W2 - Experimental Design Flashcards
Independent Variable
Aspect researcher is going to manipulate
Dependent Variable
Whats measured
Depends on the cause
What are the types of research designs?
Non-experimental / Observational
Experimental
What comes under non-experimental / observational research designs
Cross-sectional
Longitudinal / Cohort
What comes under experimental research designs?
Quasi-experimental
True experimental
Pre-experimental
Is anything manipulated in non-experimental / observational research designs?
No
They analyse the natural differences in the IV vs changes in the DV.
What do experimental research designs test?
Whether the manipulated IV causes a change in the DV.
What is the scientific method for research designs?
Observations
Research Q
Background research
Develop hypothesis
Test predictions: Observational or experimental research
Analyse data
Report Findings
What constitutes a well-designed experiment?
Is there a good theoretical framework
Appropriate participants
Appropriate experimental design
Proper selection + control of IV
Appropriate selection + measurement of DV
Correct statistical model + analysis
Correct interpretation of results
Determining cause and effect
How can causality be achieved?
Cause + effect must correlate
Cause should precede the effect
All confounding variables must be ruled out.
What are the 2 types of study design?
Cross sectional
Longitudinal
What does cross sectional study design involve?
Observation
Outcome + exposure is measured at the same time
Nothing is manipulated
What does longitudinal study design involve?
Observation
Subjects followed over time
Repeated measures
Why are experimental designs potentially better at determining causality
Control confounding variables so rival hypotheses can be rejected
Able to isolate the cause (IV)
Can the independent variable be present at different levels?
YES
What is pre-experimental design?
Subject/group is observed post-treatment in order to test whether the treatment has the potential to cause change.
True-experimental design
Controls most threats to validity
Random assignment to groups
Control group
Randomised post-test only
What is the “randomised post-test only group design” under the experimental design?
Random allocation to either treatment OR control group.
Have outcome of interest measured after they receive that treatment.
How can the true cause-effect relationship be determined in the true experimental designs?
Randomised Control Trial (RCT)
Randomised Control Trial (RCT)
What does the model show?
- Random allocation
- Pre-measurement
- Treatment OR no treatment
- Post measurement
How can the Randomised Control Trial (RCT) be expanded?
Through the cross-over design
What happens in the cross-over design of the RCT
The treatment group crosses to the 2nd row to become NO-treatment –> Post 2 measurement
No treatment group crosses to 1st row to become Treatment –> Post 2 measurement
What is the period called between the post 1 measurement of the RCT and the crossover?
Wash out treatment
What is the point of the crossover in the RCT?
Provides more reliability
Purpose of the quasi-experimental designs
When they fit the study into a more real world setting
In quasi-experimental designs how do they fit the study into a more real world setting?
Control for internal validity
Maximise external validity
Confidence to conclude causality continuum
Order from least controlled study design to most controlled
Cross-sectional
Pre-experimental
Causal comparative
Quasi-experimental
True-experimental
Give 3 examples of pre-experimental designs
1-shot study
1-group pretest-posttest design
Static group comparison
PRE-EXPERIMENTAL DESIGN
1-shot study
Treatment then test
PRE-EXPERIMENTAL DESIGN
1-group pretest-posttest design
Very weak but better than 1-shot study.
Observation pretest -> Treatment -> Observation posttest
PRE-EXPERIMENTAL DESIGN
Static group comparison
Compares 2 groups
- 1 receives treatment
- other doesn’t
Give 3 examples of true-experimental designs
Randomized-Groups design
Pretest-posttest randomised groups design
Solomon 4-Group Design
TRUE-EXPERIMENTAL DESIGN
Randomized-Groups design
Compares 2 groups: 1 receives treatment, other doesn’t.
Groups randomly formed.
TRUE-EXPERIMENTAL DESIGN
Pretest-posttest randomised groups design
Randomly formed groups
Both are given pre + post tests.
To determine amount of change prod by treatment
TRUE-EXPERIMENTAL DESIGN
Solomon 4-Group Design
Specifically evaluates 1 threat to external validity: reactive or interactive effects of testing.
TRUE-EXPERIMENTAL DESIGN
Solomon 4-Group Design
Show the design
R - O1 - T - O2
R - O3 - .. - O4
R - …… - T - O5
R - …… - … - O6