Scientific Replication
Jun. 3rd, 2021 08:44 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
ysabetwordsmith... is facing a crisis. Nonreplicable hype studies are cited far more often than replicable studies. See, this is why I say that adding money to science gets you economics, not science. Much the same is true if you add politics. People are producing and publishing bunk because it is exciting and rewarding. They do not publish as much truth, because it makes less money and gets less attention. When you set up science to require money and attention, you get more bunk than real science. That's a problem.
There is a high degree of inaccuracy in studies to begin with. Among the most ubiquitous flaws are bad sampling and poor research design. This isn't a matter of one bad lab or university or journal. It's the system. This becomes especially obvious in secondary research, where a collating study may throw out many previous studies in its topic range due to visible flaws, poor recordkeeping, and related issues.
You can deal with this yourself by triangulating data. Anything that has not been replicated should be considered a point of interest, not a fact. You need at least 3 solid studies to consider something even somewhat reliable. And remember that a lot of advice gets reversed 5-10 years down the line anyway.
There is a high degree of inaccuracy in studies to begin with. Among the most ubiquitous flaws are bad sampling and poor research design. This isn't a matter of one bad lab or university or journal. It's the system. This becomes especially obvious in secondary research, where a collating study may throw out many previous studies in its topic range due to visible flaws, poor recordkeeping, and related issues.
You can deal with this yourself by triangulating data. Anything that has not been replicated should be considered a point of interest, not a fact. You need at least 3 solid studies to consider something even somewhat reliable. And remember that a lot of advice gets reversed 5-10 years down the line anyway.
