Transforming Random Variables — AP Statistics Study Guide

For: AP Statistics candidates sitting AP Statistics.

Covers: Linear transformations of discrete and continuous random variables, rules for the mean (expected value), variance, and standard deviation of transformed variables, contrast between linear and non-linear transformations, and common AP exam multiple-choice and free-response question applications.

You should already know: Basic probability and the definition of expected value and variance. How to work with probability distributions for discrete and continuous random variables. Basic unit conversion and linear equation manipulation.

A note on the practice questions: All worked questions in the "Practice Questions" section below are original problems written by us in the AP Statistics style for educational use. They are not reproductions of past College Board / Cambridge / IB papers and may differ in wording, numerical values, or context. Use them to practise the technique; cross-check with official mark schemes for grading conventions.

1. What Is Transforming Random Variables?

Transforming a random variable means changing every outcome of the original variable by a fixed mathematical function to produce a new, transformed random variable. This technique is extremely common in real-world statistics when we need to convert units, adjust for a fixed baseline measurement, or scale raw data for analysis. For example, if you have a random variable $X$ that measures temperature in degrees Celsius, you can transform it to Fahrenheit with $Y=\frac{9}{5}X+32$ to get a new random variable $Y$ with its own mean, variance, and distribution shape. According to the AP Statistics Course and Exam Description (CED), this topic is part of Unit 4, which accounts for 12–15% of the total AP exam score. Transforming random variables appears regularly in both multiple-choice (MCQ) and free-response (FRQ) sections, often combined with other topics like normal distributions or combining random variables. Linear transformations, which take the general form $Y=aX+b$ where $a$ and $b$ are fixed constants, are the primary focus of the AP exam. We also cover the key distinction between linear and non-linear transformations, a common tested concept that trips up many students.

2. Effect of Linear Transformations on Expected Value (Mean)

When you apply a linear transformation of the form $Y=aX+b$ to a random variable $X$, the expected value (mean) of $Y$ follows a simple rule that matches the transformation itself. To build intuition, we can derive this rule for a discrete random variable: the expected value of $X$ is defined as $E(X)={\mu }_X=\sum x_{i}P(X=x_i)$. For $Y=aX+b$, every outcome $y_i=a{x}_i+b$, so: $$E(Y)=\sum y_{i}P(Y=y_i)=\sum (a{x}_i+b)P(X=x_i)=a\sum x_{i}P(X=x_i)+b\sum P(X=x_i)$$ Since $\sum P(X=x_i)=1$ for any probability distribution, this simplifies to: $$E(Y)=a{\mu }_X+b$$ This rule makes intuitive sense: multiplying every outcome by $a$ scales the average by $a$, and adding $b$ shifts the entire distribution by $b$, so the average shifts by $b$ as well. This rule works for both discrete and continuous random variables, with no exceptions for linear transformations. It is most commonly tested in unit conversion or baseline adjustment problems on the AP exam.

Worked Example

A food scale adds a 0.05 kg fixed tare weight to every measured apple weight (measured in kg before tare correction). Let $X$ be the uncorrected measured weight, with ${\mu }_X=0.33$ kg. The corrected weight (in kg) is calculated by subtracting the tare, and the orchard owner wants the mean corrected weight reported in pounds, where 1 kg = 2.205 pounds. What is the mean corrected weight in pounds?

1. Write the full linear transformation from uncorrected kg $X$ to corrected pounds $Y$: $Y=2.205(X-0.05)=2.205X-0.11025$.
2. Apply the expected value rule for linear transformations: $E(Y)=a{\mu }_X+b$.
3. Substitute the known values: $a=2.205$, ${\mu }_X=0.33$, $b=-0.11025$.
4. Calculate the result: $E(Y)=2.205(0.33)-0.11025=0.72765-0.11025=0.6174$ pounds.

Exam tip: On AP FRQs, always write out the general formula $E(aX+b)=aE(X)+b$ before plugging in values — this earns you method points even if you make an arithmetic error.

3. Effect of Linear Transformations on Variance and Standard Deviation

After finding the mean of a transformed variable, the AP exam almost always asks for the variance or standard deviation (spread) of the new variable. The rules here differ from the mean rule because adding a constant $b$ does not change the spread of the distribution — it only shifts the entire distribution left or right, so all observations stay the same distance apart, and spread remains unchanged.

We can derive this rule from the definition of variance: $Var(Y)=E[(Y-{\mu }_Y{)}^2]$. Substituting ${\mu }_Y=a{\mu }_X+b$ gives $Y-{\mu }_Y=(aX+b)-(a{\mu }_X+b)=a(X-{\mu }_X)$. Squaring both sides and taking the expectation gives: $$Var(Y)=a^{2}E[(X-{\mu }_X{)}^2]=a^{2}Var(X)$$ For standard deviation, which is the square root of variance, we get: $${\sigma }_Y=\sqrt{Var(Y)}=|a|{\sigma }_X$$ We use the absolute value of $a$ because standard deviation is always non-negative, and $a$ is almost always positive in AP problems (as it is usually a unit conversion or scaling factor).

Worked Example

In the apple orchard problem above, the uncorrected weight $X$ has standard deviation ${\sigma }_X=0.04$ kg. Find the standard deviation of the corrected weight in pounds, where $Y=2.205X-0.11025$.

1. Recall the standard deviation rule for linear transformations: ${\sigma }_{aX+b}=|a|{\sigma }_X$.
2. Identify values: $a=2.205$, ${\sigma }_X=0.04$; the shift term $b=-0.11025$ does not affect spread.
3. Calculate the standard deviation: ${\sigma }_Y=2.205\times 0.04=0.0882$ pounds.
4. If asked for variance instead: $Var(Y)=(2.205{)}^2(0.04{)}^2=0.00778$ pounds squared.

Exam tip: AP exam questions frequently ask for standard deviation, not variance. Always double-check which measure the question requests — forgetting to take the square root of variance is one of the most common mistakes on this topic.

4. Linear vs Non-Linear Transformations

All the rules we have covered so far apply only to linear transformations of the form $Y=aX+b$. Non-linear transformations (e.g., $Y=X^2$, $Y=\sqrt{X}$, $Y=\ln (X)$) do not follow the linear transformation rules. In particular, for any non-linear function $g(X)$, the expected value of the transformation is almost never equal to the transformation of the expected value: $E(g(X))\ne g(E(X))$.

To find the expected value of a non-linearly transformed random variable, you must recalculate it from scratch using the original probability distribution: for discrete $X$, $E(g(X))=\sum g(x_i)P(X=x_i)$; for continuous $X$, it is calculated via integration. The AP exam often tests this concept by asking students to identify a common mistake or justify why a student's calculation is incorrect.

Worked Example

The area of a random square plot of farm land is given by $A=s^2$, where $s$ is the side length of the plot (in meters), a random variable with mean ${\mu }_s=10$ meters. A student claims that the mean area is $10^2=100$ square meters. Is this claim correct? Justify your answer.

1. The transformation $A=s^2$ is a non-linear transformation of the random variable $s$.
2. The rule ${\mu }_{g(X)}=g({\mu }_X)$ only holds for linear transformations; it does not apply to non-linear transformations.
3. For convex functions like $x^2$, $E(X^2)>(E(X){)}^2$, so the true mean area will be greater than 100 square meters.
4. Therefore, the student's claim is incorrect, because the linear expected value rule cannot be applied to this non-linear transformation.

Exam tip: If the transformation includes any exponent, square root, logarithm, or reciprocal that cannot be rewritten as $aX+b$, it is non-linear — never apply the linear transformation rules for mean or variance.

Common Pitfalls (and how to avoid them)

• Wrong move: Calculating the variance of $aX+b$ as $aVar(X)+b^2$ instead of $a^{2}Var(X)$. Why: Students confuse the mean rule with the variance rule, and incorrectly carry over the shift term and forget to square the scale factor. Correct move: Remind yourself before starting that shifting the distribution does not change spread, so $b$ disappears, and $a$ is always squared for variance.
• Wrong move: Applying the linear expected value rule $E(Y)=a{\mu }_X+b$ to non-linear transformations like $Y=X^2$. Why: Students memorize the linear rule so well they automatically apply it to any transformation without checking the form. Correct move: First check if the transformation can be rewritten as $Y=aX+b$. If not, do not use the linear rule; calculate $E[Y]$ from the original probability distribution.
• Wrong move: Adding $b$ to the standard deviation/variance: e.g., ${\sigma }_{X+b}={\sigma }_X+b$. Why: Students assume all parts of the transformation affect all statistics, just like the mean. Correct move: Memorize that adding a constant only changes measures of center (mean, median, quartiles), never measures of spread (variance, SD, IQR, range).
• Wrong move: Reversing the scale factor when converting units: e.g., converting kg to g using $a=0.001$ instead of 1000. Why: Students mix up the direction of unit conversion. Correct move: Always write out the conversion explicitly: $Y(\text{g})=1000\times X(\text{kg})$, so $a=1000$ before calculating.
• Wrong move: Claiming a linear transformation changes the shape of the original distribution. Why: Students think stretching the distribution changes its shape, but this is incorrect. Correct move: Linear transformations only scale and shift the distribution, they do not change its shape. If $X$ is normally distributed, $aX+b$ is also normally distributed.

Practice Questions (AP Statistics Style)

Question 1 (Multiple Choice)

The random variable $X$ has mean ${\mu }_X=12$ and variance ${\sigma }_X^2=16$. A new random variable is defined as $Y=3X-5$. What is the standard deviation of $Y$? A) 12 B) 43 C) 31 D) 144

Worked Solution: First, recall the rule for variance of a linear transformation: $Var(aX+b)=a^{2}Var(X)$. We have $a=3$ and $Var(X)=16$, so $Var(Y)=3^2\times 16=144$. Standard deviation is the square root of variance, so ${\sigma }_Y=\sqrt{144}=12$. The other options correspond to common mistakes: B uses the incorrect rule $aVar(X)+b$, C gives the mean of $Y$ instead of standard deviation, D gives the variance of $Y$ instead of standard deviation. The correct answer is A.

Question 2 (Free Response)

A small business sells custom printed t-shirts. The random variable $C$ represents the total production cost (in US dollars) for one t-shirt, with mean ${\mu }_C=\$7.50$ and standard deviation ${\sigma }_C=\$1.20$. The business calculates the selling price $S$ by marking up the cost by 60% and adding a $2 fixed shipping cost per t-shirt, so $S=1.6C+2$. (a) Find the mean selling price of a t-shirt. (b) Find the standard deviation of the selling price of a t-shirt. (c) The business owner wants to convert all prices to Canadian dollars, where 1 US dollar = 1.35 Canadian dollars. Write the transformation from $S$ (US dollars) to $S_{CAD}$ (Canadian dollars), then find the mean and standard deviation of $S_{CAD}$.

Worked Solution: (a) For the linear transformation $S=1.6C+2$, apply the mean rule: $${\mu }_S=1.6{\mu }_C+2=1.6(7.50)+2=12+2=\$14.00$$ The mean selling price is $\$14.00$.

(b) Apply the standard deviation rule: the shift term of $2 does not affect spread, so: $${\sigma }_S=1.6{\sigma }_C=1.6(1.20)=1.92\text{ US dollars}$$ The standard deviation of the selling price is $\$1.92$.

(c) The transformation is $S_{CAD}=1.35S$, which simplifies to $S_{CAD}=1.35(1.6C+2)=2.16C+2.7$, a linear transformation. Mean: ${\mu }_{S_{CAD}}=1.35{\mu }_S=1.35(14.00)=18.90$ Canadian dollars. Standard deviation: ${\sigma }_{S_{CAD}}=1.35{\sigma }_S=1.35(1.92)=2.592$ Canadian dollars.

Question 3 (Application / Real-World Style)

A biologist measures the length of trout in a lake in inches, where the random variable $L$ (length of a random trout) has mean 12.5 inches and standard deviation 2.1 inches. The biologist uses two common calculations for their data: (1) converting all lengths to centimeters for publication, and (2) estimating weight from length using the empirical formula $W=0.01L^3$, where $W$ is weight in pounds and $L$ is length in inches. (a) Find the mean length in centimeters (1 inch = 2.54 cm). (b) A student intern calculates the mean weight as $0.01(12.5{)}^3=19.53$ pounds. Is this calculation correct? Justify your answer in context.

Worked Solution: (a) Converting length from inches to centimeters is a linear transformation: $L_{cm}=2.54L$. The mean length in centimeters is: $${\mu }_{L_{cm}}=2.54{\mu }_L=2.54(12.5)=31.75\text{ centimeters}$$

(b) The weight transformation $W=0.01L^3$ is non-linear, so the rule $E[g(X)]=g(E[X])$ that applies to linear transformations cannot be used here. The student incorrectly applied the linear transformation rule to a non-linear transformation, so their calculation is wrong. For this convex transformation, the true mean weight will be higher than 19.53 pounds, and we need the full distribution of lengths to calculate the correct mean weight.

In context: We cannot use the average length to directly calculate the average weight with this non-linear length-weight formula.

Quick Reference Cheatsheet

What's Next

Transforming random variables is a critical prerequisite for the next core topic in Unit 4: combining independent random variables. Without mastering the rules for linear transformations, you will not be able to correctly calculate the mean and variance of sums or differences of random variables, a heavily tested topic on both MCQ and FRQ. This topic also underpins all of inferential statistics: when we calculate confidence intervals or run hypothesis tests for means, we rely on properties of transformed sampling distributions, which build directly on the rules for linear transformations you learned here. Beyond the exam, these rules are used constantly in data analysis for unit conversion and data scaling.

Combining Independent Random Variables Normal Probability Distributions Sampling Distributions for Means