ECONOMETRICS EXAM 2 (Topics 4,5,6)

selection bias

systematically excluding groups (ex. surveying only friends, only people in one building)

non response bias

when those who skip a survey differ from those who answer (ex. students with low GPAs less willing to share grades)

researcher bias

guiding results/leading questions/ cherry picking data that fits a hypothesis

survivorship bias

focusing only on “survivors” and ignoring those who were dropped or excluded (ex. WW2 planes reinforcing where holes aren’t / dead people)

experimental data

collected in controlled setting using random assignment to isolate specific variables

observational data

collected from real world behavior, plagued by confounding factors

confounding factor

influences both the cause and effect, creating a fake relationship (ex. book ownership and longevity; education and income)

law of large numbers

about accuracy, as the sample grows, the average of the sample means will converge on the true population

central limit theorem

about distribution, if you take enough samples, the distribution of these means will become normal

mutually exclusive

the two events can’t both happen, its either one or the other, there is no overlap

expected value

weighted average, tells you the center, doesn’t have to be a real outcome, just the long run average if you repeated one event forever