How do I compute a variable that aggregate mode?

kariover

New Member
I am asking for some advice on a data management procedure that would allow me to perform a function that would aggregate variables and return mode values. Aggregate command in SPSS does not allow to aggregate by mode offering options to aggregate by mean, sum, etc.

I'm using a Likert scale from 1 to 7. I want to find a variable that shows the total mode for var1, var2, var3 and var4 combined from each respondent.

Does anyone know if there is a roundabout way to compute a variable that would return mode by break variable? I am a noob, so feel free to explain in detail. (I do not know how to write commands manually)

Thank you so much!

Berley

Member
I don't understand the question. Did each respondent give more than one answer for var1, var2, etc.?

Do you want the mode to all questions by respondent? Or do you want the mode for each question?

kariover

New Member
Hey,

thank you for responding.

Each variable is related to a simple statement / question where respondents have answered from 1 to 7. I need a variable to compute mode for several variables. I try to explain with an example:

Var1 Var2 Var3 Var4 Created mode Var5 Var6 Var7 Var8 Created mode
"Respondent 1" 4 5 6 5 *computed* 4 7 4 6 *computed*
"Respondent 2" 3 4 4 6 *computed* 3 3 3 2 *computed*

I will do this for 200 respondent and each respondent has 7 groups of variables that I need mode. I can do it manually but it is a tremendous work. Is there a way to compute the variable?

Thank you again!

Berley

Member
So you want to create a new variable that is the mode of a single respondent's answers, yes?

I don't know of a way to crerate a new variable with that calculation. There could be a way to do it in syntax, but that's not something I can help with.

If you just need to know the modes themselves and can make do with a report, then here's a way to do it:

Data > Transpose
Select each of the questions variables and move to the Variable Box
Click OK

You now have a dataset that has each question as a row and each respondent as a column

Analyze > Descriptive Statistics > Frequencies
Select all 200 respondent variables (you can shift-click to do the whole list)
Click the arrow to move them to the Variable box
Click Statistics
Select Mode and click Continue
Optional: uncheck "Display frequency tables"
Click OK

You should now have a report with columns showing each respondent, how many questions they answered, how many questions they skipped, and the Mode of their responses.

Next, right-click the table and select Edit Content > In separate window
Pivot > Transpose Rows and Columns

You now have a table that has been flipped and is easier to read: respondents in rows and mode is in the right-hand column.

Hope that helps!