A comprehensive glossary of terms for field epidemiologists

The probability of rejecting the null hypothesis when it is actually true (Type I error). Typically set at 0.05.

The probability of failing to reject the null hypothesis when it is actually false (Type II error). Statistical power = 1 - β.

A range of values that is likely to contain the true population parameter with a specified level of confidence (e.g., 95%).

The probability that a confidence interval contains the true population parameter. Common levels are 90%, 95%, and 99%.

The number of independent values that can vary in a statistical calculation. Often calculated as n-1 for a single sample.

A quantitative measure of the magnitude of a phenomenon, independent of sample size. Common measures include Cohen's d and Pearson's r.

The range of values above and below a sample statistic in a confidence interval. Represents the maximum expected difference between the sample and population parameter.

The arithmetic average of a set of values, calculated by summing all values and dividing by the count.

The middle value in a sorted dataset. Less affected by outliers than the mean.

A symmetric, bell-shaped probability distribution defined by its mean and standard deviation. Many statistical tests assume normality.

The hypothesis that there is no significant difference or effect. Statistical tests aim to reject or fail to reject this hypothesis.

The probability of obtaining test results at least as extreme as the observed results, assuming the null hypothesis is true. Values < 0.05 are typically considered statistically significant.

The probability of correctly rejecting a false null hypothesis. Calculated as 1 - β. Power of 80% or higher is generally desired.

The number of observations or units included in a study. Larger sample sizes generally provide more precise estimates.

A measure of the amount of variation in a set of values. Low values indicate data points are close to the mean.

The standard deviation of the sampling distribution of a statistic. Decreases as sample size increases.

The average of the squared differences from the mean. The square of the standard deviation.

A statistical test used to compare means among three or more groups. Tests whether group means differ significantly from each other.

A statistical test for categorical data to determine whether the observed frequencies differ significantly from expected frequencies.

A statistic measuring inter-rater agreement for categorical items, accounting for agreement by chance. Values range from -1 to 1.

A measure of the linear relationship between two variables. Ranges from -1 (perfect negative) to +1 (perfect positive).

A normality test that combines skewness and kurtosis to determine if data follows a normal distribution. Good for samples n ≥ 20.

The ratio of between-group variance to within-group variance in ANOVA. Larger values indicate greater differences between groups.

A normality test based on skewness and kurtosis. Best for large samples (n > 30). Common in econometrics.

A non-parametric alternative to one-way ANOVA for comparing medians of three or more independent groups.

A statistical method that models the relationship between a dependent variable and one or more independent variables using a linear equation.

A regression method for binary outcomes (0/1). Estimates the probability of an event occurring based on predictors.

A non-parametric test comparing two independent groups. Alternative to independent t-test when normality assumptions are violated.

A normality test that examines the correlation between data and normal scores. Most powerful for small samples (n < 50).

A statistical test comparing the means of one or two groups. Includes one-sample, independent, and paired variants.

ANOVA with two independent variables (factors). Tests main effects of each factor and their interaction effect.

A non-parametric test for paired samples. Alternative to paired t-test when normality assumptions are violated.

The area under the ROC curve. Ranges from 0.5 (no discrimination) to 1.0 (perfect discrimination).

The ratio of the probability of a test result in people with the disease to the probability in people without. LR+ > 1 indicates positive association.

The proportion of negative test results that are true negatives. Probability of not having the disease given a negative test.

The proportion of positive test results that are true positives. Probability of having the disease given a positive test.

Receiver Operating Characteristic curve. A plot of sensitivity vs (1-specificity) at various thresholds, used to evaluate diagnostic test performance.

The proportion of truly diseased persons correctly identified by the test (True Positive Rate). Sensitivity = TP / (TP + FN).

The proportion of truly non-diseased persons correctly identified by the test (True Negative Rate). Specificity = TN / (TN + FP).

A single statistic summarizing diagnostic test performance: Sensitivity + Specificity - 1. Used to find optimal cutoff points.

A measure of the 'tailedness' of a distribution. Positive kurtosis indicates heavy tails; negative indicates light tails.

A data point significantly different from other observations. May indicate measurement error or genuine extreme values.

A measure of asymmetry in a distribution. Positive skew means tail extends right; negative skew means tail extends left.

The proportion of disease cases among the exposed that can be attributed to the exposure. AF = (RR - 1) / RR.

The proportion of a population that develops a disease during a specific time period, often used during outbreaks.

The number of new cases of a disease occurring during a specified period in a population at risk.

The rate at which new cases occur in a population over time. Calculated as new cases divided by person-time at risk.

The number of patients who need to be treated to prevent one additional adverse outcome. NNT = 1 / Absolute Risk Reduction.

The ratio of the odds of exposure in cases to the odds of exposure in controls. Used in case-control studies.

A rate used for rare events occurring in a given time or space interval. Assumes events occur independently.

The proportion of a population with a disease at a specific point in time (point prevalence) or over a period (period prevalence).

The ratio of risk in the exposed group to risk in the unexposed group. RR > 1 indicates increased risk with exposure.

The absolute difference in risk between exposed and unexposed groups. Also called Attributable Risk or Absolute Risk Reduction.

Systematic error in study design, conduct, or analysis that leads to incorrect estimation of the association between exposure and outcome.

A randomization method ensuring equal allocation to groups within blocks of participants. Maintains balance throughout recruitment.

An observational study comparing people with a disease (cases) to those without (controls) to identify exposures associated with the disease.

An observational study following a group of people over time to determine how exposures affect outcomes.

When a third variable is associated with both exposure and outcome, distorting the apparent relationship between them.

A study examining a population at a single point in time to assess prevalence of outcomes and exposures simultaneously.

The extent to which study results can be generalized to other populations, settings, or conditions.

The extent to which a study establishes a trustworthy cause-effect relationship, free from systematic errors.

The process of randomly assigning study participants to different groups to minimize selection bias and confounding.

Bias arising from the way participants are selected or assigned, resulting in systematic differences between comparison groups.

Basic random assignment where each participant has equal probability of being assigned to any group, like flipping a coin.

Randomization within strata defined by baseline characteristics to ensure balanced distribution of important variables.

A graphical representation of data distribution using bars where height represents frequency of values in each interval.

A graphical display showing data distribution through quartiles, median, and outliers. Also called box-and-whisker plot.

A statistical test that makes assumptions about the underlying distribution of data (usually normal distribution).

A statistical test that makes no assumptions about the underlying distribution of data. Used when normality cannot be assumed.