Coding the Codebook

What we are building today

Lab session

• A project folder that holds this entire study in one place
• Your Chapter 8 codebook, installed in it as a real document
• The Twitch data, loaded into R and looked at for the first time
• No analysis today. Nothing gets computed. We open, we load, we look.
• Everything here becomes the Methods section of your White Paper

Why a project folder comes first

• You have designed a whole study and never seen the data
• Reproducible work lives in files in known places, not commands typed once
• A project is one folder holding the data, code, codebook, and report
• The study then travels together, and nothing depends on a file only you have
• Deciding where files go is friction. We are going to skip that decision.

Scaffolding the project

v2v::new_portfolio("twitch-portfolio")

That one line creates the whole folder for you, already named and already wired together.

Note the v2v:: prefix. It says the function comes from this book’s package, so the package never becomes a hidden assumption.

What the scaffold gives you

twitch-portfolio/
├── twitch-portfolio.Rproj
├── _quarto.yml
├── README.md
├── codebook.qmd
├── analysis.qmd
├── data/
└── .Rprofile

You did not decide where anything goes or what to name it, which is the point.

Confirming the stack

v2v::setup()

setup() checks that R, the packages this book depends on, and the project environment are all present and agree on versions.

• You met it once already, when you installed the stack
• Run it every time you open the project
• You find out about a missing piece before it interrupts you mid-analysis

The rulebook becomes a document

• Look at the scaffold again. One file is codebook.qmd.
• In Chapter 8 your codebook was an argument on paper
• Now it becomes the project’s rulebook, the reference every later step answers to
• Move the draft in, add a version line and a date, and render it
• A codebook that renders cleanly is one a second coder can actually read

Loading the data

library(tidyverse)

chat <- v2v::twitch_chat()
streams <- v2v::twitch_streams()

Line one loads the tidyverse, the toolkit we use all term; the next two lines fetch the chat table and the stream table and give each a name with <-.

The data arrives through a function call, not a file path, which removes an entire category of beginner error. What you get is the working sample from the fifty-channel corpus, not the full 2018 collection.

Looking: the chat table

glimpse(chat)

Rows: 35,267
Columns: 5
$ id      <int> 89, 551, 1033, 2094, 2803, …
$ channel <chr> "sodapoppin", "xqcow", "forsen", …
$ sender  <chr> "madzee", "prometheanow", "spectre155", …
$ message <chr> "???????????????????????", "TriEasy Clap TriEasy…
$ date    <dbl> 1542578127023, 1542578135562, 1542578143610, …

glimpse() prints every column, its type, and its first few values.

Reading that output

• 35,267 rows, five columns, and each row is one chat message
• The first message is twenty-three question marks. The second is copypasta.
• Those are not errors. That is what Twitch chat contains.
• date is a <dbl>, an ordinary number, and is completely unreadable
• It counts milliseconds since 1970. We fix that in Chapter 11.

Looking: the stream table

glimpse(streams)

Rows: 32,276
Columns: 6
$ channel <chr> "sodapoppin", "xqcow", "forsen", "giantwaffle", …
$ title   <chr> "Hello truckers\n44835158804929_2115841973", …
$ game    <chr> "Marble It Up!", "Just Chatting", "Artifact", …
$ viewers <int> 27934, 19381, 15300, 4697, 28076, …
$ date    <dbl> 1542578200214, 1542578200240, 1542578200273, …

Each row is one snapshot of one stream at one moment: its title, its category, its viewer count, and when the snapshot was taken.

Counting what channels stream

streams %>% count(game, sort = TRUE)

   game                     n
 1 Art                   5298
 2 Fortnite              3682
 3 Just Chatting         2474
 4 Hearthstone           2157
 6 League of Legends     1952
# ℹ 50 more rows

count() tallies how often each value appears, and sort = TRUE puts the biggest first.

Confirming the corpus

n_distinct(chat$channel)

[1] 50

n_distinct() counts how many different values a column holds.

• Fifty channels, exactly the corpus the book has described since Chapter 1
• n_distinct(streams$channel) returns the same fifty
• Sixty game categories across fifty channels, so most channels move around
• Counting is how you check that the data is what you were told it is

When a line breaks

• Sooner or later R will return an error instead of a result
• Misspelled function, package not loaded, a stray bracket or quote
• This happens to everyone who writes code. It is not a verdict on you.
• The Hub keeps a common-errors cheat sheet: that is the first place to look
• An error message is information. Read it before you retype anything.

The package ships the ones specific to this data: run ?v2v::common_errors for the year-50,888 timestamp, the join that matches nothing, and the surprisingly low kappa.

Checkpoint

You should now have:

• A twitch-portfolio folder, created by new_portfolio()
• A clean v2v::setup() result
• Your Chapter 8 codebook living in codebook.qmd, versioned and rendered
• chat and streams loaded, and a glimpse() of each
• Nothing changed in the data. Looking is all we did.

Before Thursday

• Finish moving your codebook into codebook.qmd and render it
• Codebook and Qual Memo is due this week: the full codebook plus a 200 to 300 word qualitative memo
• Bring your glimpse() output on Thursday. We read the data against your codebook and find out whether your study is buildable.