“…Adding a decoy potentially changes the mean (Helson, 1964), the range (Parducci, 1965(Parducci, , 1995Volkmann, 1951), and the density or frequency (Krumhansl, 1978;Parducci, 1965Parducci, , 1995 of the distribution. Adding a decoy also potentially changes reference points (Frederick & Loewenstein, 1999;Holyoak & Mah, 1982), likely groupings between items (Tversky, 1977), and the relationships between items (Choplin & Hummel, 2002). Because theories of how these factors affect attributevalue evaluation often make conflicting predictions, decoy effects allow us to pit theories of attribute-value evaluation directly against each other.…”