Here’s a (relatively) brief overview of my research with Professor Reid Hastie: Virtually every decision in business (and plenty outside of business) involves bringing together several experts to work together to solve a shared business challenge. These experts bring to bear on a problem one or both of local domain knowledge (e.g., knowledge of European versus Asian geography, culture, and population centers) and/or distinct functional expertise (e.g., expertise in accounting versus marketing).
In my research (with Professor Reid Hastie) we adopt a two stage model of collective judgment. Stage 1 (Information Acquisition) involves sharing or pooling of information, evidence, and individual judgments. Stage 2 (Solution Integration) involves the combination of information, evidence, and individual judgments through a social or ‘mechanical’ process. Stage 1 and 2 together produce a single answer which represents the collective judgment of the group.
Stage 2 has been studied extensively: Lots of research tries to determine what aggregation rules, voting schemes, and/or statistical combination equations are optimal. We focus instead on Stage 1 and explore how enhancing the exchange of information among team members can increase group accuracy.
The group-decision framework used in this study is the Delphi Method, but our learnings could be applied in prediction markets and face-to-face ad hoc groups. Our specific interest is in determining whether sharing insights or rationales (i.e., unshared, valid, judgment-relevant information) facilitates better information pooling and, if so, in what circumstances.
We find evidence that effective information sharing will matter the most under the following circumstances:
1. On “Intellective Tasks” (e.g., solving problems with objectively correct and verifiable answers);
2. When there exists a broad variety of task-relevant information;
3. Where such task-relevant information is widely distributed across group members;
4. Where group members are motivated; and
5. Where group members posses the ability to “share.”
Why does this matter? Well it has broad implications in terms of how firms, boards, and investors make group decisions. Firms, Boards, VCs all look to facilitate information pooling. A group of individuals is brought together and asked to work towards a common end. It has become a cliché that we need cross-functional teams and that diversity (whether ethnicity, nationality, gender, or merely geographic and/or industry focus) in the workforce because such diversity leads to better business outcomes. The often unstated and arguably naïve assumption on which such ideas is based is that “Two heads are better than one.” While this purported nugget of wisdom has enjoyed sufficient support over the years to become a proverb, in practice it is not always true. Our research speaks to the conditions under which information pooling works and how to maximize its effectiveness.
Anyway, that’s probably more information than any of you wanted, but Mike asked. Reid Hastie and I are looking to publish our research before embarking on the next phase of research where we look to apply our ideas to incentivizing participants in prediction markets to better share information.
Some real truth here.
This Company Made Millions Because There Was Nothing Going On (via awaldstein)
I am glad there is an effort to curb food waste. I’d love to see us do the same with packaging.
There is a battle in the Republican party and it’s playing out in the primaries.