SAMPLE

Agenda


Sampling

probability sampling

nonprobability sampling
External validity
Sampling
 Drawing
a subgroup from a population
(vs. Census)
Population
Sample

Students registered
in FAMR 380 fall ’00

Everyone in class
today

All registered HI
voters in Sept. 2000


All Adidas shoes
made in 1999
Registered voters
reached by randomdigit dialing on 9/14/00
who answered the
survey
Every 2,000th pair
produced at each plant
What is a good sample?
 A sample
that resembles the population
in characteristics
 (a representative sample)
Representativeness
Sample
Population
Representativeness
Sample
40% Males
60% Females
Population
40% Males
60% Females
Representativeness
Sample
Population
70% Satisfied
70% Satisfied
30% Dissatisfied
30% Dissatisfied
Statistics
Parameter
Why Representative Sample?
 If
characteristics of the sample is similar
to the population, the statistics of
sample are likely to be similar to the
parameters
Let’s Think …
 Research
question: How do UH
students utilize campus facilities?
 Population: UH Students
 Sample size: 200
 How will you sample?
 How can you maximize
representativeness of your sample?
Random Sampling
 Gives
everybody in the population an
equal chance to be selected as a
participant in the sample
 Requires
the list of everybody in the
population
SIMPLE RANDOM SAMPLE
Sample of 4 :
Each person 1/10 chance
Population of 40:
25%
25%
50%
Sample A
Sample B
Sample D
Sample C
Systematic Random Sampling
 Pick
up every ‘n’th subjects
 Sensitive to the way the list is ordered
SYSTEMATIC SAMPLE
Population of 40:
25%
25%
50%
For a sample of 4,
Take every 10th one
Sample B
Sample A
Stratified Random Sampling
 Divide
the population into groups (strata)
 Select subject randomly from the stratum
 Then proportion of groups in the sample is
equal to proportion of groups in population
STRATIFIED RANDOM SAMPLE
Population of 40:
25%
25%
50%
Stratify (layer, category)
by color
Stratified random sample of 4:
Randomly pick from each strata to
maintain 25%, 25%, 50% balance
Cluster Sampling
 Sample
a ready-made group within the
population (cluster) assuming it has a
similar composition to the population
 Example: Third grade classrooms
• Know exact chance of being included
BEFORE participant is picked
• E.g., 1 in 100, .003%, etc.
• Need # in
• # in sample
the population,
• DON’T know each participant’s
chance of being picked
Probability vs. Non-probability
Probability
Sampling






Simple random
Systematic random
Stratified random
Cluster
Each member of the
population has a specifiable
probability of being chosen
Population info available
Non-probability
Sampling






Convenience
Snowball
Purposive
Quota
We don’t know the
probability of a specific
member of the population
being chosen
Population info not available
Representativenss
& Generalizability
 Representativeness
= Resembles
population characteristics
 Generalizability = Able to generalize the
results of your study to the whole
population
 High
representativeness = High
generalizability
 Probability sampling allows higher
representativeness than non-probability
Non-probability Sampling
Convenience Sampling
 Get
available people in the
population
 Low representativeness /
generalizability
Snowball Sampling
 Obtain
participants through a chain
of personal networking-referrals
 Useful to locate the ‘hidden’ or
‘difficult to recruit’ population
 Low representativeness /
generalizability
Quota Sampling
 Predetermine
the proportion of
groups in the sample
 e.g.,
male 50%, female 50%
 e.g., clinical trials-drug research, etc.
Purpose of Quota Sampling

1: To ensure that the sample reflects the
proportion of the group in the population

2: To secure enough numbers of group
members for analysis

If you set quota for purpose 2, your
sample may not reflect the population as a
whole
Purposive Sampling
 Obtain
most informed / most ‘typical’
participants
 “Judgmental sampling”
 High quality of information from each
participanta
 Low representativeness /
generalizability
 Quality of sample depends on
researcher’s ability to identify group
to be studied
Why is sampling important?

Usually want to talk about a POPULATION

Easier to get a sub-set of the population
(SAMPLE)

In a good sample
 Results from a good sample should match the
parent population (REPRESENTATIVE)
 Participants should be chosen without bias
(RANDOM)

This allows you to GENERALIZE the results-what holds for the sample should also hold for
the larger group
External Validity
 Degree
that results can be extended
beyond the limited research setting
 Extent findings can be generalized to
others
 Based
on sample ( rats, college students,
whites, males, lab setting)
External validity ?
 Will
the findings from this study likely be
found
 When
other individuals are studied?
Volunteers / non-volunteers,
 Gender



Under other conditions?
In other settings?
 Psychology
 The
study of college sophomores
 People in general?
 College students - intelligent, high
cognitive skills, young, developing sense
of self-identity, social and political
attitudes in state of flux, need for peer
approval, unstable peer relationships
(Cozby, 2001).
External validity
 Related
to sample and sampling
technique