Multiple Imputation in Poisson Regression - Not Working

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

njacobson88
Posts: 1
Joined: Wed Nov 30, 2011 10:15 pm

Multiple Imputation in Poisson Regression - Not Working

Postby njacobson88 » Wed Nov 30, 2011 10:32 pm

Hi everyone,

I am having quite a lot of trouble using SPSS to analyze imputed data. I want to predict the number of births (count data), from a number of dichotomous and continuous predictor variables. I used Schafer's NORM to impute the data (k = 40 times) and I was able to get it into SPSS without any issues. I then used the Data --> Split File --> by imputation. I ran a Poisson regression in the Generalized Linear Modeling. Unfortunately, I received the following error:

"Warnings
The maximum number of step-halvings was reached but the log-likelihood value cannot be further improved. Output for the last iteration is displayed for split file Imputation_ = 16.
The GENLIN procedure continues despite the above warning(s). Subsequent results shown are based on the last iteration. Validity of the model fit is uncertain.
The maximum number of step-halvings was reached but the log-likelihood value cannot be further improved. Output for the last iteration is displayed for split file Imputation_ = 17.
The GENLIN procedure continues despite the above warning(s). Subsequent results shown are based on the last iteration. Validity of the model fit is uncertain.
The maximum number of step-halvings was reached but the log-likelihood value cannot be further improved. Output for the last iteration is displayed for split file Imputation_ = 22.
The GENLIN procedure continues despite the above warning(s). Subsequent results shown are based on the last iteration. Validity of the model fit is uncertain.
The maximum number of step-halvings was reached but the log-likelihood value cannot be further improved. Output for the last iteration is displayed for split file Imputation_ = 23.
The GENLIN procedure continues despite the above warning(s). Subsequent results shown are based on the last iteration. Validity of the model fit is uncertain.
The maximum number of step-halvings was reached but the log-likelihood value cannot be further improved. Output for the last iteration is displayed for split file Imputation_ = 26.
The GENLIN procedure continues despite the above warning(s). Subsequent results shown are based on the last iteration. Validity of the model fit is uncertain.
The maximum number of step-halvings was reached but the log-likelihood value cannot be further improved. Output for the last iteration is displayed for split file Imputation_ = 30.
The GENLIN procedure continues despite the above warning(s). Subsequent results shown are based on the last iteration. Validity of the model fit is uncertain.
The maximum number of step-halvings was reached but the log-likelihood value cannot be further improved. Output for the last iteration is displayed for split file Imputation_ = 39.
The GENLIN procedure continues despite the above warning(s). Subsequent results shown are based on the last iteration. Validity of the model fit is uncertain."

As such, I increased the maximum number of split-halves to be higher and this problem went away. However, I still had a major issue: the output does not show the findings for the last seven imputations (34-40) for the parameter estimates and the tests of the model effects. Additionally, it does not pool the 33 reported.

This ONLY happens in SPSS version 19. When using SPSS version 17, I do not have the error. Instead, it estimates all of the imputations iterations without issue, it just doesn't pool them. It says: "At least one of the vectors or matrices in the model being added is different in size from the corresponding vector or matrix in previously added models. This model cannot be added."

I am open to any way to solve this problem. I am fairly confident that I could hand compute the output of version 17 for the estimates and standard errors; however, I don't know how I would calculate the Wald Chi-Square and the p-value.

Any and all advice to solve this problem will be greatly appreciated.

Who is online

Users browsing this forum: No registered users and 1 guest

cron