Effect of Assassins

[sXs]

The problem is that I have done a hypothesis test, with H0 being that the probabilities are equal. That hypothesis had to be discarded, in favour of H1: Merchants have a significantly higher probability than other officers to be assassinated.

The problem with any type of coded random number generators is that the possibility of a null hypothesis being 0 is valid. In this aspect, Bribe and Assassin are identical because the values placed in the coding is identical. If the cards are the antithesis of each other, then the 100-sided die scenario is exactly the same for both.

Truly random generation means that any previous result does not affect the next result. the chance percentage does not change no matter the number of attempts. Don't confuse probability with statistical randomness.

[sXs]

The problem is that I have done a hypothesis test, with H0 being that the probabilities are equal. That hypothesis had to be discarded, in favour of H1: Merchants have a significantly higher probability than other officers to be assassinated.

The problem with any type of coded random number generators is that the possibility of a null hypothesis being 0 is valid. In this aspect, Bribe and Assassin are identical because the values placed in the coding is identical. If the cards are the antithesis of each other, then the 100-sided die scenario is exactly the same for both.

Truly random generation means that any previous result does not affect the next result. the chance percentage does not change no matter the number of attempts. Don't confuse probability with statistical randomness.

Just to explain a bit more on what I mean.

100-sided die scenario would mean

33 sides Mercahnt
33 sides Admiral
34 sides Captain

Cast 1 result merchant.

Now a probability approach would seem to mean that on the next cast 1 merchant side of the die is removed leaving only 99 sides. Although the result would still be random, the percentages have changed in this scenario, hence the probability has changed.

[sXs]

The problem is that I have done a hypothesis test, with H0 being that the probabilities are equal. That hypothesis had to be discarded, in favour of H1: Merchants have a significantly higher probability than other officers to be assassinated.

you can not use probability to determine randomness.

The chance of casting and H0=captain does not change no matter how many casts. The same result if H0=merchant or admiral.

Random means random.

The issue with a probability/ hypothesis test is it weighs in previous results. Previous results do not matter.

[Haron]

Of course previous results do not matter. I'm quite proficient in statistics and hypothesis testing, thank you :-)

[sXs]

Of course previous results do not matter. I'm quite proficient in statistics and hypothesis testing, thank you :-)

Data analysis is my profession.

[Haron]

Then you surely can see that when I do a test with 100 assassins, and 50 end up killing merchants, then I can discard my H0.

[sXs]

Then you surely can see that when I do a test with 100 assassins, and 50 end up killing merchants, then I can discard my H0.

Like I said, you can not use hypothesis testing in a truly random scenario.

[Haron]

Yes, you can. Or you have a very different understanding of hypothesis testing than I do.

[sXs]

Yes, you can. Or you have a very different understanding of hypothesis testing than I do.

No Randomness is a measure of uncertainty,

Probability is a measure of likelihood.

Hypothesis testing is a measure of probability(likelihood)

You are measuring probability not randomness.

[Haron]

Indeed I am. And my issue is that the probability of killing a merchant and killing a captain is not identical, as they should be, in my opinion.