Table 1.
Characteristics of different sources of food losses.
Table 2.
Models explaining the acceptability of different feeds, listed in order of their Akaike weights.
Table 3.
Models predicting support for the relegalisation of swill, amongst all respondents (n = 163).
Fig 1.
Estimates from the six models with the greatest weighting (summing to 100% of model weight) of how the acceptability of different feedstuffs varies according to their characteristics (e.g. whether or not they contain animal by-products, or their legality).
The variation between different feeds, respondents, and feed:job combinations is shown below the dashed line. Model weights are proportional to the size of the points. Error bars are 89% credible intervals.
Fig 2.
Comparison of swill and conventional feed.
Responses to the question: “Compared with feeding conventional grain- and soybean-based feed, heat-treated swill is:”.
Fig 3.
Support for the relegalisation of swill amongst pig farmers and other agricultural stakeholders.
Response to the question: “If the procedures were put in place to ensure the safety of swill (e.g. heat treatment was performed by regulated swill manufacturers), would you support the relegalisation of swill?”.
Fig 4.
Predictors of the support for the relegalisation of swill, among all respondents (n = 163).
The estimates plotted are from the five models with the greatest weighting (85% of model weight), where different colours are used for each model (listed in Table 3) and model weights are proportional to the size of the points. Error bars are 89% credible intervals. For clarity, the coefficients for age groups, which was included in two models, are not plotted here; these are shown in Figure J in S2 Appendix.
Fig 5.
Predictors of farmer support for the relegalisation of swill (n = 82).
The top three models shown had 55% of the model weight, and the structure of all models are listed in Table A in S2 Appendix. Error bars are 89% confidence intervals.
Fig 6.
Predictors of farmer willingness to use swill, if it were relegalised (n = 82).
The top two models shown had 60% of the model weight; all models are listed in Table B in S2 Appendix.