#dfstat

2024-10-24

For the next few months, Dr. Andrej-Nikolai Spiess (openalex.org/works?page=1&filt) will be a guest in my working group.

We are working on a paper where we show that 29 % of papers in top journals like Science, Nature & PNAS were skewed by a single influential data point! Time to rethink our reliance on p-values and explore alternative measures like #dfstat. #reproducibilitycrisis #linearregression #rstats

Moreover, we will work on #qPCR related software like PCRedux (joss.theoj.org/papers/10.21105)

#JOSS

Show several plots with the effect of influential data points on linear regression: How do different measures respond to outliers? (A) dfbeta(slope), (B) dffits, (C) covratio, (D) hat value, (E) Cook‘s D, and (F) p-value. Outliers in orange areas exceed cut-off values; green indicates significant p-values.

No permssion to train AI in this post.

Client Info

Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst