Develop a hypothesis. For example, you could suggest that people are more quickly able to associate masculine words with men and feminine words with women. Examples of masculine terms are strong, leader or stoic. Feminine words are usually categorized as having to do with emotion or being soft, such as emotional, delicate or weak.
Develop your study by coming up with a way to objectively quantify people's associations. For example, in 1995 Anthony Greenwald and Mahzarin Banaji developed an implicit associations test (IAT) to measure how long it took people to pair words. Researchers found that, in many cases, combinations of words that matched stereotypes were the easiest to pair. To gain a better understanding of this study you can take an IAT for yourself at implicit.harvard.edu. Follow a study similar to this format or create your own.
Make a list of common gender stereotypes to compare. If you are having trouble coming up with ideas, ask your friends and family what stereotypes they are aware of. Common stereotypes revolve around strength (male) vs. weakness (female), leadership (male) vs. following (female), ruggedness (male) vs. beauty/nurturing (female) and toughness (male) vs. emotion (female).
Bring in participants for your study and run the experiment. These could be students at a local university, friends or coworkers. Measure how well the participant performs on the task you have created. The way you measure will depend on how you set up your study. If it is computer based, then the program can record the information for you. If your study is on pen and paper, you will have to record the information yourself.
Compare the results of people when their task involved a stereotype with tasks that went against stereotypes. Revise your hypothesis, if necessary. For example, the hypothesis given in Step 1 would be confirmed if you found that people had a significantly easier time pairing words that fit common gender stereotypes.