本来就和解释变量没多大区别，放入研究也很正常

Control variables are those put into regression to control the Omitted Variable Bias (OVB).
Consider Regression Y=Xb+e
In Randomized Control Trials (RCT)  X and e are uncorrelated, therefore OLS estimates are both unbiased and inconsistent.
In social sciences studies. X are normally choices of human being, which is highly possible to be correlated with e, therefore  induce the endogeneity problem.
e.g In wage regression, wage= education*b+e.
here education is endogeneous because it is correlated with 'ability'  which is not observed. In the presence of endogeneity, b can not be consistently estimated.
Therefore, if we could put ability (or its proxy) into the regression, the OVB could then be controled and we call the variable ability-control variable.
However ability cannot be observed, that's why we use other approach- IV regression, Panel data, Difference in Differences etc to control this bias.