Predicting Professional Self-Efficacy and Imposter Syndrome in Technical Slack and Microsoft Teams Communications Using Ensemble NLP Pipelines and Transformer Architectures

• Adaan Ahsun

  Covenant University

  Author

Technical professionals increasingly rely on Slack and Microsoft Teams for daily collaboration, creating rich textual records that may encode psychological states such as professional self-efficacy and imposter syndrome. Despite the prevalence of imposter syndrome—affecting an estimated 70% of employees at some point in their careers—and its documented negative association with career advancement and job satisfaction, no validated framework exists for detecting these constructs from workplace communication data. This study develops and evaluates an ensemble NLP pipeline integrating TF-IDF features with transformer-based embeddings (RoBERTa, DistilBERT) to predict self-efficacy and imposter syndrome from Slack and Teams messages. Using a de-identified dataset of 47,832 messages from 312 technical professionals, we compare ensemble architectures against single-model baselines. The ensemble pipeline achieved 89.4% accuracy (F1=0.891) in predicting high versus low professional self-efficacy, and 87.2% accuracy (F1=0.869) for imposter syndrome indicators, outperforming standalone transformers by 4.2 percentage points. Feature importance analysis identified hedging language, collective pronoun usage, and question frequency as primary predictors. The framework offers a replicable, privacy-preserving approach for organizational psychological monitoring, with implications for early intervention and team health analytics.

• pdf

Copyright (c) 2026 Adaan Ahsun (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.