Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
19 Cards in this Set
- Front
- Back
Research Designs
|
- Framework that directs research process (who, what, when, where, why)
- In "method" section - Purpose: answer research q's |
|
Good research designs - OFACIE
|
- Objective : Lit review reflects who, what, when, where, why
-Accurate : Answers research question accurately - Feasible : Follow large enough sample size to answer q within time & $$ - Control : Able to control sources of bias to give valid results - Internal validity : results = believable - External validity : results = useful for other people/situations |
|
Validity
|
- Correct / accurate
- ie mass measuring instrument reads 55kg for an object thats 55kg - ie BP cuff is valid to measure BP, thermometer is not valid to measure BP |
|
Reliability
|
- Consistency
- ie mass measuring instrument measures same everytime - ie BP cuff should be properly calibrated to measure BP reliably |
|
Types of validity measures
|
Internal Validity
- Are changes in outcome (DV) DUE to changes in exposure (IV)? - Goal = rule out other explanations External Validity - Can findings be generalized to other pops/settings? - goal = useful beyond participants, replication |
|
Threats to internal validity
|
- Selection bias : If participant's characteristics raise possible alternate explanation
- Instrumentation: Inter rater differences, failure of instrument - History: If event occurs simultaneously with intervention that affects outxome (celebrity death example) - Maturation: external, innate changes to participant unrelated to study (ie disease progression) - Testing : Act of being tested once affects outcome of next test through memory (think alzheimer's pt's) - Mortality/attrition : Loss of participants over time |
|
Threats to External validity
|
Selection Effects:
- Study sample doesn't represent population of interest (ie Dad's at Justin Bieber concert) - Hawthorne/Observer effect : Subjects improve/modify behaviour when they know they're watched Reactivity: - Natural reactions to being studied Measurement Effects: - Actual act of being tested affects outcome |
|
Experimental (RCT)
|
- Participants randomly assigned to experimental or control group, randomized control trial
- Gold standard, control most threats - Strengths: control, participants blinded, establishes causality / causal direction, double blinding possible (not always ethical - Weakness : difficult to implement, generalizability can be low, sometimes not ethical |
|
Experimental --> After only
|
- Used for studies where there's no pre, ie surgery post op pain
|
|
If you delete initial randomization step from experimental design...
|
Quasi experimental design: Non - Equivalent Control
|
|
If you delete control group entirely...
|
Quasi experimental design: One group (pre-test post-test)
|
|
Quasi - Experimental
|
- Participants not randomly assigned, but there is a control group (think smoking, if you smoke or if you don't smoke determines the group you're in, it's not random, but theres still a control group)
- Control some threats - Strengths: Establish causality / causal direction, control - Weakness: Not clear cause + effect, may be difficult to implement, generalizability low, not always ethical - |
|
Quasi - Time Series
|
- Number of pre tests, number of post tests
|
|
Non Experimental
|
- No control group / experimental group comparison
- Control some threats, no manipulation of variable no randomization |
|
Non experimental - cross sectional
|
Strengths: Fast, heap, large # of participants / variables
Weakness: No causal association, threats to validity (many bias) |
|
Non experimental - Prospective (longitudinal cohort)
|
- Strengths: Time line established (exposure before outcome), large # of participants / variables
- Limits: VERY expensive (due to long term follow ups), large attrition due to follow up, threat to validity (selection bias, history - people change exposure over time, instrumentation - tests for cancer change over time, testing - if people aware of outcome hawthorne effect can occur) |
|
Non experimental - Retrospective
|
- Strengths: Fewer participants (know outcome), large # of variables
- Weakness : Very diff to find adequate control group, threats to validity |
|
Descriptive
|
- Control some threats
|
|
Other quantitative studies (not diff designs)
|
- Methodological: Develop new tools or test existing ones in new populations/situations
- Meta-analysis: Statistical, pool data from studies of same design - Secondary data analysis: descriptive, analysis or reanalysis of data for reason other than why it was collected |