Research Design Flashcards
What are Observational Studies?
A type of study in which individuals are observed or certain outcomes are measured. No attempt is made to affe the outcome (for example, no treatment is given)
–In other words, not a “controlled” study where we try to manipulate some variable
What are other names for Observational Studies?
Epidemiology, Prevalence study, descriptive study
How can observational studies be done?
-Can be cross-sectional or longitudinal; qualitative, quantitative or mixed-methods
What are Cross-Sectional Studies?
- Snapshot of sample
- One point in time
- Cost effective
- Can’t capture change
- More common than longitudinal
- Size of study does not matter
- It is the number of assessments or the number of points of data collection that differentiate these!
What are Longitudinal Studies?
- Two or more time periods
- More complex
- More expensive
- More powerful - captures change - aims for nomothetic causality
- Descriptive & Explanatory
What are three types of Longitudinal Studies?
- Trend
- Cohort
- Panel
What are Trend Longitudinal Studies?
Measure changes in a population over time (ex: survey of college freshman each year)
What are Cohort Longitudinal Studies?
Measures changes and follow a particular population over time.
(ex: Survey 25 MS1 students in 2012, then any 25 again in 2013 as MS2 students and so on. Same cohort, but potentially different people sampled).
What are Panel Studies?
Measure changes in the SAME people over time.
(ex: enroll 200 diabetic patients in a study in 2013. Re-interview the same patients in 2014 and again in 2015)
[OFTEN SAME THING AS COHORT IN BOARDS]
You follow a sample of 500 smokers for 10 years to determine the occurrence of COPD. You will interview participants annually. Correct identification of study design would include:
A. Panel study (in biostats books, potentially called a cohort study)
B. Longitudinal Study
C. Observational Study
Incorrect identification of study design would include: (in COPD case)
A. Cross-sectional study
B. Case-control study
C. RCT
D. Experiment
What are 3 positives about Observational Studies?
- Often incorporate probability sampling for generalizability
- Great tool for building a knowledge base and exploring new topics
- Might allow us to answer some questions that controlled suited cannot (for moral and ethical reasons)
What are 3 negatives about Observational Studies?
- Validity of measures across cultures (if using survey)
- Findings are associations or correlations, not causality (causality requires time order, non-spuriousness, and association)
- Often rely on survey self-reports, not actual behavior (but can incorporate other types of data, e.g., chart reviews, multiple reporters)
What are Qualitative studies?
-High validity, allows us to explore depth, nuance.
Examples: Focus Groups, Qualitative personal interviews, non-obstructive measures, visual methodologies, etc.
What are Quantitative studies?
- High reliability, often coupled with generalizable sampling procedures
- -> Quantitative = quantify
What are Mixed Methods?
Using both Qualitative and Quantitative studies
What are four more things on Qualitative Research?
- Entails analysis of less structured data (e.g., open ended responses, social artifacts)
- Excellent tool for exploratory research
- Growing acceptance in medication research (e.g., NIH), esp. if used in mixed-methods studies
- Many software programs to aid analysis
What is RCT?
Randomized Controlled Trials or Clinical Trials
What are Experiments good for?
Mode of scientific observation. Good for hypothesis testing -determine *causation --> Take a group --> Apply an action --> Examine effects of the action
What supports a causal association in RCT?
Reproducibility of the Data
What are the three main parts of a Classical Experiment?
- IV and DV
- Pretesting and post-testing
- Control and Experimental Groups
What are the Independent and Dependent variables?
Independent = Cause/Stimulus Dependent = Effect
What is change attributed to in experiments?
Independent Variable
What is the difference between the experimental group and the control group?
Exp. group = gets the stimulus
Control group = does not receive stimulus (may receive placebo or some other comparative treatment)
–>Compare both groups to examine effect of stimulus
–>Guard against events outside the experiment
In an experiment, what do you measure before and after adding a stimulus?
Dependent variable
What is the main purpose of Phase I Clinical Trials?
Safety and side effects in heathy patients
What is the main purpose of Phase II Clinical Trials?
Ideal dosing in diseased patients
What is the main purpose of Phase III Clinical Trials?
Experimental treatment vs. control
What is the main purpose of Phase IV Clinical Trials?
Continued evaluation of FDA- approved therapy for long term effects
What is the Hawthorne Effect?
Researcher attention affects behavior
What is Double-blind?
Researcher and respondents don’t know control or exp.
What is a One-shot case study?
Apply stimulus, measure DV afterwards
What is One-group pretest-posttest?
Measure DV, apply stimulus, measure DV
What is Static-group comparison?
- Two groups
- Only one gets the stimulus
- No pre-test
What is External Invalidity?
Applicability to the real world is lacking.
What is Internal Invalidity?
Anything other than the stimulus that influences results.
What are sources of Internal Invalidity?
- History
- -> Current Events
- Maturation
- -> People change over time
- Testing
- -> Act of testing will influence behavior
- Instrumentation
- -> Any change in instrumentation can influence results
What can invalidate Statistical Regression?
Extremes can bias stimulus results
What can invalidate Selection bias?
Groups need to be comparable
What can invalidate Experimental mortality?
Subjects drop out of experiment
What can invalidate causal time order?
Cause must come before the effect
What can diffusion of treatments do to a study?
Contamination of the control group
What can compensation do to a study?
Incentive or stimulus may bias groups
What can compensatory rivalry do to a study?
Control group may try to make up for lack of stimulus
What can demoralization do to a study?
Control group may become discouraged by the lack of stimulus
What is the weakness of experimental designs?
Artificiality
What are the strengths of experimental designs?
- Isolating IV (causal influence)
- Easy to replicate
- Scientifically rigorous
What is another name for Experimental?
Clinical trail
What is another name for Cross-sectional?
Prevalence
What is another name for Retrospective?
Case-control
What is another name for Prospective?
Cohort Study (a type of longitudinal observational design)
What is controlled (experimental) research?
In some texts and writings, compared against all other types of “non-controlled” studies
What are Observational Studies?
- Case-Control (A type of observational study)
- Also case-history study
- Often compares diseases or events or exposures that have happened already. “Cases” are compared to non-cases, or “controls” in relation to certain risk and protective factors.
- The term retrospective often accompanies this case-control design in biostats writings, although any report or point of data from the past can be viewed as retrospective
What is an example of observational studies?
A group of recently pregnant women who had miscarriages is compared to a group of women who went to full term. The groups are interviewed about caffeine use during their first trimester to look for a correlation between caffeine and miscarriage (Note potential for recall bias)
