Geoscience Reference
In-Depth Information
K
∑
µ=
NN
/
ii
µ
,
(2.22)
i
=
1
and the corresponding sample estimate is
K
∑
y yN
=
/
,
(2.23)
s
i
i
i
1
=
with estimated variance
K
∑
··
.
Var(
y NyN
)
2
Var( /
2
=
(2.24)
s
i
i
i
=
1
·
, the square root of the estimated
variance, and an approximate 100(1 − α)% confidence interval for the popu-
lation mean is given by
The estimated standard error of
y
s
is
SE()
y
/2
·
yz y
±
SE()
,
(2.25)
s
s
α
where
z
α/2
is the value exceeded with probability α/2 for the standard nor-
mal distribution.
It sometimes happens that the population being dealt with is effectively
infinite. For example, if samples are taken at point locations in a field, then the
number of possible points can be regarded as infinite. In that case, Equations
(2.22) to (2.24) can be modified by replacing
N
i
/
N
by the proportion of the
population in stratum
i
. Hence, for sampling points in a field, this would be
the proportion of the area in the whole field that is in stratum
i
. Also, in this
situation the finite population corrections would not be needed because
n
i
/
N
i
would be zero for all strata.
If the population total is of interest, then this can be estimated by
·
TNy
s
(2.26)
s
with estimated standard error
···
SE() SE(
s
.
TN y
(2.27)
Again, an approximate 100(1
− α)% confidence interval takes the form
·
·
·
Tz T
±
SE()
.
(2.28)
s
α
/2