Non-manipulable variables
Researchers should refer to relationships rather than causality in cross-sectional research; they can also argue on theoretical grounds that a variable is more likely to be an independent variable than a dependent variable. Some variables are not only 'given' -not manipulable in the way required for a real experiment- but are also almost unavoidably independent variables. So we can infer causal direction when these variables are involved in research that shows relationships between them and other variables.