lecture 5 & 6

Question 1

causation

Answer 1

Assessing the impact of a factor (X) on an outcome (Y)

Question 2

two types of causation

Answer 2

Two Types:
1. Deterministic relationship: every time x occurs, y occurs
2. Probabilistic relationship: if x, then y is more likely

Question 3

problem of causal inference

Answer 3

What would have happened without the intervention
We would have to observe the same entity getting treatment and not getting it
But → the same entity cannot both get treatment and not get treatment

Question 4

five main hurdles to causation

Answer 4

1. A credible causal mechanism
   - Why independent variables cause dependent variables. A plausibility test.
2. reverse causality (endogeneity)
   - Does x cause y, or does y cause x? Sometimes it’s easy to rule out:
   * Gender → political views
   Often hard:
   * Economic development → democracy
3. do x and y co-vary
   - They should covary, one variable moves in one direction and the other moves in the same/opposite direction
   But they do not always. To fix this you could add a third variable and plot the results, and then should see a relationship between the variables
4. confounding variables
   - There could be a third variable Z (or set of variables) (confounding variable) that affects both the dependent variable (Y) and independent variable (X).
   If we do not account for this confounding, we risk drawing biased inferences from our study.
   We would falsely determine that there is a causal relationship between X and Y, when in fact this relationship is spurious.
5. selection bias
   - The way we select our observations is crucial. Ideally, they should be selected randomly.

Question 5

controlling for confounding variables

Answer 5

1. Design of the research: engineer randomness (e.g. experiment and quasi-experiments)
   * Intervention/Treatment
   * Experimental or treatment group: receives the treatment
   * Control group: does not receive the treatment
2. Statistical control: include the confounding variables in the model.

Question 6

measures of central tendency

Answer 6

Median → value that splits the distribution in half
Mean → average written as x̅
Mode → value with the greatest frequency

Question 7

Choosing measures of central tendency:

Answer 7

▶ Nominal data: Mode (mean and median CANNOT be used)
▶ Ordinal data: Mode, median (mean if categories as numbers)
▶ Quantity data: Mean or median
Median is more representative of the data
Mean is more sensitive to extreme values

Question 8

measures of dispersion

Answer 8

Dispersion → how the data is distributed, narrow or wide
Narrow (values are all close to one another)
Wide (values are spread out)

Question 9

quantifying the spread

Answer 9

the range (difference between the largest and smallest values)
Limitations: does not explain whether data is narrow or wide

Question 10

other name for the mean

Answer 10

average of deviation

Question 11

the mean of squared deviations

Answer 11

variance σ2

Question 12

standard deviation

Answer 12

Standard deviation is a statistic that measures the dispersion of a dataset relative to its mean and is calculated as the square root of the variance.

σ = √ σ 2 (and s = √ s 2 when we talk about the sample standard deviation).

Question 13

causal inference

Answer 13

the process of determining whether an observed association truly reflects a cause-and-effect relationship.