IB Maths AI HL flashcards: statistics and probability
Free statistics and probability flashcards for IB Maths Applications and Interpretation HL: 81 cards on Confidence intervals, Central limit theorem, Poisson distribution and the central limit theorem, Linear Transformations of Random Variables and more. Click the card to reveal the answer, then say whether you got it right; cards you miss come round again.
Log in to save your answers and pick up where you left off.
All cards
Prefer to read through the cards? Open the list to see every question and answer, with links to the matching revision section and practice questions.
Statistics and probability: show all cards
-
Which method?
A random sample of values is taken from a normal population. The value of is not known; only the standard deviation of the values is available. Should the confidence interval for the mean use the normal () distribution or the -distribution?
Use the -distribution. is unknown, so must be used in its place, and this needs the -distribution even though the population is normal.
Revise: FormulasPractise: Confidence intervals -
True or false
True or false: using the same sample, a 99% confidence interval for the mean is narrower than the 95% confidence interval.
False. A higher confidence level needs a larger critical value, so the 99% interval is wider.
Revise: FormulasPractise: Confidence intervals -
Spot the error
A 90% confidence interval for the mean journey time is hours. A student says: "90% of all individual journey times lie between and hours." What is wrong?
The interval is an estimate for the mean , not for individual values. The 90% means that about 90% of intervals built from repeated samples would contain the true mean.
Revise: Key termsPractise: Confidence intervals -
True or false
True or false: with and the confidence level unchanged, doubling the sample size halves the width of a confidence interval for the mean.
False. The width is proportional to , so doubling divides it by . To halve the width, must be multiplied by .
Revise: Key termsPractise: Confidence intervals -
Which formula?
A non-normal population has mean and variance . A random sample of size is taken. State the approximate distribution of the sample mean .
, by the central limit theorem since .
Revise: FormulasPractise: Central limit theorem -
True or false
True or false: if the population is normally distributed, the sample mean is normally distributed only when .
False. If the population is normal, is normal for any sample size. The condition is needed only when the population is not normal.
Revise: Key termsPractise: Central limit theorem -
Spot the error
A student writes . What is wrong?
The variance of is . It is the standard deviation that is .
Revise: FormulasPractise: Central limit theorem -
True or false
True or false: with unchanged, increasing the sample size from to makes the standard deviation of one third of its previous value.
True. The standard deviation of is , and is times , so the standard deviation is divided by .
Revise: Key termsPractise: Central limit theorem -
Criteria
The number of emails received per day has a mean of and a variance of in a large data set. Does this support a Poisson model?
No. A Poisson variable has mean equal to variance, , but here and are far apart.
Revise: Key termsPractise: Poisson distribution and the central limit theorem -
Which formula?
. Write an exact expression for .
.
Revise: FormulasPractise: Poisson distribution and the central limit theorem -
Identify
and are independent. State the distribution of .
. The means of independent Poisson variables add.
Revise: FormulasPractise: Poisson distribution and the central limit theorem -
Identify
Each of items is independently defective with probability . Which distribution (normal, binomial or Poisson) models the number of defective items, and why?
Binomial, : there is a fixed number of independent trials, two outcomes for each, and a constant probability of a defect.
Revise: FormulasPractise: Poisson distribution and the central limit theorem -
Which formula?
and . Find and .
. .
Revise: FormulasPractise: Linear Transformations of Random Variables -
Spot the error
. A student writes . What is wrong?
A variance can never be negative. Subtracting a constant shifts every value equally and does not change the spread, so .
Revise: Key termsPractise: Linear Transformations of Random Variables -
Which formula?
and are independent with and . Find .
. Variances add, even with a subtraction.
Revise: Key termsPractise: Linear Transformations of Random Variables -
True or false
True or false: if and are independent copies of , then .
False. , but . For these are and .
Revise: Key termsPractise: Linear Transformations of Random Variables -
Interpret
: a new paint has the same mean drying time as the old paint. Describe a Type I error in this context.
Concluding that the new paint has a different mean drying time when in fact it is the same as the old paint.
Revise: Key termsPractise: Errors in hypothesis tests and confidence intervals -
Identify
A coin is tossed times to test against , where is the probability of heads. is rejected if no heads occur. Identify the probability of a Type I error.
It is when is true: .
Revise: Key termsPractise: Errors in hypothesis tests and confidence intervals -
Interpret
A test of whether the mean time to finish a puzzle has increased gives a -value of at the 5% significance level. State the conclusion in context.
, so reject . There is sufficient evidence at the 5% level that the mean time to finish the puzzle has increased.
Revise: Key termsPractise: Errors in hypothesis tests and confidence intervals -
True or false
True or false: with the same sample size, lowering the significance level from 5% to 1% makes a Type II error more likely.
True. A smaller significance level makes the critical region smaller, so is rejected less often, including when it is false. This raises the probability of a Type II error.
Revise: Key termsPractise: Errors in hypothesis tests and confidence intervals -
Criteria
Is a valid transition matrix? Give the reason.
No. Each column must sum to . The first column sums to .
Revise: FormulasPractise: Markov chains -
Identify
In a transition diagram, state A has an arrow to B labelled , an arrow to C labelled , and a loop back to A. What is the probability on the loop?
, because the probabilities leaving each state sum to .
Revise: Key termsPractise: Markov chains -
Which formula?
A Markov chain has transition matrix and initial state vector . Write the expression for the state vector after transitions.
. The matrix is raised to the power and multiplies from the left.
Revise: FormulasPractise: Markov chains -
Definition
State the Markov property.
The probability of the next state depends only on the current state, not on the earlier history of how the system reached it.
Revise: Key termsPractise: Markov chains -
Definition
What equation, and what extra condition, define the steady-state vector of a transition matrix ?
, with the entries of summing to .
Revise: FormulasPractise: Steady state and equilibrium -
Criteria
Is the matrix of a regular Markov chain? Give the reason.
No. Its powers alternate between and the identity matrix, so every power contains a zero entry and none has all entries positive.
Revise: Key termsPractise: Steady state and equilibrium -
Interpret
The steady state of a regular chain on states A, B, C is . Interpret the entry .
In the long run the system is in state B with probability , that is about half of the time.
Revise: Key termsPractise: Steady state and equilibrium -
True or false
True or false: for a regular Markov chain, the steady state depends on the initial state.
False. A regular chain settles to the same steady state whatever the initial state.
Revise: Key termsPractise: Steady state and equilibrium -
Which formula?
Write the inverse of , and state when it exists.
. It exists only when .
Revise: FormulasPractise: Solving systems with matrices -
Criteria
For which value of does have no inverse?
The determinant must be zero: , so .
Revise: FormulasPractise: Solving systems with matrices -
First step
To solve and using matrices, what is the first step?
Write the system as : .
Revise: Key termsPractise: Solving systems with matrices -
Spot the error
A student solves by writing . What is wrong?
The inverse must multiply on the left: . Matrix multiplication is not commutative, and is not even defined when is a column.
Revise: Key termsPractise: Solving systems with matrices -
Identify
The number of hours of daylight in a city is recorded each day for two years. Which type of model (linear, quadratic, exponential or trigonometric) suits the data?
A trigonometric (sine) model, because the pattern repeats in a regular cycle every year.
Revise: FormulasPractise: Non-linear regression -
Definition
What does the coefficient of determination measure?
The proportion of the variation in that is explained by the model. A value close to means the model fits the data closely.
Revise: Key termsPractise: Non-linear regression -
Spot the error
A quadratic model is fitted to a child's height for ages to . A student uses it to predict the height at age . What is wrong?
Age is far outside the data range, so this is extrapolation and the prediction is unreliable. The quadratic is likely to behave unrealistically there.
Revise: Key termsPractise: Non-linear regression -
Which formula?
A linear regression model has , and the scatter diagram shows a negative trend. What is the value of Pearson's correlation coefficient ?
. For a linear model , so , and the negative trend gives the minus sign.
Revise: Key termsPractise: Non-linear regression -
Identify
and a random sample of values is taken. State the distribution of the sample mean .
.
Revise: FormulasPractise: Inverse Normal and Parameters -
Interpret
A regression line is and a data point is . Find the residual and say where the point lies relative to the line.
Predicted . Residual actual predicted . It is positive, so the point lies above the line.
Revise: Key termsPractise: Linear Regression -
Criteria
What does a -test assume about the distribution of the data?
That the data comes from a population that is (approximately) normally distributed.
-
Which method?
Two separate classes sit the same test and you want to know whether their mean scores differ. Which hypothesis test is appropriate?
An unpaired (two-sample) -test: the two groups are independent, the data is numerical and the means are being compared.
-
Identify
A coach claims a new training plan increases mean sprint speed. Speeds of one group using the plan and a separate group not using it are compared, with means and respectively. Identify the correct null and alternative hypotheses.
and . The claim is an increase in one direction only, so the test is one-tailed.
-
Spot the error
Comparing the mean scores of two independent groups, a pupil obtains at the significance level and writes: "We accept , so the two means are definitely equal." What is wrong?
Since , the correct conclusion is "do not reject ". This means there is insufficient evidence of a difference between the means; it does not prove that they are equal.
-
Which method?
You record the favourite sport (football, tennis or swimming) of pupils and whether each pupil is in Year 10 or Year 11. You want to know whether sport preference is linked to year group. Which test should you use?
A chi-squared test for independence on a contingency table, because both variables are categorical and you are testing for an association.
Revise: Key termsPractise: Chi-squared Test -
Criteria
What condition on the expected frequencies must hold for a test to be valid?
Every expected frequency should be greater than .
Revise: FormulasPractise: Chi-squared Test -
Interpret
In a test for independence, the test statistic is greater than the critical value. What do you conclude?
Reject : there is sufficient evidence, at that significance level, of an association between the two variables.
Revise: Key termsPractise: Chi-squared Test -
Spot the error
A pupil writes : "Gender and choice of snack are associated." What is wrong?
The null hypothesis states there is no association (the variables are independent). The claim of an association belongs in .
Revise: Key termsPractise: Chi-squared Test -
Which formula?
A frequency table gives with frequency , with frequency and with frequency . Which formula gives the mean, and what is it?
.
Revise: FormulasPractise: Measures of Central Tendency -
Interpret
Why is a mean calculated from grouped data only an estimate?
The individual values are not known, so each class is represented by its midpoint, as if every value in the class lay at the midpoint.
Revise: Key termsPractise: Measures of Central Tendency -
Identify
Which measure of central tendency (mean, median or mode) can be used to describe data on favourite colour?
The mode. Colours cannot be added or ordered, so only the most frequent value makes sense.
Revise: Key termsPractise: Measures of Central Tendency -
First step
Four numbers , , and have a mean of . What is the first step to find ?
Find the total from the mean: . (Then gives .)
Revise: Worked examplesPractise: Measures of Central Tendency -
Which formula?
A data set has mean and standard deviation . Every value is multiplied by and then is subtracted. Which rules give the new mean and standard deviation, and what are they?
Both are multiplied by , but only the mean is changed by subtracting . New mean . New standard deviation , because subtracting does not change the spread.
Revise: FormulasPractise: Measures of Spread -
Interpret
Two classes have the same mean test score. Class A has standard deviation and class B has standard deviation . Which class has more consistent scores, and why?
Class A. A smaller standard deviation means the scores are clustered closer to the mean.
Revise: Key termsPractise: Measures of Spread -
Interpret
Why is the interquartile range less affected by an outlier than the range?
The range uses the minimum and maximum, so one extreme value changes it directly. The interquartile range measures only the middle of the data, so extreme values barely affect it.
Revise: FormulasPractise: Measures of Spread -
Interpret
A data set has a standard deviation of . What does this tell you about the data?
There is no spread at all: every value in the data set is the same, equal to the mean.
Revise: Key termsPractise: Measures of Spread -
Identify
On a cumulative frequency graph, which values are used as the -coordinates of the plotted points?
The upper boundaries of the classes, each plotted against the running total (cumulative frequency) up to that boundary.
Revise: Key termsPractise: Statistical Diagrams -
Interpret
In a frequency histogram with equal class widths, one bar is twice as tall as another. What does this tell you?
Its class contains twice as many values. With equal widths, the height of a bar is proportional to the frequency.
Revise: Key termsPractise: Statistical Diagrams -
Interpret
A data set has values. Roughly how many of them lie inside the box of its box and whisker diagram?
About . The box runs from to , which contains the middle of the data.
Revise: FormulasPractise: Statistical Diagrams -
Identify
How is an outlier shown on a box and whisker diagram, and where does the whisker end?
An outlier is marked separately as a cross. The whisker ends at the most extreme value that is not an outlier.
Revise: FormulasPractise: Statistical Diagrams -
Which method?
A random variable has for . Which fact do you use to find , and what is ?
The probabilities sum to : , so .
Revise: Key termsPractise: Discrete Random Variables -
Criteria
In a game you pay an entry fee, and you receive dollars with probability and nothing otherwise. What fee makes the game fair, and why?
A fair game has expected gain . The expected prize is dollars, so the fee is dollars.
Revise: Key termsPractise: Discrete Random Variables -
Interpret
is the gain in dollars from one play of a game, and . What does this mean over plays?
On average you lose dollars per play, so over plays you expect to lose about dollars. The game favours the organiser.
Revise: Key termsPractise: Discrete Random Variables -
Spot the error
takes the values , , with probabilities , , . A pupil finds . What is wrong?
The values must be weighted by their probabilities: .
Revise: FormulasPractise: Discrete Random Variables -
Identify
A pupil draws cards from a standard deck without replacement and counts the hearts. Is this number binomially distributed? Why or why not?
No. Without replacement the probability of a heart changes after each card, so the trials are not independent and the probability of success is not constant.
Revise: FormulasPractise: Binomial Distribution -
Which formula?
. Which formulas give the mean and variance of , and what are they?
and .
Revise: FormulasPractise: Binomial Distribution -
First step
. How do you write in terms of a cumulative probability of the form ?
.
Revise: Key termsPractise: Binomial Distribution -
Identify
A machine makes faulty items of the time, independently. items are checked and is the number that are faulty. State the distribution of and its mean.
. The mean is .
Revise: Key termsPractise: Binomial Distribution -
Which method?
. Which rule estimates without a normal probability calculation, and roughly what is it?
The , , rule. is standard deviations above the mean. About lies within standard deviations, so the two tails share , and the upper tail is about , or .
Revise: FormulasPractise: Normal Distribution -
Which method?
and . Which property of the normal curve finds the value with , and what is ?
Symmetry about the mean. is below , so the matching value is above: .
Revise: Key termsPractise: Normal Distribution -
True or false
True or false: for a normally distributed , , so .
True. A normal variable is continuous, so the probability of any single exact value is (the area under one point is zero).
Revise: Key termsPractise: Normal Distribution -
Identify
For you are told . Is above or below the mean, and why?
Below the mean. Since of the values are greater than , more than half lie above it, so .
Revise: Key termsPractise: Inverse Normal and Parameters -
Which method?
with and . What is , and how can you find it without any calculation?
. The two tails have equal area, so and are the same distance from the mean, and the mean is their midpoint.
Revise: Key termsPractise: Inverse Normal and Parameters -
True or false
True or false: to take a simple random sample you need a complete list of the whole population.
True. Every member must have an equal chance of being chosen, so you must be able to select from the whole population (a complete list, or sampling frame). This is its main practical disadvantage: such a list can be hard or costly to obtain.
Revise: Key termsPractise: Sampling Methods -
Identify
A council puts a transport survey online and counts only the people who choose to reply. Why might the results be biased?
The sample is self-selected. People with strong views are more likely to reply, so the respondents do not represent the whole population.
Revise: Key termsPractise: Sampling Methods -
Identify
A teacher picks pupils at random from her own class of to estimate the mean height of all pupils in the school. Identify the population and the sample, and give one source of bias.
The population is all pupils in the school. The sample is the chosen pupils. Bias: they all come from one class, so pupils from other years are not represented.
Revise: Key termsPractise: Sampling Methods -
Spot the error
A pupil says a systematic sample, taking every th name from a list after a random start, is a simple random sample because the start is random. What is wrong?
In a simple random sample every possible group of that size is equally likely. In a systematic sample names next to each other can never both be chosen, so it is not a simple random sample.
Revise: Key termsPractise: Sampling Methods -
Interpret
A regression line is , where is the hours of revision and is the test mark. Interpret and .
is the gradient: each extra hour of revision is associated with a predicted rise of marks. is the predicted mark with no revision (), meaningful only if is near the data.
Revise: Key termsPractise: Linear Regression -
Criteria
State two things that make a prediction from a regression line more reliable.
Any two of: the -value lies inside the range of the data (interpolation, not extrapolation); the correlation is strong ( close to ); the sample is large; the line of on is used to predict from .
Revise: Key termsPractise: Linear Regression -
Spot the error
. A pupil says about of values lie between and , because the standard deviation is . What is wrong?
In the second number is the variance, so and . About lie within one standard deviation of the mean, between and .
Revise: Key termsPractise: Normal Distribution -
True or false
and about of values lie between and . True or false: .
False. About of values lie within standard deviations of the mean, so , giving . The distance from the mean is , not .
Revise: FormulasPractise: Normal Distribution -
True or false
True or false: for , the value with equals , whatever the value of .
True. The curve is symmetric about , so exactly half of the area lies to the left of . Changing changes the spread of the curve, not its centre.
Revise: Key termsPractise: Inverse Normal and Parameters -
Which method?
A regression line is fitted to data with mean and mean . Which fact lets you find without seeing any data points?
The regression line of on always passes through the mean point , so substitute and into and solve for .
Revise: Key termsPractise: Linear Regression
Other AI HL areas
Quick questions
Are the IB Maths AI HL statistics and probability flashcards free?
Yes. Every flashcard on IB Maths Help is free to use, with no account needed. If you log in, your answers are saved to your account, so you can pick up where you left off and skip the cards you already know.
How do the flashcards work?
Read the question, click the card to reveal the answer, then say whether you got it right. Cards you miss come round again until you have them all.
What kinds of card are there?
Definitions, criteria (what must be true for something to apply), true or false, identify, which formula, which method, spot the error, interpret and first step. They test one quick idea at a time and need no calculator.
Which topics are in the AI HL statistics and probability flashcards?
This deck covers Confidence intervals, Central limit theorem, Poisson distribution and the central limit theorem, Linear Transformations of Random Variables, Errors in hypothesis tests and confidence intervals, Markov chains, Steady state and equilibrium, Solving systems with matrices, Non-linear regression, Inverse Normal and Parameters, Linear Regression, t-test, Chi-squared Test, Measures of Central Tendency, Measures of Spread, Statistical Diagrams, Discrete Random Variables, Binomial Distribution, Normal Distribution, Sampling Methods. The other AI HL areas each have their own deck.
How are flashcards different from practice questions?
Flashcards test one quick idea, such as a definition, a formula or which method to choose. Practice questions are full IB-style questions with worked solutions. Each flashcard links to the matching practice questions.