IB Math AA: Probability Distributions

Probability distributions are the mathematical engines that power predictions, from forecasting election results to modeling genetic inheritance. In IB Math Analysis and Approaches, mastering these distributions moves you from counting simple outcomes to quantifying real-world uncertainty with precision. This knowledge is not just exam-critical; it forms the foundational language for data science, economics, and experimental research.

Discrete Foundations: The Binomial Distribution

Many real-world scenarios involve a fixed number of independent trials, each with the same two possible outcomes: success or failure. The binomial distribution is the discrete probability model for exactly these situations. It is defined by two parameters: $n$, the fixed number of trials, and $p$, the constant probability of success on a single trial.

A random variable $X$ following a binomial distribution is written as $X\sim \text{B}(n,p)$. The probability of achieving exactly $r$ successes is given by the formula: $$P(X=r)=\left(\genfrac{}{}{0ex}{}{n}{r}\right)p^r(1-p{)}^{n-r}$$ where $\left(\genfrac{}{}{0ex}{}{n}{r}\right)=\frac{n!}{r!(n-r)!}$ is the binomial coefficient, calculating the number of ways to choose $r$ successes from $n$ trials.

The mean, or expected value, and the variance of a binomial distribution are derived from its parameters: $$E(X)=np$$ $$\text{Var}(X)=np(1-p)$$ The standard deviation is simply the square root of the variance: $\sigma =\sqrt{np(1-p)}$. For example, if a fair die is rolled 60 times ($n=60,p=\frac{1}{6}$), the expected number of fours is $E(X)=60\times \frac{1}{6}=10$, with a standard deviation of $\sqrt{60\times \frac{1}{6}\times \frac{5}{6}}\approx 2.89$.

Continuous Modeling: The Normal Distribution

While the binomial distribution handles discrete counts, many natural phenomena—like heights, exam scores, or measurement errors—are continuous and often cluster around a central value. The normal distribution models these continuous data with its iconic bell-shaped curve, symmetric about its mean.

A normal distribution is defined by its mean $\mu$ (the center) and its standard deviation $\sigma$ (the spread), denoted $X\sim \text{N}(\mu ,{\sigma }^2)$. The total area under its probability density function is 1. To find probabilities, you must standardise the variable. This transforms any normal distribution into the standard normal distribution $Z\sim \text{N}(0,1)$ using the $z$-score formula: $$z=\frac{x-\mu }{\sigma }$$ The $z$-score tells you how many standard deviations an observation $x$ is from the mean. You then use your GDC or $z$-tables to find probabilities like $P(Z<z)$.

Often, you need to work backwards from a probability to find a corresponding data value, which is an inverse normal calculation. For instance, if you know $P(X<k)=0.95$, you first find the $z$-score where $P(Z<z)=0.95$ (approximately 1.645), and then "un-standardise" using $x=\mu +z\sigma$.

A crucial application is using the normal distribution to approximate a binomial distribution when $n$ is large. This requires a continuity correction because you are approximating a discrete distribution (binomial) with a continuous one (normal). If you want $P(X\le 10)$ for a binomial variable, you approximate it as $P(X_{normal}<10.5)$.

HL Extension: The Poisson Distribution and Hypothesis Testing

At Higher Level, you encounter the Poisson distribution, which models the number of events occurring in a fixed interval of time or space. These events must occur independently and at a known constant average rate, $\lambda$. Examples include calls received by a call center per hour or typos per page. If $Y\sim \text{Po}(\lambda )$, then the probability of $r$ events is: $$P(Y=r)=\frac{e^{-\lambda }{\lambda }^r}{r!}$$ Its expected value and variance are both equal to $\lambda$: $E(Y)=\text{Var}(Y)=\lambda$.

These distributions become powerful tools for formal hypothesis testing. This statistical method allows you to make inferences about a population parameter based on sample data. A typical test for a population mean using the normal distribution involves five steps:

1. State the null hypothesis ($H_0:\mu ={\mu }_0$) and the alternative hypothesis ($H_1:\mu \ne ,>,<{\mu }_0$).
2. Choose the significance level $\alpha$ (commonly 5%).
3. Calculate the test statistic (e.g., a $z$-score from sample data).
4. Determine the $p$-value—the probability of obtaining results at least as extreme as the sample, assuming $H_0$ is true.
5. Compare the $p$-value to $\alpha$. If $p\le \alpha$, you reject the null hypothesis.

The $p$-value is the cornerstone of this decision. A small $p$-value provides evidence against the null hypothesis, suggesting your sample result is unlikely to have occurred by chance alone.

Common Pitfalls

1. Misidentifying the Distribution: Assuming data is binomial when trials are not independent (like drawing cards without replacement) is a frequent error. Similarly, applying Poisson to events that are not independent or whose rate is not constant will lead to incorrect probabilities. Always check the conditions before selecting your model.
2. Misusing the Continuity Correction: When using a normal distribution to approximate a binomial, forgetting the continuity correction leads to significant inaccuracy. Remember: for $P(X\le a)$, use $a+0.5$; for $P(X\ge b)$, use $b-0.5$; for $P(a\le X\le b)$, use $(a-0.5)$ to $(b+0.5)$.
3. Confusing $p$-value and Significance Level: The significance level $\alpha$ is a pre-set threshold for rejection (e.g., 0.05). The $p$-value is a calculated probability from your sample data. You do not "accept" the null hypothesis; you either reject it or fail to reject it based on the comparison.
4. Incorrect Inverse Normal Setup: When performing an inverse normal calculation, students often mix up which probability to input. If a question asks for the top 10%, you need to use $P(Z>z)=0.10$ or, equivalently, $P(Z<z)=0.90$ when looking up the critical $z$-value on their GDC.

Summary

• The binomial distribution $\text{B}(n,p)$ models discrete counts of success in $n$ independent trials, with $E(X)=np$ and $\text{Var}(X)=np(1-p)$.
• The normal distribution $\text{N}(\mu ,{\sigma }^2)$ models continuous data; standardisation via $z=\frac{x-\mu }{\sigma }$ is essential for finding probabilities and performing inverse calculations.
• Hypothesis testing is a structured process using a test statistic to generate a $p$-value, which is compared to a significance level $\alpha$ to decide whether to reject a null hypothesis.
• HL Only: The Poisson distribution $\text{Po}(\lambda )$ models the count of independent events occurring at a constant rate, where $E(Y)=\text{Var}(Y)=\lambda$.
• Always verify the conditions for each distribution (independence, fixed $n$ and $p$, constant $\lambda$) and remember the continuity correction when approximating binomial with normal.