A probability density curve, also
Module 7 known as a probability density
Normal function (PDF), is a curve that
describes the probability distribution
Standard of a continuous random variable.
Distribution It shows the likelihood ( ) of the variable taking on a specific value
...or, the beauty of the Bell Curve (a) or the probability that the random variable will fall within a
for populations AND for samples range of values (area under the curve within a given interval).
The normal distribution is a Standardize
PDF defined by its mean (μ)
and standard deviation (σ):
The mean, median, and mode are equal
and located at the center 950 970 1010 1030 1050 1070
990 1010 −3 −2 −1 00 +1 +2 +3
A Normal Distribution The Standard Normal Distribution
Many natural and human
phenomena (heights, test scores,
"Bell Curve"
measurement errors, etc.) follow
Aprox
68.2% Standardization tool
a normal distribution.
Standard Normal
This makes it a powerful tool for
Distribution
modeling and prediction. Empirical Rule:
About 68%, 95%, and
19.1% 19.1%
almost all of data lie
Normal is: no outliers OR
Aprox within 1, 2, and 3
no skewness 95.4% 15.0% 15.0% standard deviations of
the mean, respectively
Aprox
9.2% 9.2%
99.8% 0.5% 0.5%
4.4% 4.4%
0.1% 1.7% 1.7% 0.1%
Z-Score −4 −3.5 −3 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 2.5 3 3.5 4
Standard −4σ −3σ −2σ −1σ 0 +1σ +2σ +3σ +4σ
Deviation
0.1% 2.3% 15.9% 50% 84.1% 97.7% 99.9%
Cumulative
Percent
1% 5% 10% 20 30 40 50 60 70 80 90% 95% 99%
normalcdf (lower_bound,upper_bound,[mean,sigma]) Returns an
area with
Computes the area (probability) under a normal (bell- values
shaped) curve between two specified values (endpoints) for between 0
a given mean and standard deviation (standardized or not). and 1
invNorm (area,[mean,sigma]) For a standard normal
Given a cumulative probability (area to the distribution (μ=0, σ=1), it returns
left under the normal curve or percentile), the z-score.
it calculates the corresponding z-score For any normal distribution, it
or raw value from the distribution. returns a value of x.
Can we use the normal standard Variability: Most sample means (while
distribution for samples? different) will be close to the true
YES! When we take a sample from a population and population mean.
calculate the mean (average) of that sample, we get Shape: If the sample size is large
one value for the sample mean . If we repeat this enough (typically n>30), the
process many times—taking new samples of the distribution of these sample means
same size from the same population and calculating will be approximately normal (bell-
the mean each time—we get many sample means. shaped), even if the original
population is not normal!!!
The sampling distribution of the sample mean is the
This is the Central Limit Theorem.
probability distribution of all possible sample means
NOTE: If sample size n<30 but the
from all possible samples of a given size from the
population is normally distributed,
population.
then the normal standard can be use
for these samples.
The same powerful idea behind Center: The mean of all possible sample
the Central Limit Theorem—which explains why means is equal to the population mean
averages (sample means) from large samples tend to (μ):
follow a normal distribution—can be extended to
sample proportions if the size of the sample: Spread: The standard deviation of the
sampling distribution (called the
and
standard error) is smaller than the
standard deviation of the population.
The population proportion is the fraction of a
population with a certain trait.
The sample proportion is
the fraction of a sample with that trait.
If you take many samples and compute each
time, the distribution of those values is called σx shrinks as n increases!
the sampling distribution of the sample
proportion.
"The Central Limit Theorem
is like magic: turn MOST
OF data into a normal curve.
Just add more samples!"
