Week 2 - Computational Thinking and Reproducilibility

Reproducibility & Data Sharing - What is the difference between the evaluative and generative purposes of data sharing?

Evaluative data sharing increases the credibility of findings by allowing others to verify and interrogate evidence;

generative data sharing enables researchers to pursue new questions and maximize research value.

What distinguishes confidential data from anonymous data?

Confidential data means researchers know participants' identities but agree to keep them private

anonymous data means the researchers themselves do not know the participants' identities.

Why is stripping direct identifiers (like names and social security numbers) insufficient to make a dataset truly anonymous?

Because auxiliary data matching (such as cross-referencing sparse database ratings with public sources like IMDB or Netflix reviews) can successfully re-identify individuals and connect records to real-world identities.

Replication Packages & README Files - What four components comprise a complete data replication package according to guidelines like the AJPS?

A README file, analysis datasets, software commands, and information to reconstruct the analysis dataset.

What core information should a README file include as the first item visitors see on a repository?

What the project does

how users can get started

where they can get help,

who maintains and contributes to the project.

What is the priority search order GitHub uses when displaying a README file if multiple exist in a repository?

First the .github directory, second the repository's root directory, and finally the docs directory.

Version Control & Schematic Git Workflow - What are the four file environments in a Schematic Git Workflow?

the Remote Repository (GitHub),

the Local Repository,

the Working Folder (Local Directory), and

the Staging Area.

What are the 5 core steps and commands in the Git workflow cycle?

1. git pull (update from remote repository) Downloads the latest changes from the Remote Repository directly into your Local Repository so you are working on the most up-to-date version. 

2. git checkout (get current version). Updates your Working Folder with the latest commit or specified branch from your Local Repository.  

3. git add (stage new work) Moves modified or newly created files from your Working Folder into the "Staging Area".  

4. git commit (commit changes to local repository). Takes everything in the "Staging Area" and permanently saves it as a new version snapshot inside your Local Repository.  

5. git push (send changes to remote repository). Uploads your new local commits up to the Remote Repository on GitHub so collaborators can see your work. 

Script Editor & Package Management - Why should you never include install.packages() inside a shared R analysis script?

• Because it modifies another user's local machine setup and package library without their explicit permission.

• 
  

How does groundhog.library() ensure historical reproducibility across years?

It loads packages and their hidden dependencies matching the exact versions that existed on CRAN on a chosen historical date, preventing version-update errors.

What is the purpose of using pacman::p_load() for package management?

t streamlines batch loading by checking if multiple packages are installed, automatically downloading missing ones from CRAN (R app store) and loading them into the active session in a single command.

Computational Reproducibility & File Naming - How is computational reproducibility defined as an equation?

computational reproducibility = code + data + environment + distribution

What are the key rules for machine-readable file names?

Avoid spaces, symbols, and special characters, and do not rely on case sensitivity (e.g., distinguishing "a" from "A") to separate files. + pre-fixed with numbers to allow execution in chronological order

Project Architecture & Workspace Configuration - Why should you configure RStudio's workspace settings to never save .RData on exit?  

To prevent hidden variables from persisting across sessions, ensuring your R scripts serve as the reproducible "source of truth".  → when you send your R script to professor, they can run the code on their FRESH R SESSION

GitHub Integration & Undoing Mistakes -How does git revert [commit id] differ from pressing Command + Z when fixing a pushed mistake?

Command + Z locally erases typing invisibly; git revert creates a new permanent commit that applies the exact opposite action of a faulty past commit and is pushed publicly to maintain a transparent audit trail.

Data Import, Cleaning, & Parsing - How do you handle column names with spaces (non-syntactic names) when referencing them in R?

You must surround them in backticks (e.g., Student ID) or automatically convert them using janitor::clean_names().

What does read_csv() do by default with missing values, and how do you handle custom strings like "N/A"?

By default, it only recognizes blank strings ("") as missing; you must use the na argument (e.g., na = c("N/A", "")) to register custom missing codes as true NA values.

What function should you use when read_csv() throws a warning due to unexpected column types or missing value markers?

You use problems(df) to inspect detailed row-by-row parsing errors after forcing column types with col_types.