The ask
Does the design mechanic — not gamification broadly — determine whether users trust a FinTech application?
My role
Principal Investigator · Study Designer · Analyst
Timeline
2025 – Present
Live study · Currently collecting
Before you read the findings —
experience one of the conditions yourself.
Take the study → opens in Qualtrics10 minutes · Your response enters the live dataset · Findings below reflect current N = 90
Skills used in this study — tap to explore
“It is not gamification that erodes trust — it is the uncertainty mechanic specifically. Designers can use social recognition without damaging dependability. Mystery boxes cannot be rehabilitated.”
Synthesis · Empathy Pays · Spring 2026
Experimental conditions
Each participant saw exactly one condition and were instructed to apply a $10 "Welcome Bonus". Tap any card to experience it yourself.
Transparent information with compounding interest charts and explicit volatility warnings.
Social recognition mechanics — badges and a peer leaderboard.
A Mystery Growth Box with a variable ratio reinforcement mechanic.
Key findings
Condition C produced significantly lower dependability scores than both alternatives. Critically, this effect was specific to the uncertainty mechanic. The leaderboard condition did not differ from UCD on this measure.
F(2, 86) = 14.299, p < .001
UEQ Dependability by condition (1–7 scale)
C vs A+B: p < .001
The planned contrast between UCD and the leaderboard condition was significant. The omnibus F was marginal. This is a directional finding supported by theory-driven contrasts, not an omnibus effect.
t(86) = −2.194, p = .031
IMI Intent to Return by condition (1–7 scale)
A vs B: p = .031
Omnibus F marginal (p = .069). Finding supported by pre-specified planned contrast.
Interest and enjoyment was a strong, significant predictor of retention intent. Perceived autonomy contributed no unique variance once enjoyment was in the model. Designing for genuine engagement matters more than offering more decision points.
β = .547, p < .001 · R² = .306
Standardized regression coefficients predicting Intent to Return
*Effect driven by enjoyment only. Autonomy non-significant in multivariate model (p = .824).
Study data · N = 90
Sample composition and scale reliability across the three experimental conditions.
Sample by condition
Final analytic sample after exclusions (N=90)
Well-balanced across all three conditions
Randomization produced ≤2.2pp difference between groups
Scale reliability
Cronbach's α for primary outcome measures
Both scales met reliability thresholds prior to analysis
α ≥ .60 acceptable for exploratory research · α ≥ .80 preferred
n = 90 · Data collection ongoing through summer 2026. Findings will update as the sample grows. Last analysis: April 2026.
Get your results
Want to know what your condition revealed about how you respond to design mechanics?
No spam. One email with your personalized findings.
Next project
Gen Z wants social input on their dating experience, but only on their own terms