Foundational research on heuristics and biases supports testing specific judgment errors rather than explaining every decision with a generic label. This guide turns that principle into a pre-event record, a fixed review checklist, and clear stopping conditions.
Five influences to keep separate
| Influence | Observable question | Common overclaim to avoid |
|---|---|---|
| Anchor | Did an opening price, pundit number, or previous forecast appear before the estimate? | Every later estimate was caused by the first number |
| Gain or loss frame | Did the same choice look different when described as a gain, loss, refund, or recovery? | Loss aversion explains every refusal to stop |
| Outcome knowledge | Did the review change after the result became known? | A win proves the decision was good |
| Emotion and urgency | Did excitement, frustration, fear, or a countdown change the written rule? | One feeling predicts one action in everyone |
| Social context | Did group participation or perceived norms alter the decision? | Friends directly caused an individual's harm |
Prospect theory is a descriptive account of choices under risk in studied tasks; it is not a staking formula. The original outcome-bias experiments show why learning a result can change evaluation of a prior decision. Research on fear of missing out began in a broader social-media context, so applying it to betting requires care rather than assumption.
An Australian social-network study found associations between network characteristics, gambling, and reported harm. Its cross-sectional design does not establish that one social influence caused one outcome. The useful lesson is to record the social setting as context, not to diagnose a group or person.
Worked decision audit
In an illustrative example, suppose a person records the following before an event: maximum loss GBP 10, estimated probability 45%, accepted decimal price 2.20, and no stake change after 18:00. A group message then says the selection is "certain," the price shortens to 2.05, and the person increases the stake to GBP 25 after a previous loss.
The review should not begin with whether the selection won. It should compare the record with the action:
| Test | Pre-event rule | Observed action | Result |
|---|---|---|---|
| Maximum loss | GBP 10 | GBP 25 | Rule broken |
| Probability basis | 45% with stated evidence | No revised evidence | Not updated |
| Price | 2.20 recorded | 2.05 accepted | Contract changed |
| Social information | No defined role | "Certain" message | Unsupported input |
| Loss response | No change after a loss | Stake increased | Stop condition met |
The example identifies process changes. It does not show which psychological mechanism caused them, and it does not calculate whether either price had value.
A review method that resists hindsight
- Save the decision and information cutoff before the event.
- Record price, maximum loss, probability or reason, and no-action conditions.
- Preserve later messages and price changes separately.
- Review adherence before revealing or discussing the outcome.
- Score forecast quality only across consistently defined later events.
- Treat repeated boundary failures as a reason to stop, not a prompt to optimise the routine.
When analysis should stop
The NHS lists warning signs and practical protections in its gambling-harm guidance. If gambling is affecting essential money, health, sleep, work, study, or relationships, or if stopping feels difficult, move from decision analysis to support and protective controls.
Next step
Use Cognitive Biases Betting for the next part of this topic.
Continue learning
- Next guide: Responsible Gambling
- Related guide: Signs of Gambling Harm
Assumptions and limitations
This is educational material, not an individual psychological assessment. Laboratory tasks, surveys, and observational account data answer different questions and may not generalise to every sport, product, jurisdiction, or person. A checklist can expose changes from a written plan; it cannot identify a clinical condition or guarantee better outcomes.
