Chapter 7: Experimental design Flashcards
What are the elements to establish causality?
Time-order
Def: The cause must have occurred before the effect
- ex: adults arrested for domestic violence tend to commit fewer assault AFTER compared to those who are not arrested
- ex: motivational speakers
Co-variation
Def: there is an empirical association between the IV (X) and the DV (Y) that results in co-variation
- ex: more cigarettes smoking→ the greater the odds of lung cancer
—>need covariation for causality
Rationale/ explanation
Def: compelling explanation for the connection between the IV (X) and the DV (Y
- ex: relationship between childhood poverty and petty crime
—>increase confidence in a conclusion
—>why does relationship matter if don’t know why?
Non-spuriousness
Def: If relationship is actually due to a third variable
- ex: Grade schools with larger libraries cause or produce students who are better readers
- other explanatation→ parents’ education/views on education.
- not a confounding/ does not negate the relationship between X and Y but introduce another variable
What are the different types of control group?
Placebo group→ receive a fake treatment
- Placebo effect→ Just believing that the treatment will have an effect can cause a response
- psychosomatic effect
- Outcome research (focus only on outcome of treatment so don’t care if caused by placebo effect) VS Process research (identify active components of the treatment)
Waitlisted group→ Participants who sign up to receive same treatment
- control for motivation across groups
No treatment control group→ participants did not receive any treatment
What are the four main elements of an experimental design?
Manipulation→ Researcher manipulates one variable by changing its value to create conditions
- decide on levels of IV you want to examine
- ex: no treatment, small amount of prozac, large amount of prozac
Measurement→ second variable is measured for each condition, resulting in a set of scores in each treatment condition
Comparison→ scores in each condition are compared with the scores in other conditions
Control→ All other variables are controlled to be sure that they do not influence the variables being examined
How to deal with confounding variable?
Remove the confound→ not possible for all confounding variables
Hold the confound constant→ across conditions
- eliminates potential to become a confound
- standardized environment and procedures
- ex: same noise level
Standardize the confound
- ex: age and gender standardized (30yo females)
—>have to be careful about external validity by not being to strict
Use a Placebo Control→ experimental method can become confounding variable
- placebo control group
- ex: vaccine→ fear of needles can be a confound
Matching Across Conditions→ try to match levels of it across conditions
- ex: match age of participants from experimental and control groups
- but threat to generalizability and external validity
Can match average levels across conditions
- counterbalanced
- ex: switch rooms after half of participants undergo experiments
Randomizing Participants assigned to Conditions
- aim is to disrupt any systematic relationship between the extraneous and independent variables
- controlling many environmental and participant variables simultaneously rather than individually
- ex: Participants recruited for experiment on one of 3 testing days and assigned randomly to “Intervention” or “Control” conditions
—>block randomization→ keep size of each group similar over entire study
Manipulation checks are important when…
- Participant manipulation→ check to see if worked
- Subtle manipulations→ ex: Did you notice the expression on the experimenter’s face when she gave you the instructions?
- Placebo→ did they believe it was real?
- Simulations→ did they perceive environment as real?
What are the reasons for which an experiment might not have worked?
-IV not sensitive enough
-DV not sensitive enough
-IV or DV has a ceiling or floor effects
Measurement error
Insufficient power-> not enough participants
Wrong hypothesis
What are the threat to internal validity in experimental design?
History→ events that influence participants’ responses
- ex: metro shut down→ ask participants how they commute/ will change their answers
Maturation→ normal developmental processes
- ex: children change more quickly than adults
Statistical Regression→ scores naturally moves toward the mean
- low scores cannot go lower so will point toward the mean whereas high scores will go down
Selection→ if participants self-selected or assigned randomly
- if self-select, might not be representative of population
Experimental Attrition→ people drop out of experiment more in one condition
Testing→ previous testing affects later testing
- ex: less afraid of last vaccine so might be more effective
Instrumentation→ measurement method change during the research
- ex: interviewer gets tired
Design contamination→ participants find out something about experimental conditions
- ex: find that can win more during 1st condition so are more competitive than in 2nd condition
What are the threat to external validity in experimental design?
Unique Program features→ can be replicated ?
- ex: unusually motivated set of experimenters in some conditions
Effects of Selection→ can be replicated with different participants?
Effects of environment/ setting→ can be replicated in other labs?
Effect of History→ can be replicated in different time periods?
- ex: experiment for memory of jingles
Mundane realism VS experimental realism
- Mundane realism→ how close lab is from real world
- Experimental realism→ bring only psychological aspects into the lab/ don’t recreate environment of real world
