missing value issue
Locked
 
|
Hi Listers,
I have a scale of 28 items which correspond to 7 motivational constructs - now I run the syntax below:
COMPUTE IMTK=SUM(Motiv_2, Motiv_9, Motiv_16, Motiv_23).
COMPUTE IMTA=SUM(Motiv_6, Motiv_13, Motiv_20, Motiv_27).
COMPUTE IMTES=SUM(Motiv_4, Motiv_11, Motiv_18, Motiv_25).
COMPUTE EMID=SUM(Motiv_3, Motiv_10, Motiv_17, Motiv_24).
COMPUTE EMIN=SUM(Motiv_7, Motiv_14, Motiv_21, Motiv_28).
COMPUTE EMER=SUM(Motiv_1, Motiv_8, Motiv_15, Motiv_22).
COMPUTE AMOT=SUM(Motiv_5, Motiv_12, Motiv_19, Motiv_26).
EXECUTE.
and I find that if one of the item values is missing then the construct is also missing - two questions - how would I get the syntax to give me a sum of all valid scores, and secondly what approaches would you suggest to deal with missing data in this instance?
thanks
Muir
Dr Muir Houston
Social Justice, Place and Lifelong Education Research
School of Education
College of Social Sciences
University of Glasgow
0141-330-4699
R3L+ Project - Adult education in the light of the European Quality Strategy http://www.learning-regions.net/ GINCO Project - Grundtvig International Network of Course Organisers http://www.ginconet.eu/ |
|
|
Hi Muir, Why don’t you try the recode your variables: recode Motiv_1 to Motiv_27 (sys=0). execute. and do the sum as you were doing before. Thanks. Joan Casellas Vega Media Research Analyst Phone: +44 20 7593 1585 From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Muir Houston Hi Listers, I have a scale of 28 items which correspond to 7 motivational constructs - now I run the syntax below: COMPUTE IMTK=SUM(Motiv_2, Motiv_9, Motiv_16, Motiv_23). COMPUTE IMTA=SUM(Motiv_6, Motiv_13, Motiv_20, Motiv_27). COMPUTE IMTES=SUM(Motiv_4, Motiv_11, Motiv_18, Motiv_25). COMPUTE EMID=SUM(Motiv_3, Motiv_10, Motiv_17, Motiv_24). COMPUTE EMIN=SUM(Motiv_7, Motiv_14, Motiv_21, Motiv_28). COMPUTE EMER=SUM(Motiv_1, Motiv_8, Motiv_15, Motiv_22). COMPUTE AMOT=SUM(Motiv_5, Motiv_12, Motiv_19, Motiv_26). EXECUTE. and I find that if one of the item values is missing then the construct is also missing - two questions - how would I get the syntax to give me a sum of all valid scores, and secondly what approaches would you suggest to deal with missing data in this instance? thanks Muir Dr Muir Houston Social Justice, Place and Lifelong Education Research School of Education College of Social Sciences University of Glasgow 0141-330-4699 R3L+ Project - Adult education in the light of the European Quality |
|
|
In reply to this post by Muir Houston-3
Muir,
I think that sum is functionally the same as sum.1, which i'm
guessing is what you want, i.e., a score is computed is any one item is present.
My question is this: Does sum.1 show the same pattern as
sum?
Gene Maguin From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Muir Houston Sent: Wednesday, August 24, 2011 10:08 AM To: [hidden email] Subject: missing value issue Hi Listers,
I have a scale of 28 items which correspond
to 7 motivational constructs - now I run the syntax below:
COMPUTE IMTK=SUM(Motiv_2, Motiv_9, Motiv_16, Motiv_23).
COMPUTE IMTA=SUM(Motiv_6, Motiv_13, Motiv_20, Motiv_27).
COMPUTE IMTES=SUM(Motiv_4, Motiv_11, Motiv_18, Motiv_25).
COMPUTE EMID=SUM(Motiv_3, Motiv_10, Motiv_17, Motiv_24).
COMPUTE EMIN=SUM(Motiv_7, Motiv_14, Motiv_21, Motiv_28).
COMPUTE EMER=SUM(Motiv_1, Motiv_8, Motiv_15, Motiv_22).
COMPUTE AMOT=SUM(Motiv_5, Motiv_12, Motiv_19, Motiv_26).
EXECUTE.
and I find that if one of the item values is missing then the
construct is also missing - two questions - how would I get the syntax to give
me a sum of all valid scores, and secondly what approaches would you suggest to
deal with missing data in this instance?
thanks
Muir
Dr Muir Houston
Social Justice, Place and Lifelong Education
Research
School of Education
College of Social Sciences
University of Glasgow
0141-330-4699
R3L+ Project - Adult education in the light of the
European Quality Strategy http://www.learning-regions.net/ GINCO Project - Grundtvig International Network of Course Organisers http://www.ginconet.eu/ |
|
|
In reply to this post by joan casellas
The result of the sum function will only
be missing if all the arguments are missing. By default, it
ignores missing values.
Jon Peck (no "h") Senior Software Engineer, IBM [hidden email] new phone: 720-342-5621 From: joan casellas <[hidden email]> To: [hidden email] Date: 08/24/2011 08:54 AM Subject: Re: [SPSSX-L] missing value issue Sent by: "SPSSX(r) Discussion" <[hidden email]> Hi Muir, Why don’t you try the recode your variables: recode Motiv_1 to Motiv_27 (sys=0). execute. and do the sum as you were doing before. Thanks. Joan Casellas Vega Media Research Analyst Phone: +44 20 7593 1585 From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Muir Houston Sent: 24 August 2011 15:08 To: [hidden email] Subject: missing value issue Hi Listers, I have a scale of 28 items which correspond to 7 motivational constructs - now I run the syntax below: COMPUTE IMTK=SUM(Motiv_2, Motiv_9, Motiv_16, Motiv_23). COMPUTE IMTA=SUM(Motiv_6, Motiv_13, Motiv_20, Motiv_27). COMPUTE IMTES=SUM(Motiv_4, Motiv_11, Motiv_18, Motiv_25). COMPUTE EMID=SUM(Motiv_3, Motiv_10, Motiv_17, Motiv_24). COMPUTE EMIN=SUM(Motiv_7, Motiv_14, Motiv_21, Motiv_28). COMPUTE EMER=SUM(Motiv_1, Motiv_8, Motiv_15, Motiv_22). COMPUTE AMOT=SUM(Motiv_5, Motiv_12, Motiv_19, Motiv_26). EXECUTE. and I find that if one of the item values is missing then the construct is also missing - two questions - how would I get the syntax to give me a sum of all valid scores, and secondly what approaches would you suggest to deal with missing data in this instance? thanks Muir Dr Muir Houston Social Justice, Place and Lifelong Education Research School of Education College of Social Sciences University of Glasgow 0141-330-4699 R3L+ Project - Adult education in the light of the European Quality Strategy http://www.learning-regions.net/ GINCO Project - Grundtvig International Network of Course Organisers http://www.ginconet.eu/ |
|
In reply to this post by Muir Houston-3
If this is your own scale, one way to deal with
missing values is to assume that the score for that item is the
mean of the other items for that subject.
instead of using SUM() use MEAN.1() COMPUTE IMTK=MEAN.1(Motiv_2, Motiv_9, Motiv_16, Motiv_23). Using the mean has an advantage for interpretation since it on the same scale as the item response scale. IFF you need the score as a sum to compare to other research or to norms. something like this untested syntax should work. COMPUTE IMTK=rnd(MEAN.1(Motiv_2, Motiv_9, Motiv_16, Motiv_23)*(3/nvalid(.Motiv_2, Motiv_9, Motiv_16, Motiv_23))). If you are in the process of developing these scales, if you had a few more items on each scale on the next round, that would dilute the error introduced by missing data imputation. Again if you are developing these scales, it would be a good idea to see if some items are more frequently missing. Then do some cognitive testing such as think-aloud to see why people might be skipping each items. You then need to see if skipping an item is plausibly related to the construct you are trying to measure. Art Kendall Social Research Consultants On 8/24/2011 10:08 AM, Muir Houston 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
Art Kendall
Social Research Consultants |
Administrator
|
In reply to this post by Muir Houston-3
When sums are used as scores for sub-scales, one generally wants the total number of possible items going into the sum. If some items are missing, one usually does some kind of pro-rating. Also, there is usually some minimum number of items required to compute a sum. Your sums have 4 items going into them if there are no missing data. Suppose you only want to compute a (pro-rated) sum if there are at least 2 items. You could do this:
COMPUTE IMTK=mean.2(Motiv_2, Motiv_9, Motiv_16, Motiv_23) * 4. COMPUTE IMTA=mean.2(Motiv_6, Motiv_13, Motiv_20, Motiv_27) * 4. COMPUTE IMTES=mean.2(Motiv_4, Motiv_11, Motiv_18, Motiv_25) * 4. COMPUTE EMID=mean.2(Motiv_3, Motiv_10, Motiv_17, Motiv_24) * 4. COMPUTE EMIN=mean.2(Motiv_7, Motiv_14, Motiv_21, Motiv_28) * 4. COMPUTE EMER=mean.2(Motiv_1, Motiv_8, Motiv_15, Motiv_22) * 4. COMPUTE AMOT=mean.2(Motiv_5, Motiv_12, Motiv_19, Motiv_26) * 4.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
In reply to this post by Art Kendall
Please ignore the weighting in the second compute
command in the previous message. It takes the number of items
into account twice, which is not correct.
As Bruce pointed out that command should have been COMPUTE IMTK=rnd(MEAN.1(Motiv_2, Motiv_9, Motiv_16, Motiv_23)*(3). Art Kendall Social Research Consultants On 8/24/2011 11:27 AM, Art Kendall wrote: If this is your own scale, one way to deal with missing values is to assume that the score for that item is the mean of the other items for that subject.===================== 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 |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |