Upload
vivien-kelly
View
219
Download
0
Embed Size (px)
Citation preview
There are two main purposes in statistics;• (Chapter 1 & 2) Organization & ummarization of the data[Descriptive Statistics]• (Chapter 5) Answering research questions about some population parameters [Statistical Inference] Statistical Inference (1) Hypothesis Testing:Answering questions about the population parameters (2) Estimation:Approximating the actual values of Parameters; Point Estimation Interval Estimation (or Confidence Interval)
We will consider two types of population parameters: (1) Population means (for quantitative variables): μ =The average (expected) value of some quantitative variable. Example: • The mean life span of some bacteria.• The income mean of government employees in Saudi Arabia. (2) Population proportions (for qualitative variables)
Population in the elements of no.Total
sticcharacteri specified some with Population in the elements of no.
Example:• The proportion of Saudi people who have some disease.• The proportion of smokers in Riyadh• The proportion of females in Saudi Arabia
Estimation of Population Mean :- Population (distribution) Population mean =μ Population Variance 2
Random of sizeSample n
Sample mean Sample Variance
nXXX ,, 21 X
2S
11
1
1
222
1
2
1
n
XnXXX
nS
Xn
X
in
ii
n
ii
We are interested in estimating the mean of a population (I)Point Estimation: A point estimate is a single number used to estimate (approximate) the true value of .• Draw a random sample of size n from the population: is used as a point estimator of .
nXXX ,, 21
n
iiX
nX
1
1
(II)Interval Estimation: An interval estimate of μ is an interval (L,U) containing the true value of μ “ with probability “ is called the confidence coefficientL = lower limit of the confidence intervalU= upper limit of the confidence interval • Draw a random sample of size n from the population .
1
nXXX ,, 21
1
Result: If is a random sample of size n from a distribution with mean μ and variance , then: A 100% confidence interval for is : (i) if is known:
nXXX ,, 21 2
1
nZX
nZX
2
12
1, OR
nZX
2
1
(ii) if is unknown.
n
SZX
n
SZX
21
21
, OR n
SZX
21
Example: [point estimate of is ]Variable = blood glucose level (quantitative variable)Population = diabetic ketoacidosis patients in Saudi Arabia of age 15 or more
parameter = μ= the average blood glucose level (large)
( is unknown)
123n2.26X
3.3S 2
2.26X
(i) Point Estimation: We need to find a point estimate for μ . is a point estimate for μ .(ii) Interval Estimation (Confidince Interval):
We need to find 90% confidence interval for μ .
2.26X
90% = 100%
90% confidence interval for μ is:
1
1.09.01 05.0
2
95.0
21
645.195.0
21
ZZ
n
SZX
n
SZX
21
21
,
or
123
3.3645.12.26,
123
3.3645.12.26
4894714.02.26,4894714.02.26
689471.26,710529.25or
we are 90% confident that the mean μ lies in (25.71,26.69) or
25.72 < μ < 26.69
5.4.Estimation for a population proportion:-
• The population proportion is (π is a parameter)
where number of elements in the population with a specified characteristic “A”
N
AN
AN
N = total number of element in the population (population size)• The sample proportion is
n
Anp (p is a statistic)
where number of elements in the sample with the
same characteristic “A”n = sample size
An
Result: For large sample sizes , we have 30n
1,0
1N
n
pZ
Estimation for π : (1) Point Estimation:A good point estimate for π is p.
(2) Interval Estimation (confidence interval): A confidence interval for π is %1001
n
ppzp
1
21
or
,1
,1
21
21
n
ppzp
n
ppzp
Example 5.6 (p.156)variable = whether or not a women is obese (qualitative variable)population = all adult Saudi women in the western region seeking care at primary health centersparameter = π = The proportion of women who are obese
n = 950 women in the sample n (A) = 611 women in the sample who are obese
643.0
950
611
n
Anp
is the proportion of women who are obese in the sample. (1) A point estimate for π is p = 0.643 (2) We need to construct 95% C.I. about π .
975.02
1
025.02
05.0
195.0%1001%95
96.1975.0
21
zz
95% C.I. about π is
n
ppzp
1
21
01554.096.1643.0950
643.01643.096.1643.0
or
or 0305.0643.0
6735.0,6127.0or
We can 95% confident that the proportion of obese women, π , lies in the interval (0.61,0.67) or
0.61 < π < 0.67