multiple imputations syntax question

[Jill Adelson]

Good afternoon,

 

I have 10 multiple imputations (using SPSS 17.0 Multiple Imputations). I have a cutoff value on a variable to determine each case's inclusion in 1 of 2 groups. There are some "borderline" cases that are classified one way in some imputations and differently in other imputations. Is there some syntax I could use to create a new variable that gives a 1 to everyone consistently classified as a 1 on every imputation and a 0 to everyone else? (Those who are "borderline" would not be included in either group.) I'm having trouble figuring out the syntax that will check the same child in every imputation and come back with a variable. If it were a child with 10 variables, I could easily sum and use if statements (if 10 then in group A, if 0 then in group B, if between 0 and 10 in neither group), but I'm not sure how to do this vertically (across cases rather than across variables). Any suggestions would be appreciated!

 

Many thanks!

Jill

 

 

Jill L. Adelson
Doctoral Candidate and Research Associate

Measurement, Evaluation, & Assessment and Gifted Education

Educational Psychology Department

University of Connecticut 

Jill
 

[Reutter, Alex]

There are probably a number of solutions, but one possibility is that you could use the Restructure Data Wizard (pasting CASESTOVARS) to restructure your data from cases to variables so that each imputation is a separate variable, then when you’re done figuring out what you want to do with the borderline cases use the Restructure Data Wizard (pasting VARSTOCASES) to go back to your original structure.

 

Cheers,

Alex

 

---

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Jill Adelson
Sent: Friday, March 27, 2009 11:36 AM
To: [hidden email]
Subject: multiple imputations syntax question

 

Good afternoon,

 

I have 10 multiple imputations (using SPSS 17.0 Multiple Imputations). I have a cutoff value on a variable to determine each case's inclusion in 1 of 2 groups. There are some "borderline" cases that are classified one way in some imputations and differently in other imputations. Is there some syntax I could use to create a new variable that gives a 1 to everyone consistently classified as a 1 on every imputation and a 0 to everyone else? (Those who are "borderline" would not be included in either group.) I'm having trouble figuring out the syntax that will check the same child in every imputation and come back with a variable. If it were a child with 10 variables, I could easily sum and use if statements (if 10 then in group A, if 0 then in group B, if between 0 and 10 in neither group), but I'm not sure how to do this vertically (across cases rather than across variables). Any suggestions would be appreciated!

 

Many thanks!

Jill

 

 

Jill L. Adelson
Doctoral Candidate and Research Associate

Measurement, Evaluation, & Assessment and Gifted Education

Educational Psychology Department

University of Connecticut 

Jill
 

[Jill Adelson]

Thanks, Alex.

 

I also had a suggestion to use "aggregate" to do this.

 

I'm set now. I appreciate you replying.

 

Jill

On Fri, Mar 27, 2009 at 4:09 PM, Reutter, Alex <[hidden email]> wrote:

There are probably a number of solutions, but one possibility is that you could use the Restructure Data Wizard (pasting CASESTOVARS) to restructure your data from cases to variables so that each imputation is a separate variable, then when you’re done figuring out what you want to do with the borderline cases use the Restructure Data Wizard (pasting VARSTOCASES) to go back to your original structure.

 

Cheers,

Alex

 

---

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Jill Adelson
Sent: Friday, March 27, 2009 11:36 AM
To: [hidden email]
Subject: multiple imputations syntax question

 

Good afternoon,

 

I have 10 multiple imputations (using SPSS 17.0 Multiple Imputations). I have a cutoff value on a variable to determine each case's inclusion in 1 of 2 groups. There are some "borderline" cases that are classified one way in some imputations and differently in other imputations. Is there some syntax I could use to create a new variable that gives a 1 to everyone consistently classified as a 1 on every imputation and a 0 to everyone else? (Those who are "borderline" would not be included in either group.) I'm having trouble figuring out the syntax that will check the same child in every imputation and come back with a variable. If it were a child with 10 variables, I could easily sum and use if statements (if 10 then in group A, if 0 then in group B, if between 0 and 10 in neither group), but I'm not sure how to do this vertically (across cases rather than across variables). Any suggestions would be appreciated!

 

Many thanks!

Jill

 

 

Jill L. Adelson
Doctoral Candidate and Research Associate

Measurement, Evaluation, & Assessment and Gifted Education

Educational Psychology Department

University of Connecticut 

Jill