POSC 8410: Getting up and running in R

Will Horne

Today

By the end of today you will have:
- Survey data loaded on your own machine
- Generated a figure
- Seen your first error messages + how to deal with them
We are going to do this fast, and slightly out of order

Follow along on your machines. If something doesn’t run, stop me. You won’t be the only one having trouble!

A Warning About Today

I am going to show you working code before I explain how it works
You may not understand the first thing I show you. That is ok!
- We will start with the finished analysis, and then work back piece by piece
By the end of the week, this analysis should make sense
- You’ll have a problem set to reinforce

A Complete Analysis

Five Lines

library(tidyverse)

anes <- read_csv("Data/anes.csv")

anes |>
  filter(!is.na(votechoice), votechoice != "Other") |>
  ggplot(aes(x = ft_biden, fill = votechoice)) +
  geom_density(alpha = 0.6)

That is a complete analysis: load a tool, load data, pick the rows you want, draw a picture.

The Result

What Just Happened?

Four things, and they are the four things you will do for the rest of your career:

Load a package — borrow someone else’s code
Read a file — get data off your hard drive and into memory
Subset — keep the rows you care about
Plot — look at it

Getting Oriented

The Positron Tour

I will walk through this live. Four things to find on your screen:

Console — where code actually runs. Type here for throwaway work.
Editor — where your scripts live. Type here for work you want to keep.
Variables — everything R currently has in memory.
Plots — where figures appear.

Console vs. Script

Console: disposable. Gone when you close Positron.
- Good for: “what does this column look like?”
Script (a .R file): saved. Reproducible.
- Good for: everything that is part of your actual analysis
Run one line from a script with Cmd/Ctrl + Enter

Rule: if you would be annoyed to lose it, it goes in a script.

You Try (1)

Make a folder for this class somewhere sensible
Open that folder in Positron: File → Open Folder
Make a new script: File → New File → R Script
Save it as week2.R inside your class folder
Type 2 + 2, put your cursor on the line, hit Cmd/Ctrl + Enter

Hands up if you did not see [1] 4.

The Building Blocks

Objects

R remembers things when you assign them a name with <-

x <- 4
y <- 10

x + y

[1] 14

x now lives in memory. You can see it in the Variables pane.

Objects, continued

Names are case-sensitive and cannot start with a number.

mean_temp <- 72.5
mean_temp

[1] 72.5

Use names that say what the thing is
data2_FINAL is a name you will regret in the future

You Try (2)

Create an object called n_states and set it to 50.

Then create an object called n_senators equal to n_states times 2.

n_states <- 50
n_senators <- n_states * 2
n_senators

[1] 100

Functions

A function takes inputs and returns something.

sqrt(16)

[1] 4

The inputs are called arguments and go inside the parentheses.

round(3.14159, digits = 2)

[1] 3.14

Functions You Will Use Constantly

scores <- c(88, 92, 79, 95, 61)

mean(scores)

[1] 83

length(scores)

[1] 5

c() means “combine” - it builds a vector, an ordered set of values of the same type.

Packages

Base R can do a lot. Other people have written code that does more.
A package is a bundle of that code
Two steps:
- install.packages("tidyverse") — once, ever, in the console
- library(tidyverse) — every single session, at the top of your script

Don’t put install.packages() in a script you share. It re-downloads the internet on someone else’s machine.

Your First Error

Open a fresh session and try this before loading anything:

read_csv("Data/anes.csv")

Error in read_csv("Data/anes.csv") :
  could not find function "read_csv"

R does not know what read_csv is
It lives in the tidyverse, and we have not loaded it in this session
could not find function almost always means a missing library()

Load library(tidyverse) at the top of the script, always. Then it works.

Getting Data In

Let’s Load the ANES

Put anes.csv in a Data folder inside your class folder. Then:

anes <- read_csv("anes.csv")

Error: 'anes.csv' does not exist in current working directory ('/Users/rwhorne/Dropbox/Clemson/Teaching/POSC_8410/Fall_2026/Slides').

This is the most common thing that goes wrong. Let’s understand it.

Where Is R Looking?

R always has a working directory — the folder it treats as “here”
When you asked for "anes.csv", you said: look in “here” for a file with that name
There isn’t one. The file is in Data/, one level down.

getwd()   # what does R think "here" is?

The Rule

In Positron, the working directory is whatever folder you opened
So: open your class folder, not a random file, not your Desktop
Then every path you write starts from your folder

anes <- read_csv("Data/anes.csv")   # "from here, go into Data, get anes.csv"

This is why we set up the folder!.

Relative vs. Absolute Paths

Absolute — the full address, from the root of your hard drive:
- /Users/whorne/Dropbox/POSC8410/Data/anes.csv
Relative — directions from where you already are:
- Data/anes.csv

Always use relative paths. An absolute path works on exactly one computer: yours. It breaks for your coauthor, for a replication archive, and on any other device.

Do Not Use setwd()

You will see setwd("/Users/someone/...") in older code and tutorials
- It’s absolutely in some of my old scripts
It hard-codes an absolute path into your script
- Guaranteed to break for anyone else, including future you
Opening the folder does the same job

Model Setup

POSC_8410/
├── Data/
│   └── anes.csv
├── Scripts/
│   ├── week2.R
│   └── week3.R
├── Output/
│   └── figures/
└── ProblemSets/

Raw data goes in Data/ and never gets edited by hand
If you cannot rebuild every number from Data/ + your scripts, it is not reproducible

You Try (3)

Download anes.csv from Canvas
Put it in a Data folder inside your class folder
At the top of week2.R, write library(tidyverse)
Below that: anes <- read_csv("Data/anes.csv")
Run both lines

You should see anes appear in the Variables pane with 7,453 rows.

Looking at Your Data

First Look

glimpse(anes)

Rows: 7,453
Columns: 7
$ ID         <dbl> 200015, 200022, 200039, 200046, 200053, 200060, 200084, 200…
$ state      <dbl> 40, 16, 51, 6, 8, 48, 55, 28, 4, 6, 24, 55, 41, 51, 37, 16,…
$ vote       <chr> "No", "Yes", "Yes", "Yes", "Yes", "Yes", "Yes", "No", NA, "…
$ votechoice <chr> NA, "Other", "Biden", "Biden", "Trump", "Biden", "Trump", N…
$ ft_biden   <dbl> 0, 15, 85, 100, 0, 100, 0, 50, 85, 70, 100, 70, 0, 85, 70, …
$ ft_trump   <dbl> 100, 15, 0, 60, 90, 0, 70, 0, 0, 30, 0, 0, 100, 0, 40, 100,…
$ ft_police  <dbl> 85, 90, 40, 100, 70, 70, 60, 100, 60, 70, 50, 70, 50, 0, 60…

One row per respondent, one column per variable. This data is in tidy format, stored in a dataframe.

What Are `<dbl>` and `<chr>`?

Those tags are the column’s type, and they decide what you are allowed to do.

<dbl> — a number (“double”). ft_biden, ft_trump
<chr> — text. votechoice, vote
<lgl> — TRUE / FALSE
<fct> — a factor: categories with a fixed set of levels, and potentially order

Types Are Not Cosmetic

mean(anes$votechoice)

Warning in mean.default(anes$votechoice): argument is not numeric or logical:
returning NA

[1] NA

$ pulls one column out of a data frame
The average of “Biden”, “Trump”, “Trump” is not a thing
R warns instead of failing loudly — NA is often your first clue something is wrong

This One Works

mean(anes$ft_biden, na.rm = TRUE)

[1] 53.44922

na.rm = TRUE says “ignore the missing values”
Without it, one NA anywhere makes the whole answer NA
R is refusing to guess what a missing value should be. That is what we want it to do!

You Try (4)

Using the anes data:

What is the mean feeling thermometer toward Trump? (ft_trump)
How many rows does the data have? Try nrow()
What are the column names? Try names()
What happens if you run mean(anes$ft_trump) without na.rm = TRUE?

When Things Break

Errors Are Normal

Everyone who writes code sees errors constantly. Me included!
- I make so many typos…
An error is not a big deal. It is R telling you it could not do what you asked.
The messages are terse, but they are literal

Read the last line first. That is almost always the actual problem.

The Big Four

Message	What it means
`could not find function "x"`	You forgot `library()`
`object 'x' not found`	Typo, or you never ran the line that made it
`does not exist in current working directory`	Wrong folder open, or wrong path
`non-numeric argument to binary operator`	You did math on a text column

Object Not Found

mean(anesss$ft_biden)

Error: object 'anesss' not found

There is no object called anesss
Either you mistyped it, or you never ran the line that created it
Check the Variables pane: if it isn’t listed, R doesn’t have it

The Dangerous One

Now watch what happens when the column name is wrong:

mean(anes$ft_bidenn)

Warning: Unknown or uninitialised column: `ft_bidenn`.

Warning in mean.default(anes$ft_bidenn): argument is not numeric or logical:
returning NA

[1] NA

No error. A warning and an NA.
R does not know ft_bidenn was supposed to be ft_biden
- it just reports that nothing is there
This is worse than an error, because it will happily flow into your results

How to Get Unstuck

Read the last line of the error
Check the obvious: did you load the package? run the line above? spell it right?
?function_name opens the help page
Paste the error into a search engine (or Claude/GPT) — everyone does this, it is not cheating
Ask me

For Next Time

Make sure You Try (3) works. Next class starts with data already loaded.
Read: R4DS Chapter 2 (Data Visualization) — work through it, don’t just read
Read: R4DS Chapter 4 (Data Transformation), sections 4.1–4.3
Thursday: Quarto, the pipe, and data manipulation functions
- We finish by rebuilding the callback figure from last week