Why WE SHOULD CONTROL for theoretically relevant variables

Given that our preliminary results showed that hostile sexism
was positively correlated with benevolent sexism and that both
gender and hostile sexism influenced our outcome variables (i.e.,
support for the EE policy), we controlled for hostile sexism and
gender in our analyses below. We note that controlling for either
hostile sexism or gender did not affect the significance of our
results, but we nevertheless control for them in our analyses here
(and in all subsequent studies) in line with best practices suggesting
that inclusion of control variables is warranted (and needed) if
control variables are theoretically relevant to the phenomenon
studied and are statistically related to dependent variables (e.g.,
Becker, 2005; Becker et al., 2016; Spector & Brannick, 2011).

Hiedek & Ferris (2016)