I have a survey question which is a "select all that apply" question (there are 24 choices), and at the end there is a choice basically "none of these". Survey monkey recorded all responses as 1, and non-responses as missing. I want to write syntax so that if the respondent chose ANY of the selections, which would include the "none" option, then instead of the non-responses being missings, they would be 0s. That way I could identify true missings as those who did not respond to any of the choices. Unfortunately, I am a novice with SPSS. I think I have SAS brain. Here is the code I am trying to use. Is there an easier way to do this, and how do I get SPSS to do what I am trying to do?
Thank you.
IF (IWU1ASCHPRE=1 OR IWU1BSCHPRE=1 OR IWU1CSCHPRE=1 OR IWU1DSCHPRE=1 OR IWU1ESCHPRE=1 OR
IWU1FSCHPRE=1 OR IWU1GSCHPRE=1 OR IWU1HSCHPRE=1 OR IWU1ISCHPRE=1 OR IWU1JSCHPRE =1 OR
IWU1KSCHPRE=1 OR IWU1LSCHPRE=1 OR IWU1MSCHPRE=1 OR
IWU1NSCHPRE=1 OR IWU1OSCHPRE=1 OR IWU1PSCHPRE=1 OR IWU1QSCHPRE=1
OR IWU1RSCHPRE=1 OR IWU1SSCHPRE=1 OR IWU1TSCHPRE=1 OR
IWU1USCHPRE=1 OR IWU1VSCHPRE=1 OR IWU1WSCHPRE=1 OR IWU1XSCHPRE=1
OR IWU1YSCHPRE=1) THEN IF IWU1ASCHPRE=. IWU1ASCHPRE=0.
EXECUTE.
Use selecte "and" command with clear true-missing criteria for recode to distinguish the item-missing 0s.
Hope this helps,
Matthew
I used to run to following to replace all blanks with 0s in the full dataset, as long as there were no string variables in the mix:
RECODE ALL (SYSMIS=0).
EXECUTE.
I do not know what you want to do, but when I use Multiple Dichotomous (MDSETS) variables in SPSS, I create an new variable that holds 'no answer given'. To create it: (you might want to have better names and labels, as the label for test will appear in an MD table).
COUNT test=var01 to varXX (1).
VARIABLE LABELS test 'count var'.
EXECUTE.
RECODE test (0=1)(ELSE = 0) .
You could ask for a crosstabs between this variable and "none of these" to see if there is a difference. If people used it as intended or your Survey collecting program worked as it should, the crosstab would be empty. Unless you used RECODE var01 to varXx test (SYSMIS = 0 ). This would in pratice remove all missing values and replace them with 0 (as mentioned by Joseph)
* Define Multiple Response Sets.
MRSETS
/MDGROUP NAME=$MDvar01 CATEGORYLABELS=VARLABELS VARIABLES=Var01 to VarXX Test
VALUE=1
/DISPLAY NAME=[$MDvar01].
This will add a new variable to SPSS file with the name $MDVar01 . When you save the file it will be saved as well.
However, I do not know if you will have a version of SPSS that has the module to create and analyse MD variables. The procedure CUSTOM TABLES gives you options to test hyphotesis about MD variables and these test are Bonferroni.
Rather than many logical ORs, compute a new variable using any to identify cases with at least one response of 1. Here is a small example:
DATA LIST LIST (",")/ id a b c d .
BEGIN DATA
1, , 1, , ,
2, , , , ,
3, 1, 1, 1, ,
4, 1, , , ,
5, , , , ,
6, 1, 1, , ,
END DATA
execute.
compute any_response = any(1, a to d).
execute.
DO IF any_response = 1.
recode a to d (sysmis = 0).
END IF.
execute.
Please use the video from Dr. Todd Grande on YouTube to answer your question:
https://www.youtube.com/watch?v=XE6j39CB3qQ
Thank you everyone for all the answers. I changed all the system and user missings to 0 first, then successfully created a variable for missings that worked, using AND rather than OR, as per suggestions. I finally realized that part of my problem was that I had a variable that should have been classified as numeric, but was classified as string. A newbie mistake, I know.
- Badges
- Science topic