Like this presentation? Why not share!

Population & Sampling Procedure * Dr. A. Asgari

by Dr. Azadeh Asgari, Researcher/Lecturer on Aug 31, 2010

• 3,314 views

Views

Total Views
3,314
Views on SlideShare
3,314
Embed Views
0

Likes
5
0
0

No embeds

Categories

Uploaded via SlideShare as Microsoft PowerPoint

Population & Sampling Procedure * Dr. A. AsgariPresentation Transcript

• POPULATION & SAMPLE
Research Methodology
• Population & Sample
• POPULATION:
• all individuals in a group that has similar characteristics (one or more) to be studied by the researcher.
• e.g.: all counselors; all male teachers teaching in secondary schools; all UPM students
• Population & Sample
• SAMPLE:
• Part of a chosen population to be observed and analyzed.
• By observing the randomized samples’ characteristics, several inferences on the population may be made.
• Differences between sample, subjects, respondents.
• Parameter & Statistics
• Parameter:
• values obtained from a population.
• Statistics:
• values obtained from a sample
• Randomization
• Basic to scientific observations and research
• Assumption – even if we cannot precisely predict specific events (e.g.: Individual’s achievement), but we can precisely predict the average/mean achievement of the group
• Types of Sampling
• Probability Sampling
• Non-probability Sampling
• Types of Probability Sampling
• Simple random sampling / selection
• Systematic sampling
• Stratified sampling
• Cluster sampling
• Randomization of Sample
• BASIC TO RANDOMISATION = simple randomization = every individual in the group has equal opportunity (equal chance) to be chosen i.e. not biased
• Choosing one subject is independent of the others .
• Researcher can assume that the characteristics of the sample approximate the characteristics of total population
• Sampling Frame
• Assigning a number to all individuals in a population.
• Using the sampling frame, the sample is chosen / drawn.
• Simple Random Sampling (selection)
• Using:
• Fish Bowl Technique
• Table of Random Numbers
• Computer Generated Numbers
• Table of Random Numbers
• 1 2 3 4 5 6 7 8 9 10
• ______________________________________________________________
• 1 10480 15011 01536 02011 81647 91646 69179 14194 62590 36207
• 2 22368 46573 25595 85393 30995 89198 27982 53402 93965 34095
• 3 24130 48360 22527 97265 76393 64809 15179 24830 49340 32081
• 4 42167 93093 06243 61680 07856 16376 39440 53537 71341 57004
• 5 37570 39975 81837 16656 06121 91782 60468 81305 49684 60672
• 6 77921 06907 11008 42751 27756 53498 18602 70659 90655 15053
• 7 99562 72905 56420 69994 98872 31016 71194 18738 44013 48840
• 8 96301 91977 05463 07972 18876 20922 94595 56869 69014 60045
• 9 89579 14342 63661 10281 17453 18103 57740 84378 25331 12566
• 10 85475 36857 53342 53988 53060 59533 38867 62300 01858 17893
• Systematic Sampling
• Steps:
• Calculate the Interval
• Draw the Initial Number
• Select the Other Sample
• Systematic Sampling
• In this technique, randomization is done only on the initial number.
• Drawing the initial number, fixed the other individuals in the sampling frame.
• Weakness of Systematic Sampling
• There are numbers which do not have equal opportunity to be chosen – thus a slight biasness.
• Choice of a subject depends on another.
• Stratified Sampling
• To reduce sampling error and to increase precision without increasing sample size.
• To ensure all strata are represented (not different from the population)
• In a stratum the population is more homogenous
• e.g.: socio economic status, gender, level of intelligence, level of anxiety
• If variance is reduced and therefore, sampling error will be reduced
• Stratified Sampling
• Steps:
• Determine the ratio between the strata
• Ensure the sample size
• Divide the number of sample according to the initial ratio within the population
• Select the sample using randomisation technique
• Cluster Sampling
• Sampling is according to clusters and not individuals within each cluster
• Conducted if individuals to be sampled are not known
• This technique maintained the principles of randomisation
• Cluster Sampling
• Need not know individuals within each cluster.
• If the clusters within the population are far apart .
• Very suitable and more precise if many small clusters are chosen, therefore similar to the population.
• Not suitable if a large cluster is chosen since it may not represent the population.
• Sampling error is even larger if a big and homogeneous cluster is selected.
• Types of Non-Probability Sampling
• Sample of Convenience or Accidental Sampling
• Weak sampling procedure
• Using available cases for the research
• e.g.: Interviewing the first individual you meet; using you class students; interviewing volunteers
• Types of Non-Probability Sampling
• Purposive Sampling - Judgment Sampling
• Sampling element is decided to represent the population.
• e.g.: Interviewing all possible voters in a district, and using the result to predict the voting pattern for the whole state
• Sampling Error
• Randomized sample may not represent population.
• Variations my occur, called SAMPLING ERROR .
• This variation is not an error caused by the researcher, but it occurs as a result of the sampling process.
• Selection of Biased Sample
• From a telephone directory
• From a list of magazine subscribers
• From a list of registered vehicles
• Sampling Error ( e )
• Often occurs if the mean sample is used to estimate mean population.
• Refers to the difference between population parameter and the sample statistics.
• _
• E = x - µ
• Sample Size
• Large enough so that it is representative of the population.
• Crucial issue is representativeness & not the sample size
• e.g.: Sample of 200 which has been randomly selected is better than a randomly selected sample of 100; but a randomly selected sample of 100 is better than a biased sample of 2.5 million individuals.
• Aspects in Determining Sample Size
• ECONOMY – researcher’s financial situation
• MANAGEABLE SAMPEL SIZE by researcher – during data collection
• VALIDITY – a large enough size needed for high validity
• RELIABILITY - a large enough size needed for high reliability
• UTILIZATION OF INFERENTIAL STATISTICS – depends of the type of inferential statistics to be used
• Descriptive – large
• Inferential – correlation, minimum 30
• Inferential – comparing two groups, 30 for each group
• Inferential – comparing more two groups, 30 for each group
• Experimental – small
• Hypothesis Testing
• Testing null hypothesis using different tests based on type of measurement scales and data.
• Make decision on the null hypothesis.
• Make decision on the alternative hypothesis.
• Type I & II Error Scheme H O TRUE H O FALSE REJECT H O ACCEPT H O TYPE I ERROR CORRECT ACTION CORRECT ACTION TYPE II ERROR
• Type I & II Error
• Type I Error
• Rejecting a true null hypothesis
• e.g. Rejecting
• h o = there exist no relationship between both variables – which is true
• Type II Error
• Accepting a false null hypothesis
• e.g. Acceptin g
• h o = there exist no relationship between both variables – which is false
• Level of Significance
• Researcher needs to weigh the consequences of type I and ii errors before conducting the research (how strong the evidence must be before they would reject h o ).
• Level at which h o may be rejected = level of significance
• Level of Significance
• Researcher may avoid type I error by accepting h o all the time.
• Or avoid type II error by rejecting it all the time.
• Reducing the value of level of significance (from .05 to .01 or .001) reduces the risk of doing a type I error but increases the risk of doing a type II error.