Although researchers have used phone surveys for decades, the lack of an accurate picture of the call opening reduces our ability to train interviewers to succeed. Sample members decide about participation quickly. We predict participation using the earliest moments of the call; to do this, we analyze matched pairs of acceptances and declinations from the Wisconsin Longitudinal Study using a case-control design and conditional logistic regression. We focus on components of the first speaking turns: acoustic-prosodic components and interviewer’s actions. The sample member’s “hello” is external to the causal processes within the call and may carry information about the propensity to respond. As predicted by Pillet-Shore (2012), we find that when the pitch span of the sample member’s “hello” is greater the odds of participation are higher, but in contradiction to her prediction, the (less reliably measured) pitch pattern of the greeting does not predict participation. The structure of actions in the interviewer’s first turn has a large impact. The large majority of calls in our analysis begin with either an “efficient” or “canonical” turn. In an efficient first turn, the interviewer delays identifying themselves (and thereby suggesting the purpose of the call) until they are sure they are speaking to the sample member, with the resulting efficiency that they introduce themselves only once. In a canonical turn, the interviewer introduces themselves and asks to speak to the sample member, but risks having to introduce themselves twice if the answerer is not the sample member. The odds of participation are substantially and significantly lower for an efficient turn compared to a canonical turn. It appears that how interviewers handle identification in their first turn has consequences for participation; an analysis of actions could facilitate experiments to design first interviewer turns for different target populations, study designs, and calling technologies.

Machine learning systems are making considerable inroads in society owing to their ability to recognize and predict patterns. However, the decision-making logic of some widely used machine learning models, such as deep neural networks, is characterized by opacity, thereby rendering them exceedingly difficult for humans to understand and explain and, as a result, potentially risky to use. Considering the importance of addressing this opacity, this paper calls for research that studies empirically and theoretically how machine learning experts and users seek to attain machine learning explainability. Focusing on automated trading, we take steps in this direction by analyzing a trading firm’s quest for explaining its deep neural network system’s actionable predictions. We demonstrate that this explainability effort involves a particular form of human–machine interaction that contains both anthropomorphic and technomorphic elements. We discuss this attempt to attain machine learning explainability in light of reflections on cross-species companionship and consider it an example of human–machine companionship.

No abstract

This article examines algorithmic trading and some key failures and risks associated with it, including so-called algorithmic ‘flash crashes’. Drawing on documentary sources, 189 interviews with market participants, and fieldwork conducted at an algorithmic trading firm, we argue that automated markets are characterized by tight coupling and complex interactions, which render them prone to large-scale technological accidents, according to Perrow’s normal accident theory. We suggest that the implementation of ideas from research into high-reliability organizations offers a way for trading firms to curb some of the technological risk associated with algorithmic trading. Paradoxically, however, certain systemic conditions in markets can allow individual firms’ high-reliability practices to exacerbate market instability, rather than reduce it. We therefore conclude that in order to make automated markets more stable (and curb the impact of failures), it is important to both widely implement reliability-enhancing practices in trading firms and address the systemic risks that follow from the tight coupling and complex interactions of markets.

We describe interviewers’ actions in phone calls recruiting sample members. We illustrate (1) analytic challenges of studying how interviewers affect participation and (2) actions that undergird the variables in our models. We examine the impact of the interviewer’s disfluencies on whether a sample member accepts or declines the request for an interview as a case study. Disfluencies are potentially important if they communicate the competence or humanity of the interviewer to the sample member in a way that affects the decision to participate. Using the Wisconsin Longitudinal Study, we find that although as they begin, calls that become declinations are similar to those that become acceptances, they soon take different paths. Considering all recruitment actions together, we find that the ratio of disfluencies to words does not predict acceptance of the request for an interview, although the disfluency ratio before the turning point – request to participate or a declination – of the call does. However, after controlling for the number of actions, the disfluency ratio no longer predicts participation. Instead, when we examine actions before and after the first turning point separately, we find that the number of actions has a positive relationship with participation before and a negative relationship after.