Hello,   I am trying to run the following code:   proc logistic data=l; class trt(ref="0") group  GR1 GR1/ param=ref; model crit(event="1")= trt  group BASE GR1 GR1 / alpha=0.05 expb plcl plrl lackfit orpvalue; Run; base is continuous variable, all others are binary. I get the following NOTE: At least one element of the (projected) gradient is greater than 1e-3 and it looks that this is due to calculating p-value. if orpvalue removed from the model statement, this note is not present anymore.  Is there any other way to calculate p-value for Odds Ratio with proc logistic or example of how to do this with proc glimmix or genmode?   Thank you   ... View more

Hello,   I want to use proc mi to impute missing data and i have following question: In the code below i have in the var statement 2 categorical variables with no missing values and continuous variables that have missing data.    proc mi data=data1 seed=38467 nimpute=50 out=data_mcmc;   mcmc impute=monotone chain=multiple; var treatment gender calcium_v1 calcium_v2 calcium_v3...calcium_v20; run;    In this case do i have to include 'by' statement with these 2 categorical variables OR create a dummy variable (0,1) for each category OR no updates are needed in this case?  ... View more

Hi all. I have data: trtpn visit median  q1 q3 1 2  0   0   0 2  2 0   0   0 3  2 0   0   0 1 3 2   1  4 2 3 1  -2  3 3 3 3  1  4 1 4 -2 -3 2 2 4 2  -1  3 3 4 2 -1 3   and i am using code below to plot median and interquartile: proc sgplot data=mydata noautolegend; styleattrs datasymbols=(circlefilled asterisk tranglefilled)                Datalinepatterns=(solid solid solid)                DATACONTRASTCOLORS=(blue green magenta); series x=visitn y=median / group=trtpn groupdisplay=cluster clusterwidth=0.5; scatter x=visitn y=median/yerrorlower=q1 yerrorupper=q3 group=trtpn groupdisplay=cluster clusterwidth=0.5 errorbarattrs=(thickness=1) markerattrs=(size=7) name='s'; xaxis values=(2,3,4) fitpolicy=rotate grid ; yaxis grid values=(-80 to 20 by 20); keylegend 's'/title="Treatment Group:" down=1 location=outside position=bottom noborder;     However, the color for plot is not as indicated in the  DATACONTRASTCOLORS=(blue green magenta), but it plots everything in black. Does anyone knows how to change color? Thank you! ... View more

Hello,   I copied here an answer from previous question see below. However, in this example 3 median for same treatment is plotted at same stop, how to make that Week 12 all 3 treatment would be plotted next to each other (same for week26) to be able to see difference between values? Thanks data stats; input @1 avisit $10. @9 avisitn armn armcd $ median q1 q3; datalines; Baseline 0 60 DummyA 5639 2909 7221.5 Baseline 0 70 DummyB 3876 2929 6587 Baseline 0 80 DummyC 5112 3284 6686 Week1 12 60 DummyA 3304 1961 6157.5 Week12 12 70 DummyB 3496 1717 5021.5 Week12 12 80 DummyC 4266.5 3436 6504.5 Week26 26 60 DummyA 3762 2409 6233 Week26 26 70 DummyB 4377 3380 6735 Week26 26 80 DummyC 4261 3242 6084 ; run; proc sort data=stats out=sorted; by avisitn armcd; run; proc sgplot data=sorted noautolegend; series x=avisitn y=median / markers markerattrs=(symbol=circlefilled size=9) group=armcd curvelabel; highlow x=avisitn high=q3 low=q1 / lineattrs=(thickness=2) group=armcd; run;     ... View more