sorry to bombard the list....but addendum on ezPerm

21 views
Skip to first unread message

Seth Myers

unread,
Jul 13, 2011, 4:30:03 AM7/13/11
to ez4r
Hi,

Sorry, working against a deadline.

ezPerm works when I have treatment/control set as "within" with 50
identical subjects in each within group and subject id as "wid". I
tried setting treatment/control as "between" and got the following
error.

Error in sim[, this_perm] = ezPerm_aov(sim_data, aov_formula) :
number of items to replace is not a multiple of replacement length

I'm note sure what this is about. As I said, I have 50 measurements
in treatment and control from the same 50 subjects in each.

Thanks for any insight, I looked at the r code for ezPerm and could
not determine the issue (at least not very quickly/easily).

Seth

Mike Lawrence

unread,
Jul 13, 2011, 8:55:47 AM7/13/11
to ez...@googlegroups.com
Most ez functions fail if you tell them to treat a variable as
"between" when it is truly "within". How do they know that the
variable is truly within? The "wid" labels will overlap across levels
of the variable if it is truly between.

If you have a within variable called "really_within"and you want to
explore what would happen if it were treated as between, you need to
create a new wid variable:

my_data$dumID = factor(paste(my_data$realID,my_data$really_within))

then use dumID as the "wid" variable.

Seth Myers

unread,
Jul 13, 2011, 2:16:03 PM7/13/11
to ez4r
Thanks again. I realize that this question seriously overlapped with
another, so sorry for any time wasted. -Seth
Reply all
Reply to author
Forward
0 new messages