统计代写|抽样调查作业代写sampling theory of survey代考|Design-Based Variance Estimation

如果你也在 怎样代写抽样调查Survey sampling 这个学科遇到相关的难题,请随时右上角联系我们的24/7代写客服。抽样调查Survey sampling是数学工程这一广泛新兴领域中的一个自然组成部分。例如,我们可以断言,数学工程之于今天的数学系,就像数学物理之于一个世纪以前的数学系一样;毫不夸张地说,数学在诸如语音和图像处理、信息理论和生物医学工程等工程学科中的基本影响。

抽样调查Survey sampling是主流统计的边缘。这里的特殊之处在于,我们有一个具有某些特征的有形物体集合,我们打算通过抓住其中一些物体并试图对那些未被触及的物体进行推断来窥探它们。这种推论传统上是基于一种概率论,这种概率论被用来探索观察到的事物与未观察到的事物之间的可能联系。这种概率不被认为是在统计学中,涵盖其他领域,以表征我们感兴趣的变量的单个值之间的相互关系。但这是由调查抽样调查人员通过任意指定的一种技术从具有预先分配概率的对象群体中选择样本而创建的。

statistics-lab™ 为您的留学生涯保驾护航 在代写抽样调查sampling theory of survey方面已经树立了自己的口碑, 保证靠谱, 高质且原创的统计Statistics代写服务。我们的专家在代写抽样调查sampling theory of survey方面经验极为丰富,各种代写抽样调查sampling theory of survey相关的作业也就用不着说。

统计代写|抽样调查作业代写sampling theory of survey代考|Design-Based Variance Estimation

统计代写|抽样调查作业代写sampling theory of survey代考|Design-Based Variance Estimation

When $\left(X_i, Y_i\right)$ values are available for SRSWOR of size $n$ an alternative to the ratio estimator for $\bar{Y}$ is the regression estimator
$$
t_r=\bar{y}+b(\bar{X}-\bar{x}) .
$$
Here $b$ is the sample regression coefficient of $y$ on $x$. Its variance $V_p\left(t_r\right)$ and mean square error $M_p\left(t_r\right)$ are both approximated by
$$
V=\frac{1-f}{n} \frac{1}{N-1} \sum_1^N D_i^2
$$
where
$$
\begin{aligned}
D_i & =\left(Y_i-\bar{Y}\right)-B\left(X_i-\bar{X}\right) \
B & =\sum_1^N\left(Y_i-\bar{Y}\right)\left(X_i-\bar{X}\right) / \sum_1^N\left(X_i-\bar{X}\right)^2 .
\end{aligned}
$$
The errors in these approximations are neglected for large $n$ and $N$ although for $n, N$, and $X$ at hand it is difficult to guess the magnitudes of these errors. However, there exists evidence that $t_r$ may be more efficient than the ratio estimator $\bar{t}R$ in many situations in terms of mean square error (cf. DENG and Wu, 1987). Writing $$ \begin{aligned} d_i & =\left(Y_i-\bar{y}\right)-b\left(X_i-\bar{x}\right), \ v{l r} & =\frac{1-f}{n(n-2)} \sum_s d_i^2
\end{aligned}
$$
is traditionally taken as an estimator for $V$. DENG and WU (1987) consider a class of generalized estimators
$$
v_g=\left[\frac{\bar{X}}{\bar{x}}\right]^g v_{l r}
$$

统计代写|抽样调查作业代写sampling theory of survey代考|Model-Based Variance Estimation

Besides these ad hoc variance estimators, hardly any others are known to have been proposed as estimators for $V$ with a design-based approach. However, some rivals have emerged from the least squares linear predictive approach.

Suppose $Y, X$ are conformable to the model $\mathcal{M}_{10}^{\prime}$ (cf. section 4.1.2) for which the following is tenable:
$$
\begin{aligned}
E_m\left(Y_i\right) & =\alpha+\beta X_i, \alpha \neq 0, V_m\left(Y_i\right)=\sigma^2, \
C_m\left(Y_i, Y_j\right) & =0, i \neq j .
\end{aligned}
$$
Then the BLUP for $\bar{Y}$ is $\mathrm{t}_r$ and
$$
\begin{aligned}
B_m\left(t_r\right) & =E_m\left(t_r-\bar{Y}\right)=0 \
V_m\left(t_r-\bar{Y}\right) & =\frac{1-f}{n}\left[1+\frac{(\bar{X}-\bar{x})^2}{(1-f) g(s)}\right] \sigma^2=\phi(s) \sigma^2, \text { say, }
\end{aligned}
$$
writing
$$
g(s)=\frac{1}{n} \sum_s\left(X_i-\bar{x}\right)^2 .
$$
Then, for
$$
\hat{\sigma}^2=\frac{1}{(n-2)} \sum_s d_i^2
$$
we have
$$
E_m\left(\hat{\sigma}^2\right)=\sigma^2
$$
Consequently,
$$
v_L=\phi(s) \hat{\sigma}^2=\frac{1-f}{n(n-2)}\left[1+\frac{(\bar{X}-\bar{x})^2}{(1-f) g(s)}\right] \sum_s d_i^2
$$
is an $m$-unbiased estimator for $V_m\left(t_r-\bar{Y}\right)$ under $\mathcal{M}{10}^{\prime}$. The term $$ h(s)=\frac{(X-\bar{x})^2}{(1-f) g(s)} $$ in $v_L$ vanishes if the sample is balanced, that is, $\bar{x}=\bar{X}$, and for a balanced sample $V_m\left(t_r-\bar{Y}\right)$ is the minimal under $\mathcal{M}{10}^{\prime}$.
In general,
$$
v_L=(1+h(s)) v_{l r} \geq v_{l r}
$$
with equality only for a balanced sample. If a balanced sample is drawn, then the classical design-based estimator $v_{l r}$ based on it becomes $m$-unbiased for $V_m\left(t_r-\bar{Y}\right)$.

统计代写|抽样调查作业代写sampling theory of survey代考|Design-Based Variance Estimation

抽样调查代考

统计代写|抽样调查作业代写sampling theory of survey代考|Design-Based Variance Estimation

当大小为$n$的SRSWOR的$\left(X_i, Y_i\right)$值可用时,$\bar{Y}$的比率估计器的替代方法是回归估计器
$$
t_r=\bar{y}+b(\bar{X}-\bar{x}) .
$$
其中$b$为$y$在$x$上的样本回归系数。其方差$V_p\left(t_r\right)$和均方误差$M_p\left(t_r\right)$都近似为
$$
V=\frac{1-f}{n} \frac{1}{N-1} \sum_1^N D_i^2
$$
在哪里
$$
\begin{aligned}
D_i & =\left(Y_i-\bar{Y}\right)-B\left(X_i-\bar{X}\right) \
B & =\sum_1^N\left(Y_i-\bar{Y}\right)\left(X_i-\bar{X}\right) / \sum_1^N\left(X_i-\bar{X}\right)^2 .
\end{aligned}
$$
对于较大的$n$和$N$,这些近似中的误差可以忽略不计,尽管对于$n, N$和$X$,很难猜测这些误差的大小。然而,有证据表明,在均方误差方面,$t_r$在许多情况下可能比比率估计器$\bar{t}R$更有效(cf. DENG and Wu, 1987)。写作$$ \begin{aligned} d_i & =\left(Y_i-\bar{y}\right)-b\left(X_i-\bar{x}\right), \ v{l r} & =\frac{1-f}{n(n-2)} \sum_s d_i^2
\end{aligned}
$$
通常作为$V$的估计量。DENG和WU(1987)考虑了一类广义估计量
$$
v_g=\left[\frac{\bar{X}}{\bar{x}}\right]^g v_{l r}
$$

统计代写|抽样调查作业代写sampling theory of survey代考|Model-Based Variance Estimation

除了这些特别的方差估计器之外,几乎没有其他已知的方法被提出作为$V$基于设计的方法的估计器。然而,从最小二乘线性预测方法中出现了一些竞争对手。

假设$Y, X$符合模型$\mathcal{M}{10}^{\prime}$(参见第4.1.2节),其中以下是成立的: $$ \begin{aligned} E_m\left(Y_i\right) & =\alpha+\beta X_i, \alpha \neq 0, V_m\left(Y_i\right)=\sigma^2, \ C_m\left(Y_i, Y_j\right) & =0, i \neq j . \end{aligned} $$ 那么$\bar{Y}$的BLUP就是$\mathrm{t}_r$和 $$ \begin{aligned} B_m\left(t_r\right) & =E_m\left(t_r-\bar{Y}\right)=0 \ V_m\left(t_r-\bar{Y}\right) & =\frac{1-f}{n}\left[1+\frac{(\bar{X}-\bar{x})^2}{(1-f) g(s)}\right] \sigma^2=\phi(s) \sigma^2, \text { say, } \end{aligned} $$ 写作 $$ g(s)=\frac{1}{n} \sum_s\left(X_i-\bar{x}\right)^2 . $$ 然后,对于 $$ \hat{\sigma}^2=\frac{1}{(n-2)} \sum_s d_i^2 $$ 我们有 $$ E_m\left(\hat{\sigma}^2\right)=\sigma^2 $$ 因此, $$ v_L=\phi(s) \hat{\sigma}^2=\frac{1-f}{n(n-2)}\left[1+\frac{(\bar{X}-\bar{x})^2}{(1-f) g(s)}\right] \sum_s d_i^2 $$ 是$\mathcal{M}{10}^{\prime}$下$V_m\left(t_r-\bar{Y}\right)$的一个$m$ -无偏估计量。如果样本是平衡的,则$v_L$中的$$ h(s)=\frac{(X-\bar{x})^2}{(1-f) g(s)} $$项消失,即$\bar{x}=\bar{X}$,并且对于平衡的样本,$V_m\left(t_r-\bar{Y}\right)$是$\mathcal{M}{10}^{\prime}$下的最小值。 一般来说, $$ v_L=(1+h(s)) v{l r} \geq v_{l r}
$$
只有平衡的样本才相等。如果绘制了一个平衡样本,那么基于它的经典基于设计的估计器$v_{l r}$对于$V_m\left(t_r-\bar{Y}\right)$就变为$m$ -无偏。

统计代写|抽样调查作业代写sampling theory of survey代考 请认准statistics-lab™

统计代写请认准statistics-lab™. statistics-lab™为您的留学生涯保驾护航。统计代写|python代写代考

R语言代写问卷设计与分析代写
PYTHON代写回归分析与线性模型代写
MATLAB代写方差分析与试验设计代写
STATA代写机器学习/统计学习代写
SPSS代写计量经济学代写
EVIEWS代写时间序列分析代写
EXCEL代写深度学习代写
SQL代写各种数据建模与可视化代写

发表回复

您的电子邮箱地址不会被公开。 必填项已用 * 标注