Nicole Richardson is a Distinguished Professor of Professional Practice in Economics and Data Science. Prior to arriving at RU-N in September 2020, she spent more than two decades working as an applied behavioral economist for Fortune 500 and mid-sized private companies, managing data science teams while specializing in risk, pricing and customer analytics. She has crunched numbers and analyzed trends in a variety of fields ranging from healthcare, advertising and retail/marketing to banking, pharmaceuticals, insurance and, most recently, tech, as Senior Director of Data Science for Audible.com.
We sat down with her recently to learn more about her career, the ins-and-outs of her work, and what inspired her to move into teaching.
You got into Economics quite young, didn’t you.
Yes. I decided to major in economics while in high school. I love Economics because it combines my love of history, political science and mathematics. And my data science skills enable me to explore questions within and outside the field of Economics using big data.
You crunched and analyzed data for decades in many corporate sectors. What are some of the common threads that ran through your work?
The common thread were decisions made by agents—essentially game theory, where players or agents tried to optimize their decision(s) based on the information and constraints available and known at the time. I enjoyed decomposing decision-making by companies, hospitals, consumers and the government. I used data to investigate the who, why, when, how and what of decision-making. Fundamentally, I am eternally curious about the world around us and the people who inhabit it. When possible, I leveraged my understanding of behavioral economics, domain-industry knowledge, and the results of my analyses to improve my company’s decision-making and influence its strategic direction.
Could you offer some examples?
When I was in advertising, my focus was on brand management and strategy. I analyzed the link between brand and pricing across seven competing products. The real question was, Which type of customer, based on demographics, buys each of the products? Are some or all of the products sold consistently at the same price? If we lowered one price and kept all others at their higher prices, would customers defect from their brand and substitute another? I discovered that the same type of customer repeatedly purchased one of the products regardless of price changes, while the other products were influenced by sale prices and store location. So, I recommended that the product that exhibited brand loyalty should never be placed on sale, since that loyalty supports an inelastic price strategy.
The pervasive impact of algorithms on our personal lives and society means that every SASN undergraduate student should be aware of how data, data collection and algorithms currently, and in the future, impacts their lives.
And your second example?
In healthcare policy, I investigated Medicare and Medicaid reimbursement payments. The decision that required decomposition was, How did the federal government reimburse different types of healthcare facilities for the same medical procedures? This appears to be a simple question, but it’s complex. First, I explored the healthcare facilities: How many different types are there? Do they offer the same types of medical services? Where are they geographically located? Do different types of patients frequent different types of facilities? Do certain facilities perform a small portfolio of medical procedures? What are their relative costs? Does geography impact costs? Does it impact reimbursement? If so, how?
In short, the federal government started with a core reimbursement strategy when Medicare and Medicaid began, then added adjustment factors to account for differences in facilities, geography, patient-type mixes and so on. The governing agency produces annual rules and regulations that outline reimbursement payments. These rules were and still are based on data submitted by every medical facility that participates in their program. Economists, policy analysts and, more recently, data scientists and machine-learning engineers analyze data and develop models to understand costs and optimize reimbursements—in this case, while rewarding positive medical outcomes.
This is highly analytical work.
Yes. I enjoy decomposing interesting and impactful questions into bite-size questions, then using data to explore the bite-size questions and weave the results of my models and analyses together to shed light on the big questions, like What is the optimal federal medical-cost reimbursement strategy that results in an equitable distribution of healthcare with near-optimal medical outcomes? That was my question actually, not my company’s.
After such a long and successful career at companies such as AIG, JP Morgan and Audible.com, what inspired you to get into teaching, and how did that come about?
As the growth of what I refer to as “silent algorithms” reached an audible pitch in our lives, determining where we could live based on credit scores, what we paid for insurance, access to capital, where we could work, and what we see as output from internet searches, I pondered the historic lack of demographic and academic diversity in data-science teams throughout my career and the potential impact that this continued lack of diversity would have on future algorithms. Hiring, coaching and mentoring one to three data scientists or analysts at a time was not going to bring about the kind of change that I wanted to see in data-driven fields.
I pondered the historic lack of demographic and academic diversity in data-science teams throughout my career and the potential impact that this continued lack of diversity would have on future algorithms.
Surely not. So, what then?
I was approached by my Audible colleague Monique Jones and then RU-N faculty member Dr. Lucille Booker to consider hiring an RU-N undergraduate as an intern on the Audible Data Science Team. As the co-founder and a member of the data science leadership team, with a personal mission to diversify data science, I was presented with an opportunity to open the door for a person of color and develop a connection to RU-N, the most diverse university campus in the country. After facilitating the first undergraduate intern at Audible, RU-N approached me to consult and help re-design a Data Science minor course, Everyday Data. That re-design led to my Rutgers appointment beginning September 2020.
What courses are you been teaching this semester?
I am teaching two sections of our introductory course, Everyday Data, and one section of an advanced course, Deconstructing Machine Learning Bias). The latter was developed after participating in a speaker series at Rutgers Law School in Fall 2020.
Interesting. Are you working with undergraduates and graduate students?
I am working with undergraduates. The pervasive impact of algorithms on our personal lives and society means that every SASN undergraduate student should be aware of how data, data collection and algorithms currently, and in the future, impacts their lives.
How has your transition into teaching been—especially starting during Covid—and what have been some of the biggest challenges thus far?
Most data scientists prior to Covid had the privilege of working remotely. So, the Data Science minor curriculum was easier to adapt to an online format than other courses. But moderating my energy to suit the online format was a challenge, since I have an enormous amount of energy and enjoy interacting with students while looking over their shoulders at their code and writing on a whiteboard to aid their understanding key concepts. RU-N students are excited to learn about the field of Data Science and want to stay after class and chat.
Thanks for taking the time to talk with us.
It’s been my pleasure. Thank you.