This is due to the lurking variable. This is a kind of unnecessary variable. They are neither of the variables under study, but some additional random variables that effects one or both. This is because they go unobserved by lurking underneath the surface of the variables of interest, making the connection between them appear stronger or weaker than it really is.