collect several random samples from the same population, show their results on a dot plot, and use the spread of the results to judge how far an estimate might be off.
Phase 2: Investigate
Do Now. Yesterday different teams drew 20 cards from the same kind of bag and got different estimates. Was someone careless, or is that expected? Write one sentence.
If 10 teams each draw a random sample of 20 cards, how will their percents of ring toss cards look?
The reading opens over the lesson; close it to come back to exactly where you were.
Sampling variability. Two random samples from the same population almost never match exactly. That is not a mistake; it is chance. Our reading calls it sampling variance. The useful question is: how much do the samples vary?
A dot plot of samples. If each team records its sample percent as one dot, the class dot plot shows the pattern. Most dots bunch near the true value, and a few land farther away. The middle of the dots is a better estimate than any single dot.
Worked example 1: ten samples. Sample data: ten teams each drew 20 cards from a bag and counted red cards: 5, 6, 7, 6, 4, 8, 6, 5, 7, 6. Step 1: change each count to a percent of 20; each card is 5%, so the percents are 25, 30, 35, 30, 20, 40, 30, 25, 35, 30. Step 2: the lowest is 20% and the highest is 40%. Step 3: the mean count is 60 ÷ 10 = 6 red cards, which is 30%. Step 4: the most common result (the mode) is also 30%.
Worked example 2: how far off? Sample data: the same bag holds 200 cards. Malik uses the class results to estimate the number of red cards. Step 1: the class mean, 30% of 200, is 60 cards. Step 2: the lowest sample, 20%, gives 40 cards; the highest, 40%, gives 80 cards. Step 3: so one sample of 20 could miss by as many as 20 cards, but the middle of all ten is a strong estimate: about 60.
What the teacher reveals. When we open the bag, we can see whether the dots really bunched around the true value. In most classes they do, and that is the reason random sampling works.
Reporting a range. Survey scientists often report an estimate with a range, like “about 30%, likely between 20% and 40% for one sample this size.” We will name this idea in lesson 2.10.
Hands-on. Each team draws a random sample of 20 cards from a class bag of 200, counts the red cards, returns them and shakes the bag. Post your percent as a dot on the class dot plot (0% to 60%, marked every 5%). Use Worked example 1’s data if your class is small.
Reason about sampling variability.
Sample data: five random samples of 20 students each show 9, 11, 10, 12 and 8 students who want a Friday carnival. Change each to a percent, find the mean percent, and estimate the number for a school of 480.
NAEP says another equally good sample would give slightly different results. What do statisticians call this?
Write a Wonder Board answer: why can a survey of one sample be a little off even when it is random?
Samples vary, but they cluster. Next time we test whether bigger samples cluster more tightly.
Exit ticket: predict what our dot plot would look like if every team drew 40 cards instead of 20.
Pacing: Do Now 4 min, reading 7, mini lesson 10, sampling and dot plot 14, independent practice and exit ticket 5.
Materials: one class bag of 200 cards, 60 of them red (30%), or re-use the lesson 2.05 bags. Reveal the true count (60) at the end of the lesson.
Accuracy: results as estimates from representative samples, each student answering only some questions, a different equally good sample giving slightly different results, sampling variance, measurement variance and standard error are from the NAEP page opened 2026-11-20. Standard error is named only; it is not computed in grade 7. Checked: counts sum to 60, mean 6 = 30%; 7/20 = 35%; 20% and 40% of 200 = 40 and 80; independent counts 9, 11, 10, 12, 8 = 45%, 55%, 50%, 60%, 40%, mean 50%, 50% of 480 = 240.
Likely misconceptions: believing one random sample should equal the truth; thinking variation means random samples are useless.
Supports: a pre-drawn dot plot axis with 5% marks; percent strips (1 of 20 = 5%). Stretch: what percent would one red card change a sample of 40? (2.5%.)
Answer guide: exit ticket: with 40 cards the dots should bunch more tightly around 30%. The Wonder Board answer should say chance decides who is in the sample, so the result can be a little high or low.
National Center for Education Statistics, NAEP: Statistical Significance and Sample Size · U.S. government work, public domain
NAEP reports average scores and percents for students across the country. Those numbers are estimates, the page explains, because they come from representative samples of students, not from every student. Each student also answers only some of the test questions.
Because NAEP uses a sample, its results have some uncertainty. The page says a different, equally good sample of students could have been selected, and its results would be slightly different. Statisticians call this part of the uncertainty sampling variance. It comes only from which students happened to be chosen.
A second kind of uncertainty comes from measuring. A student might answer a little differently on another day, or on a different version of the test with different but equally good questions.
NAEP shows how much uncertainty there is with a number called the standard error. The more precise an estimate is, the smaller its standard error. Results from carefully built surveys can still be very precise.
Source: National Center for Education Statistics, NAEP: Statistical Significance and Sample Size (nces.ed.gov)