[Select cases] - command that shows me only cases that have ">= X missing values"

#1
Dear community,

That is my first post and I try my best to make it a smart one. I have used multiple sources for help already, however my problem is still unsolved, that is why I am writing the post.

Currently I am working on a missing value analysis with the goal to use a multiple imputation approach. Before I do that, I want to filter cases that have more than 5 missing values.

In SPSS 18, I do not know the correct command to do the mentioned case selection. I am sure it has something to do with the option "Select cases" and "if" but my trail & error approach is rather insufficient. Your help would be very much appreciated.

Thanks in advance and greetings from Berlin, Germany.
 
#2
Run the following from a new syntax window, inserting your variable names.

Code:
compute nummiss = nmissing(VAR1 to VARX). /** if variables are not contiguous, list separated by comma e.g. - nmissing(var1, var5, var7,...varX).
compute flag = (nummiss le 5).
filter by flag.
exe.
 
#3
Hi helicon,

Thanks for your quick reply. Unfortunately, the syntax did not work accordingly. I entered the following commands (see screenshot "syntax"). Further, I'd like to have the filtered cases in a separate file (sorry, I forgot to mention that in my entrance post). I suppose your provided commands do not include that.

Anyway, as I am rather new to SPSS and not so familiar with commands yet, I'd prefer to work with the SPSS mask rather than direct commands. Could you tell me what I need to write into the yellow circled field to run the desired analysis (see screenshot "select_case_screen)?

Once again, thanks in advance and sorry for my "dummy questions".
 
#4
You can enter the following in the 'select cases if' field:

Code:
nmissing(v03_Year_of_Establishment to v38d_Hofstede_MascFem) le 5
I will stress again that this will work only if your variables are contiguous. If they are not then you should list them each individually or re-order your data file so that they are all in one block. If it doesn't work then please post the error message.