Sampling

MAT 3103: Computational Statistics and Probability

Chapter 9: Sampling

Sampling:
Sampling is the process of selecting a small group (sample) from a larger group (population)
to learn something about the whole. Since collecting data from an entire population is often
expensive, time-consuming, or impossible, we collect data from a sample and use it to draw
conclusions.

 Public health survey: The Bangladesh Bureau of Statistics (BBS) cannot visit every
household. So, they select a sample of villages and families to collect data about health,
education, or income.
 Election polls: TV channels like Somoy TV or Channel i often report public opinion
using surveys. They don’t ask every citizen, just a representative sample of voters.
 Network performance analysis (CSE): Suppose you are developing a mobile app or
testing a Wi-Fi network on a university campus. Instead of checking every device or
location, you can sample a few devices or specific zones (like classrooms, labs, or
canteen) to test the app’s speed or connection stability.
 IoT sensor data collection (EEE): In a smart farming project in Rangpur, a large
number of temperature or humidity sensors are installed. Instead of analyzing data from
all sensors every second, you can sample data from 10 sensors per hour to monitor
performance and detect faults.
 Fault detection in circuit production (EEE): If you are producing microcontroller
boards at a small hardware lab in Dhaka, you don’t have to test every single board. You
can randomly sample 10 boards out of every 100 to check for manufacturing defects.
 User feedback for software testing (CSE): A startup team developing a ride-sharing
app like Pathao wants feedback before launch. Instead of asking all users, they select a
sample of users from Dhaka, Chattogram, and Sylhet to get early insights.

Different Methods of Sampling are:

There are several different sampling techniques available. In random sampling, we start with a
complete sampling frame of all eligible individuals from which we select our sample. In this

1
 Sampling

way, all eligible individuals have a chance of being chosen for the sample, and we will be more
able to generalize the results from our study.
i) Simple random sampling, ii) Systematic sampling, iii) Circular systematic sampling iv)
Stratified random sampling, iv) Cluster sampling, etc.

Simple Random Sampling:

Simple random sampling is a sampling technique where every unit in the population has an even
chance and likelihood of being selected in the sample. Let there be N units in a population. We
need to select a random sample of size n (n ¿ N). The possible number of samples, without
1
replacement, are N c . If any of these N c samples are selected with equal probability N , then the
n n
nc

sampling is simple random sampling. In other words, if every unit of N units is selected with
1
equal probability , then the sampling is simple random sampling. Let, there are 40 students in
N
a Math class at AIUB. The teacher wants to select a student as the class monitor. He makes 40
slips, write the IDs of the students on them distinctly and put them in a box. After shuffling the
slips, he picks one up randomly and declares the student whose ID is there on the selected slip as
1
the class monitor. Here, each single student has equal probability of being selected as the
40
class monitor.
Systematic Sampling:
Systematic sampling is a type of probability sampling method in which sample members from a
larger population are selected according to a random starting point (R) but with a fixed, periodic

interval. This interval, called the sampling interval ( k= Nn ), is calculated by dividing the
population size by the desired sample size.

2
 Sampling

Circular Systematic Sampling:

In this method, we assume the listings to be in a circle such that the last unit is followed by the
first. A random start is chosen from 1 to N. We then add the intervals k until exactly n elements
are chosen. If we come to the end of the list, you continue from the beginning.

N
N = 20, n = 4, k = = 5.
n

Random start, R = 7.

7, 12, 17, and 2 are selected.

3
 Sampling

Some formulas to estimate different statistic values:

1
Σx
The estimate of sample means, x̄ = n .
1 2 (Σx )
2
s2 = Σx − ]
The estimate of sample variance, n −1 [ n .

The estimate of the variance of sample means,

The estimated standard error of sample means, s.e. ( x̄ ) = √ v ( x̄ ).

The estimate of population total,

The estimate of variance of the estimate of population total,

4
 Sampling

Estimate of proportion: Let

N = Number of population units,
n= Number of sample units,
A = Number of units in the population possessing a particular character,
a = Number of sample units possessing that particular character,
P = A /N =Proportion of population units possessing that particular character,
P = a /n = Proportion of sample units possessing that particular character.
This p is an unbiased estimate of P. The estimate of variance of proportion is given by
N −n
pq −¿
v (p) = N (n − 1) ,q=1 p

Problem 9.1: Number of bugs found in various software modules during testing are, X: 4, 7, 2,
9, 3, 12, 5, 6, 7, 4, 7, 8, 5, 3, 11, 10, 6, 13, 8, 9, 10, 6, 12, 11, 8, 7, 9, 7. Select a random sample
of 6 modules by (i) simple random sampling, (ii) systematic sampling, (iii) circular systematic
sampling.

1. Estimate the mean number of bugs per module.

2. Estimate the total number of bugs in all 28 modules.
3. Estimate standard error of estimated mean number of bugs.
4. Estimate standard error of estimated total number of bugs.
5. Find 95% confidence interval for mean number of bugs.
6. Suggest a sample of size n to estimate a population mean with margin of error 0.5 at 95%
level of confidence, where variance of the population observations is 5.75.
7. Estimate the proportion of modules with less than 7 bugs.
8. Estimate the variance of the estimated proportion.
9. Find sample size n to estimate proportion 0.6 with margin of error 0.1 at 95% confidence.

Observations (x) 4 7 2 9 3 12 5 6 7 4 7 8 5 3
Serial Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Observations (x) 11 10 6 13 8 9 10 6 12 11 8 7 9 7
Serial Number 15 16 17 18 19 20 21 22 23 24 25 26 27 28
5
 Sampling

i) We have N = 28. We need to select a sample of size n = 6 using Random Number Table. We
can use any row or any column of the table. Let us use Column 1. Here N =28, the last serial
number is of two digits. So, we need to select a random number of two digits. The selected
random numbers and the selected number of bugs of different modules are shown below:
Random Numbers 16 11 10 19 17 9
Bugs of selected modules 10 7 4 8 6 7

ii) Let N = nk, k = N/n. In our case k = 28/6 = 4.7 ~ 5. We have to select first observation from
first k = 5 observations using Random Number Table. After that every kth =5th observation is
selected. Five is a number of one digit, so we need to select a random number of one digit first.
The selected random numbers and the selected observations are shown below: [using column 2
of random number table]

Random Numbers 3 8 13 18 23 28
Bugs of selected modules 2 6 5 13 12 7
iii) Here also k = N/n = 4.7 ~ 5. First observation is selected from all observations using random
number table. After that every kth = 5th observation is selected moving through a circle. We have
n = 28 observations and 28 is a number of two digits. So, we need to select a random number of
two digits. The selected random number and the selected observations are shown below: [Using
column 3 of random number table]
Random Numbers 10 15 20 25 2 7
Bugs of selected modules 4 11 9 8 7 5

1 42
1. Estimate of mean, x̄= Σx= =7. [Calculation is from Simple random sample]
n 6

2. Estimate of total, = 28 × 7 = 196.

3. The standard error of estimate of mean is, s.e.( x̄ ¿=√ v ( x̄ )=√ 0.5238=0.7237.
N−n 2 28−6
The variance of sample mean is v ( x̄)= s= 4=0.5238 .
Nn 28 ×6

[ ]
2 2
1 (∑ x ) 1 ( 42 )
s2= ∑ x 2− = (314 – )=4
n−1 n 5 6
4. The estimate of standard error of estimate of population total is given as follows:

6
 Sampling

v(^ X )=√ v ( ^
X )=N v ( x̄ )=2 8 0.5238=410.7∧s . e . ( ^ X ) =√ 410.7=20.27
2 2

5. 95% confidence limits for mean are given as follows:

x l = x −t n−1 s.e. ( x ) ¿ 7 – 2.571 (0.7237)=5.14 , Here, t n−1 = t 5 = 2.571
x u = x + t n−1 s.e. ( x ) ¿ 7+2.571(0.7237)=8.86
2
z v (x )
6. The sample size n is given by n= 2
=¿ ¿
d
Here z is the tabulated of normal distribution at 5% level = 1.96, d = margin of error.
7. The estimate of proportion of modules with less than 7 bugs is given by
p=a/n=2/6=0.33 . Here a = number of modules with less than 7 bugs = 2.
8. The estimated variance of p is given by

v(p) = = . Here q = 1 – p = 1 - 0.33 = 0.67

2
z pq
9. The sample size n is given by, n= 2
=¿ ¿
d

Unbiasedness of mean and variance in case of simple random sampling:

Let us consider that, in a population there are N = 3 units. The unit values are x: 2, 4, 6. We can
select a sample of size n = 2. The possible number of samples, without replacement, are 3c = 3. 2

Samples 1 ( Σ x )2
Σx 2
x̄
Sample Means, = n Σx − ]
Sample Variances, [ n
2, 4
1 2+4 1 2 2 ( 2 +4)2
x̄ 1 = Σ x = =3 s21 = 2 −1 [(2 + 4 ) − ]=2
n 2 2
2, 6 2+6 1 ( 2+ 6)2
x̄ 2 = =4 s 22 = 2 2
[ ( 2 +6 )− ] =8
2 2 −1 2
4, 6 4+ 6 1 ( 4 +6 )2
x̄ 3 = =5 s 23 = 2 2
[ ( 4 +6 ) − ]= 2
2 2 −1 2

7
 Sampling

(x)
Sample mean is an unbiased estimate of population mean as: E =

variance is an unbiased estimate of population variance as: E( s ) =

2

Sample

Exercise 9
9.1 Define sampling and simple random sampling with example.
9.2 Show, by example, mean of simple random sample is an unbiased estimate of population mean.
9.3 Show, by example, variance of simple random sample is an unbiased estimate of population variance.

9.4 A server in a telecom company in Dhaka recorded the number of incoming signals on 25
different days such as: X: 5, 8, 7, 10, 7, 6, 9, 11, 4, 2, 7, 7, 12, 9, 11, 3, 7, 8, 5, 6, 7, 6, 9, 11, 4.
You are asked to select 4 days using systematic sampling. Estimate the average number of

8
 Sampling

signals received per day along with the standard error of the estimate. Estimate the
proportion of days when less than 8 signals were received.

9.5 A radio transmission station in Rajshahi recorded the number of faded-out signals sent each
day over 30 days as: X: 4, 3, 0, 2, 6, 7, 4, 3, 2, 0, 1, 0, 3, 0, 6, 8, 0, 1, 4, 3, 2, 6, 3, 7, 5, 8, 0, 2, 3,
5. Select a simple random sample of 5 days. Estimate the total number of faded signals along
with the estimated standard error.

9.6 A researcher is studying the frequency of severe signal fading in a mobile tower in Sylhet.
From past data, it is believed that on about 45% of the days, the number of faded signals is
more than 10. What should be the sample size to estimate this proportion with a margin of
error 0.10 and 95% confidence level?

9.7 The number of emails received by a university server in Bashundhara residential area over 23
different days is recorded as: X: 10, 7, 6, 9, 11, 4, 2, 7, 7, 9, 11, 45, 8, 7, 10, 7, 6, 9, 11, 4, 2, 7, 7.
Select 4 days using simple random sampling. Estimate the average number of emails
received per day and calculate the standard error of the estimate.

9.8 It is known that on approximately 30% of the days, the number of faded signals exceeds 10
at a TV transmission station in Chattogram. What sample size is required to estimate this
proportion with a margin of error 0.05 at 95% confidence level?

9.9 An electronic device in a laboratory at BUET produced the following number of noisy bits in
20 different trials: X: 5, 8, 7, 10, 7, 6, 9, 11, 4, 2, 7, 7, 12, 9, 11, 3, 7, 8, 5, 6. Select a random
sample of 5 trials using circular systematic sampling. Construct a 95% confidence interval
for the average number of noisy bits produced by the device.