Survival analysis protocols (Hough et al. 2003)
stipulate that only consumers who accept the control
sample should be part of the analysis. This reduced the data
analyzed to that of 48 out of 68 (20C), 45 out of 72 (25C)
and 76 out of 103 (30C) respondents. A relatively high proportion
of these remaining respondents (13 out of 48, 12
out of 45 and 22 out of 76 consumers tasting bars stored at
20, 25 and 30C, respectively) accepted all samples including
those stored the longest, suggesting that either they did not
notice or did not mind the subtle differences in sample
quality.