Irrespective of whether it's random or cherry picked, you can't discount a mountain of evidence from N=1. That is the playbook Big Pharma have succeeded with so far. Find one trial with evidence of fraud (El-Gazaar DBRCT) and then discount all other clinical trials on IVM for Covid. The media bought it hook, line, and sinker.
I absolutely can discount a mountain of evidence from N=1. If you bring me 10 candidates for a job and say they can all type 100 WPM and I say, "OK, give me that guy in the middle" and give him a typing test and he types 20 WPM, I'm discounting the rest of the candidates because I no longer trust the source.
Furthermore, the issue isn't the study. It's how they misrepresented the results of the study. Another example, it would be like if you give me a paper and I go to the middle of the paper and it says, "Nazi's saved millions of Jews from the rein of terror of the Hawaiians" -- I'm going to discount the paper because I think the editorializing is misleading. I don't need to read the rest of the paper -- unless you tell me that the section I read was meant to be satire.
This is why credibility matters. If that site was Nature and I looked at one study and there was something wrong with it I might be inclined to look at a few more randomly. But given that people have a finite amount of time, when you present something it better be accurate. And when it is accurate, you'll gain credibility. But if the first thing I look at is misleading -- I don't have a lot of patience to wade through data ... especially when other sources also say that you're representations are biased.
The reason why many of the studies are poorly written and have methodological limitations is that many are done by clinicians in 2nd world countries.
Not one large DBRCT from the west has reported on IVM for Covid yet. Let that sink in.
If you don't like ivmmeta.com, have a look at this published meta analysis of ivermectin for covid. It reaches the same conclusions. And yes, there is also a cochrane meta analysis that cuts out all by 9 DBRCTs (ha!), still is positive, but not stat-sig.
