Statistics

Media Bias & Statistical Analysis

1

Misleading vs Fair Graph Techniques

Sort each data presentation technique into the correct column: Misleading or Fair.

Bar chart with a y-axis starting at 0
Bar chart with a y-axis starting at 95
Pie chart where slices add to 100%
3D pie chart that distorts slice sizes
Graph with evenly spaced time intervals
Graph with uneven time intervals on the x-axis
Survey results noting the sample size was 1 000
Survey headline with no sample size mentioned
Cherry-picked data points to support a claim
Using a pictograph where each icon represents a different amount
Misleading
Fair
2

Misleading Technique → Why It's Problematic

Draw a line from each misleading technique to the reason it distorts data.

Truncated y-axis
3D pie chart
Cherry-picked time frame
Unlabelled axes
Dual y-axes with different scales
Makes small differences look dramatic
Distorts the relative size of slices due to perspective
Hides the overall trend by showing only favourable data
The reader cannot tell what the graph actually measures
Makes two unrelated variables appear to move together
3

Spot the Problem — Graphs

Circle the main problem with each graph description.

A bar chart comparing company profits starts its y-axis at $9.5 million instead of $0.

The truncated axis exaggerates the differences between bars
Bar charts should never show money
There is no problem — this is standard practice

A pie chart has five slices that add up to 120%.

Pie chart percentages must total 100%; these slices overlap or are incorrect
Pie charts always add to more than 100%
The chart just needs a bigger circle

A line graph of 'yearly rainfall' uses data from only January to March.

Three months is not a full year — the graph cherry-picks a seasonal period
Rainfall can only be measured annually
Line graphs should not be used for rainfall data

A pictograph uses one icon = 10 units in the first row but one icon = 50 units in the second row.

Inconsistent icon scales make the comparison meaningless
Pictographs are always accurate
The second row is just more important
4

Sample Size & Representativeness

Circle the BEST critique of each survey's sampling method.

A toothpaste brand surveys 5 dentists and claims '4 out of 5 dentists recommend our product.'

The sample size is far too small to generalise
Dentists always agree with brands
5 is a perfect sample size for dentists

A political poll interviews 2 000 people, all from one suburb in Sydney.

The sample is not representative of the whole country — it's geographically biased
2 000 people is always a good sample
Sydney is the capital, so it represents everyone

An online survey asks visitors to a fitness website whether they exercise daily.

The sample is biased — people on a fitness website are more likely to exercise
Online surveys are always reliable
Everyone who visits a fitness website exercises daily

A school surveys every student in Years 7–12 about canteen food preferences.

This is a reasonable census of the school — the results apply to this school's students
You can never trust surveys done in schools
Only Year 12 opinions matter
5

Biased vs Unbiased Sampling Methods

Sort each sampling method into the correct column: Biased or Unbiased.

Randomly selecting names from the full electoral roll
Surveying only people who volunteer to respond
Selecting every 10th person from a complete class list
Asking only your friends for their opinions
Stratified random sample across age groups and regions
Surveying shoppers at one luxury store
Phone poll using randomly generated landline numbers only
Posting a poll on a partisan political Facebook page
Biased
Unbiased
6

Mean, Median, or Mode — Which Is Best?

Circle the BEST measure of centre for each data set.

House prices in a suburb: $450k, $470k, $480k, $490k, $3.2 million

Median — the extreme outlier ($3.2 m) would inflate the mean
Mean — it uses every value
Mode — it is the most common measure

Shoe sizes sold in a shop last week: 8, 9, 9, 9, 10, 10, 11

Mode — the shop needs to know the most popular size to restock
Mean — the average shoe size is most useful
Median — it splits the data in half

Test scores evenly spread from 55 to 95 with no outliers

Mean — the data is symmetric with no outliers, so mean and median are similar and the mean uses all values
Mode — the most common score matters most
None — you cannot summarise test scores

Annual salaries at a startup: $60k, $62k, $65k, $65k, $900k (CEO)

Median — the CEO's salary is an extreme outlier that skews the mean
Mean — it accounts for the CEO
Mode — $65k appears twice
7

Measure of Centre → When to Use It

Draw a line from each measure of centre to the scenario where it is most appropriate.

Mean
Median
Mode
Mean after removing outliers
Weighted mean
Symmetric data with no extreme outliers
Data with outliers or a skewed distribution
Finding the most frequently occurring category
Data set with one or two obvious anomalies you can justify excluding
Combining averages from groups of different sizes (e.g. exam components worth different percentages)
8

Who Was Asked? — Identify the Audience

Circle the statement that correctly identifies a problem with who was surveyed.

A car magazine survey finds that 90% of people think cars are better than public transport.

The readers of a car magazine are biased towards cars
90% is too high to be real
Magazines cannot conduct surveys

A soft-drink company surveys teenagers at a music festival about their favourite drinks.

Festival-goers are not representative of all teenagers, and the setting may bias answers toward soft drinks
Teenagers always prefer soft drinks
The sample is perfectly representative

A government health survey randomly selects 5 000 households across all states and territories.

This is a well-designed sample — random selection across regions reduces bias
Government surveys are always biased
5 000 households is too small for a country
9

Valid Conclusions from Survey Data

Circle the VALID conclusion that can be drawn from each scenario.

A survey of 50 students at a private school found that 80% prefer organic food. Headline: 'Most Australians prefer organic food.'

Invalid — the sample is not representative of all Australians
Valid — 80% is a large majority
Valid — students are a good sample of the population

A nationwide random survey of 10 000 adults finds that 62% support renewable energy. Margin of error: ±2%.

Valid — a large random sample with a stated margin of error supports the conclusion that a majority of adults support renewable energy
Invalid — you need to survey everyone
Invalid — margins of error mean the results are unreliable

A company surveys its own employees and finds 95% are satisfied. Headline: 'Workers love their jobs.'

Invalid — employees may feel pressured to give positive responses, and one company does not represent all workers
Valid — 95% is nearly everyone
Valid — employees know best about their own jobs
10

Identify Confounding Variables

Circle the most likely confounding variable in each scenario.

A study finds that children who eat breakfast score higher on maths tests.

Household income — families who can afford breakfast may also provide more educational support
Breakfast directly improves maths ability
Children who eat breakfast are naturally smarter

Towns with more police officers have higher crime rates.

Population size — larger towns have both more police and more crime
Police cause crime
Crime data must be wrong

People who own more books tend to live longer.

Socioeconomic status — wealthier people can afford both books and better healthcare
Reading books extends your lifespan
There is no connection at all

Countries that consume more chocolate win more Nobel Prizes.

Wealth and education levels — richer countries can afford both chocolate and research funding
Chocolate makes people smarter
Nobel Prize winners eat more chocolate to celebrate
11

Correlation vs Causation in Media Claims

Circle the correct interpretation of each media claim.

A graph shows ice cream sales and drowning rates both increase in summer.

Both are caused by a third variable (hot weather) — correlation does not mean causation
Ice cream sales cause drownings
We should ban ice cream to prevent drownings

'Countries that spend more on education have higher GDP.'

There is a correlation, but wealthier countries can afford to spend more on both — the direction of causation is unclear
Spending on education directly causes GDP growth
GDP has nothing to do with education

'People who sleep with their shoes on are more likely to wake up with a headache.'

A confounding variable (such as alcohol consumption) likely causes both behaviours
Sleeping with shoes on causes headaches
This proves shoes are bad for your health
12

Misleading Percentages

Circle the correct response to each percentage claim.

An ad claims: 'Our cereal has 50% less sugar!' But it was reduced from 2 g per serve to 1 g per serve.

The percentage is technically correct but the absolute reduction is tiny — only 1 gram
50% less is always a big deal
The claim is false because 1 g is still sugar

A headline states: 'Disease X cases doubled!' Cases went from 2 to 4 in a city of 1 million.

The percentage increase (100%) sounds alarming, but the absolute numbers are extremely small and statistically insignificant
Doubling is always dangerous
The headline is completely accurate and not misleading at all

A politician says unemployment dropped from 5.0% to 4.8% and calls it 'a massive improvement.'

A 0.2 percentage point drop may be within the margin of error and is a modest change, not 'massive'
Any decrease in unemployment is massive
Unemployment statistics are never accurate

A supplement company claims their product 'triples your chance of recovery' — from 1% to 3%.

The relative increase is large (200%), but the absolute increase is only 2 percentage points — still a 97% chance of not recovering
Tripling your chance is always significant
The product must be very effective
13

Steps to Evaluate a Statistical Claim

Put the steps for critically evaluating a statistical claim into the correct order.

?
Read the claim and identify the specific statistic being used
?
Ask: Who collected the data and do they have a vested interest?
?
Check the sample — how large is it and is it representative?
?
Identify whether correlation or causation is being claimed
?
Look for confounding variables that could explain the result
?
Check whether absolute or relative numbers are being used (and which is more informative)
?
Consider what information is missing (margin of error, time period, comparison group)
?
Form your own conclusion based on the evidence, not just the headline
14

Logical Fallacy → Example

Draw a line from each logical fallacy to its correct example.

Cherry-picking
False cause
Hasty generalisation
Appeal to authority
Straw man
Showing only the three months where sales rose and hiding the nine months where they fell
Claiming that a rooster crowing causes the sun to rise because it happens every morning
Concluding all teenagers are reckless drivers after seeing one accident involving a teen
Claiming a statistic is true because a celebrity endorses it, despite no scientific backing
Exaggerating an opponent's argument to make it easier to attack (e.g. 'You want to cut the budget? So you want children to starve?')
15

Headline vs Reality

Circle the response that best describes the gap between the headline and the actual data.

Headline: 'Teen Screen Time SKYROCKETS!' Data: Average daily screen time for teens rose from 6.8 hours to 7.1 hours over 2 years.

The headline is sensationalised — a 0.3-hour increase over 2 years is modest, not a 'skyrocket'
Any increase in screen time is a crisis
The headline is accurate because the number went up

Headline: 'MAJORITY of residents oppose the new park.' Survey: 52% opposed, 48% supported, margin of error ±4%.

The result is within the margin of error, so there may be no real majority either way
52% is a clear majority
Margins of error can be ignored

Headline: 'Miracle Fruit Cuts Cancer Risk by 50%!' Study: Participants who ate the fruit daily for 10 years had a 0.02% cancer rate vs 0.04% in the control group.

The relative reduction is 50%, but the absolute risk was already tiny — the benefit is minimal in practice
50% is a huge reduction, so everyone should eat the fruit
The study proves the fruit cures cancer
16

Questions to Ask About a Statistic

Sort each item into the correct column: Important Question to Ask or Not Particularly Relevant.

How large was the sample?
What colour was the graph?
Who funded or commissioned the study?
Was the sample randomly selected?
What font was used in the report?
What is the margin of error?
Were confounding variables controlled for?
Is the percentage a relative change or an absolute change?
What was the time period of the study?
How many pages is the report?
Important Question to Ask
Not Particularly Relevant
17

Effect of Outliers on Mean vs Median

Circle the correct statement about how outliers affect each measure.

Data set: 12, 14, 15, 16, 17, 98. What happens to the mean vs the median?

The outlier (98) pulls the mean up significantly, but the median stays near 15.5
Both the mean and median are equally affected
The median is more affected than the mean

A company reports the 'average salary' is $120 000. Which is most likely true?

A few executives earning very high salaries have inflated the mean — the median salary is probably much lower
Most employees earn close to $120 000
The mean and median are the same in any salary data

If you remove the single highest value from a data set, which measure changes more?

The mean — because it uses every value in its calculation
The median — because it depends on the middle value
Neither changes — they are both resistant to change
18

Absolute vs Relative Change

Circle the response that correctly interprets each claim using absolute and relative change.

A charity says donations rose 200% — from $50 to $150.

The relative increase is large (200%), but the absolute increase is only $100 — context matters
200% always means a huge amount of money
This claim is false because $150 is not 200% of $50

A headline says 'Shark attacks up 50%!' Last year: 2 attacks. This year: 3 attacks.

The relative change (50%) sounds alarming, but the absolute change is just 1 extra attack — the risk remains extremely low
A 50% increase in shark attacks is a major public safety crisis
Shark attack statistics are always unreliable

A drug reduces heart attack risk by 25%. The control group had a 4% risk; the drug group had a 3% risk.

The relative risk reduction is 25%, but the absolute reduction is only 1 percentage point — 100 people would need to take the drug to prevent 1 heart attack
25% risk reduction means the drug prevents a quarter of all heart attacks globally
A 1 percentage point difference proves the drug does not work
19

Critique a Truncated Y-Axis Bar Chart

Analyse the following scenario and write a detailed critique.

A news headline states: 'Crime has DOUBLED!' The bar chart shows crime incidents going from 498 to 512, but the y-axis starts at 490. Explain: (a) why the truncated y-axis makes the increase look much larger than it is, (b) what the actual percentage increase is, (c) how the graph should be presented fairly, and (d) why a reader who only glances at the chart might be misled.

20

Evaluate a Media Claim About Violent Crime

Critically evaluate the following media claim.

A social media post claims: 'Australia is getting more dangerous — violent crime rose 8% last year!' List at least 5 questions you would ask before accepting or rejecting this claim. Consider: sample and data source, definition of 'violent crime', population growth, historical trends, and whether reporting rates have changed.

21

Analyse a Misleading Advertisement Statistic

Write a critical analysis of the following advertisement.

A weight-loss supplement ad claims: '95% of users lost weight in just 4 weeks!' In small print, the study had 20 participants, was funded by the supplement company, had no control group, and 'weight loss' was defined as losing at least 100 grams. Write a critical analysis covering: (a) why the sample size is a problem, (b) why the funding source matters, (c) why a control group is needed, and (d) why the definition of 'weight loss' is misleading.

22

Compare Two News Sources Reporting the Same Data

Analyse how the same data can be framed differently.

Two news outlets report on the same employment data. Source A headline: 'Economy booming — 50 000 new jobs created!' Source B headline: 'Job crisis deepens — unemployment still above 5%.' Both are citing the same government report. Explain: (a) how each source selected different aspects of the data to support its narrative, (b) what additional context a reader would need, (c) which framing (if either) is more honest, and (d) how you would write a balanced headline.

23

Design a Fair Survey

Apply your knowledge of sampling and bias to design an unbiased survey.

You want to find out whether students in your state support later school start times. Design a fair survey by describing: (a) your target population and sample size, (b) your sampling method and why it reduces bias, (c) at least 3 unbiased survey questions (avoid leading questions), (d) how you would report the results honestly, including margin of error and limitations.

24

True or False — Statistical Literacy

Circle TRUE or FALSE for each statement.

A correlation of 0.95 between two variables proves that one causes the other.

FALSE — correlation measures association, not causation, no matter how strong
TRUE — a high correlation always means causation

If a graph's y-axis starts at a value other than 0, it is always misleading.

FALSE — sometimes starting above 0 is appropriate (e.g. showing small changes in temperature), but it should be clearly labelled
TRUE — the y-axis must always start at 0

A larger sample size generally gives more reliable results.

TRUE — larger samples reduce the effect of random variation, assuming the sampling method is unbiased
FALSE — sample size does not matter if the questions are good

The median is always a better measure of centre than the mean.

FALSE — the best measure depends on the data; the mean is preferred for symmetric data without outliers
TRUE — the median is always more accurate

If a study is funded by a company with a financial interest in the results, the findings should be treated with extra scrutiny.

TRUE — funding sources can create conflicts of interest, and the methodology should be examined carefully
FALSE — who funds a study has no effect on its validity
25

Rewrite a Misleading Headline

Rewrite each misleading headline to be accurate and fair.

Misleading headline: 'DEADLY new disease sweeps the nation — cases up 300%!' Reality: Cases rose from 3 to 12 nationwide (population 26 million), the disease is rarely fatal, and the increase is partly due to improved testing. Rewrite the headline to be accurate, then explain in 2–3 sentences why the original was misleading.

Misleading headline: 'Students today are the worst readers in history!' Reality: Average reading scores dropped by 2 points on a 500-point scale compared to 10 years ago, while participation in the assessment increased by 15% (including more students with learning difficulties). Rewrite the headline to be accurate, then explain why the original was misleading.

26

Why 'Average' Can Be Misleading

Explain how the word 'average' can be used to mislead.

A real estate agent says: 'The average home in this suburb sold for $1.2 million last year.' There were 10 sales: nine homes sold for between $600k and $750k, and one waterfront mansion sold for $6 million. (a) Calculate the approximate mean and median sale prices. (b) Explain why the agent chose to report the mean. (c) Which measure would be more useful for a buyer looking at typical homes in the suburb? (d) How could the agent present the data honestly?

27

Media Literacy Challenge

Critically examine statistics used in the media.

  • 1Find a graph or statistic in a news article, advertisement, or social media post. Identify at least 3 potential sources of bias or misleading presentation (e.g. truncated axis, small sample, missing context). Write down how you would improve the presentation.
  • 2Watch a news segment that uses statistics. Record: the claim, the evidence given, and any missing information (sample size, methodology, who funded the research, margin of error). Rate the claim's reliability out of 10 and justify your rating.
  • 3Compare how two different news sources report on the same statistic (e.g. employment, housing prices, crime rates). Note the differences in framing, headline language, and which data each source chose to highlight or omit.
28

Create Your Own Fair Infographic

Apply what you have learned to create an honest, well-designed infographic.

  • 1Choose a topic you care about (sport, environment, music, gaming, etc.) and find real data from a reliable source (ABS, government report, reputable news outlet).
  • 2Create a hand-drawn or digital infographic that presents the data fairly: use a y-axis starting at 0 (or clearly label it if not), include sample sizes and sources, use consistent scales, and avoid cherry-picking.
  • 3Write a short paragraph underneath your infographic explaining the design choices you made and why they help the reader understand the data honestly.
29

Misleading Statistics — Identify and Critique

Identify and explain how statistics can be misleading.

A shampoo advertisement claims: 'In a study of 50 people, 90% noticed reduced hair fall.' Identify at least 3 reasons this statistic might be misleading or unreliable.

A newspaper reports: 'Hospital admissions rose by 50% during the long weekend.' The actual numbers were 6 admissions on a normal day and 9 on the long weekend. Explain why this is technically correct but potentially misleading.

30

Statistical Bias Types

Draw a line from each bias type to its correct description.

Sampling bias
Confirmation bias
Response bias
Publication bias
Leading question bias
Surveys that phrase questions to lead respondents toward a particular answer
Only reporting studies that show significant results, not null results
When the sample does not represent the target population
Respondents give answers they think the interviewer wants to hear
Seeking or interpreting information that confirms existing beliefs
31

Identify the Misleading Graph Technique

Circle the technique used to mislead in each graph description.

A bar chart where the y-axis starts at 95 instead of 0, making a small difference look huge:

Truncated axis
Cherry-picking data
Reversed axis

A line graph that shows data for only the past 3 months of a 10-year period where the trend favours the claim:

Cherry-picking data
Misleading scale
Wrong chart type

A 3D pie chart where the front slice appears much larger than its true proportion:

3D distortion/perspective
Truncated axis
Lack of labels

Comparing two bar charts with different y-axis scales placed side by side:

Inconsistent scales
Cherry-picked data
Reversed axis
32

Mean, Median, Mode — Which to Use?

Choose and justify the appropriate measure of centre for each context.

Household incomes in a suburb: $45,000; $52,000; $48,000; $51,000; $1,200,000. Calculate the mean, median, and mode. Which best represents a 'typical' household income? Why does the millionaire distort the mean?

A shoe shop records sizes sold in one day: 8, 9, 9, 9, 10, 10, 11, 12, 13. Which measure of centre is most useful for the shop manager deciding which sizes to reorder? Why?

33

Sample vs Population

Sort each situation: describes a Sample or a Population.

Testing 200 batteries from a production run of 50,000
Recording the height of every student at a school for a study
Surveying 1,000 voters to estimate election results for all voters
Measuring the blood pressure of all patients in a clinic on one day
Polling 500 teenagers about their music preferences
Counting the exact number of goals scored in every AFL game in a season
Sample
Population
34

Sampling Methods — Compare and Evaluate

Compare different sampling methods and their limitations.

A researcher wants to estimate the average screen time of Year 10 students in Australia. Compare three sampling methods: (a) simple random sample, (b) stratified sample (by state), (c) convenience sample (asking students at the local library). For each, describe how the sample would be selected and identify a potential source of bias.

35

Standard Deviation — Interpretation

Circle the correct interpretation of standard deviation for each scenario.

Test scores: Mean = 70, SD = 3. This means:

Most scores are clustered closely around 70
Most scores are spread far from 70
The highest score is 73

Two classes: Class A has SD = 2, Class B has SD = 12 (both have mean 65). Which is more consistent?

Class A (smaller SD = less spread)
Class B (larger SD = more consistent)
Both are equally consistent

On a normal distribution, approximately 68% of data falls within:

1 standard deviation of the mean
2 standard deviations of the mean
3 standard deviations of the mean
36

Box Plots — Construct and Compare

Construct box plots and use them for comparison.

For the data set: 12, 15, 18, 21, 22, 23, 25, 28, 30, 35, 42 (a) Find the median, Q1, Q3, minimum, and maximum. (b) Calculate the IQR. (c) Identify any outliers using: value < Q1 − 1.5×IQR or value > Q3 + 1.5×IQR. (d) Describe the shape of the distribution.

37

Graph Type to Best Use

Draw a line from each graph type to its best statistical use.

Box plot
Scatter plot
Histogram
Bar chart
Pie chart
Compare two categorical variables or show part-to-whole
Show bivariate data and correlation between two numerical variables
Compare distributions of numerical data (median, spread, outliers)
Show the distribution and shape of continuous numerical data
Compare counts or frequencies across categories
38

Statistical Reports — Evaluate and Rewrite

Critically read and improve a flawed statistical report.

A health department report states: '75% of people who ate our new supplement felt healthier.' The study involved 20 volunteers who self-reported their health. Identify at least 4 flaws in this study (sample size, self-report, no control group, etc.). Suggest how each flaw could be corrected for a more valid study.

39

Media Statistical Analysis

Critically analyse statistics used in Australian media.

  • 1Find a news article that uses a statistic to support a claim (e.g. about health, crime, economy). Identify: the statistic used, the source, the sample size if given, and any potential bias or missing context. Write a one-paragraph critical evaluation.
  • 2Watch a political advertisement or read a political press release. Identify every statistical claim made. Look up the original source (if provided) and check whether the statistics are being used accurately and in context.
  • 3Compare how two different Australian newspapers report the same statistic or survey result. Do they emphasise different aspects? Do they use different chart types or scales?
40

Skewness and Data Distribution

Describe and interpret skewed distributions.

Explain the difference between a positively skewed and negatively skewed distribution. For each type, state: (a) the shape of the histogram, (b) which measure (mean or median) is larger, (c) give a real-world example of data with that skew.

House prices in a suburb have mean $850,000 and median $720,000. (a) Is the distribution positively or negatively skewed? (b) Which measure better represents the 'typical' house price? (c) What kind of data point would cause this skew?

41

Classify: Good Sampling Practice vs Poor Sampling Practice

Sort each practice into Good or Poor sampling practice.

Using a random number generator to select participants
Surveying only your own friend group about social media habits
Stratifying a sample by age and gender to match the population
Asking leading questions in a survey ('Don't you agree that...')
Using a large enough sample to reduce chance variation
Voluntary response survey on a political website
Good Practice
Poor Practice
42

Types of Statistical Errors Found in Media

Tally each type of statistical error you find while reviewing 20 news articles or reports.

ItemTallyTotal
Misleading graph (truncated axis, 3D)
Small or biased sample
Confusing correlation with causation
Missing context or baseline
Cherry-picked data range
43

Interquartile Range and Outliers

Use IQR to identify and reason about outliers.

A data set of daily step counts has Q1 = 5,200, Q3 = 10,800. (a) Calculate the IQR. (b) Calculate the outlier boundaries (Q1 − 1.5×IQR and Q3 + 1.5×IQR). (c) A value of 25,000 steps is recorded. Is it an outlier? Should it be excluded from the analysis? Justify your answer.

44

Statistical Literacy — True or False

Circle True or False for each statement about statistical practice.

A larger sample size always eliminates bias

False — a biased sampling method stays biased even with large samples
True — more data always reduces all errors

A statistically significant result means the result is practically important

False — significance depends on sample size; a tiny effect can be significant with a huge sample
True — statistical significance equals real-world importance

The median is always a better measure of centre than the mean

False — the mean is better for symmetric distributions without outliers
True — the median is always better

A survey with 100% response rate guarantees unbiased results

False — bias can come from question wording, sample selection, and other sources
True — 100% response eliminates bias
45

Sampling Methods — Match the Definition

Draw a line from each sampling method to its description.

Simple random sample
Stratified sample
Systematic sample
Convenience sample
Cluster sample
Voluntary response sample
Every nth person on a list is selected
Respondents choose themselves
Groups are selected, then all members surveyed
Population divided into groups, proportional sample from each
Selecting whoever is easiest to reach
Each member of population has equal chance of selection
46

Evaluating Media Statistics

Critically analyse statistical claims from media sources.

Find or recall a headline that uses statistics (e.g. '9 out of 10 dentists recommend...'). Write the headline and identify two questions you would ask before accepting the claim.

A poll reports: '65% of Australians support the new policy.' What information would you need about the poll before trusting this result?

Explain how a graph with a truncated y-axis can mislead readers. Sketch an example showing how the same data looks very different depending on where the axis starts.

Draw here
47

Statistical Bias — Identify the Type

Sort each scenario into the type of bias it demonstrates.

Asking shoppers at a mall about online shopping habits
A question: 'Do you agree that the dangerous new road is unsafe?'
Only publishing studies that show positive drug effects
A bathroom scale that reads 2 kg too low
Surveying students during school to find youth opinions
Respondents overreporting their exercise frequency
Sampling bias
Response bias
Measurement bias
Publication bias
48

Box Plots and Comparing Distributions

Construct and interpret box plots to compare data sets.

Calculate the five-number summary (min, Q1, median, Q3, max) for this data set: 12, 15, 18, 21, 24, 27, 30, 33, 36. Show all working.

Draw a box plot for the data above. Label all five key values.

Draw here

A second class has a box plot with median 20, IQR of 8, and a longer upper whisker. Compare the two distributions in terms of centre, spread, and skewness.

49

Types of Misleading Graphs Encountered

Tally each type of misleading graph technique you identified in media examples.

ItemTallyTotal
Truncated y-axis
3D distortion
Cherry-picked time period
Missing data labels
Sample size not reported
50

Standard Deviation and Spread

Calculate and interpret standard deviation.

Explain what standard deviation measures in plain language. Why is it more useful than the range?

Calculate the mean and standard deviation for: 4, 7, 7, 9, 11. Show all steps.

Class A has mean = 72 and sd = 4. Class B has mean = 72 and sd = 12. What does this tell you about the two classes?

If a new student who scored 90 joins Class A, what effect would this have on the mean and standard deviation?

51

Designing a Statistical Investigation

Plan a complete statistical investigation from question to conclusion.

Write a research question you could investigate with secondary data available online (e.g. ABS, Bureau of Meteorology).

Describe your data collection plan: source, variables, sample size, and any potential biases.

List three statistical tools (e.g. mean, scatter plot, box plot) you would use to analyse the data and explain why each is appropriate.

Describe how you would present your findings to a non-mathematical audience. What visualisations would you use and why?

52

Statistical Literacy in the Media

Become a critical consumer of statistics in everyday life.

  • 1Collect three news articles this week that include statistics. For each, write one question you would ask about the data before accepting the claim.
  • 2Compare how the same statistic is reported by two different news sources. Do they interpret it the same way? Write about any differences.
  • 3Watch or read an advertisement that uses statistics. Identify any misleading techniques used. Write a more honest version.
  • 4Research a case study where bad statistics led to a wrong conclusion (e.g. early cholesterol research, the MMR vaccine panic). Write a summary.
  • 5Interview a family member or friend about a statistic they recently heard. Help them evaluate whether the claim is trustworthy.
53

Mean, Median, and Mode in Context

Choose and justify the best measure of centre for different data sets.

The weekly wages of 6 employees are: $800, $820, $850, $900, $920, $5,500. Calculate the mean, median, and mode.

Which measure best represents the 'typical' wage? Why does the mean give a misleading picture here?

A real estate agent reports the 'average' house price in a suburb as $1.2 million. Do you think this is a mean or median? Which would you prefer to know as a home buyer?

54

Margin of Error and Confidence

Interpret margins of error in survey results.

A poll reports: '48% of voters support the policy, with a margin of error of ±3%.' Write the confidence interval this implies.

Another candidate receives 46% support. Can you conclude the first candidate is ahead? Explain.

Explain how increasing the sample size affects the margin of error. What happens to the margin of error if you quadruple the sample size?

Why do political polls sometimes give very different results from the actual election outcome? List three possible reasons.

55

Statistical Terms — Match to Definition

Match each statistical term to its definition.

Population
Sample
Parameter
Statistic
Census
Sampling frame
A measurement from a sample used to estimate a parameter
A measurement describing the entire population
The list from which a sample is drawn
A subset selected from the population
Collecting data from every member of the population
The entire group being studied