Chapter 2: Descriptive Statistics

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/94

encourage image

There's no tags or description

Looks like no tags are added yet.

95 Terms

1
New cards

What are the two main types of data?

Categorical (qualitative) and numerical (quantitative).

2
New cards

What is nominal data?

Categorical data whose categories have no natural ordering.

3
New cards

What is ordinal data?

Categorical data whose categories have a meaningful order.

4
New cards

Which averages make sense for nominal data?

Mode only; mean and median are not meaningful.

<p>Mode only; mean and median are not meaningful.</p>
5
New cards

Which averages make sense for ordinal data?

Mode and median; the mean is generally not meaningful.

<p>Mode and median; the mean is generally not meaningful.</p>
6
New cards

What is discrete numerical data?

Numerical data with separated possible values, often counts.

7
New cards

What is continuous numerical data?

Numerical data that can theoretically take any value in an interval.

8
New cards

What distinguishes ratio data from interval data?

Ratio data have a meaningful natural zero; interval data do not.

9
New cards

What five features should you examine when describing a distribution?

Location, spread, shape, outliers, and other unusual features (such as distinct groups or unexpectedly frequent values).

10
New cards

What does location describe?

The centre or typical value of a distribution.

11
New cards

What does spread describe?

How variable or dispersed the observations are.

12
New cards

What is positive skew?

A distribution with a long tail extending to the right.

<p>A distribution with a long tail extending to the right.</p>
13
New cards

What is negative skew?

A distribution with a long tail extending to the left.

<p>A distribution with a long tail extending to the left.</p>
14
New cards

What is the five-number summary?

Minimum, lower quartile, median, upper quartile and maximum.

15
New cards

What does the median represent?

The middle value in the ordered data. For numerical data, use the middle observation when n is odd and the average of the middle two when n is even. For ordinal data, averaging category labels may make no sense.

16
New cards

What are the lower and upper quartiles?

The lower quartile (qL) separates approximately the lowest 25% of observations from the rest; the upper quartile (qU) separates approximately the highest 25% from the rest.

17
New cards

How is the range calculated?

Maximum - minimum.

18
New cards

How is the interquartile range (IQR) calculated?

Upper quartile - lower quartile: qU - qL.

19
New cards

What are the range and IQR measures of?

Spread or variability.

20
New cards

What is the sample mean?

The arithmetic average: x̄ = (1/n)Σxi.

21
New cards

What does the sample variance measure?

It measures spread in squared units. For n>1, s²=[Σᵢ(xᵢ−x̄)²]/(n−1), summing over the n observations.

22
New cards

What is the sample standard deviation?

The square root of sample variance; it has the data's original units.

23
New cards

When is median often preferred to mean?

When data are strongly skewed or contain outliers.

24
New cards

Why is the mean sensitive to outliers?

Extreme values directly affect the arithmetic average.

25
New cards

When is IQR often preferred to standard deviation?

For skewed data or data containing outliers.

26
New cards

What is the mode?

For categorical or discrete data, the mode is the value or values that occur most often. It is a measure of location.

27
New cards

Why should a sample mode for rounded continuous data be interpreted carefully?

Rounding or grouping can create repeated recorded values or change which value appears most often. The observed mode can therefore reflect measurement precision.

28
New cards

When do ratios and percentage changes make sense?

Only when the variable is measured on a ratio scale.

29
New cards

Why can percentage temperature changes be misleading in Celsius?

0°C is arbitrary, so Celsius is an interval rather than ratio scale.

30
New cards

Why do numerical category labels remain categorical data?

They label the categories. The numbers chosen carry no numerical meaning in themselves, so ordinary arithmetic on the labels need not be meaningful.

31
New cards

What is grouped data, and what information does grouping lose?

Grouped data report the number of observations in each class rather than each individual value. The exact values and their positions within a class cannot be recovered from those counts alone.

32
New cards

What is the 100p% sample quantile using the STAT0002 method?

For 0<p<1, use x_(p(n+1)), where x_(i) is the ith smallest observation. Interpolate when the rank is between two observations and use the nearest endpoint when it is outside the available ranks.

33
New cards

How do you linearly interpolate a sample quantile between two ordered observations?

If i<p(n+1)<i+1, the quantile estimate is x₍ᵢ₎+[p(n+1)−i](x₍ᵢ₊₁₎−x₍ᵢ₎). For ordinal categories, numerical interpolation may be meaningless.

<p>If i&lt;p(n+1)&lt;i+1, the quantile estimate is x₍ᵢ₎+[p(n+1)−i](x₍ᵢ₊₁₎−x₍ᵢ₎). For ordinal categories, numerical interpolation may be meaningless.</p>
34
New cards

What if the course quantile position p(n+1) falls outside the available ranks?

Use the nearest observation: x₍₁₎ when the position is below 1 and x₍ₙ₎ when it is above n. An estimate further outside would require an extrapolation assumption.

35
New cards

What rank positions define the course’s sample quartiles?

For ordered data, use p=1/4, 1/2 and 3/4 for the lower quartile, median and upper quartile, respectively. The rank position is p(n+1), with linear interpolation when needed. This is the type-6 convention.

36
New cards

What is the difference between xᵢ and x₍ᵢ₎?

xᵢ is the observation in the original ith position. x₍ᵢ₎ is the ith smallest value after sorting, so x₍₁₎ is the minimum and x₍ₙ₎ is the maximum.

37
New cards

Is comparing the mean and median a reliable test for skewness?

No. Mean and median can give a misleading indication of skewness.

<p>No. Mean and median can give a misleading indication of skewness.</p>
38
New cards

What should you primarily use to judge the shape or skewness of data?

A suitable plot of the distribution.

<p>A suitable plot of the distribution.</p>
39
New cards

How does standardised sample skewness react to extreme values?

It is sensitive to very large or very small observations.

40
New cards

How does quartile skewness react to extreme values?

It is much less sensitive because it is based on quartiles.

41
New cards

Can different measures of skewness disagree?

Yes. They can disagree with each other and with the appearance of a plot.

42
New cards

Why must you consider what numerical data represent before summarising them?

The measurement scale and context determine which summaries are meaningful.

43
New cards

Why can the ordinary mean misrepresent circular measurements?

Directions wrap around, so values numerically far apart can be close on a circle. A meaningful summary must respect that geometry and the units of the data.

<p>Directions wrap around, so values numerically far apart can be close on a circle. A meaningful summary must respect that geometry and the units of the data.</p>
44
New cards

What is a frequency distribution?

A table showing how many observations fall into each of several non-overlapping classes.

<p>A table showing how many observations fall into each of several non-overlapping classes.</p>
45
New cards

What is the modal class?

The class or interval with the highest frequency.

<p>The class or interval with the highest frequency.</p>
46
New cards

How is relative frequency calculated?

Frequency of a class divided by the total number of observations (n).

<p>Frequency of a class divided by the total number of observations (n).</p>
47
New cards

What is cumulative frequency?

The running total of frequencies up to and including a class.

<p>The running total of frequencies up to and including a class.</p>
48
New cards

What is cumulative relative frequency?

Cumulative frequency divided by n; the proportion of observations up to a class boundary.

<p>Cumulative frequency divided by n; the proportion of observations up to a class boundary.</p>
49
New cards

What does the area of a histogram bar represent?

The relative frequency in that interval. In a normalised histogram, all bar areas sum to 1.

<p>The relative frequency in that interval. In a normalised histogram, all bar areas sum to 1.</p>
50
New cards

What is the height of a normalised histogram bar?

Relative frequency ÷ bin width = frequency ÷ (n × bin width).

<p>Relative frequency ÷ bin width = frequency ÷ (n × bin width).</p>
51
New cards

Why must histogram bin widths be accounted for?

With unequal widths, frequency alone gives misleading bar heights; area must reflect the amount of data.

<p>With unequal widths, frequency alone gives misleading bar heights; area must reflect the amount of data.</p>
52
New cards

What distinguishes a barplot from a histogram?

A barplot displays categorical or discrete numerical data with separated bars; a histogram displays binned continuous numerical data.

<p>A barplot displays categorical or discrete numerical data with separated bars; a histogram displays binned continuous numerical data.</p>
53
New cards

Are outliers necessarily incorrect observations?

No. Extreme values can arise naturally; do not automatically remove them.

54
New cards

What are boxplots especially useful for?

Comparing distributions’ location, spread and shape; their main purpose is not outlier detection.

<p>Comparing distributions’ location, spread and shape; their main purpose is not outlier detection.</p>
55
New cards

Why are pie charts often ineffective?

Angles and areas are hard to compare accurately, and many categories create clutter.

<p>Angles and areas are hard to compare accurately, and many categories create clutter.</p>
56
New cards

What do the box and central line in a boxplot represent?

The box runs from the lower to the upper quartile and shows the middle 50% of the data. The line in the box is the sample median.

<p>The box runs from the lower to the upper quartile and shows the middle 50% of the data. The line in the box is the sample median.</p>
57
New cards

When should a time-series plot be used?

To display change over time; put time on the horizontal axis to reveal trends, cycles and unusual changes.

<p>To display change over time; put time on the horizontal axis to reveal trends, cycles and unusual changes.</p>
58
New cards

What can exploratory data analysis reveal before a formal analysis?

Graphs, tables and summary statistics can reveal errors, unusual observations, distribution shapes and relationships. They help guide the choice of analysis and communicate what the data show.

59
New cards

What should a graph’s labels and legend tell the reader?

Identify what each axis and its units represent, and explain any symbols or groups. Use a caption or legend where needed so the graph can be understood without guessing.

60
New cards

What is the equivalent formula for the sample variance using sums of squares?

For n>1, s²=[Σᵢxᵢ²−n(x̄)²]/(n−1). Using divisor n instead of n−1 gives s²(1−1/n).

61
New cards

How does a positive affine change of units affect sample summaries?

If yᵢ=axᵢ+b with a>0, mean and median become a times the old value plus b. The range, IQR and standard deviation multiply by a; the variance multiplies by a². Both sample skewness measures are unchanged where defined.

62
New cards

What is the advantage of using log(x/y) to compare two positive values?

log(x/y)=log x−log y. It is zero for equal values, and reversing the comparison changes its sign: log(y/x)=−log(x/y).

63
New cards

What does one unit on a log10 scale mean?

A tenfold change on the original scale; 10, 100 and 1,000 are equally spaced on a log10 axis.

<p>A tenfold change on the original scale; 10, 100 and 1,000 are equally spaced on a log10 axis.</p>
64
New cards

What does a scatterplot show?

The relationship between two paired numerical variables, with each point representing one pair of observations.

<p>The relationship between two paired numerical variables, with each point representing one pair of observations.</p>
65
New cards

Which variables normally go on the x-axis and y-axis of a scatterplot?

x: explanatory or predictor variable; y: response or outcome variable.

<p>x: explanatory or predictor variable; y: response or outcome variable.</p>
66
New cards

What four features should you examine in a scatterplot?

Direction, form (linear or curved), strength of association and unusual points/outliers.

<p>Direction, form (linear or curved), strength of association and unusual points/outliers.</p>
67
New cards

Why can an outlier appear in a scatterplot but not in separate one-variable plots?

A point may have ordinary individual values but an unusual combination of the two variables.

<p>A point may have ordinary individual values but an unusual combination of the two variables.</p>
68
New cards

Why might we transform a dataset?

To make a distribution more symmetric or a relationship more linear so a simple statistical method’s assumptions fit better.

<p>To make a distribution more symmetric or a relationship more linear so a simple statistical method’s assumptions fit better.</p>
69
New cards

How can a logarithmic transformation affect positively skewed data?

It compresses large positive values relative to small ones, often reducing right skew.

<p>It compresses large positive values relative to small ones, often reducing right skew.</p>
70
New cards

Which transformations might reduce positive skew?

For positive data, try a power with 0<c<1 (such as a square root), a log, or the monotone Box–Cox form with c<0. These can compress the long right tail; replot to check.

<p>For positive data, try a power with 0&lt;c&lt;1 (such as a square root), a log, or the monotone Box–Cox form with c&lt;0. These can compress the long right tail; replot to check.</p>
71
New cards

Which transformations might reduce negative skew?

For positive data, try powers y^c with c > 1, such as y² or y³, to stretch the upper tail.

72
New cards

When is taking logarithms directly invalid?

When observations are zero or negative; a suitable positive shift may sometimes be applied first.

73
New cards

Why might we straighten a curved scatterplot?

An approximately linear relationship is easier to interpret and better suited to methods such as linear regression.

<p>An approximately linear relationship is easier to interpret and better suited to methods such as linear regression.</p>
74
New cards

Can you transform x, y or both to straighten a scatterplot?

Yes. Transform the explanatory variable, response variable or both, then inspect the new plot.

<p>Yes. Transform the explanatory variable, response variable or both, then inspect the new plot.</p>
75
New cards

How do you recognise positive or negative association on a scatterplot?

Positive association: when one variable increases the other tends to increase. Negative association: when one increases the other tends to decrease. Either can be linear or curved.

76
New cards
What does Pearson's sample correlation coefficient measure?
The strength of linear association between two variables.
77
New cards
What values can Pearson's sample correlation coefficient take?
From −1 to +1.
78
New cards

What does Spearman's rank correlation coefficient measure?

How close the relationship is to being monotone. Rank each variable separately, giving rank 1 to the largest, then calculate the product-moment correlation of the paired ranks. Both sets of ranks must have nonzero spread. Perfect increasing and decreasing order give rₛ=1 and rₛ=−1, respectively.

79
New cards
What is a monotonic association?
As x increases, y tends to increase or decrease; the relationship need not be linear.
80
New cards
Why might Spearman's correlation be preferred to Pearson's?
It is less sensitive to outliers and measures monotone association, including nonlinear relationships.
81
New cards

What is sample quartile skewness?

[(qU−m)−(m−qL)]/(qU−qL), where qL, m and qU are the sample quartiles. It is defined when qU>qL.

82
New cards

What is an advantage of a stem-and-leaf plot?

It preserves the entire dataset (perhaps rounded) in sorted order, making the five-number summary easy to calculate.

<p>It preserves the entire dataset (perhaps rounded) in sorted order, making the five-number summary easy to calculate.</p>
83
New cards

What is a dotplot?

Each observation is a dot; repeated values (perhaps after rounding) are stacked vertically.

<p>Each observation is a dot; repeated values (perhaps after rounding) are stacked vertically.</p>
84
New cards
How are the whiskers of a standard boxplot determined?
They reach the most extreme observations within 1.5 IQR of the corresponding quartiles. More extreme values are shown individually.
85
New cards

What is an empirical cumulative distribution function (ECDF)?

It gives the cumulative relative frequency: the proportion of observations less than or equal to each value.

<p>It gives the cumulative relative frequency: the proportion of observations less than or equal to each value.</p>
86
New cards

What is Tukey and Mosteller's bulging rule?

A guide that uses the curve's bulging direction to suggest transformations of x, y or both that may straighten a scatterplot.

<p>A guide that uses the curve's bulging direction to suggest transformations of x, y or both that may straighten a scatterplot.</p>
87
New cards
What is the Box-Cox transformation?
BC(x) = (x^c - 1)/c when c ≠ 0, and BC(x) = ln(x) when c = 0. It is used to transform positive data, with c controlling the transformation.
88
New cards
What are binary data?
Categorical data with only 2 categories. They can be nominal or ordinal.
89
New cards

What is standardised sample skewness?

[(1/n)Σᵢ(xᵢ−x̄)³]/s³, where s is the sample standard deviation. This requires n>1 and s>0.

90
New cards

What values can the two sample skewness measures take?

Where defined, quartile skewness is in [−1,1]. Standardised sample skewness has no fixed bounds across sample sizes. Both are zero for perfectly symmetric data; the two signs can disagree.

91
New cards

How is Pearson’s sample correlation coefficient calculated?

For n≥2 paired observations whose two variables both vary, r=Σᵢ(xᵢ−x̄)(yᵢ−ȳ)/√[Σᵢ(xᵢ−x̄)²·Σᵢ(yᵢ−ȳ)²]. It measures linear association.

92
New cards
How are ties handled when calculating Spearman’s rank correlation?
Average the ranks of tied observations. For example, the 3rd and 4th largest equal values each receive rank 3.5. Calculate the product-moment correlation of the paired ranks.
93
New cards
What formula simplifies Spearman’s rank correlation when there are no ties?
rₛ=1−6Σᵢdᵢ²/[n(n²−1)], where dᵢ is the difference between the paired ranks. This simplification assumes no ties.
94
New cards

What directions of transformation does the bulging rule suggest?

Compare the curve with the straight line joining its ends. For a positive variable that bulges down, try a power below 1, such as a square root, or try a log. For a variable that bulges up, try a power above 1. Consider x and y separately.

95
New cards
What should you check before interpreting an association in observational data?
Check who is represented, which measurements are missing, and whether other variables could explain the pattern. State the association for the observed population. A causal interpretation requires additional evidence.