Date:         Wed, 14 Apr 1999 13:28:58 -0400
Reply-To:     "Karl \"Brett\" Lloyd" <lloyd@grove.iup.edu>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         "Karl \"Brett\" Lloyd" <lloyd@POP.PITT.EDU>
Subject:      HELP!!!!!!-handling lack of variance in...
Comments: cc: "Karl \"Brett\" Lloyd" <lloyd@grove.iup.edu>
Content-Type: multipart/alternative;

Hello
 
I have a rather perplexing question about a 2 by 3 ANOVA design and if there is a way to handle it in SPSS8.0.
 
Briefly - I ran an experiment with 2 learning conditions (A and B) and 3 types of learning materials (1,2,and 3). Performance was measured by task performance via completion time and via whether the task was completed or not (binary measure). In one of the rows (one of the learning material treatments) none of the subjects completed the task! Additionally they used the entire amount of allotted time to try to complete the task (except for a small number of individuals who became frustrated and quit early).
 
My difficulty is that this leaves no variance in completion time or completion rate for the subjects who received a certain type of learning material -i.e., I have one row without any variance. 
 
I do not see how this can be run as an ANOVA (because of the lack of variance in the one row and the violations of assumptions this causes to the ANOVA model). Is there a non-parametric or other method to apply to analyze these results in SPSS under these circumstances.
 
Any suggestions will be greatly appreciated!!!
 
Karl Brett Lloyd
 
lloyd@grove.iup.edu
kblloyd@icubed.com
412-856-8571   
 
[text/html]