3 ms·

a meta analysis makes this more interesting, not less though, no? with the high rate of poor quality peer reviewed research out there I find it hard to trust a

• andagainagain 5y ago
  Meta analysis is a great way to look at data that isn't actually showing what you want, and make it show what you want. What we would want is access to both the meta study and all the studies it referenced, just to see if there actually was anything here at all. Assuming all of those studies were well done and from a reliable source, and assuming that they actually tried to disprove themselves and that there weren't any other studies that successfully disproved them... then we would look at the data. Normally we can sorta fudge that because the claim doesn't sound illogical or controversial. But... 45% reduced cancer risk from eating a whole class of foods that just happen to fit well with ancient chinese medicine? Found on a chinese medicine website? The sniff test should fail from 3 miles away.
• failwhaleshark 5y ago
  p-values always seem to get swept under the rug too.