Sample size:

 The size of the sample is an important factor in research

activity because it has direct bearing on accuracy,
estimation, cost and administration of the survey.
 Large sample size may increase the cost, time but it
reduces the sampling error. On the other hand, small
sample reduces the cost, time but it increases the
sampling error. Thus, the size of the sample should be
optimum.
 The optimum sample size is that size of the sample
which fulfills the requirement of representativeness,
efficiency, precision, reliability and flexibility.
 However, the following factors greatly influence the
sample size;
@ Dinesh Shrestha 1
Factors affecting Sample size:
 However, the following factors greatly influence the
sample size;
1. Nature of population.
2. Number of classes.
3. Nature of the study.
4. Types of the sampling.
5. Degree of accuracy.

@ Dinesh Shrestha 2
Sample Size for Quantitative Data:
❑ Estimation of sample size using mean:
➢ Let 𝑥ҧ be the mean of a random sample of size ‘n’ drawn from a
population with mean (μ) and standard deviation (σ), then we have,
ҧ
𝑥−𝜇
Z= 𝜎
𝑛

n(x − ) Where,
→Z = n = required sample size
 σ = standard deviation
Squaring on both sides, we get
d = Margin of error or, difference
→ Z = nd between sample mean and
Z population mean
→ n= = |𝑥ҧ - μ|
d α = level of significance
𝑍𝛼 = significant value of Z at α
( Z  ) 2
n = level of significance (for two
d2 tailed test).
@ Dinesh Shrestha 3
 If population standard deviation ‘σ’ is not known
then it is estimated by using sample standard
deviation ‘s’ i.e. σ = s
 For the finite population size N,

𝑛
 Sample size = 𝑛
1+
𝑁
 Where, ( Z  ) 2
n =
d2

@ Dinesh Shrestha 4
Sample Size for Qualitative Data:
❑ Estimation of sample size using proportion:
➢ Let ‘p’ be the proportion of a sample of size ‘n’ drawn from a population
having proportion (P) then we have,

Z=
𝑝−𝑃 n ( p − P)
𝑃𝑄 →Z = Where,
𝑛
PQ n = required sample size
Squaring on both sides, we get P = population proportion
Q = 1-P
→ Z PQ = nd d = |p - P|
Z Error or difference between
→ n =   PQ sample proportion and
d population proportion.
2
Z  𝑍𝛼 = significant value of Z at α
 n =    PQ level of significance (for two
 d 
tailed test).
Remark: If the value of P or Q is not given and the given condition is related
to the quality or attributes, then the value of P or Q is used as 50% or 0.50.
@ Dinesh Shrestha 5
 If population proportion ‘P’ is not given then it is
estimated by using sample proportion ‘p’.
 For the finite population size (N)
𝑛
 Sample size = 𝑛
1+
𝑁
 Where, 2
 Z 
n =   PQ
 d 
 If coefficient of variation (CV) is given then the
sample size is given by
Z 2 (CV ) 2
n=
d2
@ Dinesh Shrestha 6
 Relationship of sample size with desired
level of error (sampling error):
 In statistics, we are interested to estimate the population parameters (like
population mean (μ), population standard deviation (σ), population
proportion (P) etc.) from the sample statistic (like sample mean (𝑥), ҧ sample
standard deviation (s), sample proportion (p)etc.). In estimation, we can not
estimate the exact value of the population parameter from the sample statistic
because the various factors like sample size, level of significance and nature of
parameters affect the value of the population parameter. Thus, there may arise
sampling error (margin of error) while estimating the parameters from the
sample statistic.
 Suppose, 𝑥ҧ be the mean of a random sample of size ‘n drawn from the
𝜎
normal population with mean (μ) and standard deviation
𝑛
 We have,
ҧ
𝑥−𝜇 𝑑
 Zα = 𝜎 = 𝜎
𝑛 𝑛

𝜎
 ∴ d = 𝑍𝛼 ………….. (i)
𝑛

@ Dinesh Shrestha 7
 Where,
n = sample size
σ = population standard deviation
d = Margin of error or, difference between sample mean and
population mean
= |𝑥ҧ - μ|
α = level of significance
𝑍𝛼 = significant value of Z at α level of significance (for two tailed
test).
 Similarly,
 Suppose, ‘p’ be a sample proportion of a random sample of size ‘n’
drawn from the normal population with proportion ‘P’.
 We have,
𝑝−𝑃 𝑑
 Zα = =
𝑃𝑄 𝑃𝑄
𝑛 𝑛

𝑃𝑄
 ∴ d = 𝑍𝛼 ………….. (ii)
𝑛

@ Dinesh Shrestha 8
 Where,
n = sample size
P = population proportion
Q = 1-P
d = |p - P|
= error or difference between sample proportion
and population proportion.
𝑍𝛼 = significant value of Z at α level of significance
(for two tailed test).

 From the above two equations (i) and (ii), we can conclude
that there is inverse relationship between sampling error or
margin of error (d) and sample size (n). i.e. as the sample
size increases, the margin of error decreases.

@ Dinesh Shrestha 9