๐ Intro to Statistics ยท Lesson 4 of 7
The normal curve and z-scores
The 68โ95โ99.7 rule and comparing values on different scales.
The bell curve
Many measurements โ heights, measurement errors, many test scores โ follow a roughly normal distribution: symmetric, bell-shaped, described by its mean ฮผ and standard deviation ฯ.
The 68โ95โ99.7 rule
| Within 1 ฯ of the mean | About 68% of values |
|---|---|
| Within 2 ฯ | About 95% |
| Within 3 ฯ | About 99.7% |
Example
SAT Math scores are roughly normal with mean โ 520 and ฯ โ 115 (illustrative numbers). About 68% of students score between 405 and 635, and about 95% between 290 and 750. A score of 750 is about 2 ฯ above the mean โ better than about 97.5% of test-takers.
z-scores
z = (x โ mean) / standard deviation
Exam A: you scored 82, class mean 70, SD 6 โ z = (82 โ 70)/6 = 2.0
Exam B: you scored 90, class mean 85, SD 10 โ z = (90 โ 85)/10 = 0.5The 82 was the more impressive score relative to the class, even though 90 is higher.
Tip
๐ก A z-score says how many standard deviations a value is from the mean. Positive = above, negative = below. |z| > 2 is unusual; |z| > 3 is very rare.
โ๏ธ Practice: your turn
- Use the 68โ95โ99.7 rule on heights: mean 69 in, SD 3 in โ what range holds 95%?
- Compute z-scores for two of your own exam scores (estimate class mean and SD)
- Find what percent of a normal distribution is above z = 1
- Sketch a bell curve and mark ฮผ, ยฑ1ฯ, ยฑ2ฯ
โ๏ธ Scientific calculator with a statistics tab (mean, median, standard deviation)
๐ง Quick quiz
Answer all 4 questions correctly to mark this lesson as done.
0 of 4 correct
All lessons in Intro to Statistics
- ๐ฒ 1. Data, variables and samplingCategorical vs. quantitative, populations, samples and bias
- ๐ฏ 2. Mean, median and modeMeasures of center, and which one to use when data are skewed
- โ๏ธ 3. Spread: range, IQR and standard deviationHow spread out the data are, and how to read a box plot
- ๐ 4. The normal curve and z-scoresThe 68โ95โ99.7 rule and comparing values on different scales
- ๐ 5. Correlation and regressionScatterplots, r, the line of best fit โ and causation
- ๐ฏ 6. Confidence intervalsEstimate ยฑ margin of error, and what โ95% confidentโ means
- โ๏ธ 7. Hypothesis tests and p-valuesNull and alternative, the p-value, significance and errors