Sublime
An inspiration engine for ideas
Publication bias. Perhaps unsurprisingly, the scientific literature is full of positive results. Null or negative results have traditionally been hard to publish and are often relegated to abandoned hard drives, despite the fact that they represent valuable knowledge claims. Research users looking for evidence get a distorted view, which — like... See more
An “Omics ” Answer to the Replication Crisis
This DEI professional’s breakdown of where his industry went wrong is the most levelheaded analysis I’ve heard yet: https://t.co/OI8yLY0AZj
Rikki Schlottx.comSometimes, we discover that the interpretation of a study conflicts with another interpretation, making us realise that the study is so vague that it can be used as proof for two contradicting interpretations.
Sönke Ahrens • How to Take Smart Notes: One Simple Technique to Boost Writing, Learning and Thinking
The researcher may also look for disconfirming evidence. The process involves having the researcher first establish a theme and then look through all of the evidence for exceptions to this theme. This is evidence that presents an alternative explanation for a theme.
John W. Creswell • 30 Essential Skills for the Qualitative Researcher
The contrarian’s enemy is not only random conformists. It’s also ridiculously smart people who have studied the topic in incredible depth and concluded that they’re wrong. And as we all know from certain creative offshoots of rock, paper, scissors, high-IQ mega autists beats public intellectual.
Bentham's Bulldog • Losing Faith In Contrarianism
Scary: 73 teams tested the same hypotheses with the same data. Some found negative results, some positive, some nada. No effect of expertise or confirmation bias. "Idiosyncratic researcher variability is a threat to the reliability of scientific findings." https://t.co/pCRR2wFVWE https://t.co/bDbV8SbxSM
Drawing Conclusions From Associations Was Dangerous Making definitive health recommendations based only on associative evidence was considered dangerous because of "confounders" - other unmeasured factors that could influence the results.
What happens when RAG models are provided with documents that have conflicting information?
In our new paper, we study how LLMs answer subjective, contentious, and conflicting queries in real-world retrieval-augmented situations. https://t.co/3Re50rXEUD