Latest revision |
Your text |
Line 1: |
Line 1: |
| == 10/29 Agenda == | | == 10/27 Agenda == |
| | * Pset 5 |
|
| |
|
| * OS Ch 7 exercises
| |
| ** What to do with/about violations of assumptions? (e.g., 7.12b)
| |
| ** Causality (again)
| |
| *** 7.12c: Did the lead do it?
| |
| *** Trick-or-treating experiment from Aronow et al. (Tuesday): did the treatment do it?
| |
| * T-tests "in the wild"
| |
| ** What should/can you do with this test?
| |
| ** What concerns should you have / take into account?
| |
| * Big picture considerations revisited:
| |
| ** What to do with any test?
| |
| ** How do I know whether I'm using "the right" test when there are seemingly no rules and/or whatever rules exist seem to be breakable (sometimes)?
| |
| * Self-assessment exercise demo/introduction
| |
|
| |
| == 10/27 Agenda ==
| |
| * Pset 5 PCs
| |
| ** PC4: interpreting the results of this chi-squared test
| |
| ** plotting error bars, and walking through some of the techniques Aaron used in the PC5 solution
| |
| ** review: calculating standard errors.
| |
| *** What if don't have raw data (EQ2)
| |
| *** Should I use special formulas from a textbook (such as the formula for SE of a proportion) or the default formula (sd / sqrt(n) )?
| |
| ** Other questions?
| |
| * Pset 5 EQs
| |
| ** Base rate fallacy
| |
| *** Review the definition
| |
| *** How concerned should we be?
| |
| ** P-values
| |
| *** the "strength" of relationships
| |
| *** significance of differences rather than differences in significance
| |
| ** Chi-squared in particular
| |
| *** Goodness of fit? Test if the distribution fits the data?
| |
| * Meta-questions (many via Gus's comments)
| |
| ** The practical the purpose of a statistical test (is to find the p-value)?
| |
| ** What should you ask about any test?
| |
| ** What do we know about causality from any of these tests?
| |
| * Other questions
| |
| * Upcoming | | * Upcoming |
| ** Project plans
| |
| ** Stats material this week
| |
| ** Assessment coming. Discuss some details, goals of the activity, etc.
| |
| ** No problem set is due on Nov. 3 (election day!)
| |