मालवीय राष्ट्रीय प्रौद्योगिकी संस्थान जयपु र

Malaviya National Institute of Technology Jaipur

JLN MARG, JAIPUR – 302017 (RAJASTHAN, INDIA)
(An Institute of National Importance fully funded by the Ministry of Education, Govt. of India)
www.mnit.ac.in
21BMT901 Survey Research Method

Sampling Design
Sample Size Determination

Instructor
Deepak Verma
PhD, MBA (IT/Marketing), ADSE (DBMS)
Department of Management Studies

Sampling
…the act, process, or technique of selecting a suitable sample, or a
representative part of a population for the purpose of determining
parameters or characteristics of the whole population

These handout were prepared by Deepak Verma to be used for class discussion in the course 21BMT901 Survey Research Methods offered at DMS, MNIT Jaipur.
The contents are copyrighted under Creative Commons Attribution-Non Commercial 4.0 International. You are free to copy & redistribute material in any
medium or form, or adapt, and build upon the material, provided you use it for non-commercial purposes and give proper attribution to the original author and not
restrict others from using it. You may read the specifics of the license by using this link (http://creativecommons.org/licenses/by-nc/4.0/)

1
 ▪ Is very important for any study

▪ Multiple strategies

Determining ▪ Using census for small populations

Sample size ▪ Using sample size of a similar study

▪ Using published tables

▪ Using formula to calculate sample size

How big?

▪ Less than 1/10 of the population

Determining ▪ Big enough to produce acceptable sampling error

Sample size ▪ Big enough to give reasonable number of subsets

▪ Big enough to run statistical tests required to be run

▪ Small enough to carry out with available resources and time

2
 How big?

Major considerations

Determining ▪ How precise I want to be?

Sample size ▪ How sure do I want to be of the answer?

▪ How much variation is there in the study population?

▪ How small of an effect do you want to be able to identify?

Determining Appropriate Sample Size

How precise I want to be?
SMALLER Fairly Precise Very Precise LARGER
Sample size Sample size

How sure do I want to be of the answer?

SMALLER Fairly Sure Very Sure LARGER
Sample size Sample size

How much variation is there in the study population?

SMALLER Low High LARGER
Sample size variation variation Sample size

How small of an effect do you want to be able to identify?

SMALLER Moderate Very small LARGER
Sample size Sample size

3
 ▪ Sample size

𝒁𝟐 𝒑𝒒
𝒏𝟎 =
𝒆𝟐
Sample Size Z-value (Critical) 99% - 2.578 (2.58); 95% - 1.96; 90% - 1.645
Estimation (1.65)
For population p = proportion which we expect to possess the required
proportion attribute

q = proportion which we expect not to possess the required

attribute

e = precision

Cochran, W. G. (1963) Sampling Techniques, 2nd Ed., New York: John Wiley and Sons, Inc.

In practice we often have no idea about p or q

▪ In such cases, assume p = 50%

Sample Size
Estimation: ▪ This gives the most conservative estimate (largest required) of
the sample size
For population
proportion ▪ Its almost always better to overestimate than underestimate

4
 ▪ Suppose we wish to evaluate consumer adoption of a new
product.

Sample Size ▪ Assume there is a large population but that we do not know the
variability in the proportion that will adopt the product;
Estimation: therefore, assume p=0.5 (maximum variability).
For population
proportion ▪ Furthermore, suppose we desire a 95% confidence level and
±5% precision. The resulting sample size is

𝒁𝟐 𝒑𝒒 𝟏.𝟗𝟔 𝟐 .𝟓 (.𝟓)
𝒏𝟎 = = = 𝟑𝟖𝟓
𝒆𝟐 (.𝟎𝟓)𝟐

Finite population correction

𝒏𝟎
𝒏=
(𝒏𝟎 − 𝟏)
𝟏+
𝑵
Sample Size Where n0 is unadjusted sample size
N is the population size
Estimation:
For population
proportion Suppose our evaluation of consumers’ adoption of the new
product is limited to a market segment of 2,000 B2B customers.
The sample size that would now be
𝒏𝟎 𝟑𝟖𝟓
𝒏= (𝒏𝟎 −𝟏) = (𝟑𝟖𝟓−𝟏) = 𝟑𝟐𝟑
𝟏+ 𝟏+
𝑵 𝟐𝟎𝟎𝟎

5
 Another Method (Yamane’s formula)

𝑵
𝒏=
𝟏 + 𝑵(𝒆)𝟐

Sample Size Where n is the sample size,

N is the population size,
Estimation: e is the level of precision

For population
proportion When this formula is applied to the previous example, we get

𝑵 𝟐𝟎𝟎𝟎
𝒏= 𝟐 = = 𝟑𝟑𝟑
𝟏+𝑵(𝒆) 𝟏+𝟐𝟎𝟎𝟎(.𝟎𝟓)𝟐

Yamane, T. (1967) Statistics: An Introductory Analysis, 2 nd Ed., New York: Harper and Row.

▪ The previous methods were suitable for dichotomous situations

▪ What if we have a situation where variables are polytomous or

Sample Size continuous

Estimation: 𝒁𝟐 𝝈𝟐
For population mean 𝒏=
𝒆𝟐
Where σ2 is the variance of an attribute in the population

6
Sample Size 1. Sampling Technique used
Estimation
▪ The approaches discussed have assumed that a simple random
sample is the sampling design
Some More
▪ More complex designs, e.g., stratified random samples, must
Considerations take into account the variances of subpopulations, strata, or
clusters

2. Statistical Analysis to be done

▪ Sample size should also consider the size of data set needed for
Sample Size the data analysis
Estimation
▪ For descriptive statistics, any size will suffice

Some More ▪ For regression/co-variance analysis, 200-500 will be a good

size
Considerations
▪ For Factor Analysis/SEM, 10-12 times the number of items

The sample size should be appropriate for the analysis that is

planned

7
 3. Need for Comparative Analysis
Sample Size
▪ an adjustment may be needed to accommodate a comparative
Estimation analysis of subgroups

Some More ▪ Sudman (1976) suggests

100 for each major group or subgroup in the sample
Considerations 20 to 50 for each minor subgroup

▪ Skewed distributions can result in serious departures from

normality even for moderate size samples

Sudman, Seymour (1976). Applied Sampling. New York: Academic Press

Sample Size 4. Adjust for no-response

Estimation
▪ Many researchers commonly add 10% to the sample size to
compensate for persons that the researcher is unable to contact.
Some More
Considerations ▪ The sample size also is often increased by 30% to compensate
for nonresponse.

8
 ▪ Selection bias

▪ Pre-screening bias

▪ Self-selection bias
Sample Size
Estimation ▪ Non-response bias

▪ Under-coverage bias

Some Possible ▪ Exclusion bias

Biases ▪ Survivorship bias

▪ Overmatching bias

▪ Berkson’s fallacy

प्रबंधन अध्ययन गवभाि

मालवीय राष्ट्रीय प्रौद्योगिकी संस्थान जयपुर
(शिक्षा मं त्रालय, भारत सरकार के अधीन राष्ट्रीय महत्व का सं स्थान)
जे.एल.एन. मार्ग, जयपुर - 302 017 (राजस्थान, भारत)

Department of Management Studies

Malaviya National Institute of Technology Jaipur
(An Institute of National Importance under Ministry of Education, Government of India)
J. L. N. Marg, Jaipur - 302 017 (Rajasthan, INDIA)

© 2024 Deepak Verma.

This work is licensed under Attribution-Non Commercial 4.0 International