News
Links I showed on Friday re "must published research findings are false"
Written on 18.11.2024 13:19 by Ingmar Weber
Pimeyes: scary facial recognition service, https://pimeyes.com/en
"Why Most Published Research Findings Are False", the paper that kicked of the "replication crisis" (https://en.wikipedia.org/wiki/Replication_crisis), https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.0020124
"Chocolate promotes weight loss", the problem with trying out lots of things and only reporting the one that works, https://gizmodo.com/i-fooled-millions-into-thinking-chocolate-helps-weight-1707251800, https://www.cbsnews.com/news/how-the-chocolate-diet-hoax-fooled-millions/
Bonferroni Correction, one of the ways to deal with this "multiple hypothesis" setting: https://en.wikipedia.org/wiki/Bonferroni_correction
Same data, different analysts, different conclusions: two studies show that the _same_ data and the _same_ research question can lead to different results: https://journals.sagepub.com/doi/full/10.1177/2515245917747646, https://www.sciencedirect.com/science/article/pii/S0749597821000200
Reproducibility crisis in machine learning, partly caused by "leakage" where some information from the training data leaks into the test data: https://reproducible.cs.princeton.edu/
One example of a meta analysis of if [insert some food] is good or bad for you. Coffee in this case: https://www.bmj.com/content/359/bmj.j5024