Building models for explanation · real and simulated data
Choose your controls
Add and remove control variables. Every model you fit is kept below, so you can see how much the answer moves. Then fit every combination at once and ask the question a skeptical reader would: does the conclusion depend on which controls you happened to pick?
Data
Step 1
Pick your controls
Slope–
95% CI–
Residual SE–Lower = Y predicted better. Not the same as a better answer.
Step 2
Every model you've tried
Each row is one model, newest on top. Dots are estimates and bars are 95% CIs.
Step 3
Every possible model, at once
Hover or tap a model to see its controls. Tap to load it into Step 1.
In R