DATA AND DATA COLLECTION continued

1
DATA AND DATA COLLECTION- continued

• Lecture 4

2
Sample Selection

• Population
• Sample frame
• Sample size
• Sampling error

3
Sample Selection

• Population
• The total collection of elements about which
  inferences are to be made
• The element is the subject on which measurement
  will be made

4
Sample Selection

• Sample
• Some elements drawn from the population
• E.g. 10 students selected out of the MBA class
• Sample frame
• A listing of all elements in a population
• E.g. List of all MBA Finance students

5
Sample Selection

• Sample size
• The number of the elements selected for use
• Methods of determining size will be addressed
  later
• Sampling Error

6
Sample Selection

• Purpose of sampling
• Population too large to study not feasible to
  study the whole or lower cost
• Information drawn from sample may be better than
  from the whole population (Greater accuracy)
• Better counts and observations because of small
  numbers
• Greater speed of data collection
• Availability of population elements

7
Sample Selection

• The important thing about samples is how
  REPRESENTATIVE it is of the population

8
Principles of Sampling

• Probability
• Non-Probability

9
Probability Sampling

• Simple random sampling
• Systematic Sampling
• Stratified sampling
• Cluster/Area sampling
• Multi-stage sampling
• Multi-phase sampling

10
Probability Sampling

• Simple random sampling
• Write names on paper
• Fold
• Shuffle
• Blindfold
• Pick numbers wanted
• Use of random numbers tables

11
Probability Sampling

• Systematic Sampling
• Decide on sample size
• Express in percent
• Choose the 1st element
• e.g. 4 of population as sample
• Select 1st 25 population member (random sampling)
• then later every 25th member (not random)
• Subsequent once are in fixed intervals

12
Probability Sampling

• Pluses of systematic sampling
• Spread over the entire sample
• Easier and less costly
• Can be used even with large populations
• Minuses
• Inefficient if there is hidden periodicity in the
  population
• Useful when
• Lists are available
• Of considerable length

13
Probability Sampling

• Stratified sampling
• Population is subdivided into to subpopulations
  or strata
• these are expected to be more homogeneous than
  the whole population
• Random samples or systematic sampling is applied
  to each stratum
• Useful for
• Non homogeneous population

14
Probability Sampling

• Stratified sampling
• Strata formation
• Purposive
• Based on past experience
• Researchers judgement
• E.g.
• Regions
• Sectors
• Subsectors

15
Probability Sampling

• Stratified sampling
• Strata size allocations
• Proportionate
• Disproportionate
• E.g. we want sample size n30
• Population size N8000
• Strata is 3 N14000, N22400, N31600
• Total sample size required is 30
• Given that size of sample from each strata is
• n X Pi where i is the strata number
• P1 4000/8000
• n1 x P1 30 x 0.5 15
• Calculate for N22400 and N31600

16
Probability Sampling

• Cluster Sampling
• Divide population into subgroups called clusters
• the subgroups are randomly selected for study
• All elements of the subgroup is studied.
• Where subgroups are geographical areas of refer
  to natural boundaries, the cluster sampling
  becomes Area sampling

17
Comparing Stratified and Cluster

• Stratified
• Divide population into few groups
• Each group has many elements
• Subgroups selected according to variable under
  study
• Homogeneity within subgroups, heterogeneity b/n
  subgroups
• Randomly choose elements from subgroup

• Cluster
• Divide population into many groups
• Each with few elements
• Subgroup selected to ease data collection or
  availability
• Heterogeneity within subgroups, homogeneity b/n
  subgroups
• Randomly choose number of subgroups

18
Probability Sampling

• Cluster Sampling
• Useful when
• More economic efficiency than simple random
  sampling is required
• There is unavailability of practical sampling
  frame

19
Probability Sampling

• Multi-stage sampling
• The whole population is divided into subgroups
• Some subgroups are selected for study
• This is two-stage sampling
• If the selection goes beyond this then it is
  three stage
• Beyond this it is four-stage sampling

20
Probability Sampling

• Multi-stage sampling
• Applied in big enquiries extending to a large
  geographical area e.g, the whole of Ghana
• Advantages
• Easier to administer since sampling frame is
  developed in partial units
• The sequential nature enables large areas to be
  covered in bits

21
Probability Sampling

• Multi-Phase/Sequential sampling
• collect data from a sample
• Use information from this sample to select
  subsequent sample

22
Non-Probability Sampling

• Used for pretest
• Exploratory studies
• Convenience
• Purposive
• judgemental
• Quota
• Snowball

23
Non-Probability Sampling

• Convenient sampling
• Unrestricted non-probability design
• An available sample that which appears able to
  offer answers of interest to a study
• Least reliable
• Normally cheapest
• E.g.

24
Non-Probability Sampling

• Purposive
• Conforms to a certain criterion or the studys
  needs
• Judgemental sampling
• Quota sampling

25
Non-Probability Sampling

• Judgemental sampling
• E.g. Testing a new product, staff are used to try
  the product before the public.
• Quota
• Allocate proportional representation
• E.g. population with 55 male and 45 female.
  Then distribute sample between the sexes to cover
  the sexes.
• Calculate quota for males for sample size of 60.
  How many males should be sampled

26
Non-Probability Sampling

• Snowball
• Identify one respondent who reefers the
  researcher to the next respondent
• It is like a referral network
• Used for hard to reach respondents
• E.g. Prostitutes, Drug pushers, armed robbers
  etc.
• Similar to reverse literature search

27
Methods of Data Collection

• Census
• Survey