Hi, Friends and Colleagues!

Starting at 42'48" of a video at https://www.youtube.com/watch?v=27NSGTcWaPI

I noticed a comment on the fact that SPSS did not pool ANOVA results from imputed data through its multiple imputation. The video came from 2014, and it indicated that SPSS would fix a glitch soon.

A moment ago, I downloaded data from https://stats.idre.ucla.edu/wp-content/uploads/2017/01/hsb_mar.sas7bdat to my c: drive, and ran the following syntax

SET RNG=MT MTINDEX=54321.

GET

SAS DATA='c:\hsb_mar.sas7bdat'.

DATASET NAME DataSet1 WINDOW=FRONT.

*Impute Missing Data Values.

DATASET DECLARE ANOVA.

MULTIPLE IMPUTATION PROG READ

/IMPUTE METHOD=AUTO NIMPUTATIONS=25 MAXPCTMISSING=NONE

/MISSINGSUMMARIES NONE

/IMPUTATIONSUMMARIES MODELS

/OUTFILE IMPUTATIONS=ANOVA .

*Pool ANOVA Results across imputations

DATASET ACTIVATE ANOVA.

ONEWAY READ BY PROG

/MISSING ANALYSIS.

Although no error surfaced, I was unable to get the pooled results across multiple imputations for reporting!

I would be grateful if our friends and colleagues can offer some guidance. Thank you for your attention.

Sincerely,

JJ

I'm afraid it's still the case that SPSS Statistics does not provide pooled results for multiple degree of freedom omnibus tests such as overall F tests for a one-way ANOVA model. Complete pooled results with inference are still only available for individual contrasts (or parameter estimates in more general models).