Example

Two strains of mice: A and B.

Measure cytokine IL10 (in males all the same age) after treatment.

A ● ●● ● ●● ● ●

B ● ● ● ● ● ●●●

500 1000 1500 2000 2500

IL10

A ● ● ● ● ●● ● ●

B ● ● ● ●● ●●
●

500 1000 1500 2000 2500

IL10 (on log scale)

Point: We’re not interested in these particular mice, but in as-

pects of the distributions of IL10 values in the two strains.
1

Populations and samples

We are interested in the distribution of measurements in the

underlying (possibly hypothetical) population.

Examples: • Infinite number of mice from strain A; cytokine response to

treatment.
• All T cells in a person; respond or not to an antigen.
• All possible samples from the Baltimore water supply; concen-
tration of cryptospiridium.
• All possible samples of a particular type of cancer tissue; ex-
pression of a certain gene.

We can’t see the entire population (whether it is real or hypothetical),

but we can see a random sample of the population (perhaps a set of
independent, replicated measurements).

2
 Parameters

The object of our interest is the population distribution or, in

particular, certain numerical attributes of the population distribution
(called parameters).
Population distribution Examples:
• mean
• median
• SD
• proportion = 1
σ • proportion > 40
µ • geometric mean
0 10 20 30 40 50 60 • 95th percentile

Parameters are usually assigned greek letters (like θ, µ, and σ ).

3

Sample data

We make n independent measurements (or draw a random sample

of size n).

This gives X 1, X 2, . . . , X n independent and identically distributed

(iid), following the population distribution.

Statistic: A numerical summary (function) of the X ’s (that is, of the data).

For example, the sample mean, sample SD, etc.

Estimator: A statistic, viewed as estimating some population parameter.

(estimate)

We write: θ̂ an estimator of θ X̄ = µ̂ an estimator of µ

p̂ an estimator of p S = σ̂ an estimator of σ

4
 Parameters, estimators, estimates

µ • The population mean

• A parameter
• A fixed quantity
• Unknown, but what we want to know

X̄ • The sample mean

• An estimator of µ
• A function of the data (the X ’s)
• A random quantity

x̄ • The observed sample mean

• An estimate of µ
• A particular realization of the estimator, X̄
• A fixed quantity, but the result of a random process.

Estimators are random variables

Estimators have distributions, means, SDs, etc.

Population distribution

−→ X 1, X 2, . . . , X 10 −→ X̄
σ
µ

0 10 20 30 40 50 60

3.8 8.0 9.9 13.1 15.5 16.6 22.3 25.4 31.0 40.0 −→ 18.6
6.0 10.6 13.8 17.1 20.2 22.5 22.9 28.6 33.1 36.7 −→ 21.2
8.1 9.0 9.5 12.2 13.3 20.5 20.8 30.3 31.6 34.6 −→ 19.0
4.2 10.3 11.0 13.9 16.5 18.2 18.9 20.4 28.4 34.4 −→ 17.6
8.4 15.2 17.1 17.2 21.2 23.0 26.7 28.2 32.8 38.0 −→ 22.8
6
 Sampling distribution

Distribution of X̄
Population distribution
n=5

10 15 20 25 30 35 40

n = 10

σ
µ

0 10 20 30 40 50 60 10 15 20 25 30 35 40

n = 25

Sampling distribution depends on:

• The type of statistic 10 15 20 25 30 35 40

• The population distribution n = 100

• The sample size

10 15 20 25 30 35 40

Bias, SE, RMSE

Population distribution Dist'n of sample SD (n=10)

σ
µ

0 10 20 30 40 50 60
5 10 15

Consider θ̂, an estimator of the parameter θ.

Bias: E(θ̂ − θ) = E(θ̂) − θ

Standard error (SE): SE(θ̂) = SD(θ̂).

q p
RMS error (RMSE): E{(θ̂ − θ) } = (bias)2 + (SE)2.
2

8
 Example: heights of students

14
12 mean = 67.1
10 SD = 3.9
Frequency

8
6
4
2
0

60 65 70 75

Height (inches)

Example: heights of students

Population distribution Dist'n of Sample Ave (n=10)
mean = 67.1 mean = 67.1
SD = 3.9 SD = 1.2

60 65 70 75 60 65 70 75

Height Height (inches)

Dist'n of Sample Ave (n=5) Dist'n of Sample Ave (n=25)

mean = 67.1 mean = 67.1
SD = 1.7 SD = 0.68

60 65 70 75 60 65 70 75

Height (inches) Height (inches)

10
 The sample mean

Population distribution Assume X 1, X 2, . . . , X n are iid with

mean µ and SD σ .

σ
µ
Mean of X̄ = E(X̄ ) = µ.
0 10 20 30 40 50 60

Bias = E(X̄ ) − µ = 0.
√
SE of X̄ = SD(X̄ ) = σ/ n.

RMS error of X̄ =
p √
(bias)2 + (SE)2 = σ/ n.

11

If the population is normally distributed

Population distribution

If X 1, X 2, . . . , X n are iid
normal(µ,σ), then σ

√ µ
X̄ ∼ normal(µ, σ/ n).
Distribution of X

σ n

12
 Example
Suppose X 1, X 2, . . . , X 10 are iid normal(mean=10,SD=4)
Then X̄ ∼ normal(mean=10, SD ≈ 1.26); let Z = (X̄ – 10)/1.26.

Pr(X̄ > 12)?

1.26 ≈ 1 ≈ 5.7%
10 12 −1.58 0

Pr(9.5 < X̄ < 10.5)?

≈ ≈ 31%
10 0
9.5 10.5 −0.40 0.40

Pr(|X̄ − 10| > 1)?

≈ ≈ 43%
9 10 11 −0.80 0 0.80

13

Central limit theorm

If X 1, X 2, . . . , X n are iid with mean µ and SD σ .

and the sample size (n) is large,

√
then X̄ is approximately normal(µ, σ/ n).

How large is large?

It depends on the population distribution.
(But, generally, not too large.)

14
 Example 1

Distribution of X̄
Population distribution
n=5

10 15 20 25 30 35 40

n = 10

σ
µ

0 10 20 30 40 50 60 10 15 20 25 30 35 40

n = 25

10 15 20 25 30 35 40

n = 100

10 15 20 25 30 35 40

15

Example 2

Distribution of X̄
Population distribution n = 10

50 100 150

n = 25

0 µ 50 100 150 200 50 100 150

n = 100

50 100 150

n = 500

50 100 150

16
 Example 2 (rescaled)

Distribution of X̄
Population distribution n = 10

50 100 150

n = 25

0 µ 50 100 150 200 20 40 60 80 100

n = 100

20 25 30 35 40 45 50 55

n = 500

30 35 40

17

Example 3
Population distribution Distribution of X̄
n = 10

0 0.1 0.2 0.3 0.4 0.5

n = 25

0 1

0 0.05 0.1 0.15 0.2 0.25 0.3

{X i} iid n = 100

Pr(X i = 0) = 90%
Pr(X i = 1) = 10%
0 0.05 0.1 0.15 0.2

E(X i) = 0.1; SD(X i) = 0.3 n = 500

P
X i ∼ binomial(n, p)
0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13 0.14
X̄ = proportion of 1’s

18
 The sample SD

Why use (n – 1) in the sample SD?

sP
(xi − x̄)2
s=
n−1

If {X i} are iid with mean µ and SD σ , then

E(s2) = σ 2
n−1 n−1 2
but E{ n s2 } = n σ < σ2

In other words:
Bias(s2) = 0
but Bias( n−1 2
n s ) =
n−1 2
n σ − σ 2 = − n1 σ 2
19

The distribution of the sample SD

If X 1, X 2, . . . , X n are iid normal(µ, σ )

then the sample SD, s, satisfies (n – 1) s2/σ 2 ∼ χ2n−1

When the X i are not normally distributed, this is not true.

χ2 distributions

df=9

df=19

df=29

0 10 20 30 40 50

20
 Distribution of sample SD
(based on normal data)

n=25
n=10
n=5
n=3

0 5 10 15 20 25 30

21

A non-normal example

Distribution of sample SD
Population distribution
n=3

0 5 10 15 20 25 30

n=5
σ
µ

0 10 20 30 40 50 60
0 5 10 15 20 25 30

n = 10

0 5 10 15 20 25 30

n = 25

0 5 10 15 20 25 30

22