3 ms·

I'm not educated in how it works for this kind of (sociological? Psychological?) experiment but finding repeated instances - as per the trials reported in the a

by gota 7y ago

I'm not educated in how it works for this kind of (sociological? Psychological?) experiment but finding repeated instances - as per the trials reported in the article - seems like more than just a data point. They have many more negative data points than the original paper had positive ones In pure formal logic they have at least proved it is not true for all cases, so it's a refutation in that sense at least But in any case, "refutation" is closer in popular lingo to what happened than "replication"

scoofy 7y ago
This gets into deep philosophical questions. Logic can be deductive or inductive. Your use of formal logic implies deductive logic, but empirical science is inductive. With induction, one can only have negative knowledge (this is know as the problem of induction or the black swan problem), but it gets more complicated than that. Most studies are trying to look at probabilistic outcomes and are trying to get significant p-values. There is a myriad of problems with this beyond simply the existing probability that their sample was bias. E.g., their methodology could be improper, or even just less than ideal. There could even be unknown problems with the underlying assumptions and the model being used by researches. To suggest that an opposite result refutes a previously obtained result deeply mis-characterizes how science and human understanding works, and the problems inherent in this thought process deeply overvalues the knowledge we have and undervalues the more-than-likely seriously uncertain data we work with every day. If you're interested in these problems, i'd suggest Karl Popper's work on philosophy of science.
- gota 7y ago
  Thanks for your comment, I see it can help frame the conversation, but I meant to say I'm not versed in the terminology for social sciences. I'm about to finish my Doctorate heavily involving formal logic and logic programming, and have read a sufficient bit about the philosophy of science This is an interesting discussion but I'm unfortunately out of time now. I'd only point that most of science is abductive, rather than inductive, and that the whole concept of refutation will take us to discuss what is scientific _advance_
  - your-nanny 7y ago
    best just think in terms of Bayesian update.
    - gota 7y ago
      Indeed, abduction is a way to base your priors in Bayesian nets!
- StavrosK 7y ago
  In the same vein, a positive result doesn't prove anything, it's merely a data point in the opposite direction.
- heyoni 7y ago
  Any publications in particular you suggest?

by gota 7y ago

scoofy 7y ago

This gets into deep philosophical questions. Logic can be deductive or inductive. Your use of formal logic implies deductive logic, but empirical science is inductive. With induction, one can only have negative knowledge (this is know as the problem of induction or the black swan problem), but it gets more complicated than that. Most studies are trying to look at probabilistic outcomes and are trying to get significant p-values. There is a myriad of problems with this beyond simply the existing probability that their sample was bias. E.g., their methodology could be improper, or even just less than ideal. There could even be unknown problems with the underlying assumptions and the model being used by researches. To suggest that an opposite result refutes a previously obtained result deeply mis-characterizes how science and human understanding works, and the problems inherent in this thought process deeply overvalues the knowledge we have and undervalues the more-than-likely seriously uncertain data we work with every day. If you're interested in these problems, i'd suggest Karl Popper's work on philosophy of science.

gota 7y ago
Thanks for your comment, I see it can help frame the conversation, but I meant to say I'm not versed in the terminology for social sciences. I'm about to finish my Doctorate heavily involving formal logic and logic programming, and have read a sufficient bit about the philosophy of science This is an interesting discussion but I'm unfortunately out of time now. I'd only point that most of science is abductive, rather than inductive, and that the whole concept of refutation will take us to discuss what is scientific _advance_
- your-nanny 7y ago
  best just think in terms of Bayesian update.
  - gota 7y ago
    Indeed, abduction is a way to base your priors in Bayesian nets!
StavrosK 7y ago
In the same vein, a positive result doesn't prove anything, it's merely a data point in the opposite direction.
heyoni 7y ago
Any publications in particular you suggest?