R Guide

1. Open RStudio. Click File → New File → R Markdown…
2. Name the file and select your output type: HTML, PDF, or Word. Word is best for most users as it converts easily. HTML works without LaTeX or Microsoft Office. PDF requires LaTeX (which can be difficult to install). The output can be changed at the top of the file next to output:.
3. Every new R Markdown file starts with a setup chunk — do not delete it:

{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)

4. Save the file: File → Save As… and choose your location.
5. R does not auto-save. Save frequently with File → Save (or Ctrl+S), especially after writing major code chunks.

The quickest way to insert a code chunk is with a keyboard shortcut:

Alternatively, you can manually type the chunk delimiters:

```{r}
print("Example Code Chunk")
```

## [1] "Example Code Chunk"

Create one main folder for the course, then subfolders for data files and markdown files. If you're using a lab desktop, use a flash drive so you can access files on any machine.

Use install.packages() to install a package. The name must be spelled correctly and surrounded by quotes. Only install each package once.

install.packages("stargazer")

Use library() to load a package each time you open R. No quotes needed.

library(stargazer)

## Please cite as:
## Hlavac, Marek (2022). stargazer: Well-Formatted Regression and Summary Statistics Tables.
## R package version 5.2.3. https://CRAN.R-project.org/package=stargazer

1. Copy the file path. On Windows: click the file then Ctrl+Shift+C, or right-click → "Copy as path". On Mac: Control-click → hold Option → "Copy as Pathname".
2. Switch any backslashes (\) to forward slashes (/) in the path. Then use load() and View():

# load("C:/Users/kkpat/Desktop/Econometrics/RData/attend.Rdata")
# ExampleData1 <- attend
# View(ExampleData1)

1. Convert Excel files (.xlsx/.xls) to .csv by saving as CSV in Excel first.
2. Load tidyverse or readr, then use read_csv(). The result is a tibble — a modernized data frame:

library(tidyverse)
ExampleData2 <- read_csv("C:/Users/kkpat/Desktop/STAT320/Data/SelectedVars.csv")
View(ExampleData2)

To save the tibble as an .RData file (saves to the same folder as your R Markdown):

save(ExampleData, file = "ExampleData.RData")

Stargazer produces publication-quality tables for both data and models. Use type = "text" for working in R; switch to type = "html" with an out = "Table.htm" argument for final papers.

library(stargazer)
stargazer(as.data.frame(ExampleData), type = "text", title = "Data Summary")

stargazer(Model1, type = "text", title = "Model Summary")

stargazer(Model1, Model2, type = "html", out = "ModelTable1.htm",
          title = "Model 1 and 2 Summary")

library(stargazer)
library(sandwich)

RobustSE <- sqrt(diag(vcovHC(Model1, type = "HC1")))

stargazer(Model1, Model1,
          se = list(NULL, RobustSE),
          type = "text",
          title = "Without and With Robust Standard Errors",
          notes = "Robust standard errors on the right",
          notes.append = TRUE)

A simpler alternative that outputs text-style model tables.

library(jtools)
export_summs(Model1)

library(jtools)
Model1Robust <- summ(Model1, robust = TRUE)

export_summs(Model1, Model1Robust,
             model.names = c("Standard SE", "Robust SE"))

Use lm() to fit an Ordinary Least Squares regression. Two equivalent formats:

Reg1 <- lm(ExampleData$wage ~ ExampleData$educ
                              + ExampleData$exper
                              + ExampleData$female)

Reg1 <- lm(wage ~ educ + exper + female, data = ExampleData)
# Variable names must exactly match the dataset columns

## Coefficients:
##                    Estimate Std. Error t value  Pr(>|t|)
## (Intercept)       -1.73448    0.75362  -2.302   0.0218 *
## ExampleData$educ   0.60258    0.05112  11.788  < 2e-16 ***
## ExampleData$exper  0.06424    0.01040   6.177 1.32e-09 ***
## ExampleData$female -2.15552   0.27031  -7.974 9.74e-15 ***
##
## Multiple R-squared: 0.3093,  Adjusted R-squared: 0.3053

To estimate a log-linear model, first create the log of the variable as a new column, then run lm() as usual.

# Step 1: Create the log variable
ExampleData$Logwage <- log(ExampleData$wage)

# Step 2: Run the regression on the log outcome
Reg2 <- lm(ExampleData$Logwage ~ ExampleData$educ
                                  + ExampleData$exper
                                  + ExampleData$female)
summary(Reg2)

## Coefficients:
##                     Estimate Std. Error t value  Pr(>|t|)
## (Intercept)         0.480836   0.105016   4.579 5.86e-06 ***
## ExampleData$educ    0.091290   0.007123  12.816  < 2e-16 ***
## ExampleData$exper   0.009414   0.001449   6.496 1.93e-10 ***
## ExampleData$female -0.343597   0.037667  -9.122  < 2e-16 ***
##
## Multiple R-squared: 0.3526,  Adjusted R-squared: 0.3488

The ggplot2 package (part of the tidyverse) is far more powerful and flexible for creating publication-quality visualizations. It supports scatter plots with best-fit lines, histograms, violin plots, box-and-whisker plots, line charts, pie charts, and much more.

library(ggplot2)

# Scatter plot with a best-fit line
ggplot(ExampleData, aes(x = educ, y = wage)) +
  geom_point(alpha = 0.4) +
  geom_smooth(method = "lm", se = TRUE) +
  labs(title = "Wage vs Education",
       x = "Years of Education",
       y = "Hourly Wage ($)") +
  theme_minimal()