THINK
Should You Trust the Model?
A decision activity set during an outbreak. A model gives you a number. What do you actually do with it?
- Type
- Activity
- Difficulty
- Intermediate
- Length
- 15 min activity
During the COVID pandemic, Professor Juneja's team at TIFR Mumbai built epidemiological models to estimate how the virus was spreading through the city and what might happen next. Those numbers did not make decisions on their own. People did, under real uncertainty, with real costs on every side.
This activity puts you in that seat. For each decision, pick the option you would defend, then read what your choice quietly optimises for and what it puts at risk. There is no scored answer. The point is to see how a model informs a decision without ever making it for you.
THREE CALLS, ONE IMPERFECT FORECAST
A model projects that cases will double in the next ten days if nothing changes. The model has wide error bars: the real doubling time could be seven days or fourteen. You lead the response.
DECISION 1 OF 2
The model's central estimate says hospitals overflow in three weeks. Do you order a strict lockdown now, or wait for more data?
DECISION 2 OF 2
Two teams hand you two different models with two different numbers. How do you use them?
A model is a disciplined way to reason about an uncertain future, not a crystal ball. Its greatest value is often the range it gives you and the assumptions it forces into the open, not the single headline number.