Lecture 4

Example: A random sample of n = 2 households was drawn from a small colony of

5 households having daily income (in Taka) as follows:
Household 1 2 3 4 5
Income (in Tk.) 156 149 166 164 155

i. Calculate population mean 𝑌̅, variance 𝜎 2 and mean square error 𝑆 2 .

ii. Enumerate all possible samples of size 2 by the replacement method and
show that
a. The sample mean gives an unbiased estimate of the population mean and
find is sampling variance.
b. Sample variance 𝑠 2 is an unbiased estimate of the population variance
𝜎 2.
(𝑦1 −𝑦2 )2
c. 𝑣(𝑦̅) = is an unbiased estimator of 𝑉(𝑦̅), i.e.
4
𝐸[𝑣(𝑦̅)] = 𝑉(𝑦̅) = 𝜎 2 /𝑛
iii. Enumerate all possible samples of size 2 by without replacement method and
show that
a. The sample mean gives an unbiased estimate of the population mean and
find is sampling variance.
b. Sample variance 𝑠 2 is an unbiased estimate of the population variance
𝑆2.
c. 𝑣(𝑦̅) = 3(𝑦1 − 𝑦2 )2 /20 is an unbiased estimator of 𝑉(𝑦̅), i.e.
1 1 3
𝐸[𝑣(𝑦̅)] = 𝑉(𝑦̅) = ( − ) 𝑆 2 = 𝑆 2
2 5 10

Solution: The population mean of 5 households

𝑌̅ = (156 + 149 + 166 + 164 + 155)/5 = 158
The population variance
∑𝑁 ̅ 2
𝑖=1(𝑦𝑖 −𝑌 )
𝜎 2= = [(156 − 158)2 + (149 − 158)2 + (166 − 158)2
𝑁

+(164 − 158)2 + (155 − 158)2 ] = 38.80

𝑁 5
Hence, mean square error 𝑆 2 = 𝜎 2 = × 38.8 = 48.5
𝑁−1 4

1
ii. It can be seen that the total number of possible samples is 25 and each of 25
samples has the same probability 1/25 of being selected.
Sample Units Probability Sample Sample Sampling Sampling
# in the observations mean error variance
sample
𝑦1 𝑦2 𝑦̅ (𝑦̅ − 𝑌̅) (𝑦1 − 𝑦2 )2
4
(1) (2) (3) (4) (5) (6) (7) (8)
1 1,1 1/25 156 156 156 -2.0 0
2 1,2 1/25 156 149 152.5 -5.5 45/4
3 1,3 1/25 156 166 161 3.0 100/4
4 1,4 1/25 156 164 160 2.0 64/4
5 1,5 1/25 156 155 155.5 -2.5 ¼
6 2,1 1/25 149 156 152.5 -5.5 49/4
7 2,2 1/25 149 149 149 -9.0 0
8 2,3 1/25 149 166 157.5 -0.5 289
9 2,4 1/25 149 164 156.5 -1.5 225/4
10 2,5 1/25 149 155 152.0 -6.0 36/4
11 3,1 1/25 166 156 161.0 3.0 100/4
12 3,2 1/25 166 149 157.5 -0.5 289/4
13 3,3 1/25 166 166 166.0 8.0 0
14 3,4 1/25 166 164 165.0 7.0 4/4
15 3,5 1/25 164 155 160.5 2.5 121/4
16 4,1 1/25 164 156 160.0 2.0 64/4
17 4,2 1/25 164 149 156.5 -1.5 225/4
18 4,3 1/25 164 166 165.0 7.0 4/4
19 4,4 1/25 164 164 164.0 6.0 0
20 4,5 1/25 164 155 159.5 1.5 81/4
21 5,1 1/25 155 156 155.5 -2.5 ¼
22 5,2 1/25 155 149 152.0 -6.0 36/4
23 5,3 1/25 155 166 160.5 2.5 121/4
24 5,4 1/25 155 164 159.5 1.5 81/4
25 5,5 1/25 155 155 155.0 -3.0 0
158.0 19.40
Average

2
a. The expected value of 𝑦̅ is given the average value of column 6, which is 158.
Thus,
𝐸(𝑦̅) = 158 = 𝑌̅
Thus, the sample mean 𝑦̅ is an unbiased estimate of population mean 𝑌̅.
The sample variance is the mean of the squares of error which works out to 19.40
(38.80/2).
b. The sample variance (𝑠 2 ) is given by
(𝑦1 − 𝑦2 )2 /2 which is twice the value given in column (8). Hence 2 ×
𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑜𝑓 𝑐𝑜𝑙𝑢𝑚𝑛 (8) shows that 𝐸(𝑠 2 ) is equal to 38.80 (𝜎 2 ) which shows that
𝑠 2 is an unbiased estimator for the population variance in the simple random
sampling wr.
c. An estimator of 𝑉(𝑦̅) is given by
𝑣(𝑦̅) = (𝑦1 − 𝑦2 )2 /4
the values which are given in column (8) for possible samples. The expected value
of 𝑣(𝑦̅) is the average of the values in column (8), showing that
𝐸[𝑣(𝑦̅)] = 𝑉(𝑦̅) = 𝜎 2 /2.
Thus, the estimate 𝑣(𝑦̅) is unbiased.

iii. Assignment
Table: All possible samples of 2 units from 5 units in simple random sampling
without replacement
Sample Units Probability Sample Sample Sampling Sampling
# in the observations mean error variance
sample
𝑦1 𝑦2 𝑦̅ (𝑦̅ − 𝑌̅) 3 (𝑦𝑖 − 𝑦2 )2
5 4
(1) (2) (3) (4) (5) (6) (7) (8)
1 1,2
2 1,3
3 1,4
4 1,5

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

Sample Size Determination

Sample Size for Estimating Population Proportion
For large population, the size of the representative sample for proportion is
2
𝑧𝛼⁄
𝑝𝑞
2
𝑛0 =
𝑑2
where p is the estimated proportion of an attribute that is present in the population
q=1–p
𝑧𝛼2⁄ is abscissa of the normal curve that cuts off an area 𝛼/2 at the right tail
2

and d is the desired level of precision

If the population is small then the sample size can be reduced slightly as
𝑛0
𝑛= (𝑛 −1)
1+ 0
𝑁
where n is the sample size and N is the population size.

4
Sample Size for Estimating Population Mean
For Prespecified population Variance
➢ The simplest equation relating the precision and sample size comes from the
confidence intervals.
➢ To obtain absolute precision e, find the value of n that satisfies
𝑛 𝑆
𝑒 = 𝑧𝛼/2 √(1 − )
𝑁 √𝑛
➢ To solve this equation for n, we first find the sample size 𝑛0 that we would
use for an SRSWR:
𝑧𝛼 𝑆 2
𝑛0 = ( 2 )
𝑒
➢ Then the desired sample size is
2
𝑛0 𝑧𝛼/2 𝑆2
𝑛= 𝑛 = 𝑧2 2
1+ 0 𝛼/2 𝑆
𝑁 𝑒2+
𝑁
➢ Of course, if 𝑛0 ≥ 𝑁 we simply take a census with n = N.
The z-values and corresponding precision level is as follows:
Confidence level Z-score
Precision level= 𝛼
90% 1.64 0.10
95% 1.961 0.05
99% 2.58 .01

Example: Suppose we want to estimate the proportion of recipes in the Beter Homes
Gardens & New Cook Book that do not involve animal products. We plan to take an
SRS of the N=1251 test kitchen-tested recipes, and want to use 95% CI with margin
of error 0.03. Then,
1 1
(1.96)2 ( ) (1 − )
𝑛0 = 2 2 ≈ 1067.
(0.03) 2

5
Thus, desired sample size is
𝑛0 1067
𝑛= 𝑛0 = = 576
1+ 1067
𝑁 1 + 1251
For Prespecified Cost
Let an amount of money C is being allotted for collecting n observations, 𝐶0 be the
overhead cost and 𝐶1 be the cost of collecting observation for one sample unit. Then
the total cost C can be expressed as
𝐶 = 𝐶0 + 𝑛𝐶1

Or 𝑛 = 𝐶−𝐶
𝐶
0
1
Assignment:
Draw a simple random sample without replacement of size 20 from a population of
size 105 by using random number table.
Sl.# 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Value 75 83 56 76 58 78 56 76 87 65 45 36 37 88 65
Sl.# 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Value 37 45 67 45 65 76 67 56 65 45 54 34 43 45 54
Sl.# 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
Value 75 Sl.# 56 76 58 78 56 76 87 65 45 36 37 88 65
Sl.# 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Value 37 45 67 45 65 76 67 56 65 45 54 34 43 45 54
Sl.# 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75
Value 75 83 56 76 58 78 56 76 87 65 45 36 37 88 65
Sl.# 76 77 78 79 80 81 82 82 84 85 86 87 88 89 90
Value 37 45 67 45 65 76 67 56 65 45 54 34 43 45 54
Sl.# 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
Value 75 57 56 76 58 78 56 76 87 65 45 36 37 88 65

̅ 𝑠 2 , 𝑉(𝑦̅) of the selected sample.

Calculate: 𝑦,