Comparing Perceptual Noise in Rating Scales
Ennis, D. M. (2015). IFPress, 18(3) 3-4.
In the Southern District of New York in 2011, Church & Dwight Co. Inc. (C&D) sought a preliminary injunction barring The Clorox Company (Clorox) from continuing to air a commercial that implied that Clorox’s cat litter product was superior to C&D’s at eliminating cat waste malodor1 . This preliminary injunction was granted in early 2012. Among a number of legal and technical issues in the case was the implausibility that eleven trained panelists would report a score of zero in each of four replications to a previously sealed jar containing a carbon and cat waste sample. The rating instrument used was a 15-point malodor scale with 0.1 unit increments, effectively making it a 151-point scale and previous research from Clorox had shown that human subject ratings of cat litter with or without cat waste are quite variable. C&D argued that such a result is “highly implausible” given that the test used humans, which are known to be “noisy instruments” when it comes to olfactory perception. The court explained that this is because humans “for neurological reasons, perceive the exact same thing differently at different times and report the presence of olfactory stimuli even when they do not exist.”
Figure 1. (A) Two perceptual distributions separated by one standard deviation, d = 1. (B)Two distributions centered on the same means, but separated by two standard deviations, d = 2. The only difference between (A) and (B) is the size of the perceptual standard deviation.