r-statistics.co by Selva Prabhakaran

About

R for Stata Users: Complete Stata-to-R Reference Translation

Stata and R both excel at statistical analysis, but their syntax is worlds apart. This guide maps every common Stata command to its R equivalent so you can transition without losing productivity.

Stata users love its concise syntax — reg y x1 x2 is hard to beat for brevity. R is more verbose but also more flexible: you get 20,000+ packages, publication-quality graphics, and a full programming language. This guide helps you translate the Stata you know into idiomatic R.

Key Conceptual Differences

Before diving into syntax, understand these fundamental differences:

Concept Stata R
Data in memory One dataset at a time Multiple data frames simultaneously
Variable access Just type the name Use df$varname or with(df, ...)
Missing values . (dot) NA
String delimiter "text" "text" or 'text'
Comments // comment or /* */ # comment
Command terminator Newline Newline (or ;)
Looping forvalues, foreach for, lapply, sapply
Output Results window Console (or R Markdown for reports)

Data Management

Loading and Saving Data

Stata R Equivalent
use "data.dta" haven::read_dta("data.dta")
save "data.dta", replace haven::write_dta(df, "data.dta")
import delimited "data.csv" read.csv("data.csv")
export delimited "data.csv" write.csv(df, "data.csv", row.names = FALSE)
describe str(df)
list in 1/10 head(df, 10)
count nrow(df)
browse View(df) (in RStudio)

Creating and Modifying Variables

Stata R Equivalent
gen newvar = x + y df$newvar <- df$x + df$y
replace x = 0 if x < 0 df$x[df$x < 0] <- 0
gen log_x = log(x) df$log_x <- log(df$x)
gen age_sq = age^2 df$age_sq <- df$age^2
egen mean_x = mean(x), by(group) df$mean_x <- ave(df$x, df$group, FUN = mean)
egen total_x = total(x), by(group) df$total_x <- ave(df$x, df$group, FUN = sum)
egen rank_x = rank(x) df$rank_x <- rank(df$x)
recode x (1/3=1)(4/6=2)(7/10=3) df$x_r <- cut(df$x, c(0,3,6,10), labels=1:3)
encode strvar, gen(numvar) df$numvar <- as.numeric(factor(df$strvar))
decode numvar, gen(strvar) df$strvar <- as.character(df$numvar)
destring x, replace df$x <- as.numeric(df$x)
tostring x, replace df$x <- as.character(df$x)
label variable x "Label" attr(df$x, "label") <- "Label"
drop x y df$x <- NULL; df$y <- NULL
keep x y z df <- df[, c("x", "y", "z")]
rename old new names(df)[names(df) == "old"] <- "new"
# Stata gen/replace equivalents df <- mtcars df$car <- rownames(df) # gen efficiency = mpg / wt df$efficiency <- round(df$mpg / df$wt, 2) # replace efficiency = 0 if efficiency < 5 df$efficiency[df$efficiency < 5] <- 0 # egen mean_mpg = mean(mpg), by(cyl) df$mean_mpg <- ave(df$mpg, df$cyl, FUN = mean) cat("=== First 6 rows ===\n") print(head(df[, c("car", "mpg", "wt", "cyl", "efficiency", "mean_mpg")]))


  






Subsetting and Sorting







Stata
R Equivalent






keep if x > 10
df <- subset(df, x > 10)




drop if missing(x)
df <- df[!is.na(df$x), ]




sort x
df <- df[order(df$x), ]




gsort -x
df <- df[order(-df$x), ]




sort group x
df <- df[order(df$group, df$x), ]




duplicates drop
df <- df[!duplicated(df), ]




duplicates list
df[duplicated(df), ]







Merging and Appending







Stata
R Equivalent






merge 1:1 id using "file2.dta"
merge(df1, df2, by = "id")




merge m:1 id using "lookup.dta"
merge(df1, lookup, by = "id", all.x = TRUE)




append using "file2.dta"
rbind(df1, df2)




reshape wide val, i(id) j(time)
reshape(df, direction="wide", idvar="id", timevar="time")




reshape long val, i(id) j(time)
reshape(df, direction="long", varying=..., idvar="id")









  


    
# Stata merge equivalent
customers <- data.frame(id = 1:5, name = c("Alice","Bob","Carol","Dave","Eve"))
purchases <- data.frame(id = c(1,2,2,5), amount = c(50, 30, 75, 120))

# merge 1:m id using purchases
result <- merge(customers, purchases, by = "id", all.x = TRUE)
cat("=== Left Join (m:1 merge) ===\n")
print(result)


    


      

      

    


    


  






Statistical Analysis



Descriptive Statistics







Stata
R Equivalent






summarize x
summary(df$x)




summarize x, detail
summary(df$x) + sd(), quantile()




tabulate x
table(df$x)




tab x y
table(df$x, df$y)




tab x y, chi2
chisq.test(table(df$x, df$y))




correlate x y z
cor(df[, c("x","y","z")])




pwcorr x y z, sig
psych::corr.test(df[,c("x","y","z")])









  


    
# Stata summarize equivalent
cat("=== summarize mpg, detail ===\n")
x <- mtcars$mpg
stats <- data.frame(
  Stat = c("N","Mean","SD","Min","P25","Median","P75","Max"),
  Value = c(length(x), round(mean(x),2), round(sd(x),2),
            min(x), quantile(x,0.25), median(x), quantile(x,0.75), max(x))
)
print(stats, row.names = FALSE)


    


      

      

    


    


  






Regression







Stata
R Equivalent






reg y x1 x2
lm(y ~ x1 + x2, data = df)




reg y x1 x2, robust
lmtest::coeftest(model, vcov = sandwich::vcovHC)




reg y x1 x2, cluster(id)
lmtest::coeftest(model, vcov = sandwich::vcovCL)




logit y x1 x2
glm(y ~ x1 + x2, family = binomial)




probit y x1 x2
glm(y ~ x1 + x2, family = binomial(link="probit"))




ologit y x1 x2
MASS::polr(factor(y) ~ x1 + x2)




poisson y x1 x2
glm(y ~ x1 + x2, family = poisson)




tobit y x1 x2, ll(0)
AER::tobit(y ~ x1 + x2, left = 0)




predict yhat
fitted(model)




predict resid, residuals
residuals(model)




test x1 = x2
car::linearHypothesis(model, "x1 = x2")




vif
car::vif(model)









  


    
# Stata reg equivalent
cat("=== reg mpg wt hp ===\n")
model <- lm(mpg ~ wt + hp, data = mtcars)
print(summary(model))


    


      

      

    


    


  






T-tests and ANOVA







Stata
R Equivalent






ttest x == 0
t.test(x, mu = 0)




ttest x, by(group)
t.test(x ~ group, data = df)




ttest x == y (paired)
t.test(x, y, paired = TRUE)




oneway y group, tabulate
summary(aov(y ~ factor(group), data = df))




anova y group1 group2
summary(aov(y ~ group1 * group2, data = df))









  


    
# Stata ttest equivalent
cat("=== ttest mpg, by(am) ===\n")
result <- t.test(mpg ~ am, data = mtcars)
cat("t =", round(result$statistic, 3), "\n")
cat("p-value =", round(result$p.value, 4), "\n")
cat("Mean (auto):", round(result$estimate[1], 2), "\n")
cat("Mean (manual):", round(result$estimate[2], 2), "\n")


    


      

      

    


    


  






Panel Data







Stata
R Equivalent






xtset id time
Use plm package: pdata.frame(df, index = c("id","time"))




xtreg y x1, fe
plm(y ~ x1, data = pdf, model = "within")




xtreg y x1, re
plm(y ~ x1, data = pdf, model = "random")




hausman fe re
phtest(fe_model, re_model)









  


    
# Panel data concepts (without plm for WebR compatibility)
cat("=== Panel Data in R ===\n")
cat("For panel/longitudinal data, use the plm package:\n\n")
cat('library(plm)\n')
cat('pdf <- pdata.frame(df, index = c("id", "time"))\n')
cat('fe_model <- plm(y ~ x1 + x2, data = pdf, model = "within")\n')
cat('re_model <- plm(y ~ x1 + x2, data = pdf, model = "random")\n')
cat('phtest(fe_model, re_model)  # Hausman test\n')


    


      

      

    


    


  






Programming Constructs







Stata
R Equivalent






local x = 5
x <- 5




global x = 5
x <<- 5 (avoid this; use function args)




forvalues i = 1/10 { }
for (i in 1:10) { }




foreach v of varlist x y z { }
for (v in c("x","y","z")) { }




while condition { }
while (condition) { }




capture noisily command
tryCatch(expr, error = function(e) ...)




program define name
name <- function(...) { }




display "text"
cat("text\n")




return scalar r = val
return(val) in a function




do "script.do"
source("script.R")







Summary: Stata → R Quick Reference







Category
Stata Command
R Equivalent






Load data
use "file.dta"
haven::read_dta("file.dta")




Describe
describe
str(df)




Create var
gen x = expr
df$x <- expr




Filter
keep if cond
df <- subset(df, cond)




Sort
sort var
df <- df[order(df$var), ]




Merge
merge 1:1 id using file
merge(df1, df2, by = "id")




Summarize
summarize x
summary(df$x)




Tabulate
tab x y
table(df$x, df$y)




Regression
reg y x1 x2
lm(y ~ x1 + x2, data = df)




Logistic
logit y x1 x2
glm(y~x1+x2, family=binomial)




T-test
ttest x, by(g)
t.test(x ~ g, data = df)




ANOVA
oneway y group
aov(y ~ factor(group))







FAQ



Can R read Stata .dta files? Yes. Use haven::read_dta("file.dta"). It supports Stata versions 8 through 18 and preserves variable labels and value labels. Install with install.packages("haven").



How do I get Stata-style robust standard errors in R? Use the sandwich and lmtest packages: lmtest::coeftest(model, vcov = sandwich::vcovHC(model, type = "HC1")). The HC1 type matches Stata's robust option.



Is R slower than Stata for large datasets? For standard operations, R is comparable or faster — especially with data.table for very large data. R's vectorized operations and packages like collapse can outperform Stata on datasets with millions of rows.



What's Next




        


        
      


      

        

        
© 2016- Selva Prabhakaran.
          This work is licensed under the Creative Commons License.