Averages and spread

To see why one spread measure is preferred, we need to recall what each one measures. Both the Range and the Interquartile Range (IQR) describe how spread out data values are, but they pay attention to different parts of the dataset.

• The Range uses only the two extremes: $\text{Range}=\text{Maximum}-\text{Minimum}$

• The IQR uses the middle quartiles, so it captures the spread of the middle $50\%$ of the data: $\text{IQR}=Q_3-Q_1$

The range can change a lot because of outliers—unusually high or low values. Since it depends only on the minimum and maximum, just one extreme value can make the dataset look far more spread out than most of the values suggest.

For instance, for $\{3,9,10,11,12,13,24\}$: $\text{Range}=24-3=21$

If the $24$ is replaced by a much larger outlier such as $120$, then: $\text{Range}=120-3=117$ Even though the other numbers stayed the same, the range jumps dramatically.

The IQR is designed to be resistant to outliers. It looks only from $Q_1$ (the $25^{th}$ percentile) to $Q_3$ (the $75^{th}$ percentile), so extreme ends have little or no effect.

Using the updated set $\{3,9,10,11,12,13,120\}$:

• $Q_1=9$
• $Q_3=13$
• $\text{IQR}=13-9=4$

Even after adding the outlier ($120$), the IQR remains $4$, which better represents the usual spread of most scores.

Now we can eliminate the incorrect statements:

• Option A is wrong because the IQR describes only the middle $50\%$ of the data, not all of it.
• Option C is wrong because both range and IQR require numerical values (you subtract), and categorical labels don’t have a meaningful maximum/minimum to subtract.
• Option D is wrong because the range is a single subtraction once you know the smallest and largest values, while the IQR requires finding $Q_1$ and $Q_3$ first, which usually takes more steps.

When comparing box plots, the goal is to compare the typical spread without being fooled by a few unusual values. Since the IQR focuses on the central half of the data and is not strongly affected by outliers, it is the standard choice.

To eliminate options efficiently, start by remembering the definitions from Statistical Measures and Data Representation:

• The Range is the difference between the maximum and minimum values: $Range=x_{max}-x_{min}$.
• The Interquartile Range (IQR) is the difference between the third and first quartiles: $IQR=Q_3-Q_1$.

Check option A. Quartiles split ordered data into four equal sections. The $IQR$ runs from $Q_1$ (the $25^{th}$ percentile) to $Q_3$ (the $75^{th}$ percentile).

$75\%-25\%=50\%$

So the IQR contains the middle $50\%$ of the data, not the full dataset. That makes option A incorrect.

Now look at option C. Both the range and the $IQR$ involve subtraction, so they only make sense for numerical data. You can’t subtract categories like “soccer” and “basketball.” Therefore the statement that the range is “only used for categorical data” is false.

Consider option D. The range is found by identifying the smallest and largest values and doing one subtraction. Finding the $IQR$ typically means sorting the data, locating the median, then finding $Q_1$ and $Q_3$ from the lower and upper halves, and finally subtracting. So the IQR is not easier to compute than the range.

After removing A, C, and D, only Option B remains. This is correct because the range is controlled by the most extreme values and can be distorted by an outlier, while the $IQR$ describes the spread of the central $50\%$ and is therefore more resistant and more representative.

Start by naming the data type. Eye color (such as blue, brown, or green) is a description of a trait, not a measurement you can count or measure with numbers.

In statistics, this is called qualitative or categorical data. More specifically, because eye colors don’t have a natural ranking (one color isn’t “greater” than another), the data is nominal.

The arithmetic mean, written as $\bar{x}$, is found using:

$\bar{x}=\frac{\sum x}{n}$

This rule requires adding all the data values ($x$) and then dividing by the number of values ($n$). But you can’t do arithmetic like adding “Blue” + “Brown” + “Green,” so a mean cannot be computed for categorical data.

This idea connects to Statistical Measures and Data Representation.

Option B blames the number of observations. But even with a very large group (hundreds or millions of people), you still couldn’t find a mean for eye colors because the issue is that the values are not numbers.

Option C talks about symmetry, which describes the shape of a numerical distribution. Whether a distribution is symmetric or not doesn’t explain why a mean can’t be found for eye colors, since the main problem is that eye colors aren’t numerical values.

Option D says the information is numerical but discrete. Discrete numerical data (like the number of books someone owns) can have a mean. Eye color isn’t numerical in the first place, so this option is based on an incorrect data type.

Because eye color is categorical (qualitative) data, arithmetic operations don’t apply. That means the mean is not a suitable summary measure for this data set.

Before checking the answer choices, determine the data type. Eye color (e.g., blue, brown, green) is categorical data (also called qualitative data). This falls under Statistical Measures and Data Representation.

A mean requires that we can add the values and divide by how many there are. Now use elimination to see which option correctly explains why that can’t happen here.

Option B suggests the mean can’t be calculated because $n$ is too small. But if the data were numerical, a mean can be computed for any sample size. A small sample may be less representative, but it doesn’t make the mean impossible to calculate.

Option C focuses on symmetry. Means can be found for symmetric, skewed, or uneven distributions. Symmetry might affect how informative the mean is, but it never stops the mean from being defined when the data is numerical.

Option D claims discrete numbers can’t have a mean, which is false. Discrete numerical data often uses means. For instance, it makes sense to say the mean number of children is $1.5$, even though no family has half a child. The mean does not require continuous data.

After eliminating the other statements, the correct reason is that eye color is categorical. Words can’t be added or averaged. For example:

$\frac{\text{Blue}+\text{Brown}+\text{Green}}{3}=\text{?}$

This has no mathematical meaning, so the mean cannot be used.

We are asked how variance changes when values move nearer to the mean, so we begin with the given variance formula:

${\sigma }^2=\frac{{\sum }_{i=1}^n(x_i-\bar{x}{)}^2}{n}$

The expression $(x_i-\bar{x})$ tells us how far each data value $x_i$ is from the mean $\bar{x}$.

If the data values are "moved closer to the mean", then the size of the deviation, $|x_i-\bar{x}|$, becomes smaller.

In the numerator, the deviations are squared: $(x_i-\bar{x}{)}^2$.

If a positive distance $k$ becomes smaller, then $k^2$ also becomes smaller. So when points get closer to the mean, each squared deviation term decreases as well.

Variance depends on the total of the squared deviations: $\sum (x_i-\bar{x}{)}^2$.

Since the individual squared terms are getting smaller and the number of data points $n$ stays fixed, the entire sum decreases. Dividing a smaller numerator by the same constant $n$ gives a smaller value of ${\sigma }^2$.

Check each option against what the formula implies:

• Option A is wrong because the numerator is shrinking, so variance cannot grow.
• Option B cannot happen because squared terms are always $\ge 0$, so variance cannot be negative.
• Option D is wrong because changing how far values are from the mean changes the numerator.

With a smaller sum of squared deviations and the same $n$, the variance must go down.

In statistics, variance describes how spread out the data values ($x_i$) are from the average, the mean ($\bar{x}$).

If you want to review how we describe and quantify data sets, take a look at Statistical Measures and Data Representation: Univariate statistics.

Variance is built from deviations $(x_i-\bar{x})$.

When data values are "moved closer to the mean $\bar{x}$", those deviations become smaller, and squaring them makes $(x_i-\bar{x}{)}^2$ smaller too.

If each squared deviation $(x_i-\bar{x}{)}^2$ gets smaller (or stays $0$ for values already at the mean), then the total ${\sum }_{i=1}^n(x_i-\bar{x}{)}^2$ must decrease.

Because $n$ stays the same, dividing by the same denominator produces a smaller variance value $({\sigma }^2)$.

A helpful picture: if everyone in a line steps toward the middle, the group becomes less spread out.

Consider the incorrect choices:

• Option A is incorrect because a larger variance would mean the data are spreading out more, not clustering.
• Option B is impossible since variance is based on squared values, which cannot be negative.
• Option D is incorrect because changing distances from the mean changes the squared deviations and therefore the variance.

Bringing points closer to the mean reduces the spread, so the variance goes down.

To observe the effect directly, we can test with a small dataset (from Statistical Measures and Data Representation).

Choose three values with a simple mean: $x=\{1,5,9\}$

Compute the mean: $\bar{x}=\frac{1+5+9}{3}=5$

Use the given formula:

${\sigma }^2=\frac{{\sum }_{i=1}^n(x_i-\bar{x}{)}^2}{n}$

With $n=3$ and $\bar{x}=5$: ${\sigma }^2=\frac{(1-5{)}^2+(5-5{)}^2+(9-5{)}^2}{3}$ ${\sigma }^2=\frac{(-4{)}^2+(0{)}^2+(4{)}^2}{3}=\frac{16+0+16}{3}=\frac{32}{3}\approx 10.67$

Now move the values closer to the mean (near $5$) while keeping $n$ the same:

$x_{new}=\{4,5,6\}$

Here $n=3$ and the mean is still $5$. Recalculate: ${\sigma }_{new}^2=\frac{(4-5{)}^2+(5-5{)}^2+(6-5{)}^2}{3}$ ${\sigma }_{new}^2=\frac{(-1{)}^2+(0{)}^2+(1{)}^2}{3}=\frac{1+0+1}{3}=\frac{2}{3}\approx 0.67$

Compare the variances: $10.67\to 0.67$.

• It did not go up, so remove Option A.
• It cannot be negative because it’s built from squared terms divided by positive $n$, so remove Option B.
• It changed, so it did not stay constant; remove Option D.

Since the variance dropped from about $10.67$ to about $0.67$ when the points moved closer to the mean, the variance decreases.

To figure out what units variance has, we should begin with its definition. Variance (written as ${\sigma }^2$ or $s^2$) is the mean of the squared distances from the Mean. The formula is:

${\sigma }^2=\frac{\sum (x-\mu {)}^2}{N}$

In this expression, $x$ stands for each data value and $\mu$ is the mean of the data.

Suppose we are timing runners in seconds ($s$).

When we compute how far a value is from the mean, we calculate $(x-\mu )$. That is seconds minus seconds:

$\text{Unit: }s-s=s$

So after subtraction, the unit is still the original unit (seconds).

Variance then requires us to square that difference: $(x-\mu {)}^2$.

Squaring affects the units as well as the number. If the difference is in seconds ($s$), then squaring gives:

$\text{Unit: }(s{)}^2=s^2$

So the unit becomes square seconds, not plain seconds.

Next, we add up the squared differences and divide by $N$. Since $N$ is only a count (it has no units), it does not change the unit.

That means variance keeps the squared units (like $s^2$).

Because variance is in squared units, it doesn’t match the original measurement unit directly. That’s why we often take the square root of variance to get the Standard Deviation, which returns to the original unit ($s$).

The claim says variance is "expressed in the original units of measurement (e.g., meters)". But our unit check shows variance ends up in squared units instead.

So the option "True" cannot be correct.

Since the statement gets the units of variance wrong, the correct choice is False.

To judge the statement, we need to remember what variance does. Variance describes how spread out data values are from their average.

For a population, the variance (${\sigma }^2$) is:

${\sigma }^2=\frac{\sum (x_i-\mu {)}^2}{N}$

The important feature is the power of 2: the difference between each value ($x_i$) and the mean ($\mu$) is squared.

If the measurements are in seconds ($s$), then $(x_i-\mu )$ is also in seconds. But once we square that difference, the units must be squared too:

$\text{units of variance}=(m{)}^2=m^2$

So variance is written in squared units, not in the original units.

This directly contradicts the idea that it stays in the "original units of measurement (e.g., meters)".

To undo the squaring and get back to the original unit, we take a square root. That is exactly how standard deviation ($\sigma$) is defined:

$\sigma =\sqrt{{\sigma }^2}$

Taking the square root returns the units to the original ones (for example, $\sqrt{m^2}=m$), which is why standard deviation is often easier to interpret than variance.

The statement says variance keeps the original units, but squaring forces the units to become squared (like $s^2$ rather than $s$). So the statement is incorrect, meaning the “True” option should be eliminated.

Because variance is in squared units, the correct answer is False.

A five-number summary is a descriptive tool used in Topic: Statistical Measures and Data Representation to describe how a dataset is spread out. It is made up of five key values:

1. Minimum
2. First Quartile ($Q_1$)
3. Median ($Q_2$)
4. Third Quartile ($Q_3$)
5. Maximum

The numbers in a five-number summary are positional statistics. That means you identify them by where they fall in the ranked list of data.

For instance, the Median is the central value, and the Quartiles split the data into four equal sections. You can’t correctly locate the “middle” or the “quarter points” unless the values are placed in an ordered sequence.

Now consider the choices that focus on computing other statistics instead of locating positions:

• Option A: Squaring each value is used when computing variance, which measures spread around the mean, but it does not help you locate $Q_1$, $Q_2$, or $Q_3$.
• Option B: Standard deviation is also a spread measure, but it is not part of a five-number summary and is not needed to find quartiles.
• Option D: The arithmetic mean is a useful average, but quartiles are found by position, not by averaging all values.

To determine the median and quartiles, you must start by sorting the raw data. Once the data is in order, you can clearly identify the minimum and maximum and then use the positions of values to locate the quartiles.

So, the necessary first step is to order the data values from smallest to largest.

To answer this, first remember what a five-number summary is. It is a group of descriptive statistics that summarizes a dataset’s distribution. The five values are:

1. Minimum value
2. First Quartile ($Q_1$)
3. Median ($Q_2$)
4. Third Quartile ($Q_3$)
5. Maximum value

All of these are positional measures, meaning they depend on where a value appears once the data is ranked. This idea is part of Statistical Measures and Data Representation.

Option A talks about squaring values to compute variance.

Variance measures how far data values tend to be from the mean. Even though it can be useful, it plays no role in locating the median or quartiles. So this is not a required step.

Option B suggests finding the standard deviation.

Standard deviation, like variance, describes spread around the mean. However, it is not one of the five numbers in a five-number summary, and you do not need it in order to determine quartiles.

Option D suggests calculating the arithmetic mean.

The mean is the average of all the values. But a five-number summary uses the median and quartiles, which are based on position in an ordered list. The average is not needed to find those positions.

Because quartiles and the median correspond to the $25^{th}$, $50^{th}$, and $75^{th}$ percentiles, you can’t tell which value is at those percentiles unless the data is organized in order.

Therefore, arranging the data from smallest to largest is the essential first step.

Start with the level tags: "Low sugar", "Medium sugar", or "High sugar".

These are words used to place items into groups based on a description rather than a measured amount. Data made of names/labels is classified as categorical.

Because the labels can be arranged from lower to higher, they can be seen as ordered categories, but for this question it is enough to say they are categorical.

Now consider the "exact mass of sugar in grams".

Mass is obtained by measuring, not counting. A measurement can include decimals (for example, $4.5\text{ g}$ or $4.52\text{ g}$), so it can take any value in an interval. That makes it numerical continuous.

Discrete numerical data usually comes from counting items, where only whole-number outcomes like $1,2,3$ occur.

Use the classifications to check each option:

• Option B cannot be right because the level labels are not numbers.
• Option C cannot be right because the mass value is a measurement, not a label.
• Option D cannot be right because mass is measured continuously, not counted in whole steps.

The labels are categorical and the mass measurement is numerical continuous, so Option A is the only statement that matches.

The first variable sorts items into: "Low sugar", "Medium sugar", or "High sugar". These are group names, not values you measure or count. In statistics, this kind of variable is categorical data.

The second variable is the "exact mass of sugar in grams per $100\text{ g}$ of cereal". Since mass is measured and can include decimals (such as $5.2\text{ g}$ or $5.25\text{ g}$), it is a numerical variable. Because it can vary smoothly rather than only in whole numbers, it is continuous.

Option B says both variables are numerical discrete. This is wrong because the level descriptions ("Low", "Medium", "High") are categorical labels, not numbers. Also, mass is continuous rather than discrete, so we mark this option as eliminated.

Option C claims both variables are categorical. The first variable is categorical, but the "exact mass" is a numerical measurement that can be used for calculations, so this option is eliminated.

Option D correctly says one variable is categorical, but it calls the other numerical discrete. Discrete data comes from counting whole units (like number of students). Since mass can be fractional and recorded with decimals, it is continuous, so this option is eliminated.

After eliminating the other options, the remaining choice states that one variable is categorical (the level groups) and the other is numerical continuous (the exact mass). That matches the data types we identified.

We must decide why standard deviation is usually a stronger measure of variability (spread) than the range when datasets are large. To do that, we compare what each measure uses from the dataset and how well it reflects the overall distribution.

The range is a basic way to describe spread. It is found by subtracting the smallest value from the largest value:

$\text{Range}=\text{Maximum}-\text{Minimum}$

This calculation relies on only two data values. All the other data points between the minimum and maximum do not affect the result, so for a big dataset it may not represent the overall pattern very well.

The standard deviation ($\sigma$) looks at how far each data value is from the mean ($\bar{x}$) on average. From the formula you learn in Topic: Statistical Measures:

$\sigma =\sqrt{\frac{\sum (x_i-\bar{x}{)}^2}{n}}$

The summation sign ($\sum$) shows that the calculation includes every data value ($x_i$). That makes it much more informative about the overall spread of a large dataset than a measure based only on extremes.

Now we test the remaining options:

• Option B: Outliers affect both measures, but the range is determined completely by the most extreme values, so it is actually more easily changed by one extreme value than standard deviation.
• Option C: Standard deviation is measured in the same units as the original data; squared units belong to the variance, not the standard deviation.
• Option D: For large datasets, standard deviation involves many repeated calculations, while range is just a single subtraction, so it is not simpler by hand.

Comparing the definitions shows the key advantage: standard deviation uses information from the whole dataset, not just two endpoints. That makes it a more dependable measure of spread for large sets of data.

We must choose the statement that explains why standard deviation is usually favored over range for comparing variability in large datasets. We’ll eliminate options that are wrong about difficulty of calculation, units, or outlier effects.

Consider how much work each measure takes. The range is found with one subtraction: $\text{Range}=x_{max}-x_{min}$ Standard deviation requires finding the mean, computing each deviation, squaring, averaging, and taking a square root. With many data points, that is far more time-consuming by hand than finding the range, so Option D is not true.

Option C mixes up variance and standard deviation. Variance has squared units, but standard deviation is the square root of the variance: $\sigma =\sqrt{\frac{\sum (x_i-\bar{x}{)}^2}{n}}$ Taking the square root returns the value to the original units of the data, so it is not expressed in squared units. Therefore, Option C is incorrect.

The range depends only on the minimum and maximum values. Adding one extreme outlier can drastically change the range immediately. Standard deviation is influenced by outliers too, but it spreads that influence across all data values through averaging, so it is generally less dominated by one extreme value than the range. So Option B is not correct.

On a GDC, 1-Variable Statistics for a long list uses every entry in the list to compute standard deviation, which matches the idea that it reflects the whole dataset rather than just the endpoints.

After removing the incorrect statements, the remaining idea is the core statistical reason: range ignores all middle values, while standard deviation accounts for every value’s distance from the mean. That’s why standard deviation gives a fuller view of spread in large datasets.

To solve this using Elimination by Classification, we first decide what kind of variable we need.

• Numerical data (quantitative) is data written as numbers.
• Continuous data is numerical data that can take any value within an interval. It usually comes from measuring (such as height, mass, or time) and can include decimals or fractions.

Check each option for one that isn’t a number at all.

Option B: The brand of smartphone is a label or category (for example, different companies), not an amount. That makes it qualitative (categorical), so it cannot be numerical or continuous.

Now look for numerical data that is discrete (counted) rather than continuous.

Option A: How many books is a number, but it must be a whole number—you can’t have $20.5$ books. Since it’s counted in integers, it is discrete, not continuous.

Option D: The finishing position of a cyclist uses numbers, but those numbers mean order (1st, 2nd, 3rd). You can’t be in “$1.7^{th}$ place.” This is discrete data (and more specifically, ordinal/ranked).

The only option left is Option C: The temperature of a cup of tea.

Temperature is a measurement and can be recorded with decimals, such as $60$°C, $60.5$°C, or $60.57$°C. Because it is numerical and can vary smoothly within a range, it is numerical and continuous.

To solve this, we need to recall what numerical continuous data looks like:

1. Numerical: the values are numbers (quantitative).
2. Continuous: the variable is measured rather than counted, so it can take any value in a range, including fractions and decimals (e.g., $1.5$, $1.57$, $\mathrm{1.572...}$).

Consider the smartphone brands. A brand is a name/category, not a number. So it is qualitative (categorical) and cannot be numerical and continuous.

Now look at how many books are on the shelf. You might count $20$ or $21$, but you cannot have $20.5$ books. Since it counts separate items, it is discrete data, not continuous.

The cyclist’s finishing position ($1^{st}$, $2^{nd}$, $3^{rd}$) is a rank in order. These values jump from one integer to the next; you cannot finish in $1.27^{th}$ place. So this is also discrete.

Now consider temperature. Temperature is a measured quantity (like weight, height, or time). The exact reading depends on how precise the thermometer is. A cup of tea could be $60.75$°C or $60.752$°C.

Because it can take any value along a scale, it is continuous.

After checking each option, temperature is the only variable that is a measured numerical value that can vary continuously.