Day 10 (June 12)
Announcements
- Recommended reading
- Chapters 2 and 3 (pgs 13 - 46) in Linear Models with R
- Questions and comments from journal
- Keep up with the great journals
- “One concept I am still struggling to understand is how much confidence researchers should place in results when small changes in model assumptions can substantially alter the conclusions.”
- “Something I did not understand from class today is why we can’t use the same data for testing that we used for making coeficients.”
- “I am struggling to understand why we would use the optim() function for optimization or standardization when we already have a direct formula to calculate the coefficients.”
- “For example, to compute the least-squares estimate, we only need to define the model and the loss function to be minimized. Assumptions such as normality are not required to obtain parameter estimates. However, additional assumptions are typically needed to perform statistical inference, such as constructing confidence intervals or conducting hypothesis tests.”
- Interesting story (book and paper)