3. Multipliers and the empirical rule
The numbers you carry in on your sheet, and how to use them without a table.
Multipliers
- The five to write down: 80% is 1.282, 90% is 1.645, 95% is 1.96, 98% is 2.326, 99% is 2.576.
- α is what is left over: α = 1 − level. For 95%, α = 0.05.
- How to find a multiplier: split α evenly between the two tails. Each tail gets α/2. The multiplier is the z-value with area 1 − α/2 to its left. For 95% that is the z with 0.975 to the left, which is 1.96.
- For t: same idea, in the row for n − 1 degrees of freedom. A t multiplier is always larger than the z multiplier, and it gets closer to z as n grows.
You need the multiplier for a 98% interval and you have a Z table. Describe how to find it.review class
Answer
α = 1 − 0.98 = 0.02, so each tail has 0.01. Look inside the Z table for the area 0.99 to the left (that is 1 − 0.01) and read off the z-value, about 2.33. The interval uses plus and minus that value, which leaves 98% in the middle.
For a 95% interval, a student looks up the z-value with 0.95 to its left. What went wrong?
Answer
That leaves 5% in the upper tail alone and gives 1.645, which is the multiplier for a 90% interval. A two-sided 95% interval has 2.5% in each tail, so the area to the left must be 0.975, giving 1.96.
The same data gives a 95% t interval with n = 10 and you compare it with a z interval. Which multiplier is larger and why?
Answer
The t multiplier is larger (2.262 for 9 degrees of freedom against 1.96). Estimating σ with s adds uncertainty, and the heavier tails of the t distribution make the interval wider to account for it.
The empirical rule
- When it applies: bell-shaped distributions. That covers the normal, Z, and T once n is around 30.
- The rule: 68% within 1 standard deviation of the mean, 95% within 2, 99.7% within 3.
- Tails come from symmetry. Whatever is not in the middle is split evenly between the two tails.
| z | −3 | −2 | −1 | 0 | 1 | 2 | 3 |
|---|---|---|---|---|---|---|---|
| Area to the left | 0.15% | 2.5% | 16% | 50% | 84% | 97.5% | 99.85% |
| Region | Percent | How |
|---|---|---|
| Beyond 1 SD on one side | 16% | (100 − 68) / 2 |
| Beyond 2 SD on one side | 2.5% | (100 − 95) / 2 |
| Beyond 3 SD on one side | 0.15% | (100 − 99.7) / 2 |
| Between 1 and 2 SD on one side | 13.5% | (95 − 68) / 2 |
| Between 2 and 3 SD on one side | 2.35% | (99.7 − 95) / 2 |
| Between the mean and 1 SD | 34% | 68 / 2 |
Using it without a table
Scores are approximately normal with mean 70 and standard deviation 10. About what percent are above 90?exam review
Answer
z = (90 − 70)/10 = 2. About 95% of scores are within 2 standard deviations, between 50 and 90. The other 5% is split evenly, so about 2.5% are above 90.
Same scores. About what percent are between 80 and 90?exam review
Answer
80 is 1 standard deviation above the mean and 90 is 2 above. Between −2 and 2 is 95%, between −1 and 1 is 68%. The difference, 27%, is split between the two sides, so about 13.5% are between 80 and 90.
Same scores. About what percent are below 60?
Answer
60 is 1 standard deviation below the mean. 68% are within 1, so 32% are outside, half on each side. About 16% are below 60.
25 people each weigh on average 160 pounds with a standard deviation of 30 pounds, independently and normally. The elevator holds 4300 pounds. About what is the chance the load exceeds the limit?class example
Answer
The total has mean 25 × 160 = 4000 and variance 25 × 30² = 22500, so its standard deviation is 150. z = (4300 − 4000)/150 = 2. By the empirical rule about 2.5% of the time the load exceeds the limit. The table value is 0.0228.
A t statistic from a sample of 40 comes out at 2.1. Roughly how unusual is that?review class
Answer
With n = 40 the t distribution is close to the standard normal, so the empirical rule applies. About 95% of values fall between −2 and 2, so a value beyond 2 happens only about 2.5% of the time in that tail. It is unusual.
Why can the empirical rule be used on a t distribution?review class
Answer
The t distribution is bell-shaped and centred at zero, like the standard normal. As the sample size grows it gets closer to the standard normal, and by about n = 30 the two are nearly the same. So 68%, 95% and 99.7% fall within 1, 2 and 3 on the t scale as well.
Where does the 1.96 in a 95% interval connect to the empirical rule?
Answer
The empirical rule says about 95% of a normal distribution is within 2 standard deviations. The exact number is 1.96. A 95% interval reaches 1.96 standard errors either side of the estimate for the same reason.