Why not? Placebos also look like the real thing, don't they?
Anyhow, in cases like this it's always nice to save all the data you can get. The correlation you'd like to demonstrate might not exist at all. However, if there are many parameters, you might be able to find something else.