Revealed Preference Choice Log¶
Observational log — instantiates Demand Curve Calibration and Response Design
Reads demand from the choices people actually made under real costs, trusting behavior over stated intent.
A Revealed Preference Choice Log records, choice by choice, what people did when facing real options at real costs — which alternative they picked, how many, and what they passed up. It takes no survey and runs no test; it is a passive ledger of naturally occurring behavior, resting on the principle that a decision made with real money, time, and consequence on the line reveals a preference more honestly than any stated one. Its defining trait is exactly that passivity and reality: it neither imposes prices (as an experiment does) nor poses hypotheticals (as a conjoint does), but reads the demand signal out of choices that would have happened anyway.
Example¶
A regional expressway offers drivers a real, recurring choice: the free general-purpose lanes, or the tolled express lane whose price floats with congestion. Every trip through the gantry is a revealed choice at a known cost, and the transponder system logs it: the toll displayed at that minute, the traffic conditions, and whether the driver bought in or stayed in the free lanes. Nobody is asked what they would pay to save fifteen minutes — the log simply records that at $4.50 a third of eligible drivers paid, and at $9.00 only a tenth did.
Because each row pairs a real choice with the exact cost and the available alternative, the log yields a demand relationship grounded entirely in behavior: how many drivers chose to buy time at each toll, and which conditions pushed them across. It captures the completed choices — the drivers who actually paid — and the switching between the two lanes as the price moved. What it cannot see is the driver who never entered the corridor at all because the whole trip felt too dear; that suppressed demand leaves no row in this log.
How it works¶
- Log the choice, the cost, and the alternative. Each row records what was chosen, the generalized cost faced at that moment, and the option not taken — the minimum needed to read a preference.
- Count completed quantities. Tally what people actually took at each cost level, keeping completed demand distinct from mere interest.
- Trace switching. Where the same agents face shifting costs over time, follow how they move between the logged alternatives.
- Keep it observational. Add nothing and impose nothing; the log's value is that the behavior was unprompted and real.
Tuning parameters¶
- Granularity of the logged choice — trip-level rows capture every decision but pile up noise; session- or day-level rows are cleaner but blur individual choices.
- Cost fields captured — logging only the toll is cheap but thin; also capturing travel-time saved, weather, and trip purpose enriches the read at the cost of instrumentation.
- Identity resolution — linking rows to the same agent enables switching analysis but raises privacy stakes; anonymous rows are safer but lose the panel.
- Retention window — long histories reveal slow preference change; short ones are lighter but miss it.
When it helps, and when it misleads¶
Its strength is fidelity: real choices under real stakes sidestep the hypothetical bias that haunts stated-preference methods, and because the behavior would have occurred anyway the log imposes no cost on anyone. As raw evidence of what people actually do, nothing beats it.
Its deepest failure mode is survivorship bias: the log only sees choices that were made[1], so demand that was priced out, deterred, or blocked never appears, and reading the log as the whole demand story silently erases everyone who walked away. Observed choices are also confounded — a driver who pays the toll on rainy Fridays reveals a preference tangled with weather and payday — so the log shows correlation, not the clean causal slope an experiment gives. The classic misuse is treating "who bought" as "who wanted." The guarding discipline is to remember the log is censored at the point of transaction, pair it with a suppressed-demand probe, and resist reading causation into raw behavior.
How it implements the components¶
observed_response_evidence_base— the log is this component: a first-hand record of real behavior under real costs.quantity_sought_measure— it counts completed choices at each cost level, anchoring the measure in what was actually taken.choice_set_and_substitution_map— by logging the alternative not chosen, it records the real choice set and how agents substitute between options as cost moves.
It only sees choices that happened, so it cannot recover demand suppressed by scarcity or barriers (latent_and_suppressed_demand_probe) — that is the Waitlist and Stockout Analysis — and it reads correlations rather than manipulating price, so it does not produce a causal slope (elasticity_and_threshold_profile, demand_schedule_model), which is the Price Sensitivity Experiment.
Related¶
- Instantiates: Demand Curve Calibration and Response Design — supplies the observed, behavior-grounded evidence the schedule is validated against.
- Sibling mechanisms: Conjoint or Discrete Choice Model · Cross-Elasticity Matrix · Demand Curve Estimation Workbook · Demand Segmentation Dashboard · Equity Access Impact Review · Price Sensitivity Experiment · Scenario Demand Stress Test · Shadow Price Probe · Waitlist and Stockout Analysis
Editorial Notes¶
Form Classification¶
Form family: Record, Log & Register
Rationale: Revealed Preference Choice Log operates as a persistent ledger, log, register, or case record that preserves history and traceability because it reads demand from the choices people actually made under real costs, trusting behavior over stated intent.
Independent corroboration: The frozen evidence defines Revealed Preference Choice Log as 'Reads demand from the choices people actually made under real costs, trusting behavior over stated intent', so its operative form is Record, Log & Register.
Nearest alternative: Analysis, Modeling & Optimization — Revealed Preference Choice Log includes features of an analytical, modeling, inference, comparison, or optimization procedure that derives insight or a solution, but its defining operation is a persistent ledger, log, register, or case record that preserves history and traceability.
Review outcome: Independent reviewer agreement; medium confidence.
Origin Attribution¶
Primary origin: Economics & Finance
Origin pattern: Single lineage
Present-day reach: Multi-domain
Rationale: Inferring demand from costly observed choices is the canonical revealed-preference tradition in economics.
Related originating lineages:
- Behavioral Economics — Behavioral research materially qualifies how context and bias affect observed choice.
- Organizational & Management Science — Organizational design, management, and operational governance supplies a parallel or contributing lineage for the mechanism's defining operation: reads demand from the choices people actually made under real costs, trusting behavior over stated intent.
Review resolution: Both blind reviewers agree that economics_finance is the primary historical origin. Explicit reconciliation of alternate origin disagreement starts from reviewer_a’s mechanism-specific evidence: Inferring demand from costly observed choices is the canonical revealed-preference tradition in economics. Reviewer A proposed alternates=behavioral_economics, origin_mode=single_lineage, domain_reach=multi_domain, and encyclopedia_synthesis=false; reviewer B proposed alternates=organizational_management, origin_mode=single_lineage, domain_reach=multi_domain, and encyclopedia_synthesis=false. The final record retains every independently supported alternate from either review (behavioral_economics, organizational_management) without an arbitrary cap, selects origin_mode=single_lineage to represent the combined lineage evidence, and keeps domain_reach=multi_domain and encyclopedia_synthesis=false from the more mechanism-specific assessment. Present-day transfer is recorded as reach and is not treated as proof of historical origin.
Review outcome: Reconciled after independent review; high confidence.
References¶
[1] James J. Heckman. "Sample Selection Bias as a Specification Error". Econometrica 47(1): 153–161, 1979. Shows that behavioral outcomes can be observed only for units whose self-selection decision places them in the recorded sample. registry ↩