Title: **Simply way to calcuatle p value ?**

Post by:**sven_Bent** on **2019-02-21 15:06:45**

Post by:

i undersand the improtna of p value and waht it represent. im just not good at getting the math right and everything i look up has it way overcomplicated for my little needs

So anyone that can help my with a simple way to calcuatle it for my A/B situations ?

In my test the issue is either True or false ( kinde like in ABX testing)

I want to calcaulte thd P value for the results being true

aka let say 12 trials

trial 1 True

trial 1 Trues

trial 1 Trues

trial 1 False

trial 1 Trues

trial 1 False

trial 1 TRues

trial 1 TRues

trial 1 TRues

trial 1 TRues

trial 1 TRues

trial 1 TRues

aka 12 trials. ( 10 Trues. 2 False)

What is the simplest way to calculate the p value here? cutting out anything related to test done with other paramets

So anyone that can help my with a simple way to calcuatle it for my A/B situations ?

In my test the issue is either True or false ( kinde like in ABX testing)

I want to calcaulte thd P value for the results being true

aka let say 12 trials

trial 1 True

trial 1 Trues

trial 1 Trues

trial 1 False

trial 1 Trues

trial 1 False

trial 1 TRues

trial 1 TRues

trial 1 TRues

trial 1 TRues

trial 1 TRues

trial 1 TRues

aka 12 trials. ( 10 Trues. 2 False)

What is the simplest way to calculate the p value here? cutting out anything related to test done with other paramets

Title: **Re: Simply way to calcuatle p value ?**

Post by:**Porcus** on **2019-02-22 09:17:25**

Post by:

The "simplest way" is, arguably, to enter the figures into an online calculator: https://www.google.com/search?q=binomial+test+calculator

The null hypothesis of wild guessing would be a probability = 1/2 each time, since you have two alternatives A and B. What you are looking for, is the chance of getting "such a wild result" as*10 or more*, if the null hypothesis is true.

If you want to do it by hand: Then https://en.wikipedia.org/wiki/Binomial_distribution#Cumulative_distribution_function .

Rather than calculating up to 10, you can calculate "from 10 and up": since probabilities are 1/2 The chance of getting 10 or 11 or 12, equals the chance of getting 2 or 1 or 0. The probability of 1/2 simplifies p^k(1-p)^{n-k} to 1/2^n.

The null hypothesis of wild guessing would be a probability = 1/2 each time, since you have two alternatives A and B. What you are looking for, is the chance of getting "such a wild result" as

If you want to do it by hand: Then https://en.wikipedia.org/wiki/Binomial_distribution#Cumulative_distribution_function .

Rather than calculating up to 10, you can calculate "from 10 and up": since probabilities are 1/2 The chance of getting 10 or 11 or 12, equals the chance of getting 2 or 1 or 0. The probability of 1/2 simplifies p^k(1-p)^{n-k} to 1/2^n.