search

jjmedinaariza / modelling_book

This is the digital textbook for Modelling Criminological Data
https://jjmedinaariza.github.io/modelling_book/
4 stars 6 forks source link

Updates required for 2020 #2

Open jjmedinaariza opened 5 years ago

jjmedinaariza commented 5 years ago

1) Shift to rio as the main import/export package 2) Add something in chapter 4 about renaming factor levels 3) Add visual aid cards to explain the code in simpler terms 4) Explore greater interactivity (eg., swirl etc) 5) Get the syllabus out of BB and place it here with all the reading hyperlinks 6) Think about simplifying materials and continue ensuring we complete the shift to tydiverse 7) Change the homework for week 6 (and the guidance about t test assumptions) 8) Include mosaic plots in week 7 9) Say something in week 4 about sometime treating ordinal as numeric (as a practice in disuse) 10) Think about building hyperlinks to substantive reading/videos per week 11) Develop cumulative cheatsheet of functions in a weekly basis (with a key to section where is covered) 12) Introduce at an earlier stage how to invoke functions from particular packages without loading the package 13) Ensure appropriate storage and workflow with better instructions for setting up the working environment in a weekly basis and saving any modified data at the end of each session. 14) Review guidance for assignment 15) Ensure greater use of piping operator through the course 16) Think about how to incorporate criticisms of p values 17) Re-think use of odd ratios in week 7 18) Make the material about assumptions for ANOVA optional

jjmedinaariza commented 5 years ago

Check this out as well: https://journals.sagepub.com/doi/10.1177/0003122418806282

jjmedinaariza commented 5 years ago

https://twitter.com/analisereal/status/1108533039712235520?s=12

jjmedinaariza commented 5 years ago

In week 4 refer this: http://www.stat.columbia.edu/~gelman/research/published/multiverse_published.pdf

jjmedinaariza commented 5 years ago

In section 2, 3 and 4 we need to start signposting more what the general direction is (building models). In section 4 we need an explicit section on model building as part of the description of variable selection.

jjmedinaariza commented 5 years ago

Consider dropping sjPlot and use alternatives (stargazer for tables and something else for forest plots)

jjmedinaariza commented 5 years ago

We need to add more on explanation and interpretation of regression results (like a template). It would be helpful as well to have some code for customising ggplot2 (themes, labels, etc). In addition, I need to check and expand the section on chapter 4 for recoding factors. And further explain the code for relevel in the logistic regression week. It would also be helpful to explain how to obtain predicted values for combinations of covariates in logistic models (see Long book).

jjmedinaariza commented 5 years ago

When discussing model building we need to explicitly address endogeneity as an issue.

jjmedinaariza commented 5 years ago

We also need to discuss the design of stacked bar charts for making appropriate comparisons.