Search

This thesis explores consumers' purchasing behaviour under risk and uncertainty. Specifically, the role of subjective beliefs, risk and ambiguity attitudes, and information. I focus on food purchasing decisions with uncertainty about the quality generated by food fraud. The first essay uses an artefactual field experiment to explore bidding behaviour under risk and uncertainty in homegrown value (HGV) and induced value (IV) settings using Becker-DeGroot-Marschak mechanism (BDM) and second price Vickrey auctions (SPVA). It investigates if deviations from theoretical predications can be ascribed to bidding behaviour that is inconsistent with standard theories of decision making (i.e. Expected Utility Theory and Subjective Expected Utility Theory). In the HGV setting, non-monetary lotteries are used. Participants bid for a high-quality seafood product, but there is a chance (known or unknown) they receive a lower quality seafood product. The second essay uses the artefactual field experiment to explore consumers' willingness to pay (WTP) a price premium to avoid food fraud. The influence of subjective probabilities, risk and ambiguity preferences on premiums are also investigated. Results show that consumers are willing to pay a price premium to avoid food fraud and purchase a fully authentic PDO fish product. The third essay explores if consumers update their beliefs on food authenticity and their preferences for potentially fraudulent food after receiving information. Whether consumers are Bayesian updaters and if belief updating is correlated to a shift in their WTP is tested. I examine if changes in beliefs and WTP depend on whether news is perceived as good or bad and on the source of information. An online survey is split into two groups: one receiving information about food authenticity from other consumers, the second from experts.

This thesis aimed to provide comprehensive guidance for food fraud prevention and mitigation in the beef industry by analysing historical incident of food fraud in the beef supply chain, determining types of fraud which threatens the beef supply chain, assessing and prioritized the risk associated with these types of fraud, and exploring where there are gaps in current detection methods and so that scientific research can aim its investigations in the most needed areas. This amalgamation of industry need with academic research has been created to give the beef industry the tools it needs to protect its product and by extension, its reputation, and its consumers from the effects of food fraud using knowledge found through academic research and scientific exploration.

Food fraud has been a problem for both consumers and producers for millennia; the Roman Empire instituted some of the first legislation to protect against practices which could cause illness or death to its citizens. Governments in the years since have instituted a wide variety of safeguards to protect lives and to ensure socio-economic stability by permitting access to safe and nutritious food, concurrently, criminal entities have developed new and increasingly complex ways of defeating such safeguards. Producers are increasingly looking to the latest analytical technologies to both demonstrate that their products are safe and nutritious, and to provide a degree of protection against criminals who could cause significant reputational harm to producers. The aim of this thesis was to determine whether ambient mass spectrometry platforms could firstly be used to undertake the types of authenticity and quality confirmations that industry desires. The first project was to assess whether rapid evaporative ionisation mass spectrometry could be used by a major Northern Ireland poultry producer to confirm the authenticity of each poultry product they produce, primarily to confirm that their premium product offerings such as organic and free range were correctly labelled and contained the correct product. Six different poultry production systems were analysed by means of an electrosurgical knife and a bi-polar meat probe connected to a high-resolution quadrupole-time of flight mass spectrometer, which together form the rapid evaporation ionisation mass spectrometry (REIMS) system. Principle component analysis (PCA), linear discriminant analysis (LDA) and orthogonal partial least squares-discriminant analysis (OPLS-DA) chemometric models were built which demonstrated clear differences between some of the different production systems, most notably allowing those using different breeds of bird to be identified. The correct classification rates for REIMS varied from 69.22% to 100% depending on the modelling approach taken - a hierarchical approach based on separating out birds by breed then by production system was found to be much more effective than a single six class model which attempted to separate birds purely by production system first. The REIMS project also investigated whether the age of the poultry could be assessed by means of a chemometric analysis - using the same analysis techniques as before, it was determined that REIMS could determine with 100% accuracy the age of the poultry product, taken together with the significant potential in identifying some production system characteristics and poultry breeds, REIMS shows significant potential for confirming labelling on products or in detecting relabelling frauds. The second project was to investigate direct analysis in real time (DART) mass spectrometry as an alternative to REIMS for poultry production system identification. The analysis required a complex and lengthy sample preparation process, which would restrict the use of the system to an analytical laboratory, but it was found that DART when combined with a high-resolution Orbitrap mass spectrometer could identify poultry production systems with a 100% cross-validation accuracy when using the same PCA and OPLS-DA chemometric modelling techniques used for the REIMS analysis. The final project was to further investigate DART mass spectrometry and assess whether it could potentially be used to detect an organic v conventional mislabelling fraud in the vegetable sector. Leeks were chosen for this analysis and three sections from each leek (leaf, stem and root) were analysed by using a DART ion source connected to a compact and potentially field deployable single quadrupole mass spectrometer. It was found that when freeze drying and extracting the samples in methanol, it was possible to identify organic and conventional leeks with a correct classification rate of between 95% and 100% depending on which section of the leek was analysed. The work within this thesis confirms that ambient mass spectrometry could potentially be used by producers and government safety bodies to determine whether products are incorrectly labelled with regards to their quality or dates of production.

The herb and spice industry is increasing in value and popularity as the consumption of herbs and spices continues to grow. Valuable condiments as well as long and complex supply chains offer the motivation and the opportunity for fraudsters to carry out economically motivated adulteration (EMA) and pose economic and public health risks in doing so. Spectroscopy in conjunction with chemometrics, allow for nontargeted qualitative methods to be developed to detect the ever-evolving criminality of adulteration in herbs and spices. In this study, sage, paprika, garlic and black pepper were prioritised by industry experts in relation to their risk of adulteration as determined by herb and spice industry. The carefully selected adulterants in this study were analysed alongside the four herbs and spices for the development of classification models. The adulterants were morphologically similar to the herbs and spices or could be blended in ground form. Green plant cuttings, spent paprika and white powders were analysed alongside sage, paprika and garlic respectively. Black pepper adulterants included extraneous parts of the plant as well as papaya seeds and chili powder. The spectroscopic instruments used in this study were near infrared (NIR) and Fourier transform infrared (FTIR). The sage method, developed using FTIR, resulted in binary and multiclass supervised models with a measurement of fit (R 2 ) of 0.978 and 0.952 and a measurement of prediction (Q 2 ) of 0.975 and 0.936 respectively. Following validation, the area under the curve (AUC) of 1 indicated excellent method performance when a validation set was analysed. Additionally, it was found that sage could be classified according to its species. The detection of spent material in paprika was investigated with both NIR and FTIR instrumentation. Although separation was observed between spent material and paprika, these methods detected adulteration consistently from 50-90% using NIR spectral data and 40-90% using FTIR spectral data but was less accurate at low level adulteration. Correct classification rates for authentic paprika were 100% and 83.3% for NIR and FTIR respectively. An additional step was carried with the addition of Sudan dye and it was found to have no effect on the outcome of the results when detecting spent paprika adulteration. Binary supervised classification models were created following the collection of spectral data from garlic and its adulterants on NIR and FTIR. Validation of the methods resulted in 100% correct classification for all authentic garlic samples and all adulterated samples in the 20-90% range for both spectroscopic techniques. Also, following further investigation, it was found that extraneous parts of the garlic plant could be separated in an unsupervised model. Advances in technology mean that handheld and portable spectroscopy is being developed and adapted for the detection of adulteration. In this study, black pepper and the adulterants were all analysed using two portable NIR instruments and a benchtop NIR instrument. A comparison was carried out between the three methods and it was observed that the portable instruments were comparable in terms of performance with the benchtop NIR. All three methods had a correct classification rate of >90% for authentic black pepper and 100% correct classification for 20-90% level adulteration. These screening methods for EMA are vital to ensuring the authenticity of herbs and spices for the protection of the industry and the consumer. As these are non-targeted methods, they can also cover a wide range of adulterants, including those that are not yet known.