Floppertje said:
HannesPascal said:
Alterego-X said:
But even if Unity would have just one female protagonist, and all the co-op characters would be differently colored copies of HER, that still wouldn't come close to equal representation when looking at the whole franchise. Even if the whole Assassin's Creed franchise would be specifically about female assassins, that wouldn't come close to equal representation when looking at the whole gaming industry.
Actually the probability of the protagonist being female in the Assassin's Creed franchise is not statistically different from 50%.
Proof:
Assume that the number of female protagonists are binomial distributed (either female or male) with a 50% probability. In total there has been 7 protagonist (including the French guy).
What is the probability that one or less of the protagonists are women:
P=(7 over 0)*0.5^0*(1-0.5)^7+(7 over 1)*0.5^1*(1-0.5)^6=0.0625
Normal practice in science is to reject the hypothesis (that 50% of the protagonists are women) if P<0.05. Since P>0.05 it is said the probability of the protagonist being female in the Assassin's Creed franchise is not statistically different from 50%.
More about binomial distributions [http://en.wikipedia.org/wiki/Binomial_distribution]
Without calculation I can say that it's likely you're right about the whole gaming industry though.
Yeah, nice try. What that says is that IF the genders are equally distributed, this outcome would happen in six percent of all cases, and you say that that percentage is not low enough to reject the hypotheses that genders are equally distributed. Which is like flipping a coin 7 times, getting six heads and one tails and saying 'well, I don't think that's weird enough to assume the coin is unbalanced.'
But, first, this isn't blind chance, there are many factors contributing to the decision of the gender of the protagonist. Second, you take only the AC games as data. If you take all games, all games this year or even all games presented at E3, there would definitely be statistical evidence that there are way less female protagonists than male, and there is no reason at all to pick assassin's creed exclusively. very few games have so many entries that you would be able to get any hard statistical evidence(n should be 30). So your proof is bogus.
It's worse than that, it's like flipping a coin six times, getting six heads and one tails, and then saying you got 4 heads and 3 tails.
Yes, this outcome is possible in a random distribution. But "Science" doesn't say P>0.05, depending on the situation, there's different P's your going to use. Particularly when there are multiple hypothesis we could test. If we test P(M)=.85, I guarantee we'll get a much higher value than .0625. P>0.05 happens to be common to a lot of statistical analysis, but P>0.05 doesn't mean "SCIENCE". Particularly when the P>0.05 figure is not used to say that something is science, or that it's true, it's saying we can reject the Null Hypothesis, not accept the Hypothesis. It's about measuring something above what can be reasonably assured to be nothing. Additionally, Rejecting the Null Hypothesis is provisional acceptance of the Hypothesis at best, not confirmation.
Plus, it doesn't change the distribution. If they have less female protagonists than male, they have less female protagonists than male. So, we've still got more male protagonists than female, which is what people are taking issue with, not that it's possible for this to occur more than 5% of the time. So this is a simple red herring (Apart from the absolute failure of statistics and probability, which I intend to bust up completely). Which really shouldn't be tolerated in any applications of formal logic (Which the following is).
And it's retroactively assuming that the distribution overall is 50%, which anyone who's looked at games should know isn't true for gaming overall, and there's no reason to assume it for Assassin's Creed. If we take a P(M)>0.5, P(ACD) increases, P(ACD) is actually more likely if the P(M) is more than P(F). P(ACD) jumps to 15.9% for P(M)=.6, P(ACD) jumps to 32.9% for P(M)=.7, 57.7% for P(M)=.8, and 85 for P(M)=.9. To anyone doing any actual science: That indicates that the probability of the pool being biased towards male protagonists is higher than the pool being equally distributed or towards female protagonists. P>>>0.05. Which again, just says that it's probably not nothing. Not that it's science. We'd be more justified to suppose that the sample is biased towards male protagonists. By a lot. A conclusion Hannes has themself rejected, without validity of their rejection. The hypothesis that male protagonists are more likely is as much as or more suppoted than the hypothesis that they're equivalent. Additionally, I'm not going to ignore the fact that when social pressure does change perception, Ubi will change their act. You can't just say that the distribution was binomial when they finally cut their shit and it approaches something balanced.
Where P(M) is Probability of a Male protagonist overall, P(F) is likewise for female, and P(ACD) is the probability of the Assassin's Creed Distribution (Or a 100% distribution, as Hannes included, which actually makes his figures worse, but credit for the honesty) occurring in a Binomial model.
Which would mean that the hypothesis that the distribution of assassin's creed protagonists is biased towards men is greater (A conclusion that anyone could come to without math) probability, and more likely than that the distribution favours men and women equally. Because some people are too lazy to do all the maths beyond what they mistakenly think confirms their worldview and justifies obvious discrimination (A position which is profoundly unscientific). I suppose we're going to move on to practicing Homeopathy or skull measurements for intelligence to justify racial bias next (And oddly enough, in something like 70% of random data sets (IIRC), Researcher DOF can create a positive correlation out of white noise (Of course, this doesn't apply here, but it's a poignant critique of the P>0.05 nonsense).
Additionally, I'm not even sure I'd label the distribution binomial, since I really don't think that game developers are thinking that way. It's still the best model to my mind, but I'd hazard more than a guess that having a female or any other "Minority" protagonist decreases the odds of another for a few trials. Eventually, I'm sure they'll change their tune as well, which further weakens the binomial model. Applying it at that point would be completely dishonest.
That post really needs to go back to the 101 section of probability and statistics, they're really not ready for the binomial distribution yet. Also, protip: That's not a proof. That's not what we mean by a proof in maths, that's not proof of the principal you want to prove anyway (Since a proof does not show a possibility. Otherwise I could dismiss all manner of math based on the fact that simpler techniques or numerical techniques of varying complexity often provide the answer. Screw Cubic Splines, a Newton Polynomial will work if the function I'm modelling is a polynomial QED lololol).