A sample is useless if it's not representative of what is being looked at. If I want to determine the prevalence of a characteristic in pregnant women, would I include all people? Or just pregnant women? Your logic is so incredibly flawed, but it doesn't seem like I can get you to understand, so I'm moving on.