Module 6 Lesson: Probability, Randomness, and Variation
Classroom Explanation
Randomness produces patterns. Runs, clusters and streaks appear in data generated by processes that have not changed at all, and our instinct is to explain them rather than to expect them.
In this module, the goal is to build intuition for chance, repeated trials, noise, and signal. The important part is not only the calculation. The important part is the thinking before and after the calculation.
Why This Matters
The common mistake is treating one small block as proof that the process changed. This mistake can make a report sound more confident than the evidence deserves.
Key Ideas
1. Probability is about long-run chance, not certainty on the next attempt.
Probability describes chance across repeated situations. It does not promise what the very next result must be.
2. Small blocks can look very different even when the process has not changed.
Short runs are noisy. A block of 10 attempts can look unusually good or bad even when the underlying process is unchanged.
3. Larger repeated trials usually give a steadier estimate.
More repeated trials usually reduce random wobble. The estimate becomes steadier, though it still may not be perfect.
Worked Example
In 10 attempts, success rates range from 30% to 70%. In 100 or 200 attempts, the observed rate sits close to 51%.
Here is the dataset used in this module.
| trial_block | attempts | successes | observed_rate |
|---|---|---|---|
| B01 | 10 | 7 | 0.70 |
| B02 | 10 | 3 | 0.30 |
| B03 | 10 | 6 | 0.60 |
| B04 | 10 | 5 | 0.50 |
| B05 | 10 | 4 | 0.40 |
| B06 | 50 | 27 | 0.54 |
Work out what to expect before judging what happened. If a process fails five per cent of the time, one failure in twenty is the expectation and three is unremarkable — asking how surprising something is requires knowing what unsurprising looks like.
How To Think Through It
- State the rate you would expect.
- Compute what that means over this many observations.
- Compare the observed result with that expectation.
- Ask whether ordinary variation explains the gap.
- Say what further evidence would settle it.
Common Mistake
The common mistake is treating one small block as proof that the process changed.
How To Write The Result
The 10-attempt blocks vary a lot, so the short-run result should be treated as random variation unless more evidence supports a real change.
Practice Prompt
Compare the 10-attempt blocks with the 100- and 200-attempt blocks.
Takeaway
Before explaining a pattern, check whether the process would produce it anyway.