Difference tests. Wine consumption
habits for all participants are summarized in Table 2. Results
of the three triangle tests showed that consumers
could not detect a difference between wines bottled with
the different closures for either Chardonnay or Merlot
samples (Table 3). The power of the test was recalculated
post-hoc to determine if the number of actual test participants
was sufficient to reject the null. For all three tests,
the calculated power of the test was 90% with the
chance-adjusted percent of correct or proportion of discriminators
equal to 20%. The actual number of correct
participants was lower than the published table number
for Type I error, α = 0.05 (Lawless and Heymann 1999),
thus indicating the conclusion that consumers could not
detect a difference was statistically valid.
An estimate of the number of discriminators was calculated
for each triangle test (Table 3). Because the triangle
test is more difficult than other difference tests, a larger
perceptual difference is required to be a discriminator
(Lawless and Heymann 1999). Consequently, this more