Relationship between two numerical variables
Scatterplots, correlation coefficients and least squares lines honestly summarise the relationship between two numerical variables, Y and X.
However they can sometimes give a misleading impression about the relationship. The problem arises when other variables are also associated with both Y and X.
The marginal and conditional relationships between Y and X are often different and may even be in a different direction.
Lurking (or hidden) variables
If the marginal relationship between X and Y is different from their conditional relationship given Z, but Z has either not been recorded or is ignored when analysing the data, then Z is called a lurking variable (or a hidden variable).
Always think about whether there might be a lurking variable, Z, that is distorting the relationship that is observed between Y and X.