Professional Documents
Culture Documents
Census
Dr S. G. Kulkarni
Sampling
1
Essentials of Sampling
Dr S. G. Kulkarni
Principles of sampling
Dr S. G. Kulkarni
METHODS OF SAMPLING
RANDOM SAMPLING
RESRTICTED SAMPLING
STRATIFIED
SYSTEMATIC
CLUSTER
MULTI-STAGE
Dr S. G. Kulkarni
SAMPLING
Random sampling also referred to as probability sampling It does not mean haphazard or hit or miss method Every item has equal chance of being included
Dr S. G. Kulkarni
Random Sampling
Methods:
1) Lottery method 2) Tables of random numbers: a) Tippets random tables b) Fishers and Yates tables c) Kendall and Smith d) Rand Corporation 3) Software packages
Dr S. G. Kulkarni 6
Random Sampling
A- Simple or Unrestricted sampling Each and every item has equal and independent chance of being in the sample No personal bias No discretion or preference Replacement of sampling unit
Dr S. G. Kulkarni 7
Each unit is returned to the population, before drawing the next sample Probability of every item is - 1 / N Otherwise, Population is reduced for successive stagesnext item will have probability of 1 / N-1, 1000- 0 =1000 next will be 1/N-2, 1000-1 =999 1000-2 =998 1000-3 =997 1 / N-3 1000-4 =996
1000-5 =995 1000-6 =994
When the slip is returned to the drum before drawing the next slip, the size of population is same
Dr S. G. Kulkarni 8
B) Restricted Sampling
1) Stratified random sampling a) Proportionate stratified sampling b) Disproportionate stratified sampling 2) Systematic sampling 3) Multi-stage sampling 4) Cluster sampling
Dr S. G. Kulkarni 9
N is divided into groups according to homogeneity Adopted when there are heterogeneous features For example, to study the consumption pattern of Belgaum, the city is divided into a number of groups / wards Samples are taken from each ward
Dr S. G. Kulkarni 10
Proportionate stratified random sampling: sample is in proportion to the size of sub-population Disproportionate stratified random sampling:
Dr S. G. Kulkarni
11
Sample is formed by selecting one unit at random and then selecting additional units at evenly spaced intervals. Used when a complete list is available The list is prepared in alphabetical or numerical order and serially numbered
Dr S. G. Kulkarni 12
The first item is selected at random Then the remaining items are selected at k interval k is the sample interval, which is selected as follows: k = Size of Universe = N Size of Sample n
Dr S. G. Kulkarni
13
Systematic sampling-continued
Also known as equal interval sampling or quasi random sampling as the subsequent units are pre-determined Merits: 1) Simple and convenient to adopt 2) More efficient than simple random 3) Time, cost and work- become smaller
Dr S. G. Kulkarni 14
Systematic sampling-continued
Demerits : 1) not truly a random method as there are pre determined intervals 2) Problem of representative ness as there may be a hidden periodicities ( every k th worker may be a well paid)
Dr S. G. Kulkarni
15
Multi-stage sampling
Sampling is carried out at several stages Sample is taken from previous stage sample
Sample of one stage becomes population for next stage
Dr S. G. Kulkarni 16
Multi-stage sampling
Merits :1) Flexibility 2) No bias 3) Cost and efforts are less as N becomes smaller in successive stages Demerits: 1)Less accurate 2) More work 3) Not representative as stratified
Dr S. G. Kulkarni 17
Cluster sampling
Primarily selection of groups rather than individuals Population is divided into groups Groups are mutually exclusive and collectively exhaustive
Dr S. G. Kulkarni
18
Dr S. G. Kulkarni
19
Purposive or Judgment sampling also known as Deliberate (no purpose) sampling It is a conscious selection by an investigator on his own judgment. It requires deep and thorough knowledge and experience. Interviewer requires awareness of the characteristics of the population. Samples vary from investigator to investigator
1)
Dr S. G. Kulkarni 20
Merits : 1) used in solving economic and business problems 2) No missing characteristics 3) Properties of population are known 4) Appropriate for pilot surveys 5) Motivation to the investigator
Dr S. G. Kulkarni 21
Demerits :1) not scientific 2) Personal prejudice (injustice) 3) Difficult to calculate sampling errors 4) Inclination or convenience , but not judgment 5) Comparison of work is difficult
Dr S. G. Kulkarni 22
Chunk is a fraction of population which is selected neither by probability nor by judgment, but by convenience.
Samples are drawn from readily available lists such as automobiles registration, telephone directory, etc
Dr S. G. Kulkarni 23
Convenience sampling
Convenience sample is not random, although samples are drawn at random from the lists.
Good for Pilot studies
Dr S. G. Kulkarni
24
Quota sampling
It is a type of judgment sampling Quotas are set up on characteristics such as income group, age, political affiliation, religious affiliation ,etc The interviewer is asked to interview a certain number of persons in the quota. He is free to chose any person in the given quota.
Dr S. G. Kulkarni 25
For example: In a T V survey, the interviewer is told to interview any 500 persons and out of every 100 following should be the composition: 20 - Students 10 Housewives 25 Office goers 15 Children 20 - Farmers 10 - Businessmen
Dr S. G. Kulkarni 26
Sample is not selected by the investigator, but they themselves propose to be included in the sample
These persons have vital interest and can spare Ex: Questionnaire in newspaper
Dr S. G. Kulkarni 27
Snowball sample
Researcher asks the respondent for names of other individuals who are also to be surveyed The difficulty is close friends or colleagues tend to behave in the same way
Dr S. G. Kulkarni
28
DEFINE POPULATION
Size of sample
No hard and fast rule Depends on subject, time, cost and accuracy Considerations 1) Size of population 2) Accuracy desired 3) Homogeneity / heterogeneity
Dr S. G. Kulkarni 30
Size of sample-continued
4) Nature of study- intensive, continuous, technical, general 5) Practical considerations- time, finance, personnel 6) Type of sampling- stratified, etc 7) Size of questionnaire 8) Question on questionnaire
Dr S. G. Kulkarni 31
Size of sample-continued
Sampling Errors
Sample Size
Dr S. G. Kulkarni
32
Some characteristics of marketing population: a) Population is nether uniform nor concentrated b) Characteristics of people are not simple as there are so many factors c) Data on desired characteristics are nonexistent, or inaccurate d) Identity of specific population is difficult
Dr S. G. Kulkarni 33