4 ms·

Hiding research results should be seen as this scandalous, shameful, and fundamentally anti-scientific practice. A heavy bias towards business profits, which is

Hiding research results should be seen as this scandalous, shameful, and fundamentally anti-scientific practice. A heavy bias towards business profits, which is exactly what this is, goes against objective science. What if you kept repeating experiments, and only published when you got the results you wanted? That goes against my personal ethics and my understanding of how science is supposed to work.

jcranmer 7y ago
> A heavy bias towards business profits, which is exactly what this is, goes against objective science. Except the data shows that it's not the industry who's failing to report results, it's academia and the government who's failing.
- calibas 7y ago
  When people like Michael Taylor run the FDA, that's clearly agency capture. The industry has infiltrated government positions, I don't know how anybody can act like they're innocent in all this.
Spivak 7y ago
> What if you kept repeating experiments, and only published when you got the results you wanted? This is pretty much how it's currently done. Science is far less exploratory than you make it seem. When it comes to publishing you should think of it more like law. You make the strongest possible case for your conclusion.
- ouid 7y ago
  This is very backwards. In science, the evidence for a claim is precisely how hard you tried to disprove it. This doesn't mean that you're wrong, of course, but it's a massive problem.
  - Retric 7y ago
    That’s only a subset of experiments, the ideal experiment makes any result publishable. For example the search for a new particle, don’t find it that’s fine. Alternatively, any result that’s in doubt like salt’s impact on cardiovascular heath. Even better is really cheap experiments like analysis of existing data. That you can keep trying until you find something, sadly this means publishing datasets is a bad idea.
loceng 7y ago
Especially since then you can't compare to prior trial data to newer trial data to see how they were crafting the selection process, which could derive hints for contraindications and the like - or simply to know if some medication is poorly distributed, in the sense of making sure a person fits the exact criteria, then its side-effects and damages will be far greater than what was approved at the end.
andrewla 7y ago
Hiding research results should always be treated as though the results were very negative. There are many reasons that a study might get cancelled that are unrelated to business interests (like breaking of blinds or various other methodological reasons why a study's data is corrupted), and it's okay(ish) to not publish these results -- there are no good protocols for publishing data about a study's failure, but it would be nice to develop those. But without knowing the reason for the failure to publish the study, it should be assumed that the results provide negative evidence for all hypotheses. Pre-registration is supposed to solve this problem by listing the studies before they are conducted so that it is impossible to completely elide them. A drug approval would then look at all studies pre-registered for the treatment, and only at those studies pre-registered for the treatment -- whether those studies were complete or not.

jcranmer 7y ago

> A heavy bias towards business profits, which is exactly what this is, goes against objective science. Except the data shows that it's not the industry who's failing to report results, it's academia and the government who's failing.

calibas 7y ago
When people like Michael Taylor run the FDA, that's clearly agency capture. The industry has infiltrated government positions, I don't know how anybody can act like they're innocent in all this.

Spivak 7y ago

> What if you kept repeating experiments, and only published when you got the results you wanted? This is pretty much how it's currently done. Science is far less exploratory than you make it seem. When it comes to publishing you should think of it more like law. You make the strongest possible case for your conclusion.

ouid 7y ago
This is very backwards. In science, the evidence for a claim is precisely how hard you tried to disprove it. This doesn't mean that you're wrong, of course, but it's a massive problem.
- Retric 7y ago
  That’s only a subset of experiments, the ideal experiment makes any result publishable. For example the search for a new particle, don’t find it that’s fine. Alternatively, any result that’s in doubt like salt’s impact on cardiovascular heath. Even better is really cheap experiments like analysis of existing data. That you can keep trying until you find something, sadly this means publishing datasets is a bad idea.

loceng 7y ago

Especially since then you can't compare to prior trial data to newer trial data to see how they were crafting the selection process, which could derive hints for contraindications and the like - or simply to know if some medication is poorly distributed, in the sense of making sure a person fits the exact criteria, then its side-effects and damages will be far greater than what was approved at the end.

andrewla 7y ago

Hiding research results should always be treated as though the results were very negative. There are many reasons that a study might get cancelled that are unrelated to business interests (like breaking of blinds or various other methodological reasons why a study's data is corrupted), and it's okay(ish) to not publish these results -- there are no good protocols for publishing data about a study's failure, but it would be nice to develop those. But without knowing the reason for the failure to publish the study, it should be assumed that the results provide negative evidence for all hypotheses. Pre-registration is supposed to solve this problem by listing the studies before they are conducted so that it is impossible to completely elide them. A drug approval would then look at all studies pre-registered for the treatment, and only at those studies pre-registered for the treatment -- whether those studies were complete or not.