Query about Point bi-serial correlation coefficient

[Deepshikha Sharma]

Hello,

As per solution of Question 3 in Mock test, for sample of 10 employees ,the Point bi-serial correlation coefficient is computed using Population standard deviation.

But in the Tutorial 4 of week 4 lectures ,the point bi-serial correlation coefficient is calculated using Sample standard deviation.

Please confirm we have to use sample standard deviation or population standard deviation for calculating Point bi-serial correlation coefficient ?

Thanks,

Deepshikha

[Jai Hanani]

It depends on the question. In the mock test question, the whole population is given(10 employees and their salaries), therefore we had to use Population SD. 
In that Chennai COVID-19 question in one of the graded assignments, we will have to use Sample SD, because the whole population cannot be possibly given.

[Deepshikha Sharma]

In the mock test question ,sample is given and not population ,Still Population standard deviation is  used to calculate point bi serial coefficient.

Mock test Question 3  is " Use the following information and data given in Table S.1 to answer the questions 3 and 4

In an organization, data from a sample of 10 employees is collected. This data includes gender, age, and salary of employees and is given in Table S.1."

Please check and confirm.

Thanks,

Deepshikha Sharma

[CG]

I am also interested in knowing the answer to this. Hope the course team will respond. 

[Statistics 1 Support 1]

Hello,

You can use the sample standard deviation for calculating the point bi-serial coefficient, but in that case you need to use the sample correction in the formula. Instead of using sqrt(p0 *p1), use sqrt[(n0/n-1)(n1/n)].  But if you are using population standard deviation for calculation, then use sqrt(p0*p1).

In tutorial 4, there is a mistake, we will send an announcement.

Thanks

Nikita

Statistics course support team

[Deepshikha Sharma]

Hi Nikita,

Thank you so much for giving clarity on above query.

Thanks & Regards,

Deepshikha Sharma

[Jai Hanani]

Indeed. I was mistaken. You are right. 

[Shinas MN]

So in short in shift 2 stat bi-serial ...Should we take N or N-1when we multiply it with root p0*p1?