I am analyzing a survey dataset and would like to conduct a multivariable logistic regression analysis. The data set has the following survey questions that I am interested in:
1) Visited chiropractor in the past 12 months? (Chiro_12m)
- Yes (1)
- No (2)
- Refused (3)
- Unknown (4)
2) Visited massage therapist in the past 12 months? (MT_12m)
- Yes (1)
- No (2)
- Refused (3)
- Unknown (4)
3) Visited osteopath provider in the past 12 months? (Ost_12m)
- Yes (1)
- No (2)
- Refused (3)
- Unknown (4)
I would like to make a new variable and combine the responses to all three questions. For example, the new variable name would be CAM_provider_combined and the variable labels will be:
(0) stated NO to any of the above 3 questions
(1) stated YES to any of the above 3 questions
(2) stated Refused/Unknown to any of the above 3 questions
How would I do the recoding in SPSS?
Thank you!
Hi Rhea,
In SPSS syntax window the following would do what you ask, but is this what you really need.?
Note that if a subject gave the following responses:
'Yes' to chiropractor
'No' to therapist
'3' or '4' for osteopath,
then all the conditions would be fulfilled and you would get '0' as the value for CAM_provider_combined due to the last line of the code. Used in this order responding 'YES' to any one variable is considered more important than 'NO' which is more important than 'refused/unknown'.
if(chiropractor>2 | therapist>2 | osteopath>2) CAM_provider_combined = 2.
if(chiropractor=1 | therapist=1 | osteopath=1) CAM_provider_combined = 1.
if(chiropractor=2 | therapist=2 | osteopath=2) CAM_provider_combined = 0.
execute.
Regards,
Timo
Hi Rhea. You have a problem in that the categories for your combined variable are not mutually exclusive. As the following SPSS code illustrates, it is possible for a given subject to meet all 3 conditions you describe. In such a case, what value would you assign? HTH.
* Rhea said:
* I would like to make a new variable and combine the responses to all three questions.
* For example, the new variable name would be CAM_provider_combined and the variable labels will be:
(0) stated NO to any of the above 3 questions
(1) stated YES to any of the above 3 questions
(2) stated Refused/Unknown to any of the above 3 questions .
* PROBLEM: Conditions 0-2 are NOT mutually exclusive.
* Generate a small dataset where each person meets all 3 conditions.
DATA LIST LIST / Chiro_12m MT_12m Ost_12m (3F1).
BEGIN DATA
1 2 3
1 2 4
2 3 1
2 4 1
3 1 2
4 1 2
END DATA.
* Create 3 indicator variables to flag the conditions described above.
* For these Flag variables, 1=Meets condition, 0 = Does not meet condition.
COMPUTE Flag0 = ANY(2,Chiro_12m, MT_12m, Ost_12m).
COMPUTE Flag1 = ANY(1,Chiro_12m, MT_12m, Ost_12m).
COMPUTE Flag2 = ANY(3,Chiro_12m, MT_12m, Ost_12m) OR ANY(4,Chiro_12m, MT_12m, Ost_12m).
FORMATS Flag0 to Flag2(F1).
LIST.
* All subjects meet all 3 conditions.
* So what value would you assign to CAM_provider_combined?.
So for my outcome, responding YES to any (i.e., chiropractor, massage therapist, or osteopath) is more important than NO. So the value labels for CAM_provider_combined would:
0= No to any CAM provider
1= Yes to any CAM provider
2= stated Refused/Don't Know in the responses (but this is not as important)
I'll be using the 0(No to any CAM provider) and 1(Yes to any CAM provider) for binary logistic regression analysis.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS