As empirical social scientists, we are always looking for relationships between different phenomena. However, we need to be very careful which variables we include in our empirical models. For example, Spector and Brannick (2011) point out that common control variables (e.g., age or size of an organization) are often rather included because of “methodological urban legends” than for theoretical reasons. Therefore, many significant findings denote unobserved relationships or just correlate by chance.
I just came across a website which collects spurious correlations – just in case anyone needs further arguments, why a relationship should be established conceptually first.
Source: Spector, P. E., & Brannick, M. T. (2011). Methodological urban legends: The misuse of statistical control variables. Organizational Research Methods.