What statisticians learn when they swap stories

Aspeq is committed to supporting its people to grow, learn, and stay at the forefront of their fields. That means investing in professional development, encouraging innovation, and creating space for our specialists to connect with peers and emerging ideas.

Earlier this year, Aspeq’s Research and Statistical Analyst, Alana Docherty, attended the 2025 New Zealand Statistical Association Conference in Dunedin. In the piece below, Alana reflects on the conference as a valuable opportunity to learn, contribute and connect with the wider statistical community.

By Alana Docherty

I was lucky enough to attend the 2025 NZ Statistical Association Conference in Dunedin this week. This was my second time attending, and I always enjoy the opportunity to hear from others in my field, especially as I’m the sole statistician at Aspeq.

While I genuinely enjoy being the go-to person for statistics in my role, it was refreshing to connect with others in similar positions and talk through the ideas, challenges, and questions we each face in our work. There’s real value in stepping outside your organisation and stress-testing how others are thinking about the same problems.

This year’s conference had a strong focus on Bayesian statistics (prediction and probability) and biostatistics - neither of which I work in directly, but I still took a lot away from the sessions. Some of the standout talks included:

• Ting Wang’s presentation on geomagnetic storms in Dunedin, which explored how strong auroras can affect aircraft and power plants. By predicting these events, power plants can be shut down to prevent damage - a great example of statistics informing real-world risk mitigation!
  
  
• Richard Arnold’s talk on the rise of disinformation, including research being done at Victoria University of Wellington on climate change beliefs among farmers and conversations with flat-earthers to better understand why disinformation is increasing.
  
  
• Lisa Chen’s talk on AI and Simpson’s Paradox, where aggregated data can show one trend while disaggregated data shows the opposite. I’ve only encountered this once in my career so far, but it’s a powerful reminder of how easily data can be misunderstood, and how AI systems may struggle with these nuances.

Speaking of AI, I was lucky to be invited to sit on a panel discussion on AI, alongside Gary Dunnet (Stats NZ), Andrew Gray (University of Otago), and Priya Parmar (University of Auckland), hosted by Lisa Thomasen from Fonterra. This was my first time on a panel, and I thoroughly enjoyed it.

Within the statistics community, there’s a lot of shared concern about AI - particularly around job value, misuse of data, and hallucinations in AI-generated outputs. Several panellists spoke about having to correct or redo work that AI had attempted to replace. As statisticians, we’re trained to question everything, and that mindset feels more important than ever.

At the same time, I think AI can offer significant opportunities. Our roles are evolving, and having a strong understanding of the mathematics underpinning AI models allows us to add value in ways AI can’t replicate. Personally, being able to use AI to translate between coding languages has freed me up to focus more on ideas and mathematical thinking, rather than syntax.

The panel created space to share concerns, challenge assumptions, and reassure one another. AI still has a long way to go before it could replace statisticians, and perhaps the real question is what we offer that AI never will. One question that stuck with me was: as statisticians, how do we front-foot the AI movement to protect data and analysis within our organisations? It’s something I’ll continue thinking about.

Next year, I’m hoping to present an idea of my own at the conference (not AI-related). Stay tuned.