Experimental Design Flashcards
Unrelated design
Also known: between subject, between participants, independent groups etc
TWO DIFFERENT HROUPS ARE COMPARED
Related design
Also known: within subject, within participants, repeated measures, dependent groups
Comparison between same groups
Why experimental design
- Required for special equipment (eg. EEG)
- keeps all factors constant other than manipulated variable
- easier to draw conclusions about causalit
Non-experimental design
Not possible to rule out cometong explanations
non experimental vs experimental
Non experimental: how many drivers honked at a car parked at green light
Experimental: testing in cool and hot environments to see if aggression raises with hot environment
True (randomized) experiment
- Random assignment
- Experimental manipulation of independent variables
- Standardization
Experimental manipulation (independent vs dependent)
Indep. Variable: assume t to cause affect on another variable, manipulated by researchers
Dep. variable: measured for changes affected by experimental manipulation
Experimental manipulation
- uncertainty as to wether or not the manipulation worked
Standardization of procedures
- Ideally, experimental and control conditions are identical in every way but the variable is manipulated
- constant factors (experiment, participants characteristics)
Aim of randomization
- control differences between participants
- avoid systematic biases
Confounding variable
Uncrontrolled and systematic varying between independent variable
Matching
Controlling for difference between participants
Pre-Tests
Seeing if there are underlying differences between the group that could effect the interpretation of the experiment
Advanced experimental design
- Increasing th levels of indep. Variable
- using two or more independent variables
- using more than one dependent variable
Analytical plan
Omnibus Test: wether indep. Variable has significant effect: DOES IT TELL US WHICH CONDITIONS ATE DIFFERENT CROM EAVH OTHER
Post-hoc comparison: determine which condition differ significantly from another -> compositions are between the conditions
- family-wise/experiment-wise error rate: more comparisons-> greater likelihood of significant comparisons that are through chance (unnecessary comparisons)
