Correlation vs. Causality

Question 1

What are causal relationshops?

Answer 1

Unidirectional: A→B but NOT B→A

1. Flowers open, because the sun has risen
2. The sun rises, not because flowers open

Bidirectional: A→B and B→A

1. There are fewer rabbits, because there are a lot of foxes.
2. There are fewer foxes, because there are fewer rabbits.

Question 2

What is correlational research?

Answer 2

• One step back
• No causal link is established (yet)
• Changes in one variable occur together with
changes in a second variable → They covary

• Non-experimental research
• No variables are manipulated
• Can determine: Direction, magnitude and
form of correlation
• Value of Predictor Variable predicts value of
Criterion Variable

Question 3

What is the third-variable problem?

Answer 3

• Two variables covary, because they are both influenced by the same third (hidden) variable
Example: There is a moderate correlation between sunscreen use and skin cancer. The amount of strong sunlight is the third variable.

Question 4

What is the directional problem of third variable problem?

Answer 4

• The challenge of determining which factor causes which
Example: Watching violence on TV might cause people to have violent tendencies. People with violent tendencies might be drawn to watch violence on TV

Question 5

when do you use the third variable problem?

Answer 5

• Fast identification of potential causal relationships
• Variables are not manipulatable (eg ethical reasons)
• Research on naturally occurring phenomena (seasons, weather)

Question 6

What are different types of variables in experimental research?

Answer 6

Independent: Different conditions Dependent: Measured outcome
Extraneous variables:
• Outside of the experimental design but still
influence the dependent variable
• Controllable: Calculator used at exam
• Uncontrollable: Differences in skill

Question 7

What is a criterion of causation?

Answer 7

• Strong association
• Consistent association
• Independent V precedes outcome
• Outcome increases with increase of
Independent V
• Causation is plausible
(can also apply for quasi-experiments)

Question 8

Control condition & random assignment in experimental research

Answer 8

Control/placebo group:
• Often included to determine if true effect
exists
Random assignment:
• Solves problem of uncontrollable extraneous
variables, because it balances differences out between groups

Question 9

What are other elements in experimental research?

Answer 9

Quasi-Independent variable:
• Natural groups (gender)
• Reduces error variance Demonstration:
• Only one treatment condition
• Can demonstrate how something plays out
under certain conditions.
→Both cannot establish causality

Question 10

What is cross-sectional design?

Answer 10

• Measuring different age groups on a specific outcome variable
• Allows for efficient data collection
• Generation effects: age groups might be
influenced by events of their time

Question 11

What is longitudinal design? Problems?

Answer 11

• Following a group of people over time
• Clearly shows developmental changes
Problems:
• Cross-generational effects: results might
not apply to a different generation • Subject mortality (if nonrandom) • Carry over effects
• History

Question 12

What is the simpson´s paradox?

Answer 12

It depends…
• Whether results are counted per day or in total
• On both days a Maastricht had a higher
success rate
• In total Eindhoven had a higher success rate even tho it looked like maastricht did better

Question 13

Exam question:
A research team is interested in the question if girls acquire reading abilities earlier than boys. The randomly choose 20 girls and 20 boys from an elementary school and let them perform a reading task. The final score of each child is formed out of reading fluency and accuracy.
What type of design did they use?

A. Experimental Design
B. Correlational Design
C. Demonstrational Design
D. Developmental Design

Answer 13

Answer B