Hey all, I am using a large scale dataset with cases (students) nested in organizational structures (colleges). I had to impute missing values for four variables. I am used the mixed models – generalized linear model function in SPSS to run a multinomial logistic regression. However, I keep getting a message which states “Warning: This procedure ignores split file specifications”. I assume this means that the imputation is being ignored, but the case processing summary seems to indicate (based on the total N) that 5 samples are being employed. Any insight? ---------------------------------------------------------------------------------------------------- J. Luke Wood, PhD Assistant Professor of Administration, Rehabilitation and Postsecondary Education (ARPE) & Interwork Institute San Diego State University 3590 Camino del Rio North Office #213 San Diego, CA 92108 |
How many imputations did you do? Gene Maguin From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Jonathan Wood Hey all, I am using a large scale dataset with cases (students) nested in organizational structures (colleges). I had to impute missing values for four variables. I am used the mixed models – generalized linear model function in SPSS to run a multinomial logistic regression. However, I keep getting a message which states “Warning: This procedure ignores split file specifications”. I assume this means that the imputation is being ignored, but the case processing summary seems to indicate (based on the total N) that 5 samples are being employed. Any insight? ---------------------------------------------------------------------------------------------------- J. Luke Wood, PhD Assistant Professor of Administration, Rehabilitation and Postsecondary Education (ARPE) & Interwork Institute San Diego State University 3590 Camino del Rio North Office #213 San Diego, CA 92108 |
The dataset has 9,000+ cases nested in 240(ish) institutions. The model is complex, four primary scales and numerous controls. Missingness was only a problem in one scale. I had five variables (part of the problematic scale), missingness on these variables ranged from 10.7% to 25.0%. I did imputation.
|
Thank you for telling us that information. But you said you imputed values for several variables. I missed seeing that you said in your original message how many imputations you did. I've never used spss's imputation facility because I have mplus. You've got to be running genlinmixed. That said, I wonder if genlinmixed will work with imputed datasets. I'd guess that somebody on the list has run genlinmixed with imputed datasets. Perhaps they will respond. I'd expect that you could run a one level multiple regression model using your dataset. You could also run a one level ordinal regression model with genlinmixed. But, I wonder if you could run a model with mixed. Be interesting to know if the problem is with multilevel analyses, in general, or with genlinmixed, specifically.
Gene Maguin -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of jlukewood Sent: Thursday, May 09, 2013 7:55 PM To: [hidden email] Subject: Re: Multi-Level Modeling/Imputation Concerns The dataset has 9,000+ cases nested in 240(ish) institutions. The model is complex, four primary scales and numerous controls. Missingness was only a problem in one scale. I had five variables (part of the problematic scale), missingness on these variables ranged from 10.7% to 25.0%. I did imputation. -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Multi-Level-Modeling-Imputation-Concerns-tp5720153p5720156.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 ===================== 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 |
Administrator
|
From http://publib.boulder.ibm.com/infocenter/spssstat/v20r0m0/index.jsp?topic=%2Fcom.ibm.spss.statistics.help%2Fmi_analysis.htm:
Procedures That Support Pooling The following procedures support MI datasets, at the levels of pooling specified for each piece of output. Frequencies • The Statistics table supports Means at Univariate pooling (if S.E. mean is also requested) and Valid N and Missing N at Naïve pooling. • The Frequencies table supports Frequency at Naïve pooling. Descriptives • The Descriptive Statistics table supports Means at Univariate pooling (if S.E. mean is also requested) and N at Naïve pooling. Crosstabs • The Crosstabulation table supports Count at Naïve pooling. Means • The Report table supports Mean at Univariate pooling (if S.E. mean is also requested) and N at Naïve pooling. One-Sample T Test • The Statistics table supports Mean at Univariate pooling and N at Naïve pooling. • The Test table supports Mean Difference at Univariate pooling. Independent-Samples T Test • The Group Statistics table supports Means at Univariate pooling and N at Naïve pooling. • The Test table supports Mean Difference at Univariate pooling. Paired-Samples T Test • The Statistics table supports Means at Univariate pooling and N at Naïve pooling. • The Correlations table supports Correlations and N at Naïve pooling. • The Test table supports Mean at Univariate pooling. One-Way ANOVA • The Descriptive Statistics table supports Mean at Univariate pooling and N at Naïve pooling. • The Contrast Tests table supports Value of Contrast at Univariate pooling. Linear Mixed Models • The Descriptive Statistics table supports Mean and N at Naïve pooling. • The Estimates of Fixed Effects table supports Estimate at Univariate pooling. • The Estimates of Covariance Parameters table supports Estimate at Univariate pooling. • The Estimated Marginal Means: Estimates table supports Mean at Univariate pooling. • The Estimated Marginal Means: Pairwise Comparisons table supports Mean Difference at Univariate pooling. Generalized Linear Models and Generalized Estimating Equations. These procedures support pooled PMML. • The Categorical Variable Information table supports N and Percents at Naïve pooling. • The Continuous Variable Information table supports N and Mean at Naïve pooling. • The Parameter Estimates table supports the coefficient, B, at Univariate pooling. • The Estimated Marginal Means: Estimation Coefficients table supports Mean at Naïve pooling. • The Estimated Marginal Means: Estimates table supports Mean at Univariate pooling. • The Estimated Marginal Means: Pairwise Comparisons table supports Mean Difference at Univariate pooling. Bivariate Correlations • The Descriptive Statistics table supports Mean and N at Naïve pooling. • The Correlations table supports Correlations and N at Univariate pooling. Note that correlations are transformed using Fisher's z transformation before pooling, and then backtransformed after pooling. Partial Correlations • The Descriptive Statistics table supports Mean and N at Naïve pooling. • The Correlations table supports Correlations at Naïve pooling. Linear Regression. This procedure supports pooled PMML. • The Descriptive Statistics table supports Mean and N at Naïve pooling. • The Correlations table supports Correlations and N at Naïve pooling. • The Coefficients table supports B at Univariate pooling and Correlations at Naïve pooling. • The Correlation Coefficients table supports Correlations at Naïve pooling. • The Residuals Statistics table supports Mean and N at Naïve pooling. Binary Logistic Regression. This procedure supports pooled PMML. • The Variables in the Equation table supports B at Univariate pooling. Multinomial Logistic Regression. This procedure supports pooled PMML. • The Parameter Estimates table supports the coefficient, B, at Univariate pooling. Ordinal Regression • The Parameter Estimates table supports the coefficient, B, at Univariate pooling. Discriminant Analysis. This procedure supports pooled model XML. • The Group Statistics table supports Mean and Valid N at Naïve pooling. • The Pooled Within-Groups Matrices table supports Correlations at Naïve pooling. • The Canonical Discriminant Function Coefficients table supports Unstandardized Coefficients at Naïve pooling. • The Functions at Group Centroids table supports Unstandardized Coefficients at Naïve pooling. • The Classification Function Coefficients table supports Coefficients at Naïve pooling. Chi-Square Test • The Descriptives table supports Mean and N at Naïve pooling. • The Frequencies table supports Observed N at Naïve pooling. Binomial Test • The Descriptives table supports Means and N at Naïve pooling. • The Test table supports N, Observed Proportion, and Test Proportion at Naïve pooling. Runs Test • The Descriptives table supports Means and N at Naïve pooling. One-Sample Kolmogorov-Smirnov Test • The Descriptives table supports Means and N at Naïve pooling. Two-Independent-Samples Tests • The Ranks table supports Mean Rank and N at Naïve pooling. • The Frequencies table supports N at Naïve pooling. Tests for Several Independent Samples • The Ranks table supports Mean Rank and N at Naïve pooling. • The Frequencies table supports Counts at Naïve pooling. Two-Related-Samples Tests • The Ranks table supports Mean Rank and N at Naïve pooling. • The Frequencies table supports N at Naïve pooling. Tests for Several Related Samples • The Ranks table supports Mean Rank at Naïve pooling. Cox Regression. This procedure supports pooled PMML. • The Variables in the Equation table supports B at Univariate pooling. • The Covariate Means table supports Mean at Naïve pooling.
--
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/). |
Bruce, thanks. I didn’t know that existed and I see that it is reachable from the spss community link in the help drop-down.
-----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver Sent: Thursday, May 09, 2013 9:47 PM To: [hidden email] Subject: Re: Multi-Level Modeling/Imputation Concerns From http://publib.boulder.ibm.com/infocenter/spssstat/v20r0m0/index.jsp?topic=%2Fcom.ibm.spss.statistics.help%2Fmi_analysis.htm: Procedures That Support Pooling The following procedures support MI datasets, at the levels of pooling specified for each piece of output. Frequencies • The Statistics table supports Means at Univariate pooling (if S.E. mean is also requested) and Valid N and Missing N at Naïve pooling. • The Frequencies table supports Frequency at Naïve pooling. Descriptives • The Descriptive Statistics table supports Means at Univariate pooling (if S.E. mean is also requested) and N at Naïve pooling. Crosstabs • The Crosstabulation table supports Count at Naïve pooling. Means • The Report table supports Mean at Univariate pooling (if S.E. mean is also requested) and N at Naïve pooling. One-Sample T Test • The Statistics table supports Mean at Univariate pooling and N at Naïve pooling. • The Test table supports Mean Difference at Univariate pooling. Independent-Samples T Test • The Group Statistics table supports Means at Univariate pooling and N at Naïve pooling. • The Test table supports Mean Difference at Univariate pooling. Paired-Samples T Test • The Statistics table supports Means at Univariate pooling and N at Naïve pooling. • The Correlations table supports Correlations and N at Naïve pooling. • The Test table supports Mean at Univariate pooling. One-Way ANOVA • The Descriptive Statistics table supports Mean at Univariate pooling and N at Naïve pooling. • The Contrast Tests table supports Value of Contrast at Univariate pooling. Linear Mixed Models • The Descriptive Statistics table supports Mean and N at Naïve pooling. • The Estimates of Fixed Effects table supports Estimate at Univariate pooling. • The Estimates of Covariance Parameters table supports Estimate at Univariate pooling. • The Estimated Marginal Means: Estimates table supports Mean at Univariate pooling. • The Estimated Marginal Means: Pairwise Comparisons table supports Mean Difference at Univariate pooling. Generalized Linear Models and Generalized Estimating Equations. These procedures support pooled PMML. • The Categorical Variable Information table supports N and Percents at Naïve pooling. • The Continuous Variable Information table supports N and Mean at Naïve pooling. • The Parameter Estimates table supports the coefficient, B, at Univariate pooling. • The Estimated Marginal Means: Estimation Coefficients table supports Mean at Naïve pooling. • The Estimated Marginal Means: Estimates table supports Mean at Univariate pooling. • The Estimated Marginal Means: Pairwise Comparisons table supports Mean Difference at Univariate pooling. Bivariate Correlations • The Descriptive Statistics table supports Mean and N at Naïve pooling. • The Correlations table supports Correlations and N at Univariate pooling. Note that correlations are transformed using Fisher's z transformation before pooling, and then backtransformed after pooling. Partial Correlations • The Descriptive Statistics table supports Mean and N at Naïve pooling. • The Correlations table supports Correlations at Naïve pooling. Linear Regression. This procedure supports pooled PMML. • The Descriptive Statistics table supports Mean and N at Naïve pooling. • The Correlations table supports Correlations and N at Naïve pooling. • The Coefficients table supports B at Univariate pooling and Correlations at Naïve pooling. • The Correlation Coefficients table supports Correlations at Naïve pooling. • The Residuals Statistics table supports Mean and N at Naïve pooling. Binary Logistic Regression. This procedure supports pooled PMML. • The Variables in the Equation table supports B at Univariate pooling. Multinomial Logistic Regression. This procedure supports pooled PMML. • The Parameter Estimates table supports the coefficient, B, at Univariate pooling. Ordinal Regression • The Parameter Estimates table supports the coefficient, B, at Univariate pooling. Discriminant Analysis. This procedure supports pooled model XML. • The Group Statistics table supports Mean and Valid N at Naïve pooling. • The Pooled Within-Groups Matrices table supports Correlations at Naïve pooling. • The Canonical Discriminant Function Coefficients table supports Unstandardized Coefficients at Naïve pooling. • The Functions at Group Centroids table supports Unstandardized Coefficients at Naïve pooling. • The Classification Function Coefficients table supports Coefficients at Naïve pooling. Chi-Square Test • The Descriptives table supports Mean and N at Naïve pooling. • The Frequencies table supports Observed N at Naïve pooling. Binomial Test • The Descriptives table supports Means and N at Naïve pooling. • The Test table supports N, Observed Proportion, and Test Proportion at Naïve pooling. Runs Test • The Descriptives table supports Means and N at Naïve pooling. One-Sample Kolmogorov-Smirnov Test • The Descriptives table supports Means and N at Naïve pooling. Two-Independent-Samples Tests • The Ranks table supports Mean Rank and N at Naïve pooling. • The Frequencies table supports N at Naïve pooling. Tests for Several Independent Samples • The Ranks table supports Mean Rank and N at Naïve pooling. • The Frequencies table supports Counts at Naïve pooling. Two-Related-Samples Tests • The Ranks table supports Mean Rank and N at Naïve pooling. • The Frequencies table supports N at Naïve pooling. Tests for Several Related Samples • The Ranks table supports Mean Rank at Naïve pooling. Cox Regression. This procedure supports pooled PMML. • The Variables in the Equation table supports B at Univariate pooling. • The Covariate Means table supports Mean at Naïve pooling. Maguin, Eugene wrote > Thank you for telling us that information. But you said you imputed > values for several variables. I missed seeing that you said in your > original message how many imputations you did. I've never used spss's > imputation facility because I have mplus. You've got to be running > genlinmixed. That said, I wonder if genlinmixed will work with imputed > datasets. I'd guess that somebody on the list has run genlinmixed with imputed datasets. > Perhaps they will respond. I'd expect that you could run a one level > multiple regression model using your dataset. You could also run a one > level ordinal regression model with genlinmixed. But, I wonder if you > could run a model with mixed. Be interesting to know if the problem is > with multilevel analyses, in general, or with genlinmixed, specifically. > Gene Maguin > > > > > -----Original Message----- > From: SPSSX(r) Discussion [mailto: > SPSSX-L@.UGA > ] On Behalf Of jlukewood > Sent: Thursday, May 09, 2013 7:55 PM > To: > SPSSX-L@.UGA > Subject: Re: Multi-Level Modeling/Imputation Concerns > > The dataset has 9,000+ cases nested in 240(ish) institutions. The > model is complex, four primary scales and numerous controls. > Missingness was only a problem in one scale. I had five variables > (part of the problematic scale), missingness on these variables ranged > from 10.7% to 25.0%. I did imputation. > > > > -- > View this message in context: > http://spssx-discussion.1045642.n5.nabble.com/Multi-Level-Modeling-Imp > utation-Concerns-tp5720153p5720156.html > Sent from the SPSSX Discussion mailing list archive at Nabble.com. > > ===================== > To manage your subscription to SPSSX-L, send a message to > LISTSERV@.UGA > (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 > LISTSERV@.UGA > (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 ----- -- 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/Multi-Level-Modeling-Imputation-Concerns-tp5720153p5720158.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 ===================== 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 |
Free forum by Nabble | Edit this page |