Data analyses
The recognition threshold was determined when panellists were able to recognise salty taste of the odd sample that predominantly exhibited recognisable difference from the other two samples. A series of each panellist judgments was tabulated with a sequence containing ‘0’ for an incorrect choice or ‘+’ for a correct choice, which was arranged in the order of judgments of ascending concentrations of soy sauce odour. As the distribution is typically skewed, a geometric mean rather than an arithmetic mean was used to measure the centre location of the distribution (ASTM E679-04). Therefore, the best-estimate threshold (BET) concentration of soy sauce odour for the saltiness recognition threshold was the geometric mean of the two lowest concentrations at which correct responses occurred and a predominantly recognisable taste was identified as ‘salty’. In this study, soy sauce odour may possibly elicit other taste qualities including umami; however, we were mainly interested in the salty taste quality.
The group best-estimate threshold (GBET) was obtained by the arithmetic average of summation of the logarithm with base 10 (log10) of the individual BET values. The standard deviation log10 provided a measure of the group variation. The arithmetic average of GBETs of three independent replicates was reported for the saltiness recognition thresholds in ppb of soy sauce odour concentration. The percentage of panellists recognising specific tastes of the odd sample that exhibited recognisable difference was calculated for each soy sauce odour concentration.
The difference threshold was determined when panellists were able to choose the sample, which was saltier than the other two samples. The BET concentration for the difference threshold was the geometric mean of the two lowest concentrations at which correct responses occurred. The GBETs of three independent replicates were calculated as above. Sample calculation for the individual best-estimate (BET) and group BET (GBET) recognition thresholds of soy sauce odour (ppb) in 0.03 mm NaCl solution obtained from one independent replication is shown in Table 2. For both recognition and difference thresholds, the relative standard deviation (%RSD = [standard deviation/mean] × 100) was also reported. %RSD expresses the spread of data with respect to the mean value; a lower RSD value indicates less variation of the threshold values among the three independent replications in this study.