Statistics is the discipline of collecting, organizing, analyzing, and interpreting data to make informed decisions under uncertainty. Probability, the mathematical foundation of statistics, quantifies how likely events are to occur. Together they provide the tools for reasoning rigorously about patterns in data, testing hypotheses, and making predictions with measurable confidence.

These tools are relevant to software engineers for reasons that go beyond data science. Performance benchmarking requires confidence intervals. A/B testing requires hypothesis tests. Capacity planning requires an understanding of distributions. Anomaly detection requires knowing what “normal” looks like statistically. Any engineer who makes decisions based on data benefits from a working knowledge of probability and statistics.

This article is a reference, not a tutorial. It collects the formulas and definitions most commonly needed in applied work, organized for quick lookup. Each section includes the notation, the formula, and a brief explanation of when and why to use it. The level corresponds to an introductory undergraduate course in probability and statistics for engineers.

Software Versions

# Date (UTC)
$ date -u "+%Y-%m-%d %H:%M:%S +0000"
2026-02-11 07:34:17 +0000

# OS and Version
$ uname -vm
Darwin Kernel Version 23.6.0: Mon Jul 29 21:14:30 PDT 2024; root:xnu-10063.141.2~1/RELEASE_ARM64_T6000 arm64

$ sw_vers
ProductName:		macOS
ProductVersion:		14.6.1
BuildVersion:		23G93

# Hardware Information
$ system_profiler SPHardwareDataType | sed -n '8,10p'
      Chip: Apple M1 Max
      Total Number of Cores: 10 (8 performance and 2 efficiency)
      Memory: 32 GB

# Shell and Version
$ echo "${SHELL}"
/bin/bash

$ "${SHELL}" --version | head -n 1
GNU bash, version 3.2.57(1)-release (arm64-apple-darwin23)

# Claude Code Installation Versions
$ claude --version
2.1.37 (Claude Code)

Notation

The following notation is used throughout this reference.

Symbol Meaning
$n$ Sample size or number of trials
$k$ Number of successes or events
$p$ Probability of success on a single trial
$q$ Probability of failure, $q = 1 - p$
$\mu$ Population mean
$\sigma$ Population standard deviation
$\sigma^2$ Population variance
$\bar{x}$ Sample mean
$s$ Sample standard deviation
$s^2$ Sample variance
$\hat{p}$ Sample proportion (point estimate of $p$)
$\alpha$ Significance level
$Z$ Standard normal Z-score
$z^\star$ Critical value from the standard normal distribution
$t^\star_{df}$ Critical value from the $t$-distribution with $df$ degrees of freedom
$\chi^2$ Chi-squared statistic
$H_0$ Null hypothesis
$H_A$ Alternative hypothesis
$SE$ Standard error
$ME$ Margin of error

Probability Distributions

Binomial Distribution

The binomial distribution models the number of successes $k$ in $n$ independent trials, each with success probability $p$.

The Probability Mass Function (PMF) gives the probability of exactly $k$ successes.

\[P(X = k) = \binom{n}{k} p^k q^{n-k}, \quad \binom{n}{k} = \frac{n!}{(n-k)!\, k!} \tag{1}\]

The mean and standard deviation of the binomial distribution are

\[\mu = np, \quad \sigma = \sqrt{npq}. \tag{2}\]

Use the binomial distribution when counting discrete successes in a fixed number of independent trials with constant success probability.

Normal Distribution

The normal distribution is a continuous probability distribution defined by its mean $\mu$ and standard deviation $\sigma$. Its Probability Density Function (PDF) is

\[f(x) = \frac{1}{\sigma\sqrt{2\pi}} \exp\left(-\frac{(x - \mu)^2}{2\sigma^2}\right). \tag{3}\]

The standard normal distribution has $\mu = 0$ and $\sigma = 1$. Any normal random variable $X \sim N(\mu, \sigma^2)$ can be standardized to a Z-score.

\[Z = \frac{x - \mu}{\sigma} \tag{4}\]

The normal distribution arises throughout statistics because of the Central Limit Theorem and because many natural processes produce approximately normal data.

Poisson Distribution

The Poisson distribution models the number of events $k$ occurring in a fixed interval of time or space, given a known average rate $\lambda$.

\[P(X = k) = \frac{e^{-\lambda} \lambda^k}{k!}, \quad k = 0, 1, 2, \ldots \tag{5}\]

The mean and variance of the Poisson distribution are both equal to $\lambda$.

\[\mu = \lambda, \quad \sigma^2 = \lambda \tag{6}\]

Use the Poisson distribution for count data where events occur independently at a constant average rate. Common applications include modeling server request arrivals, error counts per time period, and rare event occurrences.

Normal Approximation to the Binomial

When the number of trials $n$ is large enough, the binomial distribution can be approximated by the normal distribution. The approximation is valid when both

\[np \geq 5 \quad \text{and} \quad nq \geq 5. \tag{7}\]

Under these conditions, the binomial random variable $X$ is approximately normal with mean $\mu = np$ and standard deviation $\sigma = \sqrt{npq}$.

A continuity correction improves accuracy when approximating a discrete distribution with a continuous one. The correction adds or subtracts 0.5 from the boundary value.

Discrete Probability Normal Approximation
$P(X < k)$ $P\left(Z \leq \dfrac{k - 0.5 - \mu}{\sigma}\right)$
$P(X \leq k)$ $P\left(Z \leq \dfrac{k + 0.5 - \mu}{\sigma}\right)$
$P(X = k)$ $P\left(\dfrac{k - 0.5 - \mu}{\sigma} \leq Z \leq \dfrac{k + 0.5 - \mu}{\sigma}\right)$
$P(X \geq k)$ $P\left(Z \geq \dfrac{k - 0.5 - \mu}{\sigma}\right)$
$P(X > k)$ $P\left(Z \geq \dfrac{k + 0.5 - \mu}{\sigma}\right)$

Descriptive Statistics

Sample Mean

The sample mean $\bar{x}$ estimates the population mean $\mu$.

\[\bar{x} = \frac{1}{n}\sum_{i=1}^{n} x_i \tag{8}\]

Sample Variance and Standard Deviation

The sample variance $s^2$ estimates the population variance $\sigma^2$. The denominator uses $n - 1$ rather than $n$ to correct for the bias introduced by estimating the mean from the same sample. This correction is known as Bessel’s correction.

\[s^2 = \frac{1}{n-1}\sum_{i=1}^{n}(x_i - \bar{x})^2 \tag{9}\]

The sample standard deviation is $s = \sqrt{s^2}$.

Population Variance and Standard Deviation

When the full population is known, the population variance divides by $n$ rather than $n - 1$.

\[\sigma^2 = \frac{1}{n}\sum_{i=1}^{n}(x_i - \mu)^2 \tag{10}\]

The population standard deviation is $\sigma = \sqrt{\sigma^2}$.

The Law of Total Probability

The law of total probability expresses the probability of an event $E$ as a weighted sum over mutually exclusive conditions. For two conditions $A$ and $B$ that partition the sample space,

\[P(E) = P(E \mid A) \cdot P(A) + P(E \mid B) \cdot P(B). \tag{11}\]

Rearranging to solve for a conditional probability or a partition probability,

\[P(E \mid A) = \frac{P(E) - P(E \mid B) \cdot P(B)}{P(A)}, \quad P(A) = \frac{P(E) - P(E \mid B) \cdot P(B)}{P(E \mid A)}. \tag{12}\]

Bayes’ Theorem

Bayes’ theorem relates a conditional probability to its reverse conditional. Given events $A$ and $B$ where $P(B) > 0$,

\[P(A \mid B) = \frac{P(B \mid A) \cdot P(A)}{P(B)}. \tag{13}\]

The denominator can be expanded using the law of total probability.

\[P(A \mid B) = \frac{P(B \mid A) \cdot P(A)}{P(B \mid A) \cdot P(A) + P(B \mid \neg A) \cdot P(\neg A)} \tag{14}\]

Bayes’ theorem is the foundation of Bayesian statistics and is used whenever prior knowledge must be updated with new evidence. Common applications include spam filtering, medical diagnostic reasoning, and anomaly detection.

The Central Limit Theorem

The Central Limit Theorem (CLT) states that the sampling distribution of the sample mean $\bar{x}$ approaches a normal distribution as the sample size $n$ increases, regardless of the shape of the underlying population distribution. Specifically,

\[\bar{X} \sim N\left(\mu, \frac{\sigma^2}{n}\right) \quad \text{approximately, for large } n. \tag{15}\]

The standard error of the mean is

\[SE_{\bar{x}} = \frac{\sigma}{\sqrt{n}}. \tag{16}\]

In practice, $n \geq 30$ is a common rule of thumb for the CLT approximation to be adequate, though the required sample size depends on how far the underlying distribution departs from normality.

The CLT justifies the widespread use of normal-based inference methods even when the underlying data is not normally distributed.

Hypothesis Testing

Framework

A hypothesis test evaluates a claim about a population parameter.

The null hypothesis $H_0$ is the default assumption, typically a statement of no effect or no difference. The alternative hypothesis $H_A$ is the claim being tested.

The significance level $\alpha$ is the probability threshold below which the null hypothesis is rejected. Common choices are $\alpha = 0.05$ and $\alpha = 0.01$.

Type I and Type II Errors

Decision $H_0$ True $H_A$ True
Reject $H_0$ Type I Error (rate $\alpha$) Correct
Fail to Reject $H_0$ Correct Type II Error (rate $\beta$)

The power of a test is $1 - \beta$, the probability of correctly rejecting $H_0$ when $H_A$ is true.

Z-Test for a Population Mean

When the population standard deviation $\sigma$ is known, the test statistic is

\[Z = \frac{\bar{x} - \mu_0}{\sigma / \sqrt{n}} \tag{17}\]

where $\mu_0$ is the hypothesized population mean under $H_0$.

T-Test for a Population Mean

When the population standard deviation is unknown and must be estimated by the sample standard deviation $s$, the test statistic follows a $t$-distribution with $n - 1$ degrees of freedom.

\[t = \frac{\bar{x} - \mu_0}{s / \sqrt{n}} \tag{18}\]

The $t$-distribution has heavier tails than the standard normal, reflecting the additional uncertainty from estimating $\sigma$ with $s$. As $n$ grows large, the $t$-distribution approaches the standard normal.

Z-Test for a Proportion

For testing a hypothesis about a population proportion $p$, the test statistic uses the null value $p_0$.

\[Z = \frac{\hat{p} - p_0}{SE_{\hat{p}}}, \quad SE_{\hat{p}} = \sqrt{\frac{p_0(1 - p_0)}{n}} \tag{19}\]

The normal approximation requires at least ten expected successes $np_0 \geq 10$ and ten expected failures $n(1 - p_0) \geq 10$.

General Form

For any point estimate $\hat{\theta}$ and null value $\theta_0$,

\[Z = \frac{\hat{\theta} - \theta_0}{SE_{\hat{\theta}}}. \tag{20}\]

Confidence Intervals

A confidence interval provides a range of plausible values for a population parameter. A $(1 - \alpha) \times 100\%$ confidence interval means that if the procedure were repeated many times, approximately $(1 - \alpha) \times 100\%$ of the intervals would contain the true parameter value.

General Form

\[\hat{\theta} \pm z^\star \cdot SE_{\hat{\theta}}, \quad ME = z^\star \cdot SE_{\hat{\theta}} \tag{21}\]

where $ME$ is the margin of error.

Confidence Interval for a Mean (Known Variance)

\[\bar{x} \pm z^\star_{\alpha/2} \cdot \frac{\sigma}{\sqrt{n}} \tag{22}\]

Confidence Interval for a Mean (Unknown Variance)

When $\sigma$ is unknown, use the sample standard deviation $s$ and the $t$-distribution with $n - 1$ degrees of freedom.

\[\bar{x} \pm t^\star_{\alpha/2,\, n-1} \cdot \frac{s}{\sqrt{n}} \tag{23}\]

Confidence Interval for a Variance

Uses the chi-squared distribution with $n - 1$ degrees of freedom.

\[\frac{(n-1)s^2}{\chi^2_{\alpha/2}} < \sigma^2 < \frac{(n-1)s^2}{\chi^2_{1 - \alpha/2}} \tag{24}\]

For the standard deviation,

\[\sqrt{\frac{(n-1)s^2}{\chi^2_{\alpha/2}}} < \sigma < \sqrt{\frac{(n-1)s^2}{\chi^2_{1 - \alpha/2}}} \tag{25}\]

Confidence Interval for a Proportion

\[\hat{p} \pm z^\star \cdot SE_{\hat{p}}, \quad SE_{\hat{p}} = \sqrt{\frac{\hat{p}(1 - \hat{p})}{n}} \tag{26}\]

The normal approximation requires at least ten successes $n\hat{p} \geq 10$ and ten failures $n(1 - \hat{p}) \geq 10$ in the sample.

Confidence Interval for Difference of Means

For two independent populations with known variances,

\[(\bar{x}_1 - \bar{x}_2) \pm z^\star_{\alpha/2} \cdot \sqrt{\frac{\sigma_1^2}{n_1} + \frac{\sigma_2^2}{n_2}} \tag{27}\]

Confidence Interval for Difference of Proportions

The standard error for the difference of two independent proportions is

\[SE_{\hat{p}_1 - \hat{p}_2} = \sqrt{\frac{\hat{p}_1(1 - \hat{p}_1)}{n_1} + \frac{\hat{p}_2(1 - \hat{p}_2)}{n_2}}. \tag{28}\]

The confidence interval is

\[(\hat{p}_1 - \hat{p}_2) \pm z^\star_{\alpha/2} \cdot SE_{\hat{p}_1 - \hat{p}_2}. \tag{29}\]

Pooled Proportion for Hypothesis Testing

When testing $H_0: p_1 = p_2$, the pooled proportion estimate uses the combined data from both samples.

\[\hat{p} = \frac{\hat{p}_1 n_1 + \hat{p}_2 n_2}{n_1 + n_2} \tag{30}\]

The pooled standard error uses $\hat{p}$ in place of the individual sample proportions.

\[SE_{\text{pooled}} = \sqrt{\hat{p}(1 - \hat{p})\left(\frac{1}{n_1} + \frac{1}{n_2}\right)} \tag{31}\]

Sample Size Determination

The required sample size to achieve a desired margin of error can be computed before data collection.

Sample Size for Estimating a Proportion

\[n \geq \frac{(z^\star)^2 \cdot p(1-p)}{ME^2} \tag{32}\]

When no prior estimate of $p$ is available, use $p = 0.5$ as the worst-case estimate. This maximizes $p(1-p) = 0.25$ and guarantees that the actual margin of error will not exceed the target.

Sample Size for Estimating a Mean

\[n \geq \left(\frac{z^\star \cdot \sigma}{ME}\right)^2 \tag{33}\]

This requires a preliminary estimate of $\sigma$, which can come from a pilot study, prior research, or the range divided by four as a rough approximation.

Summary

This reference covers the probability distributions and statistical methods most commonly used in applied work. The binomial, normal, and Poisson distributions model discrete counts, continuous measurements, and event rates respectively. Confidence intervals and hypothesis tests provide the framework for making decisions under uncertainty. The Central Limit Theorem and the normal approximation to the binomial justify the use of normal-based methods across a wide range of settings.

The formulas collected here are standard. They appear in every introductory statistics textbook and are implemented in every major statistical software package. The value of knowing them directly is that they make the assumptions and limitations of each method explicit. A confidence interval for a proportion requires at least ten successes and ten failures. A $t$-test requires approximately normal data. A chi-squared interval for variance requires normal data. Understanding the formulas means understanding when each method can and cannot be trusted.

Future Reading

Equations

This section collects the 33 display equations of the article in the order they appear, each under its number and reproduced exactly as it stands in the text. A title in bold names an established result, and a title in italics beginning with the word Bespoke names a construction made for this analysis. Of the 33, 33 are established and 0 are bespoke. Probabilities and proportions are dimensionless numbers between zero and one, a mean, standard deviation or standard error carries the units of the data, and a critical value $z^\star$, $t^\star$ or $\chi^2$ is read from a table at the stated significance level.

Equation 1, Binomial Probability Mass Function. The probability of exactly $k$ successes in $n$ independent trials is the binomial coefficient multiplied by $p^k q^{n-k}$.

\[P(X = k) = \binom{n}{k} p^k q^{n-k}, \quad \binom{n}{k} = \frac{n!}{(n-k)!\, k!} \tag{1}\]

Equation 2, Mean and Standard Deviation of the Binomial Distribution. A binomial count has mean $np$ and standard deviation $\sqrt{npq}$.

\[\mu = np, \quad \sigma = \sqrt{npq}. \tag{2}\]

Equation 3, Normal Probability Density Function. The normal density is fixed by its mean $\mu$ and standard deviation $\sigma$ and falls off as the exponential of the negative squared standardised distance.

\[f(x) = \frac{1}{\sigma\sqrt{2\pi}} \exp\left(-\frac{(x - \mu)^2}{2\sigma^2}\right). \tag{3}\]

Equation 4, Z-Score. Subtracting the mean and dividing by the standard deviation standardises a normal value to the standard normal distribution.

\[Z = \frac{x - \mu}{\sigma} \tag{4}\]

Equation 5, Poisson Probability Mass Function. The probability of $k$ events in an interval with average rate $\lambda$ is $e^{-\lambda} \lambda^k / k!$.

\[P(X = k) = \frac{e^{-\lambda} \lambda^k}{k!}, \quad k = 0, 1, 2, \ldots \tag{5}\]

Equation 6, Mean and Variance of the Poisson Distribution. Both the mean and the variance of a Poisson count equal the rate $\lambda$.

\[\mu = \lambda, \quad \sigma^2 = \lambda \tag{6}\]

Equation 7, Condition for the Normal Approximation to the Binomial. The conventional rule of thumb requires at least five expected successes and five expected failures before the binomial is approximated by the normal distribution.

\[np \geq 5 \quad \text{and} \quad nq \geq 5. \tag{7}\]

Equation 8, Sample Mean. The sample mean is the arithmetic average of the $n$ observations.

\[\bar{x} = \frac{1}{n}\sum_{i=1}^{n} x_i \tag{8}\]

Equation 9, Sample Variance with Bessel’s Correction. The sample variance divides the sum of squared deviations from the sample mean by $n-1$, which removes the bias of estimating the mean from the same sample.

\[s^2 = \frac{1}{n-1}\sum_{i=1}^{n}(x_i - \bar{x})^2 \tag{9}\]

Equation 10, Population Variance. When the whole population is known, the variance divides the sum of squared deviations from the population mean by the population size.

\[\sigma^2 = \frac{1}{n}\sum_{i=1}^{n}(x_i - \mu)^2 \tag{10}\]

Equation 11, Law of Total Probability over Two Conditions. When two events $A$ and $B$ partition the sample space, the probability of $E$ is the sum of its conditional probabilities weighted by the probabilities of the conditions.

\[P(E) = P(E \mid A) \cdot P(A) + P(E \mid B) \cdot P(B). \tag{11}\]

Equation 12, Law of Total Probability Rearranged. Solving the two-condition law of total probability for one conditional probability or for one partition probability gives these two forms.

\[P(E \mid A) = \frac{P(E) - P(E \mid B) \cdot P(B)}{P(A)}, \quad P(A) = \frac{P(E) - P(E \mid B) \cdot P(B)}{P(E \mid A)}. \tag{12}\]

Equation 13, Bayes’ Theorem. The probability of $A$ given $B$ equals the probability of $B$ given $A$ multiplied by the prior of $A$ and divided by the probability of $B$.

\[P(A \mid B) = \frac{P(B \mid A) \cdot P(A)}{P(B)}. \tag{13}\]

Equation 14, Bayes’ Theorem with an Expanded Denominator. Expanding the probability of $B$ by the law of total probability over $A$ and its complement gives Bayes’ theorem in terms of conditional probabilities alone.

\[P(A \mid B) = \frac{P(B \mid A) \cdot P(A)}{P(B \mid A) \cdot P(A) + P(B \mid \neg A) \cdot P(\neg A)} \tag{14}\]

Equation 15, Central Limit Theorem. For large $n$ the sample mean is approximately normal with the population mean and a variance of $\sigma^2/n$, whatever the shape of the population.

\[\bar{X} \sim N\left(\mu, \frac{\sigma^2}{n}\right) \quad \text{approximately, for large } n. \tag{15}\]

Equation 16, Standard Error of the Mean. The standard deviation of the sample mean is the population standard deviation divided by $\sqrt{n}$.

\[SE_{\bar{x}} = \frac{\sigma}{\sqrt{n}}. \tag{16}\]

Equation 17, Z Statistic for a Mean with Known Variance. The distance of the sample mean from the hypothesised mean is measured in standard errors computed from the known $\sigma$.

\[Z = \frac{\bar{x} - \mu_0}{\sigma / \sqrt{n}} \tag{17}\]

Equation 18, t Statistic for a Mean with Unknown Variance. Replacing $\sigma$ with the sample standard deviation gives a statistic that follows the $t$-distribution with $n-1$ degrees of freedom.

\[t = \frac{\bar{x} - \mu_0}{s / \sqrt{n}} \tag{18}\]

Equation 19, Z Statistic for a Proportion. The sample proportion is compared with the null value $p_0$ in units of a standard error computed from $p_0$.

\[Z = \frac{\hat{p} - p_0}{SE_{\hat{p}}}, \quad SE_{\hat{p}} = \sqrt{\frac{p_0(1 - p_0)}{n}} \tag{19}\]

Equation 20, General Form of a Z Statistic. Any test statistic of this kind is the distance of a point estimate from its null value divided by the standard error of the estimate.

\[Z = \frac{\hat{\theta} - \theta_0}{SE_{\hat{\theta}}}. \tag{20}\]

Equation 21, General Form of a Confidence Interval. A confidence interval is the point estimate plus or minus the margin of error, which is the critical value multiplied by the standard error.

\[\hat{\theta} \pm z^\star \cdot SE_{\hat{\theta}}, \quad ME = z^\star \cdot SE_{\hat{\theta}} \tag{21}\]

Equation 22, Confidence Interval for a Mean with Known Variance. The interval about the sample mean is the normal critical value multiplied by $\sigma/\sqrt{n}$.

\[\bar{x} \pm z^\star_{\alpha/2} \cdot \frac{\sigma}{\sqrt{n}} \tag{22}\]

Equation 23, Confidence Interval for a Mean with Unknown Variance. The interval about the sample mean uses the $t$ critical value with $n-1$ degrees of freedom and the sample standard deviation.

\[\bar{x} \pm t^\star_{\alpha/2,\, n-1} \cdot \frac{s}{\sqrt{n}} \tag{23}\]

Equation 24, Confidence Interval for a Variance. The interval for the population variance divides $(n-1)s^2$ by the upper and lower chi-squared critical values with $n-1$ degrees of freedom.

\[\frac{(n-1)s^2}{\chi^2_{\alpha/2}} < \sigma^2 < \frac{(n-1)s^2}{\chi^2_{1 - \alpha/2}} \tag{24}\]

Equation 25, Confidence Interval for a Standard Deviation. Taking square roots of the bounds of the variance interval gives the interval for the standard deviation.

\[\sqrt{\frac{(n-1)s^2}{\chi^2_{\alpha/2}}} < \sigma < \sqrt{\frac{(n-1)s^2}{\chi^2_{1 - \alpha/2}}} \tag{25}\]

Equation 26, Confidence Interval for a Proportion. The interval about the sample proportion uses a standard error computed from the sample proportion itself.

\[\hat{p} \pm z^\star \cdot SE_{\hat{p}}, \quad SE_{\hat{p}} = \sqrt{\frac{\hat{p}(1 - \hat{p})}{n}} \tag{26}\]

Equation 27, Confidence Interval for a Difference of Means with Known Variances. For two independent samples the variances of the two means add, so the standard error is the square root of the summed $\sigma^2/n$ terms.

\[(\bar{x}_1 - \bar{x}_2) \pm z^\star_{\alpha/2} \cdot \sqrt{\frac{\sigma_1^2}{n_1} + \frac{\sigma_2^2}{n_2}} \tag{27}\]

Equation 28, Standard Error of a Difference of Proportions. For two independent samples the variances of the two proportions add under the square root.

\[SE_{\hat{p}_1 - \hat{p}_2} = \sqrt{\frac{\hat{p}_1(1 - \hat{p}_1)}{n_1} + \frac{\hat{p}_2(1 - \hat{p}_2)}{n_2}}. \tag{28}\]

Equation 29, Confidence Interval for a Difference of Proportions. The interval about the difference of the sample proportions is the normal critical value multiplied by the standard error of the difference.

\[(\hat{p}_1 - \hat{p}_2) \pm z^\star_{\alpha/2} \cdot SE_{\hat{p}_1 - \hat{p}_2}. \tag{29}\]

Equation 30, Pooled Proportion. Under the null hypothesis of equal proportions the two samples are combined into one estimate weighted by their sizes.

\[\hat{p} = \frac{\hat{p}_1 n_1 + \hat{p}_2 n_2}{n_1 + n_2} \tag{30}\]

Equation 31, Pooled Standard Error. The standard error for testing equal proportions uses the pooled proportion for both samples.

\[SE_{\text{pooled}} = \sqrt{\hat{p}(1 - \hat{p})\left(\frac{1}{n_1} + \frac{1}{n_2}\right)} \tag{31}\]

Equation 32, Sample Size for Estimating a Proportion. Solving the margin of error of a proportion for $n$ gives the smallest sample that meets a target margin, which the worst case $p = 0.5$ bounds when no prior estimate exists.

\[n \geq \frac{(z^\star)^2 \cdot p(1-p)}{ME^2} \tag{32}\]

Equation 33, Sample Size for Estimating a Mean. Solving the margin of error of a mean with known $\sigma$ for $n$ gives the smallest sample that meets a target margin.

\[n \geq \left(\frac{z^\star \cdot \sigma}{ME}\right)^2 \tag{33}\]

Notation

  • $n$, the sample size or number of trials
  • $k$, the number of successes or events
  • $p$, the probability of success on a single trial
  • $q$, the probability of failure, $q = 1 - p$
  • $\mu$, the population mean
  • $\sigma$, the population standard deviation
  • $\bar{x}$, the sample mean
  • $s^2$, the sample variance, whose square root $s$ is the sample standard deviation
  • $\hat{p}$, the sample proportion, a point estimate of $p$
  • $\lambda$, the average rate of a Poisson process
  • $\alpha$, the significance level
  • $Z$, the standard normal Z-score
  • $z^\star$, the critical value from the standard normal distribution
  • $t^\star$, the critical value from the $t$-distribution with the stated degrees of freedom
  • $\chi^2$, the chi-squared critical value
  • $SE$, the standard error
  • $ME$, the margin of error
  • $\mu_0$, the hypothesised population mean under the null hypothesis
  • $p_0$, the hypothesised population proportion under the null hypothesis
  • $\hat{\theta}$, a general point estimate, with null value $\theta_0$

References

Secondary Sources

General Sources

Tertiary Sources