Proc means for two levels

[u58497688]

Hi,

i have a following problem, concerning the building of means per Recall and ID

ID    Recall    TotalScore

001   1          4

001   2          2

002   1          4

002   2          0.5

003   1          4

003   2          1

I want to create the means out of the TotalScore for the two Recalls, by every ID; so that it should be like

ID     TotalScore_mean

001   3

002   2.25

003   2.5

I thought about the proc means procedure and a class-statement for ID, but it is not working for me.

I would be really happy if someone could help me.

Best,

Nina

[PaigeMiller · Accepted solution]

proc means data=have mean;
    class id;
    var totalscore;
run;

[u58497688]

Thanks, i just added an output, so I could work with it further. If it helps anyone else:

proc means mean data=have;
class ID;
var TotalScore;
output out=means_totalscore mean=mean_totalscore;
run;

[ballardw]

---

@u58497688 wrote:

Thanks, i just added an output, so I could work with it further. If it helps anyone else:

proc means mean data=have;
class ID;
var TotalScore;
output out=means_totalscore mean=mean_totalscore;
run;

---

Your output data set will have a variable _type_ indicating combinations of the Class variable. The _type_=0 will be the mean for all the ID values. If you do not want that in the output then add NWAY to the Proc statement to select only the highest type which would have the combinations for the class variable(s) and none of the other summaries. This is actually  useful behavior but there is a learning curve about the various options controlling the output with class variables. There will also be a _freq_ variable indicating how many observations were used for each statistic.