Determinants of Online Word-of-Mouth in the IT Industry - Evidence from Employee Job Reviews Using Topic Modeling -

[Purpose] This study analyzed how text-based topics influenced the likelihood that IT professionals in Korea publicly recommended their employers in job reviews, thereby identifying the key factors that shaped employee online word-of-mouth (OWOM). [Methodology]We scraped 19,594 reviews (2014-2019) of 141 IT firms from Jobplanet.co.kr. After bespoke Korean preprocessing, Latent Dirichlet Allocation (LDA) uncovered eight positive and seven negative topics. The resulting 15 topic probabilities, together with star ratings, employment status, and business-outlook indicators, were entered into a firm- and year-fixed-effects logistic-regression model to examine the factors influencing employees’ “recommend company” decisions. [Findings]Only one positive theme—horizontal and innovative culture—significantly raised recommendation odds. Five negative themes—conflict with superiors, discrimination & unfair treatment, immature evaluation systems, unreasonable compensation, and a counter-intuitively positive high work-intensity factor—were significant; a full-range increase in the latter tripled the odds of recommendation (odds ratio ≈ 3.46). These patterns accord with Social Exchange Theory and Perceived Organizational Support: perceived fairness spurred, while perceived injustice suppressed, positive OWOM. [Implications]This study extended OWOM research into long-term employment settings and showed that topic probabilities extracted from job reviews could predict recommendation behavior. Organizations should cultivate egalitarian, innovative cultures, establish fair compensation and evaluation systems, and frame high-intensity work as opportunities for growth. Firms can also monitor emerging negative themes as early warning signals, while review platforms may leverage topic analytics to deliver firm-level insights.