Sampling

1
Sampling
2
Fair sampling produces a sample which represents
the population in all important ways.
3
Fair sampling produces a sample which represents
the population in all important ways.
4
Constructing a Sample
5
Constructing a Sample
1. Define the general universe.
6
Constructing a Sample
1. Define the general universe. 2. Identify an
observable working universe.
7
Constructing a Sample
1. Define the general universe. 2. Identify an
observable working universe. 3. Choose the
sampling unit.
8
Constructing a Sample
1. Define the general universe. 2. Identify an
observable working universe. 3. Choose the
sampling unit. 4. Develop or find a sampling
frame.
9
Constructing a Sample

• Define the general universe.

10
Constructing a Sample

• Define the general universe.
• U.S. school teachers

11
Constructing a Sample

• Define the general universe.
• U.S. school teachers
• 2. Identify an observable working universe.

12
Constructing a Sample

• Define the general universe.
• U.S. school teachers
• 2. Identify an observable working universe.
• teachers who belong to national teacher
  associations

13
Constructing a Sample

• Define the general universe.
• U.S. school teachers
• 2. Identify an observable working universe.
• teachers who belong to national teacher
  associations
• 3. Choose the sampling unit.

14
Constructing a Sample

• Define the general universe.
• U.S. school teachers
• 2. Identify an observable working universe.
• teachers who belong to national teacher
  associations
• 3. Choose the sampling unit.
• a single teacher

15
Constructing a Sample

• Define the general universe.
• U.S. school teachers
• 2. Identify an observable working universe.
• teachers who belong to national teacher
  associations
• 3. Choose the sampling unit.
• a single teacher
• 4. Develop or find a sampling frame.

16
Constructing a Sample

• Define the general universe.
• U.S. school teachers
• 2. Identify an observable working universe.
• teachers who belong to national teacher
  associations
• 3. Choose the sampling unit.
• a single teacher
• 4. Develop or find a sampling frame.
• lists of email addresses of members purchased
  from the National Education Association and the
  American Federation of Teachers

17
Survey Jargon
18
Survey Jargon
From the sampling frame, you choose who to
solicit for participation. You ask these people
to participate in your study or reply to your
survey. Of those you contact and solicit for
participation, some will participate and some
will not. Those who choose to participate are
respondents. Of all those who were solicited, the
percentage who became respondents is your
response rate.
19
Survey Jargon
From the sampling frame, you choose who to
solicit for participation. You ask these people
to participate in your study or reply to your
survey. Of those you contact and solicit for
participation, some will participate and some
will not. Those who choose to participate are
respondents. Of all those who were solicited, the
percentage who became respondents is your
response rate. Number responding to
survey Number solicited for participation
Response Rate
20
Sampling Strategies
21
Sampling Strategies
random systematic stratified random cluster
judgment convenience
22
Sampling Strategies
random Number all email addresses and randomly
produce numbers systematic stratified
random cluster judgment convenience
23
Sampling Strategies
random Number all email addresses and randomly
produce numbers systematic Pick every 50th email
address. stratified random cluster
judgment convenience
24
Sampling Strategies
random Number all email addresses and randomly
produce numbers systematic Pick every 50th email
address. stratified random Group teachers by
teaching level- elementary secondary.Randomly
select from each group. cluster
judgment convenience
25
Sampling Strategies
random Number all email addresses and randomly
produce numbers systematic Pick every 50th email
address. stratified random Group teachers by
teaching level- elementary secondary.Randomly
select from each group. cluster Start with a
working universe of all schools.Randomly select
schools and survey all teachers in that
school. judgment convenience
26
Sampling Strategies
random Number all email addresses and randomly
produce numbers systematic Pick every 50th email
address. stratified random Group teachers by
teaching level- elementary secondary.Randomly
select from each group. cluster Start with a
working universe of all schools.Randomly select
schools and survey all teachers in that
school. judgment Recruit the teachers who appear
to be in touch with todays issues. convenience
27
Sampling Strategies
random Number all email addresses and randomly
produce numbers systematic Pick every 50th email
address. stratified random Group teachers by
teaching level- elementary secondary.Randomly
select from each group. cluster Start with a
working universe of all schools.Randomly select
schools and survey all teachers in that
school. judgment Recruit the teachers who appear
to be in touch with todays issues. convenience Re
cruit the teachers in your school.
28
Sample Size
29
Sample Size
If the research goal is to describe a population,
the primary concern when determining a sample
size is describing the population with precision.
30
Sample Size
If the research goal is to describe a population,
the primary concern when determining a sample
size is describing the population with
precision. The primary concern when determining a
sample size is sampling error.
31
Sample Size
The larger the sample, the closer the sample
values are to the true population values.
32
Sample Size
The larger the sample, the closer the sample
values are to the true population values. The
larger the sample, the smaller the sampling error.
33
Sample Size
The larger the sample, the closer the sample
values are to the true population values. The
larger the sample, the smaller the sampling
error. A common formula for calculating sampling
error for surveys is the standard error of
proportion.
34
Standard Error of Proportion
35
Standard Error of Proportion
36
Standard Error of Proportion
37
Standard Error of Proportion
38
Standard Error of Proportion
Because standard errors of proportion are
normally distributed, we can create 95
confidence intervals by using /- 1.96 standard
errors.
39
Standard Error of Proportion
Because standard errors of proportion are
normally distributed, we can create 95
confidence intervals by using /- 1.96 standard
errors.
Out of all sailors, there is a 95 chance that
somewhere between 63 and 81 have knee trouble.
40
Standard Error of Proportion
Because standard errors of proportion are
normally distributed, we can create 95
confidence intervals by using /- 1.96 standard
errors.
Out of all sailors, there is a 95 chance that
somewhere between 63 and 81 have knee trouble.
I am 95 confident, that if we had surveyed all
voters, somewhere between 48 and 54 would say
they plan to vote for Frey. Margin of Error
/- 2.5
41
Sampling Problem
42
Sampling Problem
A researcher is interested in transplant
surgeons' attitudes toward a policy initiative
that would provide organs to patients most in
need of transplants rather than providing organs
to patients on the waiting list of hospitals who
harvest the organs. The researcher decides to
conduct a telephone interview with a random
sample of 250 board certified heart transplant
surgeons. There are 1000 board certified heart
transplant surgeons. Of the 250 heart surgeons
included in the sample, 240 were contacted and
220 of them agreed to the telephone interview.
20 of the heart transplant surgeons agree with
the statement Heart transplants should be
provided to patients in greatest need.
43
Sampling Problem
A researcher is interested in transplant
surgeons' attitudes toward a policy initiative
that would provide organs to patients most in
need of transplants rather than providing organs
to patients on the waiting list of hospitals who
harvest the organs. The researcher decides to
conduct a telephone interview with a random
sample of 250 board certified heart transplant
surgeons. There are 1000 board certified heart
transplant surgeons. Of the 250 heart surgeons
included in the sample, 240 were contacted and
220 of them agreed to the telephone interview.
20 of the heart transplant surgeons agree with
the statement Heart transplants should be
provided to patients in greatest need.
What is the sampling frame?
44
Sampling Problem
A researcher is interested in transplant
surgeons' attitudes toward a policy initiative
that would provide organs to patients most in
need of transplants rather than providing organs
to patients on the waiting list of hospitals who
harvest the organs. The researcher decides to
conduct a telephone interview with a random
sample of 250 board certified heart transplant
surgeons. There are 1000 board certified heart
transplant surgeons. Of the 250 heart surgeons
included in the sample, 240 were contacted and
220 of them agreed to the telephone interview.
20 of the heart transplant surgeons agree with
the statement Heart transplants should be
provided to patients in greatest need.
What is the response rate?
45
Sampling Problem
A researcher is interested in transplant
surgeons' attitudes toward a policy initiative
that would provide organs to patients most in
need of transplants rather than providing organs
to patients on the waiting list of hospitals who
harvest the organs. The researcher decides to
conduct a telephone interview with a random
sample of 250 board certified heart transplant
surgeons. There are 1000 board certified heart
transplant surgeons. Of the 250 heart surgeons
included in the sample, 240 were contacted and
220 of them agreed to the telephone interview.
20 of the heart transplant surgeons agree with
the statement Heart transplants should be
provided to patients in greatest need.
What is the standard error of proportion?
46
Sampling Problem
47
Sampling Problem
A school district wishes to conduct a survey to
assess alcohol use by its students. A random
sample of all students (equal numbers of boys and
girls) will be solicited for participation and
asked if they drink beer on the week-ends. The
evaluators consult with you to determine an
adequate total sample size. Assume you wish to
have a margin of error of /- 5. What sample
size would you suggest? Use the default
assumption of 50 (Yess or Nos) for your
estimates.
48
Sampling Problem
Assume you wish to have a margin of error of /-
5. What sample size would you suggest? Use the
default assumption of 50 (Yess or Nos) for
your estimates.
Margin of error .05 .05/1.96 .0255, so I
want a Standard Error of around .0255
49
Sampling Problem
Assume you wish to have a margin of error of /-
5. What sample size would you suggest? Use the
default assumption of 50 (Yess or Nos) for
your estimates.
Margin of error .05 .05/1.96 .0255, so I
want a Standard Error of around .0255
50
Sampling Problem
Assume you wish to have a margin of error of /-
5. What sample size would you suggest? Use the
default assumption of 50 (Yess or Nos) for
your estimates.
Margin of error .05 .05/1.96 .0255, so I
want a Standard Error of around .0255
51
Sampling Problem
52
Sampling Problem