like idiots.” This is the part that CN and others, including myself (physicist here, don’t know if I’ve ever mentioned that in my once-a-month posting) are rebelling against. Because if all the studies show no statistical benefit, then telling people that ivermectin doesn’t help is the right thing to do. If it turns out later, with more studies and greater statistical power, that ivermectin actually does something useful, then we revise our thinking. But that does NOT mean that we were wrong to say at this time that ivermectin wasn’t beneficial. It was the right call at the time with the information we had. Saying, “well, there’s a chance we still might be wrong, so let’s give the treatment a shot anyway” is bad medicine.
Now if ChrisV meant to say that “the MEDIA” will look like idiots, that’s a different story, and you’ll get no argument from me. The media is always over-amplifying the significance of any one study, and use bad language like “scientists have shown” or worse, “scientists have proven” all the time. If the media was clearer about the difference between peer reviewed work versus preprints, or “one more data point” versus “proof,” I wouldn’t have to spend so much time on questions like “what’s the difference between a theory and a law” at public talks.
But again, I don’t think anyone’s arguing that point. The argument is about the idea that we shouldn’t tell people to avoid ivermectin right now because we might be wrong in the future. Because we should. It’s the right call with the information we have.
If I’ve misstated someone’s position, then maybe take one more shot at explaining it? Because if I’m not understanding you, it’s not because of forum drama or “taking sides,” it’s because I really have no idea what you’re saying.
And let me take this opportunity to thank people like Danspartan, CN, Will1530, Riverman, suzzer99, and and several others I’m probably forgetting. To me this thread is the single most valuable thing about UP, and has been a lifeline for the last 18 months, both as a source for technical information and a place for shared life experiences.
I’ve been thinking, and it’s wild that “Student” wrote an extremely important scientific paper under an assumed name in the 20th century. Has anyone else ever done that?
Bob all of your talk about type II error is irrelevant. You can’t say ‘oh it might be type 2 error’! That’s true about literally every single paper that fails to reject the null hypothesis. You cannot use this fact to say “the actual observation was that the patients who got ivermectin recovered sooner” or that “this study found that the patients who were treated with ivermectin got better two days earlier” when the data doesn’t actually show any difference.
A couple of other thoughts:
Power is discussed in the paper. Here is their discussion.
This would provide an 80% power under a 2-sided type I error of 5% if the hazard ratio (HR) comparing ivermectin vs placebo is 1.4, corresponding to a 3-day faster resolution of symptoms in patients receiving ivermectin, assuming that time to resolution of symptoms is 12 days with placebo.24 With an HR of 1.4, 75% and 85% of patients in the placebo and ivermectin groups, respectively, would experience the outcome of interest at 21 days.
Most importantly, and this is a huge red flag, the paper changed their primary outcome measure when it failed so hard that they decided not going after it anymore. They likely changed their primary outcome to the one thing they were measuring that showed a possibility of working. That is not valid, and is in fact a massive bias towards finding a result. The fact that they still didn’t is notable.
How do you feel about ChrisV being accused of being someone who might be taking ivermectin? You think that’s fair? Do you know ChrisV? He’s been posting with some of the people here for 16 years.
My recollection from grad school is similar - he was developed the statistical work as part of his work responsibilities, so it wasn’t really his product to publish in his name. This story seems to back that up:
In 1908 Gosset published the fruit of his labours in a paper entitled The probable error of a mean in the journal Biometrica (Student, 1908), of which Pearson was Editor. However, Guinness had a policy of not publishing company data, and allowed Gosset to publish his observations on the strict understanding that he did so anonymously, and did not use any of the company’s data. Gosset complied and published under the pseudonym ‘Student’ – the name under which he would publish 19 of his 21 publications. The name Student apparently came from the cover of a notebook Gosset used at the time – The Student’s Science Notebook (Ziliak, 2008).
I don’t think ChrisV is taking ivermectin, but I think he did go a little far (maybe just for rhetorical effect) by saying that taking ivermectin is a freeroll.
I think others should be afforded the same rhetorical license to refer to it as horse paste.
Oh so a post where he quoted ChrisV saying “if you can afford [ivermectin] and take it as directed then you could argue that it’s a freeroll” is what you mean. Seems like directly quoting someone is a very fair thing to say. Seems like Wookie’s statement is accurate, and you’re really mad about it for some reason.