*3.3. Mixed Logit Model*

The mixed logit (ML) model is used to analyze the data collected in the choice experiment. The ML model (specified in the Equation (3) below) relaxes the independent of irrelevant alternative (IIA) assumption and allows individual variations in the attributes [45]. Meanwhile, the conditional logit model is typically used if the random terms follow independently identically distribution (IID) and assumes respondents having the same preference for the attributes. The likelihood ratio (LR) tests can be used to compare the two models [46]. If the null hypothesis that there is no difference between the two models is rejected, this indicates the ML model is more appropriate. In addition, a probability density function, *g*(*α*), is introduced for the coefficient of the presumed heterogeneous attributes. Namely, correlations between preferences are allowed and different respondents show different preferences for the attributes of enhanced mandatory labelling. The non-conditional probability *Pimn* of consumer *i* who chooses the *<sup>m</sup>*-*th* alternative in the *n-th* choice scenario can be ge<sup>t</sup> by calculating the integral of *g*(*α*) with respect to *α*.

$$P\_{\rm inn} = \int \frac{\exp\left(V\_{\rm inn}\alpha\right)}{\sum\_{m=1}^{M} \exp\left(V\_{\rm inn}\alpha\right)}\tag{3}$$

The model assumes that *g*(*α*) functions of all nonpaymen<sup>t</sup> attributes follow normal distributions. The price attribute with a fixed coefficient equals the given market price and the other two reference prices are slightly higher than the market price. The parameter α refers to scaled marginal utility for a mandatory labeling attribute or price, due to scale normalization. Therefore, we can only interpret the relative magnitude of the other attributes and statistical significance by the parameter estimates. The WTP can be calculated from the negative marginal utility divided by the coefficient (*<sup>α</sup>p*) of the price attribute [47]. Therefore, they are comparable across the results.

$$\mathcal{W}TP = -\frac{\alpha\_k}{\alpha\_p} \tag{4}$$
