Since your DV is Psych. disorder this would be a multinomial logistic regression NOT ordinary least squares. You would add gender as an additional IV in order to test whether or not gender affects your DV. If g is the regn coeff for gender then you want to test H0: g=0

For techniques in multinomial logistic regression see the following link. I would recommend using R but if you have a different package in mind simply replace R in the following query: https://www.google.com/search?ei=t1pXXsXeO4jmsAWP4r6wBg&q=multinomial+logistic+regression+in+r&oq=multinomial&gs_l=psy-ab.1.2.0i273l2j0i67l4j0l4.1287745.1311849..1319880...0.2..0.540.6465.25j24j2j0j1j1......0....1..gws-wiz.....6..0i71j33i22i29i30j0i22i30j0i333j0i131i273j0i131j0i362i308i154i357.lVz91Y6hP2A

Since you have only 243 individuals available using all 243 will give you the highest power possible and the most complete results rather than sampling from the 243, so a starting G power analysis is not needed. When you are finished with the analysis you may compute your power from the info at this link: https://www.google.com/search?ei=4V9XXuCmE5LctQXkkrigBQ&q=multinomial+logistic+regression+power+of+final+test+results&oq=multinomial+logistic+regression+power+of+final+test+results&gs_l=psy-ab.3...393254.413862..414727...0.2..0.159.2913.23j8......0....1..gws-wiz.......0i71j0i67j0j0i22i30j0i333j33i22i29i30j33i160j33i299.kD90WBxMukU&ved=0ahUKEwigm4_Wi_HnAhUSbq0KHWQJDlQQ4dUDCAs&uact=5

I have attached an excellent set of notes which should get you started in logistic regression in general and then go to the above link on multinomial logistic regression for your dissertation approach. if you have further questions, please feel free to ask. Best wishes, David Booth

BTW a power calculation of the type your advisor requested would not be a simple thing. For example see the link: https://www.bing.com/search?q=sample+size+calculation+for+logistic+regression&pc=U528&form=QBLH&sp=-1&ghc=1&pq=sample+size+calculation+for+logistic+regression&sc=1-47&qs=n&sk=&cvid=3C0BD85B7D2E4BF9881F0267114A9471. I think it is best to just use all 243 individuals.

Thank you for you response! I really appreciate it.

I'm wondering if I do use all 243 how will that separate equally to have a true comparison of male vs female? I have 20 females that meet my specifications of narrowing down my cases. If I were to use all of the possible cases than there is definitely going to be more pull from the male side that would give me false percentages. Does that make sense?

Statistics and I do not get along so I am on the struggle bus. Thank you for all help! @DavidEugeneBooth

@ Alice, The fact is there are many more male serial killers than female serial killers. Your research question seems to be what are the differences. If you selected say 10 males and 10 females at random and did the same regression, then you would have a severe case of sampling bias.. If someone demands that you do something like that I would use a stratified sampling scheme with allocation proportional to group size(this is a real representative sample). For details, See: https://www.bing.com/search?q=W+Cochran+Sampling+Techniques+3rd+ed&pc=U528&form=QBLH&sp=-1&pq=w+cochran+sampling+techniques+3rd+ed&sc=1-36&qs=n&sk=&cvid=EEBF2C2356764781BFB25FE53245EBD4