I blogged with #rstats simulated data on the evils of stepwise variable selection regression. False positives go up, inference after the fitting is borked, estimates are biased away from zero. Yet still we see it all the time. https://freerangestats.info/blog/2024/09/14/stepwise
