topic Re: Effect Modification logistic regression in SAS Programming

Effect Modification logistic regression

kyla123 — Fri, 18 Mar 2022 13:55:44 GMT

For an assignment I am trying to get the odds ratios of hpv shots and determining the association of poverty and using region as a confounder

The code I am attempting to use is

proc logistic data = puf.nisteenpuf19;

class cen_reg (param =ref ref= "NORTHEAST")incpov1 (param = ref ref = "ABOVE POVERTY > $75K") ;

model hpvi_group = incpov1 cen_reg incpov1*cen_reg;

run;

However in the output it doesn't give value for any northeast data and i need odds ratios for northeast_below poverty and northeast_above poverty <= $75k. it references all northeast values instead of just referencing northeast_above poverty >75$k

Hope this make sense

Re: Effect Modification logistic regression

PaigeMiller — Fri, 18 Mar 2022 13:58:13 GMT

Show us the output that you have questions about.

Re: Effect Modification logistic regression

kyla123 — Fri, 18 Mar 2022 14:00:25 GMT

Re: Effect Modification logistic regression

PaigeMiller — Fri, 18 Mar 2022 14:01:14 GMT

What happens if you include the ODDSRATIO statement in your code?

https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.4/statug/statug_logistic_syntax26.htm

Re: Effect Modification logistic regression

kyla123 — Fri, 18 Mar 2022 14:11:57 GMT

I am not 100% sure how to include it

proc logistic data = puf.nisteenpuf19;

class cen_reg (param =ref ref= "NORTHEAST")incpov1 (param = ref ref = "ABOVE POVERTY > $75K") ;

model hpvi_group = incpov1 cen_reg incpov1*cen_reg;

oddsratio cen_reg incpov1 dif = ref | all;

run;

Re: Effect Modification logistic regression

PaigeMiller — Fri, 18 Mar 2022 14:16:10 GMT

Was there a problem somewhere? I think it should be:

oddsratio cen_reg*incpov1 dif = all;

Please don't make us keep asking ... when there is a problem in the code, or a problem in the log, or a problem in the output ... show us.

Re: Effect Modification logistic regression

kyla123 — Fri, 18 Mar 2022 14:18:34 GMT

NOTE: PROCEDURE LOGISTIC used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

NOTE: The SAS System stopped processing this step because of errors.
2891 proc logistic data = puf.nisteenpuf19;
2892 class cen_reg (param =ref ref= "NORTHEAST")incpov1 (param = ref ref = "ABOVE POVERTY >
2892! $75K") ;
2893 model hpvi_group = incpov1 cen_reg incpov1*cen_reg;
2894 oddsratio cen_reg*incpov1 dif = all;
-
22
76
ERROR 22-322: Syntax error, expecting one of the following: ;, /.
ERROR 76-322: Syntax error, statement will be ignored.
2895 run;

Re: Effect Modification logistic regression

PaigeMiller — Fri, 18 Mar 2022 14:20:11 GMT

My mistake. I think it should be

oddsratio cen_reg*incpov1/dif = all;

Re: Effect Modification logistic regression

kyla123 — Fri, 18 Mar 2022 14:24:21 GMT

I received the same error message

NOTE: The SAS System stopped processing this step because of errors.
2896 proc logistic data = puf.nisteenpuf19;
2897 class cen_reg (param =ref ref= "NORTHEAST")incpov1 (param = ref ref = "ABOVE POVERTY >
2897! $75K") ;
2898 model hpvi_group = incpov1 cen_reg incpov1*cen_reg;
2899 oddsratio cen_reg*incpov1/dif = all;
-
22
76
ERROR 22-322: Syntax error, expecting one of the following: ;, /.
ERROR 76-322: Syntax error, statement will be ignored.
2900 run;

Re: Effect Modification logistic regression

PaigeMiller — Fri, 18 Mar 2022 14:27:01 GMT

Next step

When you have an error in the log, copy the log as text and paste it into the window that appears when you click on the </>. We need this every time you provide a log for us. This preserves the formatting of the log and makes it more readable.

Re: Effect Modification logistic regression

kyla123 — Fri, 18 Mar 2022 14:29:12 GMT

NOTE: PROCEDURE LOGISTIC used (Total process time):
      real time           0.00 seconds
      cpu time            0.00 seconds

NOTE: The SAS System stopped processing this step because of errors.
2910  proc logistic data = puf.nisteenpuf19;
2911  class cen_reg (param =ref ref= "NORTHEAST")incpov1 (param = ref ref = "ABOVE POVERTY >
2911! $75K") ;
2912  model hpvi_group = incpov1 cen_reg incpov1*cen_reg;
2913  oddsratio cen_reg*incpov1/dif = all;
                       -
                       22
                       76
ERROR 22-322: Syntax error, expecting one of the following: ;, /.
ERROR 76-322: Syntax error, statement will be ignored.
2914  run;

Sorry, I am very new to this

Re: Effect Modification logistic regression

kyla123 — Fri, 18 Mar 2022 14:34:15 GMT

I also tried using proc genmod but the output zeroed out all of northeast

PROC GENMOD DATA = puf.nisteenpuf19;
class CEN_REG (ref = "NORTHEAST") INCPOV1 (REF= "ABOVE POVERTY > $75K");
model HPVI_GROUP = CEN_REG INCPOV1 AGE SEX RACEETHK INCPOV1*CEN_REG/DIST = BINOMIAL LINK = LOG;
RUN;

Re: Effect Modification logistic regression

PaigeMiller — Fri, 18 Mar 2022 14:34:40 GMT

Apparently, ODDSRATIO will not work for interactions, so you should remove it. The default output from PROC LOGISTIC (without the ODDSRATIO statement) should show odds ratios for all model terms. Can you show us that odds ratio output?

Re: Effect Modification logistic regression

kyla123 — Fri, 18 Mar 2022 14:46:19 GMT

This is the output I receive when running pro logistic without the odds ratio statement

proc logistic data = puf.nisteenpuf19;
class cen_reg (param =ref ref= "NORTHEAST")incpov1 (param = ref ref = "ABOVE POVERTY > $75K") ;
model hpvi_group = incpov1 cen_reg incpov1*cen_reg;
run;

Re: Effect Modification logistic regression

PaigeMiller — Fri, 18 Mar 2022 14:56:21 GMT

Ok. I don't know how to get odds ratios for interactions. The odds ratios for your two main effects are produced by these statements:

oddsratio cen_reg;
oddsratio incpov1;

Re: Effect Modification logistic regression

PaigeMiller — Fri, 18 Mar 2022 15:05:35 GMT

Odds ratios for interactions: https://support.sas.com/kb/24/455.html