Analyze Multiple choice survey questions: many variables in one crosstable

[FelipOH]

Hi:

I am having trouble analyzing Multiple Choice questions of a survey.

Survey Monkey provided me a table with all answers to my questions.

For multiple choice questions it coded each choice into one variable.

When I analyze the data for one question, I get many tables (one for each variable).

I would like to know if it possible to get all "choice" variables in one single table:

With the current code:

Proc freq data=work.cmoj;
Tables (
Principales_act_1
Principales_act_2
Principales_act_3
Principales_act_4
Principales_act_5
Principales_act_6
Principales_act_7
Principales_act_otro) *Attendance_2019/ nocol norow;
Run;

I get one table for each variable, but what I want to do is have one table that includes all choice variable in rows, and the crossed variable choices in columns.

Thank you very much.

[Reeza]

You need to change your data structure. Transpose your data such that you have your data structured such as:


ID Question Response Attendance_2019
1 1 1 0
1 2 5 0
1 3 4 0
1 4 3 0

Then you can use a proc freq. Or you can capture those results and then use PROC TRANSPOSE on the results. Either are options.

[FelipOH]

Hi Reeza:

Thank you very much for answer. I think transposing would't do what I need.

My data structure is in this way:

Attribute__1	Attribute__2	Attribute__3	Attribute__4	Attribute__5	Type_assit
				Socialize	freq
Sessions	Congress				ocas
Sessions	Congress	Certification	Social Service	Socialize	freq
Sessions					ocas
Sessions	Congress	Certification			freq
Sessions	Congress	Certification			ocas
Sessions	Congress	Certification		Socialize	freq

What I want to read is in this way:

	freq	ocas	Total	freq	ocas	Total
Sessions	10	9	19	36%	38%	37%
Congress	7	6	13	25%	25%	25%
Certification	6	3	9	21%	13%	17%
Social Service	2	2	4	7%	8%	8%
Socialize	3	4	7	11%	17%	13%
	28	24	52	100%	100%	100%

Hope this can illustrate my issue, thanks again for your time.

[Reeza]

Transposing would do exactly what you want. Have you tried it?

proc transpose data=have out=long;
by type_assist;
var attribute1-attribute5;
run;

proc freq data=long;
table col1*type_assist;
run;

Or PROC TABULATE for percentages.
This is one approach, there are others. Feel free to choose the solution that meets your needs.

[FelipOH]

Thank you very much for the info!