Exact Test
Exact Test
Exact Tests
Exact Tests
Favorable
Favorable
Unfavorable
Unfavorable
Total
Total
Test
Test
Control
Control
10 2
10 2
2
2
4
4
12
12
6
6
Total
Total
12 6
12 6
18
18
►
A test treatment
A test treatment
and a control are
and a control are
compared to determine whether the rates of
compared to determine whether the rates of
favorable response are the
favorable response are the
same.
same.
►
T
T
he sample sizes requirements for the chi-
he sample sizes requirements for the
chi-square tests are not met by these data
►
if you can consider the margins (12, 6, 12, 6)
if you can consider the margins (12, 6, 12, 6)
to be
to be
fixed, then you can assume that the
fixed, then you can assume that the
data are distributed hypergeometrically and
data are distributed hypergeometrically and
write
write
►
Pr
Pr
(
(
n
n
ijij)
)
=
=
n
n
1+1+!n
!n
2+2+!n
!n
+1+1!n
!n
+2+2!
!
/
/
n!n
n!n
1111!n
!n
1212!n
!n
2121!n
!n
2222!
!
►
p-value is the probability of the observed
p-value is the probability of the observed
data or more extreme data occurring
data or more extreme data occurring
under
under
the null hypothesis
the null hypothesis
►
With Fisher’s exact test, determine the
With Fisher’s exact test, determine the
p
p
-
-value for this table
value for this table
by summing the
by summing the
probabilities of the tables that are as likely or
probabilities of the tables that are as likely or
less likely, given the fixed
The following table includes all possible table configurations and their
The following table includes all possible table configurations and their
associated
associated probabilities.probabilities.
Table CellTable Cell
► (1,1) (1,1) (1,2) (1,2) (2,1) (2,1) (2,2)(2,2) Probabilities Probabilities
---► 12 12 0 0 0 0 6 6 0.00010.0001 ► 11 11 1 1 1 1 5 5 0.00390.0039
---► 1010 2 2 2 2 4 4 0.05330.0533
---► 9 9 3 3 3 3 3 3 0.23700.2370 ► 8 8 4 4 4 4 2 2 0.40000.4000 ► 7 7 5 5 5 5 1 1 0.25600.2560 ► 6 6 6 6 6 6 0 0 0.04980.0498
To find the one-sided
To find the one-sided pp-value, you sum the probabilities as small or smaller -value, you sum the probabilities as small or smaller than those
than those computed for the table observed, in the direction specified by computed for the table observed, in the direction specified by the one-sided alternative. In
the one-sided alternative. In this case, it would be those tables in which this case, it would be those tables in which the Test treatment had the more favorable
►
To find the two-sided
To find the two-sided
p
p
-value, you sum
-value, you sum
all of the probabilities that are as small
all of the probabilities that are as small
or smaller
or smaller
than that observed, or
than that observed, or
►
p = 0
p = 0
.
.
0533 + 0
0533 + 0
.
.
0039 + 0
0039 + 0
.
.
0001 +
0001 +