Introduction to Probability and

Probability Distributions
 Probability
 Probability – the chance that an uncertain event
will occur (always between 0 and 1)
 Probability distribution
 A mathematical function where the area
under the curve is 1.
 Gives the probabilities of all possible
outcomes.
 The probabilities must sum (or integrate) to 1.0.
 Probability distributions can be
discrete or continuous
 Discrete: has a countable number of outcomes
 Examples: Dead/alive, treatment/placebo, dice,
counts, etc.
 Continuous: has an infinite continuum of
possible values.
 Examples: blood pressure, weight, the speed of a
car, the real numbers from 1 to 6.
Discrete example: roll of a die

p(x)

1/6

x
1 2 3 4 5 6

 P(x)  1
all x
Probability mass function (pmf)
x p(x)
1 p(x=1)=1/6

2 p(x=2)=1/6

3 p(x=3)=1/6

4 p(x=4)=1/6

5 p(x=5)=1/6

6 p(x=6)=1/6
1.0
Cumulative distribution function
(CDF)

1.0 P(x)
5/6
2/3
1/2
1/3
1/6
1 2 3 4 5 6 x
Cumulative distribution
function
x P(x≤A)
1 P(x≤1)=1/6

2 P(x≤2)=2/6

3 P(x≤3)=3/6

4 P(x≤4)=4/6

5 P(x≤5)=5/6

6 P(x≤6)=6/6
 Practice Problem:
 The number of patients seen in the ER in any given hour is a
random variable represented by x. The probability distribution
for x is:

x 10 11 12 13 14
P(x) .4 .2 .2 .1 .1

Find the probability that in a given hour:

a.    exactly 14 patients arrive  p(x=14)= .1
b.    At least 12 patients arrive p(x12)= (.2 + .1 +.1) = .4
c.    At most 11 patients arrive p(x≤11)= (.4 +.2) = .6
 Review Question 1
If you toss a die, what’s the probability that you
roll a 3 or less?

a. 1/6
b. 1/3
c. 1/2
d. 5/6
e. 1.0
 Review Question 2
Two dice are rolled and the sum of the face
values is six? What is the probability that at
least one of the dice came up a 3?

a. 1/5
b. 2/3
c. 1/2
d. 5/6
e. 1.0
 Continuous case
 The probability function that accompanies a
continuous random variable is a continuous
mathematical function that integrates to 1.
 For example, recall the negative exponential
function (in probability, this is called an
“exponential distribution”):
f ( x)  e  x
 This function integrates to 1:
 

e
x x
 e  0 1 1
0
0
 Continuous case: “probability
density function” (pdf)

p(x)=e-x

The probability that x is any exact particular value (such as 1.9976) is 0;

we can only assign probabilities to possible ranges of x.
 For example, the probability of x falling within 1 to 2:

Clinical example: Survival times

after lung transplant may
roughly follow an exponential p(x)=e-x
function.
Then, the probability that a 1
patient will die in the second
year after surgery (between
years 1 and 2) is 23%.

x
1 2

2 2


x x
P(1  x  2)  e  e  e  2  e 1  .135  .368  .23
1
1
 Example 2: Uniform
distribution
The uniform distribution: all values are equally likely.
f(x)= 1 , for 1 x 0
p(x)

x
1

We can see it’s a probability distribution because it integrates

to 1 (the area under the curve is 1): 1 1
 
1  x
0
0
1 0 1
 Example: Uniform distribution
 What’s the probability that x is between 0 and ½?

Clinical Research Example:

p(x) When randomizing patients in
an RCT, we often use a random
1 number generator on the
computer. These programs work
by randomly generating a
number between 0 and 1 (with
equal probability of every
0 ½ x
1 number in between). Then a
subject who gets X<.5 is control
and a subject who gets X>.5 is
treatment.

P(½ x 0)= ½

 Expected Value and Variance

 All probability distributions are

characterized by an expected value
(mean) and a variance (standard
deviation squared).
For example, bell-curve (normal) distribution:

One standard
deviation from the
Mean ()
mean ()
 Expected value of a random variable

 Expected value is just the average or mean (µ) of

random variable x.

 It’s sometimes called a “weighted average”

because more frequent values of X are weighted
more highly in the average.

 It’s also how we expect X to behave on-average

over the long run (“frequentist” view again).
 Expected value, formally
Discrete case:

E( X )   x p(x )
all x
i i

Continuous case:

E( X )  
all x
xi p(xi )dx
Symbol Interlude
 E(X) = µ
 these symbols are used interchangeably
 Example: expected value

 Recall the following probability distribution of ER

arrivals:

x 10 11 12 13 14
P(x) .4 .2 .2 .1 .1

 x p( x)  10(.4)  11(.2)  12(.2)  13(.1)  14(.1)  11.3

i 1
i
 Sample Mean is a special case of
Expected Value…

Sample mean, for a sample of n subjects: =

n

x i n
1
X i 1
n
 
i 1
xi ( )
n

The probability (frequency) of each

person in the sample is 1/n.
Expected Value
 Expected value is an extremely useful
concept for good decision-making!
Example: the lottery
 The Lottery (also known as a tax on people
who are bad at math…)
 A certain lottery works by picking 6 numbers
from 1 to 49. It costs $1.00 to play the
lottery, and if you win, you win $2 million
after taxes.

 If you play the lottery once, what are your

expected winnings or losses?
 Lottery
Calculate the probability of winning in 1 try:

1 1 1 “49 choose 6”
   7.2 x 10-8
 49  49! 13,983,816
  Out of 49
6 43!6!
numbers, this is
the number of
distinct
The probability function (note, sums to 1.0): combinations of 6.

x$ p(x)
-1 .999999928

+ 2 million 7.2 x 10--8

 Expected Value
The probability function
x$ p(x)
-1 .999999928

+ 2 million 7.2 x 10--8

Expected Value
E(X) = P(win)*$2,000,000 + P(lose)*-$1.00
= 2.0 x 106 * 7.2 x 10-8+ .999999928 (-1) = .144 - .999999928 = -$.86
 

Negative expected value is never good!

You shouldn’t play if you expect to lose money!
 Expected Value
If you play the lottery every week for 10 years, what are your
expected winnings or losses?
 
520 x (-.86) = -$447.20
Recent headlines: Record
Mega Millions…
 Recently Mega Millions had a jackpot of $656
million ($474 immediate payout).

 Question I received: “If the odds of winning

the Mega millions is 1 in 175,000,000 is there
a significant statistical advantage in playing
100 quick picks rather than one?
 “For a half-billion-with-a-B dollars it almost
seems worth it.”
 Answer, 1 ticket:
 Chances of losing, 1 ticket:
1-1/175,000,000=99.9999994%

x$ p(x)
-1 . 999999994

+ 500 million 6 × 10-9

Expected Value
E(X) = P(win)*$500,000,000 + P(lose)*-$1.00
= 6.0 x 10-9 * 500,000,000+ .999999994 (-1) = +2
 
 Answer, 100 tickets:
 Chances of losing, 100 tickets:
99.999943%

x$ p(x)
-1 . 99999943

+ 500 million 5.7× 10-7

Expected Value
E(X) = P(win)*$500,000,000 + P(lose)*-$1.00
= 5.7 x 10-7 * 500,000,000+ .9999994 (-1) = +285
 
So…
 One could make a case for playing, but
doesn’t account for multiple winners,
taxes, lump-sum payouts, etc…
 After all that is taken into account, payout
would still have to be >$167 million for
expected value to be positive.
 And, the fact is, you’re still going to lose
with almost near certainty!
 Probability 99.9999…%!
 Gambling (or how casinos can afford to give so
many free drinks…)

A roulette wheel has the numbers 1 through 36, as well as 0 and 00. If you
bet $1 that an odd number comes up, you win or lose $1 according to
whether or not that event occurs. If random variable X denotes your net
gain, X=1 with probability 18/38 and X= -1 with probability 20/38.
 
E(X) = 1(18/38) – 1 (20/38) = -$.053
 
On average, the casino wins (and the player loses) 5 cents per game.
 
The casino rakes in even more if the stakes are higher:
 
E(X) = 10(18/38) – 10 (20/38) = -$.53
 
If the cost is $10 per game, the casino wins an average of 53 cents per game.
If 10,000 games are played in a night, that’s a cool $5300.
Expected value isn’t
everything though…
 Take the hit new show “Deal or No Deal”
 Everyone know the rules?
 Let’s say you are down to two cases left. $1
and $400,000. The banker offers you
$200,000.
 So, Deal or No Deal?
Deal or No Deal…
 This could really be represented as a
probability distribution and a non-
random variable:
x$ p(x)
+1 .50

+$400,000 .50

x$ p(x)
+$200,000 1.0
Expected value doesn’t help…

x$ p(x)
+1 .50

+$400,000 .50

  E( X )   x p(x )  1(.50)  400,000(.50)  200,000

all x
i i

x$ p(x)
+$200,000 1.0

  E ( X )  200,000
 How to decide?
Variance!
• If you take the deal, the variance/standard
deviation is 0.
•If you don’t take the deal, what is average
deviation from the mean?
•What’s your gut guess?
 Variance/standard deviation
2=Var(x) =E(x-)2

“The expected (or average) squared

distance (or deviation) from the mean”

 2  Var ( x)  E[( x   ) 2 ]  
all x
( xi   ) 2 p(xi )
 Variance, continuous
Discrete case:

Var ( X )   (x
all x
i
2
  ) p(xi )

Continuous case?:


2
Var ( X )  ( xi   ) p(xi )dx
all x
Symbol Interlude
 Var(X)= 2
 SD(X) = 
 these symbols are used interchangeably
 Similarity to empirical variance

The variance of a sample: s2 =

 ( xi  x ) 2 N
1
i 1
n 1
 
i 1
( xi  x ) (2
n 1
)

Division by n-1 reflects the fact that we have lost a

“degree of freedom” (piece of information) because
we had to estimate the sample mean before we could
estimate the sample variance.
 Variance

 2
 (x
all x
i
2
  ) p(xi )

2  
all x
( xi   ) 2 p(xi ) 

 (1  200,000) 2 (.5)  (400,000  200,000) 2 (.5)  200,000 2

  200,000 2  200,000

Now you examine your personal risk tolerance…

Practice Problem
On the roulette wheel, X=1 with
probability 18/38 and X= -1 with
probability 20/38.
 We already calculated the mean to be = -
$.053. What’s the variance of X?
 Answer
 2
 (x   )
all x
i
2
p(xi )
 (1  .053) 2 (18 / 38)  (1  .053) 2 (20 / 38)
 (1.053) 2 (18 / 38)  (1  .053) 2 (20 / 38)
 (1.053) 2 (18 / 38)  (.947) 2 (20 / 38)
 .997

  .997  .99
Standard deviation is $.99. Interpretation: On average, you’re
either 1 dollar above or 1 dollar below the mean, which is just
under zero. Makes sense!
Review Question 3
The expected value and variance of a coin
toss (H=1, T=0) are?

a. .50, .50
b. .50, .25
c. .25, .50
d. .25, .25
 Variance example: TPMT
 TPMT metabolizes the drugs 6-
mercaptopurine, azathioprine, and 6-thioguanine
(chemotherapy drugs)
 People with TPMT-/ TPMT+ have reduced levels of
activity (10% prevalence)
 People with TPMT-/ TPMT- have no TPMT activity
(prevalence 0.3%).
 They cannot metabolize 6-
mercaptopurine, azathioprine, and 6-thioguanine,
and risk bone marrow toxicity if given these drugs.
 TPMT activity by genotype

Weinshilboum R. Drug Metab Dispos. 2001 Apr;29(4 Pt 2):601-5

 TPMT activity by genotype
The variability in TPMT
activity is much higher
in wild-types than
heterozygotes.

Weinshilboum R. Drug Metab Dispos. 2001 Apr;29(4 Pt 2):601-5

 TPMT activity by genotype
There is variability in
expression from each
wild-type allele. With
No variability in two copies of the
expression here, good gene present,
since there’s no there’s “twice as
working gene. much” variability.

Weinshilboum R. Drug Metab Dispos. 2001 Apr;29(4 Pt 2):601-5

Important discrete probability
distribution: The binomial
 Binomial Probability
Distribution
 A fixed number of observations (trials), n
 e.g., 15 tosses of a coin; 20 patients; 1000 people
surveyed
 A binary outcome
 e.g., head or tail in each toss of a coin; disease or no
disease
 Generally called “success” and “failure”
 Probability of success is p, probability of failure is 1 – p
 Constant probability for each observation
 e.g., Probability of getting a tail is the same each time we
toss the coin
Binomial distribution
Take the example of 5 coin tosses.
What’s the probability that you flip
exactly 3 heads in 5 coin tosses?
 Binomial distribution
Solution:
One way to get exactly 3 heads: HHHTT

What’s the probability of this exact arrangement?

P(heads)xP(heads) xP(heads)xP(tails)xP(tails)
=(1/2)3 x (1/2)2

Another way to get exactly 3 heads: THHHT

Probability of this exact outcome = (1/2)1 x (1/2)3
x (1/2)1 = (1/2)3 x (1/2)2
 Binomial distribution
In fact, (1/2)3 x (1/2)2 is the probability of each
unique outcome that has exactly 3 heads and 2
tails.

So, the overall probability of 3 heads and 2 tails

is:
(1/2)3 x (1/2)2 + (1/2)3 x (1/2)2 + (1/2)3 x (1/2)2
+ ….. for as many unique arrangements as
there are—but how many are there??
 Outcome Probability
  THHHT (1/2)3 x (1/2)2
HHHTT (1/2)3 x (1/2)2
TTHHH (1/2)3 x (1/2)2
HTTHH (1/2)3 x (1/2)2 The probability
ways to HHTTH (1/2)3 x (1/2)2
5 arrange 3 HTHHT (1/2)3 x (1/2)2
of each unique
  heads in THTHH (1/2)3 x (1/2)2
outcome (note:
 3 5 trials HTHTH (1/2)3 x (1/2)2
they are all
equal)
HHTHT (1/2)3 x (1/2)2
THHTH (1/2)3 x (1/2)2
10 arrangements x (1/2)3 x (1/2)2
C3 = 5!/3!2! = 10  
5

Factorial review: n! = n(n-1)(n-2)…

 5
P(3 heads
 
and 2 tails) =   x P(heads)3 x P(tails)2 =
 3

10 x (½)5=31.25%
Binomial distribution
function:
X= the number of heads tossed in 5 coin
tosses
p(x)

x
0 1 2 3 4 5
number of heads
 Binomial distribution,
generally
Note the general pattern emerging  if you have only two possible
outcomes (call them 1/0 or yes/no or success/failure) in n independent
trials, then the probability of exactly X “successes”=
n = number of trials

n X n X
  p (1  p )
X 1-p = probability
of failure
X=# p=
successes probability of
out of n success
trials
 Binomial distribution: example

 If I toss a coin 20 times, what’s the

probability of getting exactly 10 heads?

 20  10 10
 (.5) (.5)  .176
 10 
 Binomial distribution: example
 If I toss a coin 20 times, what’s the
probability of getting of getting 2 or
fewer heads?
 20  0 20 20!
 (.5) (.5)  (.5) 20  9.5 x10 7 
0 20!0!
 20  1 19 20!
  (.5) (.5)  (.5) 20  20 x9.5 x10  7  1.9 x10 5 
1 19!1!
 20  2 18 20!
 (.5) (.5)  (.5) 20  190 x9.5 x10 7  1.8 x10  4
2 18!2!
 1.8 x10  4
**All probability distributions are
characterized by an expected value and a
variance:

If X follows a binomial distribution with

parameters n and p: X ~ Bin (n, p)
Then:
Note: the variance will
E(X) = np always lie between
0*N-.25 *N
Var (X) = np(1-p) p(1-p) reaches

SD (X)= np(1  p) maximum at p=.5

P(1-p)=.25
Practice Problem
 1. You are performing a cohort study. If the
probability of developing disease in the exposed
group is .05 for the study duration, then if you
(randomly) sample 500 exposed people, how many
do you expect to develop the disease? Give a margin
of error (+/- 1 standard deviation) for your estimate.

 2. What’s the probability that at most 10 exposed

people develop the disease?
Answer
1. How many do you expect to develop the disease? Give a margin of
error (+/- 1 standard deviation) for your estimate.

X ~ binomial (500, .05)

E(X) = 500 (.05) = 25
Var(X) = 500 (.05) (.95) = 23.75
StdDev(X) = square root (23.75) = 4.87 
25  4.87
 Answer
2. What’s the probability that at most 10 exposed
subjects develop the disease?

This is asking for a CUMULATIVE PROBABILITY: the probability of 0 getting the

disease or 1 or 2 or 3 or 4 or up to 10.
 
P(X≤10) = P(X=0) + P(X=1) + P(X=2) + P(X=3) + P(X=4)+….+ P(X=10)=

 500  0 500  500  1 499  500  2 498  500  10 490

 (.05) (.95)   (.05) (.95)   (.05) (.95)  ...   (.05) (.95)  .01
 0  1  2  10 
Practice Problem:
You are conducting a case-control study of
smoking and lung cancer. If the probability of
being a smoker among lung cancer cases is .6,
what’s the probability that in a group of 8 cases
you have:

a. Less than 2 smokers?

b. More than 5?
c. What are the expected value and variance of the number
of smokers?
Answer
X P(X)
8
0 1(.4) =.00065
1 7
1 8(.6) (.4) =.008
2 28(.6)2 (.4) 6 =.04
3 5
3 56(.6) (.4) =.12
4 70(.6)4 (.4) 4 =.23
5 3
5 56(.6) (.4) =.28
6 2
6 28(.6) (.4) =.21
7 1
7 8(.6) (.4) =.090
8 1(.6)8 =.0168

0 1 2 3 4 5 6 7 8
 Answer, continued

P(<2)=.00065 + .008 = .00865 P(>5)=.21+.09+.0168 = .3168

0 1 2 3 4 5 6 7 8

E(X) = 8 (.6) = 4.8

Var(X) = 8 (.6) (.4) =1.92
StdDev(X) = 1.38
Practice Problem
You toss a coin 100 times. What’s the expected number of
heads? What’s the variance of the number of heads?
 Answer
E(X) = 100 (.5) = 50
Var(X) = 100 (.5) (. 5) = 25
StdDev(X) = square root (25) = 5
 
Interpretation: When we toss a coin
100 times, we expect to get 50 heads
plus or minus 5.
Or use computer simulation…
 Flip coins virtually!
 Flip a virtual coin 100 times; count the
number of heads.
 Repeat this over and over again a large
number of times (we’ll try 30,000 repeats!)
 Plot the 30,000 results.
Coin tosses…

Mean = 50
Std. dev = 5
Follows a normal
distribution
95% of the time, we
get between 40 and
60 heads…
 Review Question 4
In your case-control study of smoking and lung-
cancer, 60% of cases are smokers versus only 10%
of controls. What is the odds ratio between smoking
and lung cancer?

a. 2.5
b. 13.5
c. 15.0
d. 6.0
e. .05
 Review Question 5
What’s the probability of getting exactly 5
heads in 10 coin tosses?

 10  5 5
 (.50) (.50)
a. 0

b.  (.50) 5 (.50) 5
10

5
c.  10  10
 (.50) (.50)
5

5
d.  10  10 0
 (.50) (.50)
 10 
 Review Question 6
A coin toss can be thought of as an example of a binomial
distribution with N=1 and p=.5. What are the expected
value and variance of a coin toss?

a. .5, .25
b. 1.0, 1.0
c. 1.5, .5
d. .25, .5
e. .5, .5
 Review Question 7
If I toss a coin 10 times, what is the expected value
and variance of the number of heads?

a. 5, 5
b. 10, 5
c. 2.5, 5
d. 5, 2.5
e. 2.5, 10
 Review Question 8
In a randomized trial with n=150, the goal is to randomize
half to treatment and half to control. The number of people
randomized to treatment is a random variable X. What is
the probability distribution of X?

a. X~Normal(=75,=10)
b. X~Exponential(=75)
c. X~Uniform
d. X~Binomial(N=150, p=.5)
e. X~Binomial(N=75, p=.5)
 Review Question 9
In the same RCT with n=150, if 69
end up in the treatment group and 81
in the control group, how far off is
that from expected?

a. Less than 1 standard deviation

b. 1 standard deviation
c. Between 1 and 2 standard deviations
d. More than 2 standard deviations
Proportions…
 The binomial distribution forms the basis of
statistics for proportions.
 A proportion is just a binomial count divided
by n.
 For example, if we sample 200 cases and find 60
smokers, X=60 but the observed proportion=.30.
 Statistics for proportions are similar to
binomial counts, but differ by a factor of n.
 Stats for proportions
For binomial:  x  np
Differs by
2 a factor of
 x  np(1  p ) n.

 x  np (1  p )
Differs
by a
factor
 pˆ  p of n.
For proportion:
np(1  p) p (1  p)
 pˆ 2  2

n n
P-hat stands for “sample p (1  p )
proportion.”  pˆ 
n
It all comes back to normal…
 Statistics for proportions are based on a
normal distribution, because the
binomial can be approximated as
normal if np>5 (more on this next
week…)