Sampling

1
Sampling

• ???????????????

2
AC Nielsen announces the rating of Thai TVs as
follows

• Channel 7 45 (of Thai population)
• Channel 3 30
• Channel 5 10
• Channel 9 7
• ITV 6
• Channel 11 2
• How does AC Nielsen obtain such information?

3
Population in statistics and everyday term

• (Statistical) Population
• A collection of objects (target group or
  documents) that have something in common that
  others do not. ??????? ????????????? ????
  ????????????????
• Characteristics
• Fit specific need at a specific time
  ??????????????? ?????????????
• Define precisely ??????????????????????????
• No need to be human beings ???????????????????????

4
Why sampling?

• Cannot collect data from everyone
  ?????????????????????????????????????????????
• Assume that samples represent the populations
  ??????????????????????????????????????????????????
  ?????????????????????????

5
How does it work?

• Choose sample of several hundred people to
  estimate the population of several million.
  ??????????????????????????????????????
• The researcher extends the knowledge from a
  sample that she has measured to the population
  she has not. ?????????????????????????????????????
  ??????????

6
(No Transcript)
7
Generalization

• A reliable measure generalizes across similar
  situations. ?????????????????????????
  ??????????????????????????????????????????????????
  
• The Sun has risen every morning since recorded
  history. It will rise tomorrow as well.
  (????????????????????????????????????
  ?????????????????????????????????????)

8
How to choose samples?

• Define a population - Target group
  ??????????????????????????
• Select a sample from the population
  ???????????????????????
• Measure the variable of interest
  ????????????????????
• Summarize the results ??????
• Estimate the value in the population
  ??????????????????????????????????????????????????
  ???? ?????????????????????????????????????????????

9
Define a population - Target group
100,000Parameters
10
Select a sample from the population
Choose samples (Descriptive statistics)
100,000Parameters
384 Statistics
11
Measure the variables of interest

• Stats

384
12
Summarize the results
384 Statistics
13
Estimate the value in the population
To generalize (Inferential statistics)
100,000Parameters
384 Statistics
14
Where do samples come from?

• Name list
• Telephone directory
• Customer lists
• Professional associations
• Streets

15
How many people? (sampling size)

• Mathematics formula
• p anticipated measure
• se sampling error

16
Taro Yamane Table5 sampling error at 95
confidence level

• Population Sample
• 550 228
• 1100 285
• 1700 313
• 2400 331
• 4000 351
• 8000 367
• 20000 377
• 100000 384

• Population Sample
• 10 10
• 30 28
• 60 52
• 80 66
• 110 86
• 170 118
• 210 136
• 320 175

17
Sampling error ???????????????????????????????

• Degree to which a sample differs from population
  characteristics on some measure.
  ??????????????????????????????????????????????????
  ????????????????
• As the sample size gets larger and larger
  proportion of the population, the amount of
  sampling error is reduced. ltIf the whole
  population is investigated, theres no sampling
  error!gt ?????????????????????????????? ???
  Sampling error ???????? ??????????????????????????
  ???????? ??????????????? Sampling error

18
How to choose the samples (Types of sampling)

• By probability (Random) ?????????????????
• Drawing the Lot (Simple random sampling)
• nth subject(Systematic sampling)
• Population characteristics (Stratified random
  sampling)
• Population grouping (Cluster sampling)

• Not by probability (Nonrandom) ???????????????????
  ??????
• AnyoneAccidental - convenience sampling
• Quota basedQuota sampling
• Specific group of populationPurposive / Known
  group sampling
• Referred key persons Snowball sampling

19
Simple random sampling

• Drawing the Lot ???????
• Research problems do not require the samples be
  divided into subgroups ??????????????????????????
  ??????????????????????

20
Systematic sampling

• Chooses every nth subject (sampling rate) from a
  complete list of a population.
• Choosing subjects for content analysis

M
A
M
J
J
A
S
O
N
D
J
F
21
Stratified random sampling

• Population characteristics ???????????????????????
  ??????????????????????
• Men vs. women (49 vs. 51)
• Decide the sample size (200)
• 98 Men vs. 102 Women

22
Cluster sampling

• Population groupings according to categories/area
  ?????????????????????????????
• Newspapers
• English language The Nation Bangkok Post
• Tabloid ?????? ???????? ????????? ??????
• Business ?????????????
• Middle class ????? ????????
• Thais
• Chiang Mai (CNX)
• Bangkok (BKK)
• Phuket (HKT)

23
Accidental - convenience sampling

• Whoever ????? ?????????????
• Nonscientific surveys of opinions
  ???????????????????????????????????????
• Radio call in polls ?????????????????
• Surveys at department stores ????????????????????

24
Quota Sampling

• Population characteristics ?????????????
• Similar to stratified sampling ????????
  Stratified sampling
• But samples chosen from accidental sampling
  instead. ????????????????????????????????
• Analyzing advertisements impacts in 5 years
  ?????????? 12 ?????? 5 ?? ????????????????????????
  ??????????

Ad in 1996 Ad in 1997 Ad in 1998 Ad in
1999 Ad in 2000
M
A
M
J
J
A
S
O
N
D
J
F
M
A
M
J
J
A
S
O
N
D
J
F
M
A
M
J
J
A
S
O
N
D
J
F
M
A
M
J
J
A
S
O
N
D
J
F
M
A
M
J
J
A
S
O
N
D
J
F
25
Purposive / Known group sampling

• Specific group of population ??????????????
• Male PR Managers who earn an Engineering degree.
• Female CEOs
• MSM

26
Snowball sampling

• Referred key informants
• Illegal or secret issues
• Crime network
• Human trade network
• Drug uses

27
Advantages and disadvantages nonrandom sampling

• Strength
• Field and quasi experimental research
  ?????????????????????????????????????
• Allow researchers to collect data from samples
  that are hard to reach ???????????????????????????
  ????????????

• Limitations
• Bias ???????????????????????????
• Not so accurate data ?????????????????????
• Limited conclusions that researchers can draw.
  ??????????????????????????????????????????????????
  ????????????????

28
Which sampling method is best for the following
research?

• Peoples opinion on Thai government
• Foreign businessmens satisfaction of Thai
  Airways business class service
• Casino management in Thailand