Lesson 5 - Causal Research & Experimental Design Flashcards
Conditions for causality
- Time Order ( x cannot happen after y)
- Elimination of other possible causes (holidays)
- Covariation (y must co-vary with x)
Correlation
Is not causation
Example correlation ❌ causation
Ice cream sales and shark attacks are correlated because they occur on the same time but there is no causation because there is no covariation between them and there is no evidence that they co-vary
What happens at correlation
X and y can change together
What happens at causation
Changing x causes a change in y
What is an experiment?
- primary method of establishing a cause-and-effect relationship
- used for testing specific hypotheses
Steps of an experiment
- Divide participants into two or more groups ( control and experimental group)
- Expose each group to different treatments
- Measure the outcomes of each group
- Compare outcomes for differences
Example experiment Coke
- 2 groups
- control group: small thick can of coke
- experimental group: tall skinny can
- ask them which can has more coke in it
- people think tall thin ones have more coke in it and they drink it slower than the others
Basic definitions and concepts of experiments
- independent variable
- dependent variable
- experiment
Independent variable
- predictor variables
- variables that are manipulated by researcher
- ex package design
Dependent variable
- outcome variable
- variables that measure the effect of manipulating the independent variables
- ex. sales
Definition Experiment
Process of manipulating independent variables and determining their effect on dependent variables
Example experiment 2
Will students score be affected by distracting sounds in the environment?
Control group: takes the test normal
Experimental group: has a band in the room while taking the test
Example experiment 3
Manager wants to find out whether in-store phone use affects purchase
Independent variables: phone use
Dependent variable: total purchases
Field experiments
Conducted in natural setting where realism is higher
Lab experiments
Conducted in lab where extraneous variables can be controlled
Pro-Contra lab studies
- high internal validity
- controlled environment
Pro: stronger ability to establish causality
Contra: low generalizability
Pro-Contra field experiments
- high external validity
- real-world environment
Pro: natural responses are obtained
Contra: difficult to control, expensive
Extraneous variables
Any variable that you are not investigating that can potentially affect the outcomes of your research (ex. Black Friday)
Goal experiments (picture)
Manipulate:
Independent variable
(Phone use) \
Dependent variable (Consumption)
Extraneous variables /
z (Black Friday)
Threats to validity
Extraneous variables
Types of extraneous variables
- history
- maturation
- mortality
- instrument variation
- selection bias
- testing effects
- demand effects
- regression to mean
Extraneous variable history
Events or actions during experiment that may influence dependent variable
Extraneous variable maturation
Change in participants during experiment that may influence response to treatment
Extraneous variable mortality
Drop-out of participants during experiment
Extraneous variable instrument variation
Change in measurement instruments that may effect measurements
Extraneous variable selection bias
Biased participant selection that drives differences between test and control group
Extraneous variable testing effects
Mere deft of earlier responses on later responses
Extraneous variable demand effects
Effect of participants guessing your hypothesis on behavior because there trying to help you find results
Extraneous variable regression to mean
Tendency to move toward average response
Types of experimental designs
- pre-experimental
- true experimental
- quasi-experimental
-statistical
Experimental designs pre-experimental
Either a single group or multiple groups are observed subsequent to some treatment presumed to cause change
Single case is observed at 2 time points, one before the treatment and one after. Changes in outcome are presumed to be result of intervention or treatment
-> one-group-pretest-post-test
Experimental designs true experimental
Participants are randomly assigned to either the experimental or control group
Selection bias is eliminated by randomization
Experimental design quasi experiment
There is no randomization of test units to treatments
You hypothesize that a new after-school program will lead to higher grades
Ethicality of experimentation
- responsibility to protect participants from harm
- respect rights and dignity of participants
- ethical guidelines
Limitation of experimentation
- experiments can be time consuming especially during long term experiments
- expensive because control, experimental group and multiple measurements
- it may be impossible to control effect of extraneous variables (especially in field experiment)
Research designs
Exploratory research design
(Qualitative)
⬅️➡️
Conclusive research design
(Quantitative)
⤵️
Descriptive research
Causal research