Proc Freq Procedure.

[vishalrajpoot3]

I have below table with variables and their labels and there are multiple observations against each value/labels.

Variable	Value	Label
Pre_1	1	Use every day
Pre_1	2	Use some days
Pre_1	3	Don't use, but have tried in the past
Pre_1	4	Heard a lot & never tried
Pre_1	5	Heard a little & never tried
Pre_1	6	Never heard of it
Pre_2	1	Use every day
Pre_2	2	Use some days
Pre_2	3	Don't use, but have tried in the past
Pre_2	4	Heard a lot & never tried
Pre_2	5	Heard a little & never tried
Pre_2	6	Never heard of it

How can I get below output using proc freq:

Q225
pre_1_2	Frequency	Percent	Cumulative Frequency	Cumulative Percent
Use every day	30	6.32	30	6.32
Use some days	98	20.63	128	26.95
Not at all	347	73.05	475	100.00

[PaigeMiller]

Please clarify this part:

I have below table with variables and their labels

Is the column named LABEL the name of an actual variable named LABEL, or is it an actual label for a variable?

[vishalrajpoot3]

 This is label for the variable.

[PaigeMiller]

Okay, so show us a portion of the actual data.

[vishalrajpoot3]

@PaigeMiller The table I have pasted above is actual data and below one is the output of proc freq. In output both the variable (pre_1 and pre_2) grouped as pre_1_2 and last 4 value of both the variables grouped as "Not at all".

[PaigeMiller]

You can't get the PROC FREQ output from the input data set without re-arranging the data. There is no variable Pre_1_2 in your original data set, but there is in the output data set. And because of this, it isn't even clear to me how you derive the output table if you were to do this by pencil and paper. Please explain the steps between the input data and the output table, step-by-step, so I can see what you are doing.

[vishalrajpoot3]

please ignore the previous information provided.

So, I have a data set abc and it has two variable pre_1 and pre_2 with values from 1 to 6. All the 6 values has format as in below output table :

Pre_1	Pre_2
Use some days	Heard a little & never tried
Use every day	Use every day
Never heard of it	Don't use, but have tried in the past
Heard a lot & never tried	Heard a lot & never tried
Heard a little & never tried	Never heard of it
Don't use, but have tried in the past	Use some days

so, using proc freq on above dataset can I get below output:

pre_1_2	Frequency	Percent	Cumulative Frequency	Cumulative Percent
Use every day	30	6.32	30	6.32
Use some days	98	20.63	128	26.95
Not at all	347	73.05	475	100.00

[Tom]

Put the data into ONE variable. Then run PROC FREQ.

data tall;
  set have;
  pre_1_2 = pre_1;
  output;
  pre_1_2=pre_2;
  output;
run;
proc freq data=tall;
 tables pre_1_2;
run;

[vishalrajpoot3]

will it not increase the no of observation?

[Tom]

Yes. It will double the number of observations. Which is what you need to be able to produce the table you showed.

If you wanted to produce such a table without increasing the total N you would need to classify each individual observation into just one category. I could be done but you would need to decide what the rules are.

[vishalrajpoot3]

for both the variables, rules are like: value 1 (use every day) and 2 (use some days) should be displayed as it is and rest all values from 3 to 6 should be clubbed in single category "Not at all".
can you please help me with the code?

[Reeza]

Your problem as I understand it:

• 2 variables with the same labels/values
• Would like a summary of the values across both columns in one summary table
• With labels applied, not raw values

If that's correct the solution is to:

• Use PROC FORMAT to create a format for the variable, CNTLIN is a good option here
• Combine the two variables into one (PROC TRANSPOSE or data step) 
• Summarize

Or you can do it in in a slightly different order:

• Use PROC FORMAT to create a format for the variable, CNTLIN is a good option here
• Summarize and save to data set
• Combine output data sets via PROC APPEND
• Re-summarize

If you need help with code, please provide sample data in the form of a data step. 

---

@vishalrajpoot3 wrote:

I have below table with variables and their labels and there are multiple observations against each value/labels.

Variable	Value	Label
Pre_1	1	Use every day
Pre_1	2	Use some days
Pre_1	3	Don't use, but have tried in the past
Pre_1	4	Heard a lot & never tried
Pre_1	5	Heard a little & never tried
Pre_1	6	Never heard of it
Pre_2	1	Use every day
Pre_2	2	Use some days
Pre_2	3	Don't use, but have tried in the past
Pre_2	4	Heard a lot & never tried
Pre_2	5	Heard a little & never tried
Pre_2	6	Never heard of it

How can I get below output using proc freq:

Q225
pre_1_2	Frequency	Percent	Cumulative Frequency	Cumulative Percent
Use every day	30	6.32	30	6.32
Use some days	98	20.63	128	26.95
Not at all	347	73.05	475	100.00

---