combine multiple 'multiple responses'
Locked
 
|
Hi,
I have multiple responses for psychiatric diagnoses from four different sources: medical record problem list (6 variables for 6 possible diagnoses), other places in the record (4 variables for 4 possible diagnoses), other hospitals (3 variables for 3 possible diagnoses), and the doctor's diagnoses at the study visit (5 variables for 5 possible diagnoses). A participant may have one or more diagnoses from each source. There are no duplicates within sources but I expect them among sources. I have successfully used the command Mult Response to create a group for each of the four sources. Now, I want to 'combine' them so that I have the diagnoses (without duplication) from all four sources on each participant. I would also like to be able to compare the diagnoses. For instance to see whether the same diagnoses were recorded in the medical record problem list and the doctor's diagnoses at the study visit. Any help is appreciated. Jan Here is my multiple response code for each source (thank you John Hall) MULT RESPONSE /groups = diagnoses3 'diagnoses from SD medical record' ( MR_DxGrp1 xMR_DxGrp2 xMR_DxGrp3 xMR_DxGrp4 (0,7)) /variables=MR_sex (1,2) /tables= MR_sex by diagnoses3 /base = cases /cells = all. ===================== 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 |
Re: combine multiple 'multiple responses'
|
Jan
You don't need the forward slash before groups = I'm checking out what's possible in mult resp as I remember once doing some highly complex analysis which involved more than one MR group in the same table, and in whcu there was overlap in the contents of the groups. John [hidden email] http://surveyresearch.weebly.com ----- Original Message ----- From: J McClure To: [hidden email] Sent: Monday, August 30, 2010 1:47 AM Subject: combine multiple 'multiple responses' Hi, I have multiple responses for psychiatric diagnoses from four different sources: medical record problem list (6 variables for 6 possible diagnoses), other places in the record (4 variables for 4 possible diagnoses), other hospitals (3 variables for 3 possible diagnoses), and the doctor's diagnoses at the study visit (5 variables for 5 possible diagnoses). A participant may have one or more diagnoses from each source. There are no duplicates within sources but I expect them among sources. I have successfully used the command Mult Response to create a group for each of the four sources. Now, I want to 'combine' them so that I have the diagnoses (without duplication) from all four sources on each participant. I would also like to be able to compare the diagnoses. For instance to see whether the same diagnoses were recorded in the medical record problem list and the doctor's diagnoses at the study visit. Any help is appreciated. Jan Here is my multiple response code for each source (thank you John Hall) MULT RESPONSE /groups = diagnoses3 'diagnoses from SD medical record' ( MR_DxGrp1 xMR_DxGrp2 xMR_DxGrp3 xMR_DxGrp4 (0,7)) /variables=MR_sex (1,2) /tables= MR_sex by diagnoses3 /base = cases /cells = all. ===================== 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 ===================== 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 |
Re: combine multiple 'multiple responses'
|
|
In reply to this post by J McClure
Jan
You can create more than one group in mult response. Did a quick check on one of my small data sets: mult resp groups a1 (v4 to v7 (1,5)) a2 (v16 to v20 (1,5)) /tab a1 by a2 . a2a Total Typing Word- process Social statistics Survey analysis Other a1a 1 Count 93 115 56 22 66 140 2 Count 87 103 46 22 57 124 3 Count 69 90 46 21 54 109 4 Count 82 101 43 23 62 125 5 Count 90 108 50 24 69 131 Total Count 106 130 61 28 77 158 Percentages and totals are based on respondents. a. Group but it also works if you use the same source vars and (different) values, so you should be able to use this idea. If I were you I'd change my varnames to something less error-prone when typing and less likely to give you repetitive strain injuury (RSI)! v15 or whatever is much quicker and easier to type than MR_DxGrp1 John [hidden email] http://surveyresearch.weebly.com ----- Original Message ----- From: J McClure To: [hidden email] Sent: Monday, August 30, 2010 1:47 AM Subject: combine multiple 'multiple responses' Hi, I have multiple responses for psychiatric diagnoses from four different sources: medical record problem list (6 variables for 6 possible diagnoses), other places in the record (4 variables for 4 possible diagnoses), other hospitals (3 variables for 3 possible diagnoses), and the doctor's diagnoses at the study visit (5 variables for 5 possible diagnoses). A participant may have one or more diagnoses from each source. There are no duplicates within sources but I expect them among sources. I have successfully used the command Mult Response to create a group for each of the four sources. Now, I want to 'combine' them so that I have the diagnoses (without duplication) from all four sources on each participant. I would also like to be able to compare the diagnoses. For instance to see whether the same diagnoses were recorded in the medical record problem list and the doctor's diagnoses at the study visit. Any help is appreciated. Jan Here is my multiple response code for each source (thank you John Hall) MULT RESPONSE /groups = diagnoses3 'diagnoses from SD medical record' ( MR_DxGrp1 xMR_DxGrp2 xMR_DxGrp3 xMR_DxGrp4 (0,7)) /variables=MR_sex (1,2) /tables= MR_sex by diagnoses3 /base = cases /cells = all. ===================== 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 ===================== 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 |
Re: combine multiple 'multiple responses'
|
In reply to this post by J McClure
You can compare the 4 sources by having 4 /groups and /tables for 2
or 3 or 4 groups
You could also have a 5th group with all of the sources. Art Kendall On 8/29/2010 7:47 PM, J McClure wrote: Hi,===================== 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
Art Kendall
Social Research Consultants |
Re: combine multiple 'multiple responses'
|
|
In reply to this post by J McClure
Plain text doesn't show the table, so here it is
again.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
Hi,
Thanks for the additional information. I used your syntax below (mult resp) to create a supergroup of the other 4 groups and a frequency table for the supergroup.� I want to be sure the results are correct. I tried to check the results using List a1 a2 a3 a4 Dx /cases 30 but Dx is not recognized as a variable. Any suggestions? Jan � � On 8/30/2010 9:57 AM, John F Hall wrote: ===================== 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 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Re: combine multiple 'multiple responses'
|
|
Jan
Send me your syntax. I don't think you can
use LIST with MULT RESP as MR works with (hidden) tables, not data, and I
can't try it out as my machine is going berserk telling to replace my mouse
batteries even though they're still half full. Get the SPSS Help menu and
have a look at the syntal manual to see what you can and cannot do with
MR.
How have you generated DX?
John
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Re: combine multiple 'multiple responses'
|
|
Jan
because the groups are tables, not variables,
they're not saved, so you can't make groups from groups. The only way
round will be something like this:
MULT RESPONSE groups
diagnoses (PL_DxGrp1 xPL_DxGrp2 xPL_DxGrp3 xPL_DxGrp4 xPL_DxGrp5 xPL_DxGrp6 (0,7)) diagnoses2 (VA_DxGrp1 xVA_DxGrp2 xVA_DxGrp3 (0,7)) diagnoses3 ( MR_DxGrp1 xMR_DxGrp2 xMR_DxGrp3 xMR_DxGrp4 (0,7)) diagnoses4 ( MD_DxGrp2 xMD_DxGrp2 xMD_DxGrp3 xMD_DxGrp4 xMD_DxGrp5 (0,7)) /tab diagnoses by diagnoses2 to diagnoses4 / diagnoses2 by diagnoses3 diagnoses4
/ diagnoses3 by diagnoses4.
count a1 = PL_DxGrp1 xPL_DxGrp2 xPL_DxGrp3 xPL_DxGrp4
xPL_DxGrp5 xPL_DxGrp6 (0,7) .
count a2 = VA_DxGrp1 xVA_DxGrp2 xVA_DxGrp3 (0,7)
.
count a3 = MR_DxGrp1 xMR_DxGrp2 xMR_DxGrp3
xMR_DxGrp4 (0,7) .
count a4 = MD_DxGrp2 xMD_DxGrp2
xMD_DxGrp3 xMD_DxGrp4 xMD_DxGrp5 (0,7) .
*Need to remember what these are! . do repeat x = a1 to a4 /y = 1 to 4 .
compute x = y .
end repeat .
* This loses the composition of the groups, leaving
a unique value 1 to 4 for each group .
MULT RESPONSE groups
supergroup (a1 to a4 (1,4)) diagnoses (PL_DxGrp1 xPL_DxGrp2 xPL_DxGrp3 xPL_DxGrp4
xPL_DxGrp5 xPL_DxGrp6 (0,7))
diagnoses2 (VA_DxGrp1 xVA_DxGrp2 xVA_DxGrp3 (0,7)) diagnoses3 ( MR_DxGrp1 xMR_DxGrp2 xMR_DxGrp3 xMR_DxGrp4 (0,7)) diagnoses4 ( MD_DxGrp2 xMD_DxGrp2 xMD_DxGrp3 xMD_DxGrp4 xMD_DxGrp5 (0,7)) /freq supergroup /tab supergroup by diagnoses to
diagnoses4 .
Apart from that you'll need help from someone who
can do macros/Python, but you'll need to be very specific about what you want
your tables to look like and what you want in each cell. Best of
luck.
John
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |