Multiple Imputation - Negative Values
Locked
Multiple Imputation - Negative Values
 
|
I'm experiencing a problem with multiple imputation. Some of the variables that I'm imputing cannot have negative values, e.g. income. Yet, imputed data sets include negative values for some of the originally missing cases on these variables. Is there a way to prevent the imputation procedure from assigning negative values to variables?
Courtney Cronley, Ph.D. Postdoctoral Associate Center of Alcohol Studies Rutgers University [hidden email] ===================== 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: Multiple Imputation - Negative Values
|
|
Use /CONSTRAINTS MIN=0 From the syntax reference: MIN = NONE | num. Minimum allowable imputed value for scale variables. Specify a number. If an imputed value is less than the minimum, the procedure draws another value until it finds one that is greater than or equal to MIN or the MAXCASEDRAWS or MAXPARAMDRAWS threshold is reached (See METHOD subcommand). There is no default minimum. MIN is ignored when predictive mean matching is used or when applied to a categorical variable. An error occurs if MIN is greater than or equal to MAX. For date format variables, values must be enclosed in single or double quotes and expressed in the same date format as the defined date format for the variable. Alex
I'm experiencing a problem with multiple imputation. Some of the variables that I'm imputing cannot have negative values, e.g. income. Yet, imputed data sets include negative values for some of the originally missing cases on these variables. Is there a way to prevent the imputation procedure from assigning negative values to variables? Courtney Cronley, Ph.D. Postdoctoral Associate Center of Alcohol Studies Rutgers University [hidden email] |
Re: Multiple Imputation - Negative Values
|
In reply to this post by Courtney M. Cronley
Are you sure you haven't got missing data with
negative values?
|
Re: Multiple Imputation - Negative Values
|
|
Negative
or otherwise out of range values, e.g., an imputed value of 6.34 or 3.33
for a 1 to 5 likert type variable are certainly possible because of the
underlying generating process. The question is what to with them. John Graham
who wrote an early piece of software was one of the early practicioners of
MI. I believe John has written about this issue. Another person to check out is
Joe Schafer who wrote a key book on MI and developed more software. (Let me add
that in naming these two, i intend no slight to Donald
Rubin).
Gene
Maguin
From: SPSSX(r) Discussion [mailto:[hidden email]]
On Behalf Of John F Hall
Sent: Thursday, September 23, 2010 4:03 PM To: [hidden email] Subject: Re: Multiple Imputation - Negative Values Are you sure you haven't got missing data with
negative values?
|
Re: Multiple Imputation - Negative Values
Administrator
|
Although I've not read the book Gene mentions, I will second his recommendation of Schafer & Graham generally. Here is list of articles I found helpful when I first started reading about missing data, including some by Schafer & Graham. Acock, A. C. (2005). Working with missing values. Journal of Marriage and Family, 67, 1012-1028. Donders, A. Rogier T., van der Heijden, Geert J.M.G., Stijnen, T., & Moons, K. G. M. (2006). Review: A gentle introduction to imputation of missing values. Journal of Clinical Epidemiology, 59, 1087-1091. Multiple Imputation Online. http://www.multiple-imputation.com/ Schafer, J. L. (1999). Multiple imputation: A primer. Statistical Methods in Medical Research, 8, 3-15. Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147-177. HTH.
--
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/). |
Re: Multiple Imputation - Negative Values
|
|
I would be careful imposing restrictions on the imputation model such as
/CONSTRAINTS MIN=0. When a variable has no negative values but multiple imputation still imputes negative values, this could be an indication that the variable in question is heavily skewed. I would check that first. If income is heavily skewed then maybe you could do a log transformation, impute the missing data and transform the variable back after multiple imputation. Another option which is probably even better is to use Predictive Mean Matching instead of Linear Regression (Method, Imputation Method, Custom, Predictive Mean Matching (PMM)). This option automatically preserves the properties of the variables such as minimum and maximum values, increment etc. Good luck! Best regards, Joost van Ginkel Joost R. van Ginkel, PhD Leiden University Faculty of Social and Behavioural Sciences PO Box 9555 2300 RB Leiden The Netherlands Tel: +31-(0)71-527 3620 Fax: +31-(0)71-527 1721 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver Sent: Thursday, September 23, 2010 11:37 PM To: [hidden email] Subject: Re: Multiple Imputation - Negative Values Gene Maguin wrote: > > Negative or otherwise out of range values, e.g., an imputed value of > 6.34 or > 3.33 for a 1 to 5 likert type variable are certainly possible because > of the underlying generating process. The question is what to with > them. John Graham who wrote an early piece of software was one of the > early practicioners of MI. I believe John has written about this > issue. Another person to check out is Joe Schafer who wrote a key book > on MI and developed more software. (Let me add that in naming these > two, i intend no slight to Donald Rubin). > > Gene Maguin > > Although I've not read the book Gene mentions, I will second his recommendation of Schafer & Graham generally. Here is list of articles I found helpful when I first started reading about missing data, including some by Schafer & Graham. Acock, A. C. (2005). Working with missing values. Journal of Marriage and Family, 67, 1012-1028. Donders, A. Rogier T., van der Heijden, Geert J.M.G., Stijnen, T., & Moons, K. G. M. (2006). Review: A gentle introduction to imputation of missing values. Journal of Clinical Epidemiology, 59, 1087-1091. Multiple Imputation Online. http://www.multiple-imputation.com/ Schafer, J. L. (1999). Multiple imputation: A primer. Statistical Methods in Medical Research, 8, 3-15. Schafer, J. L., & Graham, J. W. (2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147-177. HTH. ----- -- Bruce Weaver [hidden email] http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." NOTE: My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Multiple-Imputation-Negati ve-Values-tp2851717p2851942.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 ********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. ********************************************************************** ===================== 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: Multiple Imputation - Negative Values
|
|
Good point, Joost, especially since the OP specifically mentioned income as a variable that was receiving negative imputed values, and income tends to be skewed. Alex |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |