topic Re: Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a lin in SAS Procedures

Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a linear

MVP — Wed, 17 Oct 2018 15:10:22 GMT

Hello,

I am running into this error which is leaving me stumped. I am not sure what is going on and it is hard to find an easy(easier) to understand explanation of the error.

The following parameters have been set to 0, since the variables are a linear combination of other variables as shown.

RHQ420YES=INTERCEPT

My code is as follows:

PROC SURVEYLOGISTIC DATA=NH.MASTER_11_14_OC_HRT NOMCAR;
STRATA SDMVSTRA;
CLUSTER SDMVPSU;
WEIGHT COMBINED_WEIGHT_OC_HRT_4YR;
CLASS RHQ420 (PARAM=REF REF='No');
DOMAIN RHQ420;
MODEL IMMEDIATERECALL (EVENT='<=13')=RHQ420/ VADJUST=NONE;
FORMAT RHQ420 YESNO.

IMMEDIATERECALL IMMEDIATERECALL.;
RUN;

and the frequency chart

SAS Output

The FREQ Procedure

FrequencyPercentRow PctCol Pct

Table of RHQ420 by IMMEDIATERECALLRHQ420(Evertaken birthcontrol pills?) IMMEDIATERECALL<=13 >=14 TotalYesNoTotalFrequency Missing = 18355

4.00

6.79

32.64

865

54.89

93.21

62.55

928

58.88

130

8.25

20.06

67.36

518

32.87

79.94

37.45

648

41.12

193

12.25

1383

87.75

1576

100.00

Can anyone help shed some light on my issue?

Thanks,

Mark

Re: Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a lin

Reeza — Wed, 17 Oct 2018 15:14:59 GMT

Run that PROC FREQ table again, but by your Strata and CLUSTER variables and see if you end up with an zero cells.

Re: Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a lin

MVP — Wed, 17 Oct 2018 15:24:13 GMT

Thank you for the reply.

I ran re-ran with the strata, cluster, and weights using proc surveyfreq and did not find any zero cells.

proc surveyfreq
data=NH.MASTER_11_14_OC_HRT;
strata sdmvstra;
cluster sdmvpsu;
TABLE RHQ420*IMMEDIATERECALL;

weight COMBINED_WEIGHT_OC_HRT_4YR;
FORMAT IMMEDIATERECALL IMMEDIATERECALL.
RHQ420 YESNO.;
RUN;

SAS Output

The SAS System

The SURVEYFREQ Procedure

Data SummaryNumber of StrataNumber of ClustersNumber of ObservationsNumber of Observations UsedNumber of Obs with Nonpositive WeightsSum of Weights

19931

19151

780

308897448

Table of RHQ420 by IMMEDIATERECALLRHQ420 IMMEDIATERECALL Frequency WeightedFrequency Std Err ofWgt Freq Percent Std Err ofPercentYes <=13 >=14 TotalNo <=13 >=14 TotalTotal <=13 >=14 TotalFrequency Missing = 17575

63	997238	153386	3.3522	0.5600
865	18095632	1145792	60.8273	1.5853
928	19092870	1122048	64.1794	1.4627
130	1928290	206343	6.4818	0.6766
518	8728040	659213	29.3387	1.3916
648	10656330	722564	35.8206	1.4627
193	2925528	290156	9.8340	1.0276
1383	26823672	1587016	90.1660	1.0276
1576	29749200	1611067	100.000

Re: Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a lin

MVP — Wed, 17 Oct 2018 15:33:30 GMT

I think it has something to do with my domain statement, however, I want/need to analyse according to that specific subgroup.

Re: Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a lin

MVP — Wed, 17 Oct 2018 15:50:57 GMT

What I want to do is have RHQ420 (Ever take birth control pills?) group as my reference group. I've run the same procedure before using other variables with no issues, and again, I have no zero values in my frequency chart, stumped me even more.

Re: Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a lin

MVP — Wed, 17 Oct 2018 16:00:04 GMT

Also, the log does not indicate any errors:

NOTE: PROCEDURE SURVEYFREQ used (Total process time):
real time 0.04 seconds
cpu time 0.04 seconds

1107 PROC SURVEYLOGISTIC DATA=NH.MASTER_11_14_OC_HRT NOMCAR;
1108 STRATA SDMVSTRA;
1109 CLUSTER SDMVPSU;
1110 WEIGHT COMBINED_WEIGHT_OC_HRT_4YR;
1111 CLASS RHQ420 (PARAM=REF REF='No');
1112 DOMAIN RHQ420;
1113 MODEL IMMEDIATERECALL (EVENT='<=13')=RHQ420/ VADJUST=NONE;
1114 FORMAT RHQ420 YESNO.
1115 IMMEDIATERECALL IMMEDIATERECALL.;
1116 RUN;

NOTE: PROC SURVEYLOGISTIC is modeling the probability that IMMEDIATERECALL='<=13'.
NOTE: Convergence criterion (GCONV=1E-8) satisfied.
NOTE: PROC SURVEYLOGISTIC is modeling the probability that IMMEDIATERECALL='<=13'.
NOTE: Convergence criterion (GCONV=1E-8) satisfied.
NOTE: PROC SURVEYLOGISTIC is modeling the probability that IMMEDIATERECALL='<=13'.
NOTE: Convergence criterion (GCONV=1E-8) satisfied.
NOTE: PROCEDURE SURVEYLOGISTIC used (Total process time):
real time 0.34 seconds
cpu time 0.32 seconds

Re: Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a lin

ballardw — Wed, 17 Oct 2018 17:05:24 GMT

@MVP wrote:

Hello,

I am running into this error which is leaving me stumped. I am not sure what is going on and it is hard to find an easy(easier) to understand explanation of the error.

The following parameters have been set to 0, since the variables are a linear combination of other variables as shown.

RHQ420YES=INTERCEPT

My code is as follows:

PROC SURVEYLOGISTIC DATA=NH.MASTER_11_14_OC_HRT NOMCAR;
STRATA SDMVSTRA;
CLUSTER SDMVPSU;
WEIGHT COMBINED_WEIGHT_OC_HRT_4YR;
CLASS RHQ420 (PARAM=REF REF='No');
DOMAIN RHQ420;
MODEL IMMEDIATERECALL (EVENT='<=13')=RHQ420/ VADJUST=NONE;
FORMAT RHQ420 YESNO.

IMMEDIATERECALL IMMEDIATERECALL.;
RUN;

and the frequency chart

SAS Output

The FREQ Procedure

FrequencyPercentRow PctCol Pct

Table of RHQ420 by IMMEDIATERECALLRHQ420(Evertaken birthcontrol pills?) IMMEDIATERECALL<=13 >=14 TotalYesNoTotalFrequency Missing = 18355

63

4.00

6.79

32.64

865

54.89

93.21

62.55

928

58.88

130

8.25

20.06

67.36

518

32.87

79.94

37.45

648

41.12

193

12.25

1383

87.75

1576

100.00

Can anyone help shed some light on my issue?

Thanks,

Mark

Try creating another variable with the same values as your domain variable and use that instead on the domain statement. If you have a format assigned to your current domain variable make sure to do the same for the new one.

Re: Proc SurveyLogistic - The following parameters have been set to 0, since the variables are a lin

MVP — Wed, 17 Oct 2018 17:19:58 GMT

I created a variable the same as my domain variable, but the same error came up:

DATA NH.MASTER_11_14_OC_HRT;
SET NH.MASTER_11_14_OC_HRT;
IF RHQ420=. THEN RHQ420X=.;
IF RHQ420=1 THEN RHQ420X=1;
ELSE IF RHQ420=2 THEN RHQ420X=2;
RUN;

PROC SURVEYLOGISTIC DATA=NH.MASTER_11_14_OC_HRT NOMCAR;
STRATA SDMVSTRA;
CLUSTER SDMVPSU;
WEIGHT COMBINED_WEIGHT_OC_HRT_4YR;
CLASS RHQ420 (PARAM=REF REF='No');
DOMAIN RHQ420X;
MODEL IMMEDIATERECALL (EVENT='<=13')=RHQ420/ VADJUST=NONE;
FORMAT RHQ420 YESNO.
RHQ420X YESNO.
IMMEDIATERECALL IMMEDIATERECALL.;
RUN;