YOUR OWN QUIZ STUDIO
Choose the numbers. Build the question.
Make a quiz from your own counts or data. See why its answer works, then change one thing and try again.
Everything you enter here is calculated in this tab. Nothing is sent to a tutor. The quiz resets when you refresh or close the page. Use Download quiz to keep a plain text copy.
Start making a quizA RECIPE YOU CAN REUSE
Build a valid question in four moves.
- Choose the skill. Ask one clear thing: a probability, a summary or a prediction.
- Write the setup. Include all counts or observations. State the group, replacement rule, units and rounding when they matter.
- Solve it before using it. Keep an exact fraction or full stored precision until the last step. Check that the answer fits the setup.
- Make a second version. Change one count, value or condition. Predict what should change, then rebuild the answer key.
Use the worked key to check your method. Reading help is practice; this page does not award an unaided score or mark a lesson mastered.
1. CHOOSE AND EDIT
Your setup controls the answer.
2. ASSEMBLE YOUR QUIZ
Keep the versions you want to try.
Preview a question, then add it here. Editing the setup does not change questions already added. Each keeps its own data and worked key. Make up to 20 questions.
No questions added yet.
3. CATCH THE TRAP
Check the setup before the arithmetic.
Can these three probability givens coexist?
P(A) = 0.1, P(B) = 0.3, P(A given B) = 0.5.
No. The last two imply P(A and B) = 0.3 × 0.5 = 0.15. A shared part cannot be larger than all of A, which is only 0.1. For your own quiz, start with actual table counts and derive the probabilities from them.
Why do reversed given questions have different answers?
With Wavy Blonde = 5, Wavy Black = 3, Straight Blonde = 2 and Straight Black = 6, P(Wavy given Blonde) = 5/7, while P(Blonde given Wavy) = 5/8. The same shared cell sits over a different whole group. Try adding these two questions to your quiz.
Can a grouped table answer an exact raw data question?
A class from 36 through 41 does not tell you how many values exceed 40. Use the raw observations when the cutoff splits a class. In the 25 value example, 17 values exceed 40, so 17/25 = 68%. A grouped mean uses midpoints and is an estimate, not the exact raw mean.
Values: 30, 32, 34, 36, 36, 37, 39, 39, 41, 41, 42, 42, 43, 43, 44, 45, 45, 45, 46, 47, 47, 49, 49, 52, 53.
Raw mean = 42.28. Classes 30 through 35, 36 through 41, 42 through 47 and 48 through 53 have frequencies 3, 7, 11 and 4. Their midpoint estimate is 42.34. These answer different questions.
Does expected count guarantee the next result?
No. If p = 0.3 and n = 20 trials, np = 6 is an average expected count across repetitions, not a promise of exactly six successes. To estimate chance from cumulative results, divide total successes by total trials. Do not average batch percentages when batch sizes differ.
How do I design a fair simulation?
Define equally likely random labels, map each label to an outcome, and state what counts as one complete trial. For a jar with 3 red, 4 green and 2 white balls, label the balls 1 through 9. For two draws without replacement, reject a repeat of the same ball label, not every repeat of the same color. Reset the full jar for the next trial. A finite run estimates probability; it does not guarantee the theoretical fraction.
Work through the full probability examples · Review standard deviation · Review regression