Statistical Modeling of Seafood Fraud Highlights Uncertainties in Products From Metro Vancouver, British Columbia, Canada: Revisiting Hu et al. (2018)

Seafood misrepresentation, encompassing product adulteration, mislabeling, and substitution, among other fraudulent practices, has risen globally over the past decade, greatly impacting both the loss of important fish species and the behavior of human consumers alike. While much effort has been spent attempting to localize the extent of seafood mislabeling within the supply chain, strong associations likely existing among key players have prevented timely management and swift action within Canada and the United States in comparison to European nations. To better address these shortcomings, herein frequentist and Bayesian logistic generalized linear models (GLMs) are developed in R and Stan for estimation, prediction, and classification of finfish product mislabeling in Metro Vancouver, British Columbia, Canada based on product source (grocery store/restaurant/sushi bar), state (cooked/raw), appearance (modified/plain), form (chopped/chunk/fillet/whole), and color (light/red). Obtained results based on odds ratios and probabilities for 281 seafood samples from 92 grocery stores, 82 restaurants, and 107 sushi bars paint a grim picture and are consistent with general trends found in past studies, especially when products have been altered. In addition to estimation and prediction, analyses are extended through applying regression models to classify four collected seafood samples whose labeling status (correctly labeled or mislabeled) was not known a priori. While all samples were deemed correctly labeled, point and confidence/credible intervals differed markedly, indicating considerable uncertainty as to true mislabeling rates. This work paves the way to rapidly assess the current state of knowledge surrounding seafood fraud nationally and on a global scale using established pedagogical statistical methodology.

Practical Applications

Seafood is one of the largest global food commodities but also among the most frequently subject to fraud, including substitution, adulteration, and mislabeling to evade tariffs or conceal illegal, unreported, and unregulated (IUU) fishing. Consumers often cannot verify what they purchase, especially when identifying features like skin, fins, and scales are removed or products are heavily processed through cooking or canning. Despite its prevalence, the timing, location, and mechanisms of seafood fraud within supply chains remain poorly understood. This study applies statistical regression modeling to investigate and address these uncertainties.