Predicting Engagement With Conversational Agents in Mental Health Therapy by Examining the Role of Epistemic Trust, Personality, and Fear of Intimacy: Cross-Sectional Web-Based Survey Study
Bibliographic record
Abstract
Background: The use of conversational agents (CAs) in mental health therapy is gaining traction due to their accessibility, anonymity, and nonjudgmental nature. However, understanding the psychological factors driving preferences for CA-based therapy remains critical to ensure ethical and effective application. Variables such as epistemic trust, attachment styles, personality traits, and fear of intimacy appear central in shaping attitudes toward these artificial intelligence (AI)-driven interventions. Objective: This study aimed to investigate the role of epistemic trust, attachment styles, personality traits, and fear of intimacy in influencing individuals' willingness to engage with CA-based therapy. Methods: An online survey was administered to 876 psychology students, yielding 736 responses (84.01% response rate). Variables measured included epistemic trust, attachment styles, personality traits, and fear of intimacy. A 5-point ordinal scale assessed willingness to engage in CA-based therapy. The data were analyzed using ordinal logistic regression models, including proportional odds models (POMs), nonproportional odds models (NPOMs), and partial proportional odds models (PPOMs), with residual deviance used to compare model fit. Results: The PPOM provided the best model fit (residual deviance=3530.47), outperforming both the NPOM (deviance=6244.01) and the POM based on Brant test results indicating violations of the proportional odds assumption (χ²105=187.8; P<.001). In the final model (n=735), epistemic trust significantly increased willingness to engage in CA-based therapy across all ordinal thresholds (odds ratio [OR] 1.75, 95% CI 1.50, 2.03; P<.001). Fear of sharing demonstrated a nonuniform effect, with stronger associations at higher levels of willingness (OR 1.086; P=.001). Among personality traits, detachment negatively predicted CA preference (OR 0.95; P=.001), while psychoticism showed a positive association (OR 1.12; P=.003). Being single emerged as a strong predictor of preference for CA-based therapy (OR 3.717; P<.001). Attachment styles showed more nuanced effects. While dismissing and fearful-avoidant individuals were descriptively less inclined to engage in traditional human-based therapy, this association was nonsignificant in the case of fearful-avoidant attachment (P=.34) and should therefore be interpreted cautiously. Conclusions: Epistemic trust and fear of intimacy emerged as pivotal factors influencing preferences for CA-based therapy, underscoring the role of interpersonal dynamics and emotional vulnerabilities. The findings suggest that individuals with avoidant attachment styles or maladaptive personality traits are more inclined toward AI-mediated interventions, driven by reduced fear of judgment and increased perceived safety. The relative homogeneity of the sample considered-particularly in terms of age, education level, and cultural exposure-limits the generalizability of these findings to broader or more diverse populations. Nonetheless, these insights highlight the need for ethical considerations and personalized approaches in deploying CA-based mental health tools to balance user reliance with human-centric therapeutic values.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.010 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".