marlio

AP Statistics, Unit 1

Exam words: Exploring One-Variable Data and Collecting Data

The 79 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. statistical question

    A question answered with data that vary from one individual to the next.

    Loses the mark: “it has a number for an answer so it is statistical”. A question with one fixed numerical answer, like the height of one door, is not statistical; the answers must vary.

    Example: How long students at our school sleep on a school night is a statistical question, because the answers differ from student to student.

  2. statistical study

    An investigation using a sample's statistic to answer a question about a population.

    Loses the mark: “the sample result is the exact population value”. A sample's result is an estimate; another random sample would give a slightly different value.

    Example: In this statistical study, 120 of the school's 1,850 students were sampled, and the percent who walk estimates the percent for the whole school.

  3. individuals

    The people, animals, objects or events a data set describes, one per row.

    Loses the mark: “the individuals are the measurements”. Measurements are variables; the individuals are who or what was measured.

    Example: In the vet clinic's records the individuals are the 120 pets seen in March, one row per pet.

  4. variable

    A characteristic recorded for every individual that can differ between individuals, one per column.

    Loses the mark: “the variables are the people surveyed”. The people are the individuals; a variable is something recorded about each of them.

    Example: Weight in kilograms is a variable recorded for each pet in the clinic's data.

  5. categorical variable

    A variable that sorts individuals into groups, such as eye color or ZIP code.

    Loses the mark: “quantitative because the values are numbers”. Numbers used only as labels, like ZIP codes or jersey numbers, still make a categorical variable.

    Example: Shirt size (small, medium, large) is a categorical variable, even though the sizes have an order.

  6. quantitative variable

    A variable recording a numerical measurement or count, so averaging it makes sense.

    Loses the mark: “it is a number so it is quantitative”. Having digits is not enough; the values must be measurements or counts that can be averaged meaningfully.

    Example: Commute time in minutes is a quantitative variable, so it makes sense to report its mean.

  7. discrete variable

    A quantitative variable with separate possible values, usually counts, such as number of siblings.

    Loses the mark: “discrete means categorical”. Discrete variables are quantitative counts; categorical variables sort individuals into groups.

    Example: The number of texts a student sends in an hour is a discrete variable because it can only be 0, 1, 2 and so on.

  8. continuous variable

    A quantitative variable that can take any value in an interval, such as height.

    Loses the mark: “continuous because it keeps going”. Continuous means any value in an interval is possible, not that data were collected over a long time.

    Example: A runner's finishing time is a continuous variable, since it could be 21.4 or 21.43 minutes.

  9. relative frequency

    The proportion or percent of individuals in a category: its count divided by the total.

    Loses the mark: “has more people so it is more popular”. A larger raw count can come from a larger group; compare relative frequencies instead.

    Example: Of the 250 cafeteria customers, 60 chose soup, a relative frequency of 60/250 = 0.24.

  10. bar chart

    A graph of a categorical variable with a bar per category, showing count or percent.

    Loses the mark: “histogram of the categories”. Histograms are for quantitative variables; categories get a bar chart with separated bars.

    Example: A bar chart of favorite lunch options shows pizza with the tallest bar at 38% of students.

  11. pie chart

    A circle split into slices, each category's slice proportional to its share of the whole.

    Loses the mark: “pie chart of the averages”. A pie chart shows parts of one whole, so its slices must be shares of a total, not separate averages.

    Example: The pie chart of commuting mode shows walking as a quarter of the circle, so about 25% of employees walk.

  12. two-way table

    A table counting individuals by two categorical variables, one in rows, one in columns.

    Loses the mark: “a table with two columns”. A two-way table classifies by two categorical variables; two columns of raw data are just a data table.

    Example: A two-way table of grade level by phone brand lets us compare brand preference among 9th and 12th graders.

  13. marginal proportion

    The proportion of all individuals in one row or column category of a two-way table.

    Loses the mark: “proportion of the row”. A share of one row is a conditional proportion; a marginal proportion divides by the grand total.

    Example: The marginal proportion of students who own a bike is 140/400 = 0.35.

  14. joint proportion

    The proportion of all individuals in one cell, having both its row and column categories.

    Loses the mark: “multiply the row and column totals”. A joint proportion is one cell count over the grand total, not a product of totals.

    Example: The joint proportion of students who are seniors and own a bike is 30/400 = 0.075.

  15. conditional proportion

    The proportion within one group only, using that group's total rather than the grand total.

    Loses the mark: “out of the grand total”. Dividing by the grand total gives a joint or marginal proportion; a conditional one divides by the group's own total.

    Example: Among the 120 seniors, the conditional proportion who own a bike is 30/120 = 0.25, compared with 0.40 among freshmen.

  16. frequency

    The count of individuals in a category or with a value, not a percent.

    Loses the mark: “the frequency is the percent”. A frequency is a count; dividing it by the total gives the relative frequency.

    Example: In the survey of 60 students, the frequency for walking to school is 18, a relative frequency of 30%.

  17. distribution

    The values a variable takes and how often it takes each one.

    Loses the mark: “how the data looks”. Too vague: a distribution tells which values occur and how often each occurs.

    Example: The distribution of resting heart rates for the 30 swimmers is roughly symmetric with a center near 62 beats per minute.

  18. dotplot

    A graph placing one dot per individual above its value on a number line.

    Loses the mark: “each dot is a category”. In a dotplot each dot is one individual's value on a number line.

    Example: The dotplot of hours slept shows two students at 4 hours, well below the rest of the class.

  19. stemplot

    A display splitting each value into a stem (leading digits) and a leaf (final digit).

    Loses the mark: “the leaves are the tens digit”. Leaves are the final digit; the stems carry the leading digits, and the key sets the place value.

    Example: A stemplot with key 3|5 = 35 seconds shows the reaction times clustered in the 30s.

  20. histogram

    A graph with touching bars showing the count or percent in each equal-width bin.

    Loses the mark: “each bar is one person”. Each bar counts all values in an interval, not a single data value.

    Example: The histogram of delivery times has its tallest bin at 20 to 25 minutes and a long tail out to 60 minutes.

  21. symmetric

    Describes a distribution whose left and right sides are roughly mirror images.

    Loses the mark: “evenly spread out”. Even spread is about variability; symmetry is about the two sides of the center matching.

    Example: The distribution of bolt lengths is roughly symmetric and single-peaked, centered near 5.0 cm.

  22. skewed right

    Describes a distribution with a long tail stretching toward larger values.

    Loses the mark: “skewed left because most of the data is on the left”. Skew is named for the side of the long tail, not for where most of the data sit.

    Example: Household incomes in the town are skewed right, with most near $50,000 and a long tail out past $300,000.

  23. skewed left

    Describes a distribution with a long tail stretching toward smaller values.

    Loses the mark: “skewed right because most of the data is on the right”. Skew is named for the side of the long tail, not for where most of the data sit.

    Example: Scores on the easy quiz are skewed left: most students scored 9 or 10, with a tail down to 3.

  24. bimodal

    Describes a distribution with two clear peaks, often from two groups mixed together.

    Loses the mark: “has two means”. A distribution has one mean; bimodal means it has two peaks.

    Example: Waiting times at the clinic are bimodal, with peaks near 10 and 45 minutes, suggesting walk-ins and booked patients differ.

  25. shape

    The form of a distribution: symmetric or skewed, number of peaks, any gaps or clusters.

    Loses the mark: “the graph is normal”. Describe the shape you see (symmetric or skewed, number of peaks); real data are at most approximately Normal.

    Example: In shape, the distribution of battery lives is skewed left with a single peak near 11 hours.

  26. center

    A typical value of the distribution, reported as the median or mean, with units.

    Loses the mark: “the center is in the middle”. Give the measure (median or mean), its value and its units in context.

    Example: The center of the commute-time distribution is a median of 22 minutes, compared with 15 minutes for cyclists.

  27. spread

    How much a distribution's values vary, measured by range, IQR or standard deviation.

    Loses the mark: “the data is spread out”. Vague: report a measure of spread with its value and units, and compare it in context.

    Example: The spread of commute times is larger for drivers, with an IQR of 18 minutes compared with 7 minutes for cyclists.

  28. outlier

    A value well outside the overall pattern of a distribution.

    Loses the mark: “an outlier is the highest value”. The maximum is an outlier only if it sits far from the rest of the data by a stated rule.

    Example: One commute of 95 minutes is an outlier, far above the rest of the times, which stay under 50 minutes.

  29. in context

    Stated using the real situation: naming the variable, its units and the individuals.

    Loses the mark: “the center is 50”. A bare number with no variable or units is not in context; say what was measured and in what units.

    Example: In context, the median delivery time for the 40 pizza orders was 28 minutes, and the times are skewed right.

  30. mean

    xˉ\bar{x}

    The sum of the values divided by how many there are; it is not resistant.

    Loses the mark: “the mean is resistant”. The mean is pulled toward outliers and long tails; the median is the resistant measure of center.

    Example: The mean wait was 12.4 minutes, pulled above the median of 9 minutes by a few very long waits.

  31. median

    The middle value of the ordered data, with half the data at or below it.

    Loses the mark: “the median is the average”. Average means the mean; the median is the middle of the ordered values.

    Example: Because the house prices are skewed right, the median of $310,000 is a better summary of a typical price than the mean.

  32. range

    The largest value minus the smallest value; it depends only on the extremes.

    Loses the mark: “the range is from”. The range is one number, maximum minus minimum, not an interval of values.

    Example: The range of the plant heights is 41 cm minus 12 cm, which is 29 cm.

  33. quartiles

    Q1Q_1, Q3Q_3

    The values cutting the ordered data into four equal parts: Q1, the median and Q3.

    Loses the mark: “Q1 is 25% of the maximum”. Q1 is the 25th percentile of the data, not a quarter of the largest value.

    Example: The quartiles of the test scores are Q1 = 64 and Q3 = 85, so the middle half scored between 64 and 85.

  34. interquartile range

    IQR=Q3−Q1IQR = Q_3 - Q_1

    Q3 minus Q1, the spread of the middle half of the data; it is resistant.

    Loses the mark: “the IQR is from”. The IQR is the single number Q3 minus Q1, not the interval between the quartiles.

    Example: The interquartile range of the commute times is 34 minus 18, which is 16 minutes.

  35. standard deviation

    ss (sample), σ\sigma (population)

    A measure of spread: the typical distance of values from their mean.

    Loses the mark: “how spread out the data is”. Too vague: interpret it as the typical distance of the values from the mean, with units and context.

    Example: The standard deviation of the bag weights is 3.2 grams, so a typical bag is about 3.2 grams away from the mean of 250 grams.

  36. five-number summary

    The minimum, first quartile, median, third quartile and maximum of a data set.

    Loses the mark: “mean, median, mode”. The five-number summary uses the minimum, Q1, median, Q3 and maximum; the mean and mode are not in it.

    Example: The five-number summary of the sprint times is 11.2, 12.0, 12.6, 13.1 and 15.8 seconds.

  37. boxplot

    A graph of the five-number summary: box from Q1 to Q3, median line, whiskers.

    Loses the mark: “the line in the box is the mean”. The line inside the box marks the median; a standard boxplot does not show the mean.

    Example: Side-by-side boxplots show a lower median sprint time for the relay team, and the relay team's box is narrower.

  38. resistant

    Describes a statistic not strongly affected by outliers or skew, such as the median.

    Loses the mark: “the mean is not affected by outliers”. The mean is pulled toward extreme values; it is the median that is resistant.

    Example: The median is resistant, so the one $2 million home barely changes it, while the mean jumps by $40,000.

  39. 1.5 x IQR rule

    Q1−1.5 IQRQ_1 - 1.5\,IQR and Q3+1.5 IQRQ_3 + 1.5\,IQR

    A value more than 1.5 times the IQR below Q1 or above Q3 is flagged.

    Loses the mark: “1.5 times the range”. The rule uses the IQR, measured out from the quartiles, not the range.

    Example: By the 1.5 x IQR rule the upper fence is 85 + 1.5(21) = 116.5, so a score of 120 is an outlier.

  40. population

    The entire group of individuals that a study wants information about.

    Loses the mark: “the population is the people surveyed”. Those surveyed are the sample; the population is the whole group the conclusion is about.

    Example: The population is all 2,400 students at the high school, and the survey reached 150 of them.

  41. sample

    The part of the population from which data are actually collected.

    Loses the mark: “the sample is everyone in the school”. If everyone is measured it is a census; a sample is a subset of the population.

    Example: The sample of 150 students was chosen at random from the school's enrollment list.

  42. census

    A data collection that attempts to measure every individual in the population, not a sample.

    Loses the mark: “a census is a large sample”. A census covers the whole population; a sample of any size is still a sample.

    Example: Counting every tree in the park is a census, so no sampling variability is involved.

  43. simple random sample

    A sample where every possible group of n individuals is equally likely to be chosen.

    Loses the mark: “everyone has an equal chance”. Equal individual chances is not enough; an SRS gives every possible group of n the same chance.

    Example: Number the 2,400 students from 1 to 2,400 and use a random number generator to choose 150 different numbers, giving a simple random sample.

  44. stratified random sample

    A sample made by splitting the population into strata, then taking an SRS of each.

    Loses the mark: “randomly choose some of the groups”. Choosing whole groups is cluster sampling; stratified sampling takes an SRS from every group.

    Example: A stratified random sample takes an SRS of 40 students from each of the four grade levels.

  45. cluster sample

    A sample of every individual in some randomly chosen groups (clusters) of the population.

    Loses the mark: “sample from every cluster”. Taking individuals from every group is stratified sampling; a cluster sample takes whole groups chosen at random.

    Example: Randomly choosing 6 of the 60 homerooms and surveying every student in them is a cluster sample.

  46. convenience sample

    A sample of whoever is easiest to reach; it usually does not represent the population.

    Loses the mark: “random because anyone could be chosen”. Easy-to-reach people are not chosen by chance and often differ from the population.

    Example: Surveying the first 50 students who arrive at the library is a convenience sample, likely to overstate study time for the school.

  47. sampling frame

    The list a sample is actually drawn from; anyone missing can never be chosen.

    Loses the mark: “the sampling frame is the sample”. The frame is the list you draw from; the sample is the part of it you actually select.

    Example: The school's enrollment list is the sampling frame, and the 50 students drawn from it at random are the sample.

  48. strata

    The groups of similar individuals a population is split into for a stratified sample.

    Loses the mark: “pick a few strata at random”. Choosing whole groups at random is cluster sampling; a stratified sample takes an SRS from every stratum.

    Example: The school was split into strata by grade, and an SRS of 25 students was drawn from each grade.

  49. cluster

    One group, ideally a small mixed copy of the population, picked in a cluster sample.

    Loses the mark: “take an SRS from every cluster”. Sampling inside every group is stratified sampling; a cluster sample picks whole clusters at random.

    Example: The district picked 4 of its 30 homerooms at random as clusters and surveyed every student in those 4.

  50. systematic random sample

    A sample taking every kth individual from an ordered list, after a random starting point.

    Loses the mark: “every kth person starting from the first”. Without a random starting point the choice is not random, so the method can be biased.

    Example: For a systematic random sample of 40 from 800 receipts, pick a random start from 1 to 20, then take every 20th receipt.

  51. bias

    A systematic tendency of a method to overestimate or underestimate the true value.

    Loses the mark: “a larger sample would remove the bias”. More data from a flawed method repeat the same error; bias is fixed by changing the method, not the size.

    Example: The library survey has bias: frequent studiers are overrepresented, so it likely overestimates average study time.

  52. undercoverage

    Bias when some of the population have no, or a reduced, chance of being sampled.

    Loses the mark: “they did not respond”. Not answering after being chosen is nonresponse; undercoverage means they could not be chosen at all.

    Example: A phone survey using only landline numbers suffers from undercoverage of young adults who own only cell phones.

  53. nonresponse

    Bias when chosen individuals cannot be reached or refuse, and differ from those who respond.

    Loses the mark: “they were left off the list”. Being left off the sampling list is undercoverage; nonresponse happens after someone is chosen.

    Example: Only 30% of the selected households returned the mailed survey, so nonresponse could bias the results if busy families did not reply.

  54. response bias

    Bias from how answers are collected, such as leading wording or misreporting of sensitive behavior.

    Loses the mark: “people answered randomly”. Response bias is a systematic push in one direction, such as underreporting, not random answering.

    Example: Asking 'Don't you agree the cafeteria food is too expensive?' creates response bias toward agreement.

  55. voluntary response sample

    A sample of people who choose themselves by answering an open invitation.

    Loses the mark: “anyone could respond so it is random”. Letting people choose to respond is self-selection, not a chance process.

    Example: The radio station's call-in poll is a voluntary response sample, so listeners with strong feelings are likely overrepresented.

  56. observational study

    A study that observes individuals and measures variables without assigning any treatment.

    Loses the mark: “an experiment because groups were compared”. Comparing groups that chose their own condition is still observational; an experiment imposes the treatment.

    Example: Comparing the grades of students who chose to play sports with those who did not is an observational study.

  57. experiment

    A study where researchers impose treatments on experimental units and measure a response.

    Loses the mark: “it is an experiment because they compared”. Comparing groups is not enough; an experiment requires the researcher to assign the treatments.

    Example: Randomly assigning 60 volunteers to sleep either 6 or 8 hours before a memory test makes this an experiment.

  58. treatment

    A specific condition applied to experimental units, such as a drug dose or teaching method.

    Loses the mark: “the treatment is the result”. The treatment is what is imposed; the result that is measured is the response.

    Example: The two treatments are 6 hours and 8 hours of sleep the night before the memory test.

  59. experimental unit

    The individual, object or group that is randomly assigned a treatment; people are called subjects.

    Loses the mark: “the units are the measurements”. Units are who or what receives a treatment; the measurements are the response.

    Example: Each of the 40 tomato plants is an experimental unit, assigned one of two fertilizers.

  60. confounding variable

    A variable tied to the explanatory variable that also affects the response, mixing their effects.

    Loses the mark: “another variable that could affect the results”. To confound, it must also be linked to the explanatory variable or groups, not just affect the response.

    Example: Family income is a confounding variable: it is linked to playing sports and also affects grades through tutoring.

  61. random assignment

    Using chance to decide which experimental units get which treatment, so groups start roughly alike.

    Loses the mark: “random sampling lets us conclude cause”. Random sampling picks who is in the study; random assignment decides who gets which treatment.

    Example: Random assignment of the 40 plants to the two fertilizers balances sunlight and soil quality across the groups on average.

  62. control

    Keeping other variables as similar as possible across groups, so differences come from treatment.

    Loses the mark: “control means having a control group”. Control also means keeping other conditions the same for every group, not only including a baseline group.

    Example: To control other variables, all plants got the same pot size, water schedule and greenhouse bench rotation.

  63. control group

    A group given no treatment, a placebo or the standard treatment, as a baseline.

    Loses the mark: “the group that is controlled”. A control group is the baseline for comparison, not a group kept under tighter rules.

    Example: The control group received the usual fertilizer so the new formula's effect could be judged against a baseline.

  64. placebo

    A fake treatment with no active ingredient, so the comparison group has the same experience.

    Loses the mark: “the placebo makes them better”. A placebo has no active effect; any improvement comes from expectation, the placebo effect.

    Example: Subjects in the comparison group took a placebo tablet that looked and tasted like the real vitamin.

  65. placebo effect

    A real response caused by believing one is being treated, even with no active ingredient.

    Loses the mark: “the placebo worked”. The pill itself did nothing; improvement came from expectation, which is why every group must believe it is treated.

    Example: Because of the placebo effect, some subjects reported less pain after taking the inactive tablet.

  66. blinding

    Keeping subjects, or those measuring the response, unaware of who received which treatment.

    Loses the mark: “hiding the results”. Blinding hides the treatment each subject received, not the results.

    Example: Blinding the judges to which fertilizer each plant got keeps their ratings from favoring the new formula.

  67. double-blind

    An experiment where neither subjects nor those measuring the response know who got which treatment.

    Loses the mark: “double-blind means two groups”. Double-blind refers to two parties being unaware of the treatments, not to the number of groups.

    Example: The vitamin trial was double-blind: neither the subjects nor the nurses recording symptoms knew who got the real tablet.

  68. replication

    Using enough experimental units in each treatment group that chance differences average out.

    Loses the mark: “repeat the experiment later”. In design, replication means many units per treatment within the study, not running the study again.

    Example: Replication with 20 plants per fertilizer makes it unlikely that one unusually healthy plant decides the result.

  69. completely randomized design

    A design assigning all experimental units to treatments entirely at random, with no blocking.

    Loses the mark: “completely randomized design randomizes within each block”. Randomizing inside blocks is a randomized block design.

    Example: In a completely randomized design, the 40 plants are randomly split into two groups of 20, one per fertilizer.

  70. randomized block design

    A design that sorts alike units into blocks, then randomly assigns treatments within each block.

    Loses the mark: “the blocks are chosen at random”. Blocks are formed by a known variable; the randomness is in assigning treatments within each block.

    Example: In a randomized block design, the plants are blocked by greenhouse bench and the fertilizers are randomly assigned within each bench.

  71. matched pairs design

    A design pairing similar units, or giving one unit both treatments, then comparing within pairs.

    Loses the mark: “pairing at random”. Pairs are formed by similarity; the randomness is which unit in each pair gets which treatment.

    Example: In a matched pairs design, each runner tries both shoe types in random order and we compare the two times for each runner.

  72. factor

    An explanatory variable the experimenter deliberately sets for each unit, such as fertilizer type.

    Loses the mark: “the factor is the response”. A factor is set by the experimenter; the response is what is measured afterward.

    Example: The garden experiment has two factors, fertilizer type and amount of water, and each treatment combines one setting of each.

  73. level

    One of the specific settings a factor takes, such as 0, 10 or 20 grams.

    Loses the mark: “the levels are the experimental units”. Levels are settings of a factor; the units are the subjects or objects that receive them.

    Example: Water has three levels, low, medium and high, so with two fertilizers the experiment has six treatments.

  74. scope of inference

    Whether a study's design lets us generalize to a population and conclude cause and effect.

    Loses the mark: “the results apply to everyone”. Generalize only to the population that was randomly sampled, and name that population.

    Example: The scope of inference is limited: subjects were volunteers, so results apply only to people like them, though random assignment supports a causal claim.

  75. random selection

    Choosing individuals from the population by a chance process, which allows generalizing to that population.

    Loses the mark: “random assignment allows us to generalize”. Random assignment supports cause and effect; generalizing needs random selection from the population.

    Example: Because the 500 voters were a random sample of the city's registered voters, the results can be generalized to all registered voters in the city.

  76. cause-and-effect conclusion

    A claim that the treatment caused the change in response; justified only with random assignment.

    Loses the mark: “random sampling allows a cause-and-effect conclusion”. Cause and effect depends on random assignment of treatments, not on random sampling.

    Example: Since the fertilizers were randomly assigned, a cause-and-effect conclusion is justified: the new fertilizer caused the larger yields for these plants.

  77. generalize to the population

    Extend a study's results to the larger population the individuals were randomly selected from.

    Loses the mark: “applies to all people”. Name the specific population that was randomly sampled; results do not reach beyond it.

    Example: Because the subjects were volunteers, we cannot generalize to the population of all adults in the city.

  78. investigative question

    The question a study sets out to answer, naming individuals, variables and the comparison.

    Loses the mark: “any data about the topic will answer it”. Data answer the question only if they record the variables it names, for the individuals it is about.

    Example: The investigative question 'Do students who play a sport sleep more?' needs each student's sport status and hours of sleep.