Variables missing SAS error code when using letters to denote treatment in ANOVA

Question

I have been having this problem with SAS all year and I can fix it, but it ends up being time-consuming.

Example: When I use the code attached (a simple ANOVA code) I get errors that say I have invalid data for the treatment code. This seems to be because I have denoted the different treatments using A, B, C, D, and E. I can fix this by switching the letters to numbers but I feel like it should be able to run with the letters.

The specific error is: ERROR: One or more variables are missing or freq or weight is zero on every observation

SAS Code

The code I am using (pictured with proper format in picture) is: Data ANOVA3; input treat val; cards; C 96 B 69 D 65 E 34 A 89 C 90 D 74 E 49 B 58 D 41 C 60 A 90 A 106 B 95 E 41 D 54 C 57 A 99 D 35 E 18 C 62 B 78 B 96 E 24 A 71 ; run; title "problem3"; proc glm; class treat; model val= treat; lsmeans treat / stderr pdiff; run;

If anyone could please tell me how to run code without changing every data point into a number that would be great. This also seems to happen to me (only sometimes!) with M and F for male and female, but it used to work so I know I must be forgetting something.

Thanks

Answer 1

If you look at the SAS log, you should see a lot of errors from the DATA step that reads in the data. When you read in a character value, you need to use an informat to tell SAS that you are reading a character value, not a numeric value.

Below I use the $1 informat:

Data ANOVA3; 
input treat $1. val; 
cards; 
C 96 
B 69
D 65
;

* Check data;
proc print data=ANOVA3;
run;

By the way, to format code you can paste it in then highlight the code, and click the {} button on the text formatting menu. It makes the code readable, separate from text. Or manually, you can place four spaces at the front of a line. The HELP for formatting text is: https://stackoverflow.com/help/formatting

Answer 2

A, B, C, D and E and not valid numeric values. But if you have included those letters in the MISSING statement then they will be read as valid representations of the special missing values .A, .B, .C, .D and .E .

If you want TREAT to be a character variable either DEFINE it before the INPUT statement.

data ANOVA3;
   length treat $1 val 8 ;
   input treat val;

Or use a CHARACTER informat with it in the INPUT statement so that SAS will change its GUESS of the type of variable you wanted.

Make sure to include the : modifier so the input statement still reads the next value on the line instead of a fixed number of bytes.

data ANOVA3;
   input treat :$1. val;