Data-Driven Decision Making
What this domain covers
This domain tests whether you can turn analysis into defensible decisions. Understand the decision cycle: define the question, identify required data, analyse, interpret, decide, act, and review. Know how to frame a measurable question and how to distinguish leading from lagging indicators. Be comfortable with uncertainty: sampling error, confidence, margin of error, and the difference between statistical significance and practical importance.
Concepts to master
Recognise cognitive biases that corrupt interpretation, including confirmation bias, anchoring, survivorship bias, and selection bias. Know the value of a control or comparison group and why A/B testing needs a clear success metric defined in advance. Understand that base rates matter: a test with high accuracy can still produce many false positives when the condition is rare. Appreciate data storytelling: lead with the finding, support with evidence, state limitations, and recommend an action.
Traps in questions
- Treating a statistically significant result as automatically worth acting on.
- Changing the success metric after seeing the results.
- Ignoring base rates when interpreting a screening test.
- Drawing conclusions from a group that differs systematically from the population (selection bias).
- Presenting a recommendation with no stated assumptions or limitations.
Worked rule of thumb
Use the so-what test: for every finding, state the decision it would change and the evidence behind it. If no decision changes, the finding is context, not insight. Pair this with an explicit limitation statement, such as small sample size or short observation window, so decision-makers can weigh the risk.