UK A-Level: Statistical Distributions

Understanding statistical distributions provides the mathematical language to model uncertainty, predict outcomes, and make informed decisions from data. For A-Level Mathematics and Statistics, mastering the binomial and normal distributions is a foundational skill for interpreting the world, from quality control in manufacturing to analyzing survey results.

The Binomial Distribution: Modelling Success or Failure

The binomial distribution is used to model experiments where you count the number of "successes" in a fixed number of independent trials. For a situation to be modelled binomially, four conditions must be met:

1. There is a fixed number of trials, $n$.
2. Each trial has only two possible outcomes: success or failure.
3. The probability of success, $p$, is constant for each trial.
4. The trials are independent of each other.

If these conditions hold, we say that the random variable $X$, the number of successes, follows a binomial distribution: $X\sim B(n,p)$.

The probability of getting exactly $k$ successes is given by the binomial formula: $$P(X=k)=\left(\genfrac{}{}{0ex}{}{n}{k}\right)p^k(1-p{)}^{n-k}$$ Here, $\left(\genfrac{}{}{0ex}{}{n}{k}\right)$ (read as "n choose k") is the binomial coefficient, calculated as $\frac{n!}{k!(n-k)!}$. It accounts for the number of different ways $k$ successes can occur in $n$ trials.

Calculating binomial probabilities often involves using this formula directly or employing cumulative tables from your formula booklet. For example, if a fair die is rolled 10 times, and success is defined as rolling a six, then $X\sim B(10,\frac{1}{6})$. The probability of getting exactly two sixes is: $$P(X=2)=\left(\genfrac{}{}{0ex}{}{10}{2}\right){\left(\frac{1}{6}\right)}^2{\left(\frac{5}{6}\right)}^8\approx 0.2907$$ To find the probability of getting at most two sixes, you would sum the probabilities for $k=0,1,$ and $2$: $P(X\le 2)=P(X=0)+P(X=1)+P(X=2)$.

The Normal Distribution: The Bell-Shaped Curve

While the binomial distribution deals with discrete counts, the normal distribution models continuous data. It is defined by two parameters: the mean, $\mu$ (which locates the centre of the distribution), and the variance, ${\sigma }^2$ (which measures its spread). We write $X\sim N(\mu ,{\sigma }^2)$.

Its key properties are:

• It is symmetrical and bell-shaped about the mean $\mu$.
• The mean, median, and mode are all equal.
• Approximately 68% of the data lies within $\mu \pm \sigma$, 95% within $\mu \pm 2\sigma$, and 99.7% within $\mu \pm 3\sigma$ (the empirical rule).
• The curve extends infinitely in both directions, approaching but never touching the horizontal axis.

To find probabilities for any normal distribution, we must first standardize it. This process converts our normal variable $X$ into a standard normal variable $Z$, which has a mean of 0 and a variance of 1: $Z\sim N(0,1)$. Standardization is done using the formula: $$Z=\frac{X-\mu }{\sigma }$$ This Z-score tells you how many standard deviations an observation $X$ is above or below the mean. You then use the standard normal distribution tables to find probabilities like $P(Z<z)$.

Working Backwards: Inverse Normal Calculations

Often, you are given a probability and need to find the corresponding value of $X$. This is an inverse normal calculation. The process is straightforward:

1. Use the standard normal probability tables in reverse to find the Z-score ($z$) that corresponds to the given cumulative probability.
2. Substitute the known $z$, $\mu$, and $\sigma$ into the standardization formula rearranged to solve for $X$:

$$X=\mu +z\sigma$$ For instance, if heights are normally distributed with $\mu =170$ cm and $\sigma =10$ cm, and we want the height that marks the top 10%, we first find the Z-score for a cumulative probability of 0.90 (approximately $z=1.2816$). Then, $X=170+1.2816\times 10=182.8$ cm. Therefore, the top 10% of people are taller than approximately 182.8 cm.

Connecting the Two: Normal Approximation to the Binomial

When the number of trials $n$ in a binomial distribution is large, calculating exact probabilities can be cumbersome. Fortunately, under certain conditions, the binomial distribution $B(n,p)$ can be approximated by a normal distribution $N(\mu ,{\sigma }^2)$, where $\mu =np$ and ${\sigma }^2=np(1-p)$.

The conditions for a valid normal approximation are that both $np>5$ and $n(1-p)>5$. When these are satisfied, $B(n,p)\approx N(np,np(1-p))$.

A critical step in this approximation is applying a continuity correction. Because the binomial is discrete (whole numbers) and the normal is continuous, we adjust our boundaries by 0.5 to improve accuracy. For example:

• $P(X\le 10)$ for the binomial becomes $P(X\le 10.5)$ for the normal.
• $P(X<10)$ becomes $P(X\le 9.5)$.
• $P(X\ge 15)$ becomes $P(X\ge 14.5)$.
• $P(X=8)$ becomes $P(7.5\le X\le 8.5)$.

You then standardize the corrected value (e.g., 10.5) using the normal parameters $\mu =np$ and $\sigma =\sqrt{np(1-p)}$ to find the Z-score and the corresponding probability from the tables.

Common Pitfalls

1. Misapplying the Binomial Conditions: The most common error is forcing a binomial model onto a scenario that violates independence or constant probability. For example, selecting cards from a pack without replacement changes probabilities, making the trials dependent. Always check all four conditions first.
2. Forgetting the Continuity Correction: When using the normal approximation to the binomial, omitting the $\pm 0.5$ adjustment leads to significant inaccuracies. Remember: you are fitting a continuous curve over discrete blocks; the correction accounts for this.
3. Confusing Variance and Standard Deviation in the Z-formula: The standardization formula uses the standard deviation $\sigma$, not the variance ${\sigma }^2$. A frequent mistake is to calculate $Z=(X-\mu )/{\sigma }^2$. Always use $Z=(X-\mu )/\sigma$.
4. Misinterpreting Inverse Normal Problems: When asked, "find the value that 30% of data exceeds," this means find $x$ such that $P(X>x)=0.30$. This is equivalent to $P(X<x)=0.70$. Carefully translate the wording into a cumulative probability less than $x$ before looking up the Z-score.

Summary

• The binomial distribution $B(n,p)$ models discrete counts of success under strict conditions: fixed $n$, constant $p$, binary outcomes, and independence. Probabilities are calculated using the formula $P(X=k)=\left(\genfrac{}{}{0ex}{}{n}{k}\right)p^k(1-p{)}^{n-k}$.
• The normal distribution $N(\mu ,{\sigma }^2)$ is a continuous, symmetric bell-shaped curve defined by its mean and variance. Probabilities are found by standardizing to a Z-score: $Z=(X-\mu )/\sigma$.
• Inverse normal calculations allow you to find a data value corresponding to a given probability by rearranging the standardization formula: $X=\mu +z\sigma$.
• For large binomial trials (where $np>5$ and $n(1-p)>5$), you can use the normal approximation $N(np,np(1-p))$. This must include a continuity correction ($\pm 0.5$) to account for the shift from discrete to continuous.
• Always double-check the conditions for each distribution and handle the transition between them with care, as this is a frequent source of examination errors.