Transcript SAMPLING - Offices and Directory

SAMPLING PRINCIPLES
Research Methods
University of Massachusetts at Boston
©2011William Holmes
1
WHAT IS A SAMPLE?
• Part of a whole. The larger whole is
a population. The subgroup is the
sample.
• Some selected by scientific
procedures
• Some selected by haphazard
procedures
• Some selected with deliberate bias
2
WHY DO YOU NEED A SAMPLE?
• To make generalizations about
a population.
• Populations are expensive to
get.
• Populations are difficult to
obtain.
• A good sample is better than a
poor population
3
HOW DO YOU GET A GOOD SAMPLE?
• Fit the sampling procedure to the
population, the resources, and the moral
and legal constraints.
• Choose the most scientific procedure
feasible.
• Choose the largest sample possible.
• Choose probability samples over nonprobability.
4
TYPES OF SAMPLES
• Non-probability Sample—
haphazard, convenient
• Probability Sample—
systematic
• Fraudulent Sample—
deliberately biased
5
WHAT ARE PROBABILITY
SAMPLES?
• Follows standard procedure
for everyone in population
• Chance of selection using
procedure is known
• Unintended, random bias is
possible
6
TYPES OF PROBABILITY
SAMPLES
•
•
•
•
Simple Random Sample
Systematic Sample
Cluster Sample
Stratified Sample
7
WHAT ARE NONPROBABILITY
SAMPLES?
• Uses Non-standardized
(Variable) procedures
• Chance of selection is
unknown
• Unintended, systematic bias
may creep in
8
TYPES OF NON-PROBABILITY
SAMPLES
• Convenience Sample—not deliberately
biased
• Purposive Sample—chosen to be similar to
a population, according to the chooser
• Quota Sample—chosen to be similar to a
population, according to known
characteristics
• Snowball Sample—using referrals from
known members of a population
9
FRAUDULENT SAMPLES
• Artificially constructed to show a
characteristic or a relationship
• Violates norms of science and
research
• Selects cases to prove a point
• Concerned with non-scientific
ends—money, promotion, ideology.
10
HOW DO YOU TELL IF YOU’VE GOT A
GOOD SAMPLE?
• Check for scientific
procedures
• Check for ethical and legal
requirements
• Compare with known
population characteristics
• Look for weirdness
11
SELECTING A RANDOM SAMPLE
• 1. Define population
• 2. Get list of random numbers or
choose a random process
• 3. Make a decision rule to select
cases
• 4. Assign random numbers
• 5. Select persons who meet criteria
12
SELECTING A SYSTEMATIC
SAMPLE
• 1. Define population.
• 2. Decide on sample size.
• 3. Divide population into groups where the
number of groups equals the sample size.
• 4. For first group, select one by simple random
sampling.
• 5. Count down on list a number equal to the group
size.
• 6. Select each person at end of count. Repeat.
13
SAMPLING EXAMPLE
Person
Age
Gender
Rdn Nbr* Grp
Random Number Criteria:
select persons with even
random numbers
1
18
1
4#
1
2
25
1
3
1^
3
21
2
7
2
4
34
2
5
2^
5
22
1
1
3
6
19
1
2#
3^
Rdm mean age=20.3
7
33
2
7
4
Rdm mean sex=0.67
8
20
1
7
4^
Syst mean age=24.4
9
21
2
5
5
Syst mean sex=0.60
10
24
2
6#
5^
Systematic Sample start:
person number 2
*from random number table. #Selected for random sample. ^Selected for systematic sample.
14