Sampling and a large data set
引入| English | 释义 | 拼音 |
|---|---|---|
| quota sampling/ˈkwəʊtə ˈsæmplɪŋ/ | 配额抽样 | pèi é chōu yàng |
| simple random sampling | 简单随机抽样 | jiǎn dān suí jī chōu yàng |
| systematic sampling/ˌsɪstəˈmætɪk ˈsæmplɪŋ/ | 系统抽样 | xì tǒng chōu yàng |
| stratified sampling/ˈstrætɪfaɪd ˈsæmplɪŋ/ | 分层抽样 | fēn céng chōu yàng |
"Dewey defeats Truman", November 1948
- On 3 November 1948, the Chicago Daily Tribune printed the headline "Dewey Defeats Truman" — but Harry Truman had won the US presidential election. The polls used quota sampling 配额抽样: interviewers chose people to fill fixed numbers in each group, and the people they chose were not typical voters.
- S3 compares sampling methods, then shows how to combine random variables — the sum of several masses, or the difference between two.
Random sampling methods
- In simple random sampling 简单随机抽样 every sample of size $n$ is equally likely. Systematic sampling 系统抽样 takes every $k$th item from a list, starting at a random point among the first $k$.
- Stratified sampling 分层抽样 takes a random sample from each group in proportion to its size: from groups of 120 and 80, a sample of 30 takes 18 and 12.
Sampling and a large data set
Identify the population, sampling unit and frame
Allocate proportionally, then explain why random selection within each group still matters.
A school has 450 students in Year 10 and 350 in Year 11. A stratified sample of 40 is taken. How many Year 11 students are in it?
难度 2/5 · 今日首次完成可获 140 金币及少量额外奖励
$40 \times \tfrac{350}{800} = 17.5$, so 17 or 18 in practice.
A list has 600 names and a sample of 20 is needed. How is a systematic sample taken?
难度 1/5 · 今日首次完成可获 100 金币及少量额外奖励
$k = \tfrac{600}{20} = 30$, with a random start.
Non-random methods
- Quota sampling fills a quota for each group but lets the interviewer choose people; it is quick and needs no list, but can be biased.
- Opportunity sampling uses whoever is available — the cheapest method and the least representative.
Quota sampling is a random sampling method.
难度 1/5 · 今日首次完成可获 100 金币及少量额外奖励
The interviewer chooses who fills each quota.
Worked example: a lift's weight limit
Adult masses are $X \sim \mathrm{N}(80, 12^2)$ kg. A lift holds 8 people and has a limit of 680 kg. Find the probability that 8 random adults exceed the limit.
- The sum: $T = X_1 + \dots + X_8$ has mean $8 \times 80 = 640$ and variance $8 \times 144 = 1152$ (independent masses).
- Standardise: $\mathrm{P}(T > 680) = \mathrm{P}\left(Z > \dfrac{40}{\sqrt{1152}}\right) = \mathrm{P}(Z > 1.179)$.
- Answer: $0.119$ — about 1 lift-load in 8.
Bags have masses $\mathrm{N}(5, 0.2^2)$ kg. Find the standard deviation of the total mass of 4 bags.
难度 2/5 · 今日首次完成可获 140 金币及少量额外奖励
方差 $4 \times 0.04 = 0.16$.
Combining random variables
- For independent $X$ and $Y$: $\mathrm{E}(aX \pm bY) = a\mathrm{E}(X) \pm b\mathrm{E}(Y)$ and $\mathrm{Var}(aX \pm bY) = a^2\mathrm{Var}(X) + b^2\mathrm{Var}(Y)$ — variances always add.
- If $X$ and $Y$ are normal, so is $aX \pm bY$. Apples $\mathrm{N}(150, 10^2)$ g and oranges $\mathrm{N}(170, 15^2)$ g: $Y - X \sim \mathrm{N}(20, 325)$, so $\mathrm{P}(\text{orange heavier}) = 0.866$.
$X \sim \mathrm{N}(30, 4)$ and $Y \sim \mathrm{N}(25, 5)$ are independent. What is the distribution of $X - Y$?
难度 1/5 · 今日首次完成可获 100 金币及少量额外奖励
Means subtract; variances add.
$3X$ is not $X_1 + X_2 + X_3$
- $3X$ is one value tripled: variance $9\sigma^2$. $X_1 + X_2 + X_3$ is three independent values: variance $3\sigma^2$.
- With $\sigma = 12$: $3X$ has standard deviation 36, the sum of three has 20.8. Independent values partly cancel each other out.
$\mathrm{Var}(X) = 4$. Find $\mathrm{Var}(3X) - \mathrm{Var}(X_1 + X_2 + X_3)$.
难度 2/5 · 今日首次完成可获 140 金币及少量额外奖励
$36 - 12 = 24$.
Subtracting variances for a difference. $\mathrm{Var}(Y - X) = \mathrm{Var}(Y) + \mathrm{Var}(X) = 225 + 100 = 325$, not $225 - 100$. A difference of two random quantities is more uncertain than either one, never less.
Random methods (simple random, systematic, stratified) give every member a known chance; quota and opportunity samples do not, and can be biased. For independent variables, means combine with the same signs and coefficients, and variances add with squared coefficients. Linear combinations of normal variables are normal. $3X$ and $X_1 + X_2 + X_3$ have different variances.
A cup holds $C \sim \mathrm{N}(250, 4^2)$ ml and a machine pours $P \sim \mathrm{N}(240, 3^2)$ ml. Find the probability that a cup overflows.
Difference of normals.