SAS Support Communities

Hi John,    I think you might have better luck if you post this to this community page.    https://communities.sas.com/t5/Base-SAS-Programming/bd-p/programming    While it isn't exactly a perfect option, it is probably the most visable community.  Might I also suggest calling SAS tech support at 1 (919) 677-8008   This is a typical question that they would be happy to help you with. ... View more

I should mention though that you can access the newest version of SAS here. Try SAS® Now! Which you can use as long as you follow the License agreement. -Ken ... View more

Hi Sonyboy, In SAS 9.22 the most advanced heteroskedasticity SE correction is the option HCCME=4.  Reference for this correction is here. SAS/ETS(R) 9.22 User's Guide There is no canned correction for autocorrelation in that version. Best of luck. -Ken ... View more

Hi Maria, Can you tell us a bit more about your "complex survey design"?  We have a portfolio of "SURVEY*" procedures but none address 2sls specifically though I am sure they could be cobbled together.  Ken ... View more

It is possible that it is due to an imbalance.  You could check this by deleting a couple time obs from a cross section and then checking against my code i provided.  I am afraid that if you need any more assistance, you will have to contact Tech Support and open a formal track. 1-800-727-0025 Good luck-Ken  ... View more

alphabeta, Our apologies.  I was using syntax that we introduced in a more recent version.   Please try the following modifications. remove the groupid= include a id statement change countreg to TCOUNTREG For example *poisson with RE and no exposure(service) option; proc tcountreg data=two dist=poisson ; id ship; model accident=op co_65_69 co_70_74 co_75_79/ errorcomp=random /*offset=lnservice*/; run; *Poisson with RE and exposure(service); proc tcountreg data=two dist=poisson; id=ship; model accident=op co_65_69 co_70_74 co_75_79/ errorcomp=random offset=lnservice; run; I have run this vs old syntax and get the same estimates from older to newer versions. ==> it will still match Stata's coefficients. Good luck-Ken ... View more

I think the coefficients are different because of your use of the exposure() options.  If you run the following example you will find that the SAS and Stata coefficients are exactly the same.  data ships; input ship time yr_con yr_op service accident op co_65_69 co_70_74 co_75_79        weight; int=1; cards; 1       1       1       1       127         0   0       0       0       0 3 1       2       1       2       63          0   1       0       0       0 4 1       3       2       1       1095    3       0       1       0       0 4 1       4       2       2       1095    4       1       1       0       0 3 1       5       3       1       1512    6       0       0       1       0 6 1       6       3       2       3353    18      1       0       1       0 7 1       7       4       1       .           .   0       0       0       1 8 1       8       4       2       2244    11      1       0       0       1 7 2       1       1       1       44882   39      0       0       0       0 4 2       2       1       2       17176   29      1       0       0       0 5 2       3       2       1       28609   58      0       1       0       0 3 2       4       2       2       20370   53      1       1       0       0 2 2       5       3       1       7064    12      0       0       1       0 3 2       6       3       2       13099   44      1       0       1       0 4 2       7       4       1       .           .   0       0       0       1 6 2       8       4       2       7117    18      1       0       0       1 5 3       1       1       1       1179    1       0       0       0       0 4 3       2       1       2       552         1   1       0       0       0 3 3       3       2       1       781         0   0       1       0       0 2 3       4       2       2       676         1   1       1       0       0 1 3       5       3       1       783         6   0       0       1       0 3 3       6       3       2       1948    2       1       0       1       0 4 3       7       4       1       .           .   0       0       0       1 5 3       8       4       2       274        1    1       0       0       1 3 4       1       1       1       251        0    0       0       0       0 2 4       2       1       2       105        0    1       0       0       0 1 4       3       2       1       288        0    0       1       0       0 5 4       4       2       2       192        0    1       1       0       0 5 4       5       3       1       349        2    0       0       1       0 4 4       6       3       2       1208   11       1       0       1       0 3 4       7       4       1       .          .    0       0       0       1 2 4       8       4       2       2051    4       1       0       0       1 4 5       1       1       1       45         0    0       0       0       0 5 5       2       1       2       .          .    1       0       0       0 6 5       3       2       1       789        7    0       1       0       0 7 5       4       2       2       437        7    1       1       0       0 8 5       5       3       1       1157   5        0       0       1       0 9 5       6       3       2       2161   12       1       0       1       0 6 5       7       4       1       .          .    0       0       0       1 4 5       8       4       2       542        1    1       0       0       1 3 ; data two; set ships; lnservice = log(service); run; proc countreg data=two dist=poisson groupid=ship; model accident=op co_65_69 co_70_74 co_75_79/ errorcomp=fixed offset=lnservice; run; proc countreg data=two dist=negbin groupid=ship; model accident=op co_65_69 co_70_74 co_75_79/ errorcomp=fixed offset=lnservice; run; *poisson with RE and no exposure(service) option; proc countreg data=two dist=poisson groupid=ship; model accident=op co_65_69 co_70_74 co_75_79/ errorcomp=random /*offset=lnservice*/; run; *Poisson with RE and exposure(service); proc countreg data=two dist=poisson groupid=ship ; model accident=op co_65_69 co_70_74 co_75_79/ errorcomp=random offset=lnservice; run; proc export data=two outfile="C:\Public\two.dta"; run; /*open stata and run this use "C:\Public\two.dta", clear xtset ship *coefficients match the "Poisson with RE and no exposure(service) option" exactly xtpoisson accident op co_65_69 co_70_74 co_75_79,  re estimates store re *coefficients matches the "Poisson with RE and exposure(service)" exactly xtpoisson accident op co_65_69 co_70_74 co_75_79, exposure(service) re estimates store re_wos outreg2 [re re_wos] using compare, addstat(log-like, `e(ll)') ... View more

Hi alphabeta, You might want to try this. SAS/ETS(R) 14.1 User's Guide The COUNTREG procedure in ETS has panel data poisson regression which is likely similar to the Stata xtpoisson command.  try proc countreg data=a groupid=id; model y = x1-x10 / errorcomp=fixed dist=poisson; run; Let me know if this helps.  -Ken ... View more

See this. ... View more

The short answer is that you do not have SAS/ETS installed.  If you are interested in licensing that module, follow this link.  How to Buy SAS Software | SAS If you are a student or using the product for human capital purposes (non revenue generating) then you can access "cloud SAS" here. SAS® Logon Manager All you need to do is sign up for an account. Best of luck-Ken ... View more