List and summary reports are often created from SAS data. To produce a variety of reports using a single report-writing tool, you can use PROC REPORT. In addition to creating list reports, PROC REPORT enables you to

You can use PROC REPORT in three ways:

This chapter shows you how to use PROC REPORT by submitting SAS statements. Although PROC REPORT enables you to create highly customized reports, we'll focus on basic statements and will add a few enhancements. By the end of the chapter, you'll create a list report and a summary report.

In this chapter, you learn to

In this program, PROC REPORT reads the Flights.Europe data set and creates a report in nonwindowing mode:

proc report data=flights.europe nowd;
run;

This is HTML output from the PROC REPORT step. Notice that by default

The following COLUMN statement specifies that only the variables Flight, Orig, Dest, Mail, Freight, and Revenue be printed, in that order.

proc report data=flights.europe nowd;
   column flight orig dest mail freight revenue;
run;

The following WHERE statement specifies that only observations that have the value LON or PAR for the variable Dest be printed.

proc report data=flights.europe nowd;
   column flight orig dest mail freight revenue;
   where dest in ('LON','PAR'),
run;

In the output that you've seen in this chapter so far, you might have noticed that PROC REPORT displays

You can enhance the report by

Let's see how to define variables.

To describe how to use and display variables in your report, you use one or more DEFINE statements. You can list DEFINE statements in any order, and you can list options (usages, attributes, and so on) in any order in a DEFINE statement.

These DEFINE statements specify usages, attributes, options, justification, and column headings for the variables Flight and Orig. In this chapter, you'll look at each of these ways of defining variables.

proc report data=flights.europe nowd;
   where dest in ('LON','PAR'),

   column flight orig dest mail freight revenue;
   define flight / order descending 'Flight Number' 
                center width=6 spacing=5; 
   define orig / 'Flight Origin' center width=6;
run;

Let's start with attributes.

In previous chapters, you learned that formats determine how data values appear in SAS output. If you do not specify a format for a variable within the PROC REPORT step, PROC REPORT uses the format that is stored in the data set. If no format is stored in the data set, PROC REPORT uses the default format for that variable type.

To assign a format to a specific report column, use the FORMAT= attribute in the DEFINE statement for that column. You can specify any appropriate SAS format or user-defined format.

The variable Revenue has no format assigned to it in the Flights.Europe data set. So, in your current report, revenue values appear as shown in the example below.

But suppose you want your revenue figures to be formatted with a dollar sign, commas, and two decimal places, in a total width of 15 positions. To do this, you can assign the DOLLAR15.2 format to Revenue.

proc report data=flights.europe nowd;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
run;

This is part of the HTML output from the program above. Notice that the format supplies the dollar sign, comma, decimal point, and decimal places. However, because the HTML table column conforms to the width of its contents, assigning the format does not increase the column width beyond the length of the data values.

By contrast, the monospace SAS listing of the report does display the increased column width.

In the previous example of SAS listing output, you might have noticed that several headings were wrapped over two lines. (In HTML output, the longest cell value determines the column width, so wrapping doesn't occur.)

If a variable in the input data set doesn't have a format associated with it, the default PROC REPORT column width is

The character variables Flight, Orig, and Dest each have a length of 3, and no format is associated with them. So 3 is their default column width.

To specify a width for columns in your report, use the WIDTH= attribute in the DEFINE statement. You can specify values from 1 to the value of the LINESIZE= system option.

The WIDTH= attribute has no effect on HTML output.

To specify column widths that accommodate the column headings for Flight, Orig, and Dest, you can use the following DEFINE statements in your PROC REPORT step:

proc report data=flights.europe nowd;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight / width=6; 
   define orig / width=4; 
   define dest / width=4;
run;

Now the headings appear on one line.

Another way to enhance your PROC REPORT output is to specify column spacing, which is the number of blank characters between the selected column and the column immediately to its left. The default column spacing is 2. To specify a different column spacing, use the SPACING= attribute in the DEFINE statement.

The SPACING= attribute has no effect on HTML output.

This is PROC REPORT output without any spacing defined. Orig and Dest have 2 blank characters preceding their columns.

To specify 5 blank spaces before the column headings for Orig and Dest, you can use DEFINE statements as shown below in your PROC REPORT step:

proc report data=flights.europe nowd;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight / width=6;
   define orig / width=spacing=5;
   define dest / width=4 spacing=5;
run;

Now the two columns display the extra spacing.

Suppose you want to label Flight as Flight Number, Orig as Flight Origin and Dest as Flight Destination. Add these column headings to the DEFINE statements, being sure to match quotation marks. You can also change the WIDTH= specifications to accommodate the new headings.

proc report data=flights.europe nowd;
   where dest in ('LON','PAR'),

   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight / width=13 'Flight Number';
   define orig /  width=13  spacing=5 'Flight Origin';
   define dest /  width=18  spacing=5 'Flight Destination';
run;

This is HTML output from the same program:

Now you have the column headings that you want. But the columns are quite wide, aren't they? Let's see what you can do to fix this.

To control how words break in column headings, you can use a split character in the column label. When PROC REPORT encounters the split character in a column heading, it breaks the heading and continues the heading on the next line. The split character itself does not appear in the heading.

To use a split character, you can do either of the following:

Suppose you want to break headings so that only one word appears on a line. Using the default slash as the split character, you can submit this PROC REPORT step. Notice that the column width has been reduced.

proc report data=flights.europe nowd;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight /  width=6 'Flight/Number';
   define orig /  width=6 spacing=5 'Flight/Origin';
   define dest /  width=11 spacing=5  'Flight/Destination';

run;

Or you can submit this program, which uses the SPLIT= option and produces exactly the same output:

proc report data=flights.europe nowd split='*';
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight / width=6 'Flight*Number';
   define orig /  width=6 spacing=5  'Flight*Origin';
   define dest /  width=11 spacing=5  'Flight*Destination';
run;

Here are both types of output from both programs:

To center headings and values for Flight, Orig, and Dest, you can specify the CENTER option as shown below in your DEFINE statements:

proc report data=flights.europe nowd;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight / width=6 'Flight/Number' center;

   define orig / width=6 spacing=5 'Flight/Origin' center;
   define dest / width=11 spacing=5 'Flight/Destination' center;
run;

In this PROC REPORT step, the PROC REPORT statement specifies both HEADLINE and HEADSKIP.

proc report data=flights.europe nowd headline headskip;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight / width=6 'Flight/Number' center;
   define orig / width=6 spacing=5 'Flight/Origin' center;
   define dest / width=11 spacing=5 'Flight/Destination' center;
run;

In the SAS listing output, the column headings are underlined and are followed by a blank line.

PROC REPORT uses each variable in one of six ways (DISPLAY, ORDER, GROUP, ACROSS, ANALYSIS, or COMPUTED). By default, PROC REPORT uses

Because you haven't explicitly defined any variable usages, your current list report contains only display and analysis variables:

In the illustration below, columns for display variables are shown in white. Columns for analysis variables are shown in gray.

How you use a variable in a report determines, among other things, the order of the rows in your report. Let's see the effect of defining a variable as an order variable.

An order variable orders the detail rows in a report according to their formatted values. For example, suppose you want to see values in your list report ordered by flight number. To use Flight as an order variable, you specify the ORDER usage option in the DEFINE statement, as shown below.

proc report data=flights.europe nowd headline headskip;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;

   define flight /  order 'Flight/Number' width=6 center;
   define orig / width=6 spacing=5 'Flight/Origin' center;
   define dest / width=11 spacing=5 'Flight/Destination'   center;
run;

This is your ordered output. Notice that PROC REPORT displays only the first occurrence of each value of an order variable in a set of rows that have the same value for all order variables.

By default, the order is ascending, but you can change it with the DESCENDING option in the DEFINE statement:

proc report data=flights.europe nowd headline headskip;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight / order  descending 'Flight/Number' width=6   center;

   define orig / width=6 spacing=5 'Flight/Origin' center;
   define dest / width=11 spacing=5 'Flight/Destination'   center;
run;

Your list report is complete. But suppose you now want to create a summary report. That is, rather than a list of the mail, freight, and revenue for each flight, you want the total mail, freight, and revenue by flight number. Here is the summary report that you want to create:

To summarize your data using PROC REPORT, you can define one or more group variables. A group variable groups the detail rows in a report according to their formatted values. If a report contains one or more group variables, PROC REPORT consolidates into one row all observations from the data set that have a unique combination of values for all group variables.

To define a group variable, you specify the GROUP usage option in the DEFINE statement. Let's see how group variables affect your report.

If you submit the following PROC REPORT step, with Flight defined as a group variable, you get the output shown below.

proc report data=flights.europe nowd headline headskip;
   where dest in ('LON','PAR'),
   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight /  group 'Flight/Number' width=6 center;
   define orig / width=6 spacing=5 'Flight/Origin' center;
   define dest / width=11 spacing=5 'Flight/Destination' center;
run;

But this output looks exactly like the list output in which Flight was an order variable. What happened?

The problem with the preceding output is that your report contains display variables. As character variables, Orig and Dest are defined as display variables by default.

All of the variables in a summary report must be defined as group, analysis, across, or computed variables. This is because PROC REPORT must be able to summarize all variables across an observation in order to collapse observations. If PROC REPORT can't create groups, it displays group variables as order variables.

To group data in your report, you need to define the character variables (Flight, Orig, and Dest) as group variables, as shown below:

proc report data=flights.europe nowd headline headskip;
   where dest in ('LON','PAR'),

   column flight orig dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight /  group  'Flight/Number' width=6 center;
   define orig /  group  width=6 spacing=5 'Flight/Origin'
                center;
   define dest /  group width=11 spacing=5
                'Flight/Destination' center;
run;

Now PROC REPORT can create groups, and your summary report displays the total mail, freight, and revenue by flight number. Remember that the default statistic for the analysis variables is SUM.

The following table compares the effects of using order variables and group variables.

As you saw in the previous example and practice, the default statistic for analysis variables is SUM. But you might want to display other statistics in your PROC REPORT output. To associate a statistic with an analysis variable, specify it as an attribute in the DEFINE statement.

Here's the previous sample output, which displays the default statistic SUM for the three analysis variables:

By specifying MEAN in the DEFINE statement for Revenue, you can display the average revenue for each flight number. The optional column heading Average Revenue clarifies that the MEAN statistic is displayed.

proc report data=flights.europe nowd headline headskip;
   where dest in ('LON','PAR'),

   column flight orig dest mail freight revenue;
   define revenue /  mean format=dollar15.2
                 'Average/Revenue';
   define flight / group 'Flight/Number' width=6 center;
   define orig / group width=6 spacing=5 'Flight/Origin'
                center;
   define dest / group width=11 spacing=5
               'Flight/Destination' center;
run;

You can use the following statistics in PROC REPORT:

So far, we've looked at display, analysis, order, and group variables. You can also define variables as across variables, which are functionally similar to group variables. However, PROC REPORT displays the groups that it creates for an across variable horizontally rather than vertically.

Let's look at an example of across variables to clarify this usage.

The following program uses group variables to produce the output shown. The table shows unique combinations of values of the group variables, and sums of each analysis variable for each combination.

proc report data=flights.europe nowd headline headskip;
   where dest in ('LON','PAR'),
   column flight dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight /  group 'Flight/Number' width=6 center;
   define dest /  group width=11 spacing=5
                 'Flight/Destination' center;
run;

Now let's suppose that you change the group variables to across variables, as in this program.

proc report data=flights.europe nowd headline headskip;
   where dest in ('LON','PAR'),
   column flight dest mail freight revenue;
   define revenue / format=dollar15.2;
   define flight /  across 'Flight/Number' width=6 center;
   define dest /  across width=11 spacing=5
                'Flight/Destination' center;
run;

In this case, for each across variable, the table cells contain a frequency count for each unique value. For each analysis variable, the table cells represent the sum of all the variable's values.

The last type of variable usage is reserved for computed variables, which are numeric or character variables that you define for the report. They are not in the input data set, and PROC REPORT doesn't add them to the input data set. You can't change the usage of a computed variable.

In the nonwindowing environment, you add a computed variable as follows:

Let's see how you create a new variable for your report.

Suppose you want to determine the number of empty seats for each flight. To do so, you can compute the variable EmptySeats by subtracting the number of passengers deplaning (Deplaned) from the plane's total number of seats (Capacity), assuming that the plane was full at the beginning of the flight.

In the following program, you

  proc report data=flights.europe nowd;
    where dest in ('LON','PAR'),
    column flight capacity deplaned  emptyseats;
    define flight / width=6;
    define emptyseats / computed 'Empty Seats'; 
    compute emptyseats; 
        emptyseats=capacity.sum-deplaned.sum; 
    endcomp;
run;

The program creates the following output.

PROC REPORT < DATA=SAS-data-set> <options>;

         COLUMN variable(s);

         DEFINE variable /< usage> <attributes(s)><option(s)>

                  <justification> <'column-heading'>;

         COMPUTE computed-variable;

              DATA step statements;

         ENDCOMP;

RUN;

proc report data=clinic.diabetes nowd;
   column sex weight fastgluc postgluc glucrange;
   where age>40;
   define weight / format=comma6.2 spacing=4
                   'Average/Weight' width=7;
   define sex / order width=7 spacing=4 center
                'Sex of/Patient';
   define fastgluc / 'Fasting/Glucose';
   define postgluc / 'Postprandial/Glucose'
                     width=12;
   define glucrange / computed 'Glucose/Range';
   compute glucrange;
      glucrange=postgluc.sum-fastgluc.sum;
   endcomp;
run;