Methods Lecture

Question 1

Correlational vs Experimental Designs

Answer 1

Correlational:

• Measuring X and Y
• Y Happens more when X happens. X Y
• How is bullying effected by the presence of teachers
• R= -1 Negative correlation
• R= 0 No correlation
• R = 1 Positive correlation

Experimental:

• Manipulation of independent variable X
• Measuring of dependent variable Y
• Control of extraneous variables
• Y happens because of X

Question 2

Why can’t you infer causation from correlational studies

Answer 2

• Confounds

- 3rd variable can effect both

Question 3

Anatomy of the experiment

Answer 3

IV: X is independent, manipulated by the researcher. Ex. Type of video game.
DV: Y is dependent, measured by researcher, effect of cause. Ex. Amount of aggression
Design: Randomization – to control for ‘third variable’ confounds, subset of population, convenience samples often used.
Control variables: Everything constant but the variable of interest

Question 4

Ways to Assess dependent variables (4 things)

Answer 4

• Observation: Person as observer
• Survey: Self Report Bias
• Behaviour: Choice between options
• Physiological: FMRI, EEG, etc.

Question 5

Internal vs External Validity

Answer 5

Internal Validity:

• Nothing other than X can affect Y
• Controlling for extraneous variables
• Random assignment and random selection of sample from population
• Measuring what it’s supposed to measure. Does tabasco sauce actually measure anger?

External Validity:

• Generalizing results of the study to the general population that your sample is pooling from
• Making study believable- greater the realism the more external validity there is

Question 6

Causes of Replicability crisis (4 things)

Answer 6

• Boundary Effects: The effect was true but might not extend to all situations. Effect only shown in that lab at that particular time 9/11 OR Cherry picking samples
• File-Drawer Problem: Studies that fail to reject the null are less likely to be published
• Questionable Research Practices: Changing things to get the outcome that you want from the study. Fabricated results
• Insufficient Power: Large effect = few participants. Small effect = large number of participants. Needing X number of participants to detect an effect.

Question 7

Possible solutions for replicability crisis

Answer 7

Replication, open data and methods, pre-registration