SIMPLE STATISTICAL PROCEDURES AND GRAPHS

CONTINUOUS VARIABLES: PROC MEANS

PROC MEANS DATA=your.data maxdec=2;    
VAR cont _var;
CLASS cat_var;     
RUN;

• Options for PROC MEANS
  • VAR - to select variables
  • CLASS – to group observations
  • MAXDEC – limits # decimal places

CATEGORICAL VARIABLES: PROC FREQ

PROC FREQ DATA=your.data
TABLES cat _var1  cat _var1  cat _var2 / MEASURES
RUN; 

• Frequency counts and crosstabs
• Options for PROC FREQ
  • TABLES to select variables
  • TABLES x * y to do cross tabs
  • / MEASURES under PROC FREQ for a 2x2 table will give you the Odds Ratio along with its confidence limits
  • / NOCUM – suppresses cumulative frequencies and percentages

CATEGORICAL AND NUMERIC: PROC TABULATE

• PROC TABULATE data = dataset;
• CLASS – categorical grouping variable(s)
• VAR – analysis variables
• TABLES – specifies what table will look like (Page – Row - Column)

• RUN;

• Must have at least a CLASS or a VAR statement, not necessarily both

TABLE statement for PROC TABULATE

• Has its own unique syntax and operators
• Comma – go to new table dimension
• Blank – concatenate table information
• Asterisk – cross, nest, subgroup information

using a space in PROC TABULATE to calculate counts and totals

title ‘counts by location’; 
proc tabulate data=your.data; 
     class location; 
     table location all; 
run; 

• space between “location” and “all” appends results
• Key word ‘all’ returns total
• Only specifying a class variable returns a count

using a comma in PROC TABULATE to create a new dimension

title1 ‘counts by location’; 
title2 ‘by job; 
proc tabulate data=your.data; 
     class location jobcode; 
     table jobcode all , location all; 
run; 

• comma between jobcode and location creates new dimension
• ‘all’ preceded by space creates totals

PROC TABULATE with a subsetting WHERE Statement

proc tabulate data=your.data; 
     where location in (‘zipcode1’ , zipcode2’); 
     class location dx; 
     table dx all , location all; 
run;

• returns a table of locations by diagnoses with total counts in margins

The VAR statement in PROC TABULATE

proc tabulate data=your.data.; 
     where location in (‘zipcode1’ , ‘zipcode2’); 
     class location jobcode; 
     var income; 
     table jobcode, location  income; 
run;

• VAR specifies a continuous variable
• here, rather than return counts (all) return statistics on the income variable for cross categories of location and income
• default statistic is the sum
• if you want a different statistic in the cell, follow the VAR analysis variable with an asterisk and the desired statistic
  • e.g. table jobcode, location income mean;
• available statistics include median, min, max, std (standard deviation) 

SAS GRAPHICS

• flexibility to produce charts in jpeg, gif, etc…
• Save your graphics as .emf files for Word documents
• SAS shows charts in a separate graph window
• need to quit out of graph procedures
• running other procs will exit you from gchart, or can type quit;

Histograms with PROC GCHART

proc gchart data=your.data;
    vbar jobcode / sumvar=salary type=mean;
run;

• VBAR to get vertical bars, HBAR to get horizontal bars
• PIE – request pie chart
• FILL= x  for cross-hatch
• EXPODE =  to pull slice away
• TYPE= - specifies that the height or length of the bar 
• SUMVAR – id variable for simple analyses 

Plot Points with PROC GPLOT

proc gplot data=
    plot var1  * var2
run;

• SYMBOL statement 
  • VALUE= STAR, DIAMOND, SQUARE, TRIANGE
  • I= JOIN, SPLINE
  • C=  W=  color and width
  • RESET= to return to default (SYMBOL global and additive)
    • goptions reset=symbol

Graphic Odds and Ends

• SAS keeps on appending graphs in the graph window
• to get rid of old graphs:
  • open ‘G seg’ folder in the Explorer Window (found in the work library) 
  • delete the graphs you don’t want
  • can also delete the entire Gseg folder to get rid of all the graphs.