Who benefits from virtual collaboration? The interplay of team member expertness and Big Five personality traits

The study addresses how individual attributes—specifically expertness levels and Big Five personality traits—and their diversity within dyadic teams influence performance gains from virtual collaborative problem-solving (CPS). Motivated by the growing prevalence of virtual and hybrid collaboration and the challenges of reduced information richness and social presence in virtual settings, the research integrates Transactive Memory Systems (TMS) theory with the Big Five model. It identifies a gap in understanding intra-domain expertness diversity (as distinct from inter-domain expertise diversity) and the role of personality traits in TMS development and CPS performance. The authors formulate four research questions: RQ1 examines how team expertness diversity influences team performance gain from virtual collaboration; RQ2 examines, at the individual level, how one’s own and a teammate’s expertness levels influence individual performance gain; RQ3 examines how team Big Five personality trait diversity influences team performance gain; and RQ4 examines, at the individual level, how one’s own and a teammate’s Big Five traits influence individual performance gain. The purpose is to provide a nuanced understanding of how individual differences and team composition shape outcomes in virtual CPS through TMS mechanisms, addressing practical needs in increasingly virtual workplaces.

The literature situates TMS as a framework for how teams encode, store, and retrieve knowledge via perceptions of credibility, specialization, and coordination, yielding differentiated versus integrated TMS structures. Differentiated structures emphasize unique, domain-specific expertise; integrated structures feature overlapping knowledge within teams. Prior work shows context-contingent benefits of each structure, with integrated TMS often benefiting intellective tasks. The review distinguishes expertise diversity (inter-domain breadth) from expertness diversity (intra-domain depth differences), noting both are common in teams but that expertness diversity is understudied, especially in virtual CPS. Virtual CPS presents challenges (lower information richness/social presence, expertise recognition difficulties, social loafing) yet TMS can still develop using ICT affordances (visibility, searchability, awareness), aligning with social information processing and communication visibility theories. Prior findings suggest expertness diversity can enable reciprocal learning but may be moderated by team dynamics. The Big Five traits can shape TMS processes and collaboration: openness supports exploration and sharing; conscientiousness aids organization and reliability but diversity may cause friction; extraversion promotes knowledge sharing but may dominate; agreeableness fosters harmony and trust yet can impede critical debate; neuroticism can hinder communication and trust, potentially exacerbated virtually. The review motivates examining both expertness and personality trait diversity for their effects on team and individual outcomes in virtual CPS.

Design: Experimental study of ad hoc dyadic teams collaborating virtually on an intellective CPS task (general science/volcano simulation) via synchronous text chat only. Participants and sampling: Recruited via Amazon Mechanical Turk. Initially 1000 invited; final analytic sample after exclusions: 377 dyads (754 individuals). Demographics: 49% male, 51% female; 77% white, 8% black, 7% Hispanic, 6% Asian, 2% American Indian or Pacific Islander; 98% native English speakers; ages 18–68 (70% 18–35; 29% 35–60; 1% >60); all with at least 2 years of college. Procedure and task: Each participant completed (1) a 37-item multiple-choice general science knowledge test (adapted from SLIM), (2) TIPI Big Five inventory, and (3) demographics. Dyads then completed a simulation-based CPS task on volcano-related questions (7 items). Collaboration occurred via synchronous text chat with pseudonyms; one member submitted the team representative answer after discussion. Measures: - Individual expertness: Score on the 37-item test (range 0–36; mean 27.40; median 29; Cronbach’s α=0.89). Dichotomized by median: High (H) >29; Low (L) ≤29. Individual-level pairing categories: HwH, HwL, LwH, LwL. - Expertness diversity (team-level): Categorical. LH (mixed: one H and one L) = high diversity; HH (both H) and LL (both L) = low diversity. Distribution across 377 teams: LL=134, LH=178, HH=65. - Big Five personality traits: TIPI (10 items; 5-point Likert), trait scores computed as mean of a trait’s standard and reverse-scored items. For each trait, median split into High/Low per individual; team-level diversity per trait coded as high (one H, one L) versus low (both H or both L). Individual-level and team-level distributions summarized (e.g., extraversion median ~2.5; others around 4). Performance outcomes and scoring process: For each of 7 items, system recorded individual initial answers (pre-collaboration) and graded them; members then chatted and could revise their answers individually (revised scores). One member was randomly selected to submit the team representative answer. Analyses focus on initial vs revised scores. - Team-level outcome: Team score change = sum of members’ revised scores − sum of members’ initial scores across items (positive = performance gain). - Individual-level outcome: Individual score change = revised − initial (positive = performance gain). Analysis: ANCOVAs. - Team level (RQ1, RQ3): DV = team score change; IVs = team expertness diversity and team Big Five trait diversities (five traits); covariate = team initial score. Significant IVs followed by post-hoc Tukey tests among HH, LH, LL. - Individual level (RQ2, RQ4): DV = individual score change; IVs = individual expertness pairing category (HwH, HwL, LwH, LwL) and Big Five pairings; covariate = individual initial score. Significant IVs followed by Tukey tests among HwH, HwL, LwH, LwL.

Team-level (ANCOVA; DV team score change): After controlling for team initial score (F=4.15, p=0.04), two predictors were significant: - Expertness diversity: F=4.29, p=0.01. Tukey post-hoc: LH (mixed high-low expertness) teams had significantly greater performance gains than HH and LL teams (both low-diversity groups). - Agreeableness diversity: F=3.89, p=0.02. Tukey post-hoc: Teams with both members low in agreeableness (LL) showed the greatest performance gains compared with HH and LH in agreeableness. Other Big Five diversities (openness, conscientiousness, extraversion, neuroticism) were not significant. Individual-level (ANCOVA; DV individual score change): After controlling for individual initial score (F=183.89, p<0.001), significant predictors: - Expertness pairing: F=6.32, p<0.001. Tukey post-hoc: Low-expertness individuals paired with a high-expertness teammate (LwH) had the largest performance gains. Differences among HwL, LwL, and HwH were not significant. - Agreeableness showed significance in the full model (F=2.73, p=0.04), but no subgroup differences were significant in Tukey tests, thus not emphasized. Overall: High expertness diversity at the team level boosts virtual CPS gains; at the individual level, low-expertness members benefit most when paired with high-expertness partners. Teams composed of two low-agreeableness members exhibit the highest gains.

Findings support the theorized role of TMS mechanisms in virtual CPS. In intellective tasks, mixed-expertness dyads capitalize on expertise storage and retrieval: low-expertness members seek and apply knowledge from high-expertness partners, while high-expertness members consolidate understanding by explaining and correcting misconceptions, yielding reciprocal learning and improved performance. Despite reduced nonverbal cues in virtual settings, teams effectively used text-based cues to coordinate knowledge, aligning with social information processing and communication visibility theories. The positive effect of expertness diversity counters concerns that experts withhold help; in this virtual, task-focused context, pairing high and low expertness enhanced outcomes. Personality composition mattered for agreeableness: dyads with both members low in agreeableness showed the greatest gains, suggesting that direct, task-focused communication and willingness to challenge ideas can be advantageous in text-only, intellective CPS, potentially mitigating groupthink. Conversely, uniformly high agreeableness or mixed agreeableness may dampen critical debate or introduce harmony-maintenance dynamics that reduce rigorous problem solving. Other Big Five traits and their diversities did not show robust effects in this design, suggesting agreeableness plays a more salient role for short-duration, text-based virtual CPS. Overall, integrating TMS with personality perspectives illuminates how intra-domain expertness composition and specific trait configurations contribute to performance gains in virtual dyads, offering a nuanced account of when and for whom virtual collaboration is most beneficial.

This study demonstrates that, in virtual dyadic CPS on intellective tasks, teams with high expertness diversity achieve greater performance gains than homogeneous high- or low-expertness dyads, and low-expertness individuals benefit most when paired with high-expertness teammates. Personality composition also matters: dyads low in agreeableness show the largest gains, indicating benefits of direct, goal-focused interaction in virtual text-based collaboration. The integration of TMS theory with the Big Five model advances understanding of how individual attributes and their diversity shape virtual collaboration outcomes. Future research should extend to larger teams, longer-term and more naturalistic settings, richer performance measures, additional personal and demographic attributes, and analyses of communication content to unpack mechanisms linking TMS development, expertness diversity, and personality to collaborative performance.

- Context and duration: Ad hoc dyads, text-only synchronous chat, and a short (~50 minutes) session may limit generalizability and may not allow full unfolding of TMS or personality effects. - Performance measure scope: Only seven multiple-choice items were used, potentially limiting sensitivity to performance changes. - Omitted variables: Analyses focused on expertness and Big Five traits; other attributes (e.g., demographics, culture) were not modeled despite potential relevance. - No discourse analysis: Communication content (chat logs) was not analyzed; thus, mechanisms linking TMS processes, expertness diversity, and personality to outcomes remain indirect.