1 / 15
Session 3 of 8
Sampling
Strategies
How do you learn about a million people without asking every single one? You take a smart sample.
Data Science for Young Minds · Grade 5 — Data Detective
2 / 15
Today's Plan
What We're Doing Today
- Hook — the mystery bead jar
- Population vs. sample
- Three sampling strategies: random, convenience, stratified
- Bead Jar Simulation — does sample size matter?
- Calculating percentages from samples
- Analysis: which sample was more accurate — and why?
3 / 15
Opening Hook
The Mystery Jar
Your teacher is holding a jar with 100 beads — four different colors mixed together.
Without counting every single bead — how could you figure out approximately how many of each color there are?
Think about it. Then we'll test your idea.
4 / 15
Lesson 1
Population vs. Sample
Population
- The ENTIRE group you want to study
- All students in your school
- All 100 beads in the jar
- Every voter in a country
- Often too large to study completely
Sample
- A smaller group selected to represent the population
- 30 students from your school
- 10 beads drawn from the jar
- 1,000 randomly selected voters
- Must be chosen carefully to be useful
The key question isn't just "how big is the sample" — it's "how was the sample chosen?"
5 / 15
Lesson 2
Three Sampling Strategies
Random Sample
Every member of the population has an equal chance of being selected. Most fair and least biased. Drawing names from a hat.
Convenience Sample
You pick whoever is easiest to reach. Fast but often biased. Asking only your friends what music everyone likes.
Stratified Sample
Divide the population into groups, then sample from each group proportionally. More complex but very representative.
A convenience sample can give you completely wrong results — even if you sample a lot of people. Who you ask matters as much as how many you ask.
6 / 15
Lesson 2
Convenience Sampling — A Famous Failure
In 1936, a magazine called Literary Digest surveyed 2.4 million people about the upcoming US election. They predicted the wrong winner — badly.
What went wrong?
- They sent surveys to phone book owners and car owners
- In 1936, mainly wealthy people had phones and cars
- Their sample was huge — but biased toward one group
- Size didn't fix the bias
The lesson
- 2.4 million people surveyed = wrong answer
- A random sample of 50,000 = correct prediction
- HOW you sample matters more than HOW MANY
- Bias in sampling = bias in conclusions
7 / 15
Lesson 3
Why Sample Size Matters Too
Once you have a randomsampling method, bigger samples give more reliable results.
- Draw 5 beads from a jar of 100: high chance of getting an unusual result
- Draw 30 beads: much more likely to reflect true proportions
- Draw 90 beads: very close to the truth — but a lot of work!
In real research, scientists calculate the minimum sample size needed to be confident in their results. More data = more certainty, but there's a point of diminishing returns.
Quick math: if you drew 30 beads and found 12 red, what % is that? → 12 ÷ 30 × 100 = 40%
8 / 15
Activity Time!
Bead Jar Simulation
Your jar contains 100 beads of 4 colors. You don't know the true distribution — yet.
- Round 1: Without looking, draw 10 beads. Count each color. Record on worksheet. Put them back and shake.
- Calculate: Convert each count to a percentage (count ÷ 10 × 100).
- Round 2: Without looking, draw 30 beads. Count each color. Record. Put back and shake.
- Calculate: Convert to percentages (count ÷ 30 × 100).
- Reveal: Your teacher will reveal the true distribution. Compare!
You have 20 minutes. Work carefully — your data is real evidence!
9 / 15
Brain Break — Sample or Population?
Stand for SAMPLE, sit for POPULATION.
"All the fish in the ocean" · "50 fish caught for a study" · "Every student at your school" · "The 30 students in this class" · "Every bead in the jar" · "10 beads you just drew"
A sample is always a subset of the population it represents!
10 / 15
Reveal!
The True Distribution
Here is what was actually in your jar:
Compare to YOUR results. Which of your samples was closer — 10 beads or 30 beads?
11 / 15
Debrief
What Did Your Data Show?
"Which sample size was closer to the true distribution? Why do you think that happened? Did anyone get a surprising result with 10 beads?"
- How far off was your 10-bead sample for each color?
- How far off was your 30-bead sample?
- If you had drawn 5 beads, what might have happened?
- Could a 10-bead sample ever be exactly right? Is that luck or skill?
This is why scientists don't trust tiny samples — even when the results look perfect.
12 / 15
Lesson 4
What Makes a Sample Representative?
- It must be selected randomly — not by who's convenient or who volunteers
- It must be large enough to capture the variation in the population
- It must reflect the proportions of different groups in the population
- Avoid self-selection bias — when only certain people choose to respond
Real-world example: A poll of 10,000 people who all watch the same TV show is less representative than a random poll of 500 people from the general population.
13 / 15
Analysis Time
Your Written Analysis
"Which sample size gave you a more accurate picture — 10 beads or 30? Use your actual data to explain. What would happen with 50 beads? 90?"
5 minutes. Use your worksheet — Part 4. Cite your actual percentages. Don't just say "30 was better" — show it with numbers.
14 / 15
Vocabulary Check
Session 3 Key Terms
Population
The entire group you want to study or learn about
Sample
A smaller group selected to represent the population
Random Sample
Every member has an equal chance of being selected
Convenience Sample
Whoever is easiest to reach — often biased
Representative
Reflects the true proportions of the population
Sample Size
The number of individuals in the sample
15 / 15
Session Close
The Sampling Detective Rule
"Ask not just how many — ask how they were chosen."
- You can explain population vs. sample
- You can compare random, convenience, and stratified sampling
- You've seen firsthand why sample size matters
- You can convert sample counts to percentages
Next session: Even a perfectly collected sample can be used to mislead. We become Data Detectives and catch the tricks.