Transcript Handbook for Health Care Research, Second Edition

Handbook for Health Care Research, Second Edition
Chapter 7
CHAPTER 7
Designing the Experiment
© 2010 Jones and Bartlett Publishers, LLC
Handbook for Health Care Research, Second Edition
Chapter 7
Samples and Populations
• Population: The entire collection of cases as defined
by a set of criteria.
- Accessible population: The collection of cases as
defined by the criteria and that are available to the
investigator.
- Target population: The entire collection of cases to
which research results (observations or conclusions)
are intended to be generalized.
• Sample: A subset of the population.
© 2010 Jones and Bartlett Publishers, LLC
2
Handbook for Health Care Research, Second Edition
Chapter 7
Selecting Samples
• Populations are defined by the researcher and
should be clearly specified.
• Populations need not be large in number.
• The accessible population from which a sample is
drawn should be clearly described by a
researcher.
• Sample size does affect the precision of
estimates, given a certain magnitude of
treatment effect, and formulas exist for
estimating sample size needed to achieve certain
risks of error and precision.
© 2010 Jones and Bartlett Publishers, LLC
3
Handbook for Health Care Research, Second Edition
Chapter 7
Obtaining a Sample
• Non-probability sampling -nonrandom
methods are used to select a sample
• Probability sampling is based on random
sampling- know the probability that each unit
in the defined population will be chosen
© 2010 Jones and Bartlett Publishers, LLC
4
Handbook for Health Care Research, Second Edition
Chapter 7
Random Sampling
• Simple random sampling- every unit in the
population has an equal and independent
chance of being selected
• Stratified sampling - selected by randomly
choosing a specified number from each
stratum
© 2010 Jones and Bartlett Publishers, LLC
5
Handbook for Health Care Research, Second Edition
Chapter 7
Random Sampling
• Systematic sampling-a random sample,
sampling interval, K, is simply N/n
• Cluster sampling – sampling units occur in
intact groups or clusters.
© 2010 Jones and Bartlett Publishers, LLC
6
Handbook for Health Care Research, Second Edition
Chapter 7
Basic Concepts of Research Design
• Research design is the plan or organization for
manipulating, observing, and controlling variables in
a research question.
• Variable: An entity that can take on different values.
• Independent variable: variable that is
manipulated/ the treatment.
• Dependent variable: variable that is
measured/the outcome
• Nuisance variables: Extraneous (usually
uncontrollable) variables, can affect the dependent
variable.
• Placebo: treatment designed to appear exactly
like a comparison© 2010
treatment
Jones and Bartlett Publishers, LLC
7
Handbook for Health Care Research, Second Edition
Chapter 7
Experimental Design Characteristics
• Manipulation of an independent variable
• Control of all other variables except for the dependent
variable
• Observation of the change in the dependent variable
 Control
• Random selection of sample and random assignment to
groups
• Matching of subjects between groups or grouping of
subjects based on a nuisance variable
• Inclusion of a nuisance variable as a treatment variable
• Statistical removal of a nuisance variable through analysis
of covariance
© 2010 Jones and Bartlett Publishers, LLC
8
Handbook for Health Care Research, Second Edition
Chapter 7
Pre-Experimental Designs
• Little or no control of extraneous nuisance
variables-the outcome variable is measured,
the treatment is given, and then the outcome
variable is measured again to see if any
change occurred (pre-test/post-test design)
• Weakness
-poor control of nuisance variables
© 2010 Jones and Bartlett Publishers, LLC
9
Handbook for Health Care Research, Second Edition
Chapter 7
Pre-Experimental Designs
Two Groups, Post- Test Only
© 2010 Jones and Bartlett Publishers, LLC
10
Handbook for Health Care Research, Second Edition
Chapter 7
Quasi-experimental
• Lack full control of all variables-example case
control studies
• Weakness:
-Randomization is lacking
© 2010 Jones and Bartlett Publishers, LLC
11
Handbook for Health Care Research, Second Edition
Chapter 7
Quasi-experimental
Case control
© 2010 Jones and Bartlett Publishers, LLC
12
Handbook for Health Care Research, Second Edition
Chapter 7
True Experimental
• Full control of variables by one or more of the
methods
• More control compared to pre and quasi
experimental design
© 2010 Jones and Bartlett Publishers, LLC
13
Handbook for Health Care Research, Second Edition
Chapter 7
True Experimental Design
True experimental design (randomized control).
© 2010 Jones and Bartlett Publishers, LLC
14
Handbook for Health Care Research, Second Edition
Chapter 7
Analysis of Variance
• Statistical technique that applies to a number
of experimental designs
• Advantage of this design is that more than
two groups can be used
Experimental Factor (Independent Variable)
subject 1
subject 2
...
subject n
subject 1
subject 2
...
subject n
subject 1
subject 2
...
subject n
subject 1
subject 2
...
subject n
© 2010 Jones and Bartlett Publishers, LLC
15
Handbook for Health Care Research, Second Edition
Chapter 7
Validity
• Correctness of conclusions- Is X related to Y?
• Internal validity: The extent to which we are
correct in our conclusion concerning the relation of
independent and dependent variables.
• External validity: The extent to which we are
correct in generalizing the conclusions from sample
results to the population.
• Threats to Internal Validity
•History and maturation: impact of passage of time
•Instrument change: Measurements obtained can be
affected by changes in instruments.
•Mortality: loss of subjects either voluntary or death
© 2010 Jones and Bartlett Publishers, LLC
16
Handbook for Health Care Research, Second Edition
Chapter 7
Validity
• Threats to External Validity
• Population validity: accessible population is not
equivalent to the intended target population
• Hawthorne effect: The awareness of being in a
study can alter a subject’s responses or behavior,
even in medical patients. Psychosomatic effects are
quite real, as evidenced by stress ulcers. similarly,
and this limits the ability to generalize beyond the
study.
• Experimenter effect: Investigators can also bias
results by consciously or unconsciously conveying
expectations
© 2010 Jones and Bartlett Publishers, LLC
Handbook for Health Care Research, Second Edition
Chapter 7
Non-Experimental Study Designs
• Do not directly manipulate or control an
independent variable
• Observation-main activity
Types:
-Retrospective Studies - attempt to reason from a
present effect, or consequence in a population back
to antecedent causes
-Prospective Studies- attempt to reason from a
present antecedent or event in a population to
future consequences or effects
© 2010 Jones and Bartlett Publishers, LLC
18
Handbook for Health Care Research, Second Edition
Chapter 7
Non-Experimental Study Designs
Types:
-Case Studies- provides a description of a single
subject, with a treatment or trait
-Survey- gathers information on a large group of
subjects or units by either written or oral
questionnaires
-Correlational Studies- provide information on the
presence and strength of a relation between two
variables
© 2010 Jones and Bartlett Publishers, LLC
19