Overview

SAS can read data in many formats, including text (ASCII) files, SPSS portable files, database files, and Excel spreadsheets. The formats allowed vary across operating systems.

Once the data are read by the SAS program, they are converted into a SAS dataset which contains information about the names, values, formats, and labels of the variables. The data set may be stored temporarily, for the duration of the program, or it may be stored permanently for re-use in other SAS programs.

The first section of this document shows how to prepare a raw (text) data set. The second section explains how to create a permanent SAS data set.

Raw Data

• The raw data file contains the values of variables in text (ASCII) format. If the data are to be read in with a SAS data step, the raw data file contains only the data. If PROC IMPORT is used to read raw data, the first line in the data file may contain the names of the variables to be read on subsequent lines.
• Data may be read inline (in the body of the SAS comand file) or from an external data file. If you plan to write several programs to analyze the data, or if the data file is more than a few lines, the data should be stored in an external file.
• The values for each unit of analysis (subject) appear on one or more (contiguous) lines.
• If possible, the data should be prepared in fixed format, with the values of each variable occupying the same column for each unit of analysis (e.g., AGE appears in columns 23-24). Spaces may be used to separate the values of the variables, but it is not required.
• If data are not read in fixed format, the values of the variables must be separated with a delimiter. The most commonly used form of delimiter is the space, but other symbols, such as the comma, ampersand (&), or tab may be used.
• Missing data may be left blank if the columns or format are specified in the data step. If data are read using freefield input (scanning for blanks), missing values may be indicated with a decimal point.
• The data file may be created using a word processor, spreadsheet, or text editor, but it must be output in text format.

SAS Data Sets

• A SAS Data Set consists of the variable names, formats, labels, and data values stored in machine-readable format. The Data Set may be created from a raw data set in a SAS data step (see the SAS syntax page ) or in a SAS PROC such as PROC REG.
• SAS data sets may be temporary (existing for the duration of the SAS run) or permanent (stored on tape or disk for later use).

How to Create SAS Data Sets in a Data step

• The SAS data step begins with the phrase DATA libref.dsn where libref is the location where the data set will reside (e.g., ~/data). If the libref is omitted, then the data set is assumed to be temporary (example: DATA STOCKS ; would create a file called stocks.sas7bdat for the duration of the execution of the command file).
• If the libref is omitted, SAS assumes that the system libref WORK is used; this is a reserved word and should not be used as a libref for a permanent SAS data set.
• If a libref is present, then a LIBNAME statement must precede the data step. Example:

      libname mydat '~/data/ ;
      data mydat.stocks ;
  
  

  This will create a permanent SAS data set called ~/data/stocks.sas7bdat.
• In version 8 of SAS, the libname and libref may be omitted by direct reference to the filename in quotes.

      data "~/data/stocks" ;
  
  

  will create a permanent SAS data set called ~/data/stocks.sas7bdat. Note that the file extension is not needed.

For examples of SAS programs using permanent SAS Data Sets, please refer to the Sample SAS Command Files Page .