Role Models in AI: Christine Pierce
Christine Pierce is the Senior Vice President of Data Science at Nielsen, where she leads the team of data scientists supporting Nielsen’s audience measurement products. With over 17 years of experience in quantitative research, Christine is an expert on demographic data and audience measurement. This expertise influenced her efforts to have the citizenship question removed from the 2020 US Census, not as an issue of policy, but of sound science.
Holding a Bachelor’s in English Literature, a Master’s of Public Policy, and a graduate certificate in Applied Statistics, Christine is vocal about allowing your interests to guide your path toward success. Christine is also awed and inspired by the current generation, and their drive to create a positive future for themselves, and the world.
Read on to learn more about Christine’s thoughts regarding inclusivity and education in the future of AI, and her data science work at Nielsen.
As told to Nicole Halmi of AI4ALL by Christine Pierce; edited by Camryn Burkins.
What do you do in your current role as Senior Vice President of Data Science at Nielsen?
As SVP of Data Science, I lead a team of scientists who support Nielsen’s audience measurement products: our television ratings, radio ratings, etc. My team has expertise in multiple fields, including quintessential data science fields — predictive modeling, machine learning, and computer science. However, I also have a number of team members with expertise in more traditional fields, like sampling, survey research, and qualitative methods. Part of my team spends time developing models, and the other part of my team spends time creating measurement panels, which serve as truth sets, or control groups of data, for our models. We have quite a few different types of talent across the organization, and it’s my job to make sure that we’re doing the best work.
You’ve been an integral voice in the push to have the citizenship question removed from the upcoming 2020 US Census. From a data scientist’s perspective, can you talk about why the citizenship question is so significant? What are some of the potential unintended consequences of flawed data, both from a Census perspective and in other applications?
Fortunately, there will not be a citizenship question on the 2020 Census, which was definitely the best outcome for data science, industry, and society in general. The Census is the benchmark for so many different datasets. The governmental uses of census data — apportionment, electoral college seat assignment, determining the number of congressional spots each state gets — are widely understood. But the data is also used by businesses to make really critical decisions, with significant impacts on economic development. Businesses use census data to determine where they should open new stores, open factories, locate employee sites, and build new homes. More relevant to the AI arena, however, is that this data serves as the truth set.
We live in a world where data is plentiful, but having representative data is still a rarity.
Having a dataset like the Census allows us to have an accurate view of the entire US population, down to city and even neighborhood-level estimates, to understand what the real population looks like. Only then can you take disparate datasets and determine how to combine them to make them representative. You have to have a benchmark like the Census to be able to understand what biases exist in datasets. If you have data that hasn’t been calibrated to a good data set like the census, you can make bad decisions. You can under-invest in an area, or under-invest in a segment of the population, because you think that it is smaller than it is.
You have a BA in English Language/Literature, a Masters of Public Policy with a concentration in Quantitative Methods, and a graduate certificate in Applied Statistics. Can you tell me a little about your journey from English to statistics? How did you get into working in data science at Nielsen? What advice do you have for students interested in multiple fields or transitioning between seemingly dissimilar industries?
I think that data science is interdisciplinary. Data science is a relatively new field, and you can find it in statistics, computer science, and a lot of different areas of higher learning. The field has really grown from a multidisciplinary approach.
In terms of my personal journey, I’ve always had an interest in social science, our culture, and storytelling. In college, what originally drew me to English literature was learning how literature and stories reflected our broader society and societal trends across the ages. I was always really good at math, but I never understood how to apply it to my interest in social science and human behavior until I did some post-baccalaureate work in statistics. When I took that first statistics class, it just clicked immediately.
Statistics allowed me to see how powerful it is to study human behavior — to understand people, their attributes, and how they change over time.
Economics and statistics provide the tools to quantify things that are really social in nature, and having actual tools to quantitatively study human behavior was really impactful for me. I got a master’s degree in public policy with a concentration in quantitative methods, and after graduating I was looking for jobs where I could use that applied quantitative method approach. I ended up finding a job at Nielsen. Over time I went back to school and got my certificate in advanced statistics while still employed at Nielsen, continuing to increase my education as I moved through the company.
My advice to others is to just follow your interests. I really think that what matters most is waking up each day and doing something that keeps you interested. If you are doing something that keeps you interested, then you’ll figure out how to do it.
What are some of the things people should be doing now to create a positive future for AI and data-driven technologies more broadly?
There have been so many instances of using artificial intelligence and ending up with a discriminatory output. These outcomes come from datasets with unknown biases in the data, or even known biases in the data that the researcher or modeler didn’t do anything to account for.
The work of making sure that the industry is inclusive, both in terms of having multiple disciplines represented, and in terms of having people from different perspectives represented, is the best way to make sure that the models we create don’t have unintended discriminatory effects.
It’s also important that as things move forward we always maintain a bit of skepticism. Where did the data come from? Is the output received the output intended? Is the output discriminatory? If so, why? These questions go back to why data from a representative source like the Census is so important. Some skepticism allows you to really examine the data that goes into models and can help prevent discriminatory outcomes.
Do you have any role models?
I’ve been really impressed with the generation that’s currently high school- and college-aged. Anytime I meet with anyone from that generation, I’m blown away by how insightful they are, but also their yearning to improve the culture. We saw in Florida, following the school shootings here, young people coming together to improve their society. And you’ve seen a lot recently with climate change too, that some of the most outspoken advocates are people from a really young generation. It’s been very inspirational for me to see people who have limited experience from an age perspective, willing to say, “This isn’t what I want for my future.” A lot of the people that I look up to now are actually of the younger generation.
What has been one of the proudest or most exciting moments in your work so far?
One of my proudest moments is the stand that Nielsen took on the citizenship question. We are one of the few private, for-profit companies that took a stand. We didn’t do it to be political. We did it because our business is data, and our businesses is being the standard of truth for audience measurement. It was an issue of good science.
About Christine
Christine Pierce is Senior Vice President of Data Science at Nielsen. She leads the team of data scientists who support Nielsen’s audience measurement products, as they tackle the industry’s toughest challenges including audience fragmentation and cross-platform measurement. Her accountabilities include developing and maintaining methods to integrate digital “big” data with traditional media panels.
Christine has over 17 years of experience in quantitative research and is an expert on demographic data and audience measurement. Christine is a frequent speaker on the business uses of U.S. Census Bureau data. She has authored articles, spoken at conferences, and participated in media interviews on the importance of an accurate 2020 Census to the private sector.
Christine is involved in several professional organizations including the Advertising Research Foundation (ARF) and she frequently represents Nielsen at research conferences. She is a graduate of Nielsen’s Global Leadership Program and was a recipient of the COO’s Leadership Award and has her Executive Black Belt certification. Christine earned her Master of Public Policy degree from the University of Minnesota and a graduate certificate in Applied Statistics from Pennsylvania State University.
