Stata Seminar

1
Stata Seminar

• Session 2 Syntax
• Francisco Jose Gonzalez Carreras
• fjg23_at_sussex.ac.uk

2
First to read an excel file

• Download the file and save.
• Open the excel file and save is as Text (tab
  delimited)
• Open stata and go to the directory where the new
  text file is. Remember cd f\ (for example)
• (In this case with data1) type set memory 5m (we
  are increasing the memory because the file is big
  and otherwise it would give us an error)
• 5 type insheet using data1.txt, clear
• Type save data1saved and is already a .dta file
• See that we lost labels, it is better to download
  from the original source.

3
Elements of Stata Commands

• How to use stata commands. Elements can be
• required
• permitted
• prohibited
• Type help summarize
• summarize varlist if in weight ,
  options
• Command summarize
• varlist means variable list
• if stands for the if qualifier (if gender1).
  Qualifiers restrict the command to a particular
  subsample of the database
• , options, it specifies the command in a more
  particular way.
• Start session and load data1 (big brother and
  use)
• log using session2.log
• use data1, clear

4
Syntax commands

• Commands can be abbreviated, see that in the help
  some letters of the command are underlined these
  are the shortest possible abbreviation of the
  command.
• This means that summarize can be written as su,
  but also sum summ summa
• Type su income and then sum income it is exactly
  the same

5
Syntax Commands
6
Syntax Variable list varlist

• varlist means that you can use a variable list,
  that you enter by writing the name of the
  variables, separated by spaces
• varlist in square brackets the variable list is
  possible but not required. summarize
• varlist with no square brackets, it is required
  if you do not write it the program will report an
  error.
• Type help drop. Typing just drop gives an error.
• After that type drop _all. All are dropped.
• We cannot reverse this command, so we have to
  upload the file again use data1, clear

7
Syntax Variable list varlist

• You can also see varname or depvar with no
  brackets. This is the case for variable lists
  that consist of one single variable. Sometimes it
  is a variable within a list of variables where
  the order is important
• Type help regress you see depvar (for dependent
  variable) and then indepvars that you can enter
  or not
• Remember session 1 regress income sex fulltime

8
Syntax Variable list varlistAbbreviation rules

• Single variables
• Abbreviate the name of the only variables that
  begin with a letter kitchen with k
• sum k summarizes kitchen
• save typing characters of a variable
• sum yh summarizes ybirth and saves you birt
• Multiple variables
• Use ? for variables with the same name except one
  character sum np940?
• Use asterisk to specify variables that share
  characters in their name sum np summarizes all
  the variables that begin with np
• Hyphen to specify a range of variables that
  should be in order.
• sum kitchen-phone is equal to write sum kitchen
  phone shower wc heating cellar balcony garden
  phone
• Lets mix resources sum rs np k-ph summarizes
  a lot of variables!!!

9
Syntax Options , options

• Commands have a default execution and options
  modify it. They are different for each command
  and are possible when you can see the word option
  after a comma.
• Different options are described below.
• We did this in the first session. Type
• summarize income, detail
• See also syntax of tabulate. Type
• tabulate gender np9506, missing row

10
Qualifiers in (order)

• The in qualifier limits the execution of the
  command to a subset of observation. ORDER
• It is composed by the word in and a range of
  observations separated by /(slash). Before the
  slash will be the first observation for which the
  command will be executed and after the / will be
  the last FIRST/LAST. If the range is a single
  observation, one its number is enough.
• Remember that it might be very important to sort
  before the data to make sure that you execute the
  commands for those observations that you want.
• Examples
• list persnr gender ybirth in 10 (only the tenth
  observation)
• list persnr gender ybirth in 10/14 (tenth to
  fourteenth)
• list persnr gender ybirth in -5/-1 (fifth from
  the last / the last)
• list persnr gender ybirth in -5/l (fifth from the
  last / the last)
• list persnr gender ybirth in 3330/-5 (3330th /
  5th from the last)

11
Qualifiers if (condition)

• Restricts the execution of the command to those
  observations that meet a particular condition,
  that has to follow the qualifier.
• We did sum income if gender 1. See it here
• list income gender in 1/5, nolabel
• list income gender in 1/5 if gender1, nolabel
  See how does if work??
• Try these others
• sum income if ybirth lt 1979
• sum income if ybirth lt 1979
• sum income if ybirth 1979
• Careful with the infinite mistake the missing
  trap!!!!!
• tab edu, missing nolabel
• sum ybirth if edugt6 see that 28 missings are
  added

12
Qualifiers if Relational operators

• Practice
• sum ybirth if edu6 edu7
• sum ybirth if edugt6 edu lt7
• sum ybirth if edugt6 edult.

13
Expressions, operators

• Allowed or required when the term exp appears in
  the syntax diagram. Type help generate This
  command needs an expression after the command
  name.
• Stata calculator display. Type
• display 22
• Operators , , /, type
• display 32
• display 23 22

14
Lists of numbers
15
Using filenames

• In stata some commands read or write a file. In
  the syntax is expressed with using filename
• Normally it consist of a directory, the name of
  the file itself and the extension.
• F\data1.dta If you type a name stata looks for
  in the current directory (bottom left hand
  corner). If it is not here, it will report it.
• If you type a filename without extension, Stata
  looks for one with an extension that is
  appropriate for the specified
• Extensions and commands in the table.

16
Repeating similar commands by prefix

• Sometimes you will need to type similar commands
  over and over again. There are two main options
  to do this
• by prefix already known, remember it executes
  the commands by the batches determined by the
  prefix command. Type
• sort gender
• by gender sum income
• by edu, sort summarize income
• Play with the by prefix
• bysort edu summarize income
• by gender edu, sort sum income (the same in
  only one command, adding one variable)

17
Repeating similar commands foreach or forvalues
loops

• Syntax DO NOT COPY-PASTE THE LOOPS, errors will
  appear because of the formats.
• foreach lname listtype list
• commands referring to lname
• The first line starts the loop and ends into a
• Then add stata commands
• Close the loop
• The element name (lname), the list type
  (listtype), and the foreach list (list)
• You state the name of the element, the list of
  parameters you want to execute the commands on,
  then close, then the command (s) and close the .
  This example a list of variables
• foreach x of varlist np9501 np9504
• tabulate x gender

18
Repeating similar commands foreach

• Other examples of foreach
• foreach var of newlist r1-r10
• gen varuniform()
• List of new variables. Uniform() creates
  uniformly distributed random number.
• foreach num of numlist 1/10
• replace rnumuniform()
• List of numbers. Replace because the variables
  already exist
• Practice lets label all the variables r1-r10
  writing a common label 1st uniform variable, 2nd
  uniform variable First, second, third typing,
  the rest with a loop.

19
Repeating similar commands forvalues

• It has a simplified syntax
• forvalues lnamerange
• commands
• forvalues num1/10
• replace rnumuniform()
• Practice replace the label of variables with a
  forvalues loop instead.

20
Repositories

• Stata saves results of statistical commands in
  r() and of estimation commands in e(). They are
  called repositories.
• Statistical summarize. Type
• summarize income and then return list and you
  will see the contents of the last r-class command
• Estimation regress. Type
• regress income yedu and then ereturn list
• You can operate with them. Type
• sum income
• display r(mean) 1.96sqrt(r(Var)/r(N))
• display r(mean) - 1.96sqrt(r(Var)/r(N))
• Stored results are deleted with a new command.
  Some commands store results into matrices but we
  will not see them

21

• Unplug big brother
• log close
• doedit
• Copy the commands in the review window and paste
  in the doedit. Save as session2
• clear
• exit