Battery lifetimes for a phone model have hours. What is , in hours, for random samples of 25 phones? Give the exact value.
The sampling distribution of a sample mean
How varies: centered at , spread , and normal when the population is normal or .
Key ideas
- Center
With random sampling, is centered at for every sample size.
- Spread
. It is always smaller than for , and population skewness does not change it.
- Shape
Normal population: is normal for any . Non-normal population: approximately normal once (the course rule of thumb; more skew needs more data).
- Reasoning backwards
If looks normal with a tiny (say 4), the population must be about normal. With you cannot tell the population's shape from the sampling distribution.
- Population versus sample versus sampling distribution
The first two show individual values (all units, or the units you drew). The third shows one per sample.
Formulas
Commute times
Commute times for a city's workers are right-skewed with minutes and minutes. Describe for random samples of 36 workers and of 144 workers. What would change for samples of 4?
Center: 28 minutes for every sample size.
n = 36: minutes. Since , the shape is approximately normal despite the skewed population.
n = 144: minute: four times the data, half the spread.
n = 4: minutes, and with only 4 workers per sample the right skew carries over, so the normal model is not justified.
at and at ; at the spread is 6 and the shape is still skewed.
Check yourself
Household water bills in a town are strongly right-skewed. A clerk draws random samples of 9 bills and plots the sample means. The sampling distribution of is most likely:
A simulation of for random samples of size 4 looks perfectly bell-shaped. What can you conclude about the population?
A researcher increases the sample size from 16 to 64 when sampling from a skewed population. Select every statement that is true.
A quality inspector weighs the 50 bags in one random sample and draws a histogram of those 50 weights. This histogram shows:
Commute times have minutes. Quantity A: the standard deviation of individual commute times. Quantity B: for random samples of 9 commutes.