Roulette: how do I analyze sample data from a discrete uniform distribution?
Hi, I have a question about analysing sample results from a process that theoretically follows the discrete uniform distribution.
I have been taking observations off roulette wheels at a casino to try to detect any biased wheels.
I have a good sample size (~1,000 observations) and a simple eyeball of the data shows strong variability around the theoretical percentage for each number (1/37 European wheel).
My problem is doing the correct statistical analysis on the data to give myself confidence that there is bias at play.
I have so far done Chi goodness of fit tests to see what extent the null hypothesis of a uniform distribution fits the data (some wheels cannot reject the null hypothesis, some wheels I can at the ~80-90 probability level).
But what I would like to do is test whether the number of say 32's (3.8% of spins) observed on a particular wheel is significantly different from the theoretical percentage (2.7% of spins). Is this a simple z-score test?
Many thanks.
Hi, I have a question about analysing sample results from a process that theoretically follows the discrete uniform distribution.
I have been taking observations off roulette wheels at a casino to try to detect any biased wheels.
I have a good sample size (~1,000 observations) and a simple eyeball of the data shows strong variability around the theoretical percentage for each number (1/37 European wheel).
My problem is doing the correct statistical analysis on the data to give myself confidence that there is bias at play.
I have so far done Chi goodness of fit tests to see what extent the null hypothesis of a uniform distribution fits the data (some wheels cannot reject the null hypothesis, some wheels I can at the ~80-90 probability level).
But what I would like to do is test whether the number of say 32's (3.8% of spins) observed on a particular wheel is significantly different from the theoretical percentage (2.7% of spins). Is this a simple z-score test?
Many thanks.