Вы находитесь на странице: 1из 4

CSS 212: Introduction to Statistics

First Semester 2011/12 Academic Year

Lecture 14: Two sample estimation and hypothesis testing

Independent samples versus paired samples

In a typical two-sample problem, there are two random variables, X1 and X2, that have the same type of unknown parameter (e.g., means). The investigator wants to know how these parameters differ. The most important factor that must be considered in selecting the two-sample techniques for inferential problems is whether the samples are independent or paired.

The optimal estimator of the difference between two population means (

Assuming you are studying the relationship between two normally distributed populations. o The first population has unknown mean , and known standard deviation ; and o The second population has unknown mean , and known standard deviation ;

If o is the random variable representing the means of random samples of size from the first population; o is the random variable representing the means of random samples of size from the second population; and o These samples are independent, Then the two-sample statistics is the optimal, unbiased estimator of ( ). drawn drawn

We can also say that in any specific instance where two such independent samples are taken and the sample means are calculated, and , then the difference between the sample means is a point estimate of the difference between the population means ( ).

1 Lecture 14 Wilson Wesley Lazaro Jere

CSS 212: Introduction to Statistics

First Semester 2011/12 Academic Year

The theoretical sampling distribution of the difference between two means

The sampling distribution of the difference between the two means is normally distributed with:

Mean: Variance:

; ; and

Standard deviation

Note: The standard deviation of a theoretical distribution for a given statistic is called the standard error of that statistic, and as such, difference between two means. is called the standard error of the

Confidence interval for the difference between means (

We will first deal with case where standard deviations ( independent and normally distributed.

and

) are know, and the samples are

We know that We can compute Z as follows:

and

2 Lecture 14 Wilson Wesley Lazaro Jere

CSS 212: Introduction to Statistics

First Semester 2011/12 Academic Year

We can present the (1 - )100% confidence interval for

as follows:

a) b)

Example 14.1. Independent random samples are taken from two normally distributed populations that have unknown means, and , and known standard deviations. For the first population, and the sample values are and . For the second population, and the sample values are and. . What is the 95 % confidence interval for ?

Solution:

= = 0.2 0.48 or (-0.28, 0.68)

Two sample hypothesis testing

For a two-tailed test:

and

For a right-tailed test, either : : and and : : , or

For a left-tailed test, either : : and and : : , or

3 Lecture 14 Wilson Wesley Lazaro Jere

CSS 212: Introduction to Statistics

First Semester 2011/12 Academic Year

Example 14.2 For the conditions in Example 14.1, do a two-tailed test of =0.05. Solution. a) : and :

using

b) =0.05 c) Reject if calculated Z is greater that d) The value of the Z statistic is:

= 0.810

e) Since the calculated z of 0.810 is less than the tabulated z of 1.96, we fail to reject the null hypothesis. f) We conclude that we sufficient evidence that the means of the two populations are not statistically different at 5 % level of significance.

Note: When you encounter small samples of (

and

), then use the t-statistic.

4 Lecture 14 Wilson Wesley Lazaro Jere

Вам также может понравиться