2025 Presidential Address
05/16/2025
Greetings, fellow AAPOR members.
Last year, our conference chair, Gina Walejko, showed remarkable foresight when she chose the theme for this year’s conference: “Reshaping Democracy’s Oracle: Transforming Polls, Surveys, and the Measurement of Public Opinion in the Age of AI.” At the time, I was so deeply immersed in AI developments that I focused mostly on the last part of the title, “in the age of AI.” But with the developments over recent months, that has definitely changed.
As you can imagine, preparing my remarks has been challenging.
It has been challenging because I cannot help but think of the people who are not here because they lost their travel privileges. No matter how generous the AAPOR community was able to be with travel support, they had to stay put.
It has been challenging to think of those in our community who lost their jobs and, despite the newly created matchmaking and networking features, haven’t found their next employment yet.
It has been challenging because I anticipated—correctly—that this conference, thanks to all of you, would showcase a phenomenal wave of innovation. Today we are transforming polls, using surveys, leveraging data collection methods, and harnessing AI in ways we couldn’t have imagined a year ago.
So before I dive in, let me take a moment for a heartfelt shoutout — to all the AAPOR first-timers (384 of you), and to the AI start-ups who joined us this year: Thank you for daring to be here, for bringing fresh ideas, and for enriching our community.
And to the AAPOR regulars: Thank you for embracing this new wind — for welcoming new perspectives, new challenges, and new opportunities.
Now, I want to use this moment to share three thoughts with you — wrapped in what I like to call three five-minute tales. You’ll understand why in just a moment.
Tale 1: A personal reflection on change
Three years ago, just before AAPOR, my mother passed away. Two years ago, shortly before AAPOR again, my father passed away. On the day of his memorial service, I learned I had been elected Vice President of AAPOR. It was an emotional moment layered on top of another — and I remember thinking, “Well, maybe this new role will help me refocus — and hopefully, it will be filled with opportunity and joy.” Boy, was I wrong… and boy, was I right.
What followed are stories I’ll share today — in those five-minute tales. Why five minutes? At my father’s service, I gave my first-ever speech without slides. And, in a strange way, I found joy in the challenge of packing meaning into a short, spoken story. It even sparked the idea of journaling my work and life in five-minute reading packages — perhaps to publish someday.
Some of you might remember Laura Young, the student who wrote the beautiful AAPOR newsletter piece about last year’s AAPOR conference. She gave me some great writing tips — which helped a bit with this year’s newsletter entries — but I’m still chasing that habit of regular five-minute reflections. So here we are — off to a new start (and yes, already two minutes in).
Why begin with such a personal story — about loss, and about the challenge of making time to develop new habits? Because I believe many of us have faced something similar over these past years. And sharing these moments — relating to each other — is the first step in navigating stormy times together.
As Ezra Klein and Derek Thompson write in their new book Abundance: “It is hard to build trust without nearness.”
I like that sentence. That need for nearness — for connection — is exactly why we’re here, in person. The world around us is changing — technologically, societally — and whether we choose to embrace these changes or not, we are confronted by them. We must find ways to adapt. And I believe we can only do that together — by trusting each other, and by trusting that we all share a common goal: to ensure that valid, transparent data about behaviors, attitudes, and characteristics continue to empower informed decision-making in our society.
I am encouraged to see on the Census Bureau’s website the strong headline: “Measuring America’s People and Economy: We Believe in the Power of Quality Data to Impact Public Life.” We do! Thank you.
Tale 2: The rise of AI as partner or disruptor
When Gina and I brainstormed this year’s conference theme, AI quickly took center stage. How central it will become is still unfolding. But this week has made one thing clear: AI is no longer just a topic of discussion. It’s becoming a partner in many respects.
AI has become a partner in data collection. We have seen talks (e.g., the idea group led by Darby Steiger) showcasing how AI bots help conduct open-ended interviews.
AI has become a partner in coding, whether it’s open-ended responses or entire scripts. Not only is sentiment captured, but answers get classified into occupational coding schemes or types of diseases. It might even help us move beyond the straightjacket of closed-ended questions altogether.
And we are not the first to realize this. Interestingly, researchers in natural language processing (NLP) have discovered the survey arena as an application area — with very little to none of the survey methodology literature cited. Some of these efforts can help slash costs — if we focus on solving certain puzzles or developing leaner models. This is certainly something the idea groups on Tuesday, led by James Wagner, focused on.
AI could also eliminate nonresponse — after all, we can generate as many synthetic personas as we like.
But here is one of the most critical questions: Would AI solve nonresponse bias? So far, the answer is a clear no.
Just last month, at the International Conference on Learning Representation (ICLR) in Singapore –one of the top three Machine Learning Conferences with 10.000 attendees— a workshop on Bidirectional Human-AI Alignment asked: How do we ensure AI systems act in line with human intentions, ethics, and societal values? Or, in AAPOR terms: How do we make sure that AI-generated synthetic or imputed data genuinely reflect the world we live in — so that the statistics we produce remain valid and reliable?
This is the cutting edge of what’s now called data-centric AI. And as Danqi Chen, one of the keynote speakers at ICLR, put it: “Data quality isn’t a technical afterthought. It’s a critical pillar of AI development — too often overlooked.”
Here’s where I think we come in. The AAPOR community brings decades of expertise in evaluating data quality, in knowing that data collection is not just engineering — it’s a scientific craft.
Two opportunities stand out.
First, we could guide AI to understand how human values and opinions shift over time — not as static facts, but as evolving narratives across diverse populations. Think of the roughly 7 million variables from over 21,000 studies at ICPSR alone. Now imagine transforming that wealth of information into structured pre-training data.
Second — our studies could become benchmarks for AI. Why should LLMs be tested only on bar exams or math problems? Why not evaluate how well they grasp societal realities? The American Community Survey already does this through Folktables — now a go-to benchmark in AI research, cited in countless papers.
So let’s think bigger. Why shouldn’t AAPOR’s legacy—decades of expertise in measuring public opinion and understanding social behavior—help shape the next generation of AI models? This would involve not just using AI tools for surveys or data analysis, but also actively setting the standards for how AI should represent human attitudes and societal complexity. And think about the possibilities for a business pivot.
To think bigger, we need to do a few things.
Most immediately, we need to make the larger research community aware of the richness of our data treasures and the richness of our knowledge. After all, our deep knowledge allows us to evaluate the quality of such data (as we all know, not all of it is flawless). We are able to identify sampling and non-sampling errors in data collection – not only in surveys but also in data use for AI training and benchmarking.
Maybe we can even get the AI powerhouses to help us finance the so badly needed high-quality surveys that can also help with adjustments for nonprobability surveys and other data collection efforts in the citizen science space?
Tale 3: Adapting AAPOR to the next chapter
This involves moving back from AI to democracy’s oracle — and democracy itself.
In a democracy, we cherish that everyone has a voice. At AAPOR, we take that a step further — we ensure those voices are heard, by building the foundation that makes the invisible visible through rigorous data collection and interpretation. As Patricia Moy noted in her presidential address: For AAPOR, voice is both a privilege and an obligation.
But democracy also comes with a reality: Most votes win. AAPOR’s tradition is grounded in polls and surveys, helping news consumers, journalists, and youth understand the role public opinion plays in a democracy – not just with respect to elections, but also the significant issues we face daily. When we look around at today’s research landscape, dominated by computer science and AI, it’s clear that this AAPOR tradition is now the minority.
There are a lot of conferences where attendees present work related to what we do, which could mean we lose the next generation. Or, we can be proactive and decide to partner and offer joint workshops, like the aforementioned conference in Singapore, bridging ICLR and CHI (the human-computer interaction conference).
Why not AAPOR as the third partner?
Much potential exists on this front. Take, for example, the Conference on Language Modeling (COLM, not related to O’Muircheartaigh). COLM will meet this October in Montreal and just last week, they accepted a joint proposal submitted by Stephanie Eckman and other members of the AAPOR community as well as researchers from the NLP community. We will have two groups – one representing AAPOR and the other, the world of NLP – and they will run a workshop called … wait for it… “NLPOR” – “Natural Language in Public Opinion Research.”
What about continuing to bolster partnerships with WAPOR, with whom we held this year’s joint conference? As the US pulls back, there is a void in guidance to individual countries on how to get the best possible information out of a multitude of data collection modes and data streams. WAPOR and AAPOR will be needed partners for this.
And what about institutional partnerships? Could AAPOR become a strategic partner to Meta, Anthropic, Amazon, OpenAI — no pressure to anyone in the room, of course — not just as a data provider, but as a guardian of data quality and societal relevance?
Could these collaborations sustain the infrastructure we need — not just for today’s market needs, but for tomorrow’s AI systems that must be aligned with human values, ethics, and realities?
As AI changes the nature of what the AAPOR community intellectually engages in against this landscape, we need to ask ourselves key operational questions. For example, can we still afford journal articles that take two years to publish? Can we afford not to offer archival publications at our conferences, while other fields accelerate ahead? Can we afford conferences that price out the very people — students, startups, innovators — who will shape the next chapter?
In her upcoming term, Jordon Peugh be asking the question: Are our current structures within AAPOR – the roles we elect, the processes we follow — still fit for the future we face? We will be revisiting our executive committees, with the roles assigned to them, their subcommittees, and task forces to ensure that they best serve AAPOR.
Let’s be clear: Change is hard. Changing a culture is even harder. But we’ve already begun. Thanks to Tristanne Staudt’s leadership, and with Jordon ready to carry the torch forward, AAPOR is moving — not just reacting, but proactively reshaping itself. I want to sincerely thank everyone who is part of this effort to keep AAPOR nimble, open, and future-ready.
The losses I mentioned at the beginning of the talk have left me uprooted, and in great uncertainty. It takes time to find new footing. I suspect we as an association will experience that too. Figuring out new rules and new tools, and realizing that what worked yesterday might not work tomorrow is disorienting. But honestly, it’s also exciting.
Because we’re not standing on shifting ground alone. We’re standing together — with the chance to shape, to adapt, and to actively design the future of our field. So when we gather in 2035 for AAPOR’s 90th conference, I don’t think we’ll just look back and say, “We survived the changes.” I believe we’ll look around and say, “We led them.” Together.
Frauke Krauter
2024-2025 AAPOR President