3
we have to do when N is unknown, the effect on the variance of the estimator is slight when N is
large. When N is small, however, the variance of the estimator can be overestimated
appreciably.
Example
. (to be added)
Estimating the Population Total
Like the number of entities per sample unit, the total number of entities in the entire population is
another attribute estimated commonly. Unlike the population mean, however, estimating the
population total requires that we know the number of sampling units in a population, N.
The population total
∑
=
==
N
i
i
Ny
1
μτ
is estimated with the sample total (
ˆ
) which has an unbiased
estimator:
∑
=
==
n
i
i
y
n
N
N
1
μτ
ˆˆ
where N is the total number of sample units in a population, n is the number of units in the
sample, and y
i
is the value measured from each sample unit.
In studies of wildlife populations, the total number of entities in a population is often referred to as
“abundance” and is traditionally represented with the symbol N. Consequently, there is real
potential for confusing the number of entities in the population with the number of sampling units
in the sampling frame. Therefore, in the context of sampling theory, we’ll use
ˆ
to represent the
population total and N to represent the number of sampling units in a population. Later, when
addressing wildlife populations specifically, we’ll use N to represent abundance to remain
consistent with the literature in that field.
Because the estimator
ˆ
is simply the number of sample units in the population N times the
mean number of entities per sample unit,
ˆ
, the variance of the estimate
ˆ
reflects both the
number of units in the sampling universe N and the variance associated with
ˆ
. An unbiased
estimate for the variance of the estimate
ˆ
is:
⎟
⎠
⎞
⎜
⎝
⎛
−
⎟
⎟
⎠
⎞
⎜
⎜
⎝
⎛
==
N
nN
n
s
NN
2
22
)
ˆ
var()
ˆ
var(
μτ
where s
2
is the estimated population variance.
Example
. (to be added)
Estimating a Population Proportion
If you are interested in the composition of a population, you could use a simple random sample to
estimate the proportion of the population p that is composed of elements with a particular trait,
such as the proportion of plants that flower in a given year, the proportion of juvenile animals
captured, the proportion of females in estrus, and so on. We will consider only classifications that
are dichotomous, meaning that an element in the population either has the trait of interest
(flowering) or it does not (not flowering); extending this idea to more complex classifications is
straightforward.