Calculations and Group selection on imputed data.

[Jon K Peck]

It's hard to guess what might be happening here.  Multiple imputation works by using the SPLIT FILES mechanism, which has no effect on transformations.  It seems that  the problem is likely to be with the imputed values themselves.  Are you sure that they are actually what you think they are?  Try increasing the number of decimals displayed in the Data Editor so that you can see the full-precision values of the variables being used in the formulas.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:        SashaLibbrecht <[hidden email]>
To:        [hidden email],
Date:        03/26/2014 10:09 AM
Subject:        [SPSSX-L] Calculations and Group selection on imputed data.
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

Hello,

I have problems working with the imputed dataset of SPSS 22:

I tried to calculate a new variable in this dataset by deviding 2 others.
All the results for the original data were normal, but for the imputed data
I got some really weird numbers. For example: if it devided 1/2 = 0,81..

The same problem with selecting a group of cases:
I wanted all the cases with LNpos > 0 (number of metastatic lymph nodes)
Select cases--> if condition is satisfied--> LNpos > 0
Yet SPSS still selected cases with LNpos=0 (all of these were imputed
value's)

Similar problem is I tried selecting other groups.

Does anyone have any idea for a workaround or how to solve this problem?

Cheers!




--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Calculations-and-Group-selection-on-imputed-data-tp5725041.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[David Marso]

"All the results for the original data were normal, but for the imputed
data....."
Isuspect the original data are formatted as F1.0 or similar.  
In other words probably NO decimals (who'd a thought).
The lack of decimals is retained in the imputed data.

Indeed, as an example:
DATA LIST FREE / v1 v2 .
BEGIN DATA
.5 1.5
1.49 1.5
END DATA.
FORMATS v1 v2 (F1).
COMPUTE varnew=v1/v2.
LIST.
 
v1 v2   varnew
 
 1  2      .33
 1  2      .99
 
 
Number of cases read:  2    Number of cases listed:  2

Jon K Peck wrote

It's hard to guess what might be happening here.  Multiple imputation
works by using the SPLIT FILES mechanism, which has no effect on
transformations.  It seems that  the problem is likely to be with the
imputed values themselves.  Are you sure that they are actually what you
think they are?  Try increasing the number of decimals displayed in the
Data Editor so that you can see the full-precision values of the variables
being used in the formulas.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:   SashaLibbrecht <[hidden email]>
To:     [hidden email],
Date:   03/26/2014 10:09 AM
Subject:        [SPSSX-L] Calculations and Group selection on imputed
data.
Sent by:        "SPSSX(r) Discussion" <[hidden email]>



Hello,

I have problems working with the imputed dataset of SPSS 22:

I tried to calculate a new variable in this dataset by deviding 2 others.
All the results for the original data were normal, but for the imputed
data
I got some really weird numbers. For example: if it devided 1/2 = 0,81..

The same problem with selecting a group of cases:
I wanted all the cases with LNpos > 0 (number of metastatic lymph nodes)
Select cases--> if condition is satisfied--> LNpos > 0
Yet SPSS still selected cases with LNpos=0 (all of these were imputed
value's)

Similar problem is I tried selecting other groups.

Does anyone have any idea for a workaround or how to solve this problem?

Cheers!




--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Calculations-and-Group-selection-on-imputed-data-tp5725041.html

Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

Jon K Peck wrote

It's hard to guess what might be happening here.  Multiple imputation
works by using the SPLIT FILES mechanism, which has no effect on
transformations.  It seems that  the problem is likely to be with the
imputed values themselves.  Are you sure that they are actually what you
think they are?  Try increasing the number of decimals displayed in the
Data Editor so that you can see the full-precision values of the variables
being used in the formulas.


Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
phone: [phone]




From:   SashaLibbrecht <[hidden email]>
To:     [hidden email],
Date:   03/26/2014 10:09 AM
Subject:        [SPSSX-L] Calculations and Group selection on imputed
data.
Sent by:        "SPSSX(r) Discussion" <[hidden email]>



Hello,

I have problems working with the imputed dataset of SPSS 22:

I tried to calculate a new variable in this dataset by deviding 2 others.
All the results for the original data were normal, but for the imputed
data
I got some really weird numbers. For example: if it devided 1/2 = 0,81..

The same problem with selecting a group of cases:
I wanted all the cases with LNpos > 0 (number of metastatic lymph nodes)
Select cases--> if condition is satisfied--> LNpos > 0
Yet SPSS still selected cases with LNpos=0 (all of these were imputed
value's)

Similar problem is I tried selecting other groups.

Does anyone have any idea for a workaround or how to solve this problem?

Cheers!




--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/Calculations-and-Group-selection-on-imputed-data-tp5725041.html

Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD