01 / 21

Introducing
Data Samples

Why is my sample not like yours? Explore sampling variability, the difference between populations and samples, and why statistics depends on the art of good sampling.

AP Statistics

What Is a Population?

In statistics, a population is the entire group of individuals or observations that you want to study. It includes every single member that fits your criteria: every student in a school, every light bulb produced in a factory, or every voter in a country.

What Is a Sample?

A sample is a subset of the population that we actually collect data from. Because it's impractical (or impossible) to measure every member, we use the sample to estimate the population parameter.

Population Sample Parameter: \(\mu\) | Statistic: \(\bar{x}\)

Why Is My Sample Not Like Yours?

Every time you draw a new random sample from the same population, you'll likely get different values. This natural variation is called sampling variability. It's not a flaw: it's an inherent feature of working with subsets.

Population: \(\mu = 70\) Sample 1 68, 72, 65, 74, 71 \(\bar{x}_1 = 70.0\) Sample 2 75, 62, 80, 69, 66 \(\bar{x}_2 = 70.4\) Sample 3 73, 77, 64, 71, 68 \(\bar{x}_3 = 70.6\) Each sample gives a different \(\bar{x}\): that's sampling variability. 68 70 72 70.0 70.4 70.6

The Law of Large Numbers

As the sample size \( n \) increases, the sample mean \( \bar{x} \) gets closer and closer to the true population mean \( \mu \). This is the Law of Large Numbers: larger samples give more reliable estimates.

Sample size \( n \) \( \bar{x} \) \(\mu\) 5 25 100 1000

The Sampling Distribution

If you repeated the sampling process thousands of times and plotted every \( \bar{x} \), you'd get the sampling distribution of \( \bar{x} \). This distribution tells us how much sample means vary from sample to sample.

\(\mu\) Spread = \(\frac{\sigma}{\sqrt{n}}\) Sampling Distribution of \(\bar{x}\)

A filling line that deposits one statistic for a whole tray

A filling line loads bottles into trays while one statistic is recorded for each tray.A tray contains individual observations, but its average contributes one value to a distribution of sample means. Repeating the sampling process reveals how that statistic varies.

One tray contributes one mean

Twenty-five labeled bottle values are combined into one sample-mean marker at 50.4 mL.Suppose each independent fill is equally likely to be 40 or 60 mL. This tray contains twelve 40-mL and thirteen 60-mL fills, totaling 1,260 mL; its 25 observations contribute one mean, 50.4 mL.

Build the distribution of tray means

Twenty-six possible sample means appear as discrete probability bars with a normal approximation drawn for comparison.For 25 independent fills equally likely to be 40 or 60 mL, K is binomial with n = 25 and p = 0.5, and the tray mean is 40 + 0.8K. The exact distribution is discrete, centered at 50 mL with standard deviation 2 mL. A normal curve approximates its shape.

Parameter vs. Statistic

Parameter Statistic
Describes Population Sample
Notation (mean) \( \mu \) \( \bar{x} \)
Notation (proportion) \( p \) \( \hat{p} \)
Notation (std. dev.) \( \sigma \) \( s \)
Value Fixed but usually unknown Varies from sample to sample
Mnemonic Population → Parameter Sample → Statistic

Types of Random Sampling

Simple Random
Every member of the population has an equal chance of being selected. Use a random number generator or drawing names from a hat.
Stratified
Divide the population into strata (groups), then take a simple random sample from each stratum. Ensures representation.
Cluster
Divide the population into clusters (often geographic), randomly select entire clusters, then survey all members within them.
Systematic
Select every \( k^{\text{th}} \) member after a random starting point. Simple to execute but risky if there's a hidden pattern.

Steps for Good Sampling

1
Define the population. Be specific about who or what you're studying.
2
Create a sampling frame. List every member of the population (or as close as possible).
3
Choose a sampling method. Simple random, stratified, cluster, or systematic: pick what fits your resources and goals.
4
Select a large enough \( n \). Bigger samples reduce variability: \( \text{SE} = \frac{\sigma}{\sqrt{n}} \).
5
Watch for bias. Voluntary response, convenience sampling, and undercoverage can all produce misleading results.

Sources of Bias

Undercoverage
Some groups in the population are left out of the sampling frame. Example: a phone survey that excludes people without phones.
Voluntary Response
Only people who feel strongly respond. Online polls and call-in surveys attract extreme opinions, skewing results.
Nonresponse
Selected individuals choose not to participate. If nonrespondents differ systematically from respondents, results are biased.

Example: School Lunch Survey

A school principal wants to estimate the proportion of students who prefer hot lunch. The school has 1,200 students. She surveys 50 students in the cafeteria during lunch period and finds \( \hat{p} = 0.72 \).

Example: Calculating Standard Error

A population of test scores has \( \mu = 500 \) and \( \sigma = 100 \). You take a random sample of \( n = 25 \) students. What is the standard error of \( \bar{x} \)?

Example: Stratified vs. SRS

A company with 300 managers and 1,700 hourly workers wants to estimate average job satisfaction. Which method is better?

SRS of 100
Draw 100 employees randomly. You might get 85 hourly workers and 15 managers: reasonable representation, but no guarantee.
Stratified: 15 + 85
Take 15 managers and 85 hourly workers (proportional to population). Guarantees both groups are represented correctly.

Variability vs. Bias

High Bias Low Variability Low Bias High Variability Bullseye = true parameter value

Common Mistakes

✕ Confusing Parameter and Statistic
Using \( \bar{x} \) when you mean \( \mu \), or saying "the population statistic." Parameters describe populations; statistics describe samples.
✕ Thinking Bigger Samples Fix Bias
A biased method stays biased no matter how large \( n \) is. Surveying 10,000 people at a gym still only captures gym-goers.
✕ Confusing Variability with Bias
Getting different results each time is variability (expected). Consistently over- or under-estimating is bias (a design problem).

Key Concepts Summary

Concept Definition Key Fact
Population Entire group of interest Described by parameters
Sample Subset actually observed Described by statistics
Sampling Variability Statistics differ sample to sample Decreases as \( n \) increases
Bias Systematic error in method Does NOT decrease with \( n \)
Standard Error \( \frac{\sigma}{\sqrt{n}} \) Measures spread of \( \bar{x} \)

Question 1 of 5

A researcher surveys every student in a school about screen time. The average screen time she calculates is best described as:

Question 2 of 5

A radio station asks listeners to call in and vote on their favorite song. This is an example of:

Question 3 of 5

A population has \( \sigma = 40 \). If the sample size increases from \( n = 16 \) to \( n = 64 \), the standard error of \( \bar{x} \):

Question 4 of 5

A polling company consistently overestimates support for a candidate by 3 percentage points across many polls. Increasing the sample size to \( n = 10{,}000 \) will:

Question 5 of 5

A school has 400 freshmen, 350 sophomores, 300 juniors, and 250 seniors. To ensure each grade is represented proportionally in a sample of 130, which method should be used?