Media & Information Literacy Grade 9 Statistical Manipulation How Numbers Lie

Study Guide: Statistical Manipulation – How Numbers Lie
Grade 9 | Media & Information Literacy

1. The Driving Question

"If a news headline says ‘90% of teens are addicted to social media,’ but your friends don’t seem addicted, how do you figure out if the number is real—or if someone twisted it to scare you? And why would anyone do that?"

This isn’t just about math—it’s about power. Numbers feel like facts, but they’re tools. Whoever controls them controls the story. How do you spot when they’re being bent?

2. The Core Idea – Built, Not Listed

Imagine you’re at a school basketball game. The announcer says, "Our team scored 20 points in the first quarter—proof they’re the best in the league!" But wait: the other team scored 30. The announcer chose to highlight one number and ignore the rest. That’s statistical manipulation—picking, framing, or distorting data to push a story, not the truth.

Here’s how it works: - Cherry-picking: Like only showing the three days your favorite stock went up, not the seven it crashed.
- Misleading scales: A bar graph where the y-axis starts at 90% to make a 92% approval rating look huge (when it’s really just a 2% bump).
- Correlation ≠ causation: "Ice cream sales rise when drowning deaths increase—so ice cream causes drowning!" (No, it’s summer. The real cause is hidden.) - Loaded language: "A shocking 60% of students failed!" vs. "40% of students passed—up 10% from last year!"

Key Vocabulary
1. Cherry-picking
- Definition: Selecting only the data that supports your claim while ignoring the rest.
- Example: A toothpaste ad says, "80% of dentists recommend our brand!" but leaves out that the survey gave dentists 10 brands to choose from—so 80% might’ve picked 3 or 4 others too.
- College shift: In research, this becomes "p-hacking"—running tests until you find any statistically significant result, even if it’s meaningless.

1. Correlation vs. causation
2. Definition: Two things happening together (correlation) doesn’t mean one caused the other (causation).
3. Example: A study finds that people who own more books live longer. But it’s not the books—it’s that wealthier people (who live longer) buy more books.

4. College shift: In epidemiology, this is the difference between "smoking is associated with lung cancer" (correlation) and "smoking causes lung cancer" (proven through controlled experiments).

5. Misleading scale

6. Definition: Using a graph’s axis to exaggerate or downplay changes in data.
7. Example: A climate change graph where the y-axis jumps from 0°F to 50°F, making a 1°F temperature rise look like a mountain instead of a blip.

8. College shift: In economics, this is called "axis manipulation"—used in stock charts to make small market shifts look like crashes (or booms).

9. Loaded language

10. Definition: Using emotionally charged words to make data sound more dramatic (or less) than it is.
11. Example: "A mere 5% of students passed the test" vs. "An encouraging 5% of students passed the test—up from 2% last year!"
12. College shift: In journalism, this becomes "framing"—how word choice shapes public perception (e.g., "undocumented immigrant" vs. "illegal alien").

3. Assessment Translation

How this appears on tests (and in real life):
- Multiple choice: "A graph shows that ice cream sales and drowning deaths both increase in June. What’s the most likely explanation?" (Distractors: "Ice cream causes drowning," "Drowning causes ice cream sales," "The graph is fake.") - Proficient answer: "Both increase because of a third factor—hot weather." - Developing answer: Picks a direct cause-effect (e.g., "Ice cream makes people swim more").

• Short answer: "A news article claims ‘Vaccines cause autism’ based on a study of 12 children. What’s one reason this claim might be misleading?"
• Proficient response: "The sample size is too small—12 kids isn’t enough to prove a link for millions. Also, correlation isn’t causation; maybe the kids had other factors in common."

• Developing response: "The study is bad" (lacks specifics).

• Evidence-based writing (SAT/ACT-style): "Read this article about a study linking video games to violence. Write a paragraph explaining whether the study’s conclusions are trustworthy, using evidence from the text."

• Proficient model:
  > "The study’s conclusion—that video games cause violence—isn’t trustworthy because it only surveyed 50 teens in one city. A sample that small can’t represent all gamers. Also, the article admits the study didn’t control for other factors, like home life or mental health. Without ruling those out, we can’t say games are the cause. The headline ‘Games Turn Kids Violent’ is loaded language—it makes the data sound scarier than it is."

What assessors look for:
- Grade 9: Can you spot how the manipulation works? (Not just "this is wrong," but "they cherry-picked the data by...").
- SAT/ACT: Do you use textual evidence to support your critique? (Quote the misleading scale, the small sample size, etc.).
- AP Seminar/Research: Can you design a better study? (E.g., "To prove causation, they’d need a control group and a larger sample.")

4. Mistake Taxonomy

Mistake 1: Falling for the "big number" trap
- Prompt: "A headline says ‘10,000 people injured by self-driving cars!’ Should you be worried?" - Common wrong response: "Yes, 10,000 is a huge number!" - Why it loses credit: No context. 10,000 injuries over how many cars? Over how many years? Without the denominator (total cars on the road), the number is meaningless.
- Correct approach: - Ask: "Out of how many total self-driving cars?" (If 10,000 injuries out of 100,000 cars, that’s bad. If out of 100 million, it’s rare.) - Compare: "How does this rate compare to human-driven cars?" (If humans cause 100,000 injuries, self-driving cars might actually be safer.)

Mistake 2: Assuming graphs are neutral
- Prompt: "This graph shows a 50% increase in crime in your town. Should you be concerned?" (Graph’s y-axis starts at 95%.) - Common wrong response: "Yes, crime is skyrocketing!" - Why it loses credit: The graph’s scale exaggerates the change. A 50% increase from 2% to 3% is tiny, but the graph makes it look huge.
- Correct approach: - Check the y-axis: "Does it start at 0? If not, the change is exaggerated." - Calculate the actual numbers: "If crime went from 20 to 30 incidents, that’s a 50% increase—but is 30 incidents a lot for this town?"

Mistake 3: Ignoring the source’s motive
- Prompt: "A study funded by a soda company finds ‘no link between soda and obesity.’ Is this study trustworthy?" - Common wrong response: "Yes, because it’s a study." - Why it loses credit: The funder has a vested interest in the outcome. Studies funded by industries often find results that benefit those industries.
- Correct approach: - Ask: "Who paid for this? What do they gain if the results are positive?" - Look for independent replication: "Have other, non-industry-funded studies found the same thing?"

5. Connection Layer

1. Within Media Literacy → Algorithmic bias

2. Why it matters: Statistical manipulation isn’t just about graphs—it’s how social media algorithms choose which numbers to show you (e.g., "90% of your friends like this post!" when they actually showed it to 100 people and only 90 clicked "like"). Understanding cherry-picking helps you spot when an algorithm is curating reality for you.

3. Across Subjects → Science (Experimental Design)

4. Why it matters: In science, a "control group" exists to prevent cherry-picking. If a drug trial only tests the drug on healthy 20-year-olds, the results are meaningless for 80-year-olds with heart conditions. Statistical manipulation in media is just bad science repackaged for clicks.

5. Outside School → Sports analytics

6. Why it matters: Ever hear a commentator say, "This player has the highest shooting percentage in the league!"? Check the sample size. A player who’s taken 10 shots and made 8 (80%) is not better than one who’s taken 100 and made 60 (60%). Sports stats are full of misleading scales and cherry-picked data—now you’ll notice.

6. The Stretch Question

"A politician says, ‘Our new policy reduced unemployment by 20%!’ The opposition says, ‘Unemployment actually went up by 1 million people!’ How can both statements be true at the same time?"

Pointer toward the answer:
- The 20% drop might be relative (e.g., from 5% to 4% unemployment—a 20% decrease in the rate).
- The 1 million increase is absolute (e.g., the total number of unemployed people rose from 5 million to 6 million because the population grew).
- The politician is using a percentage change to make the policy look successful, while the opposition is using raw numbers to make it look like a failure. Both are technically correct—but which one tells the real story? (Hint: It depends on whether you care about rates or total impact.)