The validation engine

Original Experiments

These studies are where the framework is put at risk. Each one tests a specific link in the chain and ships a full output package — executive summary, statistics, regression, and dataset — so the thesis stands or falls on evidence, not assertion.

Status badges

Each study carries a lifecycle status — Planned, In design, Collecting data, or Analysis complete. Statuses update as the project progresses; result figures shown are clearly-labeled demo data until collection is complete.

Experiment 01

Risk Under Pressure Study

Validates · Links 1→3 · Pressure → Bias → Decision Distortion

Collecting data

Tests whether different pressure conditions change risk-taking, confidence, and decision quality relative to a baseline.

What is being tested?

Whether time pressure, social evaluation, and loss framing each shift risk-taking, confidence calibration, and herding away from a control baseline.

What did we find?

Preliminary (demo) results suggest loss framing is the largest mover — pulling risk down and widening the confidence gap — while social evaluation most increases herding. To be confirmed with collected data.

Why does it matter?

It is the core validation step: if pressure distorts individual choices in a structured way, the framework's first three links hold and the market-level question becomes worth asking.

Output package

Demo data

Executive summary — readable insights

Loss framing produced the most risk-avoidant behavior (mean risk 39 vs. 52 control) — the single largest condition effect.
Confidence decoupled from accuracy most under loss framing and time pressure (gaps of +15 and +12 points).
Social evaluation drove the strongest herding response (+21 vs. +8 control), consistent with H3.
Net read: pressure does not just raise or lower risk uniformly — each condition distorts a different part of the decision.

Summary statistics

Condition	n	Risk (SD)	Conf. gap	Herding
Control	64	52 (18)	+4	+8
Time Pressure	60	61 (22)	+12	+14
Social Evaluation	62	47 (19)	+9	+21
Loss Framing	62	39 (17)	+15	+17

Regression summary

Term	β	SE
Time Pressure	+9.1*	3.6
Social Evaluation	−4.8	3.5
Loss Framing	−13.2***	3.4
Numeracy (z)	+2.1	1.5
Constant (control mean)	52.0***	2.4

risk_score ~ time_pressure + social_eval + loss_framing (OLS, control = reference, numeracy covariate)
* p<.05 ** p<.01 *** p<.001 · Coefficients are change in risk score (0–100) vs. the control group. Demo estimates.

Dataset preview

id	condition	risk	confidence	accuracy	herding
p001	Control	55	66	63	6
p002	Time Pressure	64	78	61	12
p003	Social Evaluation	44	72	60	24
p004	Loss Framing	37	74	57	18
p005	Loss Framing	41	70	59	15

Clean dataset (CSV)Coming soonSummary statistics (PDF)Coming soonRegression tables (PDF)Coming soonExecutive summary (PDF)Coming soon

Conditions

Control
Time Pressure
Social Evaluation
Loss Framing

Measures

Risk tolerance
Choice consistency
Confidence calibration
Herding tendency
Loss sensitivity
Reaction time (where available)

Hypotheses

H1 — Time pressure increases variance in risk-taking and reduces choice consistency relative to control.
H2 — Loss framing shifts choices toward risk-avoidance and raises loss sensitivity.
H3 — Social evaluation increases herding tendency and widens the confidence–accuracy gap.

Variables

Variable	Type	Description
Pressure condition	Independent	Randomly assigned: control, time pressure, social evaluation, or loss framing.
Risk score	Dependent	Composite of allocation choices across paired gamble tasks (0–100).
Confidence gap	Dependent	Stated confidence minus realized accuracy on judgment items.
Herding tendency	Dependent	Shift toward displayed group consensus after it is revealed.
Numeracy	Control	Short numeracy check used as a covariate, not a screen.

Sample decision tasks

Task 01

Choose between a guaranteed $40 and a 50% chance of $100 (otherwise $0).

Repeated across gain and loss frames to estimate reference dependence.

Task 02

Rate your confidence (0–100%) that your previous answer was the better choice.

Paired with realized outcomes to compute calibration.

Task 03

A group of prior participants mostly chose Option B. Do you want to revise your choice?

Isolates willingness to follow revealed consensus.

Participant recruitment

Financial Analytics Club
Athletic teams
School mailing lists
Social media

Sample size & power

Target N ≈ 240–280 (≈60–70 per condition). Pilot N = 48 complete. Powered to detect medium between-condition differences (d ≈ 0.5) at α = .05.

Assumptions

Simulated pressure (timers, observation, framing) meaningfully proxies real-world stress.
Composite risk score captures the construct better than any single task.
Random assignment balances unobserved traits across conditions.

Methodology note

Between-subjects random assignment to one pressure condition. Tasks are incentive-compatible where feasible; item order is randomized to limit sequence effects.

Ethics note

Voluntary participation, anonymous responses, aggregate-only analysis. No clinical or diagnostic claims. Participants may stop at any time.

Theory anchor

Prospect theory (reference dependence, loss aversion) and dual-process accounts of judgment under load.

Experiment 02

Identity and Leadership Moderation Study

Validates · Moderators on Links 1→3 · who bends, and how much

In design

Tests whether identity salience and leadership responsibility moderate risk-taking, herding, and confidence.

What is being tested?

Whether making identity salient, or assigning responsibility for others, changes how strongly the core pressure effects appear.

What did we find?

Pending — the study is in design and pre-registration. No results yet.

Why does it matter?

It identifies who is most susceptible to the pressure tax, and whether leadership cues amplify or discipline it — the bridge from individual to group outcomes.

Output package

Demo data

Results pending

Results pending. The output package — summary statistics, regression, and an executive read — will publish here once data is collected.

Pre-registration (PDF)Coming soonClean dataset (CSV)Coming soonSummary statistics (PDF)Coming soon

Conditions

Identity prime
Leadership responsibility prime
Neutral / control

Measures

Risk tolerance
Herding shift
Confidence calibration
Willingness to deviate from group consensus
Leadership accountability effect

Hypotheses

H1 — An identity prime increases the weight placed on in-group consensus (more herding).
H2 — A leadership-responsibility prime changes willingness to deviate from consensus relative to control.
H3 — Accountability cues shift confidence calibration, not only the level of risk taken.

Variables

Variable	Type	Description
Prime condition	Independent	Randomly assigned identity, leadership, or neutral prime.
Deviation from consensus	Dependent	Distance between final choice and displayed group consensus.
Herding shift	Dependent	Change in choice after consensus is revealed, by condition.
Pressure condition	Moderator	Whether a base pressure manipulation is also present.

Sample decision tasks

Task 01

Brief writing prompt that makes a personal or group identity salient before the decision tasks.

Standard priming approach; content is non-sensitive.

Task 02

You are responsible for the outcome on behalf of a small team. Make the allocation.

Operationalizes leadership responsibility.

Participant recruitment

Financial Analytics Club
Athletic teams
School mailing lists
Social media

Sample size & power

Target N ≈ 210 (≈70 per prime), reusing Study 1's task battery. Pre-registration in progress.

Assumptions

A short writing prime is sufficient to make identity momentarily salient.
Assigned 'responsibility for a team' approximates real leadership stakes.
Effects generalize beyond the specific in-group used in the prime.

Methodology note

Builds on Study 1's task battery, adding a randomized prime. Where possible, primes are validated with a brief manipulation check.

Ethics note

Primes are non-sensitive and non-deceptive beyond standard practice. Same voluntary, anonymous, aggregate-only protocol as Study 1.

Theory anchor

Social identity theory (in-group signal weighting) and accountability / responsibility effects on choice.

How we keep it honest

Methodology & credibility

The standards every study is held to — and the limits stated up front, not buried.

Design

Between-subjects randomized assignment, pre-registered primary hypotheses, incentive-compatible tasks where feasible, and randomized item order.

Sample & pools

Participants recruited from a finance club, athletic teams, school mailing lists, and social media. Pools are convenience samples — a stated limitation, not a hidden one.

Analysis

Condition contrasts against control via OLS with a numeracy covariate; effects reported with standard errors and significance, never as proof of universal laws.

Limitations & assumptions

Simulated pressure is not real-world stress; samples are modest and school-based; results are associations and condition differences. Each study lists its own assumptions.