INSTRUMENTS / SELECTION EXPLORER
Constructed exampleWiden the frame.
Keep every project in the model. Choose which ones enter the view.
01 / THE STORIES WE SEE
Set the population.
02 / WIDEN THE FRAME
The projects behind the stories.
100 projects · all known outcomes
Same successes. Now we can see what each strategy was tried against.
Inspect a strategy or project
Every visible project can be inspected without targeting a small mark.
The visible stories were only part of the population.
Count the attempts, not only the successes.
Save this experiment
MODEL & LIMITS
What the complete picture still cannot tell us
The missing projects are known because this example supplies them. These rates describe two groups; they do not establish that either strategy caused success.
- PopulationWho tried?
- SelectionWho is visible?
- InferenceWhat follows?
Inspect the complete counts and workings
| Strategy | Reached goal | Did not | Total | Success rate |
|---|---|---|---|---|
| Launch early | 20 | 60 | 80 | 25% |
| Pilot first | 10 | 10 | 20 | 50% |
Share among successes = strategy successes ÷ all successes. Strategy success rate = strategy successes ÷ all projects using that strategy. A zero denominator is undefined.
Assumptions, sources and evidence status
Original constructed fixture, constructed-project-selection-v1. Outcomes are supplied, not observed. Known unsuccessful projects are not the same as projects with unknown outcomes. Four nonnegative whole counts, total at most 200.