Quantifying consumer perceptions in smartwatch markets: integrating aspect-based sentiment analysis with panel data modelling
Rajeev Kumar Ray et al.
Abstract
Purpose This study proposes an integrated text analytics framework that investigates how aspect-specific consumer sentiments influence/correlate with perceived product quality in the smartwatch market. Design/methodology/approach The framework integrates DBSCAN (a density-based clustering algorithm), InstructABSA (Instruction Learning for Aspect Based Sentiment Analysis) and Instrument Variable Two-Stage Least Squares (IV-2SLS) estimation of panel data modelling and is tested on 71,858 Amazon reviews across 10 smartwatch manufacturers from 2021 to 2023. IV-2SLS estimation establishes correlations between review data and perceived product quality while addressing potential biases. Findings Monitoring capabilities sentiment demonstrates substantially stronger influence on perceived quality when accounting for potential statistical bias that could obscure the true relationship. Results indicate that 1% increase in sentiments on monitoring capabilities will increase perceived quality by 0.0198%. Software quality sentiment consistently drives satisfaction (ß = 0.0105, p < 0.01), and those on alerts and notification show positive effects (ß = 0.00685, p < 0.01). It is also noted that positive high-arousal emotions significantly enhance quality perceptions (ß = 0.305, p < 0.01). Research limitations/implications The research focuses on consumer smartwatches from a single platform, potentially limiting its generalizability. Practical implications Manufacturers should prioritize software quality management, enhance health monitoring capabilities as a key quality driver and foster positive emotional engagement through user experience design to leverage the impact of positive emotions on consumer perceptions. Originality/value This research empirically establishes a correlation between consumer sentiment embedded in online product reviews with perceived product quality in the smartwatch domain by integrating aspect-based sentiment analysis with instrumental variables two-stage least squares (IV-2SLS).
Evidence weight
Balanced mode · F 0.40 / M 0.15 / V 0.05 / R 0.40
| F · citation impact | 0.50 × 0.4 = 0.20 |
| M · momentum | 0.50 × 0.15 = 0.07 |
| V · venue signal | 0.50 × 0.05 = 0.03 |
| R · text relevance † | 0.50 × 0.4 = 0.20 |
† Text relevance is estimated at 0.50 on the detail page — for your query’s actual relevance score, open this paper from a search result.