| Message/Author |
|
|
| Ryan Snead posted on Monday, April 06, 2020 - 5:56 pm
|
|
|
I’m using the R3STEP method with BRFSS data. When I run the first step for 2-4 classes and output the saved cprob I have no errors. For the second step I have no errors for 2 classes but I get this when I run the same syntax for 3 and 4 classes: “Clusters with the same IDs have been found in different strata....”. I used the same data and syntax as I did for 2 classes. Any ideas what may be going on?
[SECOND STEP]
Data:
File is 'brfss_cogdec_impute_step1_c2_save.txt';
Format is FREE;
Variable: names = CDHOUSE CDASSIST CDHELP CDSOCIAL CDDISCUS SEX MARITAL EDUCAG GENHLTH EMPLOY1 INCOMG RACEG21 AGEG BMI5CAT SMOKE100 DRNKANY5 CPROB1 CPROB2 C FINALWT STSTR PSU;
usevariables = PSU STSTR finalwt CDHOUSE CDASSIST CDHELP CDSOCIAL CDDISCUS;
categorical = CDHOUSE CDASSIST CDHELP CDSOCIAL CDDISCUS;
auxiliary = (R3STEP) SEX EDUCAG GENHLTH EMPLOY1 INCOMG RACEG21 AGEG SMOKE100 DRNKANY5;
strat = STSTR;
cluster = PSU;
weight = finalwt;
missing = *;
classes = c(2);
Analysis: Type = complex mixture; |
|
|
|
|
| We need to see the full outputs for the two runs you mention - send to Support along with your data and license number. |
|
|
| Ryan Snead posted on Friday, April 10, 2020 - 5:31 am
|
|
|
| Thank you, Bengt! I have a follow up question. In the BRFSS data set the "cluster" variable is "psu" and is unique for each individual. If they are unique identifiers already, should I instead be using "ststr" for my cluster variable? Or is the cluster specification even necessary? |
|
|
|
|
| I don't understand your complex sampling situation because it would be strange if each individual has a unique psu. You may want to ask on Multilevelnet and get advice from others who have used these data. |
|
|
| Back to top |
