tidy_add_header_rows() error with continuous * categorical interaction about broom.helpers HOT 3 CLOSED

OK. I have updated tidy_add_header_row(). We additionally check if a term is different from its variable to identify such cases

from broom.helpers.

tidy_header_rows() add an header row only when a variable have at least two rows.
Here, the interaction variable has only one row. So no header row is added.
This is due to the fact that no reference row is added for interaction terms.

I understand your point. However, for this interaction terms, what would be the rule to decide when adding a header row? Because, if it would have been an interaction between two continuous variables, you would not add a header row.

from broom.helpers.

I think a good rule of thumb is that we need a header row if the label doesn't make sense without it. Or if the variable label and the label column do not match (because of terms in the label column), we need a header.

In the case above the label column is "1 * Marker Level (ng/mL)" for the interaction row. Without the header above it to indicate the variable label (which is "factor(response) * Marker Level (ng/mL)"), the label column is not clear.

If we had an interaction for two continuous variables, the var label and the label column would both be "variable1 * variable2", which would be clear.

from broom.helpers.