Scale a sample without inventing certainty
Separate a sample count from a population estimate. The method of selection determines whether a broader inference is reasonable. A large convenience sample may still be biased if it systematically excludes part of the population.
- Mean is sensitive to extreme values.
- Median is the middle value after ordering.
- Probability is favorable outcomes divided by total possible outcomes.
- A sample must be representative before generalizing.
- Correlation does not prove causation.
Estimate a count from a sample proportion
A random sample of 150 students contains 42 who walk to school. Estimate the number who walk in a school of 1,200 students.
The sample proportion is 42/150 = 0.28. Applying that proportion gives 0.28(1,200) = 336 students. Report 336 as an estimate, not an exact head count. A different random sample could produce a slightly different proportion.
Keep the setup accurate
- Using the wrong denominator in probability.
- Confusing median with mean.
- Assuming a trend line passes through every point.
Use the method while it is fresh.
Can a survey of walkers alone answer this question?
No. Asking only students arriving on foot builds the outcome into the sampling method. A population estimate needs a sampling process that gives the relevant student groups a fair opportunity to be selected.