Thanks Daryl. The heuristic I keep coming back to is to test the slope across multiple conditions. If it persists through different environments it’s probably structural. If it only shows up in one context, it’s probably event-driven. That distinction is the main guard rail against being unwise about it. But you’re very right this is hard to do.
I agree. The question is how? Very thought provoking.
Thanks Daryl. The heuristic I keep coming back to is to test the slope across multiple conditions. If it persists through different environments it’s probably structural. If it only shows up in one context, it’s probably event-driven. That distinction is the main guard rail against being unwise about it. But you’re very right this is hard to do.