Dear all,   I wonder whether you can help me with this issue.   In an experiment I have several groups of three participants each, with defined roles as D, M and O, answering questions about each other Subsequently, each participant evaluated the other two on variable Y, using a scale 0-100. The data looks like in the attached excel file.   ID= participant N Group=group N Role=Role (D, M o O) V1me=self evaluation  V1D=evaluation of the D within the group V1M=evaluation of the M within the group V1O=evaluation of the O within the group   If the participant has Role D, then V1D has missing value, since the participant did not evaluate the D, but evaluated the self, M and O. Same for the other Role(s).   I need to compute a measure of accuracy, by comparing the evaluation that one gives of the other two members within the group, to the evaluation that they give to themselves. For instance, if the participant is a D, then his/her D_accuracy missing, his M_accuracy it is the difference between his/her evaluation of M and the evaluation that M gives to himself/herself, and for O_accuracy is the difference between his/her evaluation of O and the evaluation that O gives to himself/herself.   In the data sample attached, the D accuracy for subject 1004 is missing value, for M it is 10 (10-0) and for O it is 0 (0-0).   I would like SAS to do this for me, so that three variables are ultimately added for each participants, indicating D_accuracy, M_accuracy and O-accuracy.   Any suggestion would be welcome!   Thanks in advance.   E      ID group Role V1me V1D V1M V1O   Group ... View more

Dear all, I am using PROC MIXED for a simple design with one CLASS variable with two levels, and a random factor. Browsing the postings I saw a discussion about effect size in PROC MIXED dating of 2016, so I wonder whether there has been any progress with PROC MIXED and effect size can be automatically generated by PROC MIXED in the same way in which PROC GLM now does. proc mixed data = XX ; class cond randomF; model DV= cond ; random randomF ; lsmeans cond/tdiff cl pdiff; Thank you in advance. Emanuele ... View more