There are many definitions of outliers.
If you mean values that are not valid for the variable, e.g., 7 on a 1
to 5 response scale, then if you cannot go back to the source and get
the legitimate value, use simply define such values are missing for any
group.
Otherwise, rarely does it make sense to exclude outliers at all.
Please explain in more detail why a value would be illegitimate (or
extreme) for one group and not the other?
Art Kendall
Social Research Consultants
Zhicheng Lin wrote:
> Dear All,
>
> One question puzzling me: Suppose in the Exp we have two conditions A
> and B. Do you exclude outliers in group A and B separately or pool
> them together? I think it make more sense to do it separately if you
> assume that A and B are different. What do you think? Thanks.
>
> Cheers,
>
> Zhicheng
> --
> ******************************************
> Zhicheng Lin
> Department of Psychology
> University of Minnesota
> 75 East River Rd, Elliott Hall
> Minneapolis, MN 55455
>
>
> Email:
[hidden email]
> Phone: 612-625-2470
> Fax: 612-626-2079
>
http://zhichenglin.googlepages.com/> ******************************************
>
> =====================
> To manage your subscription to SPSSX-L, send a message to
>
[hidden email] (not to SPSSX-L), with no body text except the
> command. To leave the list, send the command
> SIGNOFF SPSSX-L
> For a list of commands to manage subscriptions, send the command
> INFO REFCARD
>
>
>
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Art Kendall
Social Research Consultants
Locked
