In this case it is recommended that you combine the actual Chi-Square statistics from the Type3 tests in PHREG. This cannot be done with Proc MIANALYZE, but there is a MACRO on Dr. Paul Allison's web site called %COMBCHI that can do this directly. http://www.ssc.upenn.edu/~allison/#Macros ... View more

The GEE algorithm is an iterative algorithm that solves for both the parameter estimates and the covariance matrix so you should expect to see differences in both the parameters and the standard errors. There will be more specific details here http://support.sas.com/documentation/cdl/en/statug/68162/HTML/default/viewer.htm#statug_genmod_details29.htm   ... View more

That first ERROR usually has something to do with changes made to the output template.  I would contact technical support to ask about the second.   In the mean time you might also try simplifying your PHREG code so that only what you need for MIANALYZE will be produced.   Proc phreg data=pop1_imput1; format slevel slevel.; class slevel (param=ref ref='Primary'); model delriskc1*riskc1(0)=slevel/eventcode=1; by _Imputation_; ods output parameterEstimates = a_mvn; run; proc mianalyze parms(classvar=full)=a_mvn; class slevel; modeleffects slevel; RUN; run; ... View more

The way it actually works is that you next run whatever analysis you were planning to run (regression for example) and then use MIANALYZE to combine the estimates from that analysis.   There is a good description of the three step process here in the documentation. http://support.sas.com/documentation/cdl/en/statug/68162/HTML/default/viewer.htm#statug_mianalyze_overview.htm   ... View more

  The approach from a syntax standpoint would be similar to this example http://support.sas.com/documentation/cdl/en/statug/68162/HTML/default/viewer.htm#statug_mi_examples07.htm   The only difference in your case would be that rather than using the DISCRIM method, you would use the LOGISTIC method. fcs logistic (species/link=glogit);       ... View more

Based on the description at the beginning of the paper, it seems they are fitting something similar to this example from the Proc FMM documentation. http://support.sas.com/documentation/cdl/en/statug/68162/HTML/default/viewer.htm#statug_fmm_examples01.htm   In any regard, FMM will give you estimates of the mixing probabilties directly (or they can be modeled using additional effects with the PROBMODEL statement0. ... View more

Thinking more about this, that won't work as it will only give the individual comparisons for one of the variables.     An easier and direct way would be to use the LSMEANS statement with the DIFF, EXP and  CL option.  You will also need to make sure you use GLM coding so you can get the LSMEANS    class grouping1 (ref='0') race (ref='1') education (ref='1') ragender (ref='1.Male') health (ref='1') smokenow (ref='0') physical (ref='0') depression (ref='2')/param=glm; model duration*status(0)= grouping1 R7AGEY_M race education ragender health smokenow physical BMI depression grouping1*depression /risklimits; lsmeans grouping1*depression/cl diff exp; run;   ... View more

Are you saying that you want to compare the different cells in the two-way interaction rather than comparisons where one of the effects is held constant?   If so, then you can just put the interaction on the HAZARDRATIO statement and it will give you each of the comparisons. ... View more

There is an HAZARDRATIO statement that will allow you to get hazard ratios in the presence of an interaction.   http://support.sas.com/documentation/cdl/en/statug/68162/HTML/default/viewer.htm#statug_phreg_syntax13.htm   ... View more

You would be best served using the COLLIN option in Proc REG to assess collinearity. The numbers listed in the "Collinearity Diagnostics" table represent the number of eigenvalues extracted from the rescaled X`X matrix. They are listed from largest to smallest. You can look at the Condition Number to determine if there is any collinearity. The condition indices are the square roots of the ratio of the largest eigenvalue to each individual eigenvalue. The largest condition index is the condition number of the scaled X matrix. Belsey, Kuh, and Welsch (1980) suggest that, when this number is around 10, weak dependencies may be starting to affect the regression estimates. When this number is larger than 100, the estimates may have a fair amount of numerical error (although the statistical standard error almost always is much greater than the numerical error). For each variable, PROC REG produces the proportion of the variance of the estimate accounted for by each principal component. A collinearity problem occurs when a component associated with a high condition index contributes strongly (variance proportion greater than about 0.5) to the variance of two or more variables. The VIF option in the MODEL statement provides the Variance Inflation Factors (VIF). These factors measure the inflation in the variances of the parameter estimates due to collinearities that exist among the regressor (dependent) variables. There are no formal criteria for deciding if a VIF is large enough to affect the predicted values, although some authorities (Myers 1990) state that values exceeding 10 may be cause for concern. The variables with the larger VIF values may indicate that those variables are the ones involved in the collinearity. ... View more