Sie sind auf Seite 1von 12

CRP I

Lesson VI (cont) : Probability Distributions


==================================================

Review of Probability Formulas
------------------------------------------------------------------------------------
P(A B) P(A B)
------------------------------------------------------------------------------------
General Theorem = P(A) + P(B) P(AB) = P(A) x P(B/A)

Special Case = P(A) + P(B) = P(A) x P(B)
if A and B are mutually if A and B are
exclusive; i.e., independent; i.e.
if P(AB) = 0 if P(B/A) = P(B)
-------------------------------------------------------------------------------------

Example :
A psychiatric researcher wished to evaluate a diagnostic instrument
for possible use as screening tool for major psychotic illness.
This test instrument contained 20 questions regarding individual
psychiatric symptoms. A positive test result is one in which a score
of 16 or greater is attained, that is, if a person reports having at least
16 of the 20 symptoms.
A sample of 388 psychotic patients and independent sample of 690
nonpsychotic patients were administered the test instrument with
The following results :
Diagnosis
Psychosis (D) Not Psychotis (ND)
-------------------------------------
Positive (T) 374 10 384
Test
Negative (NT) 14 680 694
------------------------------------- -----
Total 388 690 1078

Find the probability :
a. P (Test positive/Psychosis) c. P (Psychosis/test positive)
b. P ( (Test negative/Not Psychosis) d. P (Not psychosis/test negative)

Solution :
a. P (test positive/Psychosis) = test sensitivity = 374/388 = 0.964
b. P (test negative/not psychosis) = test specificity
= 680/690 = 0.986
c. P (Psychosis/test positive) = the predictive value positive
P(D/T) = P (DT) / P(T)
= 374/384 = 0.974;
is the probability that a person who has a positive test result has the
disease.
d. P (Not Psychosis/test negative) = the predictive value negative
P(ND/NT) = P (NDNT) / P(NT)
= 680/694 =0.980;
is the probability that a person who has a negative test result does
not have the disease.

Probability Distributions

Definition :
1. A random variable is a quantity which takes any of the values of a
specified set with a specified probability.
2. The probability distribution of a random variable is a table, graph
or mathematical expression giving the probabilities with which the
random variable takes different values.
3. A discrete probability distribution is one in which the random
variable, x, takes the discrete values x
1
, x
2
, x
3
, , x
k
with
respective probabilities p
1
, p
2
, p
3
, , p
k
such that the sum of all the
probabilities is unity, i.e..


p
i
= 1
A discrete probability distribution is analogous to the relative
frequency distribution with probabilities replacing relative frequen-
cies.
Common discrete probability distributions are the:(1) Binomial
and (2) Poisson.




Binomial Distribution
The Binomial distribution is relevant in cases where there are
two outcomes to a trial, success (with probability p) or
failure (with probability 1- p) and the trial is repeated n
times, each repetition being independent. The random
variable of interest, X, is the number of successes in the n
trials. The binomial distribution is the probability
distribution of X. For a particular value of X equal to r
(r = 0,1,2, , n), it determines the probability of observing
exactly r successes in the n independent trials.
The probability distribution is given by



The probability distribution is given by :

P (X = r) = (n / r ! (n-r) !) p
r
(1 p )
n r
for r = 0,1,2,, n.
Where n ! = (n factorial) is n ! = n (n-1) (n-2) (1);
0 ! = 1, by definition.
p may be known a priori or it may be estimated empirically
as p = r/n, the ratio of the number of successes observed in
the n trials to the number of trials.
Properties of the binomial distribution :
1. Characterised by n and p.
2. Mean = np
3. Variance = np(1-p)
4. Standard deviation = np (1-p)

Examples of Binomial variables

-------------------------------------------------------------------------------------
Trial Success Failure p n r
-------------------------------------------------------------------------------------
1. Birth of a child Girl Boy 0.5 number number of
in a family of children girls in the
family
2. Tossing a fair Head Tail 0.5 number number of
coin of tosses heads
altogether
3. Drawing a woman Pregnant not Preg. Sample Number
at random in a pregnant rate in size women in
fertility survey the popu- the sample
lation of
women
-------------------------------------------------------------------------------------
Example ;
In a family of 8 children, what is the probability of getting
exactly 3 girls ?

The probability of exactly 3 girls, is found by subtituting
n = 8, r = 3, and p = 0.5 into the binomial distribution :

P (X=3) = 8 ! / (3 ! 5!) 0.5
3
(1-0.5)
8-3

= 0.219
Problem :
Suppose the probability that a child selected at random is left-handed
is 0.20. For random sample of five children, what is the probability that:
a. two are left-handed ?
b. two or less are left-handed ?
Poisson Distribution
The Poisson distribution is the probability distribution of
the number of events, X, which occur independently and
randomly in :
1. Time, at fixed rate per unit time, such that the probability of
occurrence of more than one event in short interval of time
is very small as compared to the length of the interval.
2. Space, with a fixed density per unit area (volume), such that
the probability of occurrence of more than one event in a
small area (volume) is very small as compared to the size
of area (volume).

The probability of r events occurring is :
P(X = r) = e
-

r
/ r ! for r = 0, 1, 2, ,

where e = 2.718 is the base of natural (Naperian) logarithms
and may be known a priori or it may be estimated from
sample by X (arithmetic mean).

Properties of the Poisson distribution :
1. Characterised by .
2. Mean = .
3. Variance = .
4. Standard deviation = .
5. The Poisson distribution approximates the binomial distri-
bution when the number of trials, n, is large and the proba-
bility of success in a single trial, p, is small; with =np.
Example :
Suppose there are typically five cases of bladder cancer per
year among Jakarta men 35 40 years of age. Let X be the
number of cases observed during a year. For a given year,
compute the probability of zero, one, and two cases.

Answer :
P(X = r) = e
-

r
/ r ! With = 5(1) = 5.

a. P(X=0) = e
-5
5
0
/ 0 ! = 0.00674.
b. P(x=1) = e
-5
5
1
/ 1 ! = 0.03369.
c. P(x=2) = e
-5
5
2
/ 2 ! = 12.5e
5
= 0.08422.
Problems :
1. Suppose that annual incidence rate of leukemia in Indone-
sia is 11.2 cases per 100,000 population. If 100,000 people
are followed for one year, what is the probability of no new
cases of leukemia ?
If 1000 people are followed for one year what is the proba-
bility of no new cases ?

2. Suppose the number of cases of a disease in certain region
is a Poisson random variable with mean of eight per week
and let X be the number of cases in any week. Determine :
(a) P (X 2),
(b) P (X = 3),
(c) P (X > 5),
(d) P (4 X< 15).

Das könnte Ihnen auch gefallen