Speaking Notes
PADM 5502
September 13, 2007
Dr. Neubauer
WHERE WE ARE:
Chapter 5 -- Sampling
If your research involves making inferences from a sample to a population, you most define the POPULATION, identify your SAMPLING FRAME (if available), and explain your SAMPLING STRATEGY.
POPULATIONS have PARAMETERS, which are usually unknowable because it is not possible to ask everyone in the population to participate.
SAMPLES have STATISTICS.
If the sample has been RANDOMLY SELECTED from the POPULATION then
the resulting STATISTICS should be good estimates of the corresponding PARAMETERS.
This is the meaning of INFERENTIAL STATISTICS.
If the intent is to INFER from the STATISTICS derived from a SAMPLE to estimate the PARAMETERS in a population, you need a SIMPLE RANDOM SAMPLE.
To get a SIMPLE RANDOM SAMPLE you need:
A sampling frame
A way to select from the sampling frame randomly.
The ability to find everyone so selected.
The voluntary cooperation of everyone selected.
You follow up with "call backs" to try to get everyone selected to participate.
There are other approaches to sampling that are not "as good" as a simple random sample but that may be necessary and can be "defended."
PURPOSIVE SAMPLE -- not for inferential statistics. You want to survey or interview people (or other units of analysis) because they are outstanding or special in some way.
PROPORTIONATE STRATIFIED RANDOM SAMPLING or QUOTA sample.
You know the important proportions of kinds of people (or other units of analysis) in the population and you select participants in such a way that the proportions are "right" in your sample. You can do this without a sampling frame and you don't need to do "call backs."
CLUSTER SAMPLE -- You have a multi-stage way of selecting participants. For example, a random sample of states, followed by a random sample of counties within the selected states, and then a random sample of people within the counties. The problem is that some people have a higher probability of being asked to participate than others, depending upon the populations of the states, the different number of counties in each state, and the different populations of counties.
SNOWBALL SAMPLE -- You ask participants to refer you to other possible participants. This may be necessary for certain kinds of studies but there are serious ethical implications.
RANDOM DIGIT DIALING -- You let a computer generate telephone numbers and you call the numbers and ask for an adult. This has many, many problems and limitations.
There are many HYPOTHESES. Each hypothesis anticipates a relationship (or a lack of a relationship) between an INDEPENDENT VARIABLE and the ENTIRE DEPENDENT CONCEPT. Here is an example of a hypothesis.
H5: that persons with higher family incomes are more likely than persons with lower family incomes to have positive attitudes toward hospice care.
ALL THE HYPOTHESES SHOULD END THE SAME WAY (except null hypotheses).
It is usually better to turn the polarity of the statement of the independent variable so the word, “more” appears in the hypothesis.
This is not good, although technically it is not wrong.
H7: that persons with fewer years of formal education are less likely than persons with more years of formal education to have positive attitudes toward hospice care.
THE HYPOTHESIS SHOULD NOT BE A TAUTOLOGY. IN OTHER WORDS, THE INDEPENDENT VARIABLE MUST NOT BE ESSENTIALLY THE SAME AS THE DEPENDENT CONCEPT.
This is not good.
H6: That people who like the idea of hospice care are more likely than people who dislike the idea of hospice care to have positive attitudes toward hospice care.
A tautology basically goes around in a circle and says nothing.
Each hypothesis needs a RATIONALE following it. The rationale should be a short paragraph and should probably begin with, “The rationale for hypothesis.”
H3: that men are more likely than women to have positive attitudes toward hospice care.
The rationale for hypothesis 3 is that women may be more aware of their feelings about death and other things and that their awareness of their feelings may cause them to have reservations about hospice care. A positive attitude toward hospice care implies an acceptance of the inevitability of death. If in fact men are less aware of their feelings they may be less aware of attitudes of fear related to death.
It is okay to cite literature in a rationale if you want to. Use APA style, as always.
------------------
SOME REVIEW (and new) QUESTIONS:
1. A statistic based on a random sample is an estimate of a ________________ of the relevant population.
2. A question that “flat out” asks about what it is you are trying to measure is said to have ________________ validity.
3. A question is _________________ if it in fact measures what it is intended to measure.
4. A question is __________________ if in repeated uses it is found to give consistent results.
5. A sample is ______________________ if everyone on the sampling frame has an EQUAL, KNOWN PROBABILITY of being asked to participate in the survey.
6. The list of all the members of the population is called the ____________________.
7. If you are interested in comparing two groups within a population and one group is a very small percentage of the entire population, why would you not draw a simple random sample from the entire population?
8. If you ask participants to refer you to other prospective participants in your research project, your sample is a ______________________ sample.
9. A sample selected because it is easy to survey is a _________________.
10. A hypothesis is about the possible relationship between a ________________ and a _______________________.
11. Write a hypothesis (in the form taught) regarding a possible relationship between political party affiliation and attitudes regarding preemptive war.
H5: that ___________________ are more likely than ________________ to think that preemptive wars are sometimes necessary.
12. What are the five usual demographic variables usually included in survey research projects?
13. What is a null hypothesis? When would you use one?
14. If you are asking people to offer their own opinions about something, what is the unit of analysis?
15. If you are asking people to provide information about their household, what is the unit of analysis?
16. If you are asking people to provide information about the organization in which they work, what is the unit of analysis?
17. What is it about the nature of scientific knowledge that makes inclusion of a literature review necessary in a research report?
18. Is it “wrong” to replicate a study that someone else has already done?
19. What does it mean that two variables (in a sample) appear have a direct relationship with one another?
20. What does it mean that two variables (in a sample) appear to have an inverse relationship with one another?
21. What does it mean that two variables have a spurious relationship?
22. What is the level of measurement?
a. Age in years: __________________
23. What is the level of measurement?
Religious affiliation: Catholic Jewish Protestant Other
24. What is the level of measurement?
How strongly do you feel about global climate change?
[ ] not concerned
[ ] slightly concerned
[ ] quite concerned
[ ] very concerned
Here is an example
informed consent statement:
RESEARCH SURVEY ON ATTITUDES TOWARD HOSPICE CARE
This is a survey regarding attitudes toward hospice care being conducted by a student at the University of Georgia. You have been selected to be asked to participate in this research project. If you choose to participate your participation will be anonymous. You may choose to skip any question you do not want to answer. Your participation is evidence of your voluntary informed consent. There is no reward for participation and no penalty for not participating. Please do not put you name on these pages. Thank you.
Here is the common
formats of several kinds of questions.
4. How old are you? ____________________ years
9. What is your family's approximate annual income?
[ ] less than $20,000
[ ] $20,000 to $39,999
[ ] $40,000 to $59,999
[ ] $60,000 to $79,000
[ ] $80,000 or more
For each of the following statements, please indicate whether or not you agree with the statement.
26. Hospice care for terminally ill patients should be funded by government agencies.
[ ] [ ] [ ] [ ] [ ]
strongly disagree undecided agree strongly
disagree agree