RESEARCH METHODS

Chapter four: Sampling Design

Outline

1. Census and sample survey

2. Importance of sampling
3. Steps in sampling design
4. Types and causes of incorrect inferences in sampling
5. Types of sample design
1. Census and sample survey
Census survey…
- A complete enumeration of all items in the population
- Highest accuracy is obtained
Limitation?
- Expensive
- Beyond the reach of ordinary researchers
- Only government can do it… rarely though!!
eg. Population census
Sample survey…

-What do we do when it is technically impossible to

examine all elements of a population?

-It is usually possible to obtain sufficiently accurate results

by taking part of the population.

- make sure the sample is as representative as possible.

-The process of selecting samples is called sampling

technique.
2.Importance of Sampling

-Research still possible in limited resource

- In case of product testing…. Can we test every item
produced???? 
-Quicker results!!!
-Infinite population?? …. Sampling then!!!!
- quality of the study??? … the bulkier the data harder to
manage it well… so sampling provides better quality
results
3. Steps in Sampling Design
Pay attention to the following points while developing a
sample design
 Type of universe (population)
Sampling Unit
Source list (sampling frame)
Size of sample
Parameters of interest
Budgetary constraint
Sampling procedure
 Finite… certain number of items..
1. Types of universe
Eg. No. of workers in a factory
Infinite…. Uncertain number of
items.. Eg no. of stars

2. Sampling Unit.. Geographical (kebele) or social unit of

interest (gender, family??)

3. Source list.. Also called sampling frame…. Is the list of

all items in a universe…
-Should be comprehensive, correct and reliable
-Should be as representative as possible
4. Size of sample… is the number of items in a sample
-Should be neither too large nor too small
To determine sample size,,,
-size of population variance,

size of population,
-parameter of interest in the research study and
- budgetary constraints should be considered.

Higher variability…… larger sample size

One

Necessary Sample Size = (Z-score)² *

SD*(1-SD) / (margin of error)2 or = (z.SD/e)^2

-Confidence interval (margin of error) -95% Z score is 1.96

-Standard deviation .5 ???do you have it??
-Significance level 5%
-Sample size = (1.96*1.96)* 0.5*(1-0.5))/0.05*0.05
= 3.8416*0.25/0.0025
=0.9604/0.0025
= 384.16 385
Two: Soln.

n= N/1+N(e)2 n= 23680/1+ 23680(0.05*0.05)

= 393.35
Where:
n: sample size
N:population size
e:the level of precision

Eg. Taking 95% confidence level, calculate the sample

size for a population of 23680 people.
5. Parameters of interest… the question of the specific
population parameters which are of interest. What do we want to
discover, understand about the population. (income? Food security??)

6. Budgetary constraint… Cost considerations

7. Sampling procedure… type of sample to be used?

… techniques of selecting sample??

A researcher must select that design which, for a given sample

size and for a given cost, has a smaller sampling error.
4.Types and Causes of Incorrect Inferences in Sampling

Two costs in sample survey… costs of collecting info and cost of

an incorrect inference (conclusion) resulting from the data.

Two causes for an Incorrect inference… sampling error

…systematic bias
1. Sampling errors
-are the random variations in the sample estimates around the
true population parameters.
-Can be reduced by increasing sample size
-Smaller magnitude of sampling error in case of homogeneous
population
-Precision can be increased by increasing sample size … but cost
and systematic errors will increase as sample size increases.
Eg. sample result says 75% of students score 3.8
Population 70% of students score 3.8
The difference is the result of an error (i.e. 5%)
2. Systematic bias (non sampling error)
- results from errors in the sampling procedures
- Bias enters in when a sample fails to represent the population
- it cannot be reduced or eliminated by increasing the sample
size.
- the causes responsible for these errors can be detected and
corrected
a systematic bias is the result of either
 Inappropriate sampling frame:

Defective measuring device:

Non-respondents:
Natural bias in the reporting of data
Eg. Are you a good driver

A.Yes B. No

People usually choose yes … conclusion biased!!!

Instead

How many times have u been stopped by traffic police last

week?/month/year?
Characteristics of a good sample design:

 results in a truly representative sample

results in small sampling error (estimated value closer to
population value)
be viable in the context of funds available for the research study
must enable to control the systematic bias in a better way
Be such that the results of the sample study can be applied, in
general, for the universe with a reasonable level of confidence.
Types of sample design
5. Types of Sample Design
Two types: Probability and Non- probability

1 Probability sampling techniques

- Sample units are selected randomly

-Can be simple random sampling or complex random sampling

2. Non probability sampling techniques

- Sample units are selected by the researcher purposively

Simple random sampling

-Each unit in the universe has equal probability to be

chosen
-Needs an up to date sampling frame
-Representative if population is homogeneous

 lottery method
Advantages Limitations

- Best technique for a - Time consuming

representative sample
Complex random sampling
-Probability sampling under restricted sampling techniques.
-Also called mixed sampling as it has some characteristics
from both probability and non probability methods.
Three types
1.Systematic random sampling
2.Stratified sampling
3.Cluster sampling
1. Systematic random sampling
- Randomly choose the first sampling unit.
- Then choose every ith element.
How to do it??
- Assign a sequence number for all the sample units.
- Determine the skip interval …. How??
I= P/S…. Where I is skip interval, P is population and S is the
sample size.
- Select a starting point in a random digit table (it must be
between 1 and I).
- Then select every Ith unit until sample size is reached.
Advantages Limitations

-Less costly -hidden periodicity in the

-Easier especially when the population?? Then
population is very large. systematic sampling is
-Samples will be distributed inefficient.
evenly over the entire -What if every ith unit has
population similarity?? Defect for
example. Or every ith
student has same GPA?
2. Stratified sampling
- When the population is not homogeneous
- population is divided into several subpopulations (strata)
- Sample selected from each strata
 The basic steps for stratified sampling are:
 Divide the population to be surveyed in to strata of similar
study units or into areas with which similar social,
environmental, or economic conditions exist.
 Make a separate and complete list of the stratum and from
each stratum draw a separate random sample of study units.
 
Oromia South Amhara
Yes No
No
No no
Yes
No No
Yes
No
No
Harari
no
Yes
 proportional allocation of sample size for each stratum … how?
i.e. if Pi is proportion of stratum i and n sample size, then the
number of sampling units to be selected from stratum i will be n*pi
Pi = no of Popn in strata i/ total popn

Eg. Chat consumption of AgEc students by region?

NB. 1. Each stratum, in stratified sampling, is homogeneous
internally and Heterogeneous with other strata.

Advantage Disadvantage

-most convenient way to -not useful when the

gather a representative population cannot be

data…. Reduces exhaustively partitioned into

sampling error disjoint subgroups

Video on stratified sampling
3. Cluster sampling

- When there are natural groupings within the population

-If the total area of interest is big.
-Divide the area into a number of smaller non-overlapping
areas.
-Then randomly select a number of these smaller areas
(clusters).

Advantage Disadvantage
- Cost reduction - Less precise than simple
random sampling
 NB: if clusters are based on some geographic
subdivisions more convenient to call it Area sampling
Eg. Studying internet utilization in Ethiopian
Universities

Video on cluster sampling

Multi-Stage Random Sampling…
-When we use more than one sampling procdure before getting
sampling units.
Eg. Performance of rural primary schools in Ethiopia
-First select a region
- then select a zone
-Select a Woreda and interview all teachers in the Woreda
5.2 Non-probability sampling technique
-Intentional selection of a sample.
-Conclusions drawn works only for the sample …. Generalization
to the entire population is not possible.
-Susceptible to bias.
Non-probability sampling techniques
1.Judgment (purposive) sampling
2.Convenience (accidental) sampling
3.Quota sampling
4.Referral Sampling
1. Judgment (purposive) sampling
- Researcher uses his judgment to select people of certain
characteristics.

2. Convenience (accidental) sampling

- Researcher’s convenience to select sample.
- Could be friends, family, colleagues etc.
- This method is used to save resource.
- Most likely enable to obtain accurate information.
3. Quota sampling
- researcher selects a predetermined number of individuals
from different group (i.e. based on age, gender, etc.
- is useful when a particular group or characteristics is
relatively rare in the population

4. Referral Sampling
- respondents who are initially contacted are asked to
supply the names and addresses of members of the target
population