5D Library › Choice Architecture, Cumulative Prospect Theory and Computational Behavioural Economics of Pricing Interfaces · 中文
Preface: Options, References and Time
Note
Purpose and risk notice: please read this first. This book presents mathematical models for teaching and research. It is not marketing, pricing, legal, compliance or investment advice. Every pricing page, price history, sales record, experimental data set and person in the book is fictional or simulated, not data from any real business, product or customer. The price-indication rules discussed here are used only to show the limits of the mathematics and the evidence; they are not an interpretation of the law of any jurisdiction, and real commercial decisions call for a qualified local professional. The book offers no method that “guarantees higher conversion”, “raises sales by X %” or “always wins on price”. Using choice design to mislead customers (fictitious former prices, fake countdowns, hidden exit routes, pre-ticked add-on charges) may be unlawful and runs against the position of this book; such practices are analysed here so that they can be recognised and constrained.
How many plans should a pricing page show? Should an option that “looks like a bad deal” sit next to the middle one? Should a former price be struck through beside the current one? Should the checkout button say “save $10 now” or “lose your $10 discount if you leave”? Should a sign-up form have three steps or seven? Countless pages answer these questions every day, often on the strength of widely repeated claims: “the more options, the higher the conversion”, or the reverse, “too many options and customers buy nothing”; “strike through a high price and customers will pay more”; “people feel losses 2.25 times as strongly as gains, so a loss frame always works better”; “add a decoy and the middle tier will sell”. This book asks which of these claims are supported by mathematics and evidence, under what conditions, which need qualifying and which are wrong.
Answering takes four layers of tools. The first is the scale of the evidence: how much of an effect seen in a single experiment survives in meta-analyses and large field trials. The second is value and probability: how cumulative prospect theory describes the subjective treatment of gains, losses and probabilities, where its parameters come from, and what they can and cannot tell us. The third is the choice set: whether a newly added option can change preferences among the existing ones, why standard discrete-choice models cannot do this, and what kind of model can. The fourth is time: why people keep postponing what would take a few minutes, and how each step of friction in a form is amplified.
What the book covers
Chapter 1 covers choice overload: the original numbers of the jam study, the conclusions of two meta-analyses and an author’s model of when more options help and when they hurt, and how the publication filter inflates nudge effects. Chapter 2 covers struck-through and reference prices: transaction utility, the credibility ceiling of a reference price, and the US and EU rules on “former” and “prior” prices. Chapter 3 covers the value function and two probability-weighting functions of cumulative prospect theory, and the difference between cumulative (rank-dependent) weights and weighting each probability separately. Chapter 4 covers the evidence on the loss-aversion coefficient and its identification, and the link function that is missing between a model of individual choice and a “conversion gap”. Chapter 5 covers the independence of irrelevant alternatives in the multinomial logit, the regularity theorem for random utility models, Tversky and Simonson’s context-dependent model and multialternative decision field theory, and the boundaries of the decoy effect. Chapter 6 covers exponential, quasi-hyperbolic and hyperbolic discounting, naive and sophisticated procrastination, and an author’s model of multi-step forms. Chapter 7 reads these beside the “Guo xu” and “Qingzhong yi” chapters of the Guanzi and Jiran’s policies in the Shiji. Chapter 8 offers a four-step checklist and a pre-registered online randomised experiment.
The book neighbours several others in 5D Library; where they overlap it only recaps briefly, cites the earlier book and does not repeat its programs. Book 6, The Economics of Attention, wrote the conversion funnel as an absorbing Markov chain and treated pricing among platforms, agencies and creators (this book treats only choice design on a single page). Book 14 treated dopamine reward-prediction errors and immediate rewards (Chapter 6 only recaps). Book 19, Non-Ergodic Wealth Dynamics, the Kelly Criterion and Volatility Drag, treated logarithmic utility, time averages and Jiran in the Shiji (Chapters 3 and 7 only recap). Book 26 treated probability judgement and calibration (Chapter 3’s probability weighting is related but different: the weighting of known probabilities rather than the estimation of unknown ones). Book 27 treated mechanism design, bargaining and the BATNA (Chapter 2 only recaps anchors in bargaining). Book 33, Scientific Methodology, Falsifiability and the Mathematical Statistics of the Replication Crisis, treated pre-registration, the winner’s curse, power and decision rules (Chapters 1 and 8 follow it). Book 42 treated optimal stopping and sunk costs (Chapter 6’s procrastination model is its neighbour). This book’s risk notice, disclosure and Chapter 8 follow the pattern of Book 45.
Claims corrected after checking
Several points in the author’s framework needed correcting, supplementing or qualifying after checking. Each chapter says so where it arises, and the accompanying editorial notes list them together.
- The two extremes of choice overload. “More options, more conversions” does not hold, and neither is “too many options and customers do not buy” a general law. In Iyengar and Lepper’s jam study more people stopped at the 24-jam booth (60 % against 40 %), but only 2.8 % of those who stopped bought, against 29.8 % at the 6-jam booth; the often-quoted “3 % against 30 %” is the share of those who stopped, and per passer-by it is 1.7 % against 11.9 % (Iyengar and Lepper 2000). Scheibehenne and co-authors pooled 63 conditions from 50 experiments and found a mean effect of “virtually zero” with large differences between studies (Scheibehenne et al. 2010); Chernev and co-authors then identified four moderators (choice-set complexity, task difficulty, preference uncertainty and an effort-minimising goal) (Chernev et al. 2015). The author’s model shows that the effect of the number of options can be positive or negative, depending on comparison cost and preference uncertainty (Chapter 1).
- “Striking through a former price raises willingness to pay without limit.” Reference-price effects exist, but exaggerated reference prices lose credibility (Urbany et al. 1988; Compeau and Grewal 1998; Krishna et al. 2002); in the author’s model the gain in willingness to pay has an explicit ceiling, and once suspicion is added a very high struck-through price lowers it. The US Federal Trade Commission’s Guides Against Deceptive Pricing require a “former price” to be one at which the article was offered to the public in good faith, on a regular basis, for a reasonably substantial period (U.S. Federal Trade Commission 1964); Article 6a of EU Directive 98/6/EC (inserted by Directive 2019/2161) requires the “prior price” in a price-reduction announcement to be the lowest price applied in at least the 30 days before the reduction (European Parliament and Council of the European Union 1998, 2019), which the Court of Justice confirmed in Case C-330/23 (Court of Justice of the European Union 2024). The Malaysian rules could not be verified in this round and are not cited (Chapter 2).
- “TK1992 + Prelec” and probability weighting. The probability-weighting function of Tversky and Kahneman’s 1992 cumulative prospect theory is , with median estimates for gains and for losses (Tversky and Kahneman 1992); Prelec’s 1998 is a different, axiomatically derived form (Prelec 1998), and 0.65 is the value he uses in figures, not an estimate. Combining the 1992 value function with Prelec’s weights is legitimate but should be called a combination, not “TK1992”. The program finds that the one-parameter Prelec function closest to the 1992 gain curve has parameter 0.501 (Chapter 3).
- Separate versus cumulative weights. Cumulative prospect theory does not weight each outcome’s probability on its own; it weights the cumulative probability of “at least this good” and takes differences (rank dependence). Separable weighting violates first-order stochastic dominance: in the program it prefers a dominated lottery (certainty equivalent 53.73 against 37.38), and splitting one outcome into 50 pieces lifts the certainty equivalent to 316.63; cumulative weights do neither (Chapter 3).
- Parameters are not constants. and are medians from 25 graduate students in an experiment that did not pay according to choices (Tversky and Kahneman 1992). A meta-analysis by Brown and co-authors of 607 estimates in 150 articles from 1992 to 2017 gives a mean loss-aversion coefficient of 1.955 (95 % interval 1.820 to 2.102) (Brown et al. 2024); Gal and Rucker argue that current evidence does not show losses to be generally more impactful than gains (Gal and Rucker 2018). The program shows one fictitious decision-maker (true ) whose simulated choices at small, medium and large stakes, estimated with the same linear model, yield three “loss-aversion coefficients”: 1.43, 1.13 and 0.88 (Chapters 3 and 4).
- “The conversion gap between loss and gain frames.” Cumulative prospect theory describes individual choices between lotteries; it does not deliver a checkout conversion rate. Turning values into shares needs a link function with a scale parameter, and the size of any “framing gap” is set almost entirely by that parameter; in the program the same discount gives a gap of anywhere from 4.3 to 26.7 percentage points between frames. Framing effects are reliable but small to moderate in meta-analysis (Kühberger 1998). All such numbers are marked as illustrative (Chapter 4).
- “A pure MNL cannot produce the decoy effect.” True, but too narrow. In any random utility model, adding an option cannot raise the choice probability of an existing one (regularity) (Block and Marschak 1960), so the decoy (attraction) effect violates every random utility model, including probit, nested logit and mixed logit, not just the multinomial logit; the program shows that correlated errors can at most stop the decoy from taking share, never make the target rise (Chapter 5).
- Where decoy and compromise effects take hold. Tversky and Simonson’s 1993 model is a “context-dependent preference” model: relative advantage, with disadvantages weighted in a loss-averse way (Tversky and Simonson 1993); the program shows the attraction effect appearing only once the weight on relative advantage passes a threshold (between 0.01 and 0.03 in the example; below it the dilution by the decoy dominates), while the compromise effect also needs a convex disadvantage function. In multialternative decision field theory both effects depend on lateral inhibition and both vanish without it (Roe et al. 2001); with this book’s illustrative parameters, the program finds the compromise effect only when inhibition is broad and strong. In 38 studies Frederick and co-authors found the attraction effect essentially vanishing when attributes were presented perceptually or experienced, with an occasional reverse “repulsion effect” (Frederick et al. 2014); in e-commerce field data only 11 % to 25 % of customers noticed the dominance relation (Wu and Cosguner 2020) (Chapter 5).
- The origin of β–δ and the “procrastination–default threshold”. The quasi-hyperbolic discount function was first used by Phelps and Pollak for intergenerational saving (Phelps and Pollak 1968); Laibson applied it in 1997 to one person’s selves at different dates (Laibson 1997); the naive–sophisticated distinction comes from O’Donoghue and Rabin (O’Donoghue and Rabin 1999). The “friction cost” of multi-step forms and its completion thresholds are the author’s model, not a published result (Chapter 6).
- The source of the Guanzi quotation. “夫物多则贱,寡则贵,散则轻,聚则重” is found in the “Guo xu” chapter of the Guanzi; the author’s framework gives “Qingzhong jia / Guo xu”, and the “Qingzhong jia” transcription checked in this round does not contain it. The “balance numbers” (衡数) of the framework’s subtitle come from “衡无数也” and “然则衡数不可调耶” in “Qingzhong yi”. Quotations follow Li Xiangfeng’s Guanzi jiaozhu (黎翔凤 撰,梁运华 整理 2004); this round could only compare online transcriptions, and the author is asked to check the Zhonghua Shuju print edition (Chapter 7).
- Jiran’s policies and “price-expectation anchoring”. “贵上极则反贱,贱下极则反贵” and “上不过八十,下不减三十” are in the “Biographies of the Money-Makers” of the Shiji (司马迁 撰 1959); the former was discussed in Book 19, and whether Jiran is a person or the title of a book has long been disputed, a question this book leaves open. The two sentences concern mean reversion of prices and an official price band, not anchoring in the modern psychological sense; the program shows that “extremes reverse” holds only when supply shocks are stationary, not under a random walk (Chapter 7).
- The three-tier example prices. The prices in the author’s framework, US$9.99, 49.90 and 149.00, are the same as 5D Library’s own prices for its e-book, paperback and hardcover omnibus editions. The disclosure states this; the examples in Chapter 5 only illustrate model properties and do not validate those prices; the Chapter 8 experiment uses neither the author’s products nor these prices (Chapters 5 and 8).
How to read the boxes
As in the other books of 5D Library, different kinds of statement are marked separately.
Evidence
Evidence boxes report published research findings, with sources.
Calculation
Calculation boxes give short Python programs. Each runs unchanged in a standard Python 3.8 or later environment, uses only the standard library and fixes its random seed; its actual output is printed below it. All prices, pages, choices and experimental data in the book are fictional or synthetic, and parameters are illustrative, not estimates for any real market, business or customer.
Hypothesis · not established
Hypothesis boxes contain the author’s ideas, marked “not established”, together with the result that would show them wrong. The author’s own models and definitions are labelled “the author’s model” in the text.
Source text
Source boxes quote ancient texts with a plain-language rendering.
Note
Note boxes give reminders and explain choices.
About this edition
This is version 1.0.0, the first complete edition: a preface and eight chapters, in Chinese and English. It contains 16 Python programs and 5 vector figures. The text, boxes and code were drafted with an AI writing tool from the author’s title, framework and notes; the author is responsible for reviewing and revising them and for checking sources. AI-drafted text can get bibliographic details wrong; corrections are welcome through the 5D Library website. The choice-overload model of Chapter 1, the reference-price credibility model of Chapter 2, the attribute scales of the three-tier example in Chapter 5, the multi-step form model of Chapter 6 and the price-band model of Chapter 7 are the author’s models; their mathematical properties are not evidence of predictions about any real page or market.
Disclosure
The author uses the works of 5D Library, including this book, as the methodological basis of his community courses, the 5D Ascension Cards (Dimensional Shift Card Decks) and the Yafu AI knowledge base, and therefore has a commercial interest in the subject of this book. The book promotes no course, card deck or service; the Chapter 8 trial is designed to be run and analysed by researchers independent of the author’s business and does not recruit from the author’s community, students or customers; all the code and the study protocol are published openly for anyone to check. The author founded and runs Asia Influencer X and Yafu Global, which provide influencer marketing, live-commerce and digital marketing services to brands; that work involves the pricing pages and choice design discussed in this book, so he has a direct commercial interest in its subject. The three-tier example prices in the book (US$9.99, 49.90 and 149.00) are the same as 5D Library’s own prices for its e-book, paperback and hardcover omnibus editions, and the author has a direct interest in that price structure; the book’s analysis does not validate those prices. The book uses no data from the author’s businesses or their clients as evidence and does not promote their services.
References
Fan, Y. (2026). Choice Architecture, Cumulative Prospect Theory and Computational Behavioural Economics of Pricing Interfaces: From Loss-Aversion Value Functions and Asymmetric Dominance to Hyperbolic Discounting, Read Beside the Qingzhong Chapters of the Guanzi [选择架构、累积前景理论与销售界面价格锚定的计算行为学] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23836260 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).