Statistics for Reading the News
Read any headline number, poll or study result and judge in minutes how much of it you should believe.
This course is for members
$9 unlocks every course in the library, this one included, plus a credit to build a topic of your own and at least 10 new courses every month.
What you will learn about Statistics for Reading the News
The 7-day sprint
- Every Headline Number Is an Estimate. Most numbers in the news come from a sample rather than a full count, so each one carries a range of uncertainty that the headline almost never shows.
- Why Good Numbers Change Later. Revisions are estimates being updated as more information arrives, not proof that the first number was a lie.
- What a Margin of Error Really Covers. Margin of sampling error is expressed in percentage points at a stated confidence level, it applies only to probability based samples, and it covers just one of the many ways a poll can be wrong.
- Who Never Answered, and How the Gap Was Patched. Modern poll error comes mostly from who is missing and how the pollster adjusts for them, not from the arithmetic of sample size.
- Relative Risk, Absolute Risk and Base Rates. A percentage change in risk tells you nothing until you know the risk you started from.
- Significant Does Not Mean Important. A p-value measures how incompatible the data are with a specified statistical model, not whether a finding is true, large or worth acting on.
- Judge the Producer, Then Run the Checklist. Trust in a number depends on the transparency and systems of whoever produced it, and both official statistics and polling have published standards you can hold them to.
The 80/20: the 7 concepts that matter most
- The number has a range, and the range often includes no change
- Margin of sampling error is narrow in scope and written in points
- Who is missing beats how many answered
- Relative change without a base rate is theater
- Statistical significance is not truth, size or importance
- First prints get revised, and that is the system working
- Judge the producer's transparency when you cannot judge the method
The core idea, explained like you are 5
A number in the news is almost always a measurement taken from a small piece of the world, so it comes with a range around it, and reading it well means asking how that piece was chosen and how wide the range is.
Then the quiz and the ladder
10 questions that correct you when you are wrong, and beginner, intermediate, advanced levels, each with a practical exercise.
Created Sep 13, 2026. No mistakes have been reported by learners.
Written from these sources
- Employment Situation Technical Note - 2026 M07 Results, bls.gov. day 1 sampling error, standard error, confidence interval figures.
- Employment Situation News Release - 2026 M08 Results, bls.gov. live jobs release and reliability language.
- How and why the Bureau of Labor Statistics revises job numbers, journalistsresource.org. day 2 average revision size and interval vintages.
- What Payroll Revisions Reveal about the Economy | The Daily Economy, thedailyeconomy.org. day 2 revision mechanics and day 5 base rates.
- The Jobs Report Is an Estimate, Not a Verdict: News Article - Independent Institute, independent.org. July 2026 payroll figures and revision example.
- 1 Margin of Sampling Error/Credibility Interval, aapor.org. day 3 margin scope, percentage points, design effects.
- Margin of Sampling Error/Credibility Interval - AAPOR, aapor.org. day 3 worked example and horse race ranking caution.
- Disclosure Standards - AAPOR, aapor.org. day 7 what a poll writeup must disclose.
- Best Practices for Survey Research - AAPOR, aapor.org. weighting rationale and probability sample background.
- Standard Definitions - AAPOR, aapor.org. total survey error vocabulary, coverage and nonresponse.
- Understanding a “credibility interval” and how it differs from the “margin of sampling error” in a public opinion poll - AAPOR, aapor.org. credibility intervals for opt in polls.
- What Low Response Rates Mean for Telephone Surveys | Pew Research Center, pewresearch.org. weighting, benchmarks, education weighting in 2016.
- Assessing the Representativeness of Public Opinion Surveys | Pew Research Center, pewresearch.org. civic engagement bias among survey participants.
- Methodology | Pew Research Center, pewresearch.org. day 4 methodology box and two response rates.
- No Easy Fix for Bogus Respondents in Online Opt-In Polls | Pew Research Center, pewresearch.org. bogus and AI respondents in opt in polls.
- Disentangling Bias and Variance in Election Polls, sites.stat.columbia.edu. bias versus variance, nonresponse, likely voter screens.
- Full article: The ASA Statement on p-Values: Context, Process, and Purpose, tandfonline.com. day 6 ASA statement on p-values.
- AMERICAN STATISTICAL ASSOCIATION RELEASES STATEMENT ON, amstat.org. first two principles, file drawer, p-hacking.
- The ASA's Statement on p-Values: Context, Process, and ..., stat.berkeley.edu. scope of the ASA consensus principles.
- CrossFit | The ASA Statement on P-Values: Context, Process, and Purpose, crossfit.com. thresholds and p-value as start of analysis.
- About the Code, code.statisticsauthority.gov.uk. day 7 three pillars and cross cutting themes.
- Code of Practice for Statistics Ensuring official statistics serve the public, code.statisticsauthority.gov.uk. statutory background and definition of Value.
- Official Statistics | Northern Ireland Statistics and Research Agency, nisra.gov.uk. plain language pillars and legal definition.
- Our regulatory approach: how we apply the Code of Practice for Statistics – Office for Statistics Regulation, osr.statisticsauthority.gov.uk. voluntary application of the Code by non official producers.
- Absolute and relative risk, iwh.on.ca. day 5 absolute versus relative risk definitions and example.
- Reporting the findings: Absolute vs relative risk |, healthnewsreview.org. missing absolute numbers and mismatched framing.
- Reporting the findings: Absolute vs. relative risk | USC Center for Health Journalism, centerforhealthjournalism.org. mismatched framing in journals and news releases.
- A look at national polling accuracy | Pew Research Center, pewresearch.org. question order and social desirability measurement error.