Responding to skepticism toward your model

Several years ago I wrote about how researchers are skeptical towards numerical models and their results. I understand where this comes from. How do you convince people that you didn’t just tune the inputs of your model until it matched empirical data or yielded some other result you wanted?

I encountered this frequently during graduate school, from informal meetings to conference talks. I show a result, someone questions it, I suspect they think the result isn’t very good, until they finally reveal that they have the opposite concern: the model looks too good.

This is a fair question that should not offend you.

Do not become defensive. Their skepticism is valid.

The solution is simple: Share your code or simulation input files. My thesis included an appendix with a Python script for generating material model inputs as well as the relevant material definition blocks for all major finite element models I ran. I am considering posting the entire model definitions once all the papers are submitted.

So the next time you are giving a scientific talk and a listener is concerned that your results are too good to be true, simply point them to where you have made the code or simulation files publicly available, and invite them to verify your inputs and play with your model.

Otherwise, what are you going to say? “I really really promise these numbers are legit!” ?

You can read more about me, follow me on Twitter, subscribe to this blog by RSS or email, and find many more posts in the archives.