<s>
In	O
statistics	O
,	O
Cochran	B-General_Concept
's	I-General_Concept
C	I-General_Concept
test	I-General_Concept
,	O
named	O
after	O
William	O
G	O
.	O
Cochran	O
,	O
is	O
a	O
one-sided	B-General_Concept
upper	O
limit	O
variance	O
outlier	O
test	O
.	O
</s>
<s>
The	O
C	O
test	O
is	O
used	O
to	O
decide	O
if	O
a	O
single	O
estimate	O
of	O
a	O
variance	O
(	O
or	O
a	O
standard	B-General_Concept
deviation	I-General_Concept
)	O
is	O
significantly	B-General_Concept
larger	O
than	O
a	O
group	O
of	O
variances	O
(	O
or	O
standard	B-General_Concept
deviations	I-General_Concept
)	O
with	O
which	O
the	O
single	O
estimate	O
is	O
supposed	O
to	O
be	O
comparable	O
.	O
</s>
<s>
Cochran	B-General_Concept
's	I-General_Concept
C	I-General_Concept
test	I-General_Concept
should	O
not	O
be	O
confused	O
with	O
Cochran	B-General_Concept
's	I-General_Concept
Q	I-General_Concept
test	I-General_Concept
,	O
which	O
applies	O
to	O
the	O
analysis	O
of	O
two-way	O
randomized	O
block	O
designs	O
.	O
</s>
<s>
the	O
considered	O
full	O
data	B-General_Concept
set	I-General_Concept
should	O
consist	O
of	O
individual	O
data	B-General_Concept
series	I-General_Concept
that	O
all	O
have	O
equal	O
size	O
.	O
</s>
<s>
The	O
C	O
test	O
further	O
assumes	O
that	O
each	O
individual	O
data	B-General_Concept
series	I-General_Concept
is	O
normally	O
distributed	O
.	O
</s>
<s>
Although	O
primarily	O
an	O
outlier	O
test	O
,	O
the	O
C	O
test	O
is	O
also	O
in	O
use	O
as	O
a	O
simple	O
alternative	O
for	O
regular	O
homoscedasticity	B-General_Concept
tests	O
such	O
as	O
Bartlett	B-General_Concept
's	I-General_Concept
test	I-General_Concept
,	O
Levene	B-General_Concept
's	I-General_Concept
test	I-General_Concept
and	O
the	O
Brown	B-General_Concept
–	I-General_Concept
Forsythe	I-General_Concept
test	I-General_Concept
to	O
check	O
a	O
statistical	B-General_Concept
data	I-General_Concept
set	O
for	O
homogeneity	B-General_Concept
of	I-General_Concept
variances	I-General_Concept
.	O
</s>
<s>
An	O
even	O
simpler	O
way	O
to	O
check	O
homoscedasticity	B-General_Concept
is	O
provided	O
by	O
Hartley	O
's	O
Fmax	O
test	O
,	O
but	O
Hartley	O
's	O
Fmax	O
test	O
has	O
the	O
disadvantage	O
that	O
it	O
only	O
accounts	O
for	O
the	O
minimum	O
and	O
the	O
maximum	O
of	O
the	O
variance	O
range	O
,	O
while	O
the	O
C	O
test	O
accounts	O
for	O
all	O
variances	O
within	O
the	O
range	O
.	O
</s>
<s>
The	O
corresponding	O
data	B-General_Concept
series	I-General_Concept
is	O
then	O
omitted	O
from	O
the	O
full	O
data	B-General_Concept
set	I-General_Concept
.	O
</s>
<s>
According	O
to	O
ISO	O
standard	O
5725	O
the	O
C	O
test	O
may	O
be	O
iterated	B-Algorithm
until	O
no	O
further	O
exceptionally	O
large	O
variance	O
values	O
are	O
detected	O
,	O
but	O
such	O
practice	O
may	O
lead	O
to	O
excessive	O
rejections	O
if	O
the	O
underlying	O
data	B-General_Concept
series	I-General_Concept
are	O
not	O
normally	O
distributed	O
.	O
</s>
<s>
The	O
C	O
test	O
tests	O
the	O
null	B-General_Concept
hypothesis	I-General_Concept
(	O
H0	O
)	O
against	O
the	O
alternative	B-General_Concept
hypothesis	I-General_Concept
(	O
Ha	O
)	O
:	O
</s>
<s>
Ha	O
:	O
At	O
least	O
one	O
variance	O
value	O
is	O
significantly	B-General_Concept
larger	O
than	O
the	O
other	O
variance	O
values	O
.	O
</s>
<s>
The	O
sample	O
variance	O
of	O
data	B-General_Concept
series	I-General_Concept
j	O
is	O
considered	O
an	O
outlier	O
at	O
significance	B-General_Concept
level	I-General_Concept
α	O
if	O
Cj	O
exceeds	O
the	O
upper	O
limit	O
critical	O
value	O
CUL	O
.	O
</s>
<s>
CUL	O
depends	O
on	O
the	O
desired	O
significance	B-General_Concept
level	I-General_Concept
α	O
,	O
the	O
number	O
of	O
considered	O
data	B-General_Concept
series	I-General_Concept
N	O
,	O
and	O
the	O
number	O
of	O
data	O
points	O
(	O
n	O
)	O
per	O
data	B-General_Concept
series	I-General_Concept
.	O
</s>
<s>
Selections	O
of	O
values	O
for	O
CUL	O
have	O
been	O
tabulated	O
at	O
significance	B-General_Concept
levels	I-General_Concept
α	O
=	O
0.01	O
,	O
α	O
=	O
0.025	O
,	O
and	O
α	O
=	O
0.05	O
.	O
</s>
<s>
Fc	O
=	O
critical	O
value	O
of	O
Fisher	B-General_Concept
's	I-General_Concept
F	I-General_Concept
ratio	O
;	O
Fc	O
can	O
be	O
obtained	O
from	O
tables	O
of	O
the	O
F	B-General_Concept
distribution	I-General_Concept
or	O
using	O
computer	O
software	O
for	O
this	O
function	O
.	O
</s>
<s>
The	O
C	O
test	O
can	O
be	O
generalized	O
to	O
include	O
unbalanced	O
designs	O
,	O
one-sided	B-General_Concept
lower	O
limit	O
tests	O
and	O
two-sided	B-General_Concept
tests	I-General_Concept
at	O
any	O
significance	B-General_Concept
level	I-General_Concept
α	O
,	O
for	O
any	O
number	O
of	O
data	B-General_Concept
series	I-General_Concept
N	O
,	O
and	O
for	O
any	O
number	O
of	O
individual	O
data	O
points	O
nj	O
in	O
data	B-General_Concept
series	I-General_Concept
j	O
.	O
</s>
