Scribd
Upload
86 views20 pages

Estimation of Parameters

Okay, here are the step-by-step workings: * Sample size (n) = 1,200 * Number of observations in Class A (pa) = 550 * pa = 550/1200 = 0.458 * 1 - pa = 1 - 0.458 = 0.542 * z0.995/2 = 2.576 (from normal distribution table) * 99% Confidence Interval = pa ± 2.576(√(pa(1-pa)/n)) = 0.458 ± 2.576(√(0.458×0.542/1200)) = 0.458 ± 0.022 = (0.436, 0.480) So

Uploaded by

JohnRoldGleyPanaligan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
86 views20 pages

Estimation of Parameters

Okay, here are the step-by-step workings: * Sample size (n) = 1,200 * Number of observations in Class A (pa) = 550 * pa = 550/1200 = 0.458 * 1 - pa = 1 - 0.458 = 0.542 * z0.995/2 = 2.576 (from normal distribution table) * 99% Confidence Interval = pa ± 2.576(√(pa(1-pa)/n)) = 0.458 ± 2.576(√(0.458×0.542/1200)) = 0.458 ± 0.022 = (0.436, 0.480) So

Uploaded by

JohnRoldGleyPanaligan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 20

ESTIMATION OF

PARAMETERS
from the presentation of

Dr. Josefina V. Almeda


Professor, UP Diliman School of Statistics

April 1-4, 2014


INFERENTIAL STATISTICS

an area in Statistics that deals with methods used


to make generalizations/inferences about some
characteristics of the population based on
information contained in a sample

Types of Inferences
• estimating the value of the parameter of interest
• hypothesis testing - making decisions on whether
or not the sample agrees with the researcher’s
assertion regarding some characteristic of the
population
Example:
Research Problem: How effective is Minoxidil in treating
male pattern baldness?
Specific Objectives:
1. to estimate the population proportion of patients who
will show new hair growth after being treated with
Minoxidil

2. to determine whether treatment using Minoxidil is


better than the existing treatment that is known to
stimulate hair growth among 40% of patients with male
pattern baldness

Question: How do we achieve these objectives using


inferential statistics?
Forms of Estimation

Point estimation
uses information in a sample to arrive at a single
number that will serve as an estimate of the value of the
target parameter

Interval estimation
uses sample data to calculate the lower and
upper bound of an interval such that the researcher can
be highly confident that this interval contains the value
of the target parameter
• The objective of estimation is to determine the
approximate value of a population parameter on the basis
of a sample statistic.

• Point Estimator

• Interval Estimator

10.5
• For example, suppose we want to estimate the mean summer income of a
class of business students. For n=25 students,
• Mean is calculated to be 400 pesos/week.

point estimate interval estimate

• An alternative statement is:


• The mean income is between 380 and 420 pesos /week.
Table 4.1. Estimation of the Population Mean ()
Under Simple Random Sampling

Type of Sampling

Without With
Replacement Replacement

Point estimate
X X
Standard error  n 1

1 n
n N 1
Estimate of
s
standard error s
1
n
n N n
• The standard error is used to measure how well
the obtained statistic will estimate the target
parameter. Its unit is the same as the unit of the
data values.

• The smaller the standard error, the better the


statistic estimates the parameter; that is, if the
standard error is small then one can expect that
the estimate will not differ substantially from the
target parameter.

• The above estimate of the standard error tends to


underestimate the true standard error but this
becomes negligible as the sample size increases.
Table 4.2. Estimation of the Proportion of Elements in the
Population Belonging in Class A (PA) Under
Simple Random Sampling

Type of Sampling

pa (1  pa ) 
n 1 
n
 1 
N
Without With
Replacement Replacement
Point estimate no. of obsvns.in sample
that belong in Class A
pa
pa 
n
PA (1  PA )
Standard error PA (1  PA )  n  1 
 1  n
n  N  1
Estimate of pa (1  pa )
standard error
n 1
Example:

A simple random sample of 20 households was drawn


from a barangay containing 250 households. The numbers
of persons per household in the sample were as follows:

5 6 3 3 2 5 4 7 8 10
4 4 5 8 9 6 6 7 6 6

a) Estimate the mean number of persons per household of


the barangay.
b) Estimate the proportion of households in the barangay
that have more than 5 members.
c) Estimate the standard error of the estimates in (a) & (b) if
sampling is done with replacement.
d) Estimate the standard error of the estimates in (a) & (b) if
sampling is done without replacement.
(1-)100% Confidence Interval Estimation
(where 0 <  < 1)

Formulas used to construct the (1-)100%


confidence interval of a parameter were derived
in such a way that if the experiment were
repeated many times and the confidence interval
is computed each time then (1-)100% of these
intervals are expected to contain the true value of
the parameter and the remaining 100% are not
expected to contain it.
Example:
Suppose the population of interest contains 8 elements
and the measurements of the variable of interest taken
from each one of these elements are as follows: {0, 1, 2, 2,
2, 2, 3, 4}. Take note that the mean score is 2. Select a
sample of size 4 using simple random sampling with
replacement. Compute for the 95% confidence interval
for  using the following formula:
   
 x  z / 2 , x  z / 2 
Lower bound –  n n
 1.118 
x 1.96  
 4 
Upper bound
 1.118 
x 1.96  
 4 
Repeat this process 500 times; that is, repeatedly,
select a sample of size 4 and each time compute
for the 95% confidence interval for . What
percentage of these intervals contain the true
value of =2?

Note: Once the (1-)100% confidence interval is computed from


the sample, one can never be certain that the interval actually
contains the true value. Thus, one only states that “he is (1-
)100% confident that the interval contains the parameter”.
Considerations in Setting 

• a smaller  means more possible samples of size n will


yield intervals that contain the true value of the parameter

• a larger  means shorter intervals

Common Choices of : 0.01, 0.05, 0.10

(1-)100% Confidence Interval for the Population


Mean
• when  is known
   
 x  z / 2 , x  z / 2 
 n n

where z/2 is the (1- /2)100th percentile of the Normal


distribution

Common values of z/2


 0.01 0.05 0.10
z/2 2.576 1.96 1.645
(See Appendix B2 for other values of z/2.)
• when  is unknown
 s s
 x  t( / 2,v  n 1) , x  t( / 2,v  n 1) 
 n n
where t(/2,v=n-1) is the t-value with v=n-1
(See Appendix B3 for t-values.)

Remarks:

• The above formulas hold strictly if the population


from where the sample was taken is Normally
distributed. However, they provide good
approximate (1-)100% confidence intervals when the
distribution is not Normal provided the sample size is
large, that is, n>30.
• If  is unknown and n>30, use  s s 
 x  z / 2 , x  z / 2 
 n n
Examples:
1. Forty-seven sedentary men were studied over a
one-year period. These men tried to lose weight
through exercise alone. The results of the study
showed that the mean weight loss of these men was
0.7 kilograms with a standard deviation of 4.8. Find
the 95% confidence interval for the population
mean and interpret the result.
2. Assume that the weekly traffic on the South
Expressway is normally distributed. Find a
99% confidence interval for the population
mean weekday traffic volume and interpret it
if the observed volumes for five weeks were as
follows:

22,130 18,465 25,616 22,440 19,869


An Approximate (1-)100% Confidence Interval for PA

 pa (1  pa ) pa (1  pa ) 
 pa  z / 2 , pa  z / 2 
 n n 
where z/2 is the (1-/2)100th percentile of the Normal
distribution

• This approximate solution is appropriate for


large samples only (n >30).
Example:

A national poll based on interviews with 1,200 voters in a


presidential campaign showed that 550 will be voting
Candidate A. Calculate a 99% confidence interval for the
true proportion of voters supporting Candidate A.

You might also like