Why do we need to consider how 'sex differences' might be constructed when we undertake research?

A lot of very valuable scientific research has been done without taking sex differences into account. For example, the vast majority of research on child birthing has been done solely on women. Similarly, most research on prostate cancer has been done only on men.

Gender differences are important only when investigating cross-cutting issues such that people from each gender might provide relevant response data. Even then, data obtained from one or another gender might be of little interest or have a completely different type of relevance. For example, the number of severe concussions suffered by professional boxers might pertain to boxers of both genders. However, there are so few women boxers that their data might be both statistically unreliable and comparatively more influenced by self-selection biases.

In addition, severe concussions (however defined) are far more likely when a boxer is matched against another boxer who has unusually high upper body strength. For men, the danger would greatest in heavyweight matches, but there is no reasonable equivalent for women boxers. In research such as this, trying to control for gender might be of little practical value.