Research Design Flashcards
Cross-sectional
Data collected once from one point in time
Longitudinal
Data collected repeatedly over a period of time
Retrospective
Historical data
Prospective
Data collected in real time
Degrees of control
Maximal: experimental design
Moderate: Quasi-experimental design
Minimal: Non experimental design
Experimental design format
Purposeful manipulation of a variable (s); random assignment of subjects to two or more groups
Quasi experimental design format
Purposeful manipulation of a variable (s); no random assignment to groups; may have only one group
Non experimental design format
Observational without manipulation of a variable (s); no random assignment to groups; may have only one group
Descriptive research
- No manipulation by researcher
- Results are “tempered” (possibility of bias and lack of control)
- Info readily available
- Studies easy to do and little time
- Provide clues about possible risk factors of disease/conditions
- Clues about outcomes of tx
- Used to generate hypotheses
Ex. cross sectional studies
Types of descriptive research designs
- Case report, case study (both focus on individual
- Case series, cross sectional studies (both focus on groups)
Case Reports (individual)
- A-B design (pre/post measures)
- Deveop of disease or how patient effect by tx
- Need to have whole progression of patient
- Not considered research and don’t need IRB
- Makes a good starting point for future research. Cannot determine “cause-effect”
Case Series (groups)
- n>1
- Most basic research design
- Just longitudinal observation not “cause-effect”
- Detailed description (series of similar patients, their response to tx, etc)
Cross Sectional Studies (groups)
- Compare 2 factors of interest across groups
- Measure 2 factors of interest at one point in time
- Can’t establish time line
- Associating with correlation
- Can’t be used to determine “cause-effect”
Ex. is there an association between daily meat consumption and colon cancer?
Experimental research
- Manipulation by researcher
- Includes “control” or “comparison” group
- Results “definitive”
- Manipulate indep variable to produce effect on dep. variable
- Usuay tests a specific hypotheses
- Can determine “cause-effect”
- Can determine most effect intervention
- Randomization most common way of manipulation
Ex. RCT
Randomization
- Most common way of manipulation
- Pts or subjects allocated to a tx group at random (ex flip a coin)
- Each pt or subject has equal chance
- Should be completed before study starts
- Minimizes bias
Control vs Placebo/Sham
Control-no intervention condition. Accounts for time/natural history of dx
Placebo/Sham- an “inert” intervention. Accounts for time/natural history. Expectations of being treated (ex. white coat effect). Physiological.
Experimental research designs
- Single subject experiment (individual focus)
- Randomized clinical trial (groups focus)
Single subject experiment design
- A-B-A-B study design
- Controlled administration of intervention
- Controlled withdrawal of intervention
A=baseline
B=Tx administration
A=Treatment withdrawal
B=Tx administration
Random Clinical Trial Experiment Design
- Generate high quality evidence
- Evaluate efficacy of a tx for a specific condition. Determine preventative measure for a specific condition
- Inclusion/Exclusion criteria
- Manipulation occurs
- Gold standard for intervention
- Compare outcomes of groups
Quasi-Experimental Designs
-Experimental design without randomization
Prospective-cohort study
Retrospective-Case control
Cohort Study
-Type of longitudinal study where pts followed over time
Ex. population near radiation and other lives away. Compare cancer effects after 20 years.
Case Control Study
Observational study. Participants already experiences outcome.