Thank you, Daniel! Your prayers are much needed and appreciated. 💕🙏💕

Solving problems in this way (doing the work!) helps you recognize patterns in solutions, connect ideas, etc.

I can't imagine wanting to give that up!

Note that beta regression can only handle response variables that live in the (0,1) range.

Could drop some resources here for you once you confirm that your response variable lives in the [0,1] range.

The other option would be to use an ordbetareg model.

For both model options, you could go either via the Bayesian route or the Frequentist one in terms of R implementation.

Your response variable is bounded between 0 and 1?

If yes, one option would be to use a ZOIB model (zero-and-one-inflated beta regression model).

In which case, you would have to pay attention to how the sub-models making up the ZOIB are parametrized.

I’m with you, Ben! A participant in my last R course asked me for help with a coding problem he was having.

He shared the clunky ChatGPT solution which didn’t even work as advertised. 🤮

The solution provided by my brain: 2 lines of elegant code. 💪

If we want to “keep” all rows were col1’s elements are not equal to “a” AND col2’s elements are not equal to “b”, then we can do this:

filter(col1 != “a” , col2 != “b”)

The filter() function uses the comma symbol , as if it were the and operator &.

So filter(col1 == “a” & col2 == “b”)

can be specified as:

filter(col1 == “a” , col2 == “b”)

When I teach R, I encourage my course participants to replace “filter” with “keep” in their minds. That makes it a lot easier to understand we are keeping certain rows of a dataset.