Thank you - so I can't share data sorry - but I have mocked up the syntax as best I can below
I have a large dataset with 997 cases and I have run a multiple imputation on this data set to produce a dataset with the 997 cases imputed multiple times to fill in/estimate the missing values.
I have a smaller data set with the same (and more variables) 228 cases which I have used the exact process below and it worked fine ...but now with the larger data set it is returning only 417 rows of data (listed as table number so I believe these are supposed to be my cases - but obviously I am missing more than half of the original 997)
All the case numbers/ID numbers for the 997 are different - numeric 1-997 and marked as a nominal variable in the dataset - I also tried this with the unique ID no which the participants entered (combo of 4 letters and 4 numerals) and it still came back with 417...
* have already sorted multiple imputations into cases so that list is by case and each imputation is listed under that case in ascending order
* i.e., case 2 - has 5 lines of imputation 1-5 followed by case 3 1-5 lines etc. etc
did this by going to data --> sort cases - first by case no then by imputation number in ascending order
* Then syntax for the process I followed is below
* OMS.
DATASET DECLARE forsyntax1.
OMS
/SELECT TABLES
/IF COMMANDS=['Frequencies'] SUBTYPES=['Statistics']
/DESTINATION FORMAT=SAV NUMBERED=TableNumber_
OUTFILE='forsyntax1' VIEWER=NO
/COLUMNS SEQUENCE=[RALL CALL LALL].
DATASET ACTIVATE DataSet1.
SORT CASES BY ID_No.
SPLIT FILE SEPARATE BY ID_No.
FREQUENCIES VARIABLES= *here I have all my variables except imputation number & ID-code (i.e., case code as per the above)
/STATISTICS=MEAN MODE
/ORDER=ANALYSIS.
*after the frequencies are running I then go to OMS control panel and end the process above - then wait until pooled file appears - with less than half the cases every time...
my ID no variable is a nominal variable & marked as such in the parent file.... this exact process worked with a smaller file but does not work at all with this larger one with 997 cases.
Hope that is easier to see what I am doing :)
Thanks again
------------------------------
Sharon Cooksey
------------------------------
Original Message:
Sent: Tue November 26, 2024 08:22 PM
From: Jon Peck
Subject: Pooling a multiple imputation - losing half the data
there is a much better chance of a useful answer if you describe exactly what you did - pasting the syntax and a small sample of the dataset would help rather than saying you followed a video.
Original Message:
Sent: 11/26/2024 7:32:00 PM
From: Sharon Cooksey
Subject: Pooling a multiple imputation - losing half the data
Hi - sorry me again - and this time I can't find an easy solution. I have managed to make this work on one data file (228 cases) and keep all the data ... but when I try on a larger file (997 cases) I lose over half of my cases.... they are all there in all of the imputations - and I am following exactly the same procedure for both datasets but I cannot work out what I am doing incorrectly (have now tried at least 5 times :( )
So I am following the procedure detailed here - (last third of video)
Imputation of missing data - Multiple imputation using SPSS
| YouTube |
remove preview |
 |
| Imputation of missing data - Multiple imputation using SPSS |
| Uploaded by Roger Go on 2020-09-25. |
| View this on YouTube > |
|
|
is there something I should be doing differently please? Getting a bit desperate now (please keep answers simple as I am a novice at all this and my brain is threatening to quit ;) )
Many thanks
------------------------------
Sharon Cooksey
------------------------------