Research

Areas of interest: Organization Design, Innovation & Technology (online platforms, GenAI), Future of Work, Nonfinancial Incentives, Evaluations, Social Stratification

Research in progress

Finalizing draft

Job changes and engagement in online communities” with D. Kryscynski (Rutgers University). 

Data: merged individuals Stack Overflow contributions and their LinkedIn career profiles.

Data analysis

“Do Stack Overflow users have greater career benefits? Complementarity between “virtual” and “real” world work achievements” with D. Kryscynski (Rutgers University). 

Data: merged individuals Stack Overflow contributions and their LinkedIn career profiles.

Finalizing experiment, pilot scheduled

“Anthropomorphizing AI agents to achieve human-AI harmony – boom or bust?” with M. Raveendran (UC Riverside). Funded by Slack's Workforce Lab - join us in Dec 2024 webinar to hear what we found!

Data: survey and experiment with knowledge workers.

Finalizing lit review

“Learning and skill building in the AI era” with M. Raveendran (UC Riverside).

Data: perspective piece combining systematic literature review and opinion.

Data analysis

“Gender differences in scholars’ experience with self-promotion online” with M. Teplitskiy (UMich), M. Bhat & Á. Horvát (both at Northwestern University).

Data: large-scale survey with US scientists.

Data analysis

“Self-presentation language, leader-follower conversations, and leadership emergence” with C. Chambers (Johns Hopkins University) & E. Liu (UCL). 

Data: moderator elections on the Stack Overflow platform (nominations, discussions, votes).

Data analysis

“The effect of negative status changes on contributions to a large Q&A community” with M. D. Molina (NYU Abu Dhabi).

Data: Stack Overflow users' reactions to the 2011 incentive policy change.

Working papers

R&R, Research Policy

Smirnova I, Shannon A, & Teplitskiy M. “The effect of trainee career intentions on mentor’s interest in the trainee: Experimental evidence from academia.” 

Here, we test whether a trainee’s career intentions causally affect mentorship availability in the setting of PhD programs. We hypothesized that principal investigators (PIs) would view industry-focused trainees less favorably than academia-focused ones due to concerns about skills, commitment, and PIs’ own career benefits. We tested this with an audit experiment where a fictional PhD candidate emailed immunology and microbiology PIs about mentorship. Contrary to expectations, PIs responded similarly across all conditions (industry: 55%, academia: 60%, control: 59%). Treatment effects showed little variation based on the PIs’ institution prestige, industry ties, and career length. These findings suggest that PIs do not discriminate against high-skill prospective trainees based on their career interests. If trainees’ career intentions do causally decrease mentorship availability or quality, it likely occurs later in the pipeline.  

R&R, Strategic Management Journal

Smirnova I, Romero D, & Teplitskiy M. “The bias-reducing effect of voluntary anonymization of authors’ identities: Evidence from peer review.” 

Media coverage: IOP Publishing news; Nature career news

In this project we partnered with one of the world’s leading scientific publishing companies, IOP Publishing, to investigate whether the policy of “nudging” authors to anonymize their papers, i.e. strongly encouraging but not enforcing it, reduces prestige bias in manuscript evaluations made by editors and reviewers. We found that “nudging” works! For low-prestige authors, the policy increased acceptances by a substantial 5.6%; for middle- and high-prestige authors, on the other hand, the policy decreased final acceptance rates by 4.6% and 2.2% [n.s.]. We offer some first insights on an efficient policy design that is 1) low-cost, and 2) helps make evaluations fairer in settings with qualitative expert judgments. 

Working paper

Smirnova I, Reitzig M, & Mitsuhashi H. “OSS communities as complementary assetswhy and where do they work efficiently?” 

In this paper we investigate what shapes project growth in online communities like GitHub and why some project founders induce more contributions than others. Many online communities typically host a collection of projects, and community members have a choice of which projects, if any, to join. Each project founder thus can receive voluntary contributions from other community participants and can reciprocate back either by 1) making contributions to others’ projects or 2) by managing the incoming contributions to grow their own project optimally. We examine how community members perceive these two types of reciprocal behavior of a given project founder and whether they help a founder to receive more incoming help or not. Interestingly, we find that only the latter reciprocal behavior of a founder leads to project growth even though both behaviors are well-intended. 

Journal publications

Org Science 2022

Smirnova I, Reitzig M, & Sorenson O. 2022.Building status in an online community.” Organization Science 33(6): 20852540. 

Media coverage: UCLA Anderson Review

This research investigates how contributors, by collecting non-monetary awards for their contributions, attain high status within the community. The main motivation for this project comes from the observation that having community-wide recognition is one of the key drivers for why people join such collectives. Given how valuable and desirable community-wide status is, this paper explores the pathways through which one could gain it. We investigate the notion that the actions that contribute to status attainment vary as community contributors move up the status hierarchy. Combining the data from the online Q&A community Stack Overflow and two experiments, we find that contributors can build low levels of status by engaging in activities whose quality is easy to assess for the community (i.e. asking questions). As status rises, however, further increases in status come from engaging in activities that are more difficult to evaluate (i.e. providing answers) or even impossible to judge (i.e. commenting). 

Research Policy 2022

Smirnova I, Reitzig M, & Alexy O. 2022. “What makes the right OSS contributor tick? Treatments to motivate high-skilled developers.” Research Policy 51(1).

Media coverage: UMSI news

This paper explores how non-monetary incentives (an element of organizational design) effect effort (an important component of collaborative innovation). Just as individual community members differ in their preferences, so do they vary along a series of other dimensions: professional history, experience, skill, and so forth. Here, we explore how the management of an innovation project influences different contributors’ perceptions and engagement. Specifically, we examine how the skills of contributors effect their selection for an open source project in the first place, and how project owners—private individuals or companies—can motivate developers with the right skill to exert continuous effort on a project after joining. We focus on three design parameters that can be managed by founders (founders revealing their corporate affiliation, project acceptance rate and feedback provision time) and test how they impact the effort of different developers within the Stack Overflow and GitHub communities.  

Information Technology and Management 2021

Mishra B & Smirnova I. 2021. “Optimal configuration of intrusion detection systems.” Information Technology and Management 22: 231–244. 

An intrusion detection system (IDS) can protect your network and computer from a variety of threats. IDSs are designed to assist detection of computer security violations including illegal entry by outsiders and abuse of privileges by insiders. This article develops the clear connection between IDS systems and their economic/managerial impact and presents an optimization model based on game theory to assist firms in the IDS configuration process. The model is based on the fact that the frequencies of false positives and false negatives affect the way firms deal with signals from the IDS, and that configuration affects the behaviors of users significantly. Given that no one configuration will be optimal for all firms, our analysis develops interesting insights into how the firm’s cost parameters, the IDS’s quality parameters, and the external parameters (such as hacker’s benefit from intrusion and the penalty that is imposed when the intrusion is detected) affect the configuration decisions.