Resource map and question sheet

Print this page. It holds three sheets, one to a page. Sheet 1 is your question sheet. Sheet 2 is a blank resource map for your team's own question. Sheet 3 is the shared flood case, with its needs column already filled. Each team fills two maps, and the shared one comes first in the hour. All three sheets are portrait. Print CSS cannot turn one page of a document sideways in every browser, so a landscape sheet would come out right in one browser and wrong in the next; the grid is laid out to fit portrait instead. Sheet 3 is mapped from cards 1 and 2 of the dataset relay cards.

Your question sheet

Team   Date

Bring the half sheet from this afternoon. Line two goes in the role box.

Check in first. Nine minutes. One sheet per team: who is here, what you did, what is stuck, who does what next and by when. Nothing stuck yet is an answer. Envelope it when the timer stops.

Then the shared case. Every team maps the same question: is flood risk priced into house prices, and did a local flood change that? Use the two cards and one map sheet. The needs column is already filled. Write what you have and how you checked it, the gap, and who closes it. Fill the bottom lines: the row that stops you and its date, and what the question becomes if it does not close. Fourteen minutes. One team is drawn to read its map, blanks included. Attack by naming a specific row that is blank, unchecked or wrong. A point per attack that lands.

Then your own project. Write three candidate questions. Label the route of each, and make at least two of them different routes if you can. Give each its shape and the role that would act on it or pay for it, which is a role and not a company. Answer the four filters in writing for one candidate. Cross a second out and name the filter that killed it. For the third, write the one row that would kill it. Nine minutes. Then map the survivor only, twelve minutes. For every gap, write who closes it by when, or how the question moves. Unknown, with a name and date, counts.

One question a table. I come round once between 4:06 and 4:27. It is one question and one answer per table, not a consultation, so ask the one that is blocking you.

Then hand in. All four sheets go in the envelope before you leave, and your submitter brings it to the front. Your pitch is built from this sheet and the map. A question you kill today is a month you do not lose in March.

Written check-in, one sheet per teamnine minutes of writing, the first of the term
One shared case, one map, all nine teamsthen one team drawn to read it aloud
Your own question sheet and your own mapthree candidates, one killed, the survivor mapped
Four sheets in your envelopecheck-in, shared map, question sheet, own map
Candidate 1 run this one through all four filters
Route (1, 2 or 3) Shape, named and numbered Who acts on it, or pays for it

Filter 1 Could a team of three to five execute this in one spring semester?

Filter 2 Can you name the dataset, who holds it, and how you get it, checked rather than assumed?

Filter 3 Could the answer come out either way?

Filter 4 Would anybody act differently depending on it?

Candidate 2 the one you kill
Route (1, 2 or 3) Shape, named and numbered Who acts on it, or pays for it

Strike the question through. Killed by filter (circle one)   1 execute it   2 data named and checked   3 could come out either way   4 somebody acts

Candidate 3 name the row that would kill it
Route (1, 2 or 3) Shape, named and numbered Who acts on it, or pays for it

The one row that would kill it (circle one)   Data   Comparison   Skills   Time and money   Permissions   People and access

We map candidate   Map the survivor only.

If two survive all four, take the one whose worst gap closes soonest, then the one with a named person who acts.

Resource map · your own question

Team   Date
The question
Route (1, 2 or 3) Shape, named and numbered Who acts on it, or pays for it
What the question needs What we have, and how we checked Gap Who closes it, by when, or how the question moves
Data Which dataset, who holds it, and what it must contain
Comparison Where the counterfactual comes from: which units, which period Compared to what? Would those two groups have looked the same anyway?
Skills Language, method, who on the team has run one
Time and money How long: now, days, weeks, months, longer; dollars; the deadline it must meet
Permissions Ethics review by the IRB, a data-use agreement, terms of service, or none, and how long each takes; meeting 4 teaches the gates
People and access Who you know who holds the data, the decision, or the site
The row that stops you is , and it has to close by
If it does not close, the question becomes
Decision log, one line per move: what changed, when, why.
A "have" with no "how checked" counts as a need. A blank permissions row is not the same as "none"; write none if it is none. "Unknown, X checks by Friday" is an acceptable gap entry. The last column is where the plan lives.

The access ladder. Library databases, Statista, IBISWorld, SimplyAnalytics, FRED, EDGAR: now, on your netID or nothing. The Nielsen public-use extract (2019, three product categories): now, on your netID, and it is a practice dataset. Dewey: days; register with a Cornell address and wait for approval. Your own survey or online experiment: weeks; review by the IRB, the board that reviews research on people, then field it. Full Kilts-Nielsen data: months; a project application through Cornell Research Services, then space in the Regulated Research Environment. A data-use agreement with a firm: months, and it may fail; legal review on both sides. Census, BLS or NCHS confidential microdata via FSRDC: longer than this degree; federal project approval.

The map is not filled in once. A resource that turns out missing changes the question, the changed question changes the rows, and every redraw is the plan doing its job, not the plan failing.

Resource map · the shared case

Team   Date
Is flood risk priced into house prices, and did a local flood change that? Shape one, evaluation. Route three, from data, done honestly. Map it from relay cards 1 and 2. The needs column is filled for you.
Who acts on it, or pays for it a role, not a company
What the question needs What we have, and how we checked Gap Who closes it, by when, or how the question moves
Data Sale prices by parcel with coordinates and dates; a flood-risk measure by ZIP; a list of floods with dates and places
Comparison ZIPs with the same flood rating that did not flood, before and after the event Compared to what? Would those two groups have looked the same anyway?
Skills Joining two large files on ZIP; a parcel panel
Time and money Dewey days; no money; the transactions extract is large
Permissions Dewey's terms; no human subjects
People and access Nobody in particular
The row that stops you is , and it has to close by
If it does not close, the question becomes
Decision log, one line per move: what changed, when, why.
A "have" with no "how checked" counts as a need. A blank permissions row is not the same as "none"; write none if it is none. "Unknown, X checks by Friday" is an acceptable gap entry. The last column is where the plan lives.

The access ladder. Library databases, Statista, IBISWorld, SimplyAnalytics, FRED, EDGAR: now, on your netID or nothing. The Nielsen public-use extract (2019, three product categories): now, on your netID, and it is a practice dataset. Dewey: days; register with a Cornell address and wait for approval. Your own survey or online experiment: weeks; review by the IRB, the board that reviews research on people, then field it. Full Kilts-Nielsen data: months; a project application through Cornell Research Services, then space in the Regulated Research Environment. A data-use agreement with a firm: months, and it may fail; legal review on both sides. Census, BLS or NCHS confidential microdata via FSRDC: longer than this degree; federal project approval.

The map is not filled in once. A resource that turns out missing changes the question, the changed question changes the rows, and every redraw is the plan doing its job, not the plan failing.