I would like to know whether I should regard the following factor as a fixed or random effect in an ANOVA.
The situation involves a panel of consumers who are blind tasting 3 products. Each panellist tastes 3 replicates of each product making 9 tastings in all. They then have to rate each product on various attributes.
When doing an ANOVA, I am interested in seeing if the panellists are discriminating between the 3 products. I also interested to see if panellists are giving similar ratings or different ratings. Thus the 2 effects that I am testing in the ANOVA are product and panellist.
Clearly Product will be a fixed effect but what about panellist? If one is taking a random sample of consumers then clearly it will be a random effect. But in this case the panel is re-used on many occasions so that the panellists can be trained to identify undesirable flavours and off-notes. In this case I think you should regard the panellist effect as a fixed effect. My question is do you agree with me?
Thank you
Nigel Marriott
-----------------------------------------
Email provided by http://www.ntlhome.com/
|