Research Designs Flashcards
When an elderly sample is selected from a local fitness center
“Healthy-user” bias
When the population is selected from an impaired or diseased group
Berkson’s Bias
When patients with certain characteristics are excluded
-Ex: racial groups, socioeconomic groups, etc
Exclusion Bias
Bias decreases
Generalizability
What are the two major classes of research designs?
- ) Experimental
2. ) Non-experimental/observational
Is characterized by random assignment, manipulation of independent variables, and control of extraneous factors
Experimental Research
Experimental research is the best form of research for
Cause-and-effect
Experimental research is the standard for
Clinical trials
Characterized by no control over group assignment and no control over independent variables
Non-experimental/observational research
Non-experimental/observational research can not be used to determine
Causality
What is the gold standard experimental design for treatment studies?
Randomized control trial (RCT)
Equivalent groups are needed to interpret effects. We randomize because randomization removes
Selection bias
In the physicians aspirin study, randomization was accomplished through
Computer generated random numbers
When the investigator “knows” the expected results and treats groups differently
Investigator bias
One way to control investigator bias is when the people randomizing individuals into groups are blinded as to which subjects go into which group. This is called
Allocation Concealment
Another way to prevent investigator bias is to blind the investigator who is providing treatments or making measurements as to which group is which. This is called
Investigator blinding
Investigator bias can me minimized by
Allocation concealment and investigator blinding
When subjects change their behavior in a study, for example, they may want to please the investigators so they act differently. This is called
Subject bias (Hawthorne effect)
The Hawthorne effect (subject bias) effects
External and internal validity
How is the Hawthorne effect combated?
Subject “blinding” by use of a placebo group
Using a placebo group does not really work if the study involves an obvious
Physical change
When patients and clinicians are both blinded
Double Blinding
Ambiguous term, generally used to mean that the subjects are blinded
Single-blinding
Besides just controlling the allocation and outcome of a study, a researcher also needs to control
Extraneous variables (i.e. diet, non-compliance, drop out, etc.)
To help ensure compliance in the physicians study, subjects filled out a compliance questionnaire that also asked about the outcomes every
Six months
What are two ways to deal with drop-outs or treatment changes?
- ) Analysis according to treatment assigned (intention to treat)
- ) Analysis according to treatment received (explanatory) (better way)
The physicians study measured the number of new cases of disease arising during a given period of time, meaning they measured the
Incidence (absolute risk)
How do we find the Incidence (absolute risk)
Incidence = # people w/ disease / # total people
How do we calculate the relative risk or risk ratio?
incidence in exposed / incidence in unexposed
Both estimate the risk of disease (or outcome) in exposed (treated) vs. unexposed (untreated)
Risk ratios and Odds ratios
What is the risk reduction?
1 - RR
The physicians study showed a reduction in the risk of MI by
47%
However, the physicians study also showed an INCREASE in the risk for
Strokes
What are two common types of observational studies?
Case-control and cohort studies
What makes case-control and cohort studies observational?
- ) Can not manipulate independent variables
2. ) Can not control all extraneous factors
A group of people who have something in common when they are first assembled and who are then observed for a period of time to see what happens to them
Cohort
The goal of this type of study is to study “predictor variables”
Cohort (incidence) Studies
One of the problems with cohort studies may be
Internal validity
What is a major threat to the internal validity of a cohort study?
Extraneous factors
Differences between groups, i.e. another factor (other than the one of interest) that may be related to the differences between groups
Extraneous factors
The best solution to the threat that extraneous factors pose on internal validity is
randomization
To help minimize the extraneous factors, we could increase restriction, increasing restriction decreases
Generalizability (external validity)
What are the two different types of cohort studies?
Prospective and Retrospective
What is the usual statistic when comparing two groups?
Risk Ratio (RR = Incidence in exposed / Incidence in unexposed)
Like odds, Risk Ratios (RRs) are given with 95% confidence intervals. If the confidence interval includes 1, there is
No statistically significant difference
Looking backwards to compare people with and without a condition
Case-control study
Case-control studies are good for
Diseases w/ long latency and rare diseases
The big difference between case control and cohort studies is that in case-control studies, we don;t measure the
Incidence (it can only be estimated)
Measure how many people develop disease out of a total, i.e. what is the relative incidence of disease in both groups
Cohort studies
Look at people who already have the disease and determine the odds of exposure. I.e. what is the odds that the diseased group was exposed?
Case-control studies
What are three threats to the internal validity in case-control studies?
- ) Control group selection
- ) Recall bias
- ) Can’t determine risk directly
People may not remember their exposure of the event of interest. This is called
Recall bias
Since case-control studies can not measure the incidence, they instead use the
Odds ratios
