Event Recap
Wharton People Analytics Conference 2019

04 Apr 2019
Event Organizer: The Wharton School, University of Pennsylvania
Event Link
Tagged as: Attending, Philadelphia, PA

By Kristin Mueller - 25th April, 2019

PLASTARC is proud to have been a part of the Wharton People Analytics Conference (WPAC) for the last five years. The two-day event brings together experts and practitioners from a range of fields dedicated to understanding people through data. At last year’s WPAC, PLASTARC received the honor of "Most Innovative Use of Data" in the annual startup competition.

This conference began with the twin themes of building community and sharing ideas. Several of this year’s startup competition and white paper presentations centered less on individuals and more on the connections between them. This came through in innovative ways of analyzing social capital within a work environment and the clusters of connections between people, featured in a paper by Nathaniel Bulkley and Michael Arena. They emphasized that we traditionally focus on human capital while undervaluing the immense contribution of social capital—"what you know versus how well positioned you are to leverage what you know".

The potential for this to accelerate change was explored by OrgMapper in the startup competition. They compared the reach of managers to that of key influencers identified by their analytics, finding that influencers reach three times as many people directly. These insights into the communication networks within organizations can have huge implications for change management. Also on the topic of change, Brian Johnston and his team from IBM spoke on thinking critically about people skills and the overlaps and connections that can aid in understanding how best to re-skill in a changing workplace.

At PLASTARC, we talk a lot about the overlap of big data and little data—both are essential and, when combined, give us invaluable insights. Little data can at times be difficult to synthesize, which is why OrganizationView’s Workometry startup focusing on the synthesis of text analytics was so exciting. It affords the opportunity to more accurately analyze open-response text. They call this information "alternatively structured data, not unstructured data," emphasizing that human plus algorithm is greater than a human or an algorithm.

The second day began with a presentation by Ayreann Luedders, Senior Director, Academy Operations at Walmart. She spoke about using virtual reality systems to train store employees. Having a virtual reality environment makes it safer to fail. While we typically think of providing this type of training for high-stakes situations like flying an airplane, the affordability and portability of new systems has made it practical to use for more everyday tasks like making bread in the bakery or stocking shelves.

We also heard from Shuba Gopal and Andy Porter of the Broad Institute of MIT and Harvard. Their talk, "Data + Dialogue," examined how efforts to address complex problems are often stymied by a lack of robust data. For example, pay equity is hard to assess because of a lack of consistency around job types and requirements. There can be a lot of variability that’s made invisible by current data systems, like tenure experience for a given title. The equity gap itself is also distorted. For example, in a professional services organization, you might have more female administrative staff while more of the male staff may be executives, so it only makes sense to compare people with the same job. Yet, apples-to-apples comparisons of job descriptions are difficult because the sample sizes are too small. There’s a need for more research focused on how to make these comparisons robust. If we don’t really understand the gap, we can’t fix it.

The winners of this year’s case and white paper competitions were also revealed. The case competition asked for proposals to build capacity for the Nurse-Family Partnership. The white paper competition, sponsored by Google’s People Innovation Lab, sought actionable insights on a wide array of people analytics topics.

A fascinating panel on the ethics of people analytics followed. Arvind Narayanan of Princeton and Margaret Mitchell from Google Research and Machine Intelligence joined Lyle Ungar from University of Pennsylvania to discuss the human challenges of automating systems. When we build machines to do what humans do, we build in all the human biases and flaws. For example, many startups purport to have an assessment model for recruiting, but are really just speeding up reviews of resumes and CV’s. Assessing future performance is something that people struggle to do, so it may be naive to think it can be easily automated. As was later echoed in WPAC Faculty Co-Director Adam Grant’s closing remarks, without deep thinking and first developing a mental model that is an improvement upon current systems, we’re not actually making things better—we’re just making bad decisions faster.

Next, Deloitte CEO Cathy Engelbert and Angela Duckworth from University of Pennsylvania discussed issues of data and inclusion. As mentioned earlier, small sample sizes can create problems; you can’t preserve anonymity and can’t do the analysis you’d like to do. Another big takeaway from this session was that there is a greater burden on women executives to set an example through their decision making of the value and importance of balance. To address the discrepancy between how men and women are treated at work when they have children, for instance, women in leadership need to be "out" as parents. We also recalled Duckworth’s presentation of her book [Grit](https://www.goodreads.com/book/show/27213329-grit) at a previous WPAC.

Finally, Richard Thaler was interviewed by Cade Massey of Wharton. Thaler won a Nobel Prize in 2017 for his work in behavioral economics. His research examines and challenges the assumption that human beings are rational actors; instead, decisions are often influenced by multiple kinds of bias. His book Quasi Rational Economics digs into such effects in financial markets, though Thaler confessed that he was surprised when his research first took off in finance. He thought marketing would be first, owing to its interest in external variables, brand, consumer, and product. He attributes this to high stakes of finance for the constituents and also to the large amount of data about performance that they have at their fingertips. Marketing has since caught up in using data, as events like BRITE demonstrate. It all comes down to the value of volume—where we have lots information, we have more opportunities for insight. This plays out in multiple arenas; there is a need for more data on minority compensation, for example. There is also a dearth of data in real estate, design, and construction. Thaler’s work also reminds us of another of our favorite books, Thinking, Fast and Slow by Daniel Kahneman.

WPAC is a favorite of ours every spring—this year’s event did not disappoint. We look forward to seeing how researchers and practitioners continue to use and contribute to the substantial base of knowledge that WPAC has helped to build over the years.