Hoofdstuk 9: Quantitative research deel 2 Flashcards
What are quasi experimental design
Controlled trials with no randomisation and/or control group
Describe the non equivalent design (2)
1.Pretest- posttest design
O1 –> X –> O2
O1 ———-> O2
2.Posttest only design
X –> O1
…—>O1
=> Experimental and comparison groups are not assumed to be initially equivalent
Describe matching in non equivalent deisigns (3)
- Conventional matching
- Key variables related to outcome - Propensity matching
- Propensityscore (software gives a score)
3.Historical comparison group
Describe the time series design (3 + variations (2))
1.One-group pretest-posttest design
O1 –> X –> O2
2.Time-series design
O1 –> O2 –> O3 –> X –> O4 –> O5–> O6
=> strengthens the ability to attribute change to the intervention
3.Time series non equivalent control group design
O1 –> O2 –> O3 –> X –> O4 –> O5 –> O6
O1 –> O2 –> O3 ———–> O4 –> O5 –> O6
4.Variations
-withdrawing and reinstuting treatment
O1 –> O2 –> X –> O3 –> O4 –> -X –> O5 –> O6 –> X –> O7 –> O8
-Single subject experiments:
AB / ABA / ABAB (A=data gathering, B=Intervention)
Describe partially randomized patient preference and quasi experimental dose respone analysis
Partially randomized patient preference
=group with x preference = randomized, group with preference is given this preference
–> information about kind of people who prefer one condition over another
–> weaker evidence of treatment effectivness
Quasi experimental dose respone analysis
=Difference intervention doses are compared to one another
What are some conditions researcher should adhere to when using quasi experimental designs and comparison
- Develop strong interventions documented in protocols
- Understand ad document the counterfactual
- Non equiv. designs = conditions to wich comparison group is exposed
- Time series designs = conditions existing before implementing the intervention
- Use blinding when possible (single - double)
What are strenghts and limitations of quasi experimenten designs ?
V: -practical
-likely to be acceptable to a broad group of people
X: -Rival hypothesis can explain results
=> inferences (veronderstellingen) about causal effect are weakened
What are experimental/observational (correlational cause probing) designs ?
Designs where the independant variable cannot be manipulated (technically or ethicaly)
- Correlational design
- Descriptive design
What is correlational design and what are the forms this design can take ?
correlational design = to explore the relationship or association between variables (change in one cause change in other ?)
-Retrospective design( case-control): PAST (indep var) obtain compareble groups (without the cause in the past)
- Prospective non-exp. design/Prosective cohort design
- -> PRESENT (presumed cause) —-> FUTURE (effect)
- Natural experiments: participants exposed to eg natural disaster are measured with non exposed group
- Path analytic studies: path analysis: study of causal chain among a set of variables (independant & mediating)
What are the strenghts and limitations of prospective non-experimental designs
V: -stronger then retrospective,
- confirmation of initial absence of effect: effect --> cause - more representative samples - Control groups possible
X: -Costly
-Long follow-up
-Large N required
-Participants need to be free of effect at baseline
=: -Exploratory = wide range of causes are measured
-A priori hypothesis
What is descriptive research and wich forms (2)?
- Descriptive correlational studies: describing relationship among variables rather than to support inferences of causality
- Univariate studies: epidemiology
What are the strenghts and weaknesses of correlational research ?
X: -weak in supporting causal inferences
- Groups are not random, but self-selected (selection bias) - Behaviors, attitudes and character are interrelated in complex ways (alternate explanations)
V: -Many problems cant make use of experimental designs: causal inferens is possible with strong design
- Large amount of N - Strong in realism settings
