This forum is majority female:
http://forum.mrmoneymustache.com/welcome-to-the-forum/what-gender-are-you/
So, from 355 votes (and I think males vote less on that) you say that the majority of of 14682 total members is male?
Get a statistics class please ;)
(On that a sidenote: NEVER trust a poll where you dont know the excact question and how they did it. Depeding on the wording of the question alone you can get 91% for and 92% against something.)
On the topic of assumptions: Most forums have male dominance. Males do more internet, too.
Running a quick Chi square test with an assumed 50/50 distribution, we get a two-tailed P=0.064, which is barely not enough to reject the null hypothesis that there are as many men as women on this forum (P>0.05). A bigger N actually may have pushed it over to prove that there are more women than men here.
It of course does not assume any selection bias, but I wouldn't be too concerned about the phrasing of a question of gender, since it's not an ambiguous question (as opposed to "are you for or against" type questions). I would not assume that most people on this forum are men.
Or we could take a Bayesian approach to the problem. Let's say there are two people: Blue believes that most internet forums are generally evenly distributed, and Red believes that forums are typically about 70% male, with <10% of forums having 50% or more women. These are just beliefs that two people have accumulated through experience, and without any other information, it makes sense for Red and Blue to make similar assumptions about this forum. Now we've got some data, through the survey, and Bayes' rule tells each person how to update their beliefs about this forum with the given data.
For both Red and Blue, the additional data now lead them both to believe that there are likely more women than men in this forum. The outcomes are slightly different, since they had different starting beliefs, but 355 people is a lot of people, so the data are definitely dominating the outcome. Maybe someone can help me add in assumptions about women being more likely to answer the poll.
This forum is majority female:
http://forum.mrmoneymustache.com/welcome-to-the-forum/what-gender-are-you/
So, from 355 votes (and I think males vote less on that) you say that the majority of of 14682 total members is male?
Get a statistics class please ;)
(On that a sidenote: NEVER trust a poll where you dont know the excact question and how they did it. Depeding on the wording of the question alone you can get 91% for and 92% against something.)
On the topic of assumptions: Most forums have male dominance. Males do more internet, too.
There's a famous joke about this.. If you don't believe a small random sample can work, then when you need to get bloodwork done, tell your doctor to take it all. Statistics were designed precisely to help us draw inferences from small samples, and actually, 2% sampling is a huge amount compared to a lot of work being done all the time.
Something sort of on topic, because I realize I've veered way off course: today we had a going away party for a coworker during lunch, and we celebrated by everyone bringing their packed lunch and eating together in the break room. The departing coworker brought cake.