Gender is the Independent Variable and "men" or "women" are the two levels of the independent variable. The experimenters in the earlier experiments we reviewed randomly divided participants into groups who either got a name introduction from the server (or not), got a tip tray with a credit card logo (or no logo) or got a strong level fear inducing ad (or low level). This Gender correlational studier would divide the participants into male or female groups, observe the average tips for males and females and see if there was the predicted statistically significant difference. If there was the hypothesis prediction would confirmed and if there wasn't the hypothesis prediction would be disconfirmed.
But notice the big difference; the correlational studier can't randomly assign the participants to be in the male or female group; they CAME INTO THE STUDY being male or female and the experimenter cannot change them. The problem with this type of Pseudo-Independent Variable (PIV) is even if you find a difference between the experimental and control group, men and women in this case, you don't know why you got the difference. There is an inherent ambiguity in the (pseudo) independent variables of correlational studies. Men and women differ in a variety of ways ( average income levels, social experiences, gender role expectations, hormone levels, and so on) so even if the correlational studier's data hypothesis was confirmed, it is ambiguous what to make of it. Which of those many differences between men and women could be the reason for any obtained difference?