Estimation

Interval Estimation
In many practical problems, the exact knowledge of the parameter may not be necessary. It is quite adequate if an interval along with a probability statement is specified such that the probability that the random interval will cover the unknown parameter is a specified number.
Statistical Methods Usha A. Kumar, IIT Bombay

Confidence intervals

Let The population distribution be normal. The value of the population standard deviation be known.

Statistical Methods

Usha A. Kumar, IIT Bombay


2
1
Z
0

X Z

Statistical Methods

Usha A. Kumar, IIT Bombay

Random interval
X Z

X +Z

X Z

Statistical Methods

Usha A. Kumar, IIT Bombay

Random Interval

For = 0.05 , the random interval is

X 1.96 , X + 1.96 n n

The probability is 0.95 that the random interval includes or covers the true value of .

Statistical Methods

Usha A. Kumar, IIT Bombay

Confidence Interval for

X1 1, X 2 = If after observing = x= x2 ,, X n xn , we compute the observed sample mean x and then substitute x in the random interval, the resulting fixed interval is called the 100(1 )% confidence interval for .

Statistical Methods

Usha A. Kumar, IIT Bombay

Confidence Interval for

The confidence interval can be expressed as

,x +Z x Z is a 100(1- )% CI for 2 2 n n or x+Z x Z with 100(1- )% confidence. 2 2 n n

Statistical Methods

Usha A. Kumar, IIT Bombay

Example
Suppose it is known that the life of electric tubes manufactured by a company is normally distributed with a standard deviation of 50 hours. A random sample of 49 pieces observed for life estimation showed an average life of 1180 hrs. Compute a 95 percent confidence interval for the population mean of tubes manufactured by this company.
Statistical Methods Usha A. Kumar, IIT Bombay

Example
A new drug discovered to decrease blood pressure was tested on 100 patients which shows a mean decrease of 18 units and a sample standard deviation of 6 units. Find the 99% confidence interval for the mean decrease in blood pressure due to the new drug.
Statistical Methods Usha A. Kumar, IIT Bombay

Large sample Confidence Intervals

If n is sufficiently large, the standardized variable

has approximately a standard normal distribution. This implies that x z / 2 n is a large sample confidence interval for with confidence level approximately 100(1 )% . This formula is valid regardless of the shape of the population distribution.
Statistical Methods Usha A. Kumar, IIT Bombay

X Z= / n

Large sample Confidence Intervals - unknown

Consider the standardized variable

X Z= s/ n

Both X and vary in value from one sample to another. However, for large n the substitution of S for adds little extra variability and hence this variable also has approximately a standard normal distribution.
Statistical Methods Usha A. Kumar, IIT Bombay

Example
A market survey is conducted to ascertain the proportion of smokers smoking a particular brand A. Out of 100 smokers surveyed, 64 were found to be smoking that brand. Find a 95% confidence interval for the proportion of smokers smoking brand A.
Statistical Methods Usha A. Kumar, IIT Bombay

Confidence interval for Population proportion

Let p be the proportion of successes in a population. If np 10 and nq 10,

P z / 2 < p p < z / 2 1 p (1 p ) / n

Statistical Methods

Usha A. Kumar, IIT Bombay

Confidence interval for Population proportion

A confidence interval for a population proportion p with confidence level approximately 100(1 )% is
2 2 2 2 z / 2 z / 2 pq z / 2 pq z / 2 z / 2 + 2 p+ + z / 2 + p+ 2n n 4n , 2n n 4n 2 2 2 1 + ( z / 2 ) / n 1 + ( z / 2 ) / n

The approximate confidence limits are

p z / 2
Statistical Methods

pq n
Usha A. Kumar, IIT Bombay

Example
The length of time required for persons taking the civil service test is assumed to be normally distributed. A random sample of 16 persons taking the test is conducted and their test times are recorded, yielding an average test time of 60 minutes with a standard deviation of 12 minutes. Find a 95 percent confidence interval for the population mean test time.
Statistical Methods Usha A. Kumar, IIT Bombay

The t distribution

The population has a normal distribution. The value of the population standard deviation is unknown.

Statistical Methods

Usha A. Kumar, IIT Bombay

Result

When X is the mean of a random sample of n from a normal distribution with mean , the rv
T= X S/ n

has a probability distribution called a t distribution with n-1 degrees of freedom.

Statistical Methods Usha A. Kumar, IIT Bombay

The t Distribution
Developed by British statistician, William Gosset A family of distributions -- a unique distribution for each value of its parameter, degrees of freedom (d.f.) Symmetric, Unimodal, Mean = 0, Flatter than Z

Statistical Methods Usha A. Kumar, IIT Bombay

The density of T

The density of T is
2 ( v +1) / 2

[(v + 1) / 2] t = f (t ) 1 + v (v / 2) v

, < t <

where v is the degrees of freedom.

Statistical Methods

Usha A. Kumar, IIT Bombay

Standard Normal t (d.f. = 25) t (d.f. = 5) t (d.f. = 1)

-3

-2

-1

Statistical Methods

Usha A. Kumar, IIT Bombay

Confidence Interval for

Let x and s be the sample mean and sample standard deviation computed from the results of a random sample from a normal population with mean . Then a 100(1 )% confidence interval for is s s
, x + t / 2,n 1 x t / 2,n 1 n n

Statistical Methods

Usha A. Kumar, IIT Bombay

Example
A random sample of 20 teletype operators indicates that their salaries fluctuate quite a bit. The sample standard deviation of their daily salaries is Rs. 90. Construct a 90 percent confidence interval on the population standard deviation of the daily salaries.

Statistical Methods

Usha A. Kumar, IIT Bombay

Result

Let X 1 , X 2 ,, X n be a random sample from a normal distribution and 2 . Then the with parameters rv
(n 1) S
2

( X i X )

has a chi squared probability distribution with n-1 df.

Statistical Methods Usha A. Kumar, IIT Bombay

Selected 2 Distributions
df = 3 df = 5 df = 10

0
Statistical Methods Usha A. Kumar, IIT Bombay

Chi square density function

The continuous random variable X has a chi-squared distribution, with v degrees of freedom, if its density function is given by
f ( x) 1 x v / 21e x / 2 , x > 0 2v / 2 (v / 2)

where v is a positive integer.

Statistical Methods Usha A. Kumar, IIT Bombay

Confidence Interval for 2 .

2 ( n 1) S 2 2 < / 2, n 1 = 1 P 1 / 2, n 1 < 2 or ( n 1) S 2

/ 2, n 1
2

< <
2

( n 1) S 2

12 / 2, n 1

Thus ( n 1) s 2 ( n 1) s 2 , 2 2 is a 100(1- )% confidence / 2, n 1 1 / 2, n 1 interval for 2 .

Statistical Methods Usha A. Kumar, IIT Bombay