4. OLS with 1 variable.
5. hypothesis testing for 1 variable model
6. OLS with multiple variables
7. hypothesis testing for multiple variables
8 : non-linear (transforming to log-linear, log-log etc.)
1101 max likelihood
1001 panel data
1201 iv regression
1401 time series (temperature data, bodyweight)
Person's chi-sq test
anova
wilcoxon
experiment design : statistical power, how much data to collect?
decision tree : what to do when i see a new data set
multiple variables : OVB, multicollinearity, heteroskedasticity, model specs, interpreting R2
differences, OLS, multi OLS, DID, binary variables, heteroskedasticity, transforms (log linear etc.)
regression works with things that are continuous. if the data is categorical, then you should make binary variables
section 2.4 : the weighted average of 2 normally distributed random variables is also normally distributed (see sw page 38)
OLS assumptions
5.4 homoskedasticity var(u_i|x_i) = constant and heteroskedasticity
5.5 : theoretical foundations of OLS (advanced)
omitted variable bias (income on weight), or even simpsons paradox (surgery v medication on tumors. OVB, but also a probit/logit-type), but also shud have been a DID-type, because they didnt check the control group.
Frisch-Waugh theorem (appendix 6.3)
** DID : binary variable regression with interactive variable. how it is different from one of the chi2 tests?