Sensory evaluation
At the end of the experiment, all fish were killed, filleted and
washed, then they were steamed for 3 min. First, these fish were
evaluated for differences in the quality of fish fillet between the
control and experimental groups by a triangle test (2 identical
controls and 1 experimental group for each experimental sample)
(Meilgaard et al., 1999). The control sample was snakehead that was
bought at the local market. The two experimental samples named
‘trash-fish’ and ‘formulated feed’ were not known to the panelists.
Three samples (1 experimental and 2 control samples) were served
together to nine panelists who were asked to identify the odd sample
(one out of three samples). If less than 6 out of 9 detected the
odd sample correctly, we determined that there was no significant
difference and therefore no need to conduct a sensory test. A paired
test (Meilgaard et al., 1999) was run to determine if there was any
difference in sensory attributes even if they were minor. On the
other hand, if 7 out of 9 people detected the odd sample correctly,
there was a significant difference at P < 0.01 or 6 out of 9 P < 0.05. In
this case, a comprehensive pair test was done on appearance, texture
and taste for intensity and liking (1–9 points). The intensity