Oh my, did I say that!
The times are random.
The experiment I saw fairly recently (on UK TV) seemed to imply that the dogs often just visited the door randomly anyway so it couldn't be tied to the expectation, or knowledge of the owner returning. You do realise the impact that this might have if true? And since nothing has really changed over the time since Sheldrake proposed this, what are the chances of it being correct? Minimal.