Re: standard deviation for digits

On 13 Jul 2005 01:36:47 -0700, glenbarnett@xxxxxxxxxxxxx wrote:

>
>
>Stig Holmquist wrote:
>> A textbook I have found states that the one way Anova test
>> must be based on normal populations. In my example the
>> samples were taken from a discrete uniform population.
>>
>> Suppose I wish to test the digit distribution in a lottery game,
>> such as a pick-3 or pick-4 type. After a fixed number of draws
>> the mean frequency per digit must be te same but the strd.dev.
>> within each position can vary greatly. How do I justify using
>> the one way anova test to determine it the ball sets are all
>> uniform and withou bias?
>
>You don't!
>
>You're trying to assess whether each number is coming up with the same
>probability?
>
>The most common approach would be a chi-squared test. You'll need a lot
>of data to find anything, and the machines are usually pretty carefully
>made - the differences aren't likely to be large enough to pick up, or
>to help you much.
>
>A more fruitful approach might be to try to use the number of people
>winning while playing certain numbers (some lotteries will give you the
>winning numbers and the number of winners each week for some time into
>the past), so you can choose less popular combinations - it doesn't
>improve your chances of winning, but it improves your expected winnings
>when you do.
>
>Just a suggestion.
>
>Glen
I'm currently set up to calculate the std.dev. for frequency of a set
of balls in a Pick-4 game. It occurred to me that since the mean is
equal to the expected I can easily recalculate the std.dev. into
a Chi-square value. At what level of probability would I be justified
to suspect bias? And how many draws should be tested? It seems to
me testing 500 draws is too large to detect short term bias. Would 100
draws be enough?

Stig Holmquist
.

Relevant Pages

  • Re: standard deviation for digits
    ... > must be based on normal populations. ... > Suppose I wish to test the digit distribution in a lottery game, ... > the one way anova test to determine it the ball sets are all ... winning while playing certain numbers (some lotteries will give you the ...
    (sci.stat.math)
  • Re: How to Determine Sample Size with ANOVA
    ... simple hypothesis test (like equality of means) with 2 populations. ... Recently, however, I had to undertake a one-way ANOVA test where I am ...
    (sci.stat.math)
  • Re: How to Determine Sample Size with ANOVA
    ... simple hypothesis test (like equality of means) with 2 populations. ... Recently, however, I had to undertake a one-way ANOVA test where I am ... The standard social science reference is Cohen's book, ...
    (sci.stat.edu)
  • How to Determine Sample Size with ANOVA
    ... simple hypothesis test (like equality of means) with 2 populations. ... Recently, however, I had to undertake a one-way ANOVA test where I am ...
    (sci.stat.edu)
  • How to Determine Sample Size with ANOVA
    ... simple hypothesis test (like equality of means) with 2 populations. ... Recently, however, I had to undertake a one-way ANOVA test where I am ...
    (sci.stat.math)