THE PANELIST EFFECT – FIXED OR RANDOM?

Authors


ABSTRACT.

The theoretical and practical aspects of how panelists might be viewed are investigated. The situation where panelists are viewed as random selections (random effects) from a population of all possible panelists is compared to viewing panelists as the entire population of interest (fixed effects). The statistical implications of each viewpoint are investigated. Sources of variation related to the panel, panelists and the testing environment are discussed. An argument is presented concluding that in most situations panelists should be viewed as random effects. This allows results to be related to a larger population of prospective panelists.

Ancillary