Decision traps: method

What it measures

Performance on five decision formats: consistency across positive and negative framing, resistance to sunk costs, applying a decision rule, confidence calibration, and cognitive reflection; combined into one game score. No stable personal score per bias is claimed.

Procedure

17 original items in a fixed order, one at a time: two framing pairs (4 items; the positive and negative frame of each pair are far apart in the order), 3 sunk-cost items, 3 decision-rule items, 4 confidence items (a yes/no answer plus confidence 50, 60, 70, 80, 90 or 100%) and 3 reflection items (4 options each). Instruction: choose the best answer using only the information given. Untimed; the visitor can go back before submitting; explanations and the right answers appear only after the result.

Parameters

Scoring

Credit per item: framing pair = 1 if the same option is chosen in both frames, else 0; sunk cost, decision rules and reflection = 1 if correct; confidence item = 1 − (c/100 − correct)², with c the confidence in the chosen answer and correct = 1 or 0 (a Brier-type accuracy). Format score = mean credit within the format. Game score = round(100 × mean of the five format scores), so each format has equal weight; 0–100, higher is better. Comparisons with other players are shown as broad bands.

Valid-run rules

Every item needs exactly one answer, and confidence items need one of the offered confidence levels (other items must not have one); the item-set version must match; otherwise the run is rejected. A 'too_fast' quality flag (at least half of the answers under 700 ms) is stored, and flagged runs are left out of comparisons.

Differences from the classic task

The A-DMC battery (Bruine de Bruin et al., 2007) has longer subscales with its own items and scoring (for example, framing measured with rating scales in separate sections), and the Cognitive Reflection Test (Frederick, 2005) has three specific items. All items here were written for this site; the published reliability and validity of those instruments do not transfer.

Limitations

Few items per format, so the format scores are unreliable and only the combined score is reported. Seeing the explanations makes repeat runs easier. Some items need basic arithmetic. Translations are not validated. An educational game, not a validated decision-making competence score.

Languages and review status

English original; translations into 99 further languages (machine-assisted translation with automated script and spelling checks; numbers, restrictions, option order, right answers and explanations kept the same as in English).