My thesis is called Improving Bayesian Analysis of Consumer Panels.
Consumer sensory panels are the gold standard for evaluating food products, yet they often suffer from small sample sizes and sparse, ordinal data. Standard frequentist methods for analyzing this data (e.g. ANOVA, McNemar's test) often violate assumptions or lack statistical power. This thesis evaluates the efficacy of Bayesian Hierarchical Models (BHMs) with cumulative link likelihoods for analyzing sequential monadic consumer panels. We investigate whether domain-informed priors and complex correlation structures improve predictive performance and decision quality compared to frequentist baselines.
Link to thesis: https://lup.lub.lu.se/student-papers/search/publication/9222378
I was an alright statistician coming into my thesis, but I've really upped my game, both in translating "plain text" questions into mathematical tests and in translating the results back into human language. My job today is to automate exactly that, saving plant-based R&D teams the pain of speaking to a statistician.
Many people rush through their thesis just to get the paper with a title on it, but it's amazing what can happen when you treat it as a vehicle for impact rather than an obstacle to get past. Turning this into a company, Tasting Power, that can tangibly help develop better animal-free products is what inspires me most. I'm grateful to have a great collaboration with my former supervisor and current co-founder, and I'm equally grateful that I can increase animal welfare this way.
I had a few coaching calls and got access to resources on picking a topic, where the Weighted Factor Model was particularly helpful. The coaching calls were great for brainstorming and to provide accountability, which is especially important when the default is to pick the path of least resistance. My company wouldn't exist if Effective Thesis hadn't encouraged me to find a mission-aligned thesis supervisor.
The insight that you can "just do things" is one I wouldn't have applied without Effective Thesis. They broadened my horizon for what a thesis actually means, what it provides to the world and what it gives you for your career.
On a more concrete level, the frameworks for prioritising between different potential fields and thesis topics helped me eliminate a lot of uncertainty early on, so I could focus my energy on the right direction rather than agonising over which path to take.
"You can just do things." I emailed around 100 companies, random people who'd never heard of me, pitching my thesis idea. No one's going to scold you for audacity, but limiting yourself can leave enormous value on the table, both for your own growth and for those you want to help with your career.
That said, here's something I learned from my teenage years doing door-to-door fundraising for UNICEF and Greenpeace: it takes 100 knocks to get 10 talks and 1 deal. For finding a thesis topic and supervisor, the same logic applies, and it only needs to work once.
