3 · Finding the question
Wednesday, September 9, 2026
What you should be able to do after this meeting. Judge an idea against four tests: is it important, is it feasible, is it new, and is it a fit for your team. Say which of seven shapes your question takes, and say where the idea came from and what that route leaves missing.
What makes an answer valuable. Public value is shared, so nobody can be stopped from using the answer and it gets published. Private value can be kept, so somebody pays for it. Both are fine here. The test is not whether one person acts, it is whether you can keep the answer.
Ideas arrive by three routes. From the world, from a study, and from data. Route one lacks data, route two lacks novelty, route three lacks a decision.
Four questions kill most bad ideas. Can three to five people do it in a semester? Can you name the data, who holds it, and how you get it? Could the answer come out either way? Would anybody act differently?
The resource map is six rows against what the question needs: data, comparison, skills, time and money, permissions, people and access. A resource that turns out missing changes the question, and every redraw is the plan working.
Due today
Your team’s question sheet, with each candidate run through the four tests and the ones that fail killed on the record. Before you leave.
Session A · 2:05 to 3:15
The hour has three parts. First, what makes an idea good, which is four tests. Second, the shapes a project can take. Third, where ideas come from. The order is deliberate. You already have ideas, so the useful thing is not another way to generate them. It is a way to sort them quickly, before you spend a month on one that was never going to work.
Four tests, and a good idea passes all four
Important. Somebody is better off knowing the answer. Importance is not the size of the topic, it is the size of the decision that changes. A large topic with no decision attached is not important, it is broad. The test is to name the person or the organisation and say what they would do differently once they knew.
Feasible. You can get the data, in the time you have, with the skills on the team. Feasibility is about your resources, not about whether the question could be answered by somebody, somewhere. Most projects die here, and they die in March rather than in September because nobody checked in September. The test is to name the file, name who holds it, and say how you get a copy. Checked, not assumed.
Novel. Nobody has already answered it, and you know that because you went and looked. New does not mean nobody has worked on the topic. It means this answer is not already on the shelf. A marginal change to published work is new, but only just, and you have to be willing to say so out loud. The test is to say who has come closest, and what they did not do.
A good fit for you. A question can be important, feasible and new and still not be yours. Fit is two things: the skills your team actually has, and whether you will still care in March. A question nobody on the team can execute is somebody else’s project, however good it is.
The four pull against each other. The most important questions are usually the least feasible, and the most feasible are usually the least new. A project is a choice about where to sit between them, not a search for a question that scores full marks on all four.
An answer has value when somebody is stuck without it
The importance test has a second half worth stating on its own. Value comes in two kinds and the difference decides who your reader is.
Public value is shared. Many people are better off and you cannot stop anyone using the answer, so it is hard to charge for and it gets published. Private value can be kept. One person’s decision changes and that person can keep the answer to themselves, so somebody pays for it. The test is not whether one person acts, it is whether the answer can be kept from everybody else.
Both are fine in this course. Private value is easier to sell, and public value is easier to publish. Private value also decays. McLean and Pontiff collected 97 published rules for picking stocks that beat the market. On average a rule earned 26 percent less on later data than in the study that found it, and 58 percent less once it was in print. So a rule is worth money while one trading desk is the only one using it, and about half that once it is in a journal. If an answer is worth money and nobody has produced it, ask why, and read to check that nobody has.
A project can take different shapes
Seven shapes, and naming yours tells you what you have to go and get. The full table with what each shape can claim and how each one fails is below. The same interest can usually be turned into three of them, so pick the one you can answer rather than the one that sounds biggest.
Ideas arrive in three ways
From the world. A problem you saw, and it usually arrives without being sought: from the news, from a conversation with a colleague, or from a job you have done yourself. The good ones often come out of an unexpected conversation rather than out of sitting down to think. You start with the problem and go looking for data afterwards, so data is the thing most likely to be missing.
From previous work. You take an existing idea, generally published, and change it at the margin. The change is usually a new setting, a new period, or a limitation the original authors name themselves. Read the limitations section, because that is where authors write down what they could not do. This route tends to be marginal, and saying so is honest rather than modest. For one semester it is the right size. What is missing here is novelty.
From data. You start from a file you can reach and work backwards to a question it could answer. Data is usually the binding constraint, so this route makes the constraint visible on day one. Expect to go back and forth: you get the file, find what it does not contain, and revise the question. That revising is not the plan failing.
Working backwards from data is honest in exactly one form. Ask what this data could answer, which of those anybody would act on, and what it does not contain that you would need. You are allowed to walk away from the data. Starting from data is the only route where the data is already in hand, so nothing looks missing, and only the question of who would act can kill it.
Public value and private value, all seven examples
Two of these reach the screen. All seven are here, because the label is the easy half and the last column is the part you have to write on your question sheet.
| Example | Label | Why | Who pays or acts |
|---|---|---|---|
| When a stock suddenly draws a lot of attention, does its price rise above what the company’s earnings justify, and then fall back? (Team D’s sentence) | private | An answer you can trade on is one you can keep to yourself, and it is worth nothing the moment everybody knows it. The public version, whether small investors lose money, is a different question with a regulator as its reader | An investment fund pays while it is the only one holding the answer. Barber and Odean 2008 on buying what is in the news and Da, Engelberg and Gao 2011 on the fall back answered the first version of it, so the value lies in what they left open, which is what reading checks |
| When extreme weather hits a poor country, how much damage does it leave, and is that enough to change decisions? (Team G’s sentence) | both | Governments and aid agencies act on a published number and nobody can be stopped from using it, but the same number prices insurance that pays out automatically once a storm passes a threshold | Finance ministries and the World Bank act; reinsurers would pay to price the cover |
| Does the claim show up in revenue and stock price, and does the market only believe credible firms? (Team E’s sentence) | both | Protecting investors is public work and a published answer changes how filings are reviewed; the same answer tells a short seller which firms to bet against | The SEC, the US securities regulator, and auditors act; analysts and short sellers pay |
| Did salary disclosure laws change what employers post? (lead 9) | public | Every worker in the next state benefits, the answer cannot be kept, and the buyer is a legislature, which does not pay researchers | The next state’s legislators act; there is nobody to charge, so it gets published |
| Which store locations sell less than the area they draw customers from should deliver? (lead 7) | private | One operator acts on it and would never publish the answer, but the data on who lives near each store is bought from a vendor, so what stops you is access to the chain rather than the data | The chain’s head of operations pays, or lends its own transaction file in exchange for the answer |
| Does a logo aligned to the drinking spout raise recognition and intent to buy? (the soda-can project, from an earlier cohort) | private | One brand’s packaging decision, and the answer is worth nothing to anyone else | A soda company gave them thousands of cans, which is a firm paying in kind for an answer it wanted |
| Does disclosing that a video is AI-generated change how people rate it? (lead 28, a neighbour of Team J) | both | Platforms and brands would pay to know before a disclosure rule lands; the rule itself is public | A platform’s head of trust and safety pays; a regulator writing a disclosure rule acts |
The seven shapes in full
The slide shows each shape, the question it asks and an example. Here are all four columns.
| Shape | Asks | Can claim | Dies when |
|---|---|---|---|
| 1 Evaluation | Did it work? A decision was already made somewhere, and you estimate its effect | A causal effect, after the fact, if you can measure the thing you care about before and after, both for the group that got the change and for a credible comparison group | There is no credible comparison group, or the change was taken up precisely by the units that were already moving |
| 2 Preference and pricing | What would people choose, and what would they pay? The decision is not yet made, and what it needs is what people value | A causal effect on what people chose in your task, which is a real limitation and not a footnote | What people say they would do is reported as what they did, or forty classmates are generalised to a market |
| 3 Prediction and risk | What happens next, and how wrong could we be? The decision needs a number about the future, not a cause | How accurately it predicts, tested on later data it has not seen, against an honest baseline, and explicitly not a cause | It is never tested on later data, or the team quietly starts recommending what to change off a model that only predicts |
| 4 Diagnosis and benchmarking | Where is the gap, and what explains it? Something varies more than it should and nobody knows why | Description, plus a theory about why that somebody could later test, honest about being a first step | Description is dressed up as a cause, or the team stops at a dashboard without naming a decision |
| 5 Feasibility and landscape | Is this opportunity real, and what constrains it? | A synthesis of what is known, plus at least one piece of original data the team analyses itself | It becomes a summary of what other people already found, with no evidence of the team’s own |
| 6 Measurement | How much of this is there, and how would you know? Nobody has counted it, so the project is building the number | A defensible quantity, and what it shows, provided you say plainly what it counts and what it leaves out | The measure is never checked against anything outside itself, or what it excludes is never stated |
| 7 Valuation | What is this worth? You price the thing against things that already carry prices | A price, provided the things you compare against are genuinely comparable and you say what makes them so | The comparison set is chosen until the answer looks right, or the traits doing the work are not observable to anybody else |
Session B · 3:30 to 4:30
Ten-minute team check-in first, in writing, on the sheet on your table. Who is here, what you did since last week, what is stuck, and who does what next by name and by date. In week three, nothing is stuck yet is an acceptable answer.
Then your own project, which is the half you hand in. You write candidate questions, put each one through the four tests, and kill the ones that fail. The tests you cannot run on your own work are run by another team, because every team thinks its own question matters and a stranger is the only honest judge of that. Your sheet goes in before you leave, and it is what your pitch gets built from.
A question you kill in September is a month you do not lose in March.
Handouts
Resource map and question sheet. The question sheet you fill this afternoon, and a one-page template for setting what a question needs against what your team already has.
Team check-in sheet. One sheet per team, and you fill it every week from now on. Who is here, what you did, what is stuck, who does what next with a name and a date, and one question for me.
Project teams. The nine questions this class is asking, and who is working on each.
Where to read more
Adams, Khan, Raeside and White on the research cycle, and on the literature review and critical reading. That DOI resolves to the 2007 edition, titled Research Methods for Graduate Business and Social Science Students, so check the chapter numbers against whichever copy you open.
Sarasvathy on effectuation is the theory paper behind “start from your means”. Read it for the list of resources it implies, not for a law about entrepreneurs.
Arrow, Economic welfare and the allocation of resources for invention is where public value comes from. Knowledge is expensive to make and cheap to copy, so whoever makes it cannot charge the people who benefit.
McLean and Pontiff is the private-value paper, and it is the most useful twenty pages in this list if your interest is finance.
Sandberg and Alvesson separate looking for a hole in the literature, which they call gap-spotting, from challenging what the literature takes for granted, which they call problematization. It is a distinction for people writing for journals rather than for this course, so read it only if you want the research-side version of today.