Stata Basic Syntax Structure

command varlist qualifiers, options

command 1, 2, or 3 words specifying the task

varlist 0, 1, 2 or more variables
= exp A math or logical statement to set a value
if exp A logical statement to limit cases
, options Words and selectors to alter the command

Using Stata Help Challenge

Type help command What would these be if Stata syntax
were used in everyday life?
In Stata documentation and here, the following conventions
are used to describe the proper syntax for each command:
[brackets] not needed / optional element order burger, meat(2) cheese(American)
italics replace with your own name(s) or value(s) order burger if (place != "Carls Jr"), nomayo
underline the shortest possible abbreviation order burger if (price < 3), plain mustard
in this document, bold indicates the suggested abbreviation.
Answers
Common Symbols
3. Cheap burger with only mustard
all the usual toppings except mayo
| or == is equal to < less than 2. From any place but Carls Jr, with
1. Double-cheeseburger
& and != not equal to <= less than or equal to

Debby Kermer, Data Services, George Mason University, 2015-01-10

 Page 1: Open and Manipulate Data
File Manipulation Syntax Structure: command varlist qualifiers using file, options
Organize Set working directory cd “directory” This allows “relative” file names
Start a log file log using file Add a .log extension for plain text
Log Results
Save a started log file log close You must close a log file to save it
use “file” Opens file.dta in working directory
Stata (.dta)
Open and save “file” Saves file.dta in working directory
Save Files Delimited File (e.g., csv) import delim using “file.csv” , delim(“,”) Omit option if tab or comma/csv
Excel File import excel using “file.xls” See also “help import”
Use math generate newvar = exp exp is any math expression
Create
Indicator/Binary Variables generate newvar = (x == #) if !missing(x) Dummy Code for #: True = 1, False = 0
Variables
Use Functions egen newvar = function(x) See “help egen” for functions*
Use math replace x = exp exp is any math expression
Modify individual values recode x (1 2=1) (3 = 0), generate(x_r) Generates a categorical variable
Change
Group number ranges recode x (1/4=1) (5/9=2), generate(x_r) Generates a categorical variable
Values
destring x, generate(x_r) When the values are actually numbers
Convert string to numeric
encode x, generate(x_r) Creates labels from string/text values
Remove variables/columns drop / keep x Drop (remove) or Keep [only] the
Delete Data
Remove observations/rows drop / keep if exp variables or observations indicated

Variable Labels label variable x “My label” Labels are for your reference only
Labeling label define numbers 1 “One” 2 “Two” Both commands are needed, but
Value Labels
label values x numbers the defined label can be reused

* the function std makes z-scores

Debby Kermer, Data Services, George Mason University, 2015-01-10

 Page 2: View and Summarize Data
x → the name (not the label) of the variable of interest z → the name of a categorical / grouping variable

Look at Data browse Use edit to allow changes

Open Windows Variables Manager varmanage Can be used to label values
Do Files doedit Do files hold syntax for redo-ing tasks
about the variable describe x Omit the vaname for info about all vars
Display Information
about the values codebook x Add , compact for a table of variables
for Numeric Variables ci x Gives the CI around a mean
Confidence Intervals
for Proportions ci x, binomial Gives the CI around a proportion
summarize x N/Obs., Mean and Standard Deviation
Univarite for a Numeric Variable
summarize x, detail Also Median*, Skewness, Kurtosis, etc.
Descriptive Statistics
for a Categorical Variable tabulate x Frequency table of all values
for 2 Categorical Variables tabulate x1 x2, column row Contingency table with row & col percents
Bivariate
tabulate z, summarize(x) Statistics for x by grouping variable z
Descriptive Statistics for Numeric + Categorical
tabulate z, sum(x) nofreq nost Pivot Table, displays only means by group
histogram x, normal Omit the option for a clean graph
Histogram
Univariate Graphs histogram x, width(#) Specify the width of each bin
Box-and-Whiskers Plot graph box x One numeric variable
Box Plot by Groups graph box x, over(z) Separated by grouping variable z
Bivariate Graphs
Scatter Plot scatter x1 x2 For two numeric variables
any graph x, title(My Title) Appears over the graph
Graph Labeling Add Text Labels
any graph x, note(My Notes) Appears under the graph
Compare Groups “Prefix” Command bysort z : any command Use with any above to do for each value of z
* the median is labeled “50%”

Debby Kermer, Data Services, George Mason University, 2015-01-10

 Page 3: Inferential Statistics
x → Independent Variable (IV), Criterion, or Grouping Variable y → Dependent Variable (DV) or Response Variable
Nominal Chi-Square tabulate x1 x2, chi2 expect expect includes expected values
Kendall’s Rank Correlation Coef. tabulate x1 x2, taub PRE alternative, see also ktau
Examine Ordinal
Associations Spearman’s Rank Correlation spearman x1 x2 Comparable to Pearson’s Correlation
Uses pairwise deletion by default
Numeric Pearson’s Correlation Coefficient pwcorr x1 x2, sig listwise
(listwise does the same as correlate)
Compare a statistic One-Sample t-Test ttest x == # For a mean or a mean difference
to a specific value: # One-Sample Proportions Test prtest x == # For a proportion or a proportion difference
Test of Equal Variances sdtest x1 == x2 If significant, add unequal to ttest
Compare specific Paired t-Test ttest x1 == x2 Compares paired numeric values
values to each other
(e.g., before vs after) Paired Proportions Test prtest x1 == x2 Compares paired proportions
Wilcoxon Paired Sign-Rank Test signrank x1 == x2 Compares paired ranks (non-parametric)
Test of Equal Variances robvar y, by(x) If significant, add unequal to ttest
Independent t-Test ttest y, by(x) Compares 2 means
2 Groups
Independent Proportions Test prtest y, by(x) Compares 2 proportions
Compare
Wilcoxon Rank-Sum Test ranksum y, by(x) Compares 2 ranks (non-parametric)
Groups
Kruskal-Wallis Rank Test kwallis y, by(x) Compares 3+ ranks (non-parametric)
3+ Groups oneway yx Compares 3+ means (1 IV)
ANalysis Of VAriance
anova y x1 x2 … Compare 2+ means (2+ IVs)
[Multiple] Linear Regression regress y x1 x2 … Numeric y, Numeric or binary/dummy x’s
Numeric
Predict Plot Residuals vs Fitted Values rvfplot Check homoskedasticity after running reg
Values logit y x1 x2 … Reports coefficients (log-odds)
Binary Logistic Regression
logistic y x1 x2 … Reports odds-ratios

Debby Kermer, Data Services, George Mason University, 2015-01-10