Simple random sampling is a basic topic in standard survey sampling textbooks (Cochran,, 1953, e.g.,). Below I review some results for simple random sampling that are useful for design-based inference in the CRE in Chapters 3 and 4. I define simple random sampling based on the distribution of the inclusion indicators.
A simple random sample of size
consists of a subset from a finite population of
units indexed by
.
Let
be the inclusion indicators of the
units with
if unit
is sampled and
otherwise.
The vector
can take
possible permutations of a vector of
1’s and
0’s, and each has equal probability.
By Definition 32.1, simple random sampling is a special form of sampling without replacement because it does not allow for repeatedly sampling the same unit. Lemma 32.1 below summarizes the first two moments of the inclusion indicators.
Under simple random sampling, we have
![]() |
In more compact forms, we have
![]() |
where
is a
-dimensional vector of
’s, and
is the
projection matrix orthogonal to 
Let
be a finite population with mean
and variance
![]() |
let
be another finite population with mean
and variance
![]() |
their covariance is
![]() |
Under simple random sampling, the sample means are
![]() |
sample variances are
![]() |
the sample covariance is
![]() |
Lemma 32.2 below gives the moments of the sample means
and
.
Under simple random sampling, the sample means are unbiased for the population means:
![]() |
Their variances and covariance are
![]() |
In the variance formulas in Lemma 32.2, the coefficient
in Lemma 32.2 is different from
under IID sampling. The additional term
is called the finite population correction factor.
Lemma 32.3 below gives the unbiasedness of the sample variances and covariance for estimating the population analogs.
Under simple random sampling, the sample variances and covariance are unbiased for their population versions:
![]() |
An important practical question is to make inference about
based on the simple random sample. This requires a more precise characterization of the distribution of its unbiased estimator
. The finite-sample exact distribution of
depends on the whole finite population
, which is unknown. The following finite population CLT characterizes the asymptotic distribution of
based on its first two moments.
Under Under simple random sampling, as
, if
![]() |
then
![]() |
in distribution, and
in probability.
Lemma 32.4 justifies the Wald-type
confidence interval for
:
![]() |
where
is the
upper quantile of the standard Normal random variable.
Proof of Lemma 32.1:
By symmetry, the
’s have the same mean, so
![]() |
Because
is a Bernoulli random variable, its variance is
![]() |
By symmetry again, the
’s have the same variance and the pairs
’s have the same covariance, so
![]() |
which implies that
![]() |

Proof of Lemma 32.2: The unbiasedness of the sample mean follows from linearity. For example,
![]() |
The covariance of the sample means is
![]() |
||||
![]() |
![]() |
|||
![]() |
![]() |
|||
![]() |
![]() |
Because
![]() |
the covariance of the sample means reduces to
![]() |
||||
![]() |
![]() |
|||
![]() |
![]() |
The variance formulas are just special cases with
.

Proof of Lemma 32.3: We prove only the sample covariance term because the formulas for sample variances are special cases. We have the following decomposition:
![]() |
![]() |
![]() |
||
![]() |
![]() |
|||
![]() |
![]() |
Taking expectations on both sides, we have
![]() |
![]() |
![]() |
||
![]() |
![]() |
|||
![]() |
![]() |
|||
![]() |
![]() |
and the conclusion follows by dividing both sides by
.

Proof of Lemma 32.4:
Hájek, (1960) gave a proof of the CLT for simple random sampling, and Lehmann, (1975) gave a more accessible version of the proof. Li and Ding, (2017) modified the CLT as presented in Lemma 32.4, and proved the consistency of the sample variance. Due to the technical complexities, I omit the proof.

Survey sampling and experimental design have been deeply connected ever since Neyman, (1934); Neyman, (1935)’s seminal work. Li and Ding, (2017) and Mukerjee et al., (2018) made many theoretical ties between these two areas.
Consider a special case of Lemma 32.2 with
’s being binary. Assume the total number of 1’s equals
so the total number of 0’s equals
. Let
denote the total number of 1’s in the sample of size
.
Find the distribution, mean, and variance of
.
Remark:
follows a Hypergeometric distribution.
Assume the
’s are vectors and define
![]() |
Show that
![]() |
tables.
Biometrika, 34:123–138.
factorial designs by using potential
outcomes.
Journal of the Royal Statistical Society: Series B (Statistical
Methodology), 77:727–753.
men.
Journal of the American Medicial Association, 166:1159–1172,
1294–1308.
-values from covariate balance tests.
arXiv preprint arXiv:2112.10545.
本书:A First Course in Causal Inference,作者 Peng Ding。原文:arXiv:2305.18793v2,按 CC BY 4.0 许可发布。本站(yuanens.com)把 arXiv 网页版改成分章的纯 HTML,把公式渲染成图片,方便在 Kindle 等电子书阅读器上看,内容没有改动。转换难免有错,以原文为准。本站与作者和出版社无关。