L8: Causality

Question 1

Predictive models do not necessarily tell us anything about cause and effect relationships needed to understand or explain a phenomenon
TRUE/FALSE

Answer 1

TRUE

Question 2

When two variables display an increasing or decreasing trend. E.g., X and Y are correlated if observing a change in X tells you that Y will either increase or decrease

This is causation - TRUE/FALSE

Answer 2

FALSE - This defines correlation

Question 3

When one variable provides information about another variables. This is correlation -TRUE/FALSE

Answer 3

FALSE, this is ASSOCIATION - not all associations are correlations

Question 4

Correlation is one specific type of association. Association is a broader concept
TRUE/FALSE

Answer 4

TRUE

Question 5

When a change in one variable causes a change in another variable. This is causation? -TRUE/FALSE

Answer 5

TRUE
Causation tells us how or why something happens

Question 6

Association does/ does not imply causation

Choose the right word

Answer 6

Association DOES NOT imply causation

Question 7

Correlation does/ does not imply association

Answer 7

Correlation DOES imply association
Recall, association is the broader term and correlation is a specific type of association

Question 8

Correlation does/does not imply causation

Answer 8

Correlation DOES NOT imply causation

Question 9

Causation implies association, but not correlation

TRUE/FALSE

Answer 9

TRUE

Question 10

Accurately describing the causal mechanisms underpinning observations is ______. This is generally what ML is not good at.

A) Explanation
B) Prediction

Answer 10

Accurately describing the causal mechanisms underpinning observations is EXPLANATION. This is generally what ML is not good at.

Question 11

Explanatory tasks is where ASSOCIATION is NOT enough (oftentimes)
TRUE/FALSE

Answer 11

TRUE

Question 12

Predictive tasks is where association can be enough (oftentimes)
TRUE/FALSE

Answer 12

TRUE

Question 13

What are Directed Acyclic Graphs (DAGs)?

Answer 13

A way to visually represent causal assumptions as nodes (variables) and arrows (causal relationships)

Question 14

What is a confounder variable?

Answer 14

A confounder is a variable that is related to both the independent and dependent variables in a study, potentially leading to a spurious association between them

Question 15

The size of shoe is dependent on the age of the student, which also affects the SAT score of the student.
Which variable is the confounder?

Answer 15

Age

Question 16

When relevant confounding variables are unobserved, ignored, or otherwise omitted from a model, this can lead to spurious association between an exposure (dependent variable) and outcome.
TRUE/FALSE

Answer 16

TRUE

Question 17

A mediator is a variable that explains the process or mechanism through which an independent variable influences a dependent variable.
Adjusting for the mediator (Z) may remove the causation between X and Y, which is bad, cause there is causation, but this causation travels through Z

TRUE/FALSE

Answer 17

TRUE

Question 18

A collider is a variable that is a descendant of both the exposure and outcome variables - i.e., both the dependent variable and independent variable influence the collider variable

TRUE/FALSE

Answer 18

TRUE