Periodic Reporting for period 4 - BLINDSPOT (Diversity and Performance: Networks of Cognition in Markets and Teams)
Berichtszeitraum: 2021-03-01 bis 2021-08-31
Project Summary: Contemporary organizations face three interrelated, but analytically distinguishable, challenges. First, they should be alert to mistakes that could be catastrophic. Second, they need to allocate attention, especially to correct past mistakes and to make accurate predictions about future developments. Third, they should be innovative, able to stand out from existing categories while being recognized as outstanding. Research was conducted in different settings strategically chosen so that the scale and complexity of the performance challenges increased across the settings.
Project Goals: The research question that cuts across the aforementioned challenges asks whether and, if so, how diversity contributes to performance. The project has three-sub goals, addressed using different research designs:
1) We tested how social context, understood at the most basic level as the composition of a small collectivity, affects behaviour. To do so, we used experimental methods to test how racial diversity affects decision-making.
2) We tested how the social structure of attention affects valuation. The activities studied involve error correction and accuracy of prediction in estimates by securities analysts; the method is two-mode network analysis; and the structure and diversity of attention networks are the effects to be tested.
3) We examined relations within and across collaborative teams. In studying the network properties of creativity, the challenge is innovation, the activity is jazz music, the method is cultural network analysis, and the effects to be tested are stylistic diversity and social structure indicated by patterns of prior collaboration.
Project Importance: The organizational challenges we studied are particularly relevant for contemporary societies because the consequences that can arise from their mishandling can be devastating while the benefits from the organization of diversity can be significant.
Project Goals: The research question that cuts across the aforementioned challenges asks whether and, if so, how diversity contributes to performance. The project has three-sub goals, addressed using different research designs:
1) We tested how social context, understood at the most basic level as the composition of a small collectivity, affects behaviour. To do so, we used experimental methods to test how racial diversity affects decision-making.
2) We tested how the social structure of attention affects valuation. The activities studied involve error correction and accuracy of prediction in estimates by securities analysts; the method is two-mode network analysis; and the structure and diversity of attention networks are the effects to be tested.
3) We examined relations within and across collaborative teams. In studying the network properties of creativity, the challenge is innovation, the activity is jazz music, the method is cultural network analysis, and the effects to be tested are stylistic diversity and social structure indicated by patterns of prior collaboration.
Project Importance: The organizational challenges we studied are particularly relevant for contemporary societies because the consequences that can arise from their mishandling can be devastating while the benefits from the organization of diversity can be significant.
We have published 20 articles in leading scholarly journals and 7 book chapters; edited special issues for 2 leading social science journals; edited a book for Oxford University Press. Going forward, we have 10 major papers under review, and a book contract with Columbia University Press. Members of our team have presented their research for the project at more than 75 academic events.
During the project, we realized that the the problem of attention played a critical role in understanding diversity and performance across the various markets and teams we were studying. Dramatically so in the case of race, we open with our findings which we published in Science Advances.
Members of underrepresented racial groups bring diverse perspectives to the organizations in which they work. But for their contributions to affect the performance of others, majority members must attend to them. What if Whites are not paying attention to their Black colleagues? To answer this question, we conducted a large, multi-year experiment. To test how race shaped the likelihood of taking the perspective of another and follow their choices, we gave participants the task of solving a puzzle for a cash reward. The participants did not know, but a correct response required noticing the choices made by others and incorporating that into their decision. Those who failed to attend to others’ choices made poor decisions. We found that, even when it was in their interest to do so, White Americans are much more likely to ignore (and hence less likely to learn from) the choices of their Black peers.
In two further studies we explored treatments intended to ameliorate this racial attention deficit. In the first, before engaging with the puzzle, each participant received information about the peers’ prior accomplishments. This treatment did little to reduce the attention deficit. In the second treatment, instead of information about peers’ credentials in advance, participants repeatedly encountered evidence of the ongoing accomplishments of peers in the course of carrying out the task. Such experiential exposure entirely closed the racial recognition gap.
Attention also figured prominently in our network analytic study of securities analysts. Learning among rivals presents distinctive cognitive challenges. Much of social network analysis has focused on learning in communication networks among collaborators, in which actors can make direct inquiries to seek clarification about alters’ behavior or views. But such inquiries are typically not possible among rivals. Learning among competitors is usually observational learning. Our question: What network structure of attention promotes better performance under conditions of uncertainty?
Our hypothesis focused on the intersection of two types of attention structures. Dyadic closure, measuring ego’s exposure to her direct competitors in different markets, we proposed, increases the ability to interpret’ competitors’ observed behavior. Triadic openness measuring the extent to which ego’s competitors are not exposed to each other in the markets where ego is not participating, increases the diversity of views to which ego is exposed. In our analysis of a database of 442,988 earnings per share estimates issued by 11,385 analysts in the 1993–2007 period we found that the estimates issued by an analyst with multiple exposures to disconnected competitors are more accurate when confronting more challenging, high risk, high reward, volatile stocks.
In the field of network analysis this pattern of strong ties to disconnected others is so anomalous that it is referred to as the “forbidden triad.” We prefer “network tension” because we are interested in whether and how such strong ties with open triads can be productive. To further test our network analytic theory we turned our attention to study the “network history” of a jazz recording – the network pattern of who played with whom prior to the given session. Our analysis of an extraordinary dataset on the entire history of recorded jazz, encompassing 175,064 recording sessions from more than a century, found not only that network tension is more prevalent than expected but also that it predicts innovation along multiple measures of creative success.
In pursuing the topic of diversity and performance we probed problems of measuring performance, editing a special issue of the British Journal of Sociology (2020) with a lead essay “Put to the Test: For a New Sociology of Testing.” Responding to the COVID crisis, the PI wrote “Testing and Being Tested in Pandemic Times” for a wider public.
The concept of performance, how to measure it, and the proliferation of performance metrics were central to a series of workshops culminating in an edited volume, The Performance Complex: Competition and Competition in Social Life.
Finally (but far from exhausting an account of project output), the study of cognition in markets and teams should include analysis of the performance effects of new types of non-human agents. Papers include “The Impact of Bots on Human Collaboration Based on Observational Data: the Case of Wikipedia” and “Algorithmic Management in the Platform Economy.”
During the project, we realized that the the problem of attention played a critical role in understanding diversity and performance across the various markets and teams we were studying. Dramatically so in the case of race, we open with our findings which we published in Science Advances.
Members of underrepresented racial groups bring diverse perspectives to the organizations in which they work. But for their contributions to affect the performance of others, majority members must attend to them. What if Whites are not paying attention to their Black colleagues? To answer this question, we conducted a large, multi-year experiment. To test how race shaped the likelihood of taking the perspective of another and follow their choices, we gave participants the task of solving a puzzle for a cash reward. The participants did not know, but a correct response required noticing the choices made by others and incorporating that into their decision. Those who failed to attend to others’ choices made poor decisions. We found that, even when it was in their interest to do so, White Americans are much more likely to ignore (and hence less likely to learn from) the choices of their Black peers.
In two further studies we explored treatments intended to ameliorate this racial attention deficit. In the first, before engaging with the puzzle, each participant received information about the peers’ prior accomplishments. This treatment did little to reduce the attention deficit. In the second treatment, instead of information about peers’ credentials in advance, participants repeatedly encountered evidence of the ongoing accomplishments of peers in the course of carrying out the task. Such experiential exposure entirely closed the racial recognition gap.
Attention also figured prominently in our network analytic study of securities analysts. Learning among rivals presents distinctive cognitive challenges. Much of social network analysis has focused on learning in communication networks among collaborators, in which actors can make direct inquiries to seek clarification about alters’ behavior or views. But such inquiries are typically not possible among rivals. Learning among competitors is usually observational learning. Our question: What network structure of attention promotes better performance under conditions of uncertainty?
Our hypothesis focused on the intersection of two types of attention structures. Dyadic closure, measuring ego’s exposure to her direct competitors in different markets, we proposed, increases the ability to interpret’ competitors’ observed behavior. Triadic openness measuring the extent to which ego’s competitors are not exposed to each other in the markets where ego is not participating, increases the diversity of views to which ego is exposed. In our analysis of a database of 442,988 earnings per share estimates issued by 11,385 analysts in the 1993–2007 period we found that the estimates issued by an analyst with multiple exposures to disconnected competitors are more accurate when confronting more challenging, high risk, high reward, volatile stocks.
In the field of network analysis this pattern of strong ties to disconnected others is so anomalous that it is referred to as the “forbidden triad.” We prefer “network tension” because we are interested in whether and how such strong ties with open triads can be productive. To further test our network analytic theory we turned our attention to study the “network history” of a jazz recording – the network pattern of who played with whom prior to the given session. Our analysis of an extraordinary dataset on the entire history of recorded jazz, encompassing 175,064 recording sessions from more than a century, found not only that network tension is more prevalent than expected but also that it predicts innovation along multiple measures of creative success.
In pursuing the topic of diversity and performance we probed problems of measuring performance, editing a special issue of the British Journal of Sociology (2020) with a lead essay “Put to the Test: For a New Sociology of Testing.” Responding to the COVID crisis, the PI wrote “Testing and Being Tested in Pandemic Times” for a wider public.
The concept of performance, how to measure it, and the proliferation of performance metrics were central to a series of workshops culminating in an edited volume, The Performance Complex: Competition and Competition in Social Life.
Finally (but far from exhausting an account of project output), the study of cognition in markets and teams should include analysis of the performance effects of new types of non-human agents. Papers include “The Impact of Bots on Human Collaboration Based on Observational Data: the Case of Wikipedia” and “Algorithmic Management in the Platform Economy.”
Our concept of Racial Attention Deficit is already highly visible, and that of Network Tension is highly promising. Algorithmic Management in the Platform Economy was selected among thousands of submissions as one of 10 scientific “breakthroughs” at the Falling Walls Science Summit 2021 in Berlin.