marlio

AP Statistics, Unit 2

Exam words: Probability, Random Variables, and Probability Distributions

The 54 words a written answer in this unit needs. For each one: what it means, a wording that loses the mark and why, and a sentence that uses it the way a full answer would.

Study these words as a deck

Flashcards, Learn, Test and Match in Cards, with your Still learning marks kept, and the booklet to print on A4 or US Letter. Sign in to open it.

The words

  1. Normal distribution

    N(μ,σ)N(\mu, \sigma)

    A symmetric, single-peaked, bell-shaped density curve fully described by its mean and standard deviation.

    Loses the mark: “the data is normal”. Real data are at most approximately Normal; say a Normal model fits, with its mean and standard deviation.

    Example: Apple weights at the orchard follow approximately a Normal distribution with mean 150 g and standard deviation 12 g.

  2. density curve

    A smooth curve with total area 1, where area over an interval gives the proportion.

    Loses the mark: “the height of the curve is the probability”. For a density curve, probability is the area under the curve over an interval, not its height.

    Example: Because the density curve has total area 1, the area of 0.11 to the right of 165 g is the proportion of apples heavier than 165 g.

  3. z-score

    z=x−μσz = \dfrac{x - \mu}{\sigma}

    How many standard deviations a value lies above (positive) or below (negative) the mean.

    Loses the mark: “points above the mean”. A z-score counts standard deviations from the mean, not raw units.

    Example: An apple weighing 174 g has a z-score of (174 - 150)/12 = 2, so it is 2 standard deviations above the mean weight.

  4. empirical rule

    Normal data: about 68%, 95%, 99.7% lie within 1, 2, 3 SDs of the mean.

    Loses the mark: “68% above the mean”. 68% lies within one standard deviation on both sides; only about 34% lies between the mean and one SD above it.

    Example: By the empirical rule, about 95% of the apples weigh between 126 g and 174 g.

  5. percentile

    The value with p percent of the data at or below it.

    Loses the mark: “scored 90 percent”. The 90th percentile means doing as well as or better than 90% of others, not answering 90% correctly.

    Example: A newborn at the 80th percentile for weight is at least as heavy as 80% of newborns of the same age.

  6. standard Normal distribution

    N(0,1)N(0, 1)

    The Normal distribution with mean 0 and standard deviation 1, the model for z-scores.

    Loses the mark: “the standard distribution”. Name it fully: the standard Normal distribution, with mean 0 and standard deviation 1.

    Example: Standardizing turns the apple weights into the standard Normal distribution, so the area above z = 2 is about 0.023.

  7. segmented bar chart

    A chart with one full-height bar per group, divided into conditional-percent segments.

    Loses the mark: “bar chart of the counts”. Bars of raw counts mix group size with association; segmented bars show percents within each group.

    Example: A segmented bar chart shows 70% of dog owners and 45% of cat owners walk daily, so pet type and walking are associated.

  8. side-by-side bar chart

    Bars for each category placed side by side within each group of a second variable.

    Loses the mark: “bars of the total counts”. Comparing groups of different sizes needs percents within each group, not totals.

    Example: A side-by-side bar chart of music preference by grade shows the percent choosing pop falling from 52% in 9th grade to 31% in 12th grade.

  9. conditional distribution

    The distribution of one categorical variable among only those in one category of the other.

    Loses the mark: “the counts are different so they are associated”. Counts depend on group size; compare conditional percents across the groups.

    Example: The conditional distributions of walking frequency differ between dog owners and cat owners, so the variables are associated in this sample.

  10. marginal distribution

    The distribution of one categorical variable alone, from a two-way table's row or column totals.

    Loses the mark: “the marginal distribution shows an association”. Totals describe one variable at a time; association needs a comparison of conditional distributions.

    Example: The marginal distribution of pet type is 55% dog owners, 35% cat owners and 10% owners of other pets.

  11. simulation

    Imitating a random process with random numbers, repeated many times, to estimate probabilities.

    Loses the mark: “give each outcome one digit”. One digit per outcome only works when the outcomes are equally likely; digits must match the probabilities.

    Example: In the simulation, digits 0 to 6 stood for a made free throw, matching the player's 70% rate.

  12. trial

    One repetition of a random process or of a simulation of it, producing one outcome.

    Loses the mark: “a trial is the whole simulation”. A trial is one repetition; a simulation runs many trials.

    Example: Each trial read five random digits, one for each free throw in a set of five.

  13. random process

    A repeatable situation with uncertain individual results but a regular pattern over many repetitions.

    Loses the mark: “random means anything can happen with no pattern”. Single results are unpredictable, but the long-run proportions follow the probabilities.

    Example: Opening a cereal box is a random process: any one box may or may not hold a prize, but about 35% of boxes do.

  14. estimated probability

    The relative frequency of an event in many simulated or observed trials, estimating its probability.

    Loses the mark: “the simulation gives the exact probability”. A simulation's relative frequency is an estimate that varies from run to run; more trials bring it closer.

    Example: The event happened in 85 of 250 trials, so the estimated probability is 85/250 = 0.34.

  15. probability

    A number from 0 to 1 giving the long-run proportion of times an outcome occurs.

    Loses the mark: “will happen exactly 3 out of 10 times”. Probability describes the long run, not a guaranteed count in a short run of trials.

    Example: The probability that a bulb is defective is 0.02, meaning about 2% of bulbs are defective in the long run.

  16. sample space

    SS

    The list or set of all possible outcomes of a random process.

    Loses the mark: “each outcome is equally likely”. Listing outcomes does not make them equally likely; check before using 1 over the number of outcomes.

    Example: For two coin flips the sample space is HH, HT, TH and TT.

  17. event

    A collection of outcomes from the sample space; add their probabilities to find its probability.

    Loses the mark: “an event is one trial”. A trial is one repetition of the process; an event is a set of outcomes.

    Example: The event 'at least one head' contains HH, HT and TH, so its probability is 3/4.

  18. complement

    P(AC)=1−P(A)P(A^C) = 1 - P(A)

    The event that A does not happen, with probability 1 minus P(A).

    Loses the mark: “the opposite outcome”. Too vague: the complement is every outcome not in A, covering all the alternatives together.

    Example: Using the complement, P(at least one defective in 5 bulbs) = 1 - (0.98)^5, about 0.096.

  19. addition rule

    P(A∪B)=P(A)+P(B)−P(A∩B)P(A \cup B) = P(A) + P(B) - P(A \cap B)

    P(A or B) equals P(A) plus P(B) minus P(A and B), avoiding double counting.

    Loses the mark: “just add the probabilities”. Adding alone double-counts the overlap unless the events are mutually exclusive.

    Example: By the addition rule, P(junior or plays a sport) = 0.30 + 0.45 - 0.12 = 0.63.

  20. mutually exclusive

    Describes two events that cannot happen at the same time, so P(A and B) = 0.

    Loses the mark: “mutually exclusive so they are independent”. Mutually exclusive events with positive probabilities are dependent: one happening rules out the other.

    Example: Rolling a 2 and rolling a 5 on one die are mutually exclusive, so P(2 or 5) = 1/6 + 1/6 = 1/3.

  21. law of large numbers

    With more repetitions, the proportion of times an outcome occurs approaches its true probability.

    Loses the mark: “law of averages”. Short runs do not balance out; the long-run proportion settles, but no outcome is ever owed.

    Example: By the law of large numbers, after 10,000 simulated flips the proportion of heads was very close to 0.5.

  22. conditional probability

    P(A∣B)=P(A∩B)P(B)P(A \mid B) = \dfrac{P(A \cap B)}{P(B)}

    The probability of A given that B occurred: P(A and B) divided by P(B).

    Loses the mark: “the probability of A and B”. P(A and B) is a joint probability; to condition on B, divide it by P(B).

    Example: The conditional probability that a student plays a sport given that they are a junior is 0.12/0.30 = 0.40.

  23. false positive

    A test result saying a condition is present when it is actually absent.

    Loses the mark: “the test was wrong”. Too vague: name the error, a positive result for someone who does not have the condition.

    Example: Because the disease is rare, most positive results from the screening are false positives.

  24. base rate

    How common a condition is in the population before any test result is known.

    Loses the mark: “a positive result means a 99% chance”. P(condition given positive) also depends on the base rate, and is low when the condition is rare.

    Example: With a base rate of 1 in 1,000, even a test that is 99% accurate gives only about a 9% chance of disease after a positive result.

  25. independent events

    Events where knowing one happened does not change the other's probability: P(A given B) = P(A).

    Loses the mark: “they do not affect each other”. Show it with probabilities: compare P(A given B) with P(A).

    Example: Since P(late given rain) = 0.20 equals P(late) = 0.20, being late and rain are independent events for this bus route.

  26. multiplication rule for independent events

    P(A∩B)=P(A) P(B)P(A \cap B) = P(A)\,P(B)

    For independent events, P(A and B) = P(A) times P(B); check independence first.

    Loses the mark: “multiply because both can happen”. Multiplying the probabilities is valid only when the events are independent.

    Example: By the multiplication rule for independent events, two separate customers both paying cash has probability 0.3 times 0.3 = 0.09.

  27. dependent events

    Events where one happening changes the probability of the other.

    Loses the mark: “dependent means they always happen together”. Dependence only means the probability shifts; the events need not occur together.

    Example: Drawing two aces from a deck without putting the first back involves dependent events, since the first ace changes the chance for the second.

  28. random variable

    XX

    A variable whose numerical value is determined by the outcome of a random process.

    Loses the mark: “a random number”. A random variable assigns a number to each outcome of a chance process, with a probability for each value.

    Example: Let X be the random variable counting the defective bulbs in a pack of 4.

  29. discrete random variable

    A random variable whose possible values can be listed, with gaps between them.

    Loses the mark: “discrete means it has few outcomes”. Discrete means the values can be listed with gaps between them, however many there are.

    Example: The number of customers waiting in line is a discrete random variable taking the values 0, 1, 2 and so on.

  30. probability distribution

    A list of a random variable's values with their probabilities, which sum to 1.

    Loses the mark: “the values add up to 1”. It is the probabilities, not the values of X, that must sum to 1.

    Example: The table is a valid probability distribution for X because each probability lies between 0 and 1 and they sum to 1.

  31. expected value

    E(X)=μX=∑xi P(xi)E(X) = \mu_X = \sum x_i\,P(x_i)

    The mean of a random variable: the sum of each value times its probability.

    Loses the mark: “expected value is the most likely outcome”. E(X) is a long-run average and need not be the value with the highest probability.

    Example: The expected value of a raffle ticket's net winnings is -$0.40, so a buyer loses 40 cents per ticket on average in the long run.

  32. long-run average

    The average result over many repetitions of a random process, settling near the expected value.

    Loses the mark: “you will win this amount”. Expected value is a long-run average, not what happens in any one play.

    Example: A four-sided die has a long-run average of 2.5 per roll: the mean over thousands of rolls is close to 2.5, though no single roll shows 2.5.

  33. fair game

    A game in which a player's expected net gain is zero.

    Loses the mark: “fair because everyone has the same chance to win”. Fair means an expected net gain of zero, not equal chances of winning.

    Example: Paying $1 to win $6 on a single die roll is a fair game, since the expected net gain is (1/6)(5) + (5/6)(-1) = 0.

  34. variance

    σX2=∑(xi−μX)2 P(xi)\sigma_X^2 = \sum (x_i - \mu_X)^2\,P(x_i) · Units: squared units of X

    For a random variable, the probability-weighted average of squared distances from the mean.

    Loses the mark: “the variance is the standard deviation”. Variance is in squared units; take its square root to get the standard deviation.

    Example: The variance of the number of defective bulbs is 0.16 squared bulbs, so the standard deviation is 0.4 bulbs.

  35. standard deviation of a random variable

    σX=σX2\sigma_X = \sqrt{\sigma_X^2}

    The square root of a random variable's variance, its typical distance from the mean.

    Loses the mark: “how far it is from the mean every time”. The standard deviation is a typical distance over many trials, not the distance on each trial.

    Example: If X is the number of calls per hour, the standard deviation of X is 1.3 calls, so hourly counts typically fall about 1.3 calls from the mean of 4.

  36. tree diagram

    A diagram of a multistage process: multiply along paths, add the paths in the event.

    Loses the mark: “add along the branches”. Along a path the stages happen in sequence, so multiply; add only across separate paths.

    Example: A tree diagram for the screening test multiplies 0.01 by 0.95 on the disease-and-positive path, then adds the false-positive path.

  37. general multiplication rule

    P(A∩B)=P(A) P(B∣A)P(A \cap B) = P(A)\,P(B \mid A)

    P(A and B) = P(A) times P(B given A), whether or not the events are independent.

    Loses the mark: “multiply the original probabilities”. After the first stage, use the conditional probability for the second stage.

    Example: By the general multiplication rule, the chance both cards drawn are hearts is (13/52)(12/51), about 0.059.

  38. sampling without replacement

    Drawing items without returning each one before the next draw, which changes later probabilities.

    Loses the mark: “same chance on the second draw”. Without replacement the second-stage probabilities change; use the conditional values.

    Example: Because we are sampling without replacement, the second branch for a red marble uses 4/9 rather than 5/10.

  39. independent random variables

    Random variables where knowing one gives no information about the other; required to add variances.

    Loses the mark: “variances add for any two variables”. The variance rule needs independence; related variables need an extra correlation term.

    Example: The two machines' fill amounts are independent random variables, so the variance of their total is the sum of their variances.

  40. variance sum law

    σX±Y2=σX2+σY2\sigma^2_{X \pm Y} = \sigma^2_X + \sigma^2_Y

    For independent random variables, Var(X + Y) and Var(X - Y) both equal Var(X) + Var(Y).

    Loses the mark: “subtract the variances”. A difference varies more, not less; the variances still add.

    Example: By the variance sum law, the SD of the difference in weights is the square root of (3^2 + 4^2), which is 5 grams.

  41. mean of a sum or difference

    μX±Y=μX±μY\mu_{X \pm Y} = \mu_X \pm \mu_Y

    The mean of a sum or difference is the sum or difference of the means.

    Loses the mark: “the means add for a difference”. For X - Y subtract the means; it is only the variances that add.

    Example: By the rule for the mean of a sum or difference, a 250 g box with a 40 g lid has a mean total weight of 290 g.

  42. binomial setting

    A fixed number n of independent trials, each success or failure with constant probability p.

    Loses the mark: “two outcomes so it is binomial”. Also check a fixed number of trials, independence and the same p on every trial.

    Example: Shooting 10 free throws with a 0.7 chance each, independently, is a binomial setting with n = 10 and p = 0.7.

  43. binomial distribution

    P(X=k)=(nk)pk(1−p)n−kP(X = k) = \binom{n}{k}p^k(1-p)^{n-k}, μX=np\mu_X = np, σX=np(1−p)\sigma_X = \sqrt{np(1-p)}

    The distribution of the number of successes X in a binomial setting, with mean np.

    Loses the mark: “standard deviation is np(1 - p)”. np(1 - p) is the variance; take its square root for the standard deviation.

    Example: With a binomial distribution for n = 10 and p = 0.7, the mean number of made free throws is 7.

  44. binomial coefficient

    (nk)=n!k! (n−k)!\binom{n}{k} = \dfrac{n!}{k!\,(n-k)!}

    The number of ways to choose k successes from n trials, read n choose k.

    Loses the mark: “n times k”. The number of arrangements is n choose k, computed with factorials, not a product.

    Example: The binomial coefficient 10 choose 7 = 120 counts the orders in which 7 of 10 shots can be made.

  45. cumulative probability

    The probability that a random variable is at most (or at least) some value.

    Loses the mark: “binompdf for at least”. The pdf gives one exact value; at least or at most needs a cumulative sum.

    Example: The cumulative probability of making at most 5 of the 10 shots is P(X ≤ 5) = 0.150.

  46. 10% condition

    n≤0.10Nn \le 0.10N

    Sample at most 10% of the population so observations can be treated as independent.

    Loses the mark: “the 10% condition shows the sample is random”. It only lets us treat draws as independent; randomness comes from how the sample was chosen.

    Example: The 10% condition holds because 50 sampled bulbs is less than 10% of the 5,000 bulbs in the shipment.

  47. geometric setting

    Independent trials with the same success probability p, repeated until the first success.

    Loses the mark: “binomial because we count trials”. Counting trials until the first success has no fixed n, so it is geometric, not binomial.

    Example: Rolling a die until the first 6 appears is a geometric setting with p = 1/6.

  48. geometric distribution

    P(Y=k)=(1−p)k−1pP(Y = k) = (1-p)^{k-1}p, μY=1/p\mu_Y = 1/p

    The distribution of the trial number of the first success, with mean 1/p.

    Loses the mark: “the expected number of trials is p”. On average it takes 1/p trials to reach the first success.

    Example: With a geometric distribution and p = 1/6, the expected number of rolls until the first 6 is 6.

  49. parameter

    μ\mu, pp, σ\sigma

    A number describing a population, such as its mean or proportion; fixed but usually unknown.

    Loses the mark: “p-hat is the population proportion”. p-hat comes from the sample and is a statistic; the population proportion p is the parameter.

    Example: The parameter of interest is the true proportion of all voters in the county who support the new park.

  50. statistic

    xˉ\bar{x}, p^\hat{p}, ss

    A number computed from sample data to estimate a parameter; it varies between samples.

    Loses the mark: “the statistic is the true value”. A statistic is an estimate from one sample; the true value is the parameter.

    Example: The statistic is the 58% of the 400 sampled voters who support the park.

  51. sampling variability

    The natural variation of a statistic from one random sample to another.

    Loses the mark: “the sample was done wrong”. Differences between random samples are expected from chance alone, not from a mistake.

    Example: Two random samples gave 58% and 54% support; sampling variability easily explains a difference that size.

  52. sampling distribution

    The distribution of a statistic over all possible random samples of the same size.

    Loses the mark: “the distribution of the sample”. That is the data in one sample; a sampling distribution shows a statistic across many samples.

    Example: The sampling distribution of p-hat for samples of 400 voters is centered at the true proportion with standard deviation about 0.025.

  53. unbiased estimator

    A statistic whose sampling distribution has its mean equal to the parameter it estimates.

    Loses the mark: “unbiased because the spread is small”. Small spread means low variability; unbiased means the sampling distribution is centered on the parameter.

    Example: The sample proportion is an unbiased estimator of p, because the mean of its sampling distribution equals p.

  54. population distribution

    The distribution of a variable's values for every individual in the population.

    Loses the mark: “the population becomes normal”. Larger samples change the sampling distribution of the mean, not the shape of the population.

    Example: The population distribution of household sizes is skewed right, yet x-bar for samples of 50 households is approximately Normal.