How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARMS =

Ahinoa — Sun, 13 Jun 2021 04:39:26 GMT

I am fitting a mixed-effects logistic model using PROC GLIMMIX. All the variables I am working with are either binary of categorical. Because there were a few variables that had 10% or more of missing data, I used multiple imputation to be able to have a complete dataset. I fitted 5 previous models (null, level-one predictors, level-one predictors + covariates, level-two predictors, and level-two predictors + level-one variables).

Because I am interested in exploring moderation effects, I entered four interaction terms in the model statement (M_EXPANSION*PLACE_3 GINI*PLACE_3 SPP_SPECTRUM*PLACE_3 LPERCENTAGE2017*PLACE_3). Everything was working perfectly, until I entered the interaction terms into the modeling process. When I used PROC MIANALYZE to pool the estimates, I got the following error:

Within-imputation Estimate missing for effect ....in the input PARMS = dataset.

Because I am new to SAS, I cannot seem to figure out what the next steps would be to fix this error. That is why I am asking in this forum for your help. Below I am copying the code I am using to fit the models as well as section of log that is relevant to this error.

I will greatly appreciate any help you could provide!!!!

Here is my code:

TITLE "MODEL 6 - FIXED EFFECTS RANDOM INTERCEPT MODERATION ANALYSIS ";

PROC GLIMMIX DATA = DISS3.LLCP2017_I_ANALYSIS METHOD = QUAD EMPIRICAL = CLASSICAL; /* EMPIRICAL = CLASSICAL ->SANDWICH VARIANCE ESTIMATOR*/

WHERE ((_RACEGR3 = 5) AND (_RFHYPE5 = 2));

BY _IMPUTATION_;

CLASS /*STATE LEVEL VARIABLES*/

M_EXPANSION (REF = "0") GINI (REF = "0") SPP_SPECTRUM (REF = "0") LPERCENTAGE2017 (REF = "0") PLACE_3 (REF = "0")

/*INDIVIDUAL LEVEL VARIABLES*/

IRisk_Profile (REF = "0") IHC (REF = "0") IFPL (REF = "0") _EDUCAG (REF = "0")

/*DEMOGRAPHIC CHARACTERISTICS*/

SEX (REF = "0") _AGEG5YR (REF = "2") MARITAL (REF = "2") RENTHOM1 (REF = "1") EMPLOY1 (REF = "0") GENHLTH (REF = "1");

MODEL BPMEDS (DESCENDING) = IRISK_PROFILE IHC IFPL _EDUCAG SEX _AGEG5YR MARITAL RENTHOM1 EMPLOY1 GENHLTH

M_EXPANSION GINI SPP_SPECTRUM LPERCENTAGE2017 PLACE_3 M_EXPANSION*PLACE_3 GINI*PLACE_3 SPP_SPECTRUM*PLACE_3 LPERCENTAGE2017*PLACE_3/

LINK = LOGIT DIST = BINARY SOLUTION OBSWEIGHT =AW CL; /*OBSWEIGHT -> WEIGHT FOR LEVEL-ONE VARIABLES "OBSERVATIONAL LEVEL*/

RANDOM INTERCEPT / SUBJECT = _STATE TYPE = CHOL SOLUTION;

COVTEST GLM;

ODS OUTPUT SOLUTIONR = CLASSEFFECTS_6;

ODS OUTPUT PARAMETERESTIMATES =PAREST_6;

RUN;

QUIT;

PROC SORT DATA = PAREST_6;

BY _IMPUTATION_;

RUN;

PROC MIANALYZE PARMS = PAREST_6 EDF=7913; /*POOL PARAMETER ESTIMATES FOR FIXED EFFECTS*/

CLASS IRISK_PROFILE IHC IFPL _EDUCAG SEX _AGEG5YR MARITAL RENTHOM1 EMPLOY1 GENHLTH M_EXPANSION GINI SPP_SPECTRUM LPERCENTAGE2017 PLACE_3;

MODELEFFECTS INTERCEPT IRISK_PROFILE IHC IFPL _EDUCAG SEX _AGEG5YR MARITAL RENTHOM1 EMPLOY1 GENHLTH

M_EXPANSION GINI SPP_SPECTRUM LPERCENTAGE2017 PLACE_3 M_EXPANSION*PLACE_3 GINI*PLACE_3 SPP_SPECTRUM*PLACE_3

LPERCENTAGE2017*PLACE_3;

RUN;

PROC SORT DATA = CLASSEFFECTS_6;

BY SUBJECT _IMPUTATION_;

RUN;

PROC MIANALYZE PARMS = CLASSEFFECTS_6 (RENAME =(StdErrPred=STDERR)) EDF=7913; /*POLL RANDOM EFFECTS 'INTERCEPT FOR EACH STATE'*/

MODELEFFECTS INTERCEPT;

BY SUBJECT;

RUN;

QUIT;

Log:

1138 PROC SORT DATA = PAREST_6;

1139 BY _IMPUTATION_;

1140 RUN;

NOTE: Input data set is already sorted, no sorting done.

NOTE: PROCEDURE SORT used (Total process time):

real time 0.00 seconds

cpu time 0.01 seconds

1141

1142 PROC MIANALYZE PARMS = PAREST_6 EDF=7913; /*POOL PARAMETER ESTIMATES FOR FIXED EFFECTS*/

1143 CLASS IRISK_PROFILE IHC IFPL _EDUCAG SEX _AGEG5YR MARITAL RENTHOM1 EMPLOY1 GENHLTH M_EXPANSION

1143! GINI SPP_SPECTRUM LPERCENTAGE2017 PLACE_3;

1144 MODELEFFECTS INTERCEPT IRISK_PROFILE IHC IFPL _EDUCAG SEX _AGEG5YR MARITAL RENTHOM1 EMPLOY1

1144! GENHLTH

1145 M_EXPANSION GINI SPP_SPECTRUM LPERCENTAGE2017 PLACE_3 M_EXPANSION*PLACE_3

1145! GINI*PLACE_3 SPP_SPECTRUM*PLACE_3

1146 LPERCENTAGE2017*PLACE_3;

1147 RUN;

ERROR: Within-imputation Estimate missing for effect M_EXPANSION*PLACE_3 in _Imputation_= 1 in the

input PARMS= data set.

NOTE: The SAS System stopped processing this step because of errors.

NOTE: PROCEDURE MIANALYZE used (Total process time):

real time 0.00 seconds

cpu time 0.00 seconds

Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM

SAS_Rob — Mon, 21 Jun 2021 15:06:51 GMT

I suspect that the names of the variables coming out of GLIMMIX are being truncated to 20 characters. You can add the NAMELEN=100 option to the GLIMMIX statement to avoid this issue.

https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.4/statug/statug_glimmix_syntax01.htm#statug.glimmix.gmxoptnamelen

PROC GLIMMIX DATA = DISS3.LLCP2017_I_ANALYSIS METHOD = QUAD EMPIRICAL = CLASSICAL NAMELEN=100;

...

Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM

Iriska — Tue, 26 Sep 2023 20:33:18 GMT

I got the same Error after proc Mixed. The problem was in proc sort before proc mianalyze. It seems the analyzed dataset used in proc mianalyze must not ne sorted after proc MIXED or similar procedure.

Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM

YYK273 — Mon, 08 Apr 2024 14:45:28 GMT

I got the same errors. However I still got the same errors for both sorted and not sorted PARMS= data set.

These are the notes:

ERROR: Within-imputation Estimate missing for effect RACOHBYR
in _Imputation_= 1 in the input PARMS= data set.

Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM

ballardw — Mon, 08 Apr 2024 15:06:53 GMT

@YYK273

Please start your own thread. You can post a link to this question if you think it helpful.

Post your LOG with the code and the notes and messages as well as the error. Often there is diagnostic information that explains why. Copy the text from the log; on the forum open a text box and paste the text with the messages.

That sort of message where it includes something like

in _Imputation_= 1 in the input PARMS= data set.

typically indicates a data related issue related to the value of the variable, such as too few observations to model something.

topic Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM in Statistical Procedures

How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARMS =

Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM

Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM

Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM

Re: How do you solve this error: Within-imputation Estimate missing for effect ....in the input PARM